About Pearson

Pearson is the world’s learning company, with presence across 70 countries

worldwide. Our unique insights and world-class expertise comes from a long
history of working closely with renowned teachers, authors and thought leaders,
as a result of which, we have emerged as the preferred choice for millions of
teachers and learners across the world.
We believe learning opens up opportunities, creates fulfilling careers and hence
better lives. We hence collaborate with the best of minds to deliver you class-
leading products, spread across the Higher Education and K12 spectrum.
Superior learning experience and improved outcomes are at the heart of
everything we do. This product is the result of one such effort.
Your feedback plays a critical role in the evolution of our products and you can
contact us - [email protected]. We look forward to it.

A01_POWER-SYSTEM-AN00_SE_XXXX_FM.indd 1 6/2/2018 3:55:04 PM

Thispageisintentionallyleftblank
 Power System Analysis
A Dynamic Perspective

K. N. Shubhanga
Department of Electrical Engineering
National Institute of Technology Karnataka
Karnataka, India

A01_POWER-SYSTEM-AN00_SE_XXXX_FM.indd 3 6/2/2018 3:55:04 PM

 Copyright © 2018 Pearson India Education Services Pvt. Ltd

Published by Pearson India Education Services Pvt. Ltd, CIN: U72200TN2005PTC057128,

formerly known as TutorVista Global Pvt. Ltd, licensee of Pearson Education in South Asia.

No part of this eBook may be used or reproduced in any manner whatsoever without the publisher’s
prior written consent.

This eBook may or may not include all assets that were part of the print version. The publisher
reserves the right to remove any material in this eBook at any time.

ISBN 978-93-325-8466-2
eISBN 9789353063696

Head Office: 15th Floor, Tower-B, World Trade Tower, Plot No. 1, Block-C, Sector 16, Noida 201
301, Uttar Pradesh, India.
Registered Office: 4th Floor, Software Block, Elnet Software City, TS 140, Block 2 & 9, Rajiv
Gandhi Salai, Taramani, Chennai - 600 113, Tamil Nadu, India.
Fax: 080-30461003, Phone: 080-30461060
www.pearson.co.in, Email: [email protected]

A01_UGC-NET-TEACHIN_XXXX_FM.indd 1 2/23/2018 5:14:50 PM

 To My Parents,
Teachers and Students

A01_POWER-SYSTEM-AN00_SE_XXXX_FM.indd 5 6/2/2018 3:55:04 PM

Thispageisintentionallyleftblank
 Brief Contents

Foreword xvii
Preface xix
About the Author xxiii

Chapter 1 Introduction to Power System Analysis 1

Chapter 2 Basics of Power Systems 22
Chapter 3 Park-Based Transformations 113
Chapter 4 Synchronous Machine Modelling Using Primitive Parameters 143
Chapter 5 The Standard Parameters of Synchronous Machine 199
Chapter 6 Numerical Integration of ODEs 288
Chapter 7 Numerical Iterative Methods 334
Chapter 8 Fault Analysis of Power Systems 350
Chapter 9 Introduction to Sub-synchronous Resonance 419
Chapter 10 SSR Analysis of the IEEE First Benchmark Model 437
Chapter 11 Controllers for Synchronous Generator 476
Chapter 12 Power System Angle Stability 506
Chapter 13 Modal Analysis of Power Systems
with Interconnected Generators 557
Chapter 14 Transient Stability Analysis of Power Systems
with Interconnected Generators 632
Chapter 15 Dynamic Modelling of Some Electrical Machines
and their Power-flow Analysis 702

Index 779

A01_POWER-SYSTEM-AN00_SE_XXXX_FM.indd 7 6/2/2018 3:55:05 PM

Thispageisintentionallyleftblank
 Contents

Foreword xvii
Preface xix
About the Author xxiii

1. Introduction to Power System Analysis 1

1.1 Performance of Large Power Systems and their Analysis 1
1.2 Classification of Power System Stability Problems 2
1.3 Power System Stability Analysis—Computational Issues 4
1.4 Power System Stability—Modelling Issues 5
1.4.1 Modelling of Dynamical Systems 6
1.5 Electromechanical Energy Conversion 9
1.5.1 Behaviour of an Inductor with Fixed Armature Core 9
1.5.2 Behaviour of an Inductor with Movable Armature Core 10
1.5.3 Primitive Machine with Rotating Armature without Rotor Coil 12
1.5.4 Primitive Machine with Rotating Armature with a Rotor Coil 14
References 20
Review Questions 21

2. Basics of Power Systems 22

2.1 Sinor Waveform for Conventional Power Generation 22
2.2 Choice of Frequency for Power Generation 23
2.3 Concept of Phasor Analysis in AC Sinusiodal Systems 24
2.3.1 Phasor Representation 25
2.4 Power in Single-phase Circuits 33
2.4.1 Instantaneous Power in a Pure Resistor Circuit 33
2.4.2 Nature of Powers in a RL-series Circuit 34
2.4.3 Nature of Powers in a RC-series Circuit 40
2.4.4 Real Power Supply–Importance of Reactive Powers 46
2.4.5 Complex Number Representation of Real and Reactive Powers 54
2.4.6 Power Balance in Single-phase Circuits 57
2.5 Power in Three-phase Systems 64
2.5.1 Interconnection of Two Sources–Three-phase Systems 65
2.5.2 Analysis of Real and Reactive Powers in Three-phase Systems 70
2.5.3 Interpretation of Synchronous Operation in Terms of Power Angle 73
2.5.4 Complex Power Calculations in Three-phase Balanced Systems 74
2.5.5 Single-Line Diagrams of Three-phase Systems 75
2.6 Per-unit Representation 76
2.6.1 Procedure Employed for Per-unit Representation 76
2.6.2 Per-unit Model of Transformers 80

A01_POWER-SYSTEM-AN00_SE_XXXX_FM.indd 9 6/2/2018 3:55:05 PM

 x | Contents

2.6.3 Example: A Sample Power System to Demonstrate Per-unit Calculations  83

2.6.4 Per-unit Model of Transformers with Off-nominal Voltage Ratings  89
2.7  Characteristics of a Typical Power System  92
2.7.1  Characteristics of Generator and Primemover Systems  92
2.7.2  Characteristics of Transmission and Distribution Systems  97
2.7.3  Interconnected Operation of Power Systems  102
2.7.4  Characteristics of Loads  103
References 108
Review Questions  110

3.  Park-Based Transformations 113

3.1 Time-varying Parameters of a Synchronous Machine  113
3.1.1  Derivation of a Time-varying Transformation  114
3.1.2  Transformation Matrices in Different Reference-frames  118
3.1.3  Some Properties of the Transformation Matrices  119
3.1.4  A Change of Reference for Angle Measurement  123
3.1.5  An Alternate Form of Park Transformation  123
3.1.6  An Example with Park Transformation  125
3.1.7 Power-variant Park Transformation 126
3.2 Three-Phase-Based Phase-Locked Loop  129
3.2.1  Transformation from abc-frame to Synchronous-frame  129
3.2.2 Harmonic Oscillator 130
3.2.3 Some Results 131
3.3 Representation of a Transmission Line in Machine-Frame  132
3.3.1  Derivation of Capacitor Voltages in 0dq-frame 137
3.3.2  Summary of Equations in dq-frame 137
3.3.3  Summary of Equations in DQ-frame 138
References 139
Review Questions  139

4. Synchronous Machine Modelling Using Primitive Parameters 143

4.1 Modelling of a Non-salient Pole Synchronous Machine  143
4.1.1 Inductance Matrix for a Non-salient Pole Synchronous Machine  145
4.2  Modelling of a Salient Pole Synchronous Machine  146
4.2.1  Determination of Inductances of Stator Coils  147
4.2.2 Determination of Mutual-inductances Between Stator and Rotor Coils  150
4.2.3  Determination of Self- and Mutual-inductances of Rotor Coils  151
4.2.4  Voltage Equation Using the Generator Convention  152
4.2.5  Elements of Resistance Matrix  153
4.3 Flux Linkage-current Equations in Rotor-reference Frame  153
4.3.1  Determination of [LB] Matrix  155
4.4  Voltage Equations in Rotor-reference Frame  158
4.5  Expression for Torque in Rotor-reference Frame  161
4.6 Case Studies 162
4.6.1  Voltage Build-up on Open-circuit  162
4.6.2 Steady-state Short-circuit 164

A01_POWER-SYSTEM-AN00_SE_XXXX_FM.indd 10 6/2/2018 3:55:05 PM

 Contents  |  xi 

4.6.3  Natural Response Under Short-circuit  167

4.7  Rotor Mechanical Equation  184
4.8  Analysis of Linear Systems  185
4.8.1 Eigenvalue Analysis 185
4.8.2  Modal Analysis of Linear Systems  187
4.8.3  Solution of an LTI System  190
References 195
Review Questions  196

5.  The Standard Parameters of Synchronous Machine 199

5.1 Requirement of Generator Modelling Neglecting Stator Transients  200
5.1.1  Short-circuit Analysis with Stator Transients Neglected  200
5.1.2  Type of Standardised Synchronous Machine Models  202
5.2 Operational Inductance Approach 202
5.2.1 Natural Response Under Short-circuit Condition Using OI Model  205
5.3  Per-unit Representation of Generator Quantities  207
5.3.1 Base Quantities 208
5.3.2  Per-unit Stator Voltage Equations  209
5.3.3  Flux-linkage Expressions in per-unit for 1.1 Model  209
5.3.4 Derivation of a State-space Model for d-axis Rotor
Circuits without Damper Winding  210
5.3.5  Derivation of State-space Model for q-axis Rotor Circuit  216
5.4  Summary of Equations Pertaining to 1.1 Model  218
5.5  A Case Study: Voltage Build-up on No-load  220
5.6  Operational Impedance for 2.2 Model  221
5.6.1  Operational Impedance for d-axis 222
5.6.2  Operational Impedance for q-axis  227
5.7  Summary of Equations Pertaining to 2.2 Model  230
5.8 Swing Equation 232
5.8.1  Swing Equation in Form-1  233
5.8.2  Swing Equation in Form-2  235
5.8.3  Swing Equation in Form-3  235
5.8.4  Summary of Different Forms of Swing Equations  236
5.9  Steady-state Operation of Synchronous Generator  237
5.9.1 Graphical Approach to Determine the Rotor-angle in Salient-pole Machines  240
5.9.2 Example: Computation of Field Voltage Ef d 0 in a Salient-pole Machine  242
5.9.3  Expression for Developed Torque Under Steady-state  242
5.10  Steady-state Analysis: Space-phasor Approach  244
5.10.1  Space-phasor Due to Armature Currents-Revolving Field  245
5.10.2 Space-phasor Diagram for Round Rotor Synchronous Machine  248
5.10.3  Space-phasor Diagram for Salient-pole Synchronous Machine  250
5.11  Modification of 2.2 Model and Equivalent Circuits  251
5.11.1  Observations with Reduced-order Models  252
5.11.2  Dynamic Equivalent Circuits for Generators  253
5.12  Initial Condition Calculations  255
5.12.1  Example: Initial Condition Evaluation in an SMIB System  257

A01_POWER-SYSTEM-AN00_SE_XXXX_FM.indd 11 6/2/2018 3:55:05 PM

 xii | Contents

5.13 Example: Eigenvalue Analysis of a Generator Under Short-circuit Condition  259

5.13.1  Short-circuit Analysis Considering Stator Transients  260
5.13.2  Short-circuit Analysis Neglecting Stator Transients  267
5.14 Example: Synchronisation of a Standalone Generator to Mains  272
5.15 Example: Transient Stability Simulation of an Smib System  275
5.15.1  SMIB System without AVR  276
5.15.2  SMIB System with AVR  278
References 283
Review Questions  284

6.  Numerical Integration of ODEs 288

6.1 System of Differential Equations: Some Observations  288
6.2 Classification of Numerical Integration Algorithms  290
6.2.1  Taylor Series-based Methods   290
6.2.2  Single-step and Multi-step Methods  291
6.2.3  Self Starting and Non-self Starting Methods  291
6.3 Accuracy and Stability of Numerical Integration Methods  292
6.3.1  Accuracy of Numerical Integration Methods  292
6.3.2 Stability of Numerical Integration Methods Through Eigenvalue Analysis  293
6.4 Demonstration of Some Numerical Integration Methods  293
6.4.1  Forward Euler Method  293
6.4.2  Backward Euler Method  299
6.4.3  Trapezoidal Method  303
6.4.4  Runge-Kutta (RK) Fourth Order Method  304
6.4.5  Variable-step Methods  308
6.5  Example: Solution of Swing Equation  311
6.5.1 Initial Condition Calculation for the Machine Variables and System Equations  314
6.5.2 Solution of Swing Equations by Using Forward Euler Technique  316
6.5.3 Solution of Swing Equations by Backward Euler and RK-4 Techniques  320
6.5.4 Evaluation of Critical Clearing-angle Using Equal-area Criteria  324
6.5.5 Large-signal Stability Evaluation Using Energy Function Method  326
References 331
Review Questions  332

7.  Numerical Iterative Methods 334

7.1 Features of Non-linear Algebraic System of Equations  334
7.2 Fixed-Point Iteration Method  335
7.2.1  Example: A Scalar Function  335
7.2.2  Example: Power Flow in a Simple Power System-1  337
7.2.3  Example: Power Flow in a Simple Power System-2  338
7.2.4  Example: Simultaneous Non-linear Equation  340
7.3  Gauss Seidel Iteration Method  341
7.4  Newton-Raphson Iteration Method  341
7.4.1  Iterative Function for a Scalar System  341
7.4.2  Iterative Function for a System with n Variables  342

A01_POWER-SYSTEM-AN00_SE_XXXX_FM.indd 12 6/2/2018 3:55:05 PM

 Contents  |  xiii 

7.4.3 Example: Simultaneous Non-linear Equation with Newton-Raphson Method  343

7.4.4 Example: Power Flow in a Simple Power System-1 (Newton-Raphson Method)  344
7.4.5 Example: Power Flow in a Simple Power System-2 (Newton-Raphson Method)  346
References 348
Review Questions  348

  8.  Fault Analysis of Power Systems 350

8.1  Introduction to Fault Analysis  350
8.2 Symmetrical Three-phase Short-Circuit Analysis 351
8.2.1  Two-machine, Five-bus Power System  352
8.2.2 Analysis of Fault Current Using Thevenin’s Theorem  355
8.2.3 Calculation of Bus Voltages in a Faulted System  360
8.2.4 Bus Voltage Calculation without the Knowledge of the Fault Current  364
8.2.5 Calculation of System Quantities During a Fault  365
8.2.6 Effect of Synchronous Machine Models on the Fault Current  367
8.2.7 Symmetrical Fault Calculations Accounting Pre-fault Load Currents  368
8.2.8 Short Circuit Capacity at a Bus  371
8.3  Analysis of Unsymmetrical Faults  373
8.3.1 Calculation of Sequence Voltages and Currents  375
8.3.2 Sequence Impedances of Y- and ∆-connected Passive Elements  379
8.3.3 Sequence Impedances of Synchronous Machines  384
8.3.4 Sequence Impedance of Transmission Lines  385
8.3.5 Sequence Impedance of Transformers  390
8.3.6 Example: Sequence Networks for a 2-machine 5-bus Power Systems  394
8.3.7 Analysis of Unsymmetrical Short-circuit Faults  397
8.3.8  Transient Stability Analysis for Unsymmetrical Shunt Faults  410
References 416
Review Questions  416

  9.  Introduction to Sub-synchronous Resonance 419

9.1 Sub-synchronous Resonance 419
9.2  Ssr Study of a Simplified System  420
9.2.1  Turbine-generator Mechanical Systems  420
9.2.2 Electrical Systems 422
9.2.3  System Parameters and Operating Conditions  424
9.2.4  Analysis of Only Electrical Systems  424
9.2.5  Analysis of Partial Systems  425
9.2.6  Analysis of the Complete Systems  426
9.2.7  Eigenvalue Analysis of the Complete System  430
References 435
Review Questions  436

10. Ssr Analysis of the Ieee First Benchmark Model437

10.1 Turbine-Generator Mechanical System Equations  437

A01_POWER-SYSTEM-AN00_SE_XXXX_FM.indd 13 6/2/2018 3:55:05 PM

 xiv | Contents

10.1.1 The IEEE First Benchmark System and Modal Frequencies  439

10.1.2  Linearisation of Mechanical System of Equations  440
10.2 Generator Modelling 444
10.2.1 Stator and Rotor Equations  444
10.2.2 Linearisation of Te  445
10.2.3  Linearisation of Electrical Equations  446
10.3 Exciter Modelling 449
10.3.1  Single-time Constant Static Exciter  449
10.3.2 Derivation of ∆Vg in terms of vQg and vDg Components  450
10.4 Transmission Network Modelling  451
10.4.1  Derivation of Capacitor Voltage Equations  452
10.4.2  Derivation of Line Voltage Equations  455
10.5  Listing of Linearised State Equations  457
10.5.1  Interfacing of the Network to the Generator  458
10.6 Machine-frame-based Time-domain Simulation of Ssr 461
10.6.1  Computation of vgd and vgq in Simulation  461
10.6.2  Initial Condition Calculations  464
10.6.3  Case Studies and Modal-speed Calculations  467
References 474
Review Questions  475

11.  Controllers for Synchronous Generator 476

11.1 Real and Reactive Power Controllers for a Synchronous Generator  476
11.2 Functions and Types of Excitation Systems  477
11.2.1  DC Excitation Systems  479
11.2.2  AC Excitation Systems  481
11.2.3  Static Excitation Systems  484
11.2.4  State-space Model of Some System Functions  487
11.2.5  Initial Condition Calculations for Exciters  490
11.3  Prime Mover Controllers  492
11.3.1  Influence of Prime Mover Controllers on Load Sharing  493
11.3.2  Model of Hydraulic Turbines  495
11.3.3  Model of Steam Turbines  497
11.3.4  Modelling of Speed-Governing Systems  498
11.4 Windup and Non-windup Type Limiters on Integrator Blocks  500
11.4.1  Windup Type Limiter on Integrator Blocks  501
11.4.2  Non-windup Type Limiter on Integrator Blocks  501
References 502
Review Questions  503

12.  Power System Angle Stability 506

12.1 Relative-Angular Stability Analysis 506
12.1.1  Small-signal Stability Analysis of SMIB System  507
12.1.2  A Mechanical Analogy  510
12.2  Synchronising and Damping Torque Analysis  512

A01_POWER-SYSTEM-AN00_SE_XXXX_FM.indd 14 6/2/2018 3:55:05 PM

 Contents  |  xv 

12.3 Effect of a Fast-acting High-gain Static Exciter on smib System  514

12.3.1  PSS Location and its Design  515
12.4  Small-signal Stability Analysis of an Smib System  521
12.4.1 Example: Small-signal Stability of SMIB System with 0.0 Model for Generator  521
12.4.2 Example: Small-signal Stability of SMIB System with 1.0 Model for Generator  524
12.4.3 Example: Small-signal Stability of SMIB System with 2.2 Model for Generator  528
12.4.4  Modal Performance with AVR and PSS  546
References 553
Review Questions  554
13. Modal Analysis of Power Systems
with Interconnected Generators 557
13.1 Importance of Modal Analysis of Power Systems  557
13.2 Power System Oscillations  558
13.2.1 Classification of Power System Oscillation  559
13.2.2 Analysis of Small-signal Stability  560
13.3 Spring-mass System Example  561
13.3.1 Removal of Redundancy of a State  564
13.3.2 Case-1: With External F1 Without Any Damping  565
13.3.3 Case-2: With External F3 Without Any Damping  567
13.3.4 Case-3: With External F3 With Damping B1  568
13.3.5 Case-4: With External F3 as a Step-signal Without Any Damping  570
13.3.6 Case-5: With External F3 as a Step-signal With Damping B1  571
13.3.7 Performance Analysis in the COI-reference-frame  572
13.4 Linearisation of Power System Modelling Equations  578
13.4.1  Current-Injection Approach  580
13.5 Participation Matrix  602
13.5.1 Determination of Nature of Oscillatory Modes  604
13.6 Modal Analysis of a Two-machine Power System  605
13.6.1 Base Case-Without Turbine  605
13.6.2 A Reheat Turbine Enabled on Machine-1  610
13.7 Modal Analysis of Four-machine Modified Power System  612
13.7.1 Base Case Loading Condition  613
13.7.2 Base Case without Excitation Controllers  617
13.7.3 Base Case with Reduced Loading Condition  617
13.7.4 Base Case with Power System Stabiliser  621
References 627
Review Questions  629
14. Transient Stability Analysis of Power Systems
with Interconnected Generators 632
14.1  Interfacing Generator to Network  632
14.1.1  Influence of Dynamic Saliency of Generator  633
14.1.2 Dummy-coil Approach 636
14.1.3  Generator Source Current Calculations  637
14.1.4  Modelling of Network Elements  640

A01_POWER-SYSTEM-AN00_SE_XXXX_FM.indd 15 6/2/2018 3:55:05 PM

 xvi | Contents

14.2  Centre of Inertia Reference  642

14.2.1 Common-mode Frequency Calculation (Frequency Stability Analysis)  642
14.2.2  Speed and Rotor-angle Calculations with respect to COI  644
14.2.3  Swing Equations in COI-reference  645
14.3 Structure of Power System Equations and its Solution  646
14.3.1  Simultaneous Implicit Solution  647
14.3.2 Partitioned Solution 647
14.4 Load Modelling 648
14.4.1  Load Model Classification  648
14.4.2  Polynomial Load Representation  649
14.4.3  Frequency-dependent Load Models  651
14.5  Load Equivalent Circuit  652
14.5.1  Modification of Constant Power-type Load Characteristics  652
14.5.2 An Approach to Avoid Iterative Solution of Algebraic Equations  653
14.6  A Summary of the Implementation Procedure  654
14.7 Demonstration of System Frequency with Two-machine Power System  656
14.7.1  Three-phase Fault without Frequency-dependent Loads  658
14.7.2  Three-phase Fault with Frequency-dependent Loads  659
14.7.3  Line Trip without Frequency-dependent Loads  659
14.7.4  Line Trip with Frequency-dependent Loads  661
14.7.5 Line Trip, Speed-governor on M-1 Enabled without Frequency-dependent Loads  663
14.7.6 Line Trip, Speed-governor on M-1 Enabled with Frequency-dependent Loads  664
14.7.7 Line Trip, Speed-governors on M-1 and M-2 Enabled
with Frequency-dependent Loads  666
14.8  Large Disturbance Performance  670
14.8.1  Two-machine Power System  673
14.8.2  Four-machine, 10-bus Power Systems  681
14.8.3  50-machine, 145-bus the IEEE Power Systems  690
References 698
Review Questions  699

15. Dynamic Modelling of Some Electrical Machines

and their Power-flow Analysis 702
15.1 Induction Motor Model  702
15.1.1 Fifth Order Induction Motor Model  703
15.1.2 Per-unit Representation of the Induction Motor Model  704
15.1.3 Reduced Order Model for the Induction Machine  718
15.2 Modelling of dc Motor-driven Synchronous Generator  727
15.2.1 DC Motor Modelling  728
15.2.2 DC Motor Driving a Synchronous Generator  731
15.2.3 Synchronisation of Two DC Motor-driven Synchronous Generators  743
15.2.4 Synchronisation of a DC Motor-driven Generator to Mains Supply  766
References 777
Review Questions  777

Index 779

A01_POWER-SYSTEM-AN00_SE_XXXX_FM.indd 16 6/2/2018 3:55:05 PM

 Foreword

Although electrical power systems have been in existence for more than a century, challenges in this
field abound. While the structure of a bulk power system has remained more or less unchanged, the
deployment of emerging technologies like wind and solar energy systems, voltage-source (power elec-
tronic) converters, and wide-area measurement systems have had a significant impact on the operation
and control of power systems. An abiding concern of a power system engineer has been the stability
of the electrical grid, that is, its ability to withstand a disturbance and return to an acceptable equilib-
rium. Although this is stated quite simply, complex modelling, analytical and computational tools are
required to analyse, quantify, and improve stability margins.
Given the importance of maintaining stability, a good grounding in this subject is necessary for a
power system engineer. In most undergraduate curricula in electrical power engineering, students are
generally given only a brief introduction to power system stability in their first course on Electrical
Power Systems. This is often their only exposure to power system dynamics and is by itself inadequate
to analyse the nuances of the stability problem. Therefore, a follow-up course at the senior undergrad-
uate level or the postgraduate level is necessary. This course should familiarise the student to model-
ling in the d-q domain, which is not only required for machine modelling but also for understanding
the vector-control of power electronic converters. Small-signal dynamic analysis and numerical inte-
gration methods, which are the necessary tools for analysis of power system dynamic phenomena,
should also be covered in such a course.
I think that this book, Power System Analysis: A Dynamic Perspective, serves precisely as a bridge
between the undergraduate course on power systems and the complex modelling and computational
tools that are used for the dynamic analysis of practical systems.
Dr Shubhanga has worked in the area of power system dynamics for more than two decades. He has
vast experience in teaching various power system subjects, as well as in the development of laboratory
experiments and computer programmes to demonstrate concepts related to power system dynamic
phenomena. It is appropriate that he has taken on this task of organising his teaching material in the
form of this book.
Therefore, I am delighted to write the foreword for this book. I think the book will be a very useful
text and reference book for students of electrical power system engineering.
 
5th July 2017 A. M. Kulkarni
IIT Bombay

A01_POWER-SYSTEM-AN00_SE_XXXX_FM.indd 17 6/2/2018 3:55:05 PM

Thispageisintentionallyleftblank
 Preface

Power systems are characterised by a large interconnection of complex dynamic elements such as syn-
chronous machines, transformers, transmission lines, loads, network devices, etc. It is noted that even
for a simple one-machine power system, it is not easy to carry out power system analysis without the
support of a programming platform. In this regard, this book illustrates various issues related to mathe-
matical modelling of components, interfacing of elements, and assembling of equations to finally solve
them efficiently using computational techniques.
The content of the book has evolved as a result of continued teaching of the subject both for under-
graduates and postgraduates for a period of more than 15 years. Although many books have been
published in this area, it is felt that it is not easy to teach power system dynamics and related topics to
students in a class-room environment as the study requires many pre-requisites such as linear systems
theory, control theory, numerical methods, basics of power systems, etc. In an attempt to review these
areas, many examples and time-domain simulations are presented through dedicated chapters. The
main objective of this book is to cover dynamic aspects of power system analysis (in the frequency
range below power frequency), giving a thorough insight into their origin and manifestation.
To introduce the system-dynamic aspects, the generalised theory of machines has been presented.
A two-coil primitive machine is taken up to illustrate the concept of time-varying parameters in rotat-
ing electrical machines.
Both electrical and mechanical equations are given to show the nature of an energy conversion
device. These are highlighted both for motor and generator operation, which is generally skipped while
introducing the topic. The book presents power systems basics such as sinusoidal generation/operation
and control while introducing the phasor-analysis tool as a special case of the state-space analysis. In
the second chapter, the state-space modelling of dynamical systems is dealt in a tutorial manner to pre-
pare the students to take-up advanced modelling of power system components. All concepts have been
developed from fundamentals so that these topics can be introduced at a lower-semester level.
Before dynamic analysis of a power system is taken up, linear analyses techniques are briefed through
many examples including eigenvalue analysis. To appreciate the solution of ordinary differential equations
and hence to carry out time-domain simulations, in Chapter 6, a detailed introduction to numerical inte-
gration techniques is presented covering most of the popular methods giving their relevance in MATLAB/
SIMULINK software. Even the concept of variable-step numerical integration method is discussed. Chapter 7
is dedicated to cover numerical iterative techniques to obtain solution of non-linear algebraic equations. These
provide a complete base for handling load flow, fault analysis and complex stability problems.
In order to introduce the synchronous machine modelling and power system analysis with synchronous
machines, a reduced order model is derived first, in a systematic manner, before an industry-grade model is
discussed. These are covered in Chapters 4 and 5. While deriving the dynamic models, the classical transfor-
mation theory such as the Park transformation is neatly developed in Chapter 3. The generality of this trans-
formation in establishing all other transformations found in the literature, is briefed. The application of such
transformations in building a phase-lock loop (PLL) is also discussed. Many case studies and examples
about synchronous generators are included so that students can easily implement the model and reproduce
the results. This motivates self-learning, and students can design new tasks based on the understanding.
In some cases, sample plots are also shown which are acquired in the hardware lab. Having covered
the dynamic model of a synchronous machine, the steady-state models are easily derived as a special
case of the dynamic model. This, in turn, is directed to initial condition calculation, demonstrating the
importance of such a computation in dynamic studies.

A01_POWER-SYSTEM-AN00_SE_XXXX_FM.indd 19 6/2/2018 3:55:05 PM

 xx | Preface

In Chapter 8, power system fault analysis is presented with better insight to address system-level
studies, giving due importance to computational aspects. This coverage not only helps students to
learn about conventional topics, but also provides a logical continuity to model network disturbances
in the stability studies both for symmetrical and unsymmetrical faults.
Before introducing different aspects of power system stability analyses, especially related to low
frequency power oscillations, a dedicated chapter is included on subsynchronous resonance (SSR) to
demonstrate its origin in a simple power system. This tutorial chapter clarifies many issues with power
system stability analyses, in general. In the following chapter, i.e., in Chapter 10, the SSR analysis in
a slightly complicated the IEEE first benchmark system, is presented. This is found to provide a good
introduction to the conventional stability analyses as the power swing-related analysis is much more
simpler compared to the SSR analysis.
While introducing the power swing stability analyses in Chapter 12, a thorough insight is provided
to a beginner about different types of power system stability, like small-signal stability, transient sta-
bility and frequency stability. A mechanical analogy is also given to distinguish these types of stability
and its implications on the tools and techniques, and their significance. Many examples are presented
to demonstrate the analysis method in simple power systems such as the single-machine connected
to infinite bus (SMIB) systems. The effect of generator controllers such as primemover and excitation
controllers are clearly explained. All relevant equations are listed, and analysis is carried out using
vector and matrix manipulations so that students are driven to develop their own programme codes in
MATLAB. The expected sample results are also plotted which can be used for verification. In some
cases the captured plots on a lab machine are also shown.
In Chapters 13 and 14, small-signal and transient stability analyses are carried out with intercon-
nected generators. A detailed analytical and time-domain simulation analyses are carried out for a lin-
ear multi-mass spring system to illustrate relative and common-mode oscillations and their prediction.
These observations are used to get better understanding about power systems’ modal behaviour. By
taking up a two-machine power system, even frequency stability issues are covered which is generally
avoided. In this connection, the concept of centre-of-inertia (COI) variables are also explained. Using
these variables, the characteristics of different types of system loads are discussed.
Dynamic models of additional rotating machines such as induction machines, DC machine, and
DC-motor driven synchronous machines (MG sets) are discussed in Chapter 15 to understand many
hardware experiments conducted in a typical machines lab. The chapter illustrates many examples
such as loading test on a synchronous generator, synchronisation of two MG sets, synchronisation to
mains, etc., through time-domain simulation of modelling equations.
A major strength of the book is that all fundamental concepts have been explained through simple
rigorous mathematical analysis with extensive time-domain simulation results. A number of plots
have been given along with necessary elaboration in simple language for easy understanding. For
most of the examples and exercises, the associated script and model files developed in MATLAB/
SIMULINK will be provided (through website) so that they will be useful for the beginners. A mul-
ti-machine power system stability analysis programme (developed in-house) will be made available
for free-download (along with user manual) so that many examples in the book can be worked out.
These programmes cover both the small-signal and large-signal stability analyses. In the programme,
many IEEE standard-type exciters, primemovers and power system stabilisers are implemented. Many
speaking tutorials have been generated using this package, which are available on Youtube website
which will augment the class-room teaching.
The power system engineering related topics covered in the book can be used as a resource material
for teachers, UG/PG/research students, as well as utility personnel.

A01_POWER-SYSTEM-AN00_SE_XXXX_FM.indd 20 6/2/2018 3:55:05 PM

 Preface  |  xxi 

Acknowledgments
Prof. A. M. Kulkarni, Department of Electrical Engineering, IIT Bombay, India, has been a constant
source of inspiration for me through all the process of the work. His continued support and encourage-
ment guided me to conceive and draft the contents of different chapters. I also thank him for writing
the foreword for the book.
Prof. K. R. Padiyar has been my grand-guru.
The Department of Electrical Engineering, National Institute of Technology Karnataka, Surathkal,
Mangalore, India, has provided a conducive academic environment for compiling the book. I would like to
express my heartfelt thanks to my first Ph.D. student, Dr Shashidhara whose work has been utilised in some
chapters. I also thank him for his strong advise to take up book publication rather than going for open-source.
Many of my graduate students’ project work has contributed to some of the case studies presented in the
book. In particular, I wish to thank Surendra, Sreenad, Deepak, E. Prasanthi, Hema Latha, Krishna Rao, and
Santosh V. Singh, and an UG student B. Shwetha. In addition, a few students, Gajanana, Rashmi, and Teena,
helped me in preparing the manuscript and in drawing figures. The book could not have been completed
without the help and support of many friends, colleagues, teachers, staff and the head of the department.
I would like to acknowledge the support and encouragements of Prof. K. P. Vittal, Dr G. S. Punekar,
Dr Murigendrappa (of Mechanical Engineering Department), Mr. I. R. Rao, and Mr. K. Nagaraja Bhat.
Ms. R. Dheepika, Mr. Sojan Jose and Mr. M. Balakrishnan of M/s. Pearson India Education
Services were very helpful throughout the work.
The book would never have been completed without the love and support of my family members
and relatives. I would like to place on record the love and moral support of my wife Shobha PS,
daughter, Shreya SA, and my mother. I am also grateful to my uncle, Mr. B. Mohana Acharya for
his encouragement through out my academic career. I am grateful to almighty for giving me enough
strength to complete the book.

K. N. Shubhanga

A01_POWER-SYSTEM-AN00_SE_XXXX_FM.indd 21 6/2/2018 3:55:05 PM

Thispageisintentionallyleftblank
 About the Author

K. N. Shubhanga has been with the Department of Electrical Engineering, National Institute of
Technology Karnataka, Surathkal, Mangalore, India, as a faculty member, for more than 22 years.
He received the B.E. degree in Electrical Engineering and the M.Tech. degree in power systems
from Mangalore University, India, in 1991 and 1994, respectively, and Ph.D. degree from the Indian
Institute of Technology Bombay, in 2003. His research interests are in the areas of FACTS and power
system dynamics. He has setup a Power System lab in the department where a scale-down model of
a 4-machine power system has been built. This lab has provisions to inter-connect four DC-motor
driven synchronous generators so as to illustrate two-area power system operation and control and
to acquire real-time dynamic signals of machines. In this lab, UG/PG core course-related exper-
iments are conducted to augment the classroom theoretical concepts. Even in hardware lab such
as electrical machines, lab weightages are given to simulation and programming exercises. To sup-
port this, he has developed many power system analysis and simulation packages, in-house. Details
about the lab setup and other resource materials are available on Google-Drive on the web address:
https://goo.gl/l2jnld. Some speaking tutorials (MatSim phase-1) related to power system dynamics
are available in the YouTube website.
He is a senior IEEE member and a life member of the IEE (India). He has guided many UG/PG
major and minor projects and has involved in research-level activities as well. This effort has resulted
in the publication of more than 40 research articles which are indexed in the IEEE digital library.
Further, his Ph.D. and PG students have acquired prestigious POSOCO Power Systems Awards in
the recent time. He has been teaching subjects such as electrical machines, power systems, power
electronic applications to power systems, numerical methods, linear systems, etc. to both UG and
PG students. He has presented guest-lectures at various academic institutes and has been a regular
reviewer for the IEEE journals and international conferences. He has also held administrative offices
both at the institute and at the department levels.

A01_POWER-SYSTEM-AN00_SE_XXXX_FM.indd 23 6/2/2018 3:55:05 PM

Thispageisintentionallyleftblank
 Introduction to Power
System Analysis 1
CHAPTER OUTLINE

1.1 Performance of Large Power 1.4 Power System Stability—Modelling

Systems and their Analysis Issues
1.2 Classification of Power System 1.5 Electromechanical Energy
Stability Problems Conversions
1.3 Power System Stability Analysis—
Computational Issues

In this chapter, we highlight various types of power system analysis that are carried out. In addition,
we also discuss the computational issues that are to be handled. The discussion enables the reader to
choose appropriate tools and techniques to perform such analysis. A brief introduction to state-space
modelling of systems is presented.
In order to prepare a reader to take up synchronous machine modelling in the later chapters, the
fundamentals of electromechanical energy conversion process are discussed using illustrative exam-
ples which offer time-varying parameters.

1.1 PERFORMANCE OF LARGE POWER SYSTEMS

AND THEIR ANALYSIS
A large power system is generally characterised by the following:
1. Large power transfer over large distance from generation centres to load centres.
2. Lack of power flow controllability due to heavy AC connection where power flow in various lines
gets decided by physical AC laws. Hence, transmission systems are unevenly utilised.
3. Heavy AC interconnection with limited high-voltage direct current (HVDC) links between con-
trol areas. This leads to unscheduled power flow.
4. Due to limited right-of-way, it is not easy to take up new expansion planning to augment the system.

M01_POWER-SYSTEM-AN00_SE_XXXX_CH01.indd 1 6/2/2018 11:08:00 AM

 2  |  Power System Analysis

5. Stability issues largely constrain power transfer. This may require large load margins for equip-
ment, making the system bulky and non economical.
6. While transferring large power over long distance, voltage control becomes a major control
problem.
7. The system possesses very poor damping for electromechanical oscillations, thus restricting the
loading level.
In order to develop an understanding of such a system, it is necessary to carry out a detailed power
system analysis under steady-state and dynamic conditions. This generally involves the following
types of analysis:
1. Reliability analysis: It involves a probabilistic study and denotes the ability to supply load
adequately with as few interruptions as possible. It is a function of time-average performance of
the power system over a long period [1].
2. Security analysis: It involves the determination of the ability of the power system to meet the
load demand without any violation of apparatus-operating limits against possible contingencies.
It is to be noted that for the system to be reliable, it must be secure most of the time. Security and
stability issues are strongly coupled.
3. Stability analysis: It determines the ability of the system to remain intact without losing syn-
chronous operation. For a system to be secure, it must be stable. Security and stability are time-­
varying attributes and are functions of the operating state and a candidate contingency. Therefore,
when security issues are analysed, it is customary to make a distinction between security analysis
carried out at an operating point, referred to as steady-state security, and those studies in which
transition to a new operating condition, is analysed, referred to as dynamic security analysis
(DSA) [2].

1.2  CLASSIFICATION OF POWER SYSTEM STABILITY PROBLEMS

It is well-known that stability of linear systems is neither dependent on the operating point nor on the
forcing function, whereas for non-linear systems, like power systems, the stability of the system is
decided by the equilibrium point at which it is operating and the magnitude of the forcing function.
Therefore, as per [3, 5], a classification is done to perform power system stability studies as shown in
Table 1.1 based on the time-scale and driving force of instability.
Some features of the tabulated-type of power systems stability are briefly discussed below [4].

Table 1.1  Classification of Power System Stability Analysis.

Time range Generator-driven Load-driven
(Rotor-angle stability)
Small-signal Transient
Short-term
stability stability Voltage
Long-term Frequency stability
stability

M01_POWER-SYSTEM-AN00_SE_XXXX_CH01.indd 2 6/2/2018 11:08:00 AM

 Introduction to Power System Analysis  |  3 

1. Rotor-angle stability: This analysis is mainly concerned with electromechanical oscillations and
large angle variation of the generator rotors. Small-signal stability analysis deals with the oper-
ating point stability with respect to low-frequency oscillations which can range from 0.1–3 Hz.
This stability is the fundamental requirement of any system. In case of transient stability analysis,
the system performance is studied for large disturbances. In this short-term time frame, there is
no common frequency throughout the system. However, in the long-term time scale when the
short-term dynamics are stable, the frequency excursions gain importance since both under- and
over-frequency system operations are detrimental to system integrity [6]. This long-term time
scale, which typically lasts for several minutes, necessitates modelling of slow dynamical devices
such as tap changers, prime-movers controllers, boilers and their controls, etc. Such long-term
dynamics largely depend on the generation-load imbalance irrespective of network connectivity
within each connected area.
  In addition to the above listed synchronous generator rotor-related stability performances,
another type of small-signal stability analysis that is studied is the subsynchronous resonance
(SSR) problem. This analysis involves modelling of combined steam turbine-generator mechan-
ical systems, including the dynamics of capacitor-compensated electrical transmission network.
Here, the frequency of interest lies in the range of 15–45 Hz, i.e., subsynchronous range [7, 8].
With the integration of induction generator- and doubly-fed induction generator-based wind
energy conversion systems into the conventional AC grid, it has led to another type of stability
problem, referred to as speed instability [9].
2. Load-driven stability: In contrast to angle stability where we are concerned about relative
and common angle oscillations, here, the analysis mainly involves determination of voltage-­
maintaining-ability of power systems. This analysis requires full network representation along
with modelling of dynamic load components (induction motors, electronically controlled loads,
HVDC interconnections), voltage control devices such as exciters, on-load tap-changers’ con-
trollers, any reactive power sources, and so on. The time scale of analysis and complexity of
modelling is often similar to the short-term rotor-angle stability analysis. Sometimes, in long-
term voltage stability studies, for simpler and faster analysis, the possibility of voltage instability
is understood by detecting the absence of equilibrium points using a static tool like power flow
programmes, even if the voltage stability problem is inherently dynamic in nature.
It should be noted that it is not easy to carry out power system stability analysis without performing
a categorisation of stability problems since such a classification facilitates the following:
1. To choose appropriate degree of details for system representation: Time frame of study and time
response of components helps us make appropriate approximations in any analysis. For example,
while investigating the impact of the control set point of boiler, it is not necessary to solve the
complex transmission line wave equations. This is obvious because the boiler response time is in
the range of minutes, whereas the wave travel time is in the order of milliseconds. Such engineer-
ing approximation is very useful in system studies [10].
2. To select an appropriate analytical tool: For example, when small-signal stability analysis is car-
ried out, the non-linear system equations are linearised around an operating point whose stabil-
ity performance is to be investigated. The linear control theory is then applied, assuming that
the system trajectories do not trigger non-linear behaviour. Therefore, one may just perform the
eigenvalue analysis and predict the stability without solving the system’s differential-algebraic

M01_POWER-SYSTEM-AN00_SE_XXXX_CH01.indd 3 6/2/2018 11:08:00 AM

 4  |  Power System Analysis

equations (DAEs). If it is required to obtain the large-signal performance, one is left with no
choice but to carry out numerical solution of non-linear DAEs.
3. To identify key factors that contribute to instability and to devise methods for improving stable
operation.

1.3 POWER SYSTEM STABILITY ANALYSIS—

COMPUTATIONAL ISSUES
To perform power system stability analysis, a knowledge of the following topics is desirable.
1. Linear algebra: The theory of vector spaces, basis, the theory of matrices, similarity transforma-
tions, time-variant and time-invariant systems, and so on [12].
2. Systems theory: Identification of linear and non-linear systems, state-space modelling, etc [11].
3. Computational techniques: This includes the following system of equations:
(a) Algebraic system of equations:
(i) Linear algebraic system of equations: Generally, these equations will be in the form,
Ax = b . Solution techniques may employ either elimination techniques such as Gauss or
decomposition methods such as [L][U], [L][L]T, and so on [13]. In power system analy-
sis, we come across this form of equations in tasks such as (i) steady-state fault analysis,
(ii) in each iteration of power flow analysis where deviations are calculated, (iii) in each
time step when implicit numerical integration technique is employed to solve linear dif-
ferential equations, etc.
(ii) Non-linear algebraic system of equations: Generally, these equations will be in the
form, F ( x ) = 0. For such systems, only numerical solution is possible, and solutions are
obtained by using numerical iteration techniques. Solution techniques may be in form of
fixed point iteration (also referred to as Gauss method), Gauss-Seidel, or Newton-based
methods. A classical power system analysis task where we use this type equations is
power flow analysis.
(b) Differential system of equations: The following are the two major kinds of differential equa-
tions depending on the nature of equations:
(i) Linear differential equations: Here, differential equations govern the dynamics of linear
systems. Linear systems are those which contain linear elements which exhibit a linear
relationship between its excitation and response. Such system of differential equations is
amenable for closed form of analytical solutions. They need not be solved using numer-
ical integration techniques. Such differential equations are easily recognisable as lin-
ear if they are written in state-space form [14]. For example, in an RL-series circuit, if
the inductor is air-core type, its inductance value remains constant independent of flux,
hence the current. In other words, the flux-current relationship is linear, resulting in a
constant inductance or we state that the inductor remains unsaturated. Therefore, the
differential equation governing the dynamics of such a circuit remains linear.
(ii) Non-linear differential equations: Here, differential equations govern the dynamics of
non-linear systems. In the above example of a series RL circuit, if the inductor saturates,
we can seen that the flux-current relationship is not linear. This leads to a non-linear
differential equation. An important difficulty in solving such differential equations is that

M01_POWER-SYSTEM-AN00_SE_XXXX_CH01.indd 4 6/2/2018 11:08:02 AM

 Introduction to Power System Analysis  |  5 

their solution cannot be obtained analytically, or in other words, the solution cannot be
written in a closed form of expression. The solution values of variables can only be deter-
mined at discrete time-instants by applying numerical integration techniques. Therefore,
through a numerical processing, the sampled-version of the solution flow is obtained.
(c)
Differential-algebraic equations (DAEs): Here, coupled differential and algebraic equations
are solved simultaneously. Such requirement arises when systems such as power systems,
weather systems, chemical systems, and so on, are modelled. The following approaches may
be employed [10]:
(i) Partitioned solution approach: Here, the differential and algebraic equations are solved
separately in an alternate fashion. To solve differential equations, one may employ either
explicit or implicit technique.
(ii) Simultaneous implicit solution approach: In this case, the differential equations are
algebraiced using an implicit integration technique and are augmented to the algebraic
equations. The resulting algebraic equations are collectively solved using Newton-based
iterative schemes.
The above methods will be further briefed in later sections.
(d) Evaluation of eigenvalues and eigenvectors: It is known that eigenvalue analysis is one of
the widely followed methods to understand the operating point stability of power systems. In
addition, eigenvectors facilitate easy prediction of dominance of a frequency component in a
state-variable, thus guiding the placement and tuning of controllers. The eigenvalues are the
roots of the characteristic polynomial of a system. It is well-understood that beyond the fourth
order, it is not easy to determine the roots of the polynomial. Hence, one is forced to use
numerical techniques to estimate eigenvalues and eigenvectors. In this regard, QR technique
[13] and modified Arnoldi methods (MAM) [15] are more widely used.

1.4 POWER SYSTEM STABILITY—MODELLING ISSUES

Developing an accurate mathematical model for a system and its constituent components is an impor-
tant step in power system analysis. A model is said to be suitable if it can faithfully reproduce the
real-world observations adequately, i.e., to a desired level of human satisfaction, in the computational
domain. A mathematical model development for a component starts by applying known physical laws
which govern the operation of the component. However, the final usage of a model for a component
depends on some of the following factors:
1. The possibility of determining the parameters involved in the model.
2. The human effort required to implement the model.
3. The availability of a suitable computational tool.
4. The computational effort to obtain the solution.
5. The time-response of the performance desired.
6. The computational accuracy required.
Therefore, there is always a trade-off between all these factors, which decide to what detailed level a
component model is used in system analysis. In any such studies, suitable engineering approximations
are made to obtain a response (in the mathematical domain) which adequately match the real-world

M01_POWER-SYSTEM-AN00_SE_XXXX_CH01.indd 5 6/2/2018 11:08:02 AM

 6  |  Power System Analysis

physical response. A good example for demonstrating the importance of approximations in modelling
is the methodology employed for modelling of synchronous machines. To support this, Prof. M. A. Pai
acknowledges in his book [10] by stating ‘There is probably more literature on synchronous machines
than on any other device in electrical engineering. Unfortunately, this vast amount of material often
makes the subject complex and confusing.’
As per literature, it is noted that no experimental test can be conducted on a synchronous machine
in the laboratory to determine the preliminary parameters desired by the basic model. In such a case,
one is required to obtain hybrid parameters of a machine. These parameters are commonly referred to
as the standard parameters of a synchronous machine. To make the basic model usable, it is modified
in terms of the standard parameters. Therefore, obtainability or availability of machine data often
constrains one from using a detailed model for the synchronous machine. In some cases, depending
on the acceptable level of accuracy, we can use reduced order model as well.
In power system analysis, the following two modelling approaches are generally employed:
1. Distributed parameter representation: Here, time and space are treated as independent variables.
For example, while modelling high-frequency transients such as lightning or switching surges,
it is desirable to consider the spacial distribution of basic parameters R, L, and C because, in
this case, the length of the line is comparable to the electromagnetic wave-length [16] leading
to c­ onsiderable wave-travel time. While modelling such components, we invariably use partial
differential equations (PDE) and they require dedicated numerical methods to capture the wave-
travel effect. Bergeron’s method [10] is one such method employed in electromagnetic transient
packages such as PSCAD/EMTDC and EMTP RV [17].
2. Lumped parameter representation: Here, only time is treated as an independent variable. The dis-
tributed influence of the basic effects, such as resistance, inductance and capacitance, are assumed
to be lumped at a place leading to a circuit model for the component. This enables us to write
ordinary differential equations (ODEs) to describe their dynamics as the wave-travel time is neg-
ligible. The power system stability analysis uses such representation for components.

1.4.1  Modelling of Dynamical Systems

Generally, there are two kinds of modelling approaches:
1. System function-based approach: In this approach, a given ODE is algebraiced and written in
the form of a function involving the ratio of polynomials. Such representation is applicable only
to linear systems. While algebraicing, the initial values on the variables are set to zero. This rep-
resentation is generally employed for ease of depiction of dynamics of a system. A typical system
function is given by
X ( s) Np( s)
=
U ( s) Dp( s)
d
where s is an operator which represents with zero initial conditions.
dt
For example, a simple single-time constant delay circuit is depicted as
X ( s) 1
H ( s) = = (1.1)
U ( s) (1 + sT )
where T is in seconds.

M01_POWER-SYSTEM-AN00_SE_XXXX_CH01.indd 6 6/2/2018 11:08:04 AM

 Introduction to Power System Analysis  |  7 

2. State-space-based approach: In this approach, a given nth order differential equation is writ-
ten as a set of n numbers of first order differential equations by appropriately choosing a state-­
variable vector [12]. An important advantage of this approach is that this representation can be
used for either linear or non-linear systems with desired initial value on states.

1.4.1.1  State-space Modelling of a System

Every physical system is causal. Causal or non-anticipatory system is one in which the present output
depends on the past and the present inputs and not on future inputs. Defining a state for such a system
overcomes the problem of tracking the input u(t) from t = −∞, for determining the present output. The
state x of a system is that variable whose value is known at time t0 is the information at t = t0 that,
along with the input, for t ≥ t0, uniquely determines the output y(t) for all t ≥ t0. Here, x is referred to
as a state-variable and it possess the following properties:
1. State-variables are continuous in time, i.e., x(t0− ) = x(t0+ ).
2. The selection of state-variables is not unique.
3. The number of state-variables chosen for a system indicates the degrees-of-freedom associated
with the system. It also represents the order of a system.
Let us examine another classification of lumped systems which is given below [18]:
1. Time-invariant systems: If the initial state and the input are the same, no matter at what time
they are applied, the output waveform will be the same. For such systems, the parameters of the
­systems are constant and independent of time.
2. Time-varying systems: Here, parameters of the systems are not constants and are functions
of time. Examples are change of parameters due to aging, primitive inductance of rotating
machines, etc.

Note:  Most physical systems are modelled as time-invariant systems as it is not easy to handle
time-varying systems in terms of solution of systems of equations.
The state-space description of a lumped linear time-invariant (LTI) systems is given by
x = Ax + Bu

y = Cx + Du (1.2)

For a p-input and q-output system, u is p × 1 vector and y is a q × 1 vector. If the system has
n state-variables, then x is an n × 1 vector.
• A → n × n: state-feedback matrix
• B → n × p: input matrix
• C → q × n: output matrix
• D → q × p: direct-feed matrix

It should be noted that the A matrix is constant and is independent of time for LTI systems. Also,
note that the state-space description involves a set of n first-order linear differential equations and
q algebraic equations.

M01_POWER-SYSTEM-AN00_SE_XXXX_CH01.indd 7 6/2/2018 11:08:07 AM

 8  |  Power System Analysis

The above description of the state-space model, i.e., in the matrix notation see eq. (1.2), is feasible
only for linear systems. For non-linear systems, the state model cannot be written in matrix form.
Then, it can only be written in the following form:
x = F ( x , u , t ) (1.3)

where x denotes the stat-variable vector, u -represents the input vector.

If the time derivative of the state is not an explicit function of time t, such a system is referred to
as an autonomous system. As stated earlier, for non-linear systems, the solution cannot be obtained by
analytical means. Only a numerical solution is possible. It is also important to note that to apply any
numerical integration technique, the system should be modelled only in state-space form.

Example 1.1
For the system function shown in eq. (1.1), a state-space model can be written as
1 1
x = − x + u(t ) (1.4)
T T
For a unit-step input, the zero-state response is given by
 t
 − 
y (t ) = x (t ) = 1 − e  T

1
Let z = x , then x = Tz . Making this substitution in eq. (1.4) we get,
T
1 1
z = − z + 2 u(t ) (1.5)
T T
and the desired output is given by
x(t ) = Tz (t ) (1.6)
From eq. (1.5), if the system function is obtained, then we get
1
Z ( s) T
H1 ( s) = =
U ( s) (1 + sT )

Now using eq. (1.6) in the above equation, we get

1 1
X ( s)
H1 ( s) = T = T
        (1.7)
U ( s) (1 + sT )

The above equation, clearly demonstrates that for a given system function there is no unique state-
model. This is because H(s) given in eq. (1.1) can be obtained from either eq. (1.4) or eq. (1.5).

Example 1.2
Consider another system function given by
X ( s) sT
H ( s) = = (1.8)
U ( s) (1 + sT )

M01_POWER-SYSTEM-AN00_SE_XXXX_CH01.indd 8 6/2/2018 11:08:22 AM

 Introduction to Power System Analysis  |  9 

In the above case, if we simply cross-multiply LHS and RHS terms to get a state-model, then it results
in taking a derivative of the input u which should be strictly avoided in building a state-model. A pro-
cedure to avoid this is given below.
The above function is rewritten as
sT × U ( s)
X ( s) =
(1 + sT )
= sT × Z(s)(1.9)
where
U ( s)
Z ( s) =
(1 + sT )
A state-model for the above system function can be easily written as
1 1
z = − z + u(t ) (1.10)
T T
From eq. (1.9), the output equation can be written as
x(t ) = Tz (1.11)
Using eq. (1.10) in the above equation, we get the desired output equation as
x(t) = −z + u(t)(1.12)

Therefore, for the system function given in eq. (1.8), the complete state-model is given by
eqs. (1.10)  and  (1.12).

1.5  ELECTROMECHANICAL ENERGY CONVERSION

The process of electromechanical energy conversion involves two forms of energy conversion:
1. Conversion from electrical to mechanical which is referred to as motor operation.
2. Conversion from mechanical to electrical which is referred to as generator operation.
In both the processes, there exists a magnetic field which acts as a media between electrical and
mechanical systems. This media should be filled with energy to facilitate the conversion1. This implies
that electrical circuit is dominantly inductive, and its inductance may be time-varying. To understand
this complex energy exchange and interaction, a few examples are presented [19, 20].

1.5.1 Behaviour of an Inductor with Fixed Armature Core

Consider a coil as shown in Figure 1.1, where the armature core is fixed and is assumed to be unsaturated.
Applying Kirchhoff’s Voltage Law (KVL), we can write the circuit equation as
dy
v = Ri + (1.13)
dt
1This clarifies the concept of reactive power loss at fundamental frequency under sinor steady-state. An analogy to this concept
is the water in a pipe. In order to have water discharge at a constant rate at the other end of the pipe, the pipe should be filled
with water all the time as long as we need water delivery. This amount of water is not available for utilisation. This is equivalent
to a loss. This represents the reactive power and the water discharge at a given rate denotes the real power transferred.

M01_POWER-SYSTEM-AN00_SE_XXXX_CH01.indd 9 6/2/2018 11:08:26 AM

 10  |  Power System Analysis

i where, the flux linkage y = Li.

+ v
Assuming no saturation, L is a constant and we can write,
di
v = Ri + L (1.14)
dt
Figure 1.1  Inductor Circuit To calculate the instantaneous power delivered to the circuit, we have
with Fixed Armature.
di
vi = Ri 2 + Li (1.15)
dt

1 2
We know that the magnetic field energy is given by Wmg = Li . Let us consider
2
1 
d  Li 2 
2  1 di di
= 2 Li = Li (1.16)
dt 2 dt dt

Using the above result in eq. (1.15), we have

1 
d  Li 2 
 2  (1.17)
vi = Ri 2 +
dt 1 
d  Li 2 
2 2 
where Ri denotes power loss in the circuit due to resistive component and represents the
dt
rate at which energy stored in the magnetic field Wmg, is increased.

1.5.2 Behaviour of an Inductor with Movable Armature Core

Consider a coil as shown in Figure 1.2, where the armature core can move linearly in the direction
shown and is assumed to be unsaturated.

x F

i
− v +

Figure 1.2  Inductor Circuit with Linearly Movable Armature.

For the circuit, by applying KVL, we can write,

dy
v = Ri + (1.18)
dt

M01_POWER-SYSTEM-AN00_SE_XXXX_CH01.indd 10 6/2/2018 11:08:37 AM

 Introduction to Power System Analysis  |  11 

where, the flux linkage y = L(x)i. Here, it should be noted that the coil inductance is not constant, but
varies depending on the relative position of the coil axis and the that of the core, i.e., displacement x.
Hence the above equation is rewritten as
d ( Li )
v = Ri +
dt
di dL
= Ri + L + i (1.19)
dt dt
Instantaneous power delivered to the circuit is given by

di dL 2
vi = Ri 2 + Li + i (1.20)
dt dt
Now considering the magnetic energy and its rate of change, we have

1 
d  Li 2 
2  1 di 1 dL
= 2 Li + i 2
dt 2 dt 2 dt

di 1 2 dL
= Li + i (1.21)
dt 2 dt
di
Substituting for Li in eq. (1.20) from eq. (1.21), we get
dt
1 
d  Li 2 
 2  1 2 dL dL 2
vi = Ri 2 + − i + i
dt 2 dt dt

1 
d  Li 2 
 2  1 2 dL
= Ri 2 + + i (1.22)
dt 2 dt

From the above equation, it is clear that

 power converted
 Instantaneous  Power loss  rate of increase 
 power input  =  + + from electrical 
in R   of energy in L   
 to mechanical 

Let F be the force (in N) exerted on the armature, then,

dx 1 2 dL
F = i
dt 2 dt

1 2 dL dx
= i . (1.23)
2 dx dt

M01_POWER-SYSTEM-AN00_SE_XXXX_CH01.indd 11 6/2/2018 11:08:47 AM

 12  |  Power System Analysis

Therefore, the force, F acting on the armature is given by,

1 2 dL
F= i (1.24)
2 dx
Note: The direction of F is independent of the polarity of the current, whether i is DC or AC.

1.5.3 Primitive Machine with Rotating Armature without Rotor Coil

Consider a system as shown in Figure 1.3, where the armature core can exhibit angular motion.

i
− v +

Figure 1.3  A Primitive Machine with a Rotating Armature without Rotor Coil.

We know that inductance of a coil is given as the flux linking the coil per unit current. In this system,
for a given constant coil current, the flux linking the coil depends on the the rotor position, q. This is
due to the fact that when the rotor position changes, it alters the reluctance of the magnetic circuit.
Therefore, the inductance of the coil due to its current, referred to as self-inductance, is expressed as
L(q ). The flux-current relationship is given by

y = L(q )i (1.25)

where L(q ) is given by

L(q ) = Ls + Lm cos(2q )

The above expression considers terms up to the fundamental component, neglecting the higher order
terms. Here, q varies between 0 and 2p in electrical rad, i.e., over a pair of poles. A typical variation
of the inductance is shown in Figure 1.4.
The model equation is given by,
dy
v = Ri +
dt

d [ L(q )i ]
= Ri +
dt

di dL(q )
= Ri + L(q ) +i (1.26)
dt dt

M01_POWER-SYSTEM-AN00_SE_XXXX_CH01.indd 12 6/2/2018 11:08:52 AM

 Introduction to Power System Analysis  |  13 

0.18

0.16

0.14

0.12
Inductance (mH)

0.1

0.08 Ls
Lm
0.06

0.04

0.02

0
0 1 2 3 4 5 6
Rotor position (elec. rad)

Figure 1.4  A Typical Variation of L(q   ) for One Coil System.

Instantaneous power delivered to the circuit is given by,

di 2 dL(q )
vi = Ri 2 + L(q )i +i (1.27)
dt dt

Now, considering the rate of change of the magnetic energy, we have,

1 
d  L(q )i 2 
2  di 1 dL(q )
= L(q )i + i 2 (1.28)
dt dt 2 dt

di
Substituting for L(q )i in eq. (1.27) from the above expression, we have,
dt

1 
d  L(q )i 2 
 2  1 2 dL(q )
vi = Ri 2 + + i (1.29)
dt 2 dt

Similar to the previous case, from eq. (1.29), it can be identified that the rate at which the energy is
converted from electrical to mechanical is given by,

M01_POWER-SYSTEM-AN00_SE_XXXX_CH01.indd 13 6/2/2018 11:08:59 AM

 14  |  Power System Analysis

1 2 dL(q )
Temw r = i (1.30)
2 dt
where Tem denotes the air-gap mechanical torque developed in Nm and wr represents the actual rotor
speed in mechanical rad/s.
To obtain an expression for the mechanical torque developed, the above equation is rewritten as
1 2 dL(q )
Temw r = i
2 dt

1 2 dL(q ) dq m (1.31)
= i
2 dq m dt
dq m
Recognising that = w r , with qm denoting the mechanical rad, we can obtain an expression for
dt
the torque as
1 2 dL(q )
Tem = i
2 dq m (1.32)

2
For a P-pole machine, we know that, q m = q and we have,
P
P  1 2 dL(q ) 
Tem =  i  (1.33)
2 2 dq 
Note:
• L(q  ) denotes waveform for an equivalent two-pole machine.
• Tem represents the torque developed in a motor. By similarity, the torque developed in a generator
can be written as
P  1 2 dL(q ) 
Teg = −  i 
2 2 dq 

1.5.4  Primitive Machine with Rotating Armature with a Rotor Coil

In this section, before the electromechanical energy conversion - related expressions are derived for
a conventional synchronous machine. The conventions used in the analysis are described as follows:
1. Motor convention: Here, current i enters the winding at the positive terminal (Figure 1.5(a)).
This denotes a condition where the winding absorbs an instantaneous power, p = v × i. Therefore,
it is also known as load convention. By applying KVL to the loop, we can write,
v − Ri − e = 0 (1.34)

or

v = Ri + e(1.35)

M01_POWER-SYSTEM-AN00_SE_XXXX_CH01.indd 14 6/2/2018 11:09:05 AM

 Introduction to Power System Analysis  |  15 

 dy 
where, v represents the applied voltage, R is the resistance of the winding and e  = denotes
 dt 
the back-emf caused by rate of change of flux-linkage.
2. Generator convention: Here, current i leaves the winding at the positive terminal – see Figure 1.5(b).
This represents a condition where the winding supplies an instantaneous power, p = v × i
(to an external circuit at terminal voltage v). For this condition, the voltage equation is
obtained as

v + Ri − e = 0 (1.36)

or

v = −Ri + e(1.37)

As per Faraday’s law, e represents the induced emf in a generator and is given by
dy
e=−
dt

Note that this emf maintains the terminal voltage v while driving a current i.

i i
+ +
v
v

− −
(a) Motor convention (b) Generator convention

Figure 1.5  Motor and Generator Conventions.

1.5.4.1 Motor Operation of a Primitive Machine

Consider a system as shown in Figure 1.6. Here, unlike in the previous case, a rotor coil is mounted
on the rotor which can exhibit angular motion. This denotes the field winding in a conventional syn-
chronous machine.
The voltage equations are given by,
dy 1
v1 = i1R1 +
dt
dy 2
v2 = i2 R2 +
dt (1.38)

where the flux linkage-current expressions are obtained as

y 1 = L11 (q )i1 + L12 (q )i2

y 2 = L21 (q )i1 + L22 (q )i2 (1.39)

The above equations can be written in the vectorised form as

M01_POWER-SYSTEM-AN00_SE_XXXX_CH01.indd 15 6/2/2018 11:09:19 AM

 16  |  Power System Analysis

θ
𝜙2

i2
+ Rotor
v2
−

Stator

𝜙1

i1

+ v1 −

Figure 1.6  A Primitive Machine Having a Rotating Armature with a Rotor Coil.

dy
v = [ R] i + (1.40)
dt

where y = [ L (q )] i with v = [v1 v2 ]T ; i = [i1 i2 ]T ; y = [y 1 y 2 ]T and

 R1 0   L11 (q ) L12 (q ) 
[ R] =   and [ L(q )] =   (1.41)
0 R2   L21 (q ) L22 (q ) 

Note that as in the previous case, the inductance matrix is time-varying, i.e., the self and mutual
inductances are functions of the rotor angular position q. For the system shown in Figure 1.6, a typical
variation of inductances is depicted in Figure 1.7.
Using [L(q  )] from eq. (1.41) in eq. (1.40) we can write the voltage equation as

d i d [ L(q )]
v = [ R] i + [ L(q )] + i (1.42)
dt dt

Note that for a constant angular speed of the rotor, the above differential equation is linear but
time-variant.
The instantaneous power input to the system is given by,

di d [ L(q )] (1.43)
p = i T v = i T [ R] i + i T [ L(q )] + iT i
dt dt

For this coupled-coil configuration, the magnetic field energy is given by,

M01_POWER-SYSTEM-AN00_SE_XXXX_CH01.indd 16 6/2/2018 11:09:42 AM

 Introduction to Power System Analysis  |  17 

1.4
L11(θ) = Ls − Lm cos (2θ)
1.2

0.8

(pu) 0.6

0.4

L22(θ) = Ns − Nm cos (2θ)

0.2

0
L12(θ) = L21(θ) = −M sin(θ)
−0.2

−0.4

−0.6
0 1 2 3 4 5 6 2π 7

(rad)

Figure 1.7  A Typical Variation of L(q  ) for a Two-coil System.

1 T
Wmg = i [ L (q )] i (1.44)
2

Considering the rate of change of the magnetic energy, we have,

dWmg 1 diT 1 d [ L (q )] 1 di
= [ L (q )] i + i T i + i T [ L (q )] (1.45)
dt 2 dt 2 dt 2 dt

Noting that energy is a scalar quantity and [L(q   )] is a symmetric matrix, consider the transpose
of the first term in the above expression, we get,

T
1 d i T  1 di
 [ L (q )] i  = i T [ L (q )] (1.46)
 2 dt  2 dt

Using the above result in eq. (1.45), we get,

dWmg d i 1 T d [ L (q )]
= i T [ L (q )] + i i (1.47)
dt dt 2 dt

M01_POWER-SYSTEM-AN00_SE_XXXX_CH01.indd 17 6/2/2018 11:09:47 AM

 18  |  Power System Analysis

T di
Substituting for i [ L(q )] from eq. (1.47) in eq. (1.43) we get,
dt
dWmg 1 T d [ L(q )]
p = i T v = i T [ R] i + + i i (1.48)
dt 2 dt
From the above expression, the mechanical power developed can be identified as
1 T d [ L(q )]
Pem = Temw r = i i
2 dt
1 T d [ L(q )] dq m
= i i
2 dq m dt

1 T d [ L(q )]
= i i w r (1.49)
2 dq m

Therefore, we can obtain an expression for the mechanical torque developed in a motor as

1 T d [ L(q )]
Tem = i i or
2 dq m
 P 1 d [ L(q )]
=   iT i (1.50)
 2 2 dq

where P represents the number of poles in the machine.

From eq. (1.48) the power balance in the electrical part of the machine can be written as
dWmg
p = i T v = i T [ R] i + + Pem (1.51)
dt
From the principle of energy conservation, for mechanical part of the machine, we can write that
dWKE
+ PFR = Pem − PLm (1.52)
dt
dWKE
where denotes the rate of change of kinetic energy in the rotor mass.
dt
PFR represents frictional losses.
PLm denotes mechanical load on the machine which opposes the motion.
It should be noted that eqs. (1.51) and (1.52) completely define the operation of a motor. For exam-
ple, when a three-phase symmetrical motor is under steady state, both the electrical and mechanical
systems should have reached their respective steady-state so that

Pinput = i T v = i T [ R] i + Pem (1.53)

and

Pem = PFR + PLm(1.54)

M01_POWER-SYSTEM-AN00_SE_XXXX_CH01.indd 18 6/2/2018 11:09:57 AM

 Introduction to Power System Analysis  |  19 

1.5.4.2 Generator Operation of a Primitive Machine

For generator action, using the convention given in eq. (1.37), the voltage equation in (1.42) is
modified as

d i d [ L(q )]
v = −[ R] i − [ L(q )] − i (1.55)
dt dt

The instantaneous power output of the system is given by,

di d[ L(q )]
p = i T v = − i T [ R] i − i T [ L(q )] − iT i (1.56)
dt dt

Using the result given in eq. (1.47), we have,

d i dWmg 1 T d [ L (q )]
i T [ L (q )] = − i i (1.57)
dt dt 2 dt

Substituting the above expression in eq. (1.56) we get,

dWmg 1 T d [ L(q )]
p = i T v = − i T [ R] i − − i i (1.58)
dt 2 dt

The above expression can be rewritten as

dWmg
p = i T v = − i T [ R] i −
+ Peg (1.59)
dt
where, Peg denotes the electrical power developed at the air-gap, and is given by

1 T d [ L(q )]
Peg = Teg w r = − i i (1.60)
2 dt
where wr denotes the rotor speed in mechanical rad/s.
Proceeding as described above for the motor case, we can obtain an expression for the air-gap
torque in a generator as

 P 1 d[ L(q )]
Teg = −   i T i (1.61)
 2 2 dq

This is the electromagnetic torque exerted on the rotor (in a generator) in a direction opposite to that
of the mechanical torque. From eq. (1.59), it is clear that

 Electrical   Rate of increase

 power  −  Copper loss −  of energy in  =  Power supplied 
   in R     
 developed    magnetic field 
to load

M01_POWER-SYSTEM-AN00_SE_XXXX_CH01.indd 19 6/2/2018 11:10:04 AM

 20  |  Power System Analysis

However, under sinusoidal steady-state, for a three-phase generator, expression (1.59) simplifies to

PLe = i T v = − i T [ R] i + Peg (1.62)

and

Peg = −PFR + Pm(1.63)

where Pm denotes the mechanical power input to the generator and PLe represents the electrical
(real power) load connected to the generator terminals.
Further, note that for a conventional three-phase synchronous machine, which has 3-phase stator
windings and a rotor field winding, a steady-state condition denotes the following:
1. A balanced sinusoidal steady-state condition of the stator windings and a DC steady-state condi-
tion for the field winding.
2. For stator windings, we have,
ua ia + ub ib + uc ic = PLe

and the air-gap power, Peg is such that

Peg = PLe + I2Ra losses of stator windings.

For field winding, we have,

uF iF = iF2RF

where Ra denotes armature resistance and RF represents field winding resistance.

3. No currents in the damper circuits in a salient-pole machine.
4. Balancing of reactive power demand at the generator terminals.
5. In general, a fixed frequency and fixed voltage 3-phase balanced sinusoidal voltage source.

References
 [1] R. Billinton, Power System Reliability Evaluation, Gordon and Breach, New York, 1970.
 [2] Neal Balu et al., On-line Power System Security Analysis, Proceedings of the IEEE, vol. 80(2),
262–280, Febraury 1992.
 [3] P. Kundur et al., Definition and Classification of Power System Stability, IEEE Trans. on Power
System, vol.19(2), pp. 1374–1389, August 2004.
 [4] A. M. Kulkarni, Power System Dynamics and Control [Online]. Available at: http://nptel.iitm.
ac.in/courses/108101004/.
 [5] T. V. Cutsem and C. Vournas, Voltage Stability of Electric Power Systems, Kluwer Academic
Publishers, Boston, 1998.
 [6] P. Kundur, Power System Stability and Control, McGraw-Hill Inc., NY, 1994.
 [7] IEEE Committee Report Reader’s guide to subsynchronous resonance.’ IEEE Trans. Power Syst.,
vol. 7(1), pp. 150–157, 1992.
 [8] Padiyar, K.R., Analysis of subsynchronous resonance in power systems, Kluwer Academic
Publishers, Norwell, MA, USA, 1999.

M01_POWER-SYSTEM-AN00_SE_XXXX_CH01.indd 20 8/21/2017 4:47:25 PM

 Introduction to Power System Analysis  |  21 

 [9] O. Samuelsson and S. Lindahl, On speed stability, IEEE Trans. Power Systems, vol. 20(2),
pp. 1179–1180, May 2005.
[10]  P. W. Sauer and M. A. Pai, Power System Dynamics and Stability, Prentice Hall, Upper Saddle
River, New Jersy, 1998.
[11]  K. Ogata, State Space Analysis of Control Systems, Prentice-Hall, Englewood cliff, NJ, 1967.
[12]  Chi-Tsong Chen, Linear Systems Theory and Design, Oxford University Press, NY, 1999.
[13]  S. A. Soman, S. A. Khaparde, and Shubha Pandit, Computational Methods for Large Sparse power
Systems Analysis, An Object Oriented Approach, Kluwer Academic Publishers, Netherlands,
2002.
[14]  Ogata K, Modern Control Engineering, Prentice-Hall India Pvt. Ltd, New Delhi, 2001.
[15]  P. Kundur, G. R. Rogers, D. Y. Wong, L. Wang, and M. G.Lauby, A Comprehensive Computer
Program Package for Small-signal Stability Analysis of Power System,’ IEEE Trans. on Power
Systems, vol. 5(4), November 1990.
[16]  John D. Ryder, Networks, Lines and Fields, Prentice-Hall India Pvt. Ltd., 1995.
[17]  R. Ramanujam, Computational Electromagnetic Transients, I. K. International Publishing House
Pvt. Ltd., New Delhi, 2014.
[18]  Thomas Kailath, Linear Systems, Prentice-Hall, Inc., Englewood, New Jersy, 1980.
[19]  Charles V. Jones, The Unified Theory of Electrical Machines, Butterworth, London, 1967.
[20]  A. R. Bergen and V. Vittal, Power System Analysis, Pearson Education Asia, India, 2001.

Review Questions
1. What is security analysis?
2. Classify power system stability analysis. What is the need of such a classification?
3. How does an analytical solution differ from a numerical solution?
4. Differentiate an EMTP-kind of simulation package from a state-space model-based package
(e.g., SIMULINK).
5. What decides degrees-of-freedom for a dynamical system?
6. List some properties of LTI systems.
7. Enumerate two types of conventions employed for circuit analysis.
8. For a two-winding transformer, write the dynamic equations neglecting no load current and satu-
ration. Bring out the structure of parameter matrices and comment on the nature of equations.
9. System functions are written only for LTI systems. Is this a true statement?
10. For the following system function, write a state-space model:
Y ( s) K (1 + c1s)
=
U ( s) (1 + b1s + b2 s2 )

11. Explain the significance of motoring and generated torque in a synchronous generator.

M01_POWER-SYSTEM-AN00_SE_XXXX_CH01.indd 21 6/2/2018 11:10:08 AM

 Basics of Power Systems 2
CHAPTER OUTLINE

2.1 Sinor Waveform for Conventional 2.4 Power in Single-phase Circuits

Power Generation 2.5 Power in Three-phase Systems
2.2 Choice of Frequency for Power 2.6 Per-unit Representation
Generation 2.7 Characteristics of a Typical Power
2.3 Concept of Phasor Analysis in AC System
Sinusiodal Systems

In this chapter, the reader is introduced to the fundamentals of a conventional power system. Many
concepts such as phasor analysis, and real and reactive powers are discussed for single-phase and
three-phase systems. It is clarified that such a representation is valid only under sinor steady-state con-
dition. This is important because this condition also indicates the synchronous operation of a power
system. In this regard, the significance of power angle is also discussed.
To introduce the dynamic analysis of systems, the state–space modelling of some circuits are pre-
sented. Further, steady–state sinor analysis is carried out as a special case of the state-model. To valid
the existence of a steady-state in a system, the real and reactive power balances are brought out along
with their conventions. In addition, per-unit representation of quantities, its advantages and usage in
single-phase and three-phase systems have been explained.
For the benefit of the reader, the characteristics of a typical power system are highlighted so as to
appreciate their steady-state and dynamic behaviours.

2.1 SINOR WAVEFORM FOR CONVENTIONAL POWER GENERATION

A conventional power system is constituted mainly by an interconnected set of alternating current
(AC) generators, transformers, a network of lines, and loads. An effort is made to generate sinusoi-
dally varying voltage wave of a known frequency due to the following factors:
1. The most basic form of wave observed in nature is sinudoidal, i.e., the harmonic oscillation is
fundamentally in the form of sinusoids.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 22 4/23/2018 5:35:52 PM

 Basics of Power Systems  |  23 

2. The response of linear circuit elements such as resistor, inductor, and capacitor for sinusoidal
excitation remains sinors of the same frequency as the excitation. This also implies that with
sinusoids, only the fundamental frequency sinors contribute to net power (termed as real or active
power). This enables easy design and operation of power system components.
Therefore, an ideal AC generator generates voltage having a pure sinor waveform and sources power
only at fundamental frequency although it can deliver a non-linear load current.
Any node or point in the system is characterised by two important variables such as voltage and
frequency. Transformers enable us to work with different voltage levels at different parts in the network.
However, the frequency of the supply is identical at any points in the interconnected network under the
sinor steady-state. Hence, it is common to attribute voltage as the local variable and frequency as the
global variable. This is also true due to the fact that by using local means, the voltage magnitude at a
point can be controlled, whereas by no means, can the frequency of the supply be varied as desired using
conventional equipment. Therefore, a coordinated control of generator frequency is essential among the
generators which are interconnected.

2.2  CHOICE OF FREQUENCY FOR POWER GENERATION

There has been a detailed discussion on the choice of power system frequency in many literature [1].
The major factors which govern the choice of frequency of operation are described as follows:
1. Size of equipment and their cost: At low frequencies, the equipment are heavier and costlier
as they require more active materials such as copper and magnetic material. For example, the
diameter D and length L of a generator are related to the VA-rating, S and the speed of rotation ns
(in revolution per second (rps)) of the machine, by an expression given by [2],

1 S
( D 2 L) = (2.1)
Go  ns 
Where Go is an output coefficient decided by the specific values of the active material, and ns is
related to frequency of power generation (in Hz) as
 P
f = ns ×   (2.2)
 2
Where P denotes the number of poles.
From the above expressions, it can be seen that if speed is higher, the frequency is also higher,
and for a given rating of the machine, the volume of active material denoted by (D2L) decreases.
Therefore, the frequency of air-craft generators is chosen in the range of 400 Hz.
2. Voltage drop in lines and apparatus: If a higher frequency is chosen, the impedance of generators,
lines, transformers, etc., increases, leading to more voltage drop in the system. This demands a
lower frequency to be chosen for the system.
3. Flicker of lamps: If a lower frequency is chosen, the flicker of lamps becomes perceptible, thus
causing discomfort in their usage.
4. Design and operational issues: At higher frequencies, it is noted that maintenance requirement is
greater and insulation coordination issues are more involved. Sometimes, it may even constrain
the design of inherently low-speed generators (e.g., hydro generators), as they have to be provided
with more number of poles.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 23 4/23/2018 5:35:54 PM

 24  |  Power System Analysis

These factors are weighed against one another and finally, an operating frequency is standardised so
as to facilitate large production, standardisation of component design, and system interconnection.
In this context, 50 Hz and 60 Hz are the two widely preferred frequencies for the power system. It is
learnt that other than minor cost implications, there are no major reasons due to which one frequency
is chosen among the two for power system.

2.3 CONCEPT OF PHASOR ANALYSIS IN AC SINUSIODAL SYSTEMS

In order to illustrate the phasor concept, consider two sinusoidally varying quantities designated as
voltage, v(t), and current, i(t), as

v(t) = 1.0sin (w t)

 p
i(t) = 0.5sin  w t −  (2.3)
 4

Note that in the above expressions, w represents an angular frequency measured in rad/s. This is also
referred to as radian frequency and is equal to 2 × p × f. In this case, f is set to 50 Hz. Also, note that
p
the second signal i(t) maintains an angle of radian relative to v(t). This implies that these two sig-
4
nals have the same time-reference, t, which means that they are time-synchronised to one another. In
this circumstance, it can be seen that the angle between them remains constant independent of time.
This angle is denoted as the phase-angle difference. This notion also conveys that the phase-angle is a
single-frequency attribute defined only for sinusoidal waveforms.
For the two quantities in (2.3), a time–domain plot is shown in Figure 2.1.
y−axis

Snapshot B
1 90
1.0
0.8 v(t) 120 60

0.6 Snapshot B
Instantaneous value (V or A)

150 30 wt
0.4 i(t) 0.3535 0.5 increase

0.2

0 180 0
Snapshot A x−axis
−0.2

−0.4 −0.3535
210 330
−0.6

−0.8 240 300

−1 270
Snapshot A
−5 0 5 10 15 20
Time (s) x 10 −3

Figure 2.1  Phasor Diagram for Voltage and Currents.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 24 4/23/2018 5:35:55 PM

 Basics of Power Systems  |  25 

From Figure 2.1, it can be seen that if we track the v(t)-signal, its instantaneous negative maximum,
−1 occurs at t = −0.005 s and it again becomes −1 later at t = 0.015 s. This means that it completes
one cycle in 0.02 s, implying a 50 Hz signal. A similar verification can be made with i(t)-signal as
well. Further, it can be noted that the positive peak value of i(t)-signal occurs at t = 0.0075 s, whereas
the v(t)-signal reaches its positive peak before this instant, i.e., at t = 0.005 s. This relative variation
of signals is generally stated as ‘v(t) leading i(t)’. Note that this conclusion has been drawn using the
time–domain plot. However, this way of dealing with the sinors is not straightforward when many such
signals are present in a system.

2.3.1  Phasor Representation

If it is required to analyse a system under a single-frequency sinusoidal steady-state condition, it is
possible to use phasor-analysis instead of time–domain-based representation.
With the understanding of the phase-angle, a phasor is defined as a revolving vector of constant
amplitude rotating at a uniform angular speed, w. It is used to represent single-frequency sinusoidal
quantities such as voltage and current under steady-state in a linear time-invariant (LTI) system [3, 4].
Note: The phasor concept is employed even under dynamic conditions to analyse complex time-­
varying systems by defining dynamic phasors [5].
These phasors are assumed to rotate in the counter-clockwise direction where w t increases as time
progresses. One revolution of the phasor depicts completion of one cycle in the time–domain. At any
time t, if the instantaneous value of the signal in time–domain is desired, the vertical component of
the phasor is determined. For example, at snapshot-A, from which point, time t is counted, i.e., t = 0,
v(t) = 0 V, and i(t) = −0.3535 A. When snapshot-B is considered, time advances to 0.005 s, v(t) =
1.0 V, and i(t) = +0.3535 A. This shows that whether snapshot-A or snapshot-B is considered, the
p
phase-­angle difference between v(t) and i(t) remains constant at − . Therefore, following a phasor
4
representation, we obtain the instantaneous values of signals given in (2.3) as

v(t ) = Imag Vm e jw t 

 

  p   (2.4)
 jw t − 
i (t ) = Imag  I m e  4 


Since w is constant and the wave is purely sinusoidal in nature, only relative phase-angle information
is retained in phasor notations. Therefore, we have the voltage and current phasors represented as

V = Vm e j 0 = V ∠0°
p
−j
I = I me 4 = I ∠ − 45°
(2.5)

Note that in the above expressions, the peak amplitude of the signals are used. However, it is custom-
ary to employ root mean square (RMS) value in the representation. Therefore, the phasor notation
becomes

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 25 4/23/2018 5:35:58 PM

 26  |  Power System Analysis

V = Ve j 0 = V ∠0°
p
−j
I = Ie 4 = I ∠ − 45° (2.6)
Vm Im
where, V = and I = .
2 2

2.3.1.1  Cosine and Sine Waveforms for System Analysis

In literature [6, 7], there is widespread usage of cosine waveform for circuit analysis. In order to
understand the relationship between cosine-based reference and sine-based reference, consider the
following sinors of the same amplitude:

v1 (t ) = Vm cos(w t )

v2 (t ) = Vm sin(w t )

These waveforms are plotted in Figure 2.2, where it is seen that

 p
v1 (t ) = Vm cos(w t) = Vm sin  w t + 
 2

Using the sine-based reference, the corresponding peak-phasors can be written as,

p
j
V1 = Vm e 2 (2.7)

V2 = Vm e j 0 = Vm (2.8)

and are also shown in Figure 2.2. Since the amplitude of the signals is equal, we can write V1 in terms
of V2 as
p
j
V1 = V2 e 2 = jV2

Notes:
1. If sine-wave is used, x-axis is taken as the reference, whereas for cosine-wave, y-axis is taken
as the reference. These references denote the beginning of the time instant for the respective
waveforms.
2. In certain cases, cosine waveform is employed as it provides even symmetry for a function.
3. When phasor representation is employed for circuit analysis, there is no major difference between
these two references.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 26 4/23/2018 5:36:03 PM

 Basics of Power Systems  |  27 

y−axis
Reference for cos(wt)
v1(t)
1 v2(t) 90 1
120 60
0.8
0.6 Phasor V1
150 0.5 30
0.4 Reference for sin(wt)
v1(t) and v2(t)

0.2
0 180 0
Phasor V2 x−axis
−0.2
−0.4
210 330
−0.6
−0.8
240 300
−1 Snapshot point 270

−5 0 5 10 15 20
Time (s) x 10 − 3

Figure 2.2  Phasor Diagram for Cosine and Sine Waveform References.

2.3.1.2  Phasor Representation in a Simple Resistive Circuit

In Figure 2.3, a circuit schematic is shown to depict a single-phase generator delivering a pure resistive
element. The source voltage is given by v(t ) = Vm sin(w t ) , with w  = 2p f rad/s.
Notes:
1. For a source, Vm (or V) and frequency, f are generally assumed to be constant, which signify the
presence of suitable controllers on the generators. These details which bring ideality to the source
are omitted to simplify the analysis.
2. In the source representation in Figure 2.3, a potential of v(t) exists at terminal A when meas-
ured relative only to terminal B. Therefore, for ease of measurement, and also due to operational
issues, a common reference is chosen where terminal B is grounded. This implies that terminal A
is at higher potential with respect to the ground and the waveform at point A is given by v(t) and
is denoted by a ‘positive’-sign. This also signifes a condition, where if this terminal is touched,
one may get shock depending on the voltage magnitude.
For the circuit shown in Figure 2.3(a), it is assumed that the instantaneous current i1(t) is leaving the
source at the ‘positive’ terminal. In Figure 2.3(b), it can be seen that current i2(t), has been assumed
to enter the source at the ‘positive’ terminal. In both cases, in the direction of the current, the voltage
across the resistor drops. Now, applying Kirchhoff ’s voltage law (KVL) to circuit (a) we have,
v(t ) − i1 (t ) R = 0
or
v (t )
i1 (t ) = = 0.5 sin(2p 50t )A
R
for v(t ) = 1.0 sin(2p 50t ) V and R = 2 Ω.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 27 4/23/2018 5:36:06 PM

 28  |  Power System Analysis

A time–domain plot of these waveforms is given in Figure 2.4 which shows that i1(t) is in ‘phase’
with v(t), denoting a power delivery from the source, since i1(t) is leaving the ‘positive’ terminal of the
source and the resistor element needs to be supplied with power. The corresponding phasor diagram
is also given in Figure 2.4 (in terms of peak amplitude).
Similarly, in Figure 2.3(b), when KVL is applied, we get,

v(t ) + i2 (t ) R = 0

or
v (t )
i2 (t ) = − = −0.5sin(2p 50t )
R

i1 (t) i2 (t)

+ A + A
+ −
v(t) R=2Ω v(t) R=2Ω

B − B
+

(b)
(a)

Figure 2.3  (a) and (b) A Simple Resistive Element Across a Source—Two Conventions.

Voltage and current in a pure R element

1
v(t) 90 1
0.8 120 60

0.6 i1(t)
150 0.5 30
0.4
v(t) and i1(t) in V and A

0.2
Phasor− V
0 180 0
Phasor − I1
−0.2

−0.4
210 330
−0.6

−0.8 240 300

Snapshot 270
−1

−5 0 5 10 15 20
Time (s ) x 10
−3

Figure 2.4  Voltage and Current Plots for Resistive Circuit (a).

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 28 4/23/2018 5:36:09 PM

 Basics of Power Systems  |  29 

v(t)
1 90 1
120 60
0.8

0.6
150 0.5 30
0.4
v(t) and i2(t) in V and A

0.2
Phasor−V
0 i2(t) 180 0
Phasor − I2
−0.2

−0.4
210 330
−0.6

−0.8
Snapshot 240 300
−1 270

−5 0 5 10 15 20
Time (s ) −3
x 10

Figure 2.5  Voltage and Current Plots for Resistive Circuit (b).

The applied voltage and current plots are shown in Figure 2.5, where current i2(t) is ‘out-of-phase’
with the voltage; refer to the phasor diagram.

2.3.1.3  Phasor Representation in a Series Resistor–Inductor Circuit

Consider a series RL circuit as shown in Figure 2.6, where a single-phase voltage, v(t ) = Vm sin(w t + a ),
is applied at t = 0.
For this LTI circuit, (which implies that the parameters, R and L are linear and constant), current
i(t) is determined by writing a state-equation for the circuit as follows:

i(t)
+
+ vR
R
−

v(t)
+
_ L vL
−
R = 1.4142 Ω and L = 4.5016 mH

Figure 2.6  A Series RL Circuit Excited by a Single-phase Voltage.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 29 4/23/2018 5:36:11 PM

 30  |  Power System Analysis

1. First, choose the possible variables in a circuit. For example, this set is given by [vR , i , v L ]T . Note
that v is treated as a constraint on the circuit [8].
2. From the above set, create a subset of variables which satisfy the principle of continuity with
respect to time. This implies that a variable should not exhibit jumps or discontinuity. Such a set
T
is given by [vR , i ] . This is due to the fact that the voltage across the inductor, vL, can have jumps,
and the current through the inductor, i, being an energy variable, cannot change abruptly.
3. In the set of variables which remain continuous, check for linear dependency among the mem-
bers. Sequentially remove all those variables which can be expressed as a linear combination of
the remaining variables. The left-over set constitutes a state-vector of minimum possible number
of variables which completely define the nature of a circuit. For example, vR is linearly related to i
(as vR = R × i ). Hence, remove vR and choose i as the final state-vector. Another possible choice
is that one can retain vR and remove the variable i. In either case, the final state vector contains
only one variable and the state equation should be written only in terms of this state variable.
4. To understand the naturality of a circuit, it is generally convenient to redraw the circuit by com-
pletely deenergising the source and replacing it by its internal impedance. For example, consider
the circuit as shown in Figure 2.7. It is clear from Figure 2.7 that it contains only one energy-­
storing element, i.e., an inductor, describing one degree of freedom in terms of inductor current.
Also, note that only one variable in the final state-vector denotes this nature and hence the order of
the circuit. Only for such first-order circuits, it is customary to define a time–constant. For exam-
L 1
ple, the time–constant is given by t = s. The natural frequency is simply given by neper/s.
R t
Notes:
• The couter part of the neper frequency in an oscillatory system is the radian frequency in rad/s [9].
• The definition of a time–constant is generally limited to first-order circuits. However, even for
higher-order systems, it is customary to use this notion when the naturality is only due to neper
frequencies and their magnitudes vary over a wide range.

i(t)
+
R vR
−

+
L vL
−

Figure 2.7  A Series RL Circuit with a Deactivated Source.

Applying KVL to the circuit shown in Figure 2.3, we have,

v = vR + v L (2.9)

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 30 4/23/2018 5:36:13 PM

 Basics of Power Systems  |  31 

In terms of the state–variable i, we can rewrite (2.9) and obtain the state-model of the circuit as
di R 1
= − i + v (t )
dt L L (2.10)

with initial current on the inductor as i(0).

The complete solution for t ≥ 0, is written as [3],

i (t ) = itr (t ) + iss (t ) (2.11)

where the transient component of current itr(t) is given by

R R
− t Vm − t
itr (t ) = i (0)e L −
|Z|
[ sin(a − f ) ] e L

and the steady-state current iss(t) is given by

Vm
iss (t ) = sin(w t + a − f )
|Z|

with Z = | Z | ∠f = ( R + jwL) , which is the impedance of the circuit.

2.3.1.4  Significance of Phasor Analysis

To understand the significance of phasor analysis, consider the following:
1. Under the sinor stead-state, in (2.10), we can substitute iss(t), in place of i(t). After rearranging,
we get,

Vm d V 
v (t ) = R sin(w t + a − f ) + L  m sin(w t + a − f ) (2.12)
|Z| dt  | Z | 
or

Vm sin (w t + a ) = RI m sin(w t + a − f ) + (w L) I m cos(w t + a − f )

 p
= RI m sin(w t + a − f ) + (w L) I m sin  w t + a − f + 
 2

Vm
where, I m = .
|Z|
2. Now, representing the sinors in terms of peak-phasor quantities, we get,

 p
Vm ∠a = RI m ∠(a − f ) + (w L) I m ∠  a − f + 
 2
= RI m ∠(a − f ) + j (w L) I m ∠(a − f ) (2.13)

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 31 4/23/2018 5:36:34 PM

 32  |  Power System Analysis

3. Using RMS-phasor representation and taking the voltage phasor as the reference, for sinudoidal
steady-state of the circuit we obtain,

V = RI + j (w L) I (2.14)

Therefore, the current phasor, I , can be computed as

V V ∠0 o
I = =
( R + jw L) Z

V
= ∠ − f (2.15)
|Z|

Note that for sinor steady-state calculations, a may be chosen arbitrarily or even set to zero.
4. In the above analysis, Z denotes the equivalent impedance of the RL-series circuit and it is equal
to ( R + jwL) = | Z | ∠f . Here, w represents the supply frequency in rad/s. The impedance, Z, is
valid for a given angular frequency w rad/s under sinor steady-state condition only. In other words,
impedance is defined only for a single-frequency sinudoidal wave and its unit is ohms or Ω.
5. From the above analysis, it can be concluded that if it is required to obtain the nature of the cur-
rent under sinor steady-state (for LTI systems), one can avoid solving the differential equation
(given in (2.10)). It should be noted that even if it is obtained through time–domain solution of
differential equations, one needs to determine the response for a longer time until a steady-state is
reached. Therefore, it involves more computational effort. Instead, we can easily solve the current
under sinor steady-state in terms of phasor quantities as given in (2.15). From this, one can write
an expression for current in time–domain as

 V 
iss (t ) =  2sin(w t − f ) (2.16)
 | Z | 

In the above expression, an arbitrarily chosen a may be inserted to get,

 V 
iss (t ) =  2sin(w t + a − f ) (2.17)
 | Z | 

6. At t = 0, we get,

 V 
iss (0) =  2sin(a − f ) (2.18)
 | Z | 

If we set i(0) = iss(0), i.e., as the initial current in itr(t), the transient component of i(t) dis-
appears. This means that when v(t ) = Vm sin(w t + a ) is switched to the circuit at t = 0, the
circuit settles to a sinor steady-state immediately after switching. This procedure is generally
followed in time–domain analysis to avoid the initial part of simulation so as to obtain the sinor

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 32 4/23/2018 5:36:40 PM

 Basics of Power Systems  |  33 

steady-state response, to begin with. This saves computational effort to impart the operating point
on state-vectors.

Note: In (2.12), all quantities pertain to instantaneous values of sinors at a given frequency, under
steady-state. With this understanding, when they are expressed as phasors as in (2.13), it can be seen
di (t )
that a term containing L ss , is written as j (w L) I , where I is the phasor representation of iss(t).
dt
d
This implies that when [ f (t )] , where f(t) is a single-frequency sinor, is to be represented as a
dt
d F
phasor, simply replace by jw and then write the resulting phasor as jw F with F = m ∠b , where
dt 2
b is the angle associated with f(t). This is valid even for cosine-based reference.
p
For the parameters shown in Figure 2.6, Z = 1.4142 + j × 4.5016 × 10 −3 × (2p 50) = 2∠ Ω . For
4
v(t ) = 1.0sin(2p 50t ) V, the sinor steady-state RMS current is given by

1
p
I = 2 ∠−
2 4
p
= 0.35355∠ − A (2.19)
4
Therefore, the instantaneous value of i(t) under steady-state is obtained as

 p  p
i (t ) = 0.35355 × 2sin  w t −  = 0.5sin  2p 50t −  A (2.20)
 4  4

The voltage and current plots are shown in Figure 2.1, where a short duration negative time is
also indicated to depict the prior existence of sinor steady-state.

2.4  POWER IN SINGLE-PHASE CIRCUITS

In this section, the nature of power in single-phase circuits and their significance are discussed.

2.4.1  Instantaneous Power in a Pure Resistor Circuit

For Figure 2.3 (a), instantaneous power associated with the source is calculated as

Vm I m
p1 = v(t ) × i1 (t ) = Vm I m sin 2w t = (1 − cos(2w t ))
2
= V × I (1 − cos(2w t))
= P (1 − cos(2w t )) (2.21)

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 33 4/23/2018 5:36:47 PM

 34  |  Power System Analysis

Vm Im
where V = and I = denote the RMS values.
2 2

Similarly, for Figure 2.3 (b), instantaneous power associated with the source is calculated as

Vm I m
p2 = v(t ) × i2 (t ) = −Vm I m sin 2w t = − (1 − cos(2w t ))
2
= − P (1 − cos(2w t)) (2.22)

Also, note that in the above analysis, i2(t) = −i1(t).

In (2.21), P denotes the average power, whereas in (2.22), −P represents the average power. In
these circuits, it is well known that a resistor consumes power, or its power requirement must be sup-
plied. This understanding provides us two conventions for calculating the power associated with any
element, as follows:

1. Generation convention: Here, current leaves the positive terminal of the element, signifying
supply of power by the element. If the calculated power is positive, it means that the element is
acting as a power source.
For circuit (a), the power is positive, and hence, the source is supplying an average power to
the resistor.
2. Load convention: Here, current enters the positive terminal of the element, signifying absorption
of power by the element. If the calculated power is positive, it means that the element is acting as
a load.

For circuit (b), the computed power is negative, and hence, the source is absorbing a negative average
power, which means that it is delivering power to the resistor.
The instantaneous power, p1(t), calculated for a pure resistor has the following features (Figure 2.8):

1. It possesses an average value which denotes the power dissipated in the resistor. This value is
generally referred to as real or active power.
2. It oscillates at twice the fundamental frequency.

This example clearly demonstrates that when current direction is to be decided in a circuit, it is purely
arbitrary. Further, to adopt a convention to decide the nature of an element, it is also desired to have
prior knowledge about some elements in the circuit. For example, in the present case, the resistor
brings out the real power delivering ability of the source element. In practice, when transducers are
used for measurement, other settings made related to the elements guide us to understand the nature
of elements.

2.4.2  Nature of Powers in a RL-series Circuit

Consider the RL series circuit shown in Figure 2.6. For a sinor steady-state, taking the applied voltage
as the reference, we have,

v(t ) = Vm sin(w t ) and i (t ) = I m sin(w t − f ) (2.23)

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 34 4/23/2018 5:49:29 PM

 Basics of Power Systems  |  35 

Plots for circuit (a)

1
v(t)
v(t) and i1(t) in V and A

0.5

i1(t)
0

−0.5

−1
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Time (s )

0.5
p1(t)
0.4
p(t) and P in VA and W

0.3
P = 0.25 W

0.2

0.1

0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Time (s )

Figure 2.8  Time-domain Plot of p1(t) for Circuit (a).

In terms of RMS phasor quantity, they are represented as

V = V ∠0 o and I = I ∠ − f (2.24)

V V I
with I = , V = m and I = m .
|Z| 2 2
For the circuit, note that Z =| Z | ∠f with | Z |= R2 + X L2 and XL = wL. The corresponding pha-
sor diagram is shown in Figure 2.9 which shows that the voltage across the resistor and inductor is
written in phasor form as follows:

p 
VR = Vcosf∠ − f and VL = Vsinf∠  − f  (2.25)
2 

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 35 4/23/2018 5:36:53 PM

 36  |  Power System Analysis

For the example, Vm = 1.0 V and V = 0.7071 V. Using this, VR and VL are obtained as

p p p
VR = 0.7071cos ∠ − = 0.5∠ − V
4 4 4
p p p
VL = 0.7071sin ∠ = 0.5∠ V (2.26)
4 4 4

VL

o
(90 − f ) 0
V=V

VL

VR

Figure 2.9  The Phasor Diagram for the Series RL Circuit.

In order to understand the nature of powers, consider the time–domain plot of vR(t) and vL(t) along
with i(t). From (2.23) and (2.26), we have

 p  p
vR (t ) = 0.5 × 2sin  w t −  = 0.7071sin  w t −  V
 4  4
 p  p
v L (t ) = 0.5 × 2sin  w t +  = 0.7071sin  w t +  V
 4   4
 p
i (t ) = 0.5sin  w t −  A (2.27)
 4

Now, consider the instantaneous power, p(t) = v(t) × i(t). Since v(t) = vR(t) + vL(t), we have,

p(t ) = [ vR (t ) + v L (t ) ] × i (t )

Using (2.27) in the above expression, we get,

 p  p  p  p
p(t ) = VRm I m sin  w t −  sin  w t −  + VLm I m sin  w t +  sin  w t −  (2.28)
 4   4   4   4

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 36 4/23/2018 5:36:57 PM

 Basics of Power Systems  |  37 

Using the trigonometric identity,

1
sin( A)sin( B) = [cos( A − B) − cos( A + B )]
2
We can simplify (2.28) as

VRm I m   p   VLm I m
p(t ) =
2 1 − cos2  w t − 4   − 2 cos(2w t ) (2.29)
 
In terms of RMS value of quantities, we have,

  p 
p(t ) = VR I 1 − cos2  w t −   − VL Icos(2w t )
  4

= pR (t ) + pL (t ) (2.30)
where
  p 
pR (t ) = VR I 1 − cos2  w t −  
  4

pL (t ) = −VL Icos(2w t )

These waveforms are shown in Figure 2.10.

The following observations can be made:
1. It is noted in Figure 2.8 that for a pure resistor, the current which is perfectly ‘in-phase’ with the
voltage wave denotes power dissipation, whose average value is given by P = VR × I. Similarly, in
a series RL circuit, vR(t) is ‘in-phase’ with i(t). Therefore, the corresponding average power given
by P = VR × I, represents the real or active power.
2. In comparison to the real power, when vL(t) is observed, it leads the current wave by exactly 90°.
This implies that like a resistor, it does not dissipate power as it possesses no ‘in-phase’ compo-
nent. However, the presence of an inductor continues to load the source by drawing current. This
means that even if it does not consume any real power, the source must be designed to deliver
a current demanded by the an inductor. This rating requirement is simply computed (similar to
the resistor-case) as QL = VL × I. This is attributed as reactive power associated with a reactive
element.
3. With the above representation for real and reactive powers, the instantaneous power waveform
expression given in (2.30) is rewritten as

  p 
p(t ) = P 1 − cos2  w t −   − QL cos(2w t ) (2.31)
  4

It can also be seen that pR(t)-component possesses an average value, whereas the remaining compo-
nent of the instantaneous power, pL(t), which is associated with the reactive element does not have
any average value.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 37 4/23/2018 5:37:01 PM

 38  |  Power System Analysis

Power plots in RL−series circuit

1
p(t)
v(t)
v(t), i(t)and p(t) in V, A, VA
i(t)
0.5

−0.5

−1
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Time (s )
1
i(t), vrR(t) and vL(t) in A, V, V

VR(peak) = 0.70711, VL(peak) = 0.70711

vR(t)
0.5

0 i(t)
vL(t)
−0.5

−1
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Time (s )

0.4
p(t), pR(t), and pL(t) in VA

p(t) pR(t)
0.3
0.2 P=0.17678 W

0.1
0
QL=0.17678 VAR
−0.1
pL(t)
−0.2
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Time (s )

Figure 2.10  The Time–Domain Plots of Various Waveforms Related to Series RL Circuit.

4. Using load convention, the net power adsorbed by the source is obtained as

ps (t ) = v(t ) × − i (t ) (2.32)
In the above equation, −i(t) denotes a current entering the source element. Assume that v(t) = vp(t) +
vq(t). Here, the in-phase component, vp(t), is identical to vR(t) and, the quadrature component,
vq(t), is same as vL(t). Therefore, we can write,
ps (t ) = (v p (t ) + vq (t )) × ( −i (t )) (2.33)

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 38 4/23/2018 5:37:02 PM

 Basics of Power Systems  |  39 

Now, using (2.27), the components of instantaneous power are given by

ps (t ) = p p (t ) + pq (t ) (2.34)

where
  p 
p p (t ) = − P 1 − cos2  w t −  
  4

pq (t ) = QL cos(2w t )

Power plots in RL−series circuit for source element

1
v(t), −i(t) and ps(t) in V, A, VA

v(t)
−i(t)
0.5

0 ps(t)

−0.5

−1
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Time (s )
1
−i(t), vp(t) and vq(t) in A, V, V

vp(t)
0.5
vq(t)

0
−i(t)

−0.5

−1
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Time (s )
ps(t), pp(S), and pq(t) in VA

pq(t) QL = 0.17678 VAR

0.1
0
−0.1
P=−0.17678 W
−0.2
−0.3
ps(t)
−0.4 pp(t)

0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Time (s )

Figure 2.11  The Time–domain Plots of Various Waveforms Related to Source-­element in RL

Circuit.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 39 4/23/2018 5:37:04 PM

 40  |  Power System Analysis

As noted in the pure resistor case, the real power demanded by the R-element is met by the
power component, pp(t), which possesses an average value (Figure 2.11). This implies that the QL
demand of the L-element is supported only by the instantaneous component, pq(t), which does
not contain any average value. Furthermore, it can be seen that since load convention is employed
both for L-element and source-element, the inductor power requirement, where current lags vL(t),
is supported only by the presence of another reactive-natured source-element which absorbs a
current which leads vq(t). Therefore, it can be concluded that in order to deliver a reactive element
of one nature, another reactive element of the opposite nature should be present in the circuit.
In this example, the source-element can be said to either absorb reactive power of the opposite
nature or supply reactive power of the same nature as that of the L-element.
5. Since impedance is given by,

Z = | Z | ∠f = | Z | cos(f ) + j | Z | sin(f ) = R + jX L

and from (2.25), we write that

R XL p 
VR = V ∠ −f and VL = V ∠ −f .
|Z| Z  2 

V
Since we know that I = , we get,
|Z|
VR = IR and VL = IX L

The real power and reactive powers are also given by

VR2 VL2
P = I 2R = and QL = I 2 X L =
R XL
The unit real power is watts or W, and the reactive power is volt-amperes reactive (VAR).

2.4.3  Nature of Powers in a RC-series Circuit

In the previous example, the current, i(t), being an energy variable, decides the magnetic field energy
in the inductor. In the series RC circuit (Figure 2.12), the voltage across the capacitor, vC(t), denotes
the energy stored in the electric field. Therefore, it qualifies as a state-variable and the only state-vector
is vC(t). In terms of this variable, a voltage equation is written as,

v(t ) = vR (t ) + vC (t ) (2.35)

dvC (t )
Since vR(t) = i(t)R and i(t) is the current entering the capacitor, we have i (t ) = C . Substituting
dt
these quantities in the above expression, we get the desired state-space equation as

dvC (t ) v (t ) v(t ) (2.36)

=− C +
dt RC RC

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 40 4/23/2018 5:49:46 PM

 Basics of Power Systems  |  41 

i(t)
+
+ vR
R
−

v(t)
+
_ C vC
−

R = 1.4142 Ω and C = 2.2508 mF

Figure 2.12  A Series RC Circuit Excited by a Single-phase Voltage.

If the requirement is to obtain the sinusoidal steady–state response, one can avoid the explicit differ-
ential equation solution and perform the phasor analysis as follows:
d
Replace by jw and rewrite the equation in phasor form as
dt
1 1
jw VC = − VC + V (2.37)
RC RC
Solving for VC , we get,

V
VC = (2.38)
(1 + jw RC )

An expression for phasor current can be obtained as

1 V 
I =  V− 
R (1 + jw RC ) 

jw CV
= (2.39)
(1 + jw CR)

Rearranging the above expression, we get,

V 
I =   (2.40)
 Z

where, Z denotes the equivalent impedance of the RC-series circuit and is given by
 1  1
Z = R+  = ( R − jX C ) =| Z | ∠ − f with X C = .
 jw C  wC
For the parameters shown in Figure 2.12, we have,

1 p
Z = 1.4142 + −3
= 2∠ − Ω
j (2p 50)2.2508 × 10 4

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 41 4/23/2018 5:37:13 PM

 42  |  Power System Analysis

For v(t ) = 1.0sin(2p 50t ) V, the sinor steady-state RMS current is given by

1
p
I = 2∠
2 4
p
= 0.35355∠ A (2.41)
4

The corresponding phasor diagram is shown in Figure 2.13.

VR

VC
f

o
V=V 0 (90 − f )

Figure 2.13  The Phasor Diagram for the Series RC Circuit.

From the figure the voltage across the resistor and capacitor is written in phasor form as,

p p p
VR = 0.7071cos ∠ = 0.5∠ V
4 4 4
p p p
VC = 0.7071sin ∠ − = 0.5∠ − V (2.42)
4 4 4

Now, consider the time–domain plot of vR(t) and vC(t) along with i(t) as

 p  p
vR (t ) = 0.5 × 2sin  w t +  = 0.70711sin  w t +  V
 4  4
 p  p
vC (t ) = 0.5 × 2sin  w t −  = 0.70711sin  w t −  V
 4   4
 p
i (t ) = 0.5sin  w t +  A (2.43)
 4
To determine the instantaneous power, consider, p(t) = v(t) × i(t) or

p(t ) = [ vR (t ) + vC (t ) ] × i (t )

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 42 4/23/2018 5:37:17 PM

 Basics of Power Systems  |  43 

Using (2.43) in the above expression, we get,

p(t ) = pR (t ) + pC (t ) (2.44)
where
  p 
pR (t ) = VR I 1 − cos2  w t +  
  4
pC (t ) = −VC Icos(2w t )

These waveforms are shown in Figure 2.14.

Power plots in RC−series circuit

1
v(t), i(t) and p(t) in V, A, VA

v(t) p(t)
i(t)
0.5

−0.5

−1
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Time (s)

1
i(t), vR(t) and vC(t) in A, V,V

vR(t)
vC(t)
0.5

i(t)
0

−0.5

−1
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Time (s)
p(t), pR(t), and pC(t) in VA

0.4
p(t)
pR(t)
P=0.17678W
0.2

0
Qc=0.17678 VAR
pC(t)
−0.2
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Time (s)

Figure 2.14  The Time–domain Plots of Various Waveforms Related to Series RC Circuit.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 43 4/23/2018 5:37:19 PM

 44  |  Power System Analysis

From Figure 2.14, the following observations can be made:

1. In series RC circuit, vR(t) is ‘in-phase’ with i(t). Therefore, the corresponding average power
given by P = VR × I, represents the real or active power.
2. In comparison to the real power, when vC(t) is observed, it lags the current wave by exactly 90°.
This implies that like a resistor, it does not dissipate power as it has no ‘in-phase’ component.
However, as an inductor, the capacitor continues to draw current from the source. This rating
requirement is simply computed as QC = VC × I. This is attributed as reactive power associated
with a reactive element whose nature is just opposite to that of the inductor.
3. With the above notations for power, we can rewrite pR(t) and pC(t) as

  p 
pR (t ) = P 1 − cos2  w t +  
  4

pC (t ) = −QC cos(2w t )

4. As in the series RL circuit, using load convention, the net power adsorbed by the source is obtained as

ps (t ) = p p (t ) + pq (t ) (2.45)

where
  p 
p p (t ) = − P 1 − cos2  wt +  
  4

pq (t ) = QC cos(2w t )

These plots are depicted in Figure 2.15. Here, the in-phase component, vp(t), is identical to vR(t)
and, the quadrature component, vq(t), is the same as vC(t). Therefore, we can see that vp(t) and
–i(t) represent a negative real power absorption, whereas vq(t) leads –i(t). In other words, when
the source–­element is treated as a load, it draws a lagging current relative to vq(t), exhibiting an
inductive behaviour. This clearly demonstrates that the source–element either absorbs reactive
power of the opposite nature or supplies reactive power of the same nature as the C-element.
5. As in the series RL circuit, we can easily see that the real power and reactive powers are also given by

VR2 VC2
P = I 2R = and QC = I 2 X C =
R XC

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 44 4/23/2018 5:37:22 PM

 Basics of Power Systems  |  45 

Power plots in RC−series circuit for source element

1
v(t), −i(t) and ps(t) in V, A, VA −i(t)
v(t)
0.5
ps(t)
0

−0.5

−1
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Time (s )

1
−i(t), vp(t) and vq(t) in A, V, V

vq(t) −i(t)
0.5

0
vp(t)

−0.5

−1
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Time (s )

0.2
ps(t), pp(t), and pq(t) in VA

pq(t) Qc = 0.17678 VAR

P =−0.17678 W
−0.2

pp(t) ps(t)
−0.4
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Time (s )

Figure 2.15  The Time–domain Plots of Various Waveforms Related

to Source–element in RC Circuit.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 45 4/23/2018 5:37:23 PM

 46  |  Power System Analysis

2.4.4  Real Power Supply–Importance of Reactive Powers

The previous two examples demonstrated that whenever a reactive element of one type is used, it
necessitates the presence of another reactive element of the opposite nature. This balance is further
analysed in the following examples.

2.4.4.1  Load Resistor Delivered Through an Inductor, Circuit-1

Consider a system in which a pure, resistive load is delivered through an inductor (Figure 2.16). This
example represents a case where real power load is supplied through a line of dominant inductance.
Under sinusoidal steady-state condition of the circuit, with an earlier understanding about the
RL-series circuit, we have the following:
1. The source current is given by
V ∠0°
Is = (2.46)
Z

v(t) is (t) vL (t)

+ L − V
+ +
f
R Vq
−
Is
Vp

Figure 2.16  A Pure Resistive Load Delivered Through an Inductor, Circuit-1.

For V = 0.7071∠0o V, and L = 1.5915 mH, R = 1 Ω and frequency = 50 Hz, we get,

0.7071∠0°
Is = = 0.6325∠ − 26.56° A (2.47)
(1 + j 0.5)
2. The load voltage is given by

VL = I s × R = 0.6325∠ − 26.56° × 1 = 0.6325∠ − 26.56° V (2.48)

3. The voltage drop across the inductor is given by

V1 = I s ( jw L) = 0.3162∠63.44° V

4. The real power consumed by the resistor is given by

P = I s2 R = 0.4 W

The reactive power associated with the inductor (with load convention) is obtained as
QL = I s2 X L = 0.63252 × 0.5 = 0.2 VAR

5. To compute the reactive power associated with the source, we first obtain Vq (Figure 2.16). This is
equal to V × sin(26.56) = 0.7071sin(26.56) = 0.31617 V. Now, compute, Qq = Vq × Is = 0.2 VAR.
With the load convention, it implies that the source balances the reactive power of the inductor.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 46 4/23/2018 5:37:27 PM

 Basics of Power Systems  |  47 

6. Similarly, the Vp component, equal to 0.7071cos(26.56) = 0.6325 V, balances the load real power.
This can be verified as Pp = Vp × Is = 0.4 W.

It is desired to improve the load voltage so that real power is delivered adequately to the load. In an
attempt to achieve this, the following two possibilities are considered:
1. Connecting an inductor across the load terminals.
2. Connecting a capacitor across the load terminals.
Note: These examples are considered to demonstrate the voltage-controlling effects of the reactive
powers. Through these examples, it is possible to appreciate the ‘loading’ and ‘supporting’ influence
of reactive powers. In literature, this is generally treated as system compensation to enhance the real
power transfer ability. In these examples, low source voltage is used for ease of illustration.

2.4.4.2 Performance with an Inductor Across the Load Resistor,

Circuit-2
In this case, an inductance L2 = 12.732 mH is connected across the load resistor as shown in
Figure 2.17.

v(t) v L (t)
v1(t) i s (t)

+ + L − + R L2 +
vR(t) v2(t)
I
− II −
i 2(t)

Figure 2.17  Inductance Across the Resistive Load, Circuit-2.

In order to obtain the sinusoidal steady-state model for the circuit, a state-space model is derived.
Identifying the energy variables as is and i2, a state-vector is created as x = [is , i2 ]T . In terms of these
variables, a state model is determined as follows:
Applying KVL to loop-I, we get,
v(t ) = v1 (t ) + vR (t )
dis
Since, v1 (t ) = L and vR (t ) = R(is − i2 ) , the above expression can be rearranged to get,
dt
dis R R 1
= − is + i2 + v (2.49)
dt L L L
Similarly, by applying KVL to loop-II, we get
di2
v2 (t ) = L2 = vR (t ) = R(is − i2 ) (2.50)
dt
or
di2 R R
= is − i2 (2.51)
dt L2 L2

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 47 4/23/2018 5:37:31 PM

 48  |  Power System Analysis

From (2.49) and (2.51), a state–space model in matrix form can be written as
d
x = [ A] x + [ B]v (2.52)
dt
where
 R R 
− L L  1
[ A] =   and [ B ] =  L 
R R  
− 0 
 L2 
L2 

Using the approach followed in (2.37), from the above state-model, if a sinusoidal response is to be
d
obtained, substitute jw in place of and rewrite for X ( jw ) as
dt
−1
X ( jw ) = ( jw [U ] − [ A]) × [ B]V (2.53)

where [U ] is an identity matrix of dimension 2 × 2.

Equivalent Impedance Calculation of Circuit-2:

It is to be noted that instead of using the state–space model to arrive at the sinudoidal steady-state calcu-
d
lations (with jw replacement for ), a straightforward impedance model can be established as follows:
dt
1. Obtain the parallel equivalent impedance of R and L2 as outlined below:
In loop-II, we can see that
di2
L2 = −i2 R + is R Is I2
dt
Writing the above expression in phasor form, we get,
di di di2
L2 2 L=2 −i22 R=+−isiL2R2R jw L +Isis R 2s R-+I2is RR+ Is R I2
= I−2iI=
dt dt dt 2
di2 di2
2 = −i2 R and
LRearranging + is RL 2 Is =I2−in
writing + is Rof Is , we
i2 Rterms I2 obtain,
dt dt
R
I2 = I s (2.54)
( R + jw L2 )
di
di22 di
LLWe
22 know == −−ithat = ( I2R
i RR++ViiLRR I =×- −IjIw R +Replacing
L
i )R. i R Is I2 using (2.54), we get the parallel equivalent imped-
dt
dt Z pL22 = Rss2= dtss 222 2 s
ance of R and LI2s as ( R + jw L2 )
V R × jw L2
Z pL = R = (2.55)
I s ( R + jw L2 )

2. Now, using (2.55), the net impedance of the circuit as ‘seen’ from the source terminals can be
determined as
R × jw L2
ZTL = ( jw L) +
( R + jw L2 )

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 48 4/23/2018 5:37:36 PM

 Basics of Power Systems  |  49 

Once, the ZTL is obtained, the circuit is analysed as follows in an effort to improve the load voltage:
di
L2 2 =by−the
1. The current delivered + is R Is, is Icalculated
i2 Rsource, 2 as
dt
V
Is = (2.56)
ZTL
 j4 
where ZTL = j 0.5 +  = (0.9412 + j 0.7353) Ω .
 (1 + j 4) 
di
With V = 0.7071 L2 ∠2 0°
= V,
−i2we
R +obtain
is R Is = 0.5920
I2 ∠ -37.99° A.
Is = dt
ZTL
2. The voltage drop across the inductor L is given by
V1 = I s ( jw L) = 0.2960 ∠52o V

3. The load voltage VL is obtained as

VL = V − V1 = 0.5744 ∠ − 23.96o V (2.57)

4. The current through the inductor L2 is given by

VL
I2 = I s − = 0.1436∠ − 113.96o A
R
5. The real power consumed by the resistor is given by

VL2
P= = 0.3299 W
R
The reactive power associated with the inductors L and L2 (with load convention) is obtained as

QL = I s2 X L = 0.59202 × 0.5 = 0.1752 VAR

and
VL2 0.57442
QL 2 = = = 0.0825 VAR
XL 4

Note that both the inductors absorb reactive power of the same nature.
6. To compute the reactive power associated with the source we first obtain Vq as in circuit-1. This
is obtained as V × sin(37.99) = 0.4352V. Now, compute, Qq = Vq × I s = 0.2577 VAR. With the
load convention, it implies that the source balances the reactive power of both the inductors since
Qq = QL + QL 2 .
7. Similarly, it can be verified that Pp = V p × I s = Vcos(37.99) × 0.5920 = 0.3299W.
In comparison to the previous circuit, it can be concluded that by connecting an inductor across the
load, there is a reduction in the load voltage rather than aiding it. Now, let us connect a capacitor
instead of an inductor across the load resistor.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 49 4/23/2018 5:37:41 PM

 50  |  Power System Analysis

2.4.4.3 Performance with a Capacitor Across the Load Resistor,

Circuit-3
In this case, a capacitor of C = 1.5915 mF is connected across the load resistor as shown in Figure 2.18.

v(t) vL (t)
v1 (t) iS (t)

+ L −
+ +
+
R vC(t)
− C
−
i (t) i (t)
R
C

Figure 2.18  Capacitor Across the Resistive Load, Circuit-3.

T
For the circuit, a state-vector is formed using the energy variables as x = [is , vC ] .
A state-space model is obtained as follows:
dis
We know that v = L + vC . From this we get,
dt
dis 1 1
= − vC + v (2.58)
dt L L
dvC
Now, applying Kirchhoff ’s current law (KCL) at the load bus, we have is = iR + iC . Since iC = C
dt
vC
and iR = , we get another state-equation as
R
dvC 1 1
= is − vC (2.59)
dt C RC
Writing (2.58) and (2.59) in matrix form, we obtain
d
x = [ Ac ] x + [ Bc ]v (2.60)
dt
where
 1 
0 − 1
[ Ac ] =  L  and [ B ] =  
 c L
1 1   
− 0
 
 C RC 

Equivalent Impedance Calculation for Circuit-3:

Instead of using the state-equations to determine the sinudoidal response the impedance model is
derived as follows:
1. Obtain the parallel equivalent impedance of R and C. We know that at load bus,

is (t ) = iR (t ) + iC (t ) (2.61)

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 50 4/23/2018 5:37:45 PM

 Basics of Power Systems  |  51 

vC dvC
Note that iR (t ) = , iC (t ) = C and vC (t ) = v L (t ) . Now, writing the above expression
R dt
in phasor form we get,
VL
Is = + jw CVL
R
1 
= VL  + jw C 
R 

1 1 
= VL  +
 R − jX C 

VL
= (2.62)
Z pC
R( − jX C ) 1
where Z pC = with X C = .
( R − jX C ) wC
2. Now using (2.62), the net impedance of the circuit as ‘seen’ from the source terminals can be
determined as
R( − jX C )
ZTC = ( jw L) +
( R − jX C )

Once ZTC is obtained, Circuit-3 can be analysed as follows:

di
L2 2 =by−the
1. The current delivered + is R Is, is Idetermined
i2 Rsource, 2 as
dt
V
Is = (2.63)
ZTC

 − j2 
where ZTC = j 0.5 +  = (0.8 + j 0.1) Ω .
 (1 − j 2) 
di2
With V = 0.7071L2∠ 0° =
V, −the + is R Is = 0.8770
i2 Rcurrent, I2 ∠ -7.125° A.
Is = dt
ZTL voltage V is obtained as
2. The load L

VL = V − I s ( jwL) = 0.7845∠ − 33.69° V (2.64)

3. Using VL, the capacitor current can be obtained as

VL
IC = = 0.3922∠56.31° A
− jX C
Similarly,
V1 = I s ( jwL) = 0.4385∠82.88° V

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 51 4/23/2018 5:37:52 PM

 52  |  Power System Analysis

These phasors are plotted in Figure 2.19.

90 90
1 1
120 60 120 60

150 0.5 30 150 0.5 30

V1 Ic
Vq
V V
180 −Is 0 180 0

Is Is

VL
210 330 210 330

240 300 240 300

270 270

Figure 2.19  Phasor Diagram for Circuit-3.

4. The real power consumed by the resistor is given by

VL2
P= = 0.6154 W
R
5. The reactive power associated with the inductor, L (with load convention), is obtained as

QL = V1 × I s = 0.4385 × 0.8770 = 0.3846 VAR

and the reactive power associated with the capacitor C, (with load convention), is given by

QC = VL × I C = 0.7845 × 0.3922 = 0.3077 VAR

From the phasor diagram, note that the current through the inductor  Is lags the voltage V1,
whereas for the capacitor, the current IC leads the voltage VL. This clearly shows that the nature
of reactive powers is just opposite to one another.
6. To compute the reactive power associated with the source, we first obtain Vq. From Figure 2.19,
we can see that Vq = V × sin(7.125) = 0.0877 V. Now, compute Qq = Vq × Is = 0.0769 VAR. With
the load convention, the current - Is leads Vq. This demonstrates that the nature of source reactive
power is identical to that of the capacitor. If we consider the reactive power balance for the circuit,
we can write that
QL = 0.3846 = Qq + QC = 0.0769 + 0.3077

7. Similarly, it can be verified that Pp = V p × I s = Vcos( 7.125) × 0.8770 = 0.6154 W.

The above analysis repeated for C = 3.1831 mF, i.e., for Xc = 1.0 Ω. It is seen that ZTc = 0.5 Ω and
Is = 1.4142 ∠ 0° A. This indicates that Is is ‘in-phase’ with V. The load voltage, VL = 1.0 ∠ -45° V.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 52 4/23/2018 5:37:55 PM

 Basics of Power Systems  |  53 

The real power consumed by the resistor is 1 W. The reactive power associated with the inductor L, is
1 VAR, and that associated with the capacitor is 1 VAR. Also, note that since Vq = 0, the source does
not supply any reactive power.

From the above analyses, the following observations are made:

1. In Circuit-1, while delivering real power requirement of the resistor, the source also supplies
inductive reactive power. This means that the source should have reactive power-sourcing ability.
2. In Circuit-2, in an attempt to improve the load voltage magnitude, the inductor L2 reduced the load volt-
age. Therefore, for any L2, it is not possible to increase the load voltage above the value that obtained in
Circuit-1. Further, any addition of L2 would make the source to supply extra reactive power.
3. In Circuit-3, when Xc = 2 Ω, the load voltage increased from 0.6325 V (Circuit-1) to 0.7845 V (in fact,
above the source voltage). This is feasible mainly because the capacitor balanced a part of the induc-
tive reactive power requirement, relieving the source. This is evident in Circuit-1, where the source
supplied the entire inductive reactive power, whereas in Circuit-3, the inductive reactive power supplied
by the source has reduced to 0.0769 VAR. However, when Xc is set to 1 Ω, the reactive power delivered
by the source is reduced to zero and the capacitor supplied the entire inductive reactive power.
4. Due to its reactive power demand-imposing nature, an inductor generally causes bus voltage to
drop. Therefore, an inductor is referred to as a reactive power load on the system. In this context,
a capacitor is treated as a source of reactive power. However, the source-element can either sup-
ply/source or absorb reactive power. In these attributions, a load-convention is employed. This
implies that in this convention, if inductive reactive power is noted as a positive Q (as in a resistor
which consumes real power), a negative Q represents the reactive power associated with a capac-
itor, denoting reactive power generation.
5. Circuit-3 illustrates that when voltage at a bus is to be regulated, it can be locally achieved by a
reactive power-supporting device. However, real power requirement of a load can be met only by
a source-element through an energy conversion process. In a conventional generator, this energy
conversion process involves conversion of energy from mechanical form to electrical form, at a
given frequency. This implies that a real power delivery influences the network-wide frequency
and demands a constancy of frequency. Further, it should be noted that frequency regulation can-
not be locally done as it is a mechanical variable.
6. If reactive power support is provided locally, the generator capacity may be available for deliver-
ing additional real power without overloading it.
7. In Circuit-1, while delivering the desired real power, the source is required to support the reactive
power drawn by the inductor L, which represents a line (a media) which simply conveys energy.
Therefore, a portion of the generator volt-ampere capacity is used up in conveying real power to the
load. This is treated similar to a ‘loss’ which takes place in a resistor, but is referred to as reactive
power loss with VAR as its unit. For example, in Circuit-1, this loss is equal to 0.2 VAR. In Circuit-2,
this loss increases to 0.2577 VAR. In Circuit-3, when Xc = 2 Ω, the inductor L, consumes 0.3486
VAR. However, the source supplies only 0.0769 VAR (inductive). Therefore, the reactive loss is only
0.0769 VAR. Now, for Xc = 1.0 Ω, the capacitor balances the inductor reactive power completely,
placing no demand on the source-element. Hence, in this case, the reactive power loss is zero.
8. Unless the magnitude of the load bus voltage is too small as against the source-bus voltage, it can be
generally said that real power flows from a bus of ‘leading’ voltage phasor to a bus of ‘lagging’ voltage

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 53 9/12/2017 7:32:40 AM

 54  |  Power System Analysis

phasor. For example, in Circuit-1, V leads VL. This observation is valid for other two circuits as well.
However, reactive power (inductive) flows from a bus of higher voltage magnitude towards a bus hav-
ing lower voltage. For example, in Circuit-3, a capacitor at the load bus delivers a major reactive power.
9. In Circuit-3, with capacitor connection, the load voltage increases above the source voltage
depending on the value of XC. Therefore, while compensating a system for voltage support, care
should be taken to see that the load voltage is well within the permissible limit.

2.4.5  Complex Number Representation of Real and Reactive Powers

With knowledge of real and reactive powers in a system under a single-frequency sinusoidal steady-
state condition, it necessary to represent them in terms of voltage and current phasors for a given
element. In order to establish this, the following steps are employed:
1. Mark the known voltage phasors in a circuit. The phasor current directions are marked arbitrar-
ily. Applying KVL and KCL, obtain the governing equations for the circuit. Solve these system
of equations for unknowns. If the circuit is LTI, the solution can be determined analytically.
However, if the system is non-linear (arising due to specification of known quantities), an iterative
solution scheme should be employed.
2. Obtain phasor quantities pertaining to voltage across the element and current through the ­element,
for all elements.

2.4.5.1  Complex Power Representation in an RL-series Circuit

Consider an RL series circuit as shown in Figure 2.20. This circuit is identical to the one shown in
Figure 2.16, except that the instantaneous values of variables are replaced by phasor quantities.

V Is VL
o
+ − V=V 0 (90 −f )
+ L +
V1 R Vq
−
Is
Vp

Figure 2.20  Circuit-1 is Marked with Phasor Quantities.

Let us consider the powers associated with the source-element. From the phasor diagram given in
Figure 2.20, we can write that the real power, P = V p × I s and the reactive power, Q = Vq × I s .
Consider,
V × conj( I s ) = (V p + Vq ) × I s* (2.65)
From the phasor diagram, we have,
p 
V p = V p ∠ − f ; Vq = Vq ∠  − f 
2 
and I s = I s ∠ − f .
We also know that conj( I s ) = I s* = I s ∠f .

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 54 9/12/2017 7:32:46 AM

 Basics of Power Systems  |  55 

Using the above quantities in (2.65), we get,

p
V × I s* = V p × I s + Vq × I s ∠
2
= V p × I s + jVq × I s

S = P + jQ (2.66)
Therefore, while representing the real power, P, and reactive power, Q, they are simply written as a
complex number, S = P + jQ , where, S is referred to as the complex power. Note that in the circuit,
the phasor current, I s , is marked as ‘flowing out’ of the positive terminal of the source. In this proce-
dure, both P and Q are positive real numbers. This is because the source has to deliver a resistive load
and in addition, it needs to supply an inductive load. This is referred to as the ‘generator convention’.
The above phasor current, I s , can also be viewed as current ‘flowing into’ the load’s positive ter-
minal. As in the source-element case, the complex power can be computed as

S L = V × I s* = PL + jQL = VR × I s + jVL × I s (2.67)

where, VR is same as V p and VL is identical to Vq .

Note that S L is the complex power consumed by the load with both PL and QL as positive real
numbers. Here, it can be seen that the inductive reactive power, QL , is also treated as a ‘load’ on the
system similar to the real power PL , which is associated with the resistor. This method of determining
power is referred to as ‘load convention’.

2.4.5.2  Complex Power Representation in an RC-series Circuit

For an RC series circuit shown in Figure 2.12, the instantaneous values of variables are replaced with
phasor quantities as shown in Figure 2.21.

I
+
Vp
+ VR
R
−
I
V
+ Vq

_ C VC
o
− V=V 0 (90− f )

Figure 2.21  RC Series Circuit Marked with Phasor Quantities.

Here also, let us obtain the powers associated with the source-element in terms of phasor quantities.
The real power, P, is given by P = V p × I and the reactive power, Q, is calculated as Q = Vq × I .
Consider,
V × conj( I ) = (V p + Vq ) × I * (2.68)

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 55 9/12/2017 7:32:57 AM

 56  |  Power System Analysis

From the phasor diagram, see Figure 2.21, we have,

p 
V p = V p ∠f ; Vq = Vq ∠ −  − f 
2 
and I = I ∠f .
We also know that conj( I ) = I * = I ∠ − f .
Using the above quantities in (2.68), we get,
p
V × I * = V p × I + Vq × I ∠ −
2
= V p × I − jVq × I
S = P + jQ (2.69)

Note that in this circuit the phasor current, I is marked as ‘flowing out’ of the positive terminal of
the source. For this indication, P is positive as the source is delivering a resistive load, whereas Q is
a negative real number, unlike in the previous case, as the source is now delivering a capacitor whose
nature is opposite to that of an inductor. This is referred as to the ‘generator convention’.
In the circuit, the phasor current, I , can also be treated as current ‘flowing into’ the load’s positive
terminal. Therefore, for the load, the complex power can be computed as
S L = V × I * = PL + jQL = VR × I − jVC × I (2.70)

where, VR is same as V p and VC is identical to Vq .

Note that S L is the complex power consumed by the load with PL as positive and QL as negative. This
is consistent with the notation where the inductive reactive power is designated as a positive load. Similarly,
in comparison to the resistive load which draws power, the capacitor draws a negative (reactive) power,
implying generation. This is referred to as ‘load convention’. In this case, the source absorbs the reactive
power generated by the capacitor behaving like an inductor. Hence, the QL of the source should be positive.
The above two conventions employed for power representation are summarised in Tables 2.1 and 2.2.

Table 2.1  Generator Convention for Source-element in RL-series and RC-series Circuits.

Element RL-series RC-series

( S = P + jQ )
Source P positive P positive
Source Q positive Q negative

Table 2.2  Load Convention for Load-element in RL-series and RC-series Circuits.

Element RL-series RC-series

(S L = PL + jQL )

Load PL positive PL positive

Load QL positive QL negative

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 56 9/12/2017 7:33:07 AM

 Basics of Power Systems  |  57 

2.4.6  Power Balance in Single-phase Circuits

In the previous sections, it is shown that the power requirement of every element in a circuit should
be supported by other elements in the circuit so that there is a real and reactive power balance. Such
calculation verifies the solution obtained for a circuit where current directions are arbitrarily chosen
to begin with. This also validates the control settings made for an element. For checking this power
balance, a certain power calculation–convention—generator or load convention—should be employed
and consistently used for all elements. This power balance is illustrated for three circuits indicated
earlier.
In Table 2.3, power balance is shown for Circuit-1.
V
In Table 2.4, power balance is shown for Circuit-2. Note that VR = V2 = VL and I R = R .
R
In Table 2.5, power balance is shown for Circuit-3. In the table note that VR = VC = VL and
V
IR = R .
R
Notes:
1. In Tables 2.3, 2.4, and 2.5, it is shown that for both the conventions, the real power associated
with each element adds to zero. Similarly, the the sum of the reactive power of each element is
also zero. This is a characteristic feature of an AC system and is generally referred to as load-­
generation balance. Such a power balance indicates the existence of a sinusoidal steady-state for
a system at a given frequency and voltage profile.

Table 2.3  Power Balance for Circuit-1.

Element Generator convention Load convention

P (W) Q (VAR) PL (W) QL (VAR)

Re[ V I s*] Im[ V I *]

s Re[ V ( − I s )* ] Im[ V ( − I )* ]
s
Source
= 0.4 = 0.2 = −0.4 = −0.2

Inductor L Re[ V1 ( − I s )* ] Im[ V1 ( − I s )* ] Re[ V1 I s* ] Im[ V1 I s* ]

=0 =−0.2 =0 = 0.2

*
Re[ VL ( − I s ) ] Im[ VL ( − I s )* ] Re[ VL I s* ] Im[ VL I s* ]
Resistor R
= −0.4 =0 = 0.4 =0

∑ P=0 ∑ Q=0 ∑ PL = 0 ∑ QL = 0
Total

2. In the above circuits, the loads are specified in terms of impedance offered by them. This ena-
bles direct solution for line currents by using analytical expressions. Once currents in different
elements are computed, power-flow calculations can be made. However, in a conventional power
system, these loads are specified as complex power required by the loads instead of their imped-
ances. In such cases, the determination of a solution is not straightforward as it involves iterative

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 57 9/12/2017 7:33:23 AM

 58  |  Power System Analysis

Table 2.4  Power Balance for Circuit-2.

Element Generator convention Load convention

P (W) Q (VAR) PL (W) QL (VAR)

Re[ V I s* ] Im[ V I s* ] Re[ V ( − I s )* ] V ( − I s )* ]

Im[
Source
= 0.3299 = 0.2577 = −0.3299 = −0.2577

Inductor L Re[ V1 ( − I s )* ] Im[ V1 ( − I s )* ] Re[ V1 I s* ] Im[ V1 I s* ]

=0 =−0.1752 =0 =0.1752

Inductor L2 Re[ V2 ( − I 2 )* ] Im[ V2 ( − I 2 )* ] Re[ V2 I 2* ] Im[ V2 I 2* ]

=0 = −0.0825 =0 = 0.0825
Resistor R Re[ VR ( − I R )* ] Im[ VR ( − I R )* ] Re[ VR I R ]
* *
Im[ VR I R ]

= −0.3299 =0 = 0.3299 =0

Total
∑ P=0 ∑ Q=0 ∑ PL = 0 ∑ QL = 0
schemes. This is because power is a function of not only voltage and current, but also a function
of the angle between them.

2.4.6.1  Notations for Generator and Load Conventions

Let Ve = Ve ∠fv be the voltage across an element with the polarity as shown in Figure 2.22. Let
I e = I e ∠f i be the current through the element.
It is to be noted that the direction of the current decides the convention adopted for represent-
ing the complex power. If I e is assumed to leave the positive terminal of the element, it corre-
sponds to generator convention and if it is taken as entering the positive terminal, it implies load
convention.
Ie Ie

+ +

Element Element
Ve Ve

− −

Generator convention Load convention

Figure 2.22  Notations for Generator and Load Conventions.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 58 9/12/2017 7:33:45 AM

 Basics of Power Systems  |  59 

Table 2.5  Power Balance for Circuit-3.

Element Generator convention Load convention

P (W) Q (VAR) PL (W) QL (VAR)

Re[ VI s* ] Im[ VI s* ] Re[ V ( − I s )* ] Im[ V ( − I s )* ]

Source
= 0.6154 =0.0769 = −0.6154 = −0.0769
Inductor L
Re[ V1 ( − I s )* ] Im[ V1 ( − I s )* ] Re[ V1 I s* ] Im[ V1 I s* ]

=0 = −0.3846 =0 = 0.3846

Re[ VC ( − I C )* ] Im[ VC ( − I C )* ] Re[ VC I C* ] Im[ VC I C* ]

Capacitor C
=0 = 0.3077 =0 =−0.3077
Resistor R
Re[ VR ( − I R )* ] Im[ VR ( − I R )* ] *
Re[ VR I R ]
*
Im[ VR I R ]
= −0.6154 =0 = 0.6154 =0

Total
∑ P=0 ∑ Q=0 ∑ PL = 0 ∑ QL = 0
Therefore, having chosen a current direction, the complex power in volt-amperes (VA) is computed as

S = Ve I e* (2.71)

= Ve ∠fv ( I e ∠fi )*

= Ve ∠fv I e ∠ − fi

P + jQ = (Ve I e )∠(fv − fi )

= (Ve I e )cos(fv − fi ) + j (Ve I e )sin(fv − fi ) (2.72)

The polarities of P/PL and Q/QL follow from Tables 2.1 and 2.2. It should be noted that an element in
the circuit can be a source unit as well.
V
Furthermore, note that if e is equal to impedance, Z =| Z | ∠f , then from (2.71) we can write
Ie

S = Ve I e*

= I e × Z × I e*

= I e2 × Z (2.73)

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 59 9/12/2017 7:34:03 AM

 60  |  Power System Analysis

Similarly, we can write

S = Ve I e*
*
V 
= Ve ×  e 
Z

V2 
=  e  ∠f (2.74)
 | Z |

2.4.6.2 Power Balance in Two Interconnected Sources

Consider a single-phase power system where two sources are interconnected by a line as shown in
Figure 2.23. The system frequency is 50 Hz. The line impedance is Zl = 0.1 + j 0.5 Ω . In the example,
each source represents a regional power system with its own generation and load. An initial value for
variables is chosen so that there exists a real power flow from one region to the another. At the operat-
ing point, V1 = 0.7071∠ − 25o V and V2 = 0.78∠0o V.

2
1 I

+ a xL = 0.5 a +
1 2
V1 V2
− −

Figure 2.23  Interconnection of Two Sources by a Line.

In order to analyse the system, I is chosen arbitrarily in the direction shown in the circuit. Now,
by applying KVL to the loop we get,

V1 − I × Zl − V2 = 0 (2.75)
Rewriting the above expression, we can determine current as

(V1 − V2 )
I =
Zl

(0.7071∠ − 25o − 0.78∠0o )

=
(0.1 + j 0.5)

= 0.6465∠166.34o A (2.76)

Using the load convention, the power associated with each element is determined and tabulated in
Table 2.6. Table 2.6 also shows the real and reactive power balance in the system. It can be seen that
real power flows from source-2 to source-1.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 60 9/12/2017 7:34:33 AM

 Basics of Power Systems  |  61 

Table 2.6  Load Convention-based Power Calculation for Two-source Interconnection.

Element SL PL (W) QL (VAR)

Source-1 V1 ( − I )* 0.4482 –0.0899

R 0.0418 0
I 2R

xL I 2 ( jx L ) 0 0.2090

Source-2 –0.49 –0.1191

V2 I *

Total ∑ SL = 0 ∑ PL = 0 ∑ QL = 0

2.4.6.3  Synchronous Operation of Two Interconnected Sources

It should be noted that since real-power generation involves interaction with mechanical systems
(such as rotor masses), any imbalance in real power balance triggers speed variation and hence,
frequency excursion. On the other hand, reactive power generation/consumption is associated with
electrical circuits of the system. Therefore, any deviation in the reactive power balance causes volt-
age variation. As long as the said real and reactive power balances are maintained, the system con-
tinues to operate at 50 Hz with a constant phase angle difference of 25° between the phasors V1 and
V2 )and constant bus voltage magnitudes (Figure 2.24). This is referred to as synchronous operation
of the two sources. Even if the real power balance is temporarily disturbed, it can affect the oper-
ating frequency of the post-disturbance system. If the disturbance persists for a longer time, it can

90 1
1 120 60
v2(t)
0.8
0.6 150 0.5 30
0.4
0.2 −i(t) I
V and A

V2
0 180 0
v1(t)
−I
−0.2
−0.4 V1
210 330
−0.6
−0.8
240 300
−1 270
0 0.0020.0040.0060.008 0.01 0.0120.0140.0160.018 0.02
Time (s)

Figure 2.24  Interconnection of Two Sources: Time–domain Plots and Phasor Diagram.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 61 9/12/2017 7:34:39 AM

 62  |  Power System Analysis

cause a large excursion in the frequency and the phase angle difference. This can be imagined as if
the phasor V1 is drifting from the phasor V2 continuously, leading to a large angle separation. This
condition is generally known as loss-of-synchronism. Under this condition, the current and the real
power exhibit huge oscillations. The only way to protect individual sources is to isolate them from
one another.

2.4.6.4  Real and Reactive Powers in Two Interconnected Sources

In Figure 2.23, the negative terminals of the sources are generally grounded so that the positive
terminal of the sources will be at higher potential with respect to the ground. Further, the polarity
shown for the sources in Figure 2.23 denotes that the instantaneous waveform for the sources are
given by

v1 (t ) = 0.7071 2sin(wt − 25o ) and v2 (t ) = 0.78 2sin(wt )V

The current waveform is given by

i (t ) = 0.6465 2sin(wt + 166.34o )A

It is also required to note − I = 0.6465∠ − 13.66o A . These plots and phasors are depicted in Figure
2.24. Using the −I (instead of I ), it is easy to understand the nature of the sources. For source-1,
employing load convention, −I enters its positive terminal. This implies that the source element-1,
as a load, draws a current which leads its voltage phasor V1 . Therefore, it can be said that source-1
absorbs real power (as its Re[ S L ] is positive) and generates reactive power, behaving like a capacitor
(as its Im[ S L ] is negative). Similarly, for source-2, with the generator convention, it can be seen that
−I leaves its positive terminal. This shows that the source element-2, as a generator, delivers a cur-
rent which lags its voltage phasor V2 . Hence, it can be inferred that source-2 generates real power (as
its Re[ S = V2 conj( − I ) ] is positive or Re[ S L ] is negative) and supplies an inductive load. Since its
Im[ S L ] is negative this source also generates reactive power behaving like a capacitor. Therefore, the
reactive power generated by two sources together supply the reactive power losses. With regard to the
system losses, we can note the following:

• While source-2 is delivering a real power of 0.49 W, it meets the real power loss component of the
system, equal to 0.0418 W such that source-1 receives a real power equal to 0.4482 W.
• The reactive power requirement of x L is shared by two sources such that

Qloss = Q1 + Q2 = 0.0899 + 0.1191 = 0.2090 VAR

In addition, the following observations are made:

1. When V1 is set to 0.78∠ − 25° V and V2 = 0.7071∠0° V, the current I = 0.6465∠ − 168.72° A.
Using the load convention, the power associated with each element is determined and tabulated
in Table 2.7.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 62 9/12/2017 7:34:49 AM

 Basics of Power Systems  |  63 

Table 2.7 Load Convention-based Power Calculation for Two Source Interconnection-bus Voltage
Magnitude Modified.

Element SL PL (W) QL (VAR)

Source-1 V1 ( − I )* 0.4065 –0.2984

R 0.0418 0
I 2R

xL I 2 ( jx L ) 0 0.2090

Source-2 –0.4483 0.0894

V2 I *

Total
∑ SL = 0 ∑ PL = 0 ∑ QL = 0
In this case, source-1 alone generates reactive power, and source-2 and line inductor together
consume the generated reactive power. However, source-2 continues to supply the real power
loss in the system. Therefore, unlike in the reactive power case, where reactive power associated
with source elements can be altered by simply setting the bus voltage magnitudes, the real power
loss sharing cannot be modified in a conventional power-flow analysis. This happens because the
supply of real power loss in a circuit gets decided by the angle difference between the bus voltages
instead of the individual bus angles.
2. When V2 is set so that V2 = V1 , i.e., V1 = 0.7071∠ − 25° V and V2 = 0.7071∠0° V with R = 0,
I = 0.6122∠167.5° A. For this condition, the reactive power shared by the sources are such that

Qloss = Q1 + Q2 = 0.0937 + 0.0937 = 0.1874 = I 2 X L VAR

Qloss
Note that Q1 = Q2 = . For this lossless system, let V1 = V1∠d and V2 = V2 ∠0 . Writing an
2
expression for the complex power associated with source-1 employing generator convention, we get,

S1 = V1 I * (2.77)
where,
(V1 − V2 )
I = (2.78)
jX L
Using the above expression in (2.77) and simplifying, we get,
*
 (V ∠d − V2 ) 
S1 = V1∠d  1 
 jX L
 (V ∠ − d − V2 ) 
= V1∠d  1 
 − jX L

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 63 9/12/2017 7:34:57 AM

 64  |  Power System Analysis

V12 VV
P1 + jQ1 = j − j 1 2 (cosd + j sin d ) (2.79)
XL XL

Equating real and imaginary parts on both sides, we have,

V V 
P1 =  1 2  sind (2.80)
 XL 
and
(V12 − V1V2 cos d )
Q1 = (2.81)
XL

Using the numerical values given, we get P1 and Q1 as

V V  0.70712
P1 =  1 2  sin d = sin( −25o ) = −0.4226 W
 L
X 0 . 5

(V12 − V1V2 cos d ) (0.70712 − 0.70712 cos( −25o ))

Q1 = = = 0.0937 VAR
R
XL 0.5
The value of P1 denotes the average value of the instantaneous power, p(t ) = v1 (t ) × i (t ) wave form
pertaining to phase-a. In (2.80) the angle d decides the real power flow between the sources for a
given source voltage magnitude and the line reactance. Hence, this angle is referred to as power
angle. As stated earlier, under synchronous operating condition of the two sources, this angle remains
constant. Any deviation from this constant value denotes a perturbation of the synchronous operation.
The maximum real power the line can transfer under sinusoidal steady-state condition corresponds to
d = 90°.

2.5  POWER IN THREE-PHASE SYSTEMS

In systems where large-scale power generation, transmission, and utilisations are involved, apparatus
are generally built for three-phase electrical circuits in a single unit. This is done to exploit the tech-
nical and economical advantages of the three-phase systems, some of which are described as follows:
1. Three-phase systems offer better utilisation of active materials related to magnetic and electrical
circuits, thus leading to higher power density. This enables three-phase units such as synchronous
generators, transformers, and inductions motors to be built for larger capacity in a single unit.
This, in turn, improves efficiency and makes the unit cheaper.
2. The three-phase electrical power circuits can be connected either in starr (Y) or delta ∆. The
Y-winding connection has the following benefits:
1
(a) For a given line-to-line voltage, the phase voltage is only times the line-to-line value.
3
1
Therefore, each phase needs to be provided with only = 58% of the number of turns
3

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 64 9/12/2017 7:35:00 AM

 Basics of Power Systems  |  65 

required for a D connection. This reduces the insulation cost as each phase is now required to
1
withstand of the line-to-line voltage, with respect to ground.
3
(b) Further, for a Y connection, one end of the winding can be grounded to facilitate a neutral
connection. Therefore, it offers a three-phase, four-wire system. This provides two voltage
levels, i.e., line-to-line voltage and line-neutral voltage (or phase-voltage) to deliver power to
the customers.
(c) With Y connection, third harmonic and all its multiples are eliminated in the line-to-line volt-
age. This is applicable for a symmetrical machine where three-phase voltages are perfectly
balanced. Consider a phase voltage with third-harmonic of amplitude Vm3 as follows:

van = Vm1sin(w t) + Vm3sin(3w t)

 2p    2p  
vbn = Vm1sin  w t −  + Vm3sin  3  w t − 
 3   3  

Now obtain the line-to-line voltage, vab as

vab = van − vbn

  2p  
= Vm1 sin(w t) − sin  w t −  + Vm3 [sin(3w t) − sin(3w t − 2p ) ]
  3  

  2p  
= Vm1 sin(w t) − sin  w t − 
  3  

 p
= 3Vm1sin  w t + 
 6

Due to this observation, while generator windings are arranged in stator slots, care is taken to
reduce/eliminate fourth- and fifth-order harmonics rather than triplen harmonics.
3. If synchronous generators or induction motors are built for three-phased units instead of sin-
gle-phase, the rotor systems experience a net torque which is constant and non-pulsating. Hence,
it leads to less vibration and noise. While in a single-phase unit, the torque produced not only pos-
sesses an average value, but also has a double frequency component. The magnitude of the torque
in a three-phase unit is equal to three times the average value that established in a single-phase
unit. Therefore, three-phase units will be smaller in size in comparison to a single-phase unit for
a given rating.

2.5.1  Interconnection of Two Sources–Three-phase Systems

Consider a system where two three-phase synchronous generators are interconnected through a
3-phase line as shown in Figure 2.25.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 65 9/12/2017 7:35:03 AM

 66  |  Power System Analysis

Ia 2
a1 1

+ R = 0.1 Ω xL =0.5Ω a2 +
V 1a V 2a
− f −
a1
f a2
f c1 f b2
f b1 f c2
V1b V2b
c1 + +
V1c c2

+
+

V2c b2
b1
Ib
1 2

R = 0.1 Ω xL =0.5 Ω
2
1

R = 0.1 Ω xL =0.5 Ω
Ic

Figure 2.25  Interconnection of Two Three-phase Synchronous Generators.

It can be seen that this system is obtained simply by extending the single-phase system shown
in Figure 2.23 for the other two phases. Here, phase-b and phase-c are added symmetrically rel-
ative to phase-a [4]. Therefore, each source is made up of three phases in a single unit, follow-
ing abc phase-sequence. This order of appearance of phases is generally referred to as positive
phase-sequence.

2.5.1.1  A Note on Phase-sequence of Three-phase Quantities

The order of appearance of phases is an important specification for a three-phase system.
If this order is different from abc, it is referred to as negative phase-sequence, denoted as acb.
From Figure 2.26, the phase-sequence can be easily identified by staying on the figure as a sta-
tionary observer and tracing the movement of the phasors while they revolve in the direction of
the increasing angle. For example, if an observer stays at point X, the observer sees phasors V1a ,
V1b , and V1c crossing the point X in the order a –b – c. Hence, the phase-sequence is abc. In a syn-
chronous generator, the phase-sequence of the generated voltage is decided by the placement of the
windings relative to one another and the direction of rotation of the rotor. For a given direction of
rotation of the rotor, with respect to the external circuit, the phase-sequence is changed simply by
interchanging any two terminals at the point connection to the load. A linear, three-phase and sym-
metric load draws the corresponding sequence currents. This phase-sequence decides the direction
of the revolving field [2] in an induction motor and hence, affects the direction of rotation of its rotor.
This direction of rotation is very critical for water pumping, blower, lifting applications, etc. This
phase-sequence is also important for interconnected operation of two AC sources. Even in power
electronic applications, the conduction of semi-conductor devices and their driver circuits require a
matched phase-sequence.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 66 9/12/2017 7:35:05 AM

 Basics of Power Systems  |  67 

2.5.1.2 Three-phase, Four-wire and Three-phase,

Three-wire Power Systems
For source-1, the phasor voltages are given by,

V1a = 0.7071∠ − 25° = V1a ∠fv V

V1b = 0.7071∠( −25° − 120°) = 0.7071∠ − 145°V

V1c = 0.7071∠( −25° + 120°) = 0.7071∠95°V (2.82)

Similarly, for source-2, the phasor voltages are given by,

V2a = 0.78∠0°V
V2b = 0.78∠ − 120°V
V2c = 0.78∠120°V (2.83)

Following the calculation steps given in (2.76), the individual phase currents can be obtained as

I a = 0.6465∠166.34°A
I b = 0.6465∠(166.34° − 120°)A

I c = 0.6465∠(166.34° + 120°)A (2.84)

or

− I a = 0.6465∠ − 13.66° = I ∠fi A

− I b = 0.6465∠( −13.66° − 120°) = 0.6465∠ − 133.66°A

− I c = 0.6465∠( −13.66° + 120°) = 0.6465∠106.34°A (2.85)

These phasors have been plotted in Figure 2.26 for each source separately.
In Figure 2.25, two terminals of each phase related to source-1 are separately connected to the
corresponding terminals of source-2. Therefore, there are six wires running between the sources.
Note that for source-1, [a1, b1, c1] represent the start of the phase windings, whereas [ f a1 , f b1 , f c1 ]
represent the finish of the windings. Each set forms the corresponding terminals of the phase wind-
ings. It can be seen that for ease of connection, one set of corresponding points, say, finish, can be
tied together for each source, without any danger of circulating current. A similar arrangement can be
made with respect to source-2 as well. Now, instead of running three additional wires from ‘finish-1’
junction of source-1 to the ‘finish-2’ junction of source-2, one single wire can be connected between
them as shown in Figure 2.27. Therefore, it forms a three-phase four-wire system, where the 4th-wire
carries the resultant current I N given by

I N = I a + I b + I c (2.86)

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 67 9/12/2017 7:35:10 AM

 68  |  Power System Analysis

90 1 90 1
120 60 120 60
V1c
−Ic V2c −Ic
150 0.5 30 150 0.5 30

180 0 180 0
−Ia V2a
−Ia
V1b V1a
−Ib
210 −Ib 330 210 330
V2b

240 300 240 300

270 270
Source−1 Source−2

Figure 2.26  Phasor Diagram for Two Three-phase Synchronous Generators.

Ia 2
a1 1

+ R= 0.1Ω xL=0.5Ω a2 +
V 1a V 2a
− f a1 IN −
f a2
f c1 f b2
f b1 f c2
c1 V1b V2b
+ +
V1c
+

c2
+

b1 V2c b2

Ib
1 2

R= 0.1Ω xL=0.5Ω
2
1

R= 0.1Ω xL=0.5Ω
Ic

Figure 2.27  Interconnection of Two Three-phase Synchronous Generators with Four Wires.

If the line impedance is equal for all three phase-lines (with negligible impedance for the return
paths) and the source voltages are balanced as given by (2.82) and (2.83), the system is said to be
symmetric. In such a symmetric network, the phase currents are also balanced (see (2.85)). Now, the
current I N is given by

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 68 9/12/2017 7:35:15 AM

 Basics of Power Systems  |  69 

I N = Ia + Ib + Ic

= 0.6465∠166.34° + 0.6465∠(166.34° − 120°) + 0.6465∠(166.34° + 120°))

(
= 0.6465 e j166.34° + e j 46.34° + e j 286.34° )
= 0.6465 [( −0.9717 + j 0.2362) + (0.6904 + j 0.7234) + (0.2813 − j 0.9597) ]

IN = 0
(2.87)
This shows that in the fourth wire. there is no current and ‘finish-1’ and ‘finish-2’ junctions are at the
same potential relative to one another. Therefore, it can be concluded that even if this wire is not con-
nected as in Figure 2.28, the system continues to work adequately as a three-phase, three-wire system.

Ia 2
a1 1

+ R= 0.1Ω xL=0.5Ω a2 +
V 1a V 2a
− f a1
−
n n
f a2
f c1 f b2
f b1 f c2
c1
V1b V2b
c2
+ +
V1c
+
+

b1 V2c b2
Ib
1 2

R= 0.1Ω xL=0.5Ω
2
1

R= 0.1Ω xL=0.5Ω
Ic

Figure 2.28  Interconnection of Two Three-phase Synchronous Generators

as a 3-phase, 3-wire System.

Notes:
1. The junction points ‘finish-1’ and ‘finish-2’ do not have a fixed potential relative to ground.
Therefore, in order to establish a common reference or zero potential point for both the sources,
it is generally recommended to connect the junctions separately to the ground as shown in Figure
2.28. This provides a means to create a neutral wire, the fourth-wire, locally, without having it to
be drawn from the source generator unit unlike the phase-wires. This enables lying of only three-
phase wires called lines, while transmitting power from the generator location to the customer
points. At the customer location, the Y-connected transformer winding is used to generate the
neutral wire locally. The starr-point of the winding is grounded so that the phase-terminals are at
higher potential with respect to ground. Such a grounded starr-point facilitates easy detection and
protection of power apparatus and human life against abnormal condition on the system.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 69 9/12/2017 7:35:19 AM

 70  |  Power System Analysis

2. In Figure 2.28, the absence of the neutral wire also ensures that triplen currents are suppressed in
the system. This is because, as shown earlier, a triplen current (say, third order harmonic), is such
that its magnitude is equal and in-phase in all three phases. Hence, these currents should add to
three times the per-phase value in the neutral current. If the neutral wire does not exist, this triplen
current must be absent in the system. This is applicable to a balanced Y-connection. However, in
case of a balanced D-connection, if the triplen exists, it is restricted to the closed D-loop.

2.5.2  Analysis of Real and Reactive Powers in Three-phase Systems

For analysing the system shown in Figure 2.28, a single-phase equivalent of the network can be
obtained as depicted in Figure 2.23. From (2.82) and (2.85) the instantaneous power associated with
source-1 can be determined as follows:
The time–domain waveforms for voltage and current, for phase-a, can be written as

v1a (t ) = Vm sin(w t + fv )
ia (t ) = I m sin(w t + fi )

where Vm = 0.7071 2 V, fv = −25° , I m = 0.6465 2 A and fi = −13.66° . Now, the instantaneous

power, pa(t), is given by

pa (t ) = v1a (t )ia (t ) = Vm sin(w t + fv ) × I m sin(w t + fi ) (2.88)

1
Using the trigonometric relationship, sin A sin B = cos ( A − B) − cos ( A + B)  , the above expres-
sion can be simplified as 2

Vm I m
pa (t ) =
2
[cos(fv − fi ) − cos(2w t + fv + fi )] (2.89)

Similarly, the time–domain waveforms for voltage and current, for phase-b and phase-c, are ­constructed
as

v1b (t ) = Vm sin(w t + fv − 120°)

ib (t ) = I m sin(w t + fi − 120°)

and

v1c (t ) = Vm sin(w t + fv + 120°)

ic (t ) = I m sin(w t + fi + 120°)

Now, the instantaneous powers, pb(t) and pc(t) are given by

Vm I m
pb (t ) = [cos(f v − fi ) − cos(2(w t − 120°) + fv + fi )] (2.90)
2

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 70 9/12/2017 7:35:24 AM

 Basics of Power Systems  |  71 

and

Vm I m
pc (t ) =
2
[cos(fv − fi ) − cos(2(w t + 120°) + fv + fi )] (2.91)

The phase-voltage plots for three-phases are shown in Figure 2.29. The time–domain plot of −I is
also depicted for three phases. An arrow indicated in each of the first three subplots denotes the time
instant at which each phase-voltages reaches its positive peak. The order in which the phase-voltages
attain the peak represents the abc phase-sequence. The last subplot shows that the instantaneous power
associated with each phase is pulsating at double the fundamental frequency. From (2.89), (2.90)

1
v(t)
v1a, (−ia)

0 −i(t)
p(t)
−1
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Time (s )

1
v1b, (−ib)

−1
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Time (s )

1
v1c, (−ic)

−1
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Time (s )
1.5
pa, pb, pc, pT

pb pa
1 pc PT=pa+pb+pc

0.5

0
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Time (s )

Figure 2.29  Three-phase Voltage, Currents, and Instantaneous Powers for Source-1:
Two Source System.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 71 9/12/2017 7:35:25 AM

 72  |  Power System Analysis

and (2.91), we can see that the each expression possesses an average component, referred to as real
power equal to

Vm I m
P= cos(fv − fi ) = 0.7071 × 0.6465 cos( − 25° + 13.66°) = 0.4482 W per phase.
2

It can be verified that if pa (t ) + pb (t ) + pc (t ) is computed, from (2.89), (2.90), and (2.91), we have,

V I 
pa (t ) + pb (t ) + pc (t ) = 3  m m cos(fv − fi ) −
 2 
Vm I m
− [cos(g ) + cos(g − 240°) + cos(g + 240°)]
2
V I 
= 3  m m cos(fv − fi ) = P3f (2.92)
 2 

where g = (2w t + fv + fi ) and it can be shown that

[cos(g ) + cos(g − 240°) + cos(g + 240°) ] = 0

From (2.92), it can be concluded that the sum of the instantaneous power for three-phases results in
a constant value equal to three times the real power, P, in each phase (Figure 2.29). Furthermore, it
should be noted that similar to real power, P, the reactive power, Q, continues to exist in each phase.
The presence of this power is evident in each phase due to a current flow which maintains an identi-
cal non-zero phase-angle with respect to their phase-voltages. Hence, to account for the presence of
reactive power, a three-phase reactive power is computed in a similar manner to the three-phase real
power. Therefore,

V I 
Q3f = 3  m m sin(fv − fi ) (2.93)
 2 

Therefore, a complex three-phase power is computed from (2.72) as
P3f + jQ3f = 3(Ve I e )sin(fv − fi ) + j 3(Ve I e ) sin(fv − fi )

S3f = 3Ve I e* (2.94)

It should be noted that the generator and load convention shown in Figure 2.22 is applicable to a
three-phase system as well. Since the network is symmetric, instead of computing the power associ-
ated with each and every element pertaining to each of the phases, the complex power calculations are
carried out for the single-phase equivalent circuit (which is identical to Figure 2.23) and the values are
simply multiplied by 3 as per (2.94). In this expression, Ve and I e denote the phasors pertaining to
phase-a. Based on this understanding, the power balance is obtained for the three-phase circuit shown
in Figure 2.28 on 3-phase basis and are tabulated in Table 2.8.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 72 9/12/2017 7:35:30 AM

 Basics of Power Systems  |  73 

Table 2.8  Load Convention-based Power Calculation for Two Source

Interconnection-three-phase System.

Element SL3f PL3f (W) QL3f (VAR)

Source-1 1.3446 –0.2697

3V1a ( − I a )*

R 3 I a2 R 0.1254 0

xL 3 I a2 ( jx L ) 0 0.6270

Source-2 3V2a I a* –1.4700 –0.3573

Total ∑S L3f = 0 ∑PL3f = 0 ∑QL3f = 0

2.5.3 Interpretation of Synchronous Operation

in Terms of Power Angle
For a lossless balanced three-phase system, the expressions (2.80) and (2.81) are modified as

V V 
P3f = 3  1 2  sind (2.95)
 XL 

and

(V12 − V1V2 cosd )

Q3f = 3 (2.96)
XL

For a balanced system, P3f denotes the numerical value of the total three-phase power PT as shown in
Figure 2.29. This, in turn, represents the net unidirectional constant torque acting on the rotor system
of the synchronous generator since the real power delivered by a generator is obtained through an
energy conversion process. In this process, the sum of the real power delivered and the armature cop-
per losses is equal to the air-gap power developed, as given by the following relationship:

w r × Teg = 3 × P1 + 3 × copper loss in each phase

= P3f + 3 × copper loss in each phase (2.97)

where wr is the rotor mechanical speed in rad/s and Teg is the electro-magnetic torque acting on the
rotor in N-m.
Under steady-state, the air-gap power is equal to the rotor mechanical power input and this equilib-
rium maintains a constant speed of the rotor and hence a constant frequency. Furthermore, it should
be noted that as long as a sinusiodal steady-state condition prevails, the power angle d remains fixed

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 73 9/12/2017 7:35:34 AM

 74  |  Power System Analysis

and a synchronous operation is maintained. From this, it is clear that if this condition/equilibrium is
disturbed, it leads to a deviation in d, rotor speed and air-gap power from a constant value perturbing
the synchronous operation.

2.5.4  C
 omplex Power Calculations in Three-phase
Balanced Systems
It is always straightforward to use (2.94) for computing the complex power in three-phase systems
where phase-quantities are employed. However, it is customary to specify quantities in terms of line-
to-line values or line values for any power apparatus. In such cases, depending on the winding connec-
tion used for the equipment, i.e., whether, the winding is connected in Y or D (see Figure 2.30), (2.94)
is rewritten in terms of the line values as follows [3]:

I L = 3 Iph
Ie I L = I ph
a a I ph
+
+
Vph= V L
Vph Ve VL = 3 Vph

c +
+ b
c + b

Y−connected elements −connected elements

Figure 2.30  Y and D-Connected 3-phase Elements.

1. For Y connection: The line-to-line voltage magnitude, VL is related to phase-voltage as VL = 3V ph .

Note that Vph is the magnitude of each phase-voltage and is identical to phase-a voltage or voltage
across the element, Ve. Furthermore, the line current magnitude IL is the same as the per phase
current Iph, which is identical to the element current, Ie. With these notations, (2.94) is written as

*
S3f = 3V ph I ph

= 3 3V ph ∠fv ( I ph ∠fi )*

Noting that VL = 3V ph and I L = I ph , we have,

S3f = 3VL ∠fv ( I L ∠fi )* (2.98)

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 74 9/12/2017 7:35:37 AM

 Basics of Power Systems  |  75 

2. For D connection: The line-to-line voltage magnitude, VL is related to phase-voltage as VL = Vph,

where Vph represents the magnitude of voltage across the element, Ve. Further, the line current
magnitude IL is related to the per phase current, Iph as I L = 3 I ph . Here, Iph is identical to the
element current, Ie. With these notations, (2.94) is written as

*
S3f = 3V ph I ph

= 3 3V ph ∠f v ( I ph ∠fi )*

= 3V ph ∠f v ( 3 I ph ∠fi )*

Noting that VL = V ph and I L = 3 I ph , we have,

S3f = 3VL ∠f v ( I L ∠fi )* (2.99)

In (2.98) and (2.99), only the magnitudes of voltage and current are changed to line values,
whereas the angles continue to be pertaining to the per phase-values. Furthermore, note that the
expressions for power are identical both for Y- and D-connections, implying that with respect to
line terminals, both the connections behave identically.

2.5.5  Single-Line Diagrams of Three-phase Systems

Many a time, it is not convenient to draw the network diagram of a three-phase system showing all
details about each phase. In such cases, it is adequate to symbolically represent each apparatus based
on a set of standardised symbols and show their interconnection through line diagrams [10]. These
diagrams are generally referred to as single-line diagrams or on-line diagrams [11, 12]. For the system
in Figure 2.28, a single-line diagram can be drawn as in Figure 2.31.

1 2

S2
S1

Figure 2.31  Single-line Diagram for the Three-phase, Three-wire Two Source System.

In the diagram, S1 and S2 denote synchronous machines, each built for symmetric three-phases in a
single unit. Each bus represents three individual phase terminals and a line depicts three-phase wires
connected between buses 1 and 2. These lines are assumed to be symmetric, and in practice, it is
achieved by transposing the lines all along the length between the buses 1 and 2 [13]. These appara-
tus, in turn, make the system balanced. Therefore, parameters of the equipment are specified only for
phase-a and are called per-phase data. In addition, the rating details of each unit are given in term of
line-to-line voltage, and three-phase total power. The nominal frequency of operation is also specified

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 75 9/12/2017 7:35:39 AM

 76  |  Power System Analysis

in Hz. For a balanced power system analysis, the parameters pertaining to positive phase-sequence
are specified, assuming the elements to be linear. Therefore, in such analysis, only positive phase-­
sequence voltages and currents are computed. If a system with unbalanced terminal conditions are to
be analysed, additional details pertaining to negative- and zero-sequence components are desired [14].
These representations are generally applicable to fundamental frequency steady-state and dynamic
analyses. In studies where high-frequency electromagnetic transient analysis is carried out, additional
system details may be necessary [15].

2.6  PER-UNIT REPRESENTATION

Power system network involves interconnection of a large number of three-phase power apparatus
such as synchronous generators, transformers, lines, and loads, which are built for varied voltage and
current ratings. While analysing such systems, the following difficulties are generally encountered:
1. The voltage and current values associated with some apparatus vary over a wide range in different
parts of the network. For example, a synchronous generator’s typical voltage ratings are 11 kV to
22 kV, and power ratings lie in the range of 500 MVA to 1000 MVA. Similarly, power transformer’s
ratings are 1000 MVA, 22/400 kV, and may be 1000 MVA, 22/750 kV. Hence, for a transformer on
low-voltage (LV) side, currents are very large compared to those on high voltage (HV) side.
2. Synchronous generators are generally connected in Y to exploit the benefits of Y connection.
When three-phase transformers are employed for transmission application, the LV-side is con-
nected in D and HV-winding is connected in Y. Similarly, when transformers are used for distri-
bution application (to step-down the voltage), the HV-side is connected in D and LV-winding is
connected in Y. These connections permit easy isolation of line sections with respect to ground-
ing-related disturbances and facilitate creation of a neutral wire locally. It is found that low volt-
age induction motors are connected in D, whereas high voltage induction motors are connected
in Y. In such cases, one should use the 3 factor carefully depending on the winding connection.
Even the per-phase parameter specifications may vary depending on the winding connection.
3. When parameters are specified at actuals, they also vary over a wide range, depending on the
power rating of the equipment. This makes the choice more difficult and erroneous when typical
values are to be assumed for parameters in the absence of their availability.
4. For a transformer, depending on the winding side to which the parameters are referred, the param-
eters may have low or high value. This makes calculations more involved.

2.6.1  Procedure Employed for Per-unit Representation

In the per-unit representation the actual values of the variables (such as voltage, current, power, speed,
torque, flux, and so on) and parameters (such as resistance, inductance, capacitance, inertia, and so
on) are divided by the corresponding base values and are made dimensionless. While choosing such
base values, care is taken to see that the quantities are scaled so that they are convenient to handle
computationally. In this per-unit representation, the steps employed are as follows:
1. While adopting the base values, the total three-phase power rating of a Y-connected equipment
is chosen as the power base, SB (in volt-amperes (VA)) and the line-to-line voltage rating of the
device is chosen as the voltage base, VB (in V). In addition, the rated frequency, fB (in Hz) of the
unit (or the entire system) is used as the frequency base. This choice scales the actual variables

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 76 9/12/2017 7:35:40 AM

 Basics of Power Systems  |  77 

such that their per-unit values remain close to 1.0, since the units are typically operated close to
their full-load values. This removes the difficulty in handling variables which vary widely when
specified at actuals. When real power (in W) or reactive power (in VAR) is divided by the base
power, SB, it gives the per-unit value of real or reactive power.
  Using the power and voltage bases, other base values are evaluated. For example, the current
base and impedance base are obtained as follows:
  From (2.98), the magnitude of the complex power, referred to as apparent power (in VA), is
obtained as

S3f = 3VL I L

= VL ( 3 I L )

where VL represents the line-to-line voltage and IL represents the line current which is also equal
to phase current, Iph, in a Y-connected winding.
  Now, denoting the base current, IB as 3 × I L , and replacing VL by VB, we get the power base as,

S B = VB × I B (2.100)

Using the above expression the current base can be determined as
SB
IB =
VB (2.101)

Note that IB accounts for the factor 3 . Therefore, while computing the actual current from a
per-unit quantity, it results in a value equal to 3 times the phase current (or line current). An
advantage of this scaling is that if losses are computed in per-unit using current-square, then it
represents the total three-phase losses.
  The impedance base (in Ω) for the Y-connected winding can be obtained using VB and IB as
VB
ZB =
I B (2.102)

Note that the 3 factor accounted in IB converts the VB into per-phase voltage base. When per-
phase resistance in Ω or reactance in Ω (for a Y-equivalent) is divided by the base impedance it
provides the per-unit value of resistance or reactance.
  If IB in (2.102) is substituted for from (2.101), we get, an alternate expression for the base
impedance as

VB2
ZB =
S B (2.103)

2. In order to avoid the dependency on the three-phase winding connection, before calculating the
parameter values in per unit, it is ensured that all parameters pertain to the per-phase Y-equivalent.
This implies that even if per-phase parameters are given for the D-connection, they are converted
Z
into Y-equivalent by employing D to Y conversion relationships ZY = ∆ [3].
3

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 77 9/12/2017 7:35:43 AM

 78  |  Power System Analysis

3. The process of conversion to per-unit quantities removes the usage of 3 or 3 -factor in the
calculations. For a Y-connected 3-phase apparatus, from (2.98), the complex power is given by,
S3f = 3VL ∠fv ( I L ∠fi )*

Now, to express it in per-unit, we divide the above expression by SB using (2.100) to get,
*
S3f VL  3I L 
= ∠fv  ∠fi 
SB VB  IB 

S3 pu = V pu ∠fv ( I pu ∠fi )*

( )*
= V pu I pu (2.104)

Notes: For D-connected equipment, before performing per-unit conversion, an equivalent

Y-connection is obtained so that IL = Iph.
From (2.104), it can be seen that an expression for complex power in per unit appears as if it
written for a single-phase apparatus (see (2.71)). This implies that when per-unit-based analyses
are carried out, all three-phase-related quantities are identical to per-phase quantities.
4. In order to show the influence of parameter choice on the performance calculations, consider two
three-phase Y-connected generators: G1 with ratings 1 kVA, 400 V and G2 with specifications 10
MVA, 11 kV. Let the per-unit resistance, Rpu of G1 is 0.1 per-unit and that of G2 is 0.05 per-unit
(pu) on their machine ratings. The related calculations are listed in Table 2.9.

Table 2.9  Performance Calculation Related to Resistance of the Generators 1 and 2.

Observations G1: 1 kVA (SB), 400 V (VB) G2: 10 MVA (SB), 11 kV (VB)
Current per phase (IB) 2.5 A 909.09 A
ZB 160 Ω 12.1 Ω
R = Rpu × ZB 16 Ω 0.605 Ω
Total Ploss (pu) 0.1 pu 0.05 pu

Total Ploss (W) 0.1 × SB 0.05 × SB

= 100 W = 0.5 MW
Total Ploss (W) 16 × 2.52 0.605 × 909.092
2
= R × IB = 100 W = 0.5 MW
Voltage drop (line-to-line) 0.1 × 400 0.05 × 11 kV
= Rpu × VB 40 V 550 V
= R × IB
Voltage drop (per-phase) 0.1 pu 0.05 pu
VB
= R pu × 23.094 V 317.54 V
3

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 78 9/12/2017 7:35:45 AM

 Basics of Power Systems  |  79 

From Table 2.9, it is clear that Rpu not only represents the total three-phase power loss, but also
denotes the voltage drop at the rated conditions. Further, it should be noted that the numerical
values of line-to-line voltage and per-phase voltage are identical when expressed in per-unit. The
per-unit power loss is relatively large in low capacity units than those in high capacity apparatus.
Hence, specifying the corresponding parameters in per-unit is more suitable than expressing them
at actuals. For example, if the resistance of G2 is stated as 6 W, then it does not denote the losses
at full-load genuinely since it leads to a total 3-phase power loss of 909.092 × 6 = 4.9587 MW
which amounts to 0.4959 per-unit as against an acceptable range of 0.01 to 0.05 per-unit.
5. Per-unit representation of aggregate models of elements:
Consider two three-phase, Y-connected, balanced loads. Load-A: a motor rated for 50 Hz, 3 kV,
10 MVA, having a resistance of Rm(pu) = 0.06 per-unit, with ZB as the base impedance. Load-B
is made up of parallel connection of two motors of each rating as in Load-A. If it is required to
obtain the resistance of the equivalent motor whose rating is 50 Hz, 3 kV, 20 MVA, then it can
be seen that its numerical value will also be equal to 0.06 per-unit on the equivalent motor rat-
ing. This is because, with parallel connection of n number of motors, the effective resistance (in
R
ohms) becomes m( Ω) . Similarly, the base impedance for the equivalent motor becomes Z B .
n n
Therefore, we have,

 Rm( Ω) 
 n  R
m( Ω )
=
 ZB  ZB
 
n 

where Rm( Ω) = Rm( pu) × Z B .

6. When (2.95) and (2.96) are expressed in per-unit, we get

 3V1 3V2 
P3f  V VB 
Ppu = = B  sind
SB  XL 
 ZB 

or
 V1 puV2 pu 
Ppu =   sind (2.105)
 X Lpu 

and

 ( 3V1 )2 3V1 3V2 

 2
− cos d 
Q3f  VB V B VB 
Q pu = =
SB XL
ZB

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 79 9/12/2017 7:35:50 AM

 80  |  Power System Analysis

or
(V12pu − V1 puV2 pu cos d )
Q pu = (2.106)
X Lpu

Therefore, from (2.105) and (2.106) we can see that the per-unit representation uses per-unit
value of line-to-line voltages and per-phase reactance in per-unit, and the expressions are much
simpler.
7. In power systems, transformers are used to interface different units which operate at different
nominal voltages. In a way, this makes the per-unit analysis simple because for certain analysis,
one need not be given the actual values of quantities or parameters. For example, in power system
stability studies, the per-unit values are adequate to understand the ability of a system to maintain
synchronous operation. However, care has to be taken to represent transformers correctly in the
per-unit model. These details are discussed in the following section.

2.6.2  Per-unit Model of Transformers

In power system analysis, in most of the studies, the transformers are treated as linear elements,
neglecting excitation current. This implies that the no-load current and non-linearities due to satura-
tion, hysteresis, and eddy currents are neglected. In large-capacity transformers, even the resistances
are neglected, since they are very small in comparison to the reactances. In cases, where power-related
analysis is carried out (e.g. in power-flow and stability studies), phase shifts experienced by voltage
and current in D/Y transformers are not generally considered [12].
Consider a three-phase, 50 Hz transformer with nominal ratings given by VHV/VLV = 400/33 kV,
Y/D, 650 MVA. The per-phase reactance referred to LV-side is xLVΩ = 0.25 Ω. This reactance is the
Y-equivalent per-phase value on the LV-side. It is customary to represent the reactance in per-unit,
taking the transformer ratings themselves as the bases values. The observations are listed below:
332
1. On LV-side, VB = 33 kV and SB = 650 MVA. The base impedance is ZB(LV) = = 1.6754 Ω.
thus, 650

0.25
x LVpu = = 0.1492 pu
1.6754
2. On HV-side, the referred value of per-phase reactance (in ohms) is obtained as
2
V 
x HV Ω = x LV Ω  HV 
 VLV 
(2.107)
2
 400 
= 0.25  = 36.7309Ω
 33 
4002
Now, choosing VB as 400 kV and SB = 650 MVA. The base impedance is ZB(HV) = =
246.1538 Ω. Converting xHVΩ in per-unit, we get, 650

36.7309
x HVpu = = 0.1492 pu
246.1538

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 80 9/12/2017 7:35:53 AM

 Basics of Power Systems  |  81 

This clearly shows that the referred value of reactances (at actuals) when expressed in per-unit
using the the transformer ratings as the base values on the respective side, their numerical values
are identical. This enables a transformer to be modelled as a simple circuit element without any
winding representation. Note that VHV and VLV denote the nominal line-to-line voltage ratings
and SB denotes the power rating of the transformer.
In order to demonstrate the condition under which such a circuit model exists for a transformer, con-
sider the following condition:
Let VBp be the new voltage base and SBp be the new power base on the HV-side. Similarly, let VBs
be the new voltage base and SBs be the new power base on the LV-side. With these base values, the
per-unit value of the referred reactances are given by
x HV Ω
x HV ( pu) =
 VBp2 
  (2.108)
 S Bp 

and

x LV Ω
x LV ( pu) =
 VBs
2 
(2.109)
S 
 Bs 

To obtain the numerical value of xHV(pu) = xLV(pu), from (2.108) and (2.109), we have,

x HV Ω x
= LV Ω
 VBp2   VBs
2 
  S 
 S Bp   Bs 

or

2
 VBp   S Bs 
x HV Ω = x LV Ω    (2.110)
 VBs   S Bp 

Since (2.110) and (2.107) show the relationship between ohmic values of the reactances, they are
identical. Hence, we can state that,

 VBp   VHV 
 V  =  V  and S Bp = S Bs (2.111)
Bs LV

From the above expressions, we can conclude that to arrive at a simple circuit model, without any
winding representation for a transformer, consider the following:

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 81 9/12/2017 7:35:54 AM

 82  |  Power System Analysis

1. The new voltage bases should be chosen such that the ratio of the new voltage bases, VBp and VBs,
for primary and secondary sides, is equal to the ratio of the nominal voltage ratings, VRp and VRs
of the respective sides. That is,

 VBp   VRp 
 V  =  V  (2.112)
Bs Rs

2. The new power base, SBN on both primary and secondary sides should be identical.
Therefore, for a transformer with nominal voltage ratings, an equivalent circuit is obtained as shown
in Figure 2.32. Here, Zeq(s) denotes the per-phase Y-equivalent impedance of the transformer referred
to secondary-side in ohms.

Z eq(s)[ins ]
p s

V V
Rp Rs

Secondary
Primary

Figure 2.32  An Equivalent Circuit of a Transformer with Nominal Voltage Ratings.

As indicated, VBp and VBs are the new voltage bases chosen on primary and secondary sides, respec-
tively, such that the condition given in (2.110) is satisfied. Here, it is assumed that the nominal power
rating, SB, is selected as the power base. With the base selection, the per-unit model of the transformer
is determined as depicted in Figure 2.33.

Z
eq (pu)
p s

V V
Rp Rs
V V
Bp Bs

Figure 2.33  The Per-unit Equivalent Circuit of a Transformer with Nominal Voltage Ratings.

The equivalent circuit given in Figure 2.33 is further modified introducing turns-ratio as shown in
Figure 2.34.
For the three-phase transformer indicated earlier, the nominal ratings are 400/33 kV, Y/D, 650
MVA. The per-phase reactance referred to LV-side is xLVΩ = 0.25 Ω. Now, the new voltage bases are
chosen as VBp = 454.5455 kV and VBs = 37.5 kV. Using the transformer power rating itself as the powe
base, (SB = 650 MVA), we have,
x
Zeq( pu) = j LV Ω
Z Bs

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 82 9/12/2017 7:35:56 AM

 Basics of Power Systems  |  83 

Z
eq (pu)
p s

V
Rp
V
Bp
1
V
Rs
V
Bs

Figure 2.34  The Per-unit Equivalent Circuit of a Transformer with Nominal Turns-ratio.

37.52
With xLVΩ = 0.25 Ω, and Z Bs = = 2.1635 Ω , we obtain Zeq(pu) = j0.1156 pu. Furthermore,
note that, 650

 VRp 
V   400 
 Bp   
454.5455 
= = 1.0
 VRs   33 
 V   
Bs
37.5 

Therefore, using Figure 2.34, the final per-unit model of a transformer with nominal tap-ratio (on new
voltage bases), is given as shown in Figure 2.35.

j0.1156 pu
p s

1 1

Figure 2.35  The Final Per-unit Equivalent Circuit of a Transformer with Nominal Tap-ratio.

2.6.3 Example: A Sample Power System to Demonstrate

Per-Unit Calculations
Consider a power system where a three-phase generator G, is connected to a 400 kV line through a
step-up transformer, T1. At the receiving end, the line is connected to a three-phase load of SL = (350 +
j100) MVA. The single-line diagram of the system is shown in Figure 2.36.

Section−1 Section−3
Section−2

3 4 Load
1 2
G T T SL
1 2
Y Y Y

Figure 2.36  Single-line Diagram of a Sample Power System.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 83 9/12/2017 7:35:58 AM

 84  |  Power System Analysis

The rating of each units and their parameters (in per-unit) on the machine ratings themselves are
tabulated in Table 2.10.

Table 2.10  Per Unit Calculation on the Machine Ratings for the Sample Power System.

Element Generator - G Transformer - T1 Line Transformer - T2

Rating 500 MVA, 22 kV 22D/400 Y kV 500 kV, 1000 MVA 400Y/33D kV
600 MVA 650 MVA
xL = 52 W
Parameters xd = 1.7424 W xLV = 0.09 W xLV = 0.25 W
rL = 4.8 W
ZL = 52.2211 ∠ 84.73oW
ZB
222 222 5002 332
ZB = ZB = ZB = Z B( LV ) =
100 100 1000 650
= 0.968 W = 0.8067 W = 250 W = 1.6754 W
Zpu xd = 1.8 pu xt1 = 0.1116 pu ZL = 0.2089 ∠ 84.73o pu xt2 = 0.1492 pu

2.6.3.1  Source Voltage Calculation Using SBN = 100 MVA

As per (2.111), for the entire system, the same power base is chosen, i.e., SBN = 100 MVA. A voltage
base of VB1 = 22 kV, is chosen in section-1 (see Figure 2.36). Now, following the condition given in
(2.112), the voltage bases for the other sections are obtained. Hence, for section-2, VB2 = 400 kV, and
for section-3, VB3 = 33 kV. Therefore, a fixed power base is chosen for all sections, whereas the voltage
bases are ‘floated’ from one side to the another, across the windings of a transformer. On these chosen
bases, the per-unit impedances are obtained as tabulated in Table 2.11.

Table 2.11  Per Unit Impedances on SBN = 100 MVA and VB1 = 22 kV in Section-1.

Element Generator - G Transformer - T1 Line Transformer - T2

Rating 500 MVA, 22 kV 22D/400 Y kV 500 kV, 1000 MVA 400Y/33D kV
600 MVA 650 MVA
Parameters xd = 1.7424 W xLV = 0.09 W xL = 52 W xLV = 0.25 W
rL = 4.8 W
ZL = 52.2211 ∠ 84.73oW
ZB
222 222 4002 332
ZB = ZB = ZB = Z B( LV ) =
100 100 100 100
= 4.84 W = 4.84 W = 1600 W = 10.89 W
Zeq
Z pu = xd = 0.36 pu xt1 = 0.0186 pu ZL = 0.003 + j0.0325 pu xt2 = 0.023 pu
ZB

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 84 9/12/2017 11:11:58 AM

 Basics of Power Systems  |  85 

On the chosen power base, the per-unit value of the complex load is given by

S L = PL + jQL = 3.5 + j1.0 = 3.64 ∠15.945°pu

If the per-unit load voltage is V4 = 1.0 ∠0° , the load current in per-unit is obtained using (2.104) as,
*
 3.64 ∠15.945° 
IL =   = 3.64∠ − 15.945° pu (2.113)
 1.0 ∠0° 

The per-unit equivalent circuit of the sample power system is written as shown in Figure 2.37.

Notes:
• For transformers, the winding section (-see Figure 2.34) whose turn-ratio is 1:1 is not shown in
Figure 2.37 since the numerical values of quantities are identical on both sides of the winding.
• For any element, if the per-unit value of impedance is specified using the machine ratings as the
base values (see Table 2.10), one can obtain its per-unit value on the new base as follows:
N
 Z Bo 
Z pu = Z opu  N  (2.114)
 ZB 

2
VBo V2
where Z Bo = and Z BN = BN
S Bo S BN

For transformer T1, note that VB0 = VBN and the above expression simplifies to
N S 
Z pu = Z opu  BN  (2.115)
 S Bo 

1 2 3 4
I
L
+
x x r x x
d t1 L L t2

S
L
E

Figure 2.37  The Per-unit (Positive-sequence) Equivalent Circuit of the Sample Power System.

Now, to compute the source voltage E , we can apply KVL to the loop and write

E = V4 + ZTotal I L (2.116)

In the above expression, ZTotal = xd + xt1 + Z L + xt 2 = 0.003 + j 0.4341 = 0.4341∠89.6° pu. Using I L
given in (2.113), we get,

E = 1.0 ∠0° + 0.4341∠89.6° × 3.64 ∠ − 15.945°

= 2.0942∠46.39°pu (2.117)

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 85 9/12/2017 11:12:02 AM

 86  |  Power System Analysis

In Table 2.12, the base values for different sections are summarised.

Table 2.12  Base Values in Different Sections in the Sample Power System.

Section no. Section-1 Section-2 Section-3

SB 100 MVA 100 MVA 100 MVA
VBi 22 kV 400 kV 33 kV
ZBi 4.84 W 1600 W 10.89 W
IBi 4.5455 kA 0.25 kA 3.0303 kA

The following observations can be made with regard to the actual values of quantities:
1. The power base remains identical for all sections. The power supplied by the source can be com-
puted as

S g = E × I L* = 2.0942∠46.39° × 3.64 ∠15.945° (2.118)

Pg + jQg = 7.6229∠62.335° = 3.5393 + j 6.7514 pu (2.119)

If Sg is multiplied by SB = 100 MVA, we get Pg = 353.93 MW and Qg = 675.14 MVAR. This value
can be verified as

Pg = PL + I L2 rL = 3.5 + 3.642 × 0.003 = 3.5397 pu

Qg = QL + I L2 xTotal = 1.0 + 3.642 × 0.4341 = 6.7517 pu

Note that I L2 rL and I L2 xTotal denote the total three-phase copper losses and reactive power losses,
respectively, in per-unit.
2. If an ith section impedance (in per-unit) is multiplied by the base impedance of that section,
i.e., ZBi, then impedance at actual results. This is true for the bus voltages as well. For exam-
ple, the magnitude of the internal voltage of the generator at actuals, is given by E = 2.0942 ×
22
22 = 46.0724 kV (line-to-line) or 2.0942 × = 26.6 kV (per-phase).
3
3. If an ith section impedance (in per-unit) is multiplied by the base impedance of section-j, i.e., ZBj,
where, j ≠ i, it results in impedance of the ith section referred to the jth section at actuals. This is
valid for the bus voltages as well. For example, if xd = 0.36 pu is multiplied by ZB3 = 10.89 Ω,
the generator reactance referred to section-3 is obtained as 0.36 × 10.89 = 3.9204 Ω.
This is identical to referring 1.7424 Ω with voltage ratio  400  to section-2 and then to sec-
 22 
2
 33 
tion-3 with ratio  33  , so that it is equal to 1.7424   . If the internal voltage of the gen-
 400   22 
erator referred to section-3, is desired, we can obtain it as E = 2.0942 × 33 kV = 69.109 kV

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 86 9/12/2017 11:12:05 AM

 Basics of Power Systems  |  87 

(line-to-line). This implies that if the load is suddenly thrown off, the terminal voltage at bus 4
raises to 69.109 kV from 33 kV, assuming that the generator controllers are left unaltered.
4. If the line current IL is multiplied by IBi of i th section, it results in actual value of current in
amperes in that section. It should be noted that the actual phase current accounts 3 factor in
it. This implies that if the per-phase current (in Y-equivalent) is to be obtained, the value should
be divided by 3 . For example, if we consider section-2, the base current is IB2 = 0.25 kA.
910
Then, 3.64 × 0.25 is equal to 910 A, but the current in each phase is = 525.39 A. Using this,
3
the total line losses can be determined as Ploss = 3 × 525.392 × 4.8 = 3.974 MW. In per-unit, it
amounts to 0.03974 pu which is also equal to 3.642 × 0.003 pu.
5. Similarly, the per-phase currents in section-1 and section-3 are

3.64 × I B1 3.64 × 4.5455

= = 9552.62 A
3 3

and
3.64 × I B3 3.64 × 3.0303
= = 6368.34 A
3 3

Now, the reactive power losses at actuals can be obtained as

Qloss = 3 × 9552.622 × (1.7424 + 0.09) + 3 × 525.392 × 52 + 3 × 6368.342 × 0.25

= (501.634 + 43.061 + 30.417) = 575.112 MVA
 

In per-unit, QLoss, is equal to

Qloss = 3.642 × (0.36 + 0.0186) + 3.642 × 0.0325 + 3.642 × 0.023

= (5.0163 + 0.4306 + 0.3047) = 5.7516

The minor numerical differences are mainly due to round-off and truncation in the computation. This
example clearly shows that the synchronous generator has to be over excited to deliver the necessary
load of lagging nature.

2.6.3.2  Source Voltage Calculation Using SBN = 500 MVA

In this case, the power rating of the generator is chosen as the base power for the system. The voltage
base in different sections is same as that in the previous case. On these chosen bases, the per-unit
impedances are obtained as tabulated in Table 2.13.
On the chosen new power base, the per-unit value of the complex load is given by

S L = PL + jQL = 0.7 + j 0.2 = 0.728∠15.945°pu

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 87 9/12/2017 11:12:08 AM

 88  |  Power System Analysis

Table 2.13  Per Unit Impedances on SBN = 500 MVA and VB1 = 22 kV in Section-1.
Element Generator - G Transformer - T1 Line Transformer - T2
Rating 500 MVA, 22 kV 22D/400 Y kV 500 kV, 1000 MVA 400Y/33D kV
600 MVA 650 MVA
Parameters xd = 1.7424 W xLV = 0.09 W xL = 52 W xLV = 0.25 W
2eq rL = 4.8 W
ZL = 52.2211 ∠ 84.73oW
ZB
222 222 4002 332
ZB = ZB = ZB = Z B( LV ) =
500 500 500 500
= 0.968 W = 0.968 W = 320 W = 2.178 W
Zeq
Z pu = xd = 1.8 pu xt1 = 0.093 pu ZL = 0.015 + j0.1625 pu xt2 = 0.1148 pu
ZB

If the per-unit load voltage is V4 = 1.0 ∠0° , the load current in per-unit is obtained as,
I L = 0.728∠ − 15.945° pu. The per-unit equivalent circuit of the sample power system on the new
power base is identical to that shown in Figure 2.37 with the impedance values as give in Table 2.13.
If we compute the source voltage, E, with

ZTotal = 0.015 + j 2.1703 = 2.1704 ∠89.60° pu

we get, E = 2.0942∠46.39° pu, which is numerically equal to the value obtained in the previous case
since the voltage bases are the same. In these two cases, it is assumed that V4 = 1.0 ∠0° . With VB3 =
33 kV, the actual load voltage is also equal to 33 kV. Employing (2.104), we can obtain the complex
source power generated as,
S g = E × I L* = 1.5246∠62.335° = 0.7079 + j1.3503 pu

In terms of actual, this amounts to

S g = (0.7079 + j1.3503) × 500 = (353.95 + j 675.15) MVA

2.6.3.3  S
 ource Voltage Calculation Using SBN = 100 MVA
with a Different Voltage Base
In this case, for a power base of SBN = 100 MVA, a voltage base of VB1 = 25 kV is chosen in section-1.
With this, the voltage bases in other sections are obtained by floating the base as per the nominal volt-
age rating of the transformers:
 400 
1. Choose VB2 in section-2 as  × 25 = 454.5455 kV.
 22 

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 88 9/12/2017 11:12:12 AM

 Basics of Power Systems  |  89 

 33 
2. Choose VB3 in section-3 as  × 454.5455 = 37.5 kV.
 400 

On these chosen bases, the per-unit impedances are obtained as tabulated in Table 2.14.

Table 2.14  Per Unit Impedances on SBN = 100 MVA and VB1 = 25 kV in Section-1.

Element Generator - G Transformer - T1 Line Transformer - T2

Rating 500 MVA, 22 kV 22D/400 Y kV 500 kV, 1000 MVA 400Y/33D kV
600 MVA 650 MVA
Parameters xd = 1.7424 W xLV = 0.09 W xL = 52 W xLV = 0.25 W
2eq rL = 4.8 W
ZL = 52.2211 ∠ 84.73oW
ZB
252 252 454.54552 37.52
ZB = ZB = ZB = Z B( LV ) =
100 100 100 100
= 6.25 W = 6.25 W = 2066.1161 W = 14.0625 W
Zeq
Z pu = xd = 0.2788 pu xt1 = 0.0144 pu ZL = 0.0023 + j0.0252 pu xt2 = 0.0178 pu
ZB

Since the load voltage, V4 , is set to 1.0 per-unit on 33 kV base in the earlier cases, for VB3 = 37.5 kV, the
new load voltage is 33 = 0.88 pu. On the new voltage base, the load current I L = 4.13636∠ − 15.945°
37.5
pu. For this case, the new ZTotal = 0.3362∠89.608° pu. Using these details, the source voltage can be
computed as E = 1.84303∠46.39° pu. Similarly, the source power, Sg is obtained as

S g = 7.6234 ∠62.335° × 100 = (353.955 + j 675.187)MVA

The above example shows that for any choice of a suitable voltage base, at any section, as long as
the voltage bases are ‘floated’ as per the ratio of the nominal voltage ratings of the transformers, the
per-unit equivalent circuit remains simple without any (ideal) windings [16].

2.6.4 Per-unit Model of Transformers with Off-nominal

Voltage Ratings
Assume that transformer T1 in Figure 2.36 has tap-settings on the HV-side, i.e., s-side, so that the rated
N
voltage is set equal to VRs , which is different from the nominal value vrs. However, the LV-side, i.e.,
p-side voltage is maintained at the nominal value vrp. With these, the new rating of the transformer
N
is VRp / VRs , and SR. Note that even with the new voltage ratings the power rating is assumed to be
N
unchanged. For this transformer, let Zeq[ s ] be the new equivalent impedance (in ohms) referred to

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 89 9/12/2017 11:12:17 AM

 90  |  Power System Analysis

secondary, and I sN be the new current rating of the HV winding. For the same power rating, this
current is given by

VRs
I sN = I s N (2.120)
VRs
N
If Zeq[s] is the equivalent impedance of the transformer with the nominal rating, then, Zeq[ s ] can be
obtained as

2
N
 I 
Zeq[ s] = Zeq[ s]  Ns  (2.121)
 Is 

The above expression assumes that the net power-loss in the transformer is a constant. Also, note that
N
since VRP is not altered, the primary current rating remains at Ip. Therefore, Zeq[ p ] = Zeq[ p ] .
Using (2.120) in (2.121), we get,

2
N
 VRs
N
Zeq[ s] = Z eq[ s ]   (2.122)
 VRs 

N
Now, if Zeq[ s ] is expressed in per-unit using the new ratings of the transformer as the base, we obtain,

2
V N 
Zeq[ s]  Rs 
N  VRs 
Zeq[ s ]( pu ) =
(VRsN )
2

SR
Zeq[ s]
= = Zeq( pu)
 VRs
2 
(2.123)
S 
 R

Furthermore, note that

Zeq[ p]
Zeq[ p]( pu) = = Zeq( pu)
 VRp
2 
  (2.124)
 S R 

The above two expressions demonstrate that even for the transformer with the new ratings, i.e.,
off-nominal voltage ratings, the numerical value of the per-unit impedance remains identical to that
of the transformer with nominal voltage ratings, when the ratings themselves are chosen as the base

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 90 9/12/2017 11:12:20 AM

 Basics of Power Systems  |  91 

values. Therefore, the per-unit model of a transformer with off-nominal voltage ratings, is obtained
using the existing voltage bases, VBp and VBs, as described below:
N , and S . This shows that
1. The transformer ratings are: VRp / VRs R

 VRp   VBp 
 N ≠ 
 VRs   VBs 

2. Employing the representation given in Figure 2.34, connect the per-unit impedance of the trans-
former in series with an ideal transformer with an off-nominal tap-ratio ‘a’ as shown in Figure 2.38.

Z
eq (pu)
p s

V
Rp
V
Bp
a= 1
VN
Rs
V
Bs

Figure 2.38  The Per-unit Equivalent Circuit of a Transformer With Off-nominal Tap-ratio.

Note that Zeq(pu) is placed on the unit-turn-side in the figure. This signifies the numerical calculation
of Zeq(pu) using the following condition:
 VRp   VBp 
 V  =  V 
Rs Bs

Consider transformer T1: 22 kV ∆ (VRp)/400 kV Y (VRs), 600 MVA with Y-equivalent xLV per-phase =
0.09 Ω. The voltage bases are VBp = 22 kV and VBs = 400 kV. The transformer is provided with a tap
setting of −10% on the HV-side so that voltage decreases from the nominal. Hence, the new rating
N
of the transformer is 22 kV Δ(VRp)/400 × 0.9 = 360 kV Y(VRs ), 600 MVA. The per-unit model is
obtained as follows:
 VRp   VBp 
1. For the transformer with new ratings, note that  N  ≠  .
 VRs   VBs 
2. As per (2.123) and (2.124), the Zeq(pu) on 600 MVA power base is j0.1116 pu (see Table 2.10).

In Figure 2.38 the off-nominal turns ratio, ‘a’ is obtained as

 VRp 
V   22 
 Bp   
22
a= = = 1.1111
 VRs  
N 360 

 V   400  
 Bs 

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 91 9/12/2017 11:12:22 AM

 92  |  Power System Analysis

Therefore, the per-unit model of the transformer with off-nominal voltage rating on the chosen base,
is as shown in Figure 2.39.

Vp Zeq = j0.1116 pu
p s
V
V I s
s1 s

a 1
= (1.1111 1)

Figure 2.39  The Per-unit Equivalent Circuit of a Transformer with Off-nominal Tap-ratio 1.1111.

From Figure 2.39, the following relationships can be obtained by applying KVL,

Vs1 = Vs + I s Zeq( pu) (2.125)

and
Vp Vs1
= (2.126)
a 1

or V p = 1.1111Vs1 .
Note that V p and Vs1 differ from one another only by magnitude since ‘a’ is real. These expres-
sions can be employed to solve for network variables where transformers are provided with off-nom-
inal tap setting. However, while analysing a large power system, where computer programs are used,
a two-port p-model of Figure 2.39 is employed [14]. It should be noted that in many power system
studies, only the power base is specified, assuming that the voltage base (which is chosen on one sec-
tion of the power system) is floated across transformer windings as per the nominal voltage ratings of
the transformer.

2.7  CHARACTERISTICS OF A TYPICAL POWER SYSTEM

A conventional power system (see Figure 2.36) is made up of interconnection of three major systems
as follows:
1. Generation systems
2. Transmission systems
3. Distribution systems

2.7.1  Characteristics of Generator and Primemover Systems

In conventional power systems, a 50/60 Hz AC power is generated using large capacity synchronous
generators. Based on the speed of operation, these generators are classified as salient pole and
non-salient pole generators (Figure 2.40).
For a power plant, the type of generator is decided based on the fuel used for driving the primemov-
ers [17]. For example, in a coal (or oil, natural gas, nuclear) -based power plant, where steam turbines

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 92 9/12/2017 11:12:24 AM

 Basics of Power Systems  |  93 

Armature b’ Rotor
coils c
N N
+
+
+
a a’ +
+
S S
c’ Field
Stator b coils
Stator
Salient pole construction Non−salient pole construction

Figure 2.40  Structure of Salient Pole and Non-salient Pole Generators.

are the primemovers, a high-speed power generation is employed, whereas in a water-based power
plant, where hydraulic turbines constitute the primemovers, a low speed power generation is used. In
these cases, non-salient pole (or round rotor) construction is used for steam-based power plants, and
salient structure for hydropower plants [18].

2.7.1.1 Some Inferences about Generator Structures and their

Capacity
For 50 Hz frequency, a high-speed, non-salient pole machine may have two or four poles. As per (2.2),
its speed is either 50 rps (3000 RPM) or 25 rps (1500 RPM). In comparison to this, a low-speed salient
pole machine may have 36 or even 48 poles. Therefore, the speed of the generator is as low as 2.7778 rps
(166.67 RPM) or 2.0833 rps (125 RPM). To limit the peripheral speed (= p Dns m/s) to safe values, the
non-salient pole generators, also known as turbo-alternators, are designed with a smaller diameter, whereas
salient pole generators are built with a large diameter to accommodate large number of poles. The instan-
taneous value of induced emf per turn in a generator is given by the expression,
e = 2Blv V
where, B denotes the instantaneous value of spacial distribution of field flux density (in tesla) over a
pair of poles, l is the length of the conductor (in m) and v is the peripheral velocity (in m/s).
Therefore, for a practical machine, the wave shape of the flux density (in the air gap) decides the
wave shape of the voltage. This is because with iron core, the vectors, B, l, and v, remain mutually per-
pendicular to one another. Therefore, to produce a sinusoidal induced emf, it is required to establish
a sinor flux density wave in the air-gap. Such a field distribution is achieved in a salient pole gener-
ator by properly shaping the pole shoes, whereas, in a non-salient pole generator, this is obtained by
adjusting the pitch of the rotor field winding and also by altering the number of turns per winding in
different rotor slots (as we move away from the polar axis). It is found that a round rotor field offers a
better flux density wave than a salient pole structure.
The typical ratings and design data of two large synchronous generators are listed [18] as follows:
1. Salient pole generator, G1:
Three-phase, 15 kV, 500 MVA, 50 Hz, nominal power factor = 0.95, 166.67 RPM, 36 poles.
Diameter, D = 9.25 m, Axial Length, L = 2.35 m

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 93 9/12/2017 11:12:25 AM

 94  |  Power System Analysis

Field winding ratings (DC): 330 V, 2.4 kA (Brushless excitation system).

Air-gap length = 3.3 cm, moment of inertia J = 4.14 kg-m2.
2. Non-salient pole generator, G2:
Three-phase, 27 kV, 1530 MVA, 50 Hz, nominal power factor = 0.9, 1500 RPM, four poles.
Diameter, D = 1.8 m, axial Length, L = 7.49 m
Field winding ratings (DC): 600 V, 11.2 kA (Brushless excitation system).
Air-gap length = 12 cm, moment of inertia J = 0.085 kg-m2.
From the above listing, the following observations are made:
1. The speed of G1 is 2.7778 rps and that of G2 is 25 rps. From (2.1), it is clear that for a given power
rating, the size of the machine for a salient pole structure is larger than that for a non-salient pole
construction, unless a high value of G0 is chosen. Since G0 depends on the electric and magnetic
loading of the machine, it cannot be increased indefinitely for a salient pole construction. Hence,
large-capacity generator units are built with non-salient pole structures.
2. For a given power rating, the size of a low-speed salient pole generator is larger than that of a
high-speed non-salient pole generator. This facilitates easier cooling of salient pole machines. For
example, for a 500 MVA salient pole generator, air-based cooling systems are adequate, whereas
a non-salient pole machine of the same capacity requires a sophisticated hydrogen-based cooling
system due to its compact design.
3. If S is the MVA rating of a synchronous generator, theoretically, it implies that the generator can
be made to deliver real power, Pg equal to the rating S at unity power factor, and with regard to
reactive power, the machine can be made to either supply (a lagging load) or absorb (i.e., deliver-
ing a leading load), Qg equal to S. In other words, it must be seen that S = ( Pg2 + Qg2 ) is satisfied
at a given power factor. However, in practice, this is not feasible due to the following constraints:
(a) Due to economical reasons and for better utilisation, the prime movers are generally rated such that
the MW capacity is equal to (nominal power factor) × the MVA rating of the generator. Since the
real power delivered by a generator has to be balanced by the mechanical power input, the deliv-
erable Pg is limited to the turbine rating (at unity power factor). If the generator is made to deliver
this amount of real power, its gets loaded to the rated MVA power at the nominal power factor.
(b) At lagging loads, until the nominal power factor is reached, the armature heating constraints
the reactive power supply. Beyond this power factor, the field heating dominates as the
machine has to be over-excited to deliver more Qg. This shows that as when Qg is increased,
the real power loading on the machine must be reduced. Therefore, when Pg is zero, the max-
imum lagging Qg the machine can generate is typically 80% of S.
(c) At leading loads, since the machine has to be under-excited to deliver a leading Qg, it is noted
the maximum deliverable Qg is less than that of the lagging region, mainly due to increased
eddy current heating of iron parts of the armature, especially, the end regions of the armature.
Therefore, when Pg is zero, the maximum leading Qg the machine can generate is typically
60% of S. For a given real power, a large leading reactive power operation is avoided as it
worsens the small signal stability of the system. This is because as per (2.105), it can be seen
that when the machine is under-excited, its internal field voltage reduces, making the power
angle to increase, for a given real power. Here, a large power angle denotes a reduction in the
stability margin as the theoretical maximum power angle feasible is 90°.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 94 9/12/2017 11:12:25 AM

 Basics of Power Systems  |  95 

These specifications are generally obtained from capability curves of a synchronous generator [2] (see
Figure 2.41).

Qg
0.8 225% excitation
Field heating
limit (A−centre) Armature heating
limit (0−centre)

Nominal power
Lagging factor (cos F )
E
F 1.0
0 Pg
Leading V
Under−excitation Primemover
limit limit
F d

100% excitation 0.6 A Small−signal stability limit

Figure 2.41  Capability-curve of a Typical Synchronous Generator.

4. When a synchronous machine is run as a motor, its field can be controlled so that it can deliver
or absorb reactive power while driving mechanical loads. In a specially designed synchronous
motor, referred to as synchronous condensor, the machine will be on no-load and the complete
rating of the machine is utilised for the reactive power support.
5. Since the voltage rating of the field systems is very low in comparison to that of the armature
windings, it is convenient to have field structures as the rotating part of the generator. This also
facilitates easy connection of the exciters to the field winding.

2.7.1.2  Features of Primemover Systems

Two major types of primemover systems are given as follows:
1. Thermal-based steam-turbine systems
2. Water-based hydro-turbine systems
The typical characteristics of these primemover systems are as follows:
1. The most efficient steam turbines exhibit an efficiency of 45% in the thermal to mechanical
conversion process. Therefore, the overall efficiency of a thermal power plant lies in the range
of 25%–30%. The efficiency of large hydro turbines is about 90% and the overall efficiency of a
hydropower plant lies in the range of 60%–80%.
2. Due to thermal systems and their large time constants, steam turbines do not quickly respond
to load changes as it varies. Therefore, thermal power stations are normally used to supply base
loads which remain almost the same over a long period. Since hydro stations have small response

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 95 9/12/2017 11:12:26 AM

 96  |  Power System Analysis

time, they are made to serve the daily load variation. Hence, these stations are referred to as peak
load stations.
3. In thermal power stations, energy extraction from steam takes place in many stages. In each
stage, extraction is done at different steam pressures, such as high pressure, intermediate pres-
sure, low pressure, and so on. Therefore, these arrangements constitute a coupled mechanical
system, having many natural frequencies of oscillations. In addition, the turbine blades have their
own mechanical vibration-related properties. Furthermore, thermal stations are characterised by
a complex set-up for handling energy flow activities supported by induction motor drives. Due
to all these factors, thermal stations have limited ability to withstand large speed deviations (and
hence frequency), following a disturbance. In this context, a frequency decline from the nominal
is more dangerous as it may lead to tripping of turbine units. A simplified structure of a steam-
based power station is shown in Figure 2.42.

Valve control P
GV

Steam
Crossover
Reheater piping

Main Generator
Boiler Main
shaft shaft
HP
turbine IP LP

P
m

Burner Mechanical
input
Water tubes

Boiler feed Condenser

Water pump

Cooling water

Figure 2.42  A Simplified Structure of a Steam-based Power Station.

In contrast, hydropower plants are designed to withstand a relatively large frequency deviations
since water flow cannot be controlled easily due to their large inertia. A simplified block sche-
matic of a hydropower station is shown in Figure 2.43.
4. Due to elaborate arrangements in a thermal station, the station auxiliaries require a large amount
of power, in the range of 1%–5%, of the station capacity. Unless this power is supported, gener-
ators cannot produce their rated output. If a station is to be started from zero power to its rated
capability, a thermal power station requires a long time in comparison to a hydropower station.
These characteristics are very crucial if a complete system-wide blackout occurs.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 96 9/12/2017 11:12:27 AM

 Basics of Power Systems  |  97 

Head
Water
Reservoir

Generator

Penstock Valve
control Generator hall

Pelton turbine
with buckets

Figure 2.43  A Simplified Block Schematic of a Hydropower Station.

5. The complex mechanical systems associated with steam turbine-driven generators may cause
interaction with the electrical system, leading to mechanical shaft damage. This is generally
referred to as subsynchronous resonance (SSR) in power system. This is remotely possible with
hydropower plants.
6. The controllers provided on the prime movers offer a means to control the mechanical input to the
generator and hence, the real power shared by a generator and to regulate system frequency.
7. The energy cost of thermal generation is generally much higher than that of hydropower generation.

2.7.2  Characteristics of Transmission and Distribution Systems

Transmission and distribution systems convey power from the location of generation to the
load centres.

2.7.2.1  Characteristics of Transmission Systems

Transmission systems are constituted by highly networked lines which operate at extra-high voltage
in the range of 220 kV and above. They are characterised by a high value of  x L  ratio. This ratio
 r 
lies in the range of 5 to 20 depending on the nominal voltage of lines. L
For the purpose of accurately modelling them in power system studies, lines are generally classified
as follows [4]:
1. Long-lines (greater than 300 km): For these line lengths, the distributed effects of parameters are
more dominant. However, for the fundamental frequency power system studies, they are repre-
sented by an equivalent p model. These lines have considerable shunt capacitance depending on
the operating voltage.
2. Medium-lines (greater 80 km and less than 300 km): To model these line lengths, lumped param-
eter representation is adequate. Hence, they are modelled with nominal p model.
3. Short-lines (less than 80 km): These lines exhibit negligible capacitance-charging effects.
Therefore, they are represented by lumped series inductance.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 97 9/12/2017 11:12:27 AM

 98  |  Power System Analysis

Note: In the literature, there is no clear-cut distinction between transmission lines and distribution
sections based on their nominal voltages alone. The lines which are highly networked whose nominal
voltage is as low as 33 kV are also treated as transmission systems depending on their  x L  ratio.
 r 
These are also referred to as sub-transmission systems. L
Another important characteristic of transmission lines is their power transfer capability. In the liter-
ature, this capability is referred to as loadability of lines [19]. The following three factors limit the
loadability of lines:
1. Stability limit: Since long-lines offer considerable reactance, to transfer a given amount of power
(over a large distance), as per (2.105), it is possible only if a large angle is maintained. However,
such an angle reduces the stability margin and hence, the stability-related criterion constraints
large power flow in long-lines.
2. Voltage drop limit: In case of medium-length lines, before stability issues become prominent, the
acceptable voltage at the receiving end restricts large power flow transfer.
3. Thermal limit: The maximum temperature withstanding ability of conductors determines the
thermal limits. This criterion limits the maximum power transfer through short-lines.
Notes:
1. Line lengths are decided based on the wavelength, l, of a power line given by [20]:
vl
l= (2.127)
f

where, vl denotes the propagation velocity (in m/s) of electromagnetic waves. This is related to
line parameters of a lossless line as,
1
vl =
LC
with L = inductance (in H) per unit length and C = shunt capacitance (in F) per unit length of the
line. f represents the power frequency in Hz. It found that for over-head lines, vl is slightly less
than 3 × 108 m/s. With this vl, for a 50 Hz wave, the wave-length is given by 6000 km.
2. While specifying the power transfer capability of a line, it is common to express it in terms of the
surge impedance loading (SIL) of a line in MW for 3-phase. This load power is obtained as

VR2
SIL = (2.128)
Zc

where, VR denotes the line-to-line voltage rating of a line (in kV) and the per-phase surge imped-
ance in ohms is given by

L
Zc =
C

The significance of this impedance is that if a lossless long line is terminated by a resistor of value
equal to Zc, the voltage profile all along the length of the line is equal to the sending-end voltage.
Such a line is referred to as a flat-line.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 98 9/12/2017 11:12:29 AM

 Basics of Power Systems  |  99 

3. It is noted that for a 500 kV rated voltage, the typical SIL of an over-head line is 1000 MW. The
thermal limit is about three times its SIL. For an uncompensated over-head line of length 500 km,
the permissible power transfer limit is close to its SIL. However, in practice, it may be required to
transfer power much more than SIL. This requires adequate compensation of lines.
4. In comparison to over-head lines, AC cables have very high shunt capacitance. Because of this,
many a times the charging current exceeds the rating of the cable, severely limiting usage of
cables for long distances.

2.7.2.2  The IEEE 14-Bus Power Systems with Network Devices

A typical network of transmission lines is shown in Figure 2.44 [21]. Here, the lines are highly net-
worked to improve reliability of supply to loads. For the purpose of illustration, the IEEE 14-bus
test power system (with SB = 100 MVA and maximum nominal voltage of 132 kV) is divided into
x 
two-­areas interconnected by four lines, 1-5, 2-5, 2-4, and 3-4, referred to as tie-lines. The  L  ratio
 rL 
of HV lines lies in the range of 2.5 to 4.5. Note that Area-2 has no generation and a total real power

13
14

12
11
10

8
6 9
C
C 7
Area−2
5 4

P = 0.2326
P = 0.7549
P = 0.4152 Loop
P = 0.5613 Flow

1 2
G
G 3
Area−1
C

G Generator

C Synchronous Condenser

Figure 2.44  The IEEE 14 Bus Power System Showing Network of Transmission Lines.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 99 9/12/2017 11:12:29 AM

 100  |  Power System Analysis

load of 1.4310 pu in Area-2, is to be supplied from Area-1 through these tie-lines. However, in this
base case, a total of 1.7314 pu flows out of Area-1 to Area-2 and a power of 0.2326 pu returns back
to Area-1 through line 3-4. This leads to a condition called loop flow in power system network. Such
a condition cannot be avoided easily since it is mainly due to the uncontrolled power flow in AC lines
which are governed by physical laws. This causes additional losses in the system and under-utilisation
of line capacity limiting power transaction schedules.
In order to regulate steady state power flow in power systems, many network devices are employed.
These may be in the form of high-voltage direct current (HVDC) lines [22] or flexible AC transmis-
sion systems (FACTS) [23]. These power electronic-based devices enable fast and reliable control of
system quantities and even line impedances. With appropriate controllers, they can be used not only
to improve steady-state performances (e.g., bus voltage/power flow regulation), but also to augment
dynamic behaviour such as small-signal stability/transient stability/load-induced stability/SSR char-
acteristics of power systems [24].

2.7.2.3  Operation with an HVDC Link

In the above power system, in order to show the influence an HVDC line, a two-terminal HVDC link
is introduced between buses 3 and 4 in place of an AC line (Figure 2.45).

1 − 5 ( P = 0.6700)

2 − 5 (P = 0.2541)

Area−1 Area−2

2 − 4 (P = 0.3500)

(Pdr = 0.2)

3 Rec. 4
Inv.
DC line

Figure 2.45  The IEEE 14 Bus Two-area Power System with an HVDC Link.

A major advantage of an HVDC link is that a desired DC-line flow, Pdr, can be set independently, by
just controlling the firing angle of the thyristor-based converters. For a Pdr = 0.2 pu, it can be seen
from the figure that the new total real power flow from Area-1 to Area-2, is 1.4741 pu, where loop
flow is eliminated.

2.7.2.4  Operation with a TCSC

Consider a thyristor-based FACTS device, say, thyristor-controlled series capacitor (TCSC) [25, 26]
for compensating the line reactance. This device consists of a thyristor-controlled reactor in parallel
with a passive capacitor of reactance, XCF. Here, by simply controlling the firing angle, it is possible

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 100 9/12/2017 11:12:30 AM

 Basics of Power Systems  |  101 

to make the device to offer a reactance, XTCSC, either in the capacitive or inductive region. In the
capacitive region, a reactance equal to two or three times more than XCF can be obtained, whereas the
inductive operation is generally avoided as it requires excessive rating for the elements.
When this device is used in line 3-4 (see Figure 2.46), it failed to prevent the loop flow demanding
inductive region of operation [27]. Instead, the TCSC is placed in a non tie-line 2-3, where the base
case real power flow is 0.7322 pu. For a capacitive compensation of XTCSC = 0.135 pu, the real power
flow in the line increased to 1.0 pu. This forced a real power of 0.015 pu in line 3-4 from bus 3 to 4,
thus removing the loop flow. This behaviour is mainly due to the reason that in an AC system, the real
power flow in a line is decided by network laws involving phase-angle difference, bus voltage magni-
tude, and line impedance. This, in turn, is governed by load-generation balance. In case of an HVDC
link, since it can interconnect two AC systems operating at different nominal frequencies, the depend-
ency of the DC line flow on the bus phase-angle difference does not arise. Therefore, a desired power
flow can be achieved in the DC line as long as the host AC system can support such a power setting. In
cases, where two AC systems are interconnected only by such power electronic-based converters, there
will not be any contribution of inertia effects from one system to the another. Such a link is referred
to as asynchronous link. It is noted that in the IEEE 14-bus power system, the loop flow is eliminated
even for an HVDC link between buses 2 and 3 instead of an AC line. Another important feature of
an HVDC link is that for its converter operation, a large reactive power should be supplied by the AC
system. This adds to operational issues and initial cost of terminal equipment. These observations,
guided by socio-economic effects, are crucial while making an appropriate choice for power system
augmentation, out of many options.

1−5

2−5

Area−1 Area−2

2−4

Reactor

3 X CF 4
TCSC

Figure 2.46  The IEEE 14 Bus Two-area Power System with a TCSC.

2.7.2.5  Characteristics of Distribution Systems

The distribution systems interface load centres to the transmission sections at the sub-station point.
Generally, loads are served at different voltages depending on the nominal power levels. For example,
large industrial loads are delivered at high voltage in the range 66 or 33 or 11 kV, commercial loads

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 101 9/12/2017 11:12:30 AM

 102  |  Power System Analysis

at 33 or 11 kV, and residential loads are at 11 kV. These voltages may be further step-down to 400 V
(with neutral to supply single-phase loads). The line diagram of a typical 11 kV distribution feeder
system is depicted in Figure 2.47.

Substation point
L1 L2 L3 L4

∆
∆ Y

33/11 kV
transformer
Load
L1 to L4 − represent feeder sections of 3 km length each

Figure 2.47  Line Diagram of a Typical 11 kV Distribution System.

These systems are characterised by dominantly radial lines without much networking unlike in
transmission section. Their nominal voltage is relatively low compared to that of transmission lines.
x
Furthermore, their L ratio is quite low, many a time, less than one. Due to these features of distri-
rL
bution systems and lack of availability of load data, it is difficult to include them in systems studies,
retaining their identity. Hence, in most of the system analysis, only the aggregate model of distribution
systems is used at the transmission end. This not only simplifies the analysis, but also facilitates better
understanding of system-wide operation, control, and planning.
In many cases, the conventional techniques developed for transmission system analysis fail for
x
radial distribution systems which have very low L ratio. In such cases, these systems are handled
rL
exclusively. For example, to carry out a load flow of radial distribution systems special techniques
based on backward–forward sweep method have been developed [28].
With development in renewable-based energy resources such as wind, solar, and other energy sys-
tems, the distribution sections provide a means for grid interconnection of these resources.

2.7.3  Interconnected Operation of Power Systems

A large power system is generally divided into many regions depending on the geographical location.
Such a region typically consists of the following:
1. A number of generating stations located at different places in the region based on the availability
of fuel, for example, coal, oil, water reservoir, nuclear, wind, and solar, etc.
2. A common transmission network which interconnects all these generating stations leading to
resource mix.
3. Load centres in the same region which are delivered through the transmission network.
4. Regional electricity boards, load dispatch centres (LDCs), power supply corporations, etc., to
manage generation and load supply.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 102 9/12/2017 11:12:31 AM

 Basics of Power Systems  |  103 

Limited energy resources and environmental issues have stressed the need for interconnected opera-
tion of regional power systems as they permit meeting of a large energy demand by utilising the energy
which is abundant in the neighbouring systems. Such interconnection has the following advantages:
1. It increases the reliability of supply. With interconnections, a load centre is connected to genera-
tors by many lines which naturally ensures continuity of supply.
2. It ensures economical operation. When many regions are interconnected, it permits resource
mixing across the regions. Therefore, a load can be supplied to minimise the cost of generation.
3. It reduces plant reserve capacity. To supply a given load, each station is required to be built with
lower installed capacity. Hence, it reduces the cost of a plant and it can be built in less time.
4. It permits exchange of peak loads. With interconnection, peak loads occurring on a region can be
met by the reserves available in other regions through regional ties. This enables better utilisation
of resources.
5. It provides emergency support. Following a major disturbance, if generators lose synchronism with
neighbouring systems, then power plants have to be tripped to save them, and generators are shut
down. This condition is called blackout. Following this, if a thermal power plant has to be restarted to
generate nominal power, such plants can be provided with start-up power through interconnections.
6. It improves dynamic performances of the power system. The interconnection of many regional
power systems by AC lines effectively connects all generators in parallel although they are phys-
ically located at far-away places. This leads to a large synchronous grid where the inertia of indi-
vidual generators adds to a large system inertia. Therefore, when a disturbance occurs, it causes a
smaller speed deviation than that would result in an isolated region. Therefore, an interconnected
power system will be able to withstand a large disturbance in comparison to a regional system.
Furthermore, with interconnections, the reserve capacity of the system increases. This enables the
system to survive a large disturbance, say, a huge load increase. It also permits grid connection of a
large capacity intermittent generation in the form of renewable resources.
Although interconnection of regional power systems offers many benefits, it is not easy to operate
and maintain such grids due to the following reasons:
1. In order to evolve such a grid, it requires a coordinated control of regional systems.
2. Load-generation balance has to be maintained in each region so that under steady-state, the
frequency of the interconnected system is identical all over the grid.
3. Due to AC interconnections, following a perturbation at any point in the grid, the disturbance
spreads over the entire grid where all generators respond.
4. The severity of fault level increases.
5. Each regional system should follow grid discipline with respect to tie-line scheduled power flows.
A control centre is required to ensure that the overall grid is secure and stable.

2.7.4  Characteristics of Loads

Power system loads mainly draw real and reactive powers from the grid. In a conventional grid the real
power demand should be met by the synchronous generators through an energy conversion process,
where mechanical power is converted into electrical power. Hence, this process strongly influences
the frequency of operation than the generated voltage. On the other hand, the reactive power, which

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 103 9/12/2017 11:12:31 AM

 104  |  Power System Analysis

is related to field energy and being dominantly voltage dependent, can be supplied locally. Therefore,
for a power system to attain sinor steady-state, equilibrium should be reached both in the mechani-
cal system and in the electrical field system. Of these, the real power balance is more critical and is
commonly referred to as load-generation balance. If this balance is not achieved, it causes frequency
excursion, leading to rotor-angle stability issues. Similarly, depending on the reactive power balance,
system voltage may either increase or decrease. However, if such an equilibrium is not reached, then
it may lead to voltage excursion leading to voltage stability issues.

2.7.4.1  Balancing of Real and Reactive Power Loads

In power system operation, these loads are treated as demands which must be matched in terms of
power generation, in real-time, as and when they appear on the system since energy cannot be stored in
large quantities (except in small capacity batteries). Such a constraint exists for real power as reactive
power need not be transferred all the way from generating station. The random nature of loads further
complicate this issue as loads vary continuously through out the day, over a week, and exhibit seasonal
variations. To handle this nature of loads, utilities perform the following:
1. A rough estimate of future real power load that is likely to appear on the system is obtained by car-
rying out forecast analysis using historical data. This analyses spans over an hour, a day, a week,
or a few months ahead. Based on this, it commonly known as short-term or long-term forecast.
2. Based on the forecast, the LDC plans the generation pattern of a region so that the cost of meeting
the load demand, is minimised. It involves two tasks: (a) to decide a set of generators that is to be
kept ready to take up the load and (b) to determine the real power that each committed generator
has to produce. This process is referred to as optimum scheduling of generators.
Once this plan of load share among the generators is established, each generator is supplied with an
appropriate speed reference setting for its primemover controller so that the generator takes up its
share of real power load. In addition, a speed governor in the primemover control system regulates
the mechanical input to the generator so that the frequency excursion is prevented. This controller uses
the speed deviation signal as its input. A maximum of ±5% frequency deviation is permissible from the
nominal value. This means that for a 50 Hz nominal frequency, the system frequency should be between
47.5 Hz and 52.5 Hz. The regulation characteristics of speed governors of all scheduled generators are
coordinated in such a way that the sum of all real power loads on the system (including the real power
losses in the system) is shared among the generators at a common system frequency. Therefore, sys-
tem frequency can be controlled only by a coordinated global effort involving many generators spread
across many regions. In other words, frequency of the system cannot be controlled at the customer end.
The speed regulation of a governor has a droop characteristic so that when frequency decreases
(due to loading), it increases the mechanical input, depending on the slope setting of the droop char-
acteristic. In other words, if frequency increases, the governor reduces the mechanical input to the
generator. Such a droop naturally provides coordination among different speed governor controllers.
The slope setting of the speed governor is expressed in per-unit using the ratings of the generator as
the base. This per-unit speed regulation of a governor is defined as
 ∆f R 
 f 
o
s = pu frequency deviation/ pu load change. (2.129)
 ∆PL 
 P 
o

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 104 9/12/2017 11:12:33 AM

 Basics of Power Systems  |  105 

where, D fR denotes the frequency change (in Hz) = ( fnL – f ), with  fnL representing the no-load fre-
quency and f the system frequency, following a load change equal to D PL in MW, f0 is the nominal
frequency in Hz, P0 denotes the real power rating of the generator which is equal to (MVA rating ×
nominal power factor).
The typical value of s lies in the range of 4 to 5% for thermal generators, and even 10% for hydro
generators. For example, if a 200 MW, 50 Hz generator has 4% speed (governor) regulation, then it
indicates a speed regulation of (0.04 × 50)/200 = 0.01 Hz per MW. This implies that for a load change
equal to its full-load rating, the frequency of the unit changes by 0.01 × 200 = 2 Hz. If the no-load fre-
quency of the generator is set to 50 Hz, when the generator is made to deliver 200 MW, its frequency
falls to 48 Hz.
A typical plot of system frequency for Indian grid is shown in Figure 2.48, which also shows an
event where the southern grid is synchronised to the rest of the grid [29]. Once they are synchronised,
the two frequencies merge to become a single grid frequency. Note that the grid frequency does not
remain as a constant, it shows variation depicting random load changes.

WAFMS Filtered data of 2 Jan 2014

51

50.8
Southern Grid
Frequency (Hz) −−−−−−−−−−−−−>

50.6
Instant of synchronization
50.4

50.2

50

49.8
Rest of India Grid
49.6

49.4

49.2

49
5 10 15 20 25 30 35 40 45
Time elapsed after 20:23:50 GMT (in minutes) −−−−−−−−−>

Figure 2.48  Sample Plot of Indian Grid Frequency [Courtesy: WAFMS Group, IIT Bombay, India].

While balancing the real power loads through speed governor action, the reactive power requirement is
supplied by controlling the system voltage at different load points. The efforts start from the generator
end where the reference setting for the excitation controllers is modified to change the reactive power
shared by a generator. However, many a time, this alone may not be adequate to set a desired voltage at
the load end. In such cases, local voltage control is exercised in the form of transformer tap settings or
capacitor connections as most loads are inductive in nature. Therefore, unlike frequency, bus voltage
is amenable for local control at the customer end.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 105 9/12/2017 11:12:36 AM

 106  |  Power System Analysis

2.7.4.2 Frequency- and Voltage-dependency of Loads and their

Benefits
Another important feature of loads is that the magnitude of both real and reactive power drawn by
loads vary as a function of system frequency and bus voltage, with respect to their nominal values
[19]. For example, in induction motors (which constitute nearly 60% of system loads), the real power
delivered to the motor depends on the rotor speed. This, in turn, is decided by the synchronous speed,
and hence the system frequency. Therefore, if the system frequency decreases, the real power drawn
by the induction motor also decreases. In addition, if bus voltage magnitude decreases, the air-gap
torque reduces, in turn, causing the real power to drop from the nominal value.
In addition to induction motor loads, many other types of loads exhibit real power dependency on
frequency and voltage. For system analysis, as an approximation, an aggregate model of all type of loads
is considered at the transmission point. A typical frequency-dependent power system load is given by

 ∆f 
PL = PLo 1 + k pf 
 f o  (2.130)

where, Δf = (f − fo), represents the frequency deviation from the nominal value (in Hz), f denotes the sys-
tem frequency in Hz, kpf is the frequency sensitivity coefficient and PLo is the nominal value of load at fo.
In the above expression, the voltage dependency can be included as
b
 ∆f   V 
PL = PLo 1 + k pf  (2.131)
 f o   Vo 

where, V is the bus voltage, Vo denotes the nominal voltage and b is a constant.
Such frequency- and voltage-dependent characteristics of loads are beneficial for load-generation
balance in a system with limited generation. To understand this, consider a system with fixed real power
generation delivering a nominal load at 50 Hz. This implies that the mechanical input power settings are
held constant and speed governors are disabled. If real power load increases by a small amount, D P, then,
it tends to reduce the speed of the generators as the electrical output becomes higher than the mechanical
input. This causes the system frequency to decrease. If the old loads (which were being supplied prior to
load increase) are frequency-dependent, then these loads start drawing lesser power than the earlier value.
This leads to a ‘release’ of load on the system. If the frequency independent load D P, is equal to this
released amount, then the system ends up in delivering the new load as well as the old loads. Therefore, a
load-generation balance is achieved at a new frequency lower than 50 Hz, but with the same generation.
In the above case, if the old load is considered as frequency-independent, then following a load
increase, system frequency would decrease continuously, leading to system collapse due to loss of
synchronism. With regard to load-generation balance, the dependency of real power on frequency and
voltage helps systems accommodate small load changes without activating the primemover control-
lers. This is much more meaningful when power system size increases due to interconnections. Since
most system loads are frequency- and voltage-dependent, in a large system, connection/disconnection
of small loads may not disturb the load-generation balance even when the mechanical inputs are set
constant (without governor action). In addition, these frequency-dependent real power loads impart
positive damping to the system, thereby, improving the small-signal stability performance of systems.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 106 9/12/2017 11:12:37 AM

 Basics of Power Systems  |  107 

This happens because with frequency-dependent loads, a deviation in speed brings a corresponding
change in the load. Since a change in the electrical power output causes an opposite change in gener-
ator speed, it can be seen that such a load naturally damps the speed deviation.

2.7.4.3 Example: Frequency-dependent Real

Power Loads and Load Share
Two 50 Hz generators rated 200 MW and 400 MW are operating in parallel on a bus. The speed
governor of both the generators has a droop characteristic of 4%. Assume that the generators are oper-
ating at 50 Hz on no-load. It is required to determine the load shared by the generators and the bus
frequency in the following cases:
1. A frequency-independent nominal load of 600 MW is connected to the bus.
2. A load of 600 MW at nominal frequency 50 Hz, which is frequency-dependent with frequency
sensitivity factor of 2, is connected to the bus.
Notes: The expression for speed regulation can be rewritten to calculate the per-unit change in load
on the generator as
∆f R
∆PL f
= o (2.132)
Po s pu

From the above expression, it can be seen that for a given frequency change, ΔfR, if the per-unit reg-
ulations, spu, of generators are equal, then they share a common load proportional to their ratings,
in otherwords,

∆f R ∆f R
∆PL1 ∆PL 2 fo f
= = = o
Po1 Po2 s pu1 s pu2

1. Load is frequency-independent:
(a) By rearranging (2.132), we can determine the regulation of the generators in Hz/MW, from
the per-unit values as

s pu × f o ∆f R
s act = =
Po ∆PL (2.133)

   Therefore,

0.04 × 50
s act1 = = 0.01 Hz / MW
200
0.04 × 50
s act2 = = 0.005 Hz / MW
400

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 107 9/12/2017 11:12:38 AM

 108  |  Power System Analysis

(b) Now, for a given change in system frequency, ΔfR, the load shared by the generators are such that,
ΔPL1 + ΔPL2 = PL (2.134)
where, PL is the common total load on the bus at actuals.
  Using (2.133) in the above equation we get,
( f nL1 − f ) ( f nL 2 − f )
+ = PL (2.135)
s act1 s act 2

Using the numerical values, we have,
(50 − f ) (50 − f )
+ = 600
0.01 0.005
Solving for f, we get the system frequency as 48 Hz.
(c) Substituting f = 48 Hz, in the individual ΔPL expressions, the load shared by generators 1 and
2 can be computed as 200 MW and 400 MW, respectively.
2. Load is frequency-dependent:
(a) The frequency-dependent load is represented by
 ( f − fo ) 
PL = 600 1 + 2  (2.136)
 fo 

Using the above equation in (2.135), and substituting the numerical values, we get,
(50 − f ) (50 − f )  ( f − 50) 
+ = 600 1 + 2 (2.137)
0.01 0.005  50 
Solving the above equation, we get the new system frequency with frequency-dependent load
as 48.148 Hz.
(b) Substituting f = 48.148 Hz, in the individual ΔPL expressions, the load shared by generators
1 and 2 can be computed as 185.185 MW and 370.37 MW, respectively. From (2.136) the
common load on the bus can be obtained as 555.5556 MW.
From the example, the following observations are made:
1. When the load is frequency-independent the system frequency is 48 Hz, whereas if the load is
frequency-dependent, there is an improvement in the system frequency.
2. An intended load of 600 MW at the nominal frequency reduces to 555.5556 MW with frequen-
cy-dependent load characteristics. This leads to a release of load enabling the generators to supply
additional customers.

References
 [1] Gerhard Neidhofer, 50-Hz frequency, The IEEE Power and Energy Magazine, pp. 66 87,
July–August, 2011.
 [2] M. G. Say, The Performance and Design of Alternating Current Machines, CBS Publishers and
Distributors, New Delhi, 1983.
 [3] R. M. Kerchner and G. F. Corcoran, Alternating Current Circuits, Wiley Eastern Limited, New
Delhi, 1991.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 108 9/12/2017 11:12:40 AM

 Basics of Power Systems  |  109 

 [4] Olle I. Elgerd, Basic Electric Power Engineering, Addison–Wesley Publishing Co., London, 1977.
 [5] Shashidhar M. Kotian and K. N. Shubhanga, Dynamic Phasor Modelling and Simulation,
Proceedings of the IEEE Conference, INDICON-2015, India, December 2015.
 [6] Hadi Saadat, Power System Analysis, McGraw-Hill International, Singapore, 1999.
 [7] R. Ramanujam, Power System Dynamics - Analysis and Simulation, PHI - Learning Pvt. Ltd.,
New Delhi, 2010.
 [8] E. A. Guillemin, Introductory Circuit Theory, 2nd ed., New York, John Wiley and Sons, 1955.
 [9] Chi-Tsong Chen, Linear Systems Theory and Design, Oxford University Press, NY, 1999.
[10]  Graphic Symbols for Electrical and Electronics Diagrams, IEEE Standard-315, 1975.
[11]  IEEE Recommended practice for industrial and commercial power Systems Analysis, IEEE
Standard-399, 1997.
[12]  J. J. Grainger and W. D. Stevenson Jr, Power System Analysis, McGraw-Hill International Edition,
Singapore, 1994.
[13]  O. I. Elgerd, An Introduction Electric Energy Systems Theory, Tata McGraw-Hill Publishing
Company Limited, Mumbai, 1983.
[14]  M. A. Pai, Computer Techniques in Power System Analysis, Tata McGraw-Hill Publishing
Company Ltd, New Delhi, 1979.
[15]  R. Ramanujam, Computational Electromagnetic Transients, I. K. International Publishing House
Pvt. Ltd, New Delhi, 2014. 1995.
[16]  J. Duncan Glover and Mulukutla S. Sarma, Power System Analysis and Design, Thomson
Learning Brooks Cole Pvt. Ltd, Asia, 2002.
[17]  Steven W. Blume, Electric Power System Basics, The Institute of Electrical and Electronics
Engineers, Inc, 2007.
[18]  Theodore Wildi, Electrical Machines, Drives, and Power System, Pearson Education Asia Pvt.
Ltd, 2001.
[19]  P. Kundur, Power System Stability and Control, McGraw-Hill Inc., NY, 1994.
[20]  John D. Ryder, Networks, Lines and Fields, Prentice-Hall of India Pvt. Ltd., New Delhi, 1995.
[21]  Power System Test Archive - UWEE (University of Washington). Available at: http://www
.ee.washington.edu/research/pstca.
[22]  K. R. Padiyar, HVDC Power Transmission Systems Technology and System Interactions, Wiley
Eastern Limited, New Delhi, 1990.
[23]  Dusan Povh, Use of HVDC and FACTS, Proceedings of the IEEE, vol. 88(2), February 2000.
[24]  Narain G. Hingorani and Laszlo Gyugyi, Understanding FACTS, Concepts and Technology of
Flexible AC transmission systems, Standard Publishers Distributors, First Indian edition, 2001.
[25]  K. R. Padiyar, FACTS Controllers in Power Transmission and Distribution. New Age International
Publishers, New Delhi, India, 2007.
[26]  Ritty Raju and K. N. Shubhanga, Laboratory Implementation of a Thyristor Controlled Series
Capacitor, The IEEE International Conference on Signal Processing, Informatics, Communication
and Energy Systems, (IEEE SPICES 2015), India, February 2015.
[27]  K. Udaya Kishore and K. N. Shubhanga, Modeling of FACTS Controllers in Power Flow
Programmes to Analyze Prevention of Loop Flows, The IEEE International Conference,
INDICON-2015, India, December 2015.
[28]  G. W. Chang, S. Y. Chu, and H. L. Wang, An Improved Backward-Forward Sweep Load Flow
Algorithm for Radial Distribution System, The IEEE Trans. on Power Systems, vol. 22(2),
pp. 882–884, May 2007.
[29]  Prof. A. M. Kulkarni, Department of Electrical Engineering, IIT Bombay, India, Wide area
frequency measurement systems, Available at: http://www.ee.iitb.ac.in/anil.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 109 9/12/2017 11:12:40 AM

 110 | Power System Analysis

Review Questions
1. Can an ideal fundamental frequency sinusoidal voltage source supply a non-linear load which
draws harmonic current? At what frequency does the generator deliver real power?
2. What is the basic difference between vector analysis and phasor analysis?
3. In a 3-phase, 2-pole alternator, phase-a winding having T number of turns is
(a) placed in a single-slot as a bunched coil.
(b) placed in 3 slots per pole with a full-pitch coil arrangement.
Derive the wave shape of the resultant voltage in the above two cases by carrying out analytical
calculations. Given that the peak values of 3rd, 5th and 7th harmonic voltages are 0.2, 0.2 and 0.2
T
per-unit, respectively, per coil of turns. The fundamental voltage peak per coil can be take as
3
1.0 pu. This exercise shows the importance of distributing the coils in different slots in a genera-
tor, pertaining to a given phase.
4. In a 3-phase, balanced Y -connected alternator, show that even if triplen harmonics are present
in the phase voltages, they will not be present in the line-to-line voltages. This is true only for
co-phasor nature of triplen harmonics in a 3-phase generator.
5. The circuit elements such as r, L and C are generally treated as passive loads. Then, what are
active loads?
N
6. For a two-winding transformer with turns ratio 1 , using the parameter matrices [r] and [L],
N2
establish an equivalent circuit (referred to side-1) for a conventional fundamental frequency
analysis.
7. For a 3.3 kV, 1250 kVA, 50 Hz, 300 rpm, Y -connected three-phase engine-driven alternator has
the following specifications:
Specific magnetic loading = 0.58 T
Specific electric loading = 33000 A/m
Core length = 1.1 × pole-pitch
Winding factor = 0.955
(a) Obtain the preliminary dimensions of the rotor core.
(b) Estimate the number of turns/phase required to generate the rated voltage on open circuit.
8. A 15 hp, 400 V, 1420 rpm, 50 Hz, three-phase induction motor develops a torque of 50 Nm at
1460 rpm. Compute the per-unit value of torque, speed, and power output.
Given that
S B = input VA rating,
VB = 400 V and
Base speed = Synchronous speed (mechanical)
Full load efficiency = 0.8 and full-load power factor = 0.9 lagging. Take 1 hp = 735.5 W.
9. What do you mean by power flow balance and how to identify attainment of such a condition?

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 110 9/12/2017 11:12:41 AM

 Basics of Power Systems | 111

10. Consider a single-phase, 50-Hz system as shown in Figure 2.49.

vac

R1 = 0.01 Ω L1 = 0.01 H L2 = 0.01 H R2 = 0.01 Ω

+ ia1 ia2 +

va1 C = 31µF va2

Figure 2.49 Single-phase 50-Hz System.

(a) Write a state-space model choosing the state vector as x = [i1 i2 vc ]T .

(b) Modify the above mentioned state model so that it becomes applicable to sinusoidal steady-
state condition and the resulting phasor equations appear as follows, in terms of, I a1 , I a2 ,
and VaC :
( R1 + jx L1 ) 0 1  I a1  Va1 
    
 0 − ( R2 + jx L2 ) 1   I a2  = Va2 
 − jxc jxc −1 VaC   0 

Assume that Va1 = 100 ∠30° V and Va2 = 100 ∠0° V.

(c) Compute the complex power associated with each element including the sources and verify
the real and reactive power balance.
(d) Construct waveforms va1 (t ), va2 (t ), ia1 (t ), ia2 (t ), vaC (t ) , and inductor voltage vaL1 (t ).
(e) Construct pa1 (t ) = va1 (t )ia1 (t ) waveform.
11. Rewrite Figure 1 so that it is applicable to a three-phase balanced system assuming Y -connected
elements.
(a) Obtain I abc1, I abc2 , and VabcC .
(b) Compute three-phase complex powers associated with the sources.
12. Choose a three-phase base, S B = 1000 VA and VB = 175 V, obtain the per-unit impedance diagram
on single-phase basis for Figure 1. Determine the real and reactive power losses in the system.
13. Enumerate the need of grounding neutrals in EHV systems.
14. Explain, why thermal power stations are regarded as base-load stations and the hydro plants are
referred to as peak-load stations.
15. Is it feasible to increase the power transfer capability of a line indefinitely, by raising the transmis-
sion voltage rating?
16. Using a simplified model, show that by inserting a series capacitor in a transmission line the
power transfer capability can be increased.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 111 9/12/2017 11:12:48 AM

 112 | Power System Analysis

17. Why is a large frequency deviation detrimental to power system operation in comparison with
voltage deviation?
18. A 500 kW, 60 Hz, 2300 V, six-pole alternator A is paralleled with a 300 kW, 60 Hz, 2300 V, four-
pole machine B. Both machines have a frequency (speed) regulation of 2.43%. The machines are
carrying an equal share of a total 400 kW bus load at a frequency of 60.5 Hz.
(a) Determine the no-load frequencies of the machines.
(b) If the bus load increases to a total of 500 kW, calculate the operating frequency of the bus
at the new load. Assume that the load is frequency independent. Further, determine the load
carried by each machine.
(c) The new load that consumes 500 kW at f o = 60 Hz is frequency-dependent and its character-
istic is given as
 ∆f 
PL = 500 1 + 2.5  kW,
 fo 

where ∆f = f − f o in Hz represents the frequency deviation.

Compute the new bus frequency and the load shared by each generator.

M02_POWER-SYSTEM-AN00_SE_XXXX_CH02.indd 112 9/12/2017 11:12:49 AM

 Park-Based Transformations 3
CHAPTER OUTLINE

3.1 Time-Varying Parameters of a 3.3 Representation of a Transmission

Synchronous Machine Line in Machine-Frame
3.2 Three-Phase-Based Phase-Locked
Loop

In this chapter, different types of time-varying transformations are discussed. Park transformation is
first derived and other transformation matrices such as Clarke’s transformation (stationary-frame),
Kron’s transformation (synchronous-frame), and rotor–reference-frame are obtained as a special case
of the Park transformation.
In addition, power-invariant and power-variant nature of transformations are highlighted. Some
applications of these transformations, in system analysis, are presented with illustrative examples.

3.1 TIME-VARYING PARAMETERS OF A SYNCHRONOUS MACHINE

Consider a three-phase conventional synchronous machine as shown in Figure 3.1.
The combined voltage equations (for the stator and the rotor) can be expressed as
dy
v = [ R] i + (3.1)
dt
Since flux-linkages of the windings depend on the rotor positions, we have

y = [ L(q )] i
(3.2)
Using eq. (3.2) in eq. (3.1) we have,
d i d [ L(q )] (3.3)
v = [ R] i + [ L(q )] + i
dt dt

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 113 6/2/2018 12:13:10 PM

 114  |  Power System Analysis

Magnetic axis of aˇ -a
(Ref-axis)
θ q-axis
d-axis c
bˇ

iF
a
ω
aˇ

cˇ b

Figure 3.1  A Three-phase Synchronous Machine Showing Winding Details.

The above equation can also be written as

d i dq d [ L(q )]
v = [ R] i + [ L(q )] + i (3.4)
dt dt dq

dq
Note that eq. (3.4) is time-varying even when is constant and hence the solution of such
dt
systems of equation is not straight forward. This is due to the fact that the inductance matrix [L] is
time-varying. It would be advantageous if the time-varying machine equations can be transformed
into a time-invariant set. This would result in the simplification of calculations for steady-state and
transient conditions. R. H. Park introduced a time-varying transformation known as ‘Park transfor-
mation’ which transforms time-varying quantities into time-invariant quantities [6]. The objective
of this transformation is to diagonalise the parameter matrices, especially the inductance matrix,
and help transform stator quantities into rotor quantities, removing the dependency on q. The deri-
vation of this transformation and some observations about this transformation are discussed in the
following section.

3.1.1  Derivation of a Time-varying Transformation

In this section, a time-varying transformation matrix is derived to facilitate easy analysis of time-­
varying system of equations.
In Figure 3.2, a, b, and c represent the magnetic axes of the stator three-phase windings distributed
symmetrically in space by 120° electrical with respect to each other and are fixed in space (Figure 3.1).
Let fabc be the quantities associated with abc windings. From Figure 3.2, it can be seen that we have
followed a convention where d-axis leads the q-axis by an angle 90° and q is the angle between the
magnetic axes of a-a′ coil and the d-axis, which is the magnetic axis of the field coil, following the
a−b−c sequence.

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 114 6/2/2018 12:13:12 PM

 Park-Based Transformations  |  115 

)
eld
t h e fi
f
is o Magnetic axis of phase−a-a'
c ax -axis
et i d
n a
ag
(M

xis
q-a

1
θ
f)

=
(Re

θ)
−
θ

0
(9
(120 + θ1) (120 − θ )
1

c
b

Figure 3.2  Axes Representation for the Derivation of the Park Transformation.

Now, obtaining the components of fa, fb, and fc along d- and q-axes, we can get,
fd = fda + fdb + fdc
fq = fqa + fqb + fqc
Writing the above equations in complex notation, we get,
fq + jfd = ( fqa + jfda) + ( fqb + jfdb) + ( fqc + jfdc)
Referring to Figure 3.2 and rewriting the above equations using the angle q1 = 90° - q, we have,

f q + jf d = f a e jq1 + f b e j (120 +q1 ) + f c e − j (120 −q1 )

= fa [cos q1 + jsin q1] + fb [cos (120 + q1) + j sin (120 + q1)]
+ fc [cos (120 − q1) - j sin (120 − q1)]

Substituting q 1 = (90° – q ) in the above equation we get,

fq + jfd = fa[sin q + jcos q ] + fb [cos (90 – ( q −120)) + j sin (90 – (q – 120))]

+ fc [cos (–90 + (q + 120)) – j sin (–90 + (q + 120))]

= fa [sin q + j cos q ] + fb [sin(q −120) + j cos (q −120)] + fc [sin (q + 120)

+ j cos (q + 120)]

From the above equation, we can write,

  2p   2p  
f q = k q  f asinq + f b sin q −  + f c sin q + 
  3   3  

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 115 6/2/2018 12:13:13 PM

 116  |  Power System Analysis

  2p   2p  
f d = k d  f a cos q + f b cos  q −  + f c cos  q +  
 3 3 

where kq and kd are arbitrary constants.

In addition to fd and fq, a zero-sequence quantity, f0 is also considered to account for asymmetry
in fabc as
f0 = k0 [ fa + fb +fc]
where k0 is an constant.
Writing the above equations in matrix form,

 f0   fa 
   
   
 f d  = [ P ]  f b  (3.5)
   
   
 fq   f c 
 

where [P] is the transformation matrix given by

 k0 k0 k0 
 
 
  2p   2p  
k cosq k d cos  q - k d cos  q +  
[ P ] =  d  3
  3  (3.6)
 
 
 k sinq  2p   2p 
k q sin  q −  k q sin  q +  
 q  3  3   

The inverse transformation is given by
 fa   f0 
   
   
 f b  = [ P ]−1  f d  (3.7)
   
   
 f c   fq 
 
For the transformation matrix to be orthogonal, i.e., [P]−1 = [P]T, we have,

 k0 k d cosq k q sinq 
 
 
  2p   2p  
−1  k0 k d cos  q −  k q sin  q - 
[P] =   3  3 
 
 
k  2p   2p  
k d cos  q +  k q sin  q + 
 0  3  3  

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 116 6/2/2018 12:13:15 PM

 Park-Based Transformations  |  117 

Using the following identities,

 2p  2 2p  3
cos2q + cos2 q −  + cos q + =
 3 3  2
 2p  2 2p  3
sin 2q + sin 2 q −  + sin q + =
 3 3  2
 2p   2p 
cosq + cos q −  + cos q +  =0
 3  3 
 2p   2p 
sinq + sin q −  + sin q +  =0
 3   3 
 2p   2p   2p   p
2p
sinq cosq + sin q −  cos q −  + sin q +  cos q +  = 0 (3.8)
 3 3 3 3 

we can note that [P][P]T is a diagonal matrix and if [P] is to be an orthogonal matrix, then [P][P]T= [I].
Therefore, we have
3k02 0 0 
  1 0 0 
   
 3 2   
[ P ][ P ]T = 0 2
kd 0  = 0 1 0
   
   
 
 0 3 2  0 0 1 
0 kq
 2 
From the above equation, we can write
1
3k02 = 1 ⇒ k0 =
3
3 2 2
kd = 1 ⇒ kd =
2 3
3 2 2
kq = 1 ⇒ kq =
2 3
Therefore, the transformation matrix simplifies to
 1 1 1 
 2 2 2 
 
 
2   2p   2p  
[P] = cosq cos q −
 
3 
cos q +
   (3.9)
3 
3
 
 
  2p   2p  
 sinq sin q −  sin q + 
  3   3  

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 117 6/2/2018 12:13:18 PM

 118  |  Power System Analysis

The above transformation is known as Park transformation in the power-invariant form. This transfor-
mation is also referred to as arbitrary reference-frame transformation since q can be set to w t, where
w represents an arbitrary speed.

Notes:
• fa, fb, and fc represent a set voltages/currents/flux-linkages which excite the three-phase symmet-
rical winding and produce effects along the magnetic axis of the windings.
• The transformation is not restricted to sinusoidal variation of fabc quantities. fabc can represent any
time-variation of the signal, even balanced or unbalanced.
• If fabc is a balanced set of sinusoidal quantities and possesses an angular speed equal to w, then
f0 = 0 and fqd are DC values. For example, if f represents currents in the three-phase windings,
it establishes a revolving field rotating at speed w. When it is viewed through the transformation
matrix, it provides a field effect which remains stationary with respect to the rotor which itself is
running at speed w.
• f0qd quantities represent variations in the mutually perpendicular directions.

3.1.2  Transformation Matrices in Different Reference-frames

From the Park transformation, different transformation matrices are derived by just substituting
different values for q. The various transformation matrices used in the literature are as follows:

3.1.2.1  Rotor Reference-frame

For this reference frame, in eq. (3.9), q is set to w0t + d, where w0 represents the synchronous speed.
This reference-frame is also called machine reference-frame and is applicable to synchronous
machines. See Figure 3.3 for details on various angles.

α
Magnetic axis a−a' coil
D ω ot (Stationary frame)

q
d δ

θ
Q

D−Q: Synchronous-frame
d−q : Rotor-frame

Figure 3.3  Orientation of Different Reference-frames.

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 118 6/2/2018 12:13:19 PM

 Park-Based Transformations  |  119 

3.1.2.2  Synchronous Reference-frame

In this reference-frame, in eq. (3.9), q is set to w0t and the transformation matrix [PK] is given by

 1 1 1 
 2 2 2 
 
 
2   2p   2p 

[ PK ] = 3
cosw 0 t cos  w 0 t − 
 3
cos  w 0 t +  
 3   (3.10)
 
 
  2p   2p  
 sinw 0 t sin  w 0 t − 3  sin  w 0 t +  
 3 


This transformation is also known as Kron’s reference-frame and is represented as f0DQ = [PK] fabc

3.1.2.3  Stationary Reference-frame

In this reference-frame, in eq. (3.9), q is set to zero and the transformation matrix [PC] is given by

 1 1 1   1 1 1 
   3 3 3 
2 2 2  
 
   
2 1 1   
2 1 1  (3.11)
[ PC ] =  1 − −  =
 3
− −
3 2 2 6 6
   
   
 3 3  1 1 
 0 −  0 − 
2 2   2 2 

This reference-frame is also known as Clarke’s reference frame or ab reference frame and is
represented as
f0ab = [PC] fabc

3.1.3  Some Properties of the Transformation Matrices

Some of the important properties of the transformation matrices are as follows:
1. The transformation [P] is orthogonal. This implies that
If ui and uj represent ith and jth columns of [P], respectively,
then we have,
 u T . u = 1, if i = j
i j
and
 uiT . uj = 0, if i ≠ j

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 119 6/2/2018 12:13:20 PM

 120  |  Power System Analysis

or in other words, [P]¯1 = [P]T, the transformation is power-invariant. This is shown below:
Using the identity, [P]¯1 = [P]T in eq. (3.7) we have,

fabc = [P]T f0dq

Therefore, we can write
vabc = [P]T v0dq
iabc =[P]T i0dq

Instantaneous power is given by

p = iTabc vabc
= ([P]T i0dq) T[P]T v0dq
= iT0dq [P] [P]T v0dq
p = iT0dq v0dq
Thus, we have

p= ∑ vi = ∑ vi (3.12)
abc 0 dq

Therefore, we can conclude that transformation is power-invariant,

2. Using the identities
 1  3
cos(q − 2p / 3) = −   cos q +   sinq
 2  2 

 1  3
cos(q + 2p / 3) = −   cosq −   sinq
 2  2 

 1  3
sin(q − 2p / 3) = −   sinq −   cosq
 2  2 

 1  3
sin(q + 2p / 3) = −   sinq +   cosq
 2  2 

we can write the transformation matrix [P] in eq. (3.9) as

 1 1 1 
 2 2 2
1 0 0   
   
  2  1 1
[ P ] = 0 cosq −sinq  1 − − 
  3  2 2
   
0 sinq cosq   
 3 3
0 −
2 2 

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 120 6/2/2018 12:13:22 PM

 Park-Based Transformations  |  121 

With q = w0t + d, the matrix further simplifies to

 1 1 1 
 3 3 3 
1 0 0  1 0 0  
   
   2 1

1 
[ P ] = 0 cosd −sind  0 cosw 0 t −sinw 0 t   − −
   3 6 6
   
0 sind cosd  0 sinw 0 t cosw 0 t   
 1 1 
 0 − 
 2 2 

Therefore, we can write

 1 1 1 
 3 3 3 
 f0  1 0 0  1 0 0    fa 
      
     2 1

1   
 f d  = 0 cosd −sind  0 cosw 0 t −sinw 0 t   − − fb
     3 6 6 
      
 f q  0 sind
   cosd  0 sinw 0 t cosw 0 t   f 
1  
c
 1
 0 − 
 2 2 

1 0 0  1 0 0   f0 
   
     (3.13)
= 0 cosd −sind  0 cosw 0 t −sinw 0 t   fa 
   
   
0 sind cosd  0 sinw 0 t cosw 0 t   f b 

1 0 0   f0 
  
    (3.14)
= 0 cosd −sind   f D 
  
  
0 sind cosd   f Q 
where
 f0  1 0 0   f0 
    
    
 f D  = 0 cosw 0 t −sinw 0 t   fa 
    
    
 f Q  0 sinw 0 t
   cosw 0 t   f b 

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 121 6/2/2018 12:13:25 PM

 122  |  Power System Analysis

Considering only D-and Q-axes in the above equation, we have,

 f D  cosw 0 t −sinw 0 t   fa 
    
 =  
 f Q   sinw 0 t
   cosw 0 t   f b 

One may write the above equation in the complex form given by

f Q + jf D = ( f b + jfa )e − jw 0t (3.15)

Rewriting eq. (3.14), we get

 f0  1 0 0   f0 
    
    
 f d  = 0 cosd −sind   f D 
    
    
 f q  0 sind cosd   f Q 
  

Note that the above equations are valid for any time-variation of fabc and even for balance/
unbalance condition on fabc.
Suppose we consider only d- and q- axes, then we have

 f d  cosd −sind   f D 
    
 =    (3.16)
 f q   sind cosd   f Q 
  

The above equation in the complex form is given by

( )
f q + jf d = f Q + jf D e − jd (3.17)

Notes:
· The expression given in eq. (3.17) is generally referred to as ‘Park transformation in phasor
form’. This is valid only under balanced sinusoidal steady-state condition on fabc.
· The (  fQ + jfD ) denotes a single-frequency ‘true’ phasor, f , since it traces a sine-wave in the
abc-frame. However, (  fq + jfd ) is a vector assumed to be fixed in the rotor which is running at
speed w0. For ease of visualisation, we can state that id and iq are vectors which produce effects
similar to the field current and revolve in the space as the rotor rotates. Therefore, it represents a
space-phasor.

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 122 6/2/2018 12:13:27 PM

 Park-Based Transformations  |  123 

3.1.4  A Change of Reference for Angle Measurement

It should be noted that in Figure 3.3, all angles are measured taking the magnetic axis of coil a - a′
as the reference. This axis coincides with the a-axis. However, in stability studies, the rotor angle,
d which is measured relative to the synchronous-frame, is of major importance. In Figure 3.3, the
synchronous-frame is constituted by the D- and Q-axes. Hence, instead of taking the D-axis as the ref-
erence, it is customary to shift the reference to Q-axis and measure the rotor angle as an angle between
q- and Q-axes. Since D- and Q-axes are symmetrical relative to the a- and b-axes, the angle (w0t) is
represented as an angle between Q- and b-axes. This is shown in Figure 3.4 where b-axis is taken as
the reference. Accordingly, in eq. (3.17), the q- and Q-axes are taken as the references. Further, note
that the Park transformation given in eq. (3.9) remains unaltered.

α Magnetic axis a−a' coil

D (Stationary frame)

q
d

δ Q
θ
ω ot
β

D−Q: Synchronous-frame
d−q : Rotor-frame

Figure 3.4  Modified Representation of Angle References.

3.1.5  An Alternate Form of Park Transformation

This transformation differs from the earlier Park transformation given in eq. (3.9) only by the angle
used in the transformation. The angle g is defined as shown in Figure 3.5.

α Magnetic axis a−a' coil

D (Stationary frame)

q
d

γ δ Q
θ
ωot
β

D−Q: Synchronous-frame
d−q : Rotor-frame

Figure 3.5  An Alternate Angle Reference.

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 123 6/2/2018 12:13:28 PM

 124  |  Power System Analysis

To obtain the corresponding transformation matrix, we just replace q in eq. (3.9) by g to get an alter-
nate form of the Park’s transformation. This is given by

 1 1 1 
 2 2 2 
 
 
2   2p   2p  
[ PA ] = 3 
cosg cos  g − 
 3
cos  g

+   (3.18)
3 
 
 
  2p   2p  
 sing sin  g − 
 3
sin  g

+ 
3 


p
where g = q + with q = w 0 t + d .
2
It is to be noted that the above transformation leads to the same ‘phasor form’ as given in eq. (3.17).
The only difference between the above and the previous transformation is that for a given constant DC
value of fq with fd = 0 and f0 = 0, eq. (3.9) assumes a balanced sine-wave, whereas eq. (3.18) assumes
a balanced cosine-wave as the reference waves. This is further clarified below:
Let us obtain f0dq using [PA] with d = 0, such that

f 0 dq = [ PA ] f abc

where,
T
  2p   2p  
f abc =  f m cos(w 0 t ), f m cos  w 0 t −  , f m cos  w 0 t +   (3.19)
  3   3 

In terms of sine-wave, the above fabc can be written as

T
  p  p 2p   p 2p  
f abc =  f m sin  w 0 t +  , f m sin  w 0 t + −  , f m sin  w 0 t + +   (3.20)
  2  2 3  2 3 

Using the properties given in eq. (3.8), we can see that f0 = 0, fd = 0 and the only non-zero term is the
3
fq = f m. A similar result is obtained with the Park transformation (given in eq. (3.9)) when fabc is
2
equal to
T
  2p   2p  
f abc =  f m sin(w 0 t ), f m sin  w 0 t −  , f m sin  w 0 t + 

  3 3   (3.21)

Therefore, these two transformations differ only when the quantities are obtained in the abc-frame.

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 124 6/2/2018 12:13:31 PM

 Park-Based Transformations  |  125 

3.1.6  An Example with Park Transformation

Using the Park transformation given in eq. (3.9), let us obtain the n0dq such that
v0 dq = [ P ] vabc
where
T
  2p   2p  
vabc = Vm sin(w t + f ), Vm sin  w t + f −  , Vm sin  w t + f + 
  3 3  

In the Park transformation, q = w0t + d and let z = w t + f . Since nabc denotes a balanced set of three-
phase quantities, n0 = 0. Now, nd and nq components are obtained as

2   2p   2p   2p   2p  
vd = Vm cos(q )sin(z ) + cos q −  sin z −  + cos q +  sin z + 
3   3 3 3 3  

1
Now, using the trigonometric identity, cosAsinB = [sin( A + B) − sin( A − B)], we get
2

2 1  4p   4p  
vd = 3
Vm sin(q + z ) − sin(q − z ) + sin  q + z −
2  3
 − sin(q − z ) + sin  q + z +
3
 − sin(q − z ) 

(3.22)
Note that
 4p   4p 
sin(q + z ) + sin q + z −  + sin q + z +  =0
 3   3 
Therefore, eq. (3.22) simplifies to
2 3
vd = − Vm [sin(q − z ) ]
3 2
3
= Vm [sin(z − q ) ]
2
3
= Vm sin [(w − w 0 )t + f − d ] (3.23)
2
1
Similarly, using the trigonometric identity, sinAsinB = [cos( A − B) − cos( A + B)] , we get
2
3
vq = Vm cos [(w − w 0 )t + f − d ] (3.24)
2
For a condition, where w = w0, expressions eq. (3.23) and eq. (3.24) are rewritten in complex notation
as,

3
(vq + jvd ) = Vm [ cos(f − d ) + jsin(f − d ) ]
2

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 125 6/2/2018 12:13:35 PM

 126  |  Power System Analysis

3
= Vm e j (f −d )
2
3
= Vm e jf e − jd (3.25)
2
3
Note that Vm e jf represents a phasor and it is written in rectangular form as
2
3
Vm e jf = VLL (cosf + jsinf ) = (vQ + jvD ) (3.26)
2
The following observations are made:
3
1. In eq. (3.26), the magnitude of the phasor, VLL = Vm. This represents the line-to-line value of
2
the voltage in a Y-connected system. Similarly, if fabc is a current vector, iabc with peak ampli-
3
tude Im, the application of the Park transformation results in I m. This magnitude is equal to
2
3 times the per-phase RMS current (or line current) in a Y-connected system. This scaling is

mainly due to the power-invariant nature of the Park transformation. This is another reason for
including 3 factor in the base current, IB in per-unit representation.
2. If d = 0, in eq. (3.25), then ndq represents the quantities in synchronous-frame, nDQ, and is equal
to that given in eq. (3.26). Therefore, with w = w0, an analysis in the synchronous-frame simply
denotes the phasor analysis on single-phase basis, however, the quantities are scaled by a factor 3 .
3. The expression given in eq. (3.25) is used in power-swing related stability analysis, where a low
frequency oscillation of rotors is studied. In this study, for each machine, its d is used to transform
the stator quantities, such as voltage and current, from the synchronous-frame to the respective
rotor-reference-frame. This also implies that on the network side, to which the machine is inter-
faced, all calculations are done in the synchronous-frame, neglecting high frequency transients.
This avoids calculation in the abc-frame which demands a large computational effort. Thus, in a
low frequency stability analysis, the network is assumed to be in quasi-sinusoidal steady-state,
where a low frequency modulation of the phasors is represented.

3.1.7  Power-variant Park Transformation

In eq. (3.5), if we choose [P] as given below, the transformation is referred to as power-variant
transformation.
 1 1 1 
 
 2 2 2 
2  2p   2p  (3.27)
[ P ] = cosq cos q −  cos q + 
3  3   3 
 
  2p   2p  
 sinq sin q −
 
3 
sin q

+ 
3  


M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 126 6/2/2018 12:13:39 PM

 Park-Based Transformations  |  127 

Also, [P]–1 is given by

1 cos q sin q 
 2p 2p 
[ P ]−1 = 1 cos q −  sin q −  
 3 3 
 
1 cos q + 2p  sin q + 2p  
  3   3  
Instantaneous power is given by
T
p = iabc vabc

{[ P ] }
−1 T
= i0 dq [ P ]−1 v0 dq
−1 −1
= i0Tdq ([ P ] )T [ P ] v0 dq
Consider the term ([P]–1)T[P]–1:

 1 1 1  1 cosq sinq 
  
  
  2p   2p     2p   2p  
−1 T −1  cosq cos  q −  cos  q +   1 coss  q −  sin  q −  
([ P ] ) [ P ] =   3  3   3  3 
  
  
 sinq  2p   2p   2p   2p 
sin  q −  sin  q +   1 cos  q +  sin  q +  
  3  
3    3  3  

 
3 0 0 
 
3
= 0 0
 2 
 3
0 0 
 2

Therefore, we have
3 0 0 
  v0 
  
 3  
0 0  v 
p = i0 id iq   2 d
  
  
 3  vq 
0 0 
 2

p=
3
2
(2 v0 i0 + vd id + vq iq )

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 127 6/2/2018 12:13:42 PM

 128  |  Power System Analysis

In a balanced three-phase circuit, the zero-sequence current does not exist; hence, the power is repre-
sented by
3
(
p = vd id + vq iq
2
)
As ∑vi ≠ ∑vi, we can conclude that the transformation is power-variant.
abc 0 dq

3.1.7.1  Calculation of `a-components

For the stationary reference-frame, by setting q = 0 in eq. (3.27), we get,
1 1 1 
2 2 2 
 
 
2 1 1
[ Pc ] = 3  1 −
2
− 
2

 
 
0 3 3
−
 2 2 
Therefore, we have,
1 1 1 
2 2 2  v 
v0    a
    
  2 1 1  
va  =  1 − −  vb 
  3 2 2  
    
 vb    v 
 
0 3 3  c 
−
 2 2 
Making use of the identity va + vb + vc = 0 and solving for va and vb , we get

 1 2
va = va − (vb + vc ) 
 2 3
  v  2
=  va −  − a  
  2  3

or
va = va
vc − vb
vb =
3
The above results will be useful in a phase-locked loop (PLL) implementation as discussed in the
following lines.

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 128 6/2/2018 12:13:45 PM

 Park-Based Transformations  |  129 

3.2 THREE-PHASE-BASED PHASE-LOCKED LOOP

A conventional three-phase phased-locked loop (PLL) system implemented in synchronous-­reference-
frame (SRF) is shown in Figure 3.6. A PLL is a closed-loop control system that drives the phase angle
error between its output signal and its input signal to a fixed value so that a wave is generated which
tracks the input signal. In the SRF-type PLL, a PI controller is used to force the fD-component of the
input signal to zero. This calculation uses the generated waves obtained out of a harmonic oscillator.
Therefore, it ensures that the generated wave has a phase shift exactly equal to 90° with respect to the
input reference.

PLL controller
Transformation Module

i xcw
La
PI A
iLb αβ DQ Oscillator
abc αβ iD − Controller
iLc iD error +
+ + x
sw

0 ω

Figure 3.6  Block Diagram of PLL.

The different sections in synchronous reference fame PLL (SRF-PLL) are briefly explained below.

3.2.1  Transformation from abc-frame to Synchronous-frame

Here, the input signal in the abc-frame is required to be represented in the synchronous-frame using
the generated waves. To carry out this, the transformation is done in two stages: first, the input signals
are transformed into stationary-frame and then, using the generated waves they are converted into
synchronous-frame.
Let iLa, iLb, and iLc be the three-phase line current input to the PLL. These input signals in the abc-
frame are converted to the stationary- frame (i.e., a b-frame) using the power-variant transformation
as shown below:

 1 1  iLa 
ia  1 − −  
2 2  
  2
 =3  iLb  (3.28)
ib    
  0 − 3 3  
 2 2  iLc 

Since iLa + iLb + iLc = 0, and from the above relationships, we can show that

ia = iLa

iLc − iLb
ib =
3

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 129 6/2/2018 12:13:46 PM

 130  |  Power System Analysis

Now, using the relationship given in eq. (3.15), the DQ-components of i- in synchronous-frame are
obtained from ab-frame using the following transformation:

iD   xcw − xsw  ia 

    
 =    (3.29)
iQ   xsw
   xcw  ib 

It is to be noted that the xcw and xsw are the signals obtained by solving the differential equation per-
taining to the harmonic oscillator, which will be discussed later. The obtained iD signal is negated to
obtain the error signal that should be driven to zero. It is fed to a PI controller whose gain values are
chosen as: Kp= 250 and Ki = 100.

3.2.2  Harmonic Oscillator

The differential equation governing the harmonic oscillator is given by,
..
x = −w2x(3.30)
where w is the frequency of oscillation in rad/s.
For this second order system, two state variables are defined as
xsw = w x
.
xcw = x

Hence, the state space model of the oscillator is given by

 xsw   0 w   xsw 
    
 =   (3.31)
 xcw   −w 0   xcw 

The initial conditions on the states are chosen as xsw (0) = 0 and xcw (0) = 1.
The state model of the harmonic oscillator given by eq. (3.31) is represented schematically in
Figure 3.7.

A xsw
1
s

xcw
1
s
−1

Figure 3.7  Schematic Diagram of Harmonic Oscillator (in Figure 3.6).

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 130 6/2/2018 12:13:49 PM

 Park-Based Transformations  |  131 

Using xcw and xsw signals, a balanced set of three-phase waves can be generated as
 0 1 
 f a′   
   3 1  xsw 
 f b′  =  2 −  
2   xcw  (3.32)
 f c′   
− 3 1
− 
 2 2 

3.2.3  Some Results

In this case study, a three-phase sinusoidal input is considered. They are given by
T
  2p   2p  
f abc = 1.0 sin(w t ), 1.0 sin  w t −  , 1.0 sin  w t + 
  3   3   (3.33)

The signal plots are shown in Figure 3.8. It is clear that the controller takes around three cycles to track
the input. Once the steady-state is reached, for a sinor input, xsw settles as a negative sine-wave and xcw
as a negative cosine-wave. This corresponds to a stable condition of the controller.

0
fa

−1
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
1
iD−Erro r

0.5

−0.5
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
1

0
Xsw

−1
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
1

0
X cw

−1
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Time (s )

Figure 3.8  Sample Plots of the PLL.

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 131 6/2/2018 12:15:39 PM

 132  |  Power System Analysis

The three-phase tracking waves are obtained by using eq. (3.32). These are plotted in Figure 3.9. Note
that under steady-state, it constitutes a three-phase cosine-wave given by
T
′   2p   2p  
f abc =  −1.0 cos(w t ) − 1.0 cos  w t −  − 1.0 cos  w t + 

  3   3   (3.34)

This represents a three-phase sine-wave set which lags the input fabc by 90°.

0.8

f'a f'b f'c

0.6

0.4

0.2
f'abc

−0.2

−0.4

−0.6

−0.8

−1
0.16 0.165 0.17 0.175 0.18 0.185 0.19 0.195 0.2
Time (s)

Figure 3.9  Output Three-phase Reference Wave.

3.3 REPRESENTATION OF A TRANSMISSION LINE

IN MACHINE-FRAME
Consider a symmetrical three-phase transmission line compensated by a fixed series capacitor. Figure 3.10
shows the single-phase equivalent circuit of this system connected between two nodes, where the
instantaneous phase-a node voltages are va and ea. It is given that the per-phase per-unit value of line
resistance is R and line reactance is xL. The per unit reactance of the capacitor is xC. These reactances
are provided at the fundamental frequency, wB = 2pfB rad/s, fB is the base frequency in Hz. Now,

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 132 6/2/2018 12:15:41 PM

 Park-Based Transformations  |  133 

writing the circuit equations for phase-a, in terms of the line current ia, and the capacitor voltage vCa,
we get the following relationships:

R L C
+
i v
a Ca
v e
a a

Figure 3.10  Single-phase Equivalent Circuit of a Three-phase Symmetrical Transmission Line.

By applying KVL, we have

dia
va = Ria + L + vCa + ea (3.35)
dt

and by applying KCL at the capacitor junction point, we get,

dvCa
C = ia (3.36)
dt
Notes:
· In the above equations, all quantities are in per-unit for a starr-connected network.
· The inductance L and capacitance C are obtained as
xL 1
L= and C =
wB xC w B

· If the circuit is energised by a sinusoidal voltage source of frequency w rad/s, under steady-state,
 d 
the above equations are written in the phasor form  with = jw  as,
 dt 

Va = RI a + jwLI a + VCa + Ea

jwCVCa = I a

· For representing the quantities and parameters in per-unit, the line ratings are used as the base
values.
· For analysis, purpose, consider eq. (3.36) at actuals and converting it into per-unit, we get,

 vC 
d  at 
Z dvCat  Vb  iat V
Cact × b = Cact × Zb = Ib = b
Vb dt  vC  dt I b Zb
d  at 
Z dvCat  Vb  iat V
CactIb×arebthe base
where, Zb, Vb, and = values
Cact × for
Zb each phase,= and, I b = b .
Vb dt dt Ib Zb

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 133 6/2/2018 12:15:44 PM

 134  |  Power System Analysis

1
Since Zb = , we get,
w B Cb
1 Cact dvCa
= ia
w B Cb dt

In per-unit, we have,
C pu dvCa
= ia (3.37)
wB dt
Note that
1
Cact xCt w B 1
C pu = = =
Cb 1 xC
Zbw B

The expression (3.36) can be written in an alternate form as given in eq. (3.37).
It is required to transform the equations in abc-frame to 0dq-frame. In order to carry this out,
eq. (3.35) is rewritten in matrix form accounting three phases as,

 va   R 0 0  ia   L 0 0  ia  vCa  ea 

       d      
vb  = 0 R 0  ib  + 0 L 0  dt ib  + vCb  + eb 
vc  0 0 R  ic  0 0 L  ic  vC  ec 
 c

Note that in the above equation, all quantities are in the abc-frame and represent instantaneous values.
In a compact form, the above equation can be rewritten as
d
vabc = [ R]iabc + [ L] iabc + vCabc + eabc (3.38)
dt
We know that
v0dq = [P]vabc
where
 1 1 1 
 2 2 2 
 
 
2   2p   2p 


[P] = cosq cos  q − 
 3
cos  q +  
 3   (3.39)
3
 
 
  2p   2p  
 sinq sin  q − 
 3
sin  q +  
 3 


which is the Park transformation matrix in power-invariant form, and also implies that [P]−1 = [P]T.

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 134 6/2/2018 12:15:46 PM

 Park-Based Transformations  |  135 

Hence, we can write that

vabc = [P]Tv0dq
iabc = [P]Ti0dq

vCabc = [ P ] vC0 dq
T

eabc = [P]Te0dq(3.40)

Substituting the above equations in eq. (3.38), we get,

[ P ]T v0 dq = [ R][ P ]T i0 dq + [ L]
d
dt
( )
[ P ]T i0 dq + [ P ]T vC0 dq + [ P ]T e0 dq

Premultiplying both sides by [P], we have,

v0 dq = [ P ][ R][ P ]T i0 dq + [ P ][ L]
d
dt
( )
[ P ]T i0 dq + vC0 dq + e0 dq

d  d 
= [ P ][ R][ P ]T i0 dq + [ P ][ L]  [ P ]T  i0 dq + [ P ][ L][ P ]T  i0 dq  + vC0 dq + e0 dq
 dt  dt

Since resistance and inductance matrices are diagonal, we have,

d  d
v0 dq = [ R]i0 dq + [ P ][ L]  [ P ]T  i0 dq + [ L] i0 dq + vC0 dq + e0 dq (3.41)
 dt  dt

Now consider,
d d dq
[ P ]T = [ P ]T
dt dq dt

 1 
 cosq sinq 
2
 
 
d 2  1  2p   2p   d
= cos q −
  sin  q −   (w t )
dq 3 2 3   3   dt
 
 
 1  2p   2p  
 cos q +  sin q + 
 2  3   3  

0 −sinq cos q 
d T 2 
[P] = w 0 −sin(q − 120) cos(q − 120)  (3.42)
dt 3
0 −sin(q + 120) cos((q + 120) 

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 135 6/2/2018 12:15:49 PM

 136  |  Power System Analysis

We can also see that,

L 0 0 
 
[ L](3×3) = 0 L 0 
0 0 L 

1 0 0 
 
= L 0 1 0
0 0 1 

= L[ I ](3×3) (3.43)

d
Using eqs. (3.39), (3.42), and (3.43) in [ P ][ L] [ P ]T , we get,
dt

 1 1 1 
 2 2 2  0 −sinq cosq 
d 2  
[ P ][ L] [ P ]T = w L cosq cos(q − 120) cos(q + 120)  0 −sin(q − 120) cos(q − 120) 
dt 3
 sinq sin(q − 120) sin(q + 120)  0 −sin(q + 120) cos(q + 120) 

0 0 0 
 
= w L 0 0 1 
0 −1 0 
(3.44)
Substituting eq. (3.44) in eq. (3.41), we have,
0 0 0 
  d
v0 dq = [ R]i0 dq + w L 0 0 1  i0 dq + [ L] i0 dq + vC0 dq + e0 dq
dt
0 −1 0 

or

v0   R 0 0  i0  0 0 0  i0   L 0 0  i0  vC0  e0 

         d      
vd  = 0 R 0  id  + 0 0 w L  id  + 0 L 0  id  + vCd  + ed 
vq  0 0 R  iq  0 −w L 0  iq  0 0 L  dt iq  vC  eq 
             q  

Combining the first two terms on the RHS, we get the desired equation in the 0dq-frame as

v0   R 0 0  i0   L 0 0  i0  vC0  e0 

      d      
 vd  =  0 R w L  id  + 0 L 0  id  + vCd  + ed  (3.45)
vq   0 −w L R  iq  0 0 L  dt iq  vC  eq 
          q  

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 136 6/2/2018 12:15:52 PM

 Park-Based Transformations  |  137 

3.3.1  Derivation of Capacitor Voltages in 0dq-frame

In eq. (3.45), nC0dq can be obtained as follows:
Writing eq. (3.36) in matrix form accounting all three phases, we have,

d 1
vC = iabc
dt abc C

Using eq. (3.40) in the above equation, we get,

d
dt
( 1
)
[ P ]T vC0 dq = [ P ]T i0 dq (3.46)
C
or

d
dt
( ) d 1
( )
[ P ]T vC0 dq + [ P ]T vC0 dq = [ P ]T i0 dq (3.47)
dt C
Taking the first term on the LHS to the RHS and premultiplying the resulting equation by [P] we get,

d d 1
(vC0 dq ) = −[ P ] ([ P ]T )vC0 dq + [ P ][ P ]T i0 dq (3.48)
dt dt C

From the previous analysis, we see that

0 0 0
d
( ) 
[ P ] [ P ] = 0 0
dt
T
w

0 −w 0 

Substituting the above result in eq. (3.48), we get,

0 0 0
d   1
(vC0 dq ) = − 0 0 w  vC0 dq + i0 dq (3.49)
dt C
0 −w 0 

3.3.2  Summary of Equations in dq-frame

From eq. (3.45), the dq-equations are separated as
did 1 R 1 1
= vd − id − w iq − vCd − ed
dt L L L L

diq 1 R 1 1
= vq − iq + w id − vCq − eq (3.50)
dt L L L L
From eq. (3.49), the dq-equations are separated as
dvCd 1
= id − w vCq
dt C

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 137 6/2/2018 12:15:55 PM

 138  |  Power System Analysis

dvCq 1
= iq + w vCd (3.51)
dt C

3.3.3  Summary of Equations in DQ-frame

Note that if the equations are to be transformed from abc-frame to the synchronous-frame then, in
eqs. (3.50) and (3.51) replace w by w0 and replace dq by DQ. Therefore, the equations in the synchro-
nous-frame are given by
diD 1 R 1 1
= vD − iD − w 0 iQ − vCD − eD
dt L L L L
diQ 1 R 1 1
= vQ − iQ + w 0 iD − vCQ − eQ (3.52)
dt L L L L
For the capacitor voltage, we have,
dvCD 1
= iD − w 0 vCQ
dt C
dvCQ 1
= iQ + w 0 vCD (3.53)
dt C
Under sinor steady-state, in eq. (3.52) the derivatives are set to zero and rearranged to get,
vD = RiD + w 0 LiQ + vCD + eD
vQ = RiQ − w 0 LiD + vCQ + eQ

Writing the above equations in complex form, we get,

(vQ + jvD ) = R(iQ + jiD ) + jw 0 L(iQ + jiD ) + (vCQ + jvCD ) + (eQ + jeD )

V = ( R + jw 0 L) I + VC + E (3.54)

The above expression represents the voltage equation for the circuit in the phasor form. Similarly,
from eq. (3.53), we can write that
1
iD = w 0 vCQ
C
1
iQ = −w 0 vCD
C
In the complex form, we have,
(iQ + jiD ) = jw 0C (vCQ + jvCD )

I = jw 0CVC

Therefore,
I
VC =
jw 0C (3.55)

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 138 6/2/2018 12:16:00 PM

 Park-Based Transformations | 139

The following observations are made:

1. Note that expressions (3.54) and (3.55) are in per-unit. If actual values are to be obtained, SB
and VB bases should be used in the usual way. If quantities are desired in the abc-frame, Kron’s-
inverse transformation should be used as follows:
fabc = [PK]−1 f0DQ
where, q = w0t.
2. Since fabc is obtained through a power-invariant transformation, the base values remain the same
even for phase quantities.
3. If abc-frame quantities are to be obtained from 0dq-frame, then use Park-inverse transformation
as follows:
fabc = [P]−1 f0dq
where, q = w0t + d.

References
[1] K. R. Padiyar, Power System Dynamics - Stability and Control, BS Publications, Hyderabad,
India, 2002.
[2] P. Kundur, Power System Stability and Control, McGraw-Hill Inc., New York, 1994.
[3] A. R. Bergen and V. Vittal, Power System Analysis, Pearson Education Asia, India, 2001
[4] P. M. Anderson and A. A. Fouad, Power System Control and Stability, Iowa State University Press,
Ames, Iowa, 1977.
[5] K. R. Padiyar, Analysis of Subsynchronous Resonance in Power Systems, Kluwer, Academic
Publishers, Boston, 1999.
[6] P. M. Anderson, B. L. Agrawal, and J. E. Van Ness, Subsynchronous resonance in power systems,
IEEE Press, Piscataway, New Jersy, 1990.

Review Questions
1. Enumerate different forms of the Park transformation matrix.
2. Realise the following KRON’s transformation using the basic SIMULINK blocks as a subsystem
ioDQ = [ PK ] iabc

where

 1 1 1 
 2 2 2 
 
 
2   2p   2p  
[ PK ] = 3
cosw 0 t cos  w 0 t −
 
3 
cos  w 0 t +
 
3 
 
 
  2p   2p  
sinw 0 t siin  w 0 t − 3  sin  w 0 t +
 
3  


M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 139 6/2/2018 12:16:01 PM

 140 | Power System Analysis

3. If uabc is given by
T
  2p   2p  
u abc = 10 sin(w 0 t ) 9 sin  w 0 t −  10 sin  w 0 t + 
  3   3  

and [PK] is the transformation matrix, then determine the frequency components in u0DQ.
4. Using SIMULINK, perform the following:
Treating
T
  2p   2p 
vabc = Vm sin (w 0 t − f ) , Vm sin  w 0 t − − f  , Vm sin  w 0 t + −f 
  3   3 

as the applied voltage, determine iabcL in a balanced RL-series circuit (R = 1 Ω and L = 0.1 H)
connected in Y for the following cases:
(a) f = 0 with iabcL (0) = 0. Here, iaL (0), ibL (0 ) , and icL (0) denote the initial conditions on the
integrator blocks.
(b) For a given value of f , set iabcL (0) so that iabcL (t ) reaches steady state without any dynamics.
In each of the above mentioned cases, obtain ioDQ using KRON’s transformation. Verify the
power-invariance property of the transformation under steady state.
5. Consider the transformation given by vodq = [ P ] vacb
where
 1 1 1 
 2 2 2 
 
 
2   2p   2p  
[P] = cosq cos q −
  cos q + 
3 3   3 
 
 
  2p   2p  
sinq sin q −
 
3 
sin q +
 
3  

and
T
  2p   2p  
vacb = Vm sin(g ) Vm sin  g +  Vm sin  g − 
  3 3  
Determine v0 dq for q = w 0 t and g = w 0 t.
6. For a balanced RL-series circuit with R = 1 Ω and L = 0.1 H, connected in Y , a balanced
three-phase voltage given by
T
  2p   2p  
vabc = 100sin(w 0 t ), 100sin  w 0 t −  , 100sin  w 0 t + 
  3 3  
is applied. Using the Park transformation with q = (w 0 t + d ) , determine the steady-state voltage
and currents in the machine frame for the following cases:

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 140 6/2/2018 12:16:11 PM

 Park-Based Transformations | 141

(a) when d = 0.
p
(b) when d = .
3
Assume w 0 = 2p × 50 rad/s.
7. Consider the following transformations in their usual notations
f 0 dq = [ P ] f abc

and
f 0 DQ = [ PK ] f abc

Derive the transformation matrix [ PD ] so that

f 0 dq = [ PD ] f 0 DQ

8. For a three-phase balanced system, it is given that

1
( vd2 + vq2 ) 2 = F (vabc )

Determine an expression for the function F in terms of abc variables.

9. Consider the following transformations:

f abc = [C P1 ] f 0 dq1 and f abc = [C P 2 ] f 0 dq2

where
 1 
 cosf 1 sinf 1 
2
 
 
2  1  2p   2p  
[CP1 ] = 3 2
cos  f 1−
 
3 
sin  f 1−
 
3 
 
 
 1  2p   2p  
 cos  f 1+  sin  f 1+ 
 2  3   3  

and

 1 
 cosf 2 sinf 2 
2
 
 
2  1  2p   2p  
[CP2 ] = 3 2
cos  f 2 −
 
3 
sin  f 2 −
 
3 
 
 
 1  2p   2p  
 cos  f 2 +  sin  f 2 + 
 2  3   3  

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 141 6/2/2018 12:16:19 PM

 142 | Power System Analysis

Obtain the transformation matrix [C P ] such that

f 0 dq2 = [C P ] f 0 dq1

10. For the circuit shown in Figure 3.11, obtain the dynamic equations in 0DQ frame. Determine the
eigenvalues of the system in the abc-frame and 0DQ frame.

ia
a

R
van

vbn L vcn
ib L R
b
R
ic
c

Figure 3.11 Star-connected RL-series Circuit.

M03_POWER-SYSTEM-AN00_SE_XXXX_CH03.indd 142 6/2/2018 12:16:22 PM

 Synchronous Machine
Modelling Using Primitive
Parameters 4
CHAPTER OUTLINE

4.1 Modelling of a Non-salient Pole 4.5 Expression for Torque in

Synchronous Machine Rotor-reference Frame
4.2 Modelling of a Salient Pole 4.6 Case Studies
Synchronous Machine 4.7 Rotor Mechanical Equation
4.3 Flux Linkage-current Equations 4.8 Analysis of Linear Systems
in Rotor-reference Frame
4.4 Voltage Equations in Rotor-reference
Frame

In this chapter, a detailed model of a synchronous generator is derived, based on the generalised
theory of machines. To begin with, the elements of the time-varying inductance matrix are deter-
mined, showing their complete evaluation.
Using the Park transformation matrix in the machine-frame-of-reference, the voltage and torque
equations are determined. To illustrate the usage of the model in terms of primitive parameters, case
studies such as open-circuit and short-circuit calculations are carried out. Some results have been
validated by conducting lab experiments as well.
This chapter also develops sufficient material to understand machine model in terms of the stand-
ard parameters which will be taken up in the next chapter.

4.1 MODELLING OF A NON-SALIENT

POLE SYNCHRONOUS MACHINE
In the process of modelling a synchronous machine, the following assumptions are generally made [1]:
1. Saturation is neglected: This makes the electrical part of the system linear.
2. Field distribution due to field and armature mmf is assumed to be sinusoidal: This enables the
fundamental frequency modelling of the system.

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 143 6/2/2018 2:50:20 PM

 144  |  Power System Analysis

Consider a non-salient pole synchronous machine as shown Figure 4.1 with three-phase symmetri-
cal stator windings and a field winding on the round rotor.

Magnetix axis of coil a-a'

Rotor position
θ=0 Negative
Rotor Magnetix axis of coil b-b'
b'
c
N
+

Armature +
+
coils
a + a'
+
Magnetix axis of coil b-b' S

Field b
θ = 120° coils. c'
Stator
Non-salient pole construction

Figure 4.1  A Three-phase Non-salient Pole Synchronous Machine.

Writing the voltage equations for the stator and rotor coils in the abc-frame, using the ‘motor conven-
tion’, we get,
dy
v = [ R] i + (4.1)
dt

where
v( 7×1) = [va vb vc v F 0 0 0]T

and

y ( 7×1) = [ L(q )]( 7× 7) i ( 7 ×1)

(4.2)
with
y ( 7×1) = [y aa′ y bb′ y cc ′ y FF ′ y HH ′ y GG ′ y KK ′ ]T

i ( 7×1) = [ia ib ic iF iH iG iK ]T

Note that the vector v denotes the voltage across the coils, y represents the flux-linkages of all coils,
and i indicates the current through the coils with motor convention, in the abc -frame [2]. The damp-
ing effects offered by the conducting paths of the rotor mass are denoted by fictitious short-circuited
windings, H, G, and K coils. Here, it is assumed that the H-coil is located along the rotor polar- or
d-axis, and G and K coils along the quadrature- or q-axis. This model detail is normally referred to as
the 2.2 model.

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 144 6/2/2018 2:50:23 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  145 

As shown earlier for primitive machines, the first step in modelling a machine is to obtain the
elements of the inductance matrix, [L(q)]. For a synchronous machine, employing the above modelling
details, the structure of the matrix is given by,
[ L11 (q )](3×3) [ L12 (q )](3× 4) 
[ L(q )]( 7× 7) = 
[ L21 (q )]( 4 ×3) [ L22 (q )]( 4 × 4)  (4.3)

where q is in electrical radian and hence is independent of number of poles.

4.1.1  Inductance Matrix for a Non-salient Pole

Synchronous Machine
In the inductance matrix, (4.3), each sub-matrix denotes the following:
[L11(q )] = Self and mutual inductances of the stator coils
[L12(q )] = Mutual inductances between stator and rotor coils
[L21(q )] = Mutual inductances between rotor and stator coils = [L12(q )]T
[L22(q )] = Self and mutual inductances of rotor coils
It should be noted that in a non-salient pole machine, the length of the air-gap is constant unlike in
a primitive machine considered earlier. Therefore, we can conclude that the sub-matrix [L11(q )] is a
constant matrix and its elements remain independent of the rotor position, and hence the angle, q. This
is because, irrespective of the position of the rotor (relative to the magnetic axis of coil a−a ′), the self-
and mutual- flux-linkages of the stator coils do not change. The elements of [L11(q )], for a non-salient
pole machine, can be written as
 Lss −Ms −Ms 
 
[ L11 (q )](3×3) =  − M s Lss − M s  (4.4)
 − M s −Ms Lss 

The following observations are made:
1. The self-inductance, Lss, is identical for all three phases. This is based on the assumption that the
magnetic and electrical circuits associated with each phase are symmetric.
2. While determining the mutual-inductance between two stationary coils, their flux linkage is obtained
relative to their respective magnetic axes. For example, to determine the mutual-inductance
between phases a and b, i.e., Lab, the following steps are employed:
(a) For phase-b, determine the flux, fb, produced by ib alone, along its magnetic axis b−b′.
(b) Compute the component of fb along the magnetic axis of coil a−a′. Denote this component
of flux which links phase-a as, fab. This is a constant and is independent of the rotor position.
(c) Note that the flux, fab, is in a direction opposite to that established by ia alone. In other words,
fab lies along the ‘-ve’ direction of the magnetic axis a−a′. Hence, the mutual-inductance
value is given a negative sign. Now, Lab is given by
y ab
Lab = = −Ms
ib
with all other currents set to zero.

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 145 6/2/2018 2:50:25 PM

 146  |  Power System Analysis

Note: Another way to compute fab is to obtain the magnitude of fb along the ‘negative’ mag-
netic axis of coil b−b′ (due to ib alone) and then take its component along the magnetic axis
of coil a−a′. Then, add a negative sign to it.
(d) The value of Lba is also equal to −Ms as the coils share a common magnetic circuit symmetrically.
3. Since the magnetic and electrical circuits are identical for each phase, the value of the self- and
mutual-inductances is the same for all coils.
Similarly, [L22(q )] is also a constant matrix, since it depicts the inductances of coils which are housed
on the rotor itself. Hence, their values are also independent of the rotor position. However, the sub-­
matrix, [L12(q )] (and hence [L21(q )]), is not a constant matrix, i.e., its elements depend on the rotor
position. This is because the magnetic axes of the rotor coils, themselves, move relative to the sta-
tor coils, depending on the rotor position. Hence, the mutual flux-linkages between them change as
a function of the rotor angle, q . The structure of [L12(q )] and [L22(q )] is identical for salient and
non-salient pole machines. The elements of these sub-matrices are derived in the following sections.

4.2  MODELLING OF A SALIENT POLE SYNCHRONOUS MACHINE

Figure 4.2 shows a three-phase, salient pole synchronous machine with three-phase stator windings
and the remaining four rotor windings. The rotor coils, H along the d-axis, and two coils, i.e., G and
K coils, along q-axis, represent the damping effects of pole-face windings.

Magnetic axis of a-a'

(Ref−axis)
c'
θ '
vcc q-axis
d-axis c
ic

K
ia G
a H
iF
'
vaa
F
ω vF
a'

b'
ib
'
vbb
b

Figure 4.2  Three-phase Salient Pole Synchronous Machine with Motor Convention.

Applying the ‘motor convention’, the voltage equation is obtained, which is identical to the one shown
in (4.1). For the model derivation, the elements of the inductance matrix, [L(q )](7×7), have to b­ e deter-
mined. Unlike in the non-salient pole machine, since the air-gap length is not constant, the [L11(q )](3×3)
matrix becomes a function of the rotor angle, q  and the elements differ from that shown in (4.4). These
elements are obtained in the following sections.

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 146 6/2/2018 2:50:25 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  147 

4.2.1  Determination of Inductances of Stator Coils

As stated earlier, this sub-matrix denotes the self and mutual inductances among the stationary
stator coils.

4.2.1.1  Determination of Self-inductances of Stator Coils

As far as self-inductances are concerned, they are non-negative terms. Since length of the air-gap is
not uniform, as the rotor rotates, the reluctance ‘seen’ by the coils changes, leading to a modulation
of the flux linking the coil, for a constant current (or mmf). This results in variation of the self-in-
ductance offered by the coil between a maximum and a minimum value. This is accounted in the
inductance expression by considering a first-order modulation with respect to an average value. This
has been demonstrated previously, in case of primitive machines. Based on this understanding, the
self-­inductance terms are stated below:
In order to understand the expression for self-inductances of stator coils, consider a diagram as
shown in Figure 4.3 where the magnetic axis is indicated for each of the three-phase windings, following
abc-phase-sequence. Note that this sequence is shown for the counter-clockwise rotation of the rotor.
The angular position of the rotor is measured taking the magnetic axis of coil a−a′, as the reference.

(Rotor position A)
q=0
Mag. axis of a-a′ q = −p /6
Mean negative
Mag. axis of b-b′

b′ c q = −p /3
Negative
N Mag. axis of b-b′
q = p /2
Mean negative
Mag. axis of c-c′
+
a a′

S
q = 2p /3 b q = −2p /3
Mag. axis of b-b′ Mag. axis of c-c′
c′

q = −5p /6
Mean negative
Mag. axis of a-a′

Figure 4.3  Three-phase Salient Pole Synchronous Machine with Magnetic Axis of Coils.

The magnetic axis of a coil is the axis along which it sets up flux for a given current in the coil. From
Figure 4.3, it is clear that the magnetic axes for phase-a, phase-b, and phase-c, are displaced symmetri-
cally in space, by an angle equal to 120° electrical, over a pair of poles, and are assumed to be fixed in
space. When the rotor is at position-A, which corresponds to the magnetic axis of phase-a, the phase-a

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 147 6/2/2018 2:50:26 PM

 148  |  Power System Analysis

coil sees a minimum reluctance and hence, establishes a maximum flux for a given current. This, in
turn, leads to a maximum self-inductance. This is accounted in the following expression for Laa:

y aa′
Laa = |(ib ,ic ,iF ,iH ,iG ,iK ) = 0
ia
= Ls + Lm cos 2q

Also, note that after half-a-revolution, the south-pole occupies the position-A. Hence, Laa reaches
its maximum, again showing double frequency variation in the self-inductance, over a pole pair. The
factor 2 along with the angle signifies this fact.
2p
When the rotor reaches a position corresponding to q = radian, the rotor polar-axis lies along
3
the magnetic axis of phase-b. This enables the phase-b coil to set up a maximum flux for a given cur-
rent ib. This occurrence is accommodated in Lbb by adding a delay angle as shown below:

y bb′
Lbb = |(ia ,ic ,iF ,iH ,iG ,iK ) = 0
ib
= Ls + Lm cos2(q − 2p /3)

Similarly, the self-inductance of phase-c is obtained as,

y cc ′
Lcc = = Ls + Lm cos 2(q + 2p /3)
ic

4.2.1.2  Determination of Mutual-inductances of Stator Coils

While obtaining the mutual terms, care is taken to see that the magnitude of the mutual inductance is maxim-
ised, depending on the rotor position. For ease of analysis, the mutual-inductances Lab, Lbc, and Lca are deter-
mined in the order, for abc-phase-sequence. Once these are obtained, the other elements can be evaluated.
Let us consider the mutual-inductance, Lab, between phases a and b. In an effort to determine the rotor
position for which the mutual flux, and hence, Lab, is a maximum, the following steps are employed:
1. Set the first term in Lab to an average value equal to −M, which is similar to that for a non-salient
pole machine.
2. Now, a modulation term is added to this average value M so that the net inductance value reaches
a maximum, depending on the rotor position. It is preferable to choose the ‘negative’ magnetic
axis of phase-b, relative to the magnetic axis of phase-a, to calculate fab, due to ib.
p
3. When the rotor is at position, q = − , the rotor polar-axis lies along the ‘negative’ magnetic axis
3
of phase-b, where the current ib is highly effective in setting up of flux, fb. However, a small part
of fb is directed towards the phase-a coil to link it.
4. Now, consider the rotor position at q = 0, where it lies along the magnetic axis of phase-a coil. For
this location of the rotor, the situation is just opposite to that explained in the previous point, i.e.,
the current ib is less effective in setting up of flux, fb, and most of the flux is directed towards the
phase-a coil.

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 148 6/2/2018 2:50:29 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  149 

5. Therefore, to maximise the mutual flux, fab, a rotor angle,

 −p 
0+
 3  p
q = =−
2 6

is chosen for constructing an expression for the mutual-inductance, Lab, as

y ab
Lab = = − ( M + Lm cos2(q + p /6))
ib

p
Note that when the rotor occupies an angular position, q = − , the value of Lab reaches a
6
maximum. This event repeats after half-a-revolution since at this instant, the south-pole would
p
come at q = − -position. This double frequency variation of the modulation (over a pole-pair)
6
is accounted by the factor 2 along with the angle, in Lab expression.
p
6. This, q = − -position, is denoted as the ‘mean negative’ magnetic axis of b−b′ in Figure 4.3.
6
Once the ‘mean negative’ magnetic axis of b−b′ is located, the ‘mean negative’ magnetic axis of
c−c′ can be identified (in order to determine Lbc), by setting
p
q =
2
Therefore, an expression for Lbc is obtained as,
y bc
Lbc = = − ( M + Lm cos2(q − p /2))
ic

p
Note that when the rotor is at position, q = , i.e., at the ‘mean negative’ magnetic axis of c−c′, the
2
Lbc attains a maximum value. Similarly, in order to determine Lca, the ‘mean negative’ magnetic axis
5p
of a−a′ is located at q = − . Therefore, Lca can be determined as
6

y ca
Lca = = − ( M + Lm cos2(q + 5p /6))
ia

Noting that Lba = Lab, Lcb = Lbc, Lac = Lca, the matrix [L11(q)](×3) is given by,

 Ls + Lm cos2q −( M + Lm cos2(q + p / 6)) −( M + Lm cos2(q + 5p / 6)) 

 
[ L11 (q )](3×3) =  −( M + Lm cos2(q + p / 6)) Ls + Lm cos2(q − 2p / 3) −( M + Lm cos2(q − p / 2)) 
 
 −( M + Lm cos2(q + 5p / 6)) −( M + Lm cos2(q − p / 2)) Ls + Lm cos2(q + 2p / 3) 

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 149 6/2/2018 2:50:33 PM

 150  |  Power System Analysis

Notes:
• Lm in self- and mutual inductances are identical as the magnetic circuit seen is the same for both
the inductances.
• In the above notations, the values Ls and M are assumed to be higher than that of Lm.

4.2.2 Determination of Mutual-inductances
Between Stator and Rotor Coils
Here, inductances demonstrate a sinusoidal variation over a pole-pair as the magnetic axes of the rotor
coils, themselves, move relative to the magnetic axes of stator coils depending on the rotor position.
Hence, the mutual flux-linkages between them change as a function of the rotor angle, q.

4.2.2.1  Mutual-inductances Between Stator and d-axis Rotor Coils

Since the magnetic axis of a−a′ is taken as the reference for angle measurement, for those rotor coils
whose magnetic axes coincide with the rotor polar-axis (or d-axis), a cosine function is used as follows:

y aF
LaF = = M F cos q
iF

y bF
LbF = = M F cos(q − 2p / 3)
iF

y cF
LcF = = M F cos(q + 2p / 3)
iF

y aH
LaH = = M H cosq
iH

y bH
LbH = = M H cos(q − 2p / 3)
iH

y cH
LcH = = M H cos(q + 2p / 3)
iH

Notes:
• In the above expressions, MF and MH denote the effective value of the mutual-inductances which
are derived as space-phasor quantities, since they represent fields distributed in the space.
• The value of the mutual-inductance between the stator and d-axis rotor coils is decided by cosine
2p
of the angle between their magnetic axes. For example, when the rotor is at position q = , the
3
magnetic axis of the field winding lies along the magnetic axis of b−b′, leading to a maximum
mutual flux-linkage between phase-b and F-coil. Therefore, LbF also reaches a maximum. A sim-
ilar observation can be made with respect to the remaining coils.

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 150 6/2/2018 2:50:36 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  151 

4.2.2.2  Mutual-inductances Between Stator and q-axis Rotor Coils

For those rotor coils, whose magnetic axes coincide with the quadrature-axis, a sine function is used
as follows:
y aG
LaG = = M G sin q
iG
y bG
LbG = = M G sin(q − 2p / 3)
iG
y cG
LcG = = M G sin(q + 2p / 3)
iG
y aK
LaK = = M K sinq
iK
y bK
LbK = = M K sin(q − 2p / 3)
iK
y cK
LcK = = M K sin(q + 2p / 3)
iK

Notes:
• In the above expressions, MG and MK denote the effective value of the mutual-inductances which
are derived as space-phasor quantities.
• The value of the mutual-inductance between the stator and q-axis rotor coils is decided by sine
p
of the angle between their magnetic axes. For example, when the rotor is at position q = , the
2
magnetic axis of the G- coil lies along the magnetic axis of a−a′, leading to a maximum mutual
flux-linkage between phase-a and G-coil. Therefore, LaG also reaches a maximum. A similar
observation can be made with respect to the remaining coils.
Therefore, matrix [L12(q )](3×4) is given by,

 M F cosq M H cos q M G sinq M K sinq 

 
[ L12 (q )](3× 4) =  M F cos(q − 2p / 3) M H cos(q − 2p / 3) M G sin(q − 2p / 3) M K sin(q − 2p / 3) 
 
 M F cos(q + 2p / 3) M H cos(q + 2p / 3) M G sin(q + 2p / 3) M K sin(q + 2p / 3) 

Also, note that [L21(q )](4×3) = [L12(q )]T(3×4) .

4.2.3  Determination of Self- and Mutual-inductances of Rotor Coils

Since these coils are located on the rotor itself, their self- and mutual-fluxes do not change with the
rotor position. Therefore, these inductances remain constant independent of q. The corresponding
inductance matrix is given by

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 151 6/2/2018 2:50:39 PM

 152  |  Power System Analysis

 LF LFH 0 0 
L LH 0 0 
= 
FH
[ L22 (q )]( 4 × 4)
0 0 LG LGK 
 
0 0 LGK LK 

In the above matrix, all elements are constant and are independent of q.

4.2.4  Voltage Equation Using the Generator Convention

In Figure 4.4, the current directions are reversed with respect to Figure 4.2, to account for the ‘genera-
tor convention’. In this notation, the new current, i, leaves the ‘dot’ (or positive terminal). The voltage
equation for the synchronous generator operation is obtained by using – i in (4.1). This is given by
dy
v = −[ R] i − (4.5)
dt
where
v( 7×1) = [va vb vc − v F 0 0 0]T

However, the flux linkage-current expression remains the same as in (4.2), but in terms of the new i.
y ( 7×1) = [ L(q )]( 7× 7) i ( 7×1)

Also, note that in (4.5), the polarity of the field voltage vF, is reversed to match the generator
convention so that in the vector notation, the field winding continues to receive power from an
external source.

Magnetic axis of a-a'

(Ref-axis)
c'
θ '
vcc q-axis
d-axis c
ic
K
ia G
a H
iF
vaa
'

ω
F
a' vF

b'
ib
'
vbb
b

Figure 4.4  Three-phase Synchronous Machine with Generator Convention.

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 152 6/2/2018 2:50:41 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  153 

4.2.5  Elements of Resistance Matrix

In (4.5), the elements of [R] matrix is given by

 Ra 0 0 0 0 0 0 
0 Ra 0 0 0 0 0 
 
0 0 Ra 0 0 0 0 
  [ Rs ](3×3) [0](3× 4) 
[ R]( 7× 7) = 0 0 0 RF 0 0 0 = (4.6)
[0]( 4 ×3) [ Rr ]( 4 × 4) 
0 0 0 0 RH 0 0  
 
0 0 0 0 0 RG 0 
0 0 0 0 0 0 RK 


In the above matrix, Ra denotes the effective resistance of each stator coil, RF-resistance of the field
winding, RH -resistance of the H-coil, RG -resistance of the G-coil, and RK -resistance of the K-coil.

4.3  F
 LUX LINKAGE-CURRENT EQUATIONS
IN ROTOR-REFERENCE FRAME
In order to transform the flux-linkage-current equation, (4.2) from the abc-frame to the rotor-refer-
ence-frame, in vector form, the transformation matrix is re-arranged in a compact form as follows:
Noting that the first three elements in y and i are the stator quantities which correspond to abc-
frame, a transformation matrix [B] is defined in terms of the [P] matrix such that it transforms only
these abc-quantities to the rotor-frame-of-reference without altering the remaining quantities. This is
given by
[ P ](3×3) [0](3× 4) 
[ B]( 7× 7) =  T 
[0] ( 4 ×3) [ I ]( 4 × 4) 

where [P] denotes the Park transformation, a time-varying and power-invariant transformation,
and is given by

 1 1 1 
 
 2 2 2 
2  2p   2p  
[ P ](3×3) =  cos(q ) cos q −  cos q + 
3  3   3 
 
  2p   2p  
sin(q ) sin q − 3  sin q +
 
3  


In the above matrix, q = (w0t + d ) denotes the electrical angle displacement between the stationary
magnetic axis of a−a′ coil and the magnetic axis of the field coil, i.e., d-axis. w0 represents the
rated speed (or the synchronous speed) of the rotor under steady-state. d signifies the angle between
the d-axis and the D-synchronous frame (Figure 4.5 (a)). However, for ease of analysis, as stated in
Chapter 3, the angle d is denoted between the q-axis and the Q-synchronous frame (Figure 4.5 (b)),

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 153 6/2/2018 2:50:43 PM

 154  |  Power System Analysis

Magnetic axis a-a' coil Magnetic axis a-a' coil

D ωo t (Stationary frame) D (Stationary frame)

q q
d δ d
θ δ
Q Q
θ ωo t

(a) D−Q: Synchronous frame (b)

d−q : Rotor frame

Figure 4.5  Relation Between D-Q and d-q Frames.

since it is customary to take q (or Q) axis as the reference for angle measurement. Further, note that

q (at t = 0) = d0 and dq = w 0 under steady-state.

dt
Now consider,

i B( 7×1) = [ B]( 7× 7) i ( 7×1) (4.7)

y B( 7×1) = [ B]( 7× 7) y ( 7×1) (4.8)

where
i B( 7×1) = [i0 id iq iF iH iG iK ]T

and
y B( 7×1) = [y 0 y d y q y F y H y G y K ]T

Notes:
• In (4.7) and (4.8), we can see that only the iabc and yabc quantities have been transformed to the
rotor frame, whereas the iFHGK and yFHGK quantities remain in the rotor frame.
• [B] is also an orthogonal matrix like [P] and hence,

[ P ]−1 [0](3× 4) 
[ B](−71× 7) =  (3×3) 
[0]( 4 ×3) [ I ]( 4 × 4) 

[ P ]T [0](3× 4) 
=  (3×3) T
 = [ B ]( 7× 7)
[0]( 4 ×3) [ I ]( 4 × 4) 

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 154 6/2/2018 2:50:47 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  155 

• It is customary to use [P1] = [P]–1 as the similarity transformation matrix. However, in this text,
[P] is used in the transformation. In this context, it can be shown that [P1] is the matrix of
right-eigenvectors of [L11(q )]3×3. Further, such transformation results in a diagonal matrix in the
0dq-frame which are the eigenvalues of [L11(q )]3×3 [7].
• The time-varying transformation allows us to imagine an observer sitting on the rotor which is
running at speed, wo . Under a steady-state condition, a balanced set of positive-sequence currents
iabc exciting a symmetrical abc-stator winding produces a revolving field whose speed relative
to the rotor is zero. With respect to the observer, this field appears stationary and enables us to
conceive two fictitious windings (on d- and q-axes) on the rotor, which produce the same field
effects by carrying DC currents equivalent to iabc. This visualisation facilitates us to construct
time-invariant system of equations which are equivalent to the original time-varying system of
equations.

4.3.1  Determination of [LB] Matrix

From (4.7) and (4.8), we get,

i ( 7×1) = [ B]T( 7× 7) i B( 7×1) (4.9)

and
y ( 7×1) = [ B]T( 7× 7) y B( 7×1) (4.10)

Note that the above equations provide the change of variable, i.e., the linear transformation, from
the abc frame to rotor frame. Now, using (4.9) and (4.10) in (4.2), we have,

[ B]T( 7× 7) y B( 7×1) = [ L(q )]( 7× 7) [ B]T( 7× 7) i B( 7×1) (4.11)

Pre-multiplying the above equation by [B](7×7), we have,

y B( 7×1) = [ B]( 7× 7) [ L(q )]( 7× 7) [ B]T( 7× 7) i B( 7 ×1) (4.12)

or
y B( 7×1) = [ LB ]( 7× 7) i B( 7×1) (4.13)

where
[ LB ]( 7× 7) = [ B]( 7× 7) [ L(q )]( 7× 7) [ B ]T( 7× 7)

[ P ] [ 0 ] [ L11 (q )] [ L12 (q )]  [ P ]T [ 0 ]
= T  [ L (q )] [ L (q )]  
[0] [ I ]  21 22  [0]T [ I ]

[ P ] [0] [ L11 (q )][ P ]T [ L12 (q )] 

= T  
[0] [ I ] [ L21 (q )][ P ]T [ L22 (q )]

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 155 6/2/2018 2:50:50 PM

 156  |  Power System Analysis

or
[ P ][ L11 (q )][ P ]T [ P ][ L12 (q )]
[ LB ]( 7× 7) = 
[ L21 (q )][ P ]T [ L22 (q )] 

Notes:

• ([ P ][ L12 (q )]) = [ L21 (q )][ P ]T

T

• [LB] is a symmetric matrix, i.e., [LB]T = [LB]. This property is due to the usage of the power-­
invariant Park transformation.
• For physical realisation of a system, i.e., to have an equivalent circuit representation of the
system, the parameter matrix must be symmetrical and positive definite (SPD).
• If a power-variant transformation is used, [LB] matrix will no longer be symmetrical. In such
cases, the [LB] matrix, when expressed in per unit, is made symmetrical by appropriately selecting
the base values [3].
Considering the term,

 L0 0 0 
T  
[ P ][ L11 (q )][ P ] = 0 Ld 0 
0 0 Lq 

where

L0 = Ls − 2 M
3
Ld = Ls + M + Lm
2
3
Lq = Ls + M − Lm
2
Notes:
• In the above calculations, L0 denotes the zero-sequence inductance, and Ld and Lq represent the
d- and q- axes synchronous inductances of the generator, respectively. These values include the
leakage inductance as well.
• The above evaluation involves a lengthy simplification procedure. One can use Symbolic Math
tool box (in MATLAB) for carrying out this evaluation.
Now consider,
 1 
 M F cos q z M F cosq z 2 M F cosq   cosq sinq 
   2 
 M H cosq z M H cosq z 2 M H cosq  2  1
[ L21 (q )][ P ]T =   z cosq z sinq 
 M G siinq z M G sinq z 2 M G sinq  3  2
 
   1
 M K sinq z M K sinq z 2 M K sinq  z 2 cosq z sinq 
2
 2 

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 156 6/2/2018 2:50:53 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  157 

 3 
0 MF 0 
 2 
 3 
0 MH 0 
2
[ L21 (q )][ P ]T =  
 3 
0 0 MG 
 2 
 3 
0 0 MK 
 2 

In the above equation, z cos q denotes cos(q – 2p /3) and z 2 cos q denotes cos(q + 2p /3). Similarly,
z sin q denotes sin(q – 2p /3) and z 2 sin q denotes sin(q + 2p /3).
The [LB] elements are given by

 L0 0 0 0 0 0 0 
0 Ld 0 M dF M dH 0 0 
 
0 0 Lq 0 0 M qG M qK 
 
LB = 0 M dF 0 LF LFH 0 0 
0 M dH 0 LFH LH 0 0 
 
0 0 M qG 0 0 LG LGK 
0 0 M qK 0 0 LGK LK 


where

3
M dF = MF
2
3
M dH = MH
2
3
M qG = MG
2
3
M qK = MK
2

Now, (4.13) can be rewritten separately for d- and q- axes as follows:

d-axis flux linkage equations:

y d   Ld M dF M dH  id 
    
y F  =  M dF LF LFH  iF  (4.14)
y H   M dH LFH LH  iH 

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 157 6/2/2018 2:50:55 PM

 158  |  Power System Analysis

q-axis flux linkage equations:

y q   Lq M qG M qK  iq 
    
y G  =  M qG LG LGK  iG  (4.15)
y K   M qK LGK LK  iK 

The above expressions simply depict the relationships between currents and flux-linkages. The gener-
ator or motor operation is decided by the respective voltage equations.

4.4  VOLTAGE EQUATIONS IN ROTOR-REFERENCE FRAME

Writing the voltage equations for the stator and rotor coils in the abc frame using the ‘generator
­convention’, we get
dy
v = −[ R] i − (4.16)
dt

where
v ( 7×1) = [va vb vc − vF 0 0 0]T

Now, using (4.9) and (4.10) and v ( 7×1) = [ B]T( 7× 7) v B ( 7 ×1) , we have

[ B ]T v B = −[ R][ B ]T i B −
d
dt
( )
[ B]T y B (4.17)

where
v B ( 7 ×1) = [v0 vd vq − vF 0 0 0]T

Pre-multiplying both sides of (4.17) by [B], we have,

v B = −[ B][ R][ B ]T i B − [ B]
d
dt
(
[ B ]T y B )
dy B d [ B]T
= −[ R] i B − [ B ][ B ]T − [ B] yB
dt dt
dy B d [ B]T
= −[ R] i B − − [ B] yB
dt dt

Note that from (4.6), [Rs] can be written as Ra × [ I ](3×3) . Hence, [ P ][ Rs ][ P ]T = [ Rs ] . Further, [Rr]
matrix is left unaltered. Therefore, [ B][ R][ B ]T = [ R] . The above expression can be rewritten as,

dy B d [ B]T dq
v B = −[ R] i B − − [ B] yB (4.18)
dt dq dt

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 158 6/2/2018 2:51:01 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  159 

d [ B]T
Consider the term [ B ] :
dq

 T 
d [ B]T [ P ] [ 0 ]  d [ P ] [0]
[ B] = T  dq
dq [0] [ I ]  T 
[0] [0]
 d [ P ]T 
[ P ] [0]
= dq  (4.19)
[0] T
[0]


 1 1 1 
 2 2 2   0 −sinq cosq 
d [ P ]T 2   
[P] = cosq cos(q − 120) cos(q + 120)   0 −sin(q − 120) cos(q − 120) 
dq 3
sinq sin(q − 120) siin(q + 120)   0 −sin(q + 120) cos(q + 120) 
 
 
0 0 0 
 
= 0 0 1  (4.20)
0 −1 0 

Using (4.20) in (4.19), we have,
0 0 0 
0 0 1 [0](3× 4) 
d
[ B] [ B] = 
T 
dq 0 −1 0  (4.21)
 
 [0]( 4 ×3) [0]( 4 × 4) 

Also note that
dq
= w (4.22)
dt

Substituting (4.21) and (4.22) in (4.18), we get,

0 
y 
 q 
 −y d 
  dy B
v B = −[ R] i B − 0  w − (4.23)
dt
0 
 
0 
0 
 

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 159 6/2/2018 2:51:04 PM

 160  |  Power System Analysis

Expanding the above equation, we can write:

dy 0
v0 = − R0 i0 − (4.24)
dt
dy d
vd = − Raid − wy q − (4.25)
dt
dy q
vq = − Raiq + wy d − (4.26)
dt

Also,
dy F
−v F = − RF iF − (4.27)
dt
dy H
0 = − RH iH − (4.28)
dt
dy
0 = − RG iG − G (4.29)
dt
dy K
0 = − RK iK − (4.30)
dt

From the above equations, it can be seen that abc- windings get transformed into a set of fictitious
d- and q- windings on the rotor (Figure 4.6) and a zero-sequence network with R0 = Ra.

Magnetic axis of a-a'

(Ref-axis) iq
di

vq
d
v

K q-axis
d-axis
G
H
iF

F
ω
vF

Figure 4.6  Representation of a Three-phase Synchronous Machine in the Rotor-frame

(0-winding is not Shown).

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 160 6/2/2018 2:51:07 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  161 

4.5  EXPRESSION FOR TORQUE IN ROTOR-REFERENCE FRAME

We know that
P 1 T d [ L(q )]
Teg = − i i (4.31)
22 dq

From (4.9), we have

i = [ B]T i B
(4.32)

and from the expression for [ LB ] = [ B][ L(q )][ B]T , we get [ L(q )] as

[ L(q )] = [ B]T [ LB ][ B ] (4.33)

Using (4.32) and (4.33) in (4.31), we get,

Teg = −
P
2
1
2
i TB [ B]
d
dq
(
[ B]T [ LB ][ B] ) [ B]T i B
(4.34)

Consider the term

d
dq
(
[ B]T [ LB ][ B] : )
d [ B]T

d
dq
(
[ B]T [ LB ][ B] =
dq
) [ LB ][ B ] + [ B ]T
d [ LB ]
dq
[ B ] + [ B ]T [ LB ]
d[ B]
dq
d [ B]T d[ B]
= [ LB ][ B ] + [ B ]T [ LB ]
dq dq

d
Note that [ LB ] = 0 .
dq
Hence (4.34) reduces to,

P 1  T d [ B]T T T T d[ B] T 
Teg = −  i B [ B] dq [ LB ][ B ][ B ] i B + i B [ B ][ B ] [ LB ] dq [ B] i B  (4.35)
2 2  

or
P 1  T d [ B]T d[ B] T 
Teg = − i
 B [ B ] [ LB ] i B + i TB [ LB ] [ B] i B  (4.36)
2 2  dq dq 

Note that both the terms in (4.36) are scalars and we can see that
T
 T d[ B] T  d [ B]T d [ B]T
 i B [ LB ] [ B] i B  = i TB [ B] [ LB ]T i B = i TB [ B] [ LB ] i B (4.37)
dq  dq dq

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 161 6/2/2018 2:51:13 PM

 162  |  Power System Analysis

Using (4.37) in (4.36), we get,

P  T d [ B]T 
Teg = −  i B [ B] dq [ LB ] i B  (4.38)
2  

Since y B = [ LB ] i B , we have,

P  T d [ B]T 
Teg = − i
 B [ B ] y B
2  dq 
(4.39)
Using the result given in (4.21), we get,

Teg = −
P
2
(
idy q − iqy d )
or

Teg =
P
2
( )
y d iq −y q id (4.40)

where P represents the number of poles.

4.6  CASE STUDIES

4.6.1  Voltage Build-up on Open-circuit
The generator speed is assumed to be constant at its nominal value i.e., w = w0. Damper winding
dynamics are also neglected in the study. This implies that [2]

y H = 0, y G = 0, and y K = 0

Hence, the equations (4.28) to (4.30) need not be considered in our study. Further, since the generator
is on open-circuit, iabc = 0 which implies that, i0dq = 0.
Therefore, (4.25) to (4.27) reduce to
dy d
vd = −w 0y q − (4.41)
dt
dy q
vq = w 0y d − (4.42)
dt
dy F
v F = iF RF + (4.43)
dt
v0 = 0, since the generator is symmetric and generates a balanced set of voltages.
Also, from (4.14) and (4.15), we have,

y d = M dF iF (4.44)

y F = LF iF (4.45)

y q = 0 (4.46)

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 162 6/2/2018 2:51:18 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  163 

Using the above relations, we have,

dy d di
vd = − = − M dF F (4.47)
dt dt
vq = w 0y d = w 0 M dF iF (4.48)

diF
v F = iF RF + LF
dt (4.49)
Assuming, iF (0) = 0, with vF(t) as a step input of value v0F, and solving for iF (t) we have,
 − ′ 
t
v F0  Tdo 
i F (t ) = 1− e (4.50)
RF  
 

′ L
where, Tdo = F is referred to as d-axis transient open-circuit time constant.
RF

Notes:
• T ′d0 is effective under open-circuit condition of the machine and is valid during the period
when the H coil influence is negligible. This duration is designated as the transient period of the
machine.
• Since T ′d0 relates machine’s primitive parameters, LF and RF, it leads to a way to modify the
machine equations in terms of hybrid parameters (also known as the standard parameters) so as
to make them usable.
Also from (4.50) we have,
t
diF v F0 − Tdo′
= e (4.51)
dt LF

Substituting (4.51) in (4.47) and (4.50) in (4.48), we get,
t
v F0 − Tdo′
vd = − M dF e (4.52)
LF

 − ′ 
t
v F0  Tdo 
vq = w 0 M dF 1− e (4.53)
RF  
 

To evaluate the voltage in abc frame, we have
vabc = [ P ]T vodq
(4.54)
Using (4.52) and (4.53) in (4.54) and evaluating va(t), we get
 t    − ′ 
t 
2  v F0 − Tdo′   v0 1 − e Tdo   sinq  (4.55)
va ( t ) =  − M dF e  cosq + w 0 M dF F
3  LF   RF   
    

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 163 6/2/2018 2:51:22 PM

 164  |  Power System Analysis

 v0 
Since the numerical value of the term  M dF F  is relatively small when compared to that of the
 LF 
 v F0 
term  w 0 M dF  , an approximate expression for va(t) is given by
 RF 
  − ′ 
t
 2 v0 1 − e do   sin q (4.56)
T
va (t ) =  w 0 M dF F
 3 RF  
  

The steady state expression for va(t) is
 2 v0 
va (t ) =  w 0 M dF F  sinq (4.57)
 3 RF 

 v0 
Note that the term  w 0 M dF F  represent the line-to-line RMS value.
 RF 

Notes: From the above observations, we can infer the following:

• The actual voltage applied to the field, vF, is referred to the stator q-winding as per the following
relationship and is denoted as Efd:

w M 
E fd =  0 dF  v F (4.58)
 RF 
0 0
Under steady-state, the value of Efd indicates the open-circuit line-to-line voltage, i.e., E fd = VLL .
• The above expression implies that if vF value at actuals is not desired in the analysis, one need not
define any base value for the rotor circuits.
• A voltage proportional to iF is denoted in the stator side as Ea and is given by

Ea = (w 0 M dF )iF (4.59)

• Under steady state, E0fd = E0a.

4.6.1.1  Voltage Build-up Result for a Lab Synchronous Generator

In Figure 4.7, the building up of voltage, va(t), is shown for a DC-motor driven synchronous generator
in the lab [10], following an application of DC voltage to its field winding. Once the field voltage is
applied suddenly, va(t) builds up almost exponentially in a time duration of approximately 180 ms.
This period is largely decided by T′d0 of the machine. Figure 4.7 also depicts the voltage induced due
to a residual field in the machine.

4.6.2  Steady-state Short-circuit

Initially, the generator is assumed to be on steady-state open-circuit condition with a constant rotor
v F0
speed correspond to w0, and a constant v0F and iF0 = . A three-phase symmetrical short-circuit
RF

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 164 6/2/2018 2:51:27 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  165 

0.5
5T' = 180 ms
do
0.4

0.3

0.2

0.1
v(t) in pu

−0.1

−0.2

−0.3

−0.4

−0.5

2.5 2.6 2.7 2.8 2.9 3 3.1

Time (s)

Figure 4.7  Open-circuit Voltage Build Up for a Lab Generator.

is then applied at the generator terminals and the generator is assumed reach the steady-state short-­
circuit condition. Under this condition, we have vabc = 0 and hence v0dq = 0.
Also, in steady-state we have,

dy d dy q dy F
= 0, = 0 and =0
dt dt dt

Hence, equations (4.25) to (4.27) reduce to

0 = − Raid − w 0y q (4.60)

0 = − Raiq + w 0y d (4.61)

v F = RF iF (4.62)

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 165 6/2/2018 2:51:29 PM

 166  |  Power System Analysis

Also from (4.14) and (4.15) we have,

y d = Ld id + M dF iF (4.63)

y F = M dF id + LF iF (4.64)

y q = Lq iq (4.65)

Substituting (4.65) in (4.60), we get,

Raid + w 0 Lq iq = 0 (4.66)

Similarly, substituting (4.63) in (4.61), we have,

Raiq − w 0 Ld id = w 0 M dF iF (4.67)

0
Writing the above equations in the matrix form with iF = iF , we have,

 Ra w 0 Lq  id  0 
 −w L    = 0  (4.68)
 0 d i
Ra   q  w 0 M dF iF 

Solving for id and iq, we get,

w 02 M dF Lq iF0
id = − (4.69)
Ra2 + w 02 Lq Ld

Raw 0 M dF iF0
iq = (4.70)
Ra2 + w 02 Lq Ld

If Ra = 0 is assumed, then

M dF iF0
id = −
Ld
(4.71)
iq = 0

For a balanced condition, in the abc-frame, ia(t) can be written as,

2
ia (t ) = id cosq (4.72)
3

2 M dF iF0
ia (t ) = − cosq (4.73)
3 Ld

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 166 6/2/2018 2:51:36 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  167 

Writing the above equation in terms of reactance xd = Ldw0, we get,

2 w 0 M dF iF0
ia = − cosq (4.74)
3 xd

Notes:
• With Ra = 0, the iq component is zero and the short-circuit current is limited only by the d-axis
synchronous reactance, xd.
• From (4.57) and (4.74), it is clear that steady-state short-circuit current ia(t) lags va(t) by 90º.

4.6.3  Natural Response Under Short-circuit

This section deals with the electrical circuit dynamics of the synchronous machine following the
occurrence of short-circuit when the machine is on open-circuit. Like in the steady-state short-­circuit
condition, initially, the generator is assumed to be on steady-state open-circuit condition with a con-
0 v F0
stant rotor speed correspond to w0, and a constant v0F and iF = . A three-phase symmetrical
RF
short-circuit is then applied at the generator terminals at t = 0. In this case, for ease of analysis, all
damper winding dynamics are neglected.
Under a symmetrical three-phase short-circuit, we have, vabc = 0 which implies that v0dq = 0.
From (4.14) and (4.15), we have,

y d = Ld id + M dF iF (4.75)

y F = M dF id + LF iF (4.76)

y q = Lq iq
(4.77)
Making use of the above equations, (4.25) to (4.27) reduces to
did di
0 = − Raid - w 0 Lq iq − Ld − M dF F (4.78)
dt dt
di di
v F = iF RF + M dF d + LF F (4.79)
dt dt
diq
0 = − Raiq + w 0 Ld id + w 0 M dF iF − Lq (4.80)
dt

Writing the above equations in a matrix form, we get,

 did 
 
 Ld M dF 0   dt   − Ra 0 −w 0 Lq  id  0 
  di  = 0    
 M dF LF 0  F − RF 0  iF  + 1  v F (4.81)
 dt  
0
 0 Lq   di  w 0 Ld w 0 M dF − Ra  iq  0 
 q 
 dt 

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 167 6/2/2018 2:51:40 PM

 168  |  Power System Analysis

Let us choose the state-variables as x = [id , iF , iq ]T and writing the above equations in the standard
state-space model, we get,

x = [ A] x + [ B]u
(4.82)
Let us define,

 Ld M dF 0   − Ra 0 −w 0 Lq 
   
[ L] =  M dF LF 0  [ A1 ] = 0 − RF 0 
0 0 Lq  w 0 Ld w 0 M dF − Ra 

Consider,

 LF M dF 
 − 0 
 ∆ ∆ 
−1  M dF Ld 
[ L] =  − 0  (4.83)
∆ ∆
 
0 1 
0
 Lq 

with
2
∆ = Ld LF − M dF

 M2 
= LF  Ld − dF 
 LF 

′
 2 
M dF
Denoting Ld =  Ld −  we have
 LF 

∆ = L′d LF
(4.84)
−1
Therefore, [ L] simplifies to

 1 M dF 
 ′ − 0 
 Ld L′d LF 
 M Ld 
[ L]−1 =  − ′ dF 0  (4.85)
 Ld LF L′d LF 
 
0 1 
0
 Lq 

Assuming that Ra = 0 and RF = 0 in (4.82), the matrix [ A] = [ L]−1[ A1 ] , and is given by

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 168 6/2/2018 2:51:46 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  169 

 w 0 Lq 
0 0 − 
 L′d 
 M dF w 0 Lq 
[ A] = 0 0 
 L′d LF 
 
 w 0 Ld w 0 M dF
0 
 Lq Lq 
 

Since we are interested in the natural response of the synchronous machine under the short-circuited
condition, we set vF = 0 and obtain the solution of x = [id , iF , iq ]T for the initial condition on the
state-variables at t = 0, i.e., x (0) = [0 iF0 0]T . Thus, (4.82) reduces to

x = [ A] x

For above equation, the natural response is governed by

x (t ) = e At x (0)

The state transition matrix eAt is given by

e At = Lp −1{( p[ I ] − [ A]) −1}

where Lp−1 denotes the equivalent Laplace-inverse operation.

 M2 w2 w 02 M df w 0 Lq 
 p2 − dF 0 − − p 
 L′d LF L′d L′d 
 
1  M dF w 02 Ld w 02 Ld w 0 Lq M dF 
( p[ I ] − [ A]) −1 = p2 + p
∆1  L′d LF L′d L′d LF
 
 w 0 Ld w 0 M dF 
 p p p2 
 Lq Lq 

where ∆1 = p( p2 + w 02 ) , the roots which are p = 0, ± jw 0 .

Noting that to determine x(t), since x (0) = [0 iF0 0]T , it is sufficient to perform the Laplace-
inverse operation of only a few elements in the above matrix as given below. Thus we have,
 w 02 M dF iF0 
− ′ 2 2 
 Ld p( p + w 0 ) 
 
 L  iF0
x ( p) =  p2 + w 02 ′d  2

2  (4.86)
 Ld  p( p + w 0 ) 
 pw M iF0 
 0 dF 
 Lq p( p2 + w 02 ) 

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 169 6/2/2018 2:51:51 PM

 170  |  Power System Analysis

Expression for id (t):

From (4.86), id (p) is given by
w 02 M dF iF0
id ( p) = −
L′d p( p2 + w 02 )

Let us define
1 K1 K2 K3
G ( p) = 2 2
= + +
p( p + w 0 ) p p − jw 0 p + jw 0

where
1
K1 = G( p) p | p = 0 =
w 02
1
K 2 = G( p)( p − jw 0 ) | p = jw 0 = −
2w 02
1
K3 = G( p)( p + jw 0 ) | p =− jw 0 = −
2w 02

After substituting the above constants in G(p) and then in id(p) and finally solving for id(t), we have,

w 02 M dF iF0  1 1 jw 0t 1 − jw 0t 
id (t ) = − ′  2− 2
e − e 
Ld  w 0 2w 0 2w 02 

M dF iF0  e jw 0t + e − jw 0t 
=− 1 − 
L′d  2 

M dF iF0
id (t ) = − (1 − cos w 0t )
L′d (4.87)
Similarly, solving for iF(t) and iq(t), we have,

L  L  
iF (t ) =  ′d + 1 − ′d  cosw 0 t  iF0 (4.88)
 Ld  Ld  

M dF iF0
iq (t ) = sinw 0 t (4.89)
Lq

4.6.3.1  Example 4.1: Loss-less Synchronous


Machine
 Ld  Ld  0
Let Ld = 4.98 mH, Lq = 4.84 mH, MdFiF=(t49  ′ L+F1=−577
) =mH, mH,
′  0 t  i F =10 A (say) [4]. Hence,
coswand
L’d = 0.8188 mH. Then for w0 = 376.99 rad/s, from  (4.88)
 Ld(4.87), Ld  and (4.89),
 we get,

id (t ) = −598.42 (1 − cosw 0 t ) (4.90)

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 170 6/2/2018 2:51:56 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  171 

iF (t ) = (60.82 − 50.82 cos w 0 t ) (4.91)

iq (t ) = 101.24 sin w 0 t (4.92)

For a loss-less system, the DC component in id and iF and fundamental frequency component in id, iF,
and iq are sustained forever. If we calculate the waveform for ia(t) in the abc-frame, we get,

ia (t ) =
2
3
{ }
id (t )cosq + iq (t )sinq (4.93)

Note that i0 is zero as the system is symmetric. Using (4.90) and (4.92) in the above expression, we
get ia(t) in A as,

2
ia (t ) = [ −598.42cosq + 598.42cos(w 0 t )cosq + 101.24sin(w 0 t )sinq ] (4.94)
3
The above expression represents the phase-a current in a case where a three-phase symmetrical
short-circuit is applied at the terminals of an unloaded generator. In this case, the absolute position
of the rotor indicated by q is set as w 0 t + g . Here, g denotes the fault-inception angle on the phase-a
voltage wave at which the fault is applied. This also indicates the position of the rotor relative to the
magnetic axis of phase-a coil at the time of application of the fault, i.e., at t = 0. Now, using the trig-
onometric relationships
1
cosAcosB = [ cos( A + B ) + cos( A − B ) ]
2

and
1
sinAsinB = [cos( A − B) − cos( A + B)] ,
2

we can rewrite (4.94) as

2 598.42
ia (t ) = 
3
−598.42cos(w 0 t + g ) +
2
[cos(2w 0t + g ) + cosg ] +
101.24 
+ [ cosg − cos(2w 0 t + g ) ]
2 
2
= [ −598.42cos(w 0 t + g ) + 349.83cosg + 248.59cos(2w 0 t + g )] (4.95)
3

The above equation shows that ia(t) consists of fundamental frequency component, DC component,
and double the fundamental frequency components. From (4.91), we can see that the fundamental fre-
quency component gets injected into the field current which is, otherwise, a pure DC component under
steady-state. Such a frequency component produces a pulsating air-gap field which can be resolved
into two fundamental frequency revolving fields of equal amplitude. Of these two revolving fields, one
revolves in the forward direction and the other revolves in the backward direction with respect to the
rotor mass. When the rotor is running at synchronous speed, the stator circulates a double frequency
current due to the forward-field and a DC current due to the backward-field. Therefore, it can be

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 171 6/2/2018 2:55:49 PM

 172  |  Power System Analysis

concluded that a DC and a double frequency component in the currents in the abc-frame are due to the
fundamental frequency components in the rotor windings. This component decays to zero with time
in a lossy machine which implies decaying DC and double frequency components in the abc-frame.

4.6.3.2  Example 4.2: Loss-less Synchronous Machine- Flux Linkages

T
Assuming that Ra = 0 and RF = 0, we can determine an expression for [y d y F y q ] using the
relationship given by

y d (t )   Ld M dF 0  id (t ) 
    
y F (t )  =  M dF LF 0  i F (t ) 
y q (t )  0 0 Lq  iq (t ) 
  

Using the results given in (4.90), (4.91), and (4.92), we can see that

y d (t ) = 0.49 cos w 0 t (4.96)

y F (t ) = 5.77 (4.97)

y q (t ) = 0.49 sin w 0 t (4.98)

To understand the effect of either armature resistance and/or field resistance on the damping perfor-
mance of machine currents, the following cases are presented:

4.6.3.3 Example 4.3: Lossy Synchronous Machine with a Non-zero Ra

In Example 4.1, in addition to other parameters, if Ra = 0.0031Ω and RF = 0 , the expression for
id(t), iF(t), and iq(t), becomes

id (t ) = −598.4108 + (598.4211)e −2.2132t cos(376.99t − 0.3364)

iF (t ) = 60.8182 + (50.8191)e −2.2132t cos(376.99t + 179.6636)

iq (t ) = 1.0167 + (101.2397)e −2.2132t cos(376.99t − 90.5754)

4.6.3.4 Example 4.4: Lossy Synchronous Machine

with a Non-zero Ra and RF
In Example 4.1, in addition to other parameters, if Ra = 0.0031 Ω and RF = 0.0715 Ω , the expres-
sion for id(t), iF(t), and iq(t), becomes

id (t ) = ( −598.4326)e −0.7536t + (598.4379)e −2.2132t cos(376.98t − 0.2407)

iF (t ) = (60.8203)e −0.7536t + (50.8206)e −2.2132t cos(376.98t + 179.7782)

iq (t ) = (1.0167)e −0.7536t + (101.2426)e −2.2132t cos(376.98t − 90.5754)

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 172 6/2/2018 2:55:54 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  173 

Notes:
• In Example 4.4, when RF is made non-zero (in addition to Ra), it can be seen that it causes the DC
component to decay without altering the decrement factor (–2.2132) associated with the oscilla-
tory term. In Example 4.3, when Ra alone is made non-zero it added a decaying term only to the
oscillatory component which is absent in Example-4.1. This clearly shows that Ra offers damping
only to the oscillatory term.
′ LF 0.577
• For the Tdo = = = 8.0699 s. In addition to the definition of T ′d0, we can also define
RF 0.0715
T ′d referred to as d-axis transient short-circuit time constant. This is related to T ′d0 as
L′d
Td′ = Tdo
′
(4.99)
Ld
where L’d is referred to as d-axis transient inductance.
 0.8188 × 10 −3 
For example, Td′ = 8.0699   = 1.3268 s. Note that the decrement factor associated
 4.98 × 10 −3 

with the DC component is equal to 1 = 0.75367 neper/s. Therefore, the DC component in id(t) and
Td′
iF(t) and hence the fundamental frequency component in ia(t) decay with a time-constant T ′d during
the transient period before steady-state is reached.

4.6.3.5  Example 4.5: Synchronous Machine with a Non-zero Ra, RF, and vF
In this case, a forced response is obtained with Ra = 0.0031Ω , RF = 0.0715Ω and vF is set to
v F0 = iF0 × RF = 0.715 V. The state-equation given in (4.82) is solved numerically using RK-4th method
with step-size = 0.1 ms. The initial condition of the state is x (0) = [0 iF0 0]T = [0 10 0]T A.
Various matrices used on the state-mode are given below:

 Ld M dF 0   − Ra 0 −w 0 Lq 
   
[ L] =  M dF LF 0  [ A1 ] = 0 − RF 0 
0 0 Lq  w 0 Ld w 0 M dF − Ra 


and

0 
−1  
[ A] = [ L] [ A1 ] and [ B] = [ L]−1 v F0 
0 
 
where v0F is accounted as a constant in [B] matrix.
It is to be noted that short-circuit is applied on an unloaded generator with v0F maintained for
the field and the initial value of field current is 10 A. Once the fault is applied at t = 0, the fault is
not removed. The waveforms for id(t), iF(t), and iq(t) are obtained as shown in Figure 4.8. Once the

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 173 6/2/2018 2:55:58 PM

 174  |  Power System Analysis

0
−200
−400
id (A )

−600
−800
−1000
0 1 2 3 4 5 6
Time (s )

100
80
if (A )

60
40
20
0
0 1 2 3 4 5 6
Time (s )

100

50
iq (A )

−50

−100
0 1 2 3 4 5 6
Time (s )
Figure 4.8  The idFq(t) Plots for a Short-circuited Un-loaded Generator.

steady-state is reached idFq(t) becomes constant and is obtained as idFqS = –inv([A])*[B]. Therefore,
idFq (t ) = [ −98.3933, 10, 0.1671]T A.
An expanded view of the idFq plot is shown in Figure 4.9. It is clear that DC and the fundamental
components in idFq(t) decay as time progresses relative to the fault instant.
Assuming that the fault is applied at the positive zero crossing of the va waveform (which is
obtained by setting g = 0), the idFq(t) quantities are transformed into abc-frame using the Park-inverse
transformation. The resulting iabc(t) are depicted in Figure 4.10.
An expanded version of the iabc(t) plots are shown in Figure 4.11. It is apparent that decaying DC
and second harmonic components are present in iabc(t) in addition to the fundamental frequency
component. Once the steady-state is eventually reached, iabc(t) will have only fundamental sinor
components with peak value equal to 80.559 A (at the end of 10 s) dominantly decided by id, i.e.,
 2
I sm =   × 98.664 = 80.559A.
 3

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 174 6/2/2018 2:56:00 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  175 

−500
id (A )

−1000

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Time (s )

100
if (A )

50

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (s )

100

50
iq (A )

−50

−100
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (s )

Figure 4.9  The idFq(t) Plots for a Short-circuited Un-loaded Generator (Expanded View).

4.6.3.6  Determination of the Fundamental Frequency Component of ia(t)

Here, a procedure is outlined to obtain the fundamental frequency component from a measured phase
current oscillogram [8]. For this purpose, the ia(t) waveform shown in Figure 4.10, is employed.
1. From the ia(t) waveform obtain all the peaks pertaining to both positive and negative peaks. This
provides a positive envelope, iaP+ and a negative envelope, iaP– as shown in Figure 4.12.
2. Using the positive and negative peak envelopes of the waveform, the positive peak envelope of the
fundamental frequency component is determined as

(iaP + − iaP − )
iacP = (4.100)
2

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 175 6/2/2018 2:56:01 PM

 176  |  Power System Analysis

1000
800
600
ia (A )

400
200
0
−200
0 1 2 3 4 5 6
Time (s )

400
200
0
ib (A )

−200
−400
−600
−800
0 1 2 3 4 5 6
Time (s )

400
200
0
ic (A )

−200
−400
−600
−800
0 1 2 3 4 5 6
Time (s )

Figure 4.10  The iabc(t) Plots for a Short-circuited Un-loaded Generator.

The above process separates the fundamental frequency component in ia(t), removing the DC compo-
nent. This calculation assumes that the second harmonic component in ia(t) is negligible. The positive
peak envelope of only the fundamental frequency component, thus obtained, is depicted in the third
subplot of Figure 4.12. A negative peak envelope is also drawn in the fourth subplot as –(iacP).
In Figure 4.13, the envelopes of the fundamental frequency peaks in ia(t), i.e., both the (iacP) and
–(iacP) envelopes, are plotted on the ia(t) waveform. Figure 4.13 clearly shows that at later part of
cycles, i.e., after t > 2.5 s, the envelopes coincide with the peaks of ia(t), denoting that the envelope
eventually tends to a steady-state where the peak value of ia(t) becomes 80.559 A. However, immedi-
ately after the occurrence of fault, the fundamental frequency peak values (represented by the dotted
envelopes) are quite large and decay with time-constant T ′d until the steady-state is reached. Therefore,
when the fundamental frequency component alone is considered in ia(t), (which will be the case when
the fundamental frequency component in idFq(t) is neglected), we can infer that the machine, to begin
with, offers a low inductance and once the steady-state is reached, its inductance changes to a new
high value. The low inductance is referred to as transient inductance and the steady-state inductance

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 176 6/2/2018 2:56:02 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  177 

1000
800
600
ia (A )

400
200
0
−200
0 0.05 0.1 0.15 0.2 0.25 0.3
Time (s )

400
200
0
ib (A )

−200
−400
−600
−800
0 0.05 0.1 0.15 0.2 0.25 0.3
Time (s )

400
200
0
ic (A )

−200
−400
−600
−800
0 0.05 0.1 0.15 0.2 0.25 0.3
Time (s )

Figure 4.11  The iabc(t) Plots for a Short-circuited Un-loaded Generator (Expanded View).
is termed as synchronous inductance. Therefore, a synchronous generator is deemed to offer different
inductances at different time-frames unlike in an RL series circuit (Figure 4.14).
A sinusoidal voltage, va (t ) = Vm sin(w t + g ) is applied to an RL series circuit with R = 0.1 Ω and
L = 0.01 H. If ia(0) is the initial value of the inductor current then the expression for ia(t) for t ≥ 0 is
given by

 t  t
V V −  − 
ia (t ) = m sin(w t + g − f ) − m sin(g − f )e  t  + ia (0)e  t  (4.101)
|Z| |Z|

L
where Z = ( R + jw L) =| Z | ∠f and t = .
R
For g = 0, Vm = 10V, w = 2p 60 rad/s and ia(0) = 0, the ia(t) plot is shown in Figure 4.14. Here, the
arrow lengths clearly demonstrate that the amplitude of the sinusoidal component in ia(t) does not
change (except a DC shift at the initial part of the waveform).

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 177 6/2/2018 2:56:04 PM

 178  |  Power System Analysis

1000
iaP+

500

0
0 1 2 3 4 5 6 7 8 9 10

−100
iaP−

−200
0 1 2 3 4 5 6 7 8 9 10

400
iacP

200

0
0 1 2 3 4 5 6 7 8 9 10

500
iacP+−

−500
0 1 2 3 4 5 6 7 8 9 10
Time in seconds

Figure 4.12  The Peak Envelopes for ia(t) Plot.

4.6.3.7  Determination of the d-axis Standard Parameters Using ia(t)

Since the basic machine parameters such as MdF, LF, and RF cannot be easily determined, these basic
parameters are expressed in terms of experimentally determinable parameters referred to as hybrid
parameters. These hybrid parameters are also referred to as standard parameters. For the case in
hand, these are d-axis- T ′d0, T ′d, L’d, Ld, and q-axis- Lq. Further, note that the d-axis parameters are
related by an expression given in (4.99).
In the following lines, a procedure is described to estimate the d-axis parameters from the recorded
oscillogram of ia(t). However, it is to be noted that the q-axis inductance cannot be estimated easily
using ia(t), which is the short-circuit current of an unloaded generator. This is because on the appli-
cation of a short-circuit on the generator, the q-axis winding does not receive appreciable excitation
since on no-load and on reaching steady-state open-circuit condition (i.e., before the application of a
fault), the entire induced voltage is due to the d-axis component and the q-axis induced voltage is neg-
ligibly small. Therefore, in order to determine the d- and q-axes parameters accurately, an elaborate
test procedure known as standstill frequency response test has to be conducted [9, 3].

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 178 6/2/2018 2:56:05 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  179 

1000

i (t)
a

500
ia and iacP+−

iac +ve and −ve peaks

−500

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Time in seconds

Figure 4.13  The iacP Peak Envelopes and ia(t) Plot.

The steps involved to estimate the d-axis parameters using the short-circuit currents of an un-loaded
generator are as follows:
1. The positive peak envelope pertaining to the fundamental frequency component in ia(t) is obtained
employing (4.100). This is given in the third subplot in Figure 4.12. This is denoted as iacP(t).
2. From the iacP(t) plot, note the steady-state value once the wave settles to a constant. This repre-
sents the steady-state positive peak value of ia(t) and the value is denoted as Ism = 80.559 A.
3. Now, from the iacP plot, subtract Ism to obtain the transient part of the fundamental frequency
component peaks. This is denoted as iP(t) and is shown in the first subplot in Figure 4.15. From
the earlier analysis, it is clear that the decay time constant of this wave is largely due to T ′d.
4. The ip(t) data points are plotted in a semilog-Y plot. This is because we know that when an expo-
nentially decaying waveform with a dominant time constant (in a linear scale) is plotted using the
semilog-scale (to the base-10), it appears as a linear plot. This facilitates easy fitment of a slope to
determine the time constant of decay. In the current case, since only one dominant time constant

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 179 6/2/2018 2:56:06 PM

 180  |  Power System Analysis

2
ia (A )

−1

−2

−3
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
Time (s )

Figure 4.14  ia(t) Plot for an RL Series Circuit with Sinor Voltage Switching.

is involved, the resulting line (in the semilog-scale) is a prefect straight-line. In Figure 4.15, the
second subplot depicts the desired plot.
  It is noted that the first peak or the maximum value of the transient component ip(t) is Ipm =
440.7 A. The time instant t = t1 at which the ip(t) amplitude becomes equal to 0.3679 (= e–1) times
the first peak, i.e., ip(t1) = 0.3679 × 440.7 = 162.13 A, denotes the estimate of the time constant
T ′d. From Figure 4.15, it can be seen that this point occurs at t1 = T ′d = 1.225 s.
5. Just prior to the short-circuit, the phase-a peak voltage is given by

2
Vm = w 0 M dF iF0 = 150.8279 V (4.102)
3

The synchronous inductance is obtained as

1 Vm 1 150.8279
Ld = = = 4.966 mH (4.103)
w 0 I sm w 0 80.559

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 180 6/2/2018 2:56:07 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  181 

400
ip transient above lsm (A)

300

200

100

0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

3
10
ip transient (semilogY)

2
10 X: 1.225
Y: 162

1
10

0
10
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time in seconds

Figure 4.15  iP(t) Plot for Estimating T ′d.

The transient inductance is determined as

1 Vm 1 Vm
L′d = = = 0.7675 mH (4.104)
w 0 i( acPm) w 0 ( 440.7 + 80.559)

where i(acPm)denotes the value of the first peak or the maximum of iacP(t).
The specified value of Ld is 4.98 mH and L ′d is 0.8188 mH. Using the relationship given in (4.99),
we can obtain T ′d0 as

′
L 
Tdo =  ′d  Td′ = 7.93 s
 Ld 
′
The given value of Tdo is 8.0669 s.
From the above analysis, we can conclude that the time-variation of the positive peak envelope of
only the fundamental component in ia(t) can be written as
 t 
− ′ 
 Td 
iacP (t ) = I pm e + I sm

 t 
− ′ 

(
= i( acPm) − I sm e )  Td 
+ I sm
(4.105)

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 181 6/2/2018 2:56:10 PM

 182  |  Power System Analysis

Using (4.103) and (4.104) in the above equation, we have,

 t 
 V V  − T ′  V
iacP (t ) =  m ′ − m  e  d  + m
 w o Ld w L
o d w o Ld

 t 
V  1 1  − T ′  V 1
= m  ′ − e  d + m
w o  Ld Ld  w o Ld

  t 
Vm  1  1 1  −  Td′  
= + − e  (4.106)
w o  Ld  L′d Ld 
 
 

Substituting for Vm from (4.102) in the above equation, we obtain,

  t 
− ′ 
2  1  1 1  T 
iacP (t ) = M dF iF0  +  ′ −  e  d   (4.107)
3 L Ld 
 d  Ld 
 

It is to be noted that in the above equation, the RHS has terms which are similar to that in (4.71) when
the wave tends to be in a steady-state, and similar to (4.87), neglecting the fundamental frequency
terms, at t = 0. Therefore, we write that
2
iacP (t ) = [ −id (t ) without sinor terms] (4.108)
3

where id(t) is given by

  t 
 1  1 1  −  Td′  
id (t ) = − M dF iF0 L + ′ − L  e  (4.109)
 d  Ld d 
 

Now, taking the Laplace transform of the above equation, we have,

M dF iF0 1  1 1  Td′
id ( s) = − − M dF iF0  ′ − 
Ld s  Ld Ld  (1 + sTd′ )

M dF iF0 1 M dF iF0  1 1  sTd′

=− −  − 
Ld s s  L′d Ld  (1 + sTd′ )

M dF iF0  1  1 1  sTd′ 
=−  + ′ −   (4.110)
s  Ld  Ld Ld  (1 + sTd′ ) 

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 182 6/2/2018 2:56:13 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  183 

Multiplying the numerator and denominator by w0, we get,

M dF iF0 w o 1  1  1 1  sTd′ 
id ( s) = −  + ′ −  
s w o  Ld  Ld Ld  (1 + sTd′ ) 

E fd 0 1
=−
s X d ( s) (4.111)

where

1 1  1 1  sTd′ 
=  + ′ −  
X d ( s)  xd  xd xd  (1 + sTd′ )  (4.112)

E fd0
and − denotes the application of a negative step voltage so that it forces the net loop voltage to
s
zero representing a short-circuit.
The above equation, (4.111), can be written in the transformed domain as
1
id ( s) = vq ( s)
X d ( s) (4.113)

or
1
id ( s) = y d ( s) (4.114)
Ld ( s)

where

1  1  1 1  sTd′ 
= + ′ −   (4.115)
Ld ( s)  Ld  Ld Ld  (1 + sTd′ ) 

and denotes the operational admittance function of a synchronous machine for the d-axis rotor
circuits.
Note: In the above expression for id(s), it is to be noted that the field applied voltage, v0F , is left unal-
tered. This implies that vF(s) = 0

4.6.3.8  Short-circuit Current Plots for a Lab Machine

For a 230 V, 1 kVA, 50 Hz, four-pole, starr-connected DC-motor-driven synchronous generator,
the short-circuit test is conducted at a reduced voltage. The oscillogram of the three-phase cur-
rents along with the field current is recorded as shown in Figure 4.16. From Figure 4.16, it can be
seen that the phase currents have a dominant decaying DC component during the transient period
before they reach the sinor steady-state, whereas the filed current exhibits a dominant 50 Hz com-
ponent, (which can be verified by noting the time period of oscillation) before a DC steady-state
is reached [10].

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 183 6/2/2018 2:56:16 PM

 184  |  Power System Analysis

ib ic

5
Iabc (A)

−5

ia
17.98 18 18.02 18.04 18.06 18.08 18.1 18.12 18.14 18.16 18.18
Time (s )

0.6

0.5

0.4

0.3
If (A )

0.2

0.1

−0.1
17.98 18 18.02 18.04 18.06 18.08 18.1 18.12 18.14 18.16 18.18
Time (s )

Figure 4.16  iabc(t) and iF(t) Plots for a Lab Machine Under Short-circuit Condition.

4.7 ROTOR MECHANICAL EQUATION

In addition to the set of flux linkage-current expressions (4.14), (4.15), and y 0 = L0 i0 , the voltage
equations (4.24) to (4.30) and the following mechanical equations completely define the synchronous
generator model in terms of the basic parameters.
dq r
= w r (4.116)
dt
dw r
J = − Dw r + Tm − Teg (4.117)
dt
2
where q r = q , represents rotor angle displacement in mechanical radian.
P
wr denotes the rotor speed in mechanical rad/s.
J denotes the inertia of the rotor-mass in kg-m2.
D denotes the damping component, Tm Mechanical torque in N-m.
Teg is the electrical torque developed corresponding to wr and is given by (4.40).

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 184 6/2/2018 2:56:18 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  185 

The electrical power developed at the air-gap is given by

(
Peg = Teg w r = y d iq −y q id ) P2 w r
Also, note that under steady-state, Peg = electrical power output of the machine + I2Ra

4.8  ANALYSIS OF LINEAR SYSTEMS

It is noted that the presence of decaying second harmonic component and DC component in abc-
frame quantities have negligible influence on the low-frequency power swing performance1 of a
power system. Therefore, it is customary to neglect the fundamental frequency components in the
dq-frame quantities as it also avoids a choice of a small step-size for carrying out numerical integra-
tion. Therefore, if one does not want to have the fundamental frequency components in the dq-frame
dy d dy q
quantities, we need to set = 0 and = 0 . In order to justify this, a systematic eigenvalue
dt dt
and eigenvector analysis is carried out in the following lines.

4.8.1  Eigenvalue Analysis

Eigenvalues
The eigenvalue of a matrix is given by the value of the scalar parameter l for which there exist
non-trivial solution (i.e. other than u = 0) to the equation

Au = lu (4.118)
where A is an (n × n) matrix (real for a physical system such as a power system) and u is an (n × 1)
vector referred to as eigenvector.
To find the eigenvalue, (4.118) may be written in the form

( A − lI ) u = 0 (4.119)
where I is an identity matrix of dimension (n × n).
For a non-trivial solution,

det ( A − lI ) = 0 (4.120)

Expansion of the determinant gives the characteristic equation. The n solutions of l = l1, l2, … ln are
referred to as the eigenvalues of the matrix A. The eigenvalues may be real or complex, and a complex
eigenvalue always occur in conjugate pair. In general, li = s i + jw i , where s i is referred to as neper
frequency (neper/s), and wi is referred to as radian frequency (rad/s).

1 The power swings are nothing but the electromechanical oscillations associated with the rotor
­mechanical systems whose natural frequency of swings is relatively low in the range of 0.1 to 3 Hz.

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 185 6/2/2018 2:56:21 PM

 186  |  Power System Analysis

Eigenvectors
For any eigenvalue li, the n element column vector ui, which satisfies (4.118) is called the right-­
eigenvector of A associated with eigenvalue li, Therefore, we have

Aui = li ui i = 1, 2, , n
(4.121)
The eigenvector ui has the form
u1i 
u 
ui =  
2i
 
 
uni 

Since (4.119) is homogeneous, k ui (where k is a scalar) is also a solution. Thus, the eigenvectors are
determined only to within a scalar multiplier.
Similarly, the n element row vector w j which satisfies
w j A = l jw j j = 1, 2, , n
(4.122)
is called the left-eigenvector associated with the eigenvalue lj, and has the form

w j =  w j1 w j 2  w jn 

The left- and right-eigenvector corresponding to different eigenvalues are orthogonal. In other words,
if li is not equal to lj, we have,

w j ui = 0
(4.123)
However, in case of eigenvectors corresponding to the same eigenvalue li, we have,
wi ui = Ci
(4.124)
where Ci is a non-zero constant.
Since, as noted above, the eigenvectors are determined only to within a scalar multiplier, it is
common practice to normalize these vectors so that

wi ui = 1
(4.125)

Eigenvalues and Stability

The time-dependent characteristic of a mode corresponding to an eigenvalue li is given by elit.
Therefore, the stability of the system is determined by the eigenvalues as follows:
1. A real eigenvalue corresponds to a non-oscillatory mode. A negative real eigenvalue represents
a decaying mode. The larger its magnitude, the faster the decay. A positive real eigenvalue repre-
sents aperiodic monotonic instability.
2. Complex eigenvalues occur in conjugate pairs and each pair corresponds to an oscillatory mode.
The real component of the eigenvalues gives the damping, and the imaginary component gives the

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 186 6/2/2018 2:56:24 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  187 

frequency of oscillations. A negative real part represents damped oscillations, whereas a positive real
part represents oscillation of increasing amplitude. Thus, a complex pair of eigenvalues is given by,
l = s ± jw (4.126)

The frequency of oscillation in Hz is given by
w
f = (4.127)
2p

The damping ratio is given by

s
z =−
(4.128)
s 2 +w2
The damping ratio z determines the rate of decay of the amplitude of the oscillation correspond-
ing to frequency w. The above equation can be rearranged to give

1  1 1−z 2 
= T
s  2p z 

where T denotes the period of the oscillation in seconds.

  Now, consider,
 1
 1 1−z 2 
s 
 T   2p
=
z
 (4.129)

  

1
This ratio denotes the number of cycles of the waveform in time t = where the amplitude
1 s
decays to or 37% of the initial amplitude. For example, for a damping ratio of 5%, from
e
(4.129), we get 3.1791 cycles. This implies that in the following equation
 z t
− 2p 
−s t  1−z
2 T 
e =e

if we set z = 0.05 and t =3.1791, we get e −s t = e −1. Further, we can say that in 3 cycles, the
T
amplitude reduces to 38.92% of its initial value. If we state that in five cycles of oscillation, if the
amplitude has to reduce to 37% of its initial value then the damping factor, z should be 3.2%.

4.8.2 Modal Analysis of Linear Systems

For an nth order linear time-invariant theory (LTI) system, the zero-input-response, i.e., the natural
response can be described in state space form as [6].
x = A x (4.130)

with initial value of states, x(0) and the state matrix A is of dimension (n × n).

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 187 6/2/2018 2:56:28 PM

 188  |  Power System Analysis

Consider the transformation given by,

x =U y
(4.131)
where U is assumed to be a matrix of right-eigenvectors of A, pertaining to distinct eigenvalues,
 l1 , l2 ,  ln  of A.
From (4.130), we have

Uy = AUy
−1
y = U AUy
(4.132)
The operation, U −1 AU represents the similarity transformation such that

WAU = Dl (4.133)

where W = U−1, is a matrix of left-eigenvector of A.

    
 
U = u1 u2  un  with Aui = li ui i = 1, 2, 3....n
    

 w1 
 w2 
W =  with w A = l w j = 1, 2, 3....n
j j j
   
 
 wn 

and

 l1 
 l2 
Dl =  
  
 
 l n

Notes:
• ui is a column vector and wj is a row vector.
• ui and wj are orthonormal vector, i.e.,

w j ui = 1 for i = j
(4.134)
= 0 for i ≠ j

Using (4.133) in (4.132) we have,

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 188 6/2/2018 2:56:33 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  189 

y = Dl y
(4.135)
From (4.131), the initial value of y is given by

y (0 ) = U −1 x (0 ) = W x (0 )
(4.136)
The solution of (4.135) is obtained as
l1t
 y1 (t )  e   y (0 ) 
    1 
 y2 (t )  e l2t   y2 (0 )
y (t ) = =
    
  
   
 y n ( t )   e lnt   yn (0 )


From (4.131) and (4.136), we have,

l1t
 x1 (t )    e w1 x (0 ) 
    
 x2 (t ) =  u1 u2  un  e l2t w2 x (0 )
 
    
    
 x n (t )   e lnt wn x (0 )

or

 x1 (t ) 
       
 x2 (t ) =  u  e l1t w x (0 ) +  u  e l2t w x (0 ) +  +  u  e lnt w x (0 )
   1 1  2 2  n n
       
 x n (t )
n
or x (t ) = ∑ (wi x (0)) el t ui (4.137)
i

  i =1

Notes:
• (wi x(0)) is a scalar and it gives the contribution of the initial condition x(0) to the ith mode. In
other words, wi determines to what extent the ith mode gets excited (in a state) for a given initial
condition vector x(0). Thus, a left-eigenvector carries mode controllability information.
• ui describes the activity of each state variable in ith mode. In other words, it shows how the ith
mode of oscillation is distributed among the system states. Thus, it is said to describe the mode
shape of each state variable in the ith mode. The magnitude, |uki| gives the relative magnitude of
activity and the angle, ∠uki represents relative phase displacement of kth state in constituting
the ith mode. The angle information will be useful to group machines which swing together in
a mode. A right-eigenvector carries information regarding on which state variables the mode is
more observable.

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 189 6/2/2018 2:56:36 PM

 190  |  Power System Analysis

4.8.3  Solution of an LTI System

Consider the following LTI system with zero-input:

 x1   −1 2   x1 
 x  =  −49 −99  x  (4.138)
 2   2
with initial value of states given by: x1(0) = 1 and x2(0) = 0. Let us first obtain the solution by
classical approach.

4.8.3.1  Classical Approach

The above equation is written as [11],

x = Ax

( pU − A) x = (0)k
or
−1
x = ( pU − A)  (0)k
 

where k is a constant. Also, note that in the above equation, ( pU − A)−1  (0) represents the ‘oper-
 
ation on (0)’ leading to the natural response of the system. Thus, a general solution is given by
x (t ) = e At k and k to be evaluated for the given initial condition. The steps followed are listed below:

−1
1. Computation of ( pU − A) :

−1
 p + 1 −2 
( pU − A)−1 = 
 49 p + 99

1  p + 99 2 
=
( p + 1)( p + 99) + 98  −49 p + 1

1  p + 99 2 
= 
p + 100 p + 197  −49
2 p + 1

To obtain the roots, consider (p2 + 100p +197) = 0, we get,

−100 ± 1002 − 4 × 197

p1,2 =
2
−100 ± 95.979
=
2
= −2.0104, −97.9895

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 190 6/2/2018 2:56:42 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  191 

Thus,

 p + 99 2 
( p − p1 )( p − p2 ) ( p − p1 )( p − p2 ) 
( pU − A)−1 =  
 −49 p +1 
( p − p1 )( p − p2 ) ( p − p1 )( p − p2 ) 


2. Evaluation of residues:
Consider,
p + 99 A11 B11
G11 = = +
( p − p1 )( p − p2 ) ( p − p1 ) ( p − p2 )

where,
−2.0104 + 99
A11 = G11 ( p − p1 ) | p = p1 = = 1.0105
−2.0104 + 97.9895
−97.9895 + 99
B11 = G11 ( p − p2 ) | p = p2 = = −0.0105
2.0104 − 97.9895

Consider
2 A12 B12
G12 = = +
( p − p1 )( p − p2 ) ( p − p1 ) ( p − p2 )

where

A12 = G12 ( p − p1 ) | p = p1 = 0.02083

B12 = G12 ( p − p2 ) | p = p2 = −0.02083

Consider
−49 A21 B21
G21 = = +
( p − p1 )( p − p2 ) ( p − p1 ) ( p − p2 )

where,

A21 = G21 ( p − p1 ) | p = p1 = −0.51052

B21 = G21 ( p − p2 ) | p = p2 = 0.51052

Consider
p +1 A22 B22
G22 = = +
( p − p1 )( p − p2 ) ( p − p1 ) ( p − p2 )

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 191 6/2/2018 2:56:47 PM

 192  |  Power System Analysis

where,

A22 = G22 ( p − p1 ) | p = p1 = −0.01052

B22 = G22 ( p − p2 ) | p = p2 = 1.01052

Thus, we have,

1.0105e p1t − 0.0105e p2t 0.02083e p1t − 0.02083e p2t 

[ pU − A]−1 (0) = e At = 
 −0.51052e p1t + 0.51052e p2t −0.01052e p1t + 1.01052e p2t 

Now, x = Ax leads us to a general solution given by

x (t ) = e At k

1 
With x1 (0) = 1 and x2 (0) = 0 , we have x (0) =   = k .
0
  Thus, the natural response of the system is given by

1.0105  p1t  −0.0105 p2t

x (t ) =   e + 0.5105  e
 −0.5105  

or
1.0105  −2.0104t  −0.0105 −97.9895t
x (t ) =  e + e (4.139)
 −0.5105 0.5105 
Now, let us obtain the solution by eigenvalue and eigenvector-based analysis.

4.8.3.2  Eigenvalue and Eigenvector Based Analysis

Using the eig command in MATLAB, the eigenvalues and eigenvectors are obtained as
[ V D] = eig(A). They are given below:
 −2.0104 0 
The eigenvalue is given by D = 
0 −97.9896

0.8926 −0.0206
The right eigenvector is [V ] =  
 −0.4509 0.9998 
The left eigenvector is obtained by using the command, W = inv(V). It is given by,

1.1322 0.0233
W = inv(V ) =  
0.5106 1.0107 

Choose x = [V ] y such that y = [V ]−1 x = [W ] x . Now consider,

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 192 6/2/2018 2:56:53 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  193 

x = Ax
[V ] y = A[V ] y
y = [V ]−1 A[V ] y

Since eigenvalues are distinct, we have D = [V ]−1 A[V ] and

y = Dy

The solution of this equation is

y = e Dt y(0)

In terms of the original variable, we have,

x = [V ]e Dt [W ] x (0)
Expanding the solution in terms of vectors, we get

 x1 
x =   = [ v1 ] e p1t w1 x (0) + [ v2 ] e p2t w2 x (0)
 x2 
T
With x (0) = 1 0  , we have

0.8926   −0.0206
x=  (1.1322)e p1t +   (0.5106)e
p2t
 −0 .4509   0 .9998 
1.0105  −2.0104t  −0.0105 −97.9895t
= e + e (4.140)
 −0.5105 0.5105 
Note that this solution is identical to that obtained using the classical approach.

Notes:
• Since the roots p1(=−2.0104) and p2(=−97.9895) are real and |p2| is relatively large compared
to |p1|, the notion of time-constants can be used to explain their behaviour. The corresponding
time units are
1 1
= 0.4974 s and = 0.0102 s
2.0104 97.9895

In such system where

max(| p1,2 |) | p2 |
=
min(| p1,2 |) | p1 |

is very large, the system is attributed as a stiff system. An important implication of this is that if
the system is solved using a fixed-step numerical integration technique of explicit-type, then the
step-size should be chosen relatively small compared to the smallest time-constant, otherwise, the
numerical method would become unstable.

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 193 6/2/2018 2:56:58 PM

 194  |  Power System Analysis

• If it is desired to retain only the slow dynamics, so as to make the system less stiff, the following
procedure can be employed:
(a) By noting the right eigenvector, with respect to p2 = 97.9895, i.e., the fast mode, in [V]
matrix, it can be seen that the participation of state x2 is large in p2 mode or in other words,
mode-p2 is dominantly seen in x2 state.
(b) If we are interested only in slow mode, we can ignore the effect due to fast mode by setting
x2 = 0 . Thus,

x2 = −49 x1 − 99 x2 = 0.
49
x2 = − x1
99

Using the above equation in x1 = −1x1 + 2 x2 , we get,

 −49 
x1 = − x1 + 2  x
 99  1
x1 = −1.9898 x1

with x1(0) = 1.0

The solution to the above equation is

x1 (t ) = 1.0e −1.9898t
(4.141)
Using the above relationship between x1 and x2, we obtain
x2 (t ) = −0.4949e −1.9898t (4.142)

In Figure 4.17, the plots of x1(t) and x2(t) are shown for (4.140). It also shows the plots obtained from
(4.141) and (4.142). From Figure 4.17, it is clear that the modified system completely removes the
fast response, providing only the slow variation. This permits the choice of a larger step-size than that
should be chosen for (4.138) in case of numerical solution.

4.8.3.3 Generator Stator Transient Participation

From (4.25) to (4.27) with Ra = 0 and RF = 0, we can write the state-model valid for the short-circuit
condition considering the response only due to the initial conditions [0 y 0F 0]T , we get

y d  0 0 −w 0  y d 
    
y F  = 0 0 0  y F 
y q  w 0 0 0  y q 
  
The eigenvalues of the above system are p1 , p2 = ± jw 0 and p3 = 0. The right- and left-eigenvector
matrices are as indicated below:

0.7071 0.7071 0  0.7071 0 j 0.7071 

   
[V ] = 0 0 1  [W ] = 0.7071 0 − j 0.7071
 − j 0.7071 j 0.7071 0  0 1 0 

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 194 6/2/2018 2:57:03 PM

 Synchronous Machine Modelling Using Primitive Parameters  |  195 

1
X1 Actual

Modified

0.5
x1 and x2

X
2

−0.5
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (s )

Figure 4.17  Plot of x1 and x2 for the Actual and the Modified LTI System.

From the above right-eigenvectors, it can be inferred that the oscillatory mode corresponding to ± jw 0
is observable only in y d and y q and it is absent in y F as is evident from (4.96), (4.97), and (4.98).
Therefore, it is clear that the generator stator transients are due to the state variables, Yd and Yq ,
and a setting of their time-derivative equal to zero, removes the fundamental components in the
dq-frame quantities.

References
[1] K. R. Padiyar, Power System Dynamics - Stability and Control, BS Publications, Hyderabad,
India, 2002.
[2] A. R. Bergen and V. Vittal, Power System Analysis, Pearson Education Asia, India, 2001.
[3]  P. Kundur, Power System Stability and Control, McGraw-Hill Inc., NY, 1994.
[4] P. M. Anderson and A. A. Fouad, Power System Control and Stability, Iowa State University
Press, Ames, Iowa, 1977.
[5]  K. R. Padiyar, Analysis of Subsynchronous Resonance in Power Systems, Kluwer, Academic
Publishers, Boston, 1999.
[6]  K. Ogata, State Space Analysis of Control Systems, Prentice-Hall, Englewood cliff, NJ, 1967.
[7] P. M. Anderson, B. L. Agrawal, and J. E. Van Ness, Subsynchronous resonance in power systems,
IEEE Press, Piscataway, New Jersey, 1990.
[8] E. W. Kimbark, Power System Stability: Synchronous machines, Volume III, John Wiley and
Sons, 1956.

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 195 6/2/2018 2:57:06 PM

 196 | Power System Analysis

[9] IEEE Guide for Test Procedures for Synchronous Machines, IEEE Std 115-2009, May 2010.
[10] D. Surendra and K. N. Shubhanga, Development of A Power System Laboratory Supported By
Real-time Systems, International Conference on Power and Energy Systems (ICPS)-2011.
[11] I.R. Rao and Shubhanga KN, Circuit Analysis in the Time-domain: Operational Approach to
form the System-matrix Exponentials and to obtain the Natural Response therefrom. 1st inter-
national conference on Advances in Electrical Engineering, (ICAEE-2014), Vellore, India,
Jan 9-11, 2014.

Review Questions
1. Compare the structure of the inductance matrices for non-salient and salient pole synchronous
machines and comment on their dependency on the rotor-angle.
2. For the single-cage induction motor shown in Figure 4.18, obtain the elements of the inductance
matrix, L[q ](6×6).

ra
e
as
ph
sA

of
is
ax
tic
ne

rb
ag

sC
M

wr sB

qr

Magnetic axis of phase sA

120°
rc

Rotor ra

Stator

sA

Figure 4.18 Induction Motor Model.

3. For a salient pole machine, derive an expression for L0, Ld and Lq in terms of basic primitive
parameters.
4. In their usual notations, derive the stator voltage equations in the machine-frame.
5. Discuss the significance of machine field voltage, Efd.
6. For the following primitive parameters, perform the following tasks:
Ld 4.98mH
Mdf 49mH
Lf 577mH
Lq 4.84mH

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 196 6/2/2018 2:57:07 PM

 Synchronous Machine Modelling Using Primitive Parameters | 197

ifo 10 A
RF 0.0715 ohms
Ra 0.0031 ohms
Vfo ifo*RF;
wB 2*pi*60 rad/s
(a) Plot vabc under open-circuit condition of the generator when a field voltage, Vfo is suddenly
applied by keeping the speed of the machine constant at wB.
(b) Assume that the machine is on open circuit with an initial value of field current, ifo with
Vfo = 0.
(c) Assume that the machine is on open circuit with an initial value of field current, ifo with field
voltage equal to Vfo.
In the above mentioned (b) and (c) cases, at a certain time instant, a three-phase symmetrical short
circuit is applied at the terminals. Determine i0 dq and iabc for the following cases:
i. Ra = 0 and RF = 0.
ii. Ra = 0 and RF ≠ 0.
iii. Ra ≠ 0 and RF = 0.
iv. Ra ≠ 0 and RF ≠ 0.
Assume that the machine speed is held constant.
For case (c)-iv, perform the Fourier series analysis of ia (t ) over selected cycles (of a fundamen-
tal period) from the fault instant to the point of reaching a steady state. The equations related to
the Discrete Fourier series analysis are given as follows:
Let N be the number of samples over one fundamental time period T . For an nth -order har-
monic, the coefficients are given as:
N −1
2
an =
N ∑ f (k )cos(nb k )
k =0
N −1
2
bn =
N ∑ f (k )sin(nb k )
k =0
and
N −1
1
a0 =
N ∑ f (k )
k =0
 2p 
where b =   denotes sample angle and f ( k ) represents the sample value of the signal at k th
 N
instant.
Draw a bar chart of harmonic amplitudes for each cycle with n = 0,1, 2, 3, 4.

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 197 6/2/2018 2:57:15 PM

 198 | Power System Analysis

7. Consider the circuit shown in Figure 4.19.

R = 0.1Ω L1 = 10 mH

i1
Vs = 1.0 V + vc
L2 = 1 H C = 100 µF

i2

Figure 4.19 Circuit for Participation Factor Analysis.

Perform the following:

T
(a) Obtain the response of x = [i1 i2 vc ] for x(0) = 0 using SIMULINK.
(b) Obtain the eigenvalues and right and left eigenvectors, and identify the dominant states, that
is, y (a subset of x) that are associated with the fast transients.
dy
(c) Set = 0 and obtain the response of i1 (t ), i2 (t ) , and vc (t ), and compare them with x (t ).
dt

8. For the circuit shown in Figure 4.20, R1 = 1 Ω, R2 = 0.5 Ω, L = 0.5 H, C = 1 F. Determine the
eigenvalues of the system. Given that iL(0) = 0 and vc(0) = 0, obtain the response for a source
voltage v (t) = u(t) applied at t = 0. Estimate the natural frequency and decrement factor from the
time-domain plots.
iL
R1 L

C
i1 R2 vc
v iL

I II

Figure 4.20 RLC Circuit for Decrement Factor Calculation.

M04_POWER-SYSTEM-AN00_SE_XXXX_CH04.indd 198 6/2/2018 2:57:27 PM

 The Standard Parameters
of Synchronous Machine 5
CHAPTER OUTLINE

5.1 Requirement of Generator Modelling 5.9 Steady-state Operation

Neglecting Stator Transients of Synchronous Generator
5.2 Operational Inductance Approach 5.10 Steady-state Analysis:
5.3 Per-unit Representation of Generator Space-phasor Approach
Quantities 5.11 Modification of 2.2 Model
5.4 Summary of Equations Pertaining and Equivalent Circuits
to 1.1 Model 5.12 Initial Condition Calculations
5.5 A Case Study: Voltage Build-up 5.13 Example: Eigenvalue Analysis
on No-load of a Generator Under Short-circuit
5.6 Operational Impedance for 2.2 Condition
Model 5.14 Example: Synchronisation of a
5.7 Summary of Equations Pertaining Standalone Generator to Mains
to 2.2 Model 5.15 Example: Transient Stability
5.8 Swing Equation Simulation of an SMIB System

In this chapter, a state-space model of a synchronous machine is developed by employing the standard
parameters of a synchronous machine since it is practically not easy to estimate the basic or primitive
parameters (such as MdF, LF, RF, etc.) by conducting any experiments on the machine. Further, it is
shown that the kind of state-model derived previously, in terms of basic parameters, does not allow
to make approximations, say, to retain rotor winding transients while neglecting stator transients.
An operational inductance approach is employed to derive a state-model which offers mapable state
variables with respect to the actual machine variables and avoids an explicit choice of base values for
circuits other than the stator winding.
A detailed derivation is carried out to arrive at a state-model from the standard transfer functions of
the generator. The steady-state model of the generator is obtained as a special case of the state-model.
This is further supported by presenting a space-phasor-based analysis.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 199 6/2/2018 3:55:38 PM

 200  |  Power System Analysis

Examples such as short-circuit analysis of an unloaded generator and interfacing of a generator

to the mains are presented to understand the usage of the derived synchronous generator model
with and without considering the stator transients. In the later part of the chapter, case studies are
introduced to analyse transient stability of an single-machine infinite bus (SMIB) system, including
observations from a lab machine.

5.1 REQUIREMENT OF GENERATOR MODELLING

NEGLECTING STATOR TRANSIENTS
In rotor-angle stability studies, it is customary to neglect stator transients as the natural frequency of
oscillations of stator dynamics is much larger in comparison to that of electromechanical systems. In
this regard, the following observations are made:
1. Such an assumption is found to aid the computational speed since the system becomes less stiff.
The step-size, Dt, for numerical integration of ODEs, can be increased by 50 times than required
if stator transients were to be considered [1]. This is important with explicit-type numerical inte-
gration methods because the numerical method itself may become unstable if the step-size is not
chosen adequately.
2. The loss of accuracy in the electromechanical oscillation-related study due to the said
approximation is negligible.
3. If the stator transients are neglected, it brings consistency while interfacing synchronous machine
equations to the network equations (in the form of algebraic constraints), which represent the
quasi-sinusoidal steady-state behaviour of network elements.
However, such an assumption is valid only in cases where electromechanical oscillations of rotor
systems are of interest whose frequency of oscillations lies in the range of 0.5–3 Hz [2]. While car-
rying out such a simplification, one may end up with inconsistent generator equations, if they are not
handled carefully. To illustrate this, the following example is considered.

5.1.1  Short-circuit Analysis with Stator Transients Neglected

In this section, a case study is presented, where stator transients are neglected while attempting to
retain the slow rotor dynamics. Reproducing the stator voltage equations, we have

dy d (5.1)
vd = −id Ra − w oy q −
dt
dy q
vq = −iq Ra + w oy d − (5.2)
dt

From the participation factor analysis, it is shown earlier that high frequency oscillations are domi-
nantly associated with states yd and yq. Therefore, to eliminate these high frequency transients, i.e.,
dy d dy q
the stator dynamics, we can set = 0 and = 0 in the above equations. Hence, the stator
voltage equations reduces to dt dt

vd = −id Ra − w oy q

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 200 6/2/2018 3:55:41 PM

 The Standard Parameters of Synchronous Machine  |  201 

vq = −iq Ra + w oy d

Note that vd = 0 and vq = 0 under short circuit conditions. Also assuming Ra = 0, from the above
equations we have

yq = 0 and y d = 0 (5.3)

Reproducing the flux linkage equations

y d = Ld id + M dF iF
y F = M dF id + LF iF
y q = Lq iq

Making use of eq. (5.3) in the above equations we get,

yq = 0 implies that iq = 0
M dF
yd = 0 implies that id = − iF
Ld

Therefore, we have
 M 
y F = M dF  − dF iF  + LF iF
 Ld 
 M2 
=  LF − dF  iF
 Ld 
2
M dF
y F = LÄ F iF with LÄ F = LF − (5.4)
Ld

Voltage equation corresponding to field-coil is given by

dy F v Fo
v F = iF RF + with iFo =
dt RF

Using eq. (5.4) in the above equation we can write,

diF R vo
= − Ä F iF + ÄF (5.5)
dt LF LF
Solving the above differential equation for iF (t), we get,

−
t  t
− Ä 
o TdÄ v Fo  Td 
i F (t ) = iF e + 1− e
RF  
 

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 201 6/2/2018 3:55:48 PM

 202  |  Power System Analysis

v Fo
With iFo = , we have
RF
v Fo
i F (t ) = (5.6)
RF
LÄ F
Where TdÄ = is defined as the d-axis short-circuit transient time-constant.
RF
From eq. (5.6), it can be seen that there are no transients in the field current and hence in id (t). This
example clearly demonstrates that when the stator transients are neglected in the usual way, the slow
rotor dynamics also disappear, making the procedure inapplicable.

5.1.2  Type of Standardised Synchronous Machine Models

With the synchronous machine models in terms of primitive parameters, the following difficulties are
generally faced:
1. When the stator transients are neglected, the rotor electrical circuit dynamics also disappear in
the process.
2. The basic machine parameters such as LF, MdF, MdH, LFH, RF, etc., cannot be readily determined
by any standard experimentation procedure.
Therefore, to derive a state-space model which uses the measurable standard parameters [3] and
simultaneously permits to neglect stator transients, the following approaches are employed [4, 5]:
1. Operational inductance (OI) approach: Here, rotor circuit dynamics are represented by a set of
short- or open-circuit time-constants by employing operational impedance functions without
explicitly evaluating the rotor circuit parameters [2, 6].
2. Equivalent circuit (EC) approach: Here, an equivalent circuit is chosen for the rotor circuit and
the circuit elements are then determined from the standard parameters by employing a data con-
version procedure [1, 7].
It should be noted that the standard parameters of a synchronous machine are determined by employ-
ing the IEEE specified test procedures and correspond to the fundamental frequency components in
an abc-frame. In the following section, the operational inductance (OI) approach is discussed.

5.2  OPERATIONAL INDUCTANCE APPROACH

For the purpose of illustrating the OI method in terms of the standard parameters, a simplified model
of synchronous machine is chosen by neglecting the H-winding along the d-axis and K-coil along the
q-axis. Such a modelling detail is referred to as 1.1 model for a synchronous machine. For the d-axis
rotor circuit, the related equations are given by

y d = Ld id + M dF iF (5.7)

y F = M dF id + LF iF (5.8)

dy F
v F = iF RF + (5.9)
dt

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 202 6/2/2018 3:55:51 PM

 The Standard Parameters of Synchronous Machine  |  203 

Taking the Laplace transform of the above equations with yF(0) = 0, we get

y d ( s) = Ld id ( s) + M dF iF ( s) (5.10)

y F ( s) = M dF id ( s) + LF iF ( s) (5.11)

v F ( s) = RF iF ( s) + s y F ( s) (5.12)

Using eq. (5.11) in eq. (5.12) and solving for iF (s) we get,

v F ( s) − s M dF id ( s)
i F ( s) = (5.13)
( RF + sLF )
Using eq. (5.13) in eq. (5.10), we have

 2
sM dF   M dF 
y d ( s) =  Ld −  id ( s) +  v F ( s)
 RF + sLF   RF + sLF 
2
M dF M dF
s
RF RF
= Ld id ( s) − Ä
id ( s) + Ä
v F ( s) (5.14)
sTdo +1 sTdo +1
Ä LF
where Tdo =
RF
2
Ä M dF  M dF 
Ld ( sTdo + 1) − s
RF  R 
y d ( s) = Ä
id ( s) +  Ä F  v F ( s)
sTdo +1  sTdo + 1
 

 Ä
2
M dF Ä
  M dF 
L (
 d dosT + 1) − s Tdo 
LF  
=  id ( s) +  RF  v F ( s) (5.15)
 Ä  Ä
sTdo +1  sTdo + 1
   
 

2
M dF
Since LÄd = Ld − , we have
LF
 M dF
2 
Ä
 −  = ( Ld − Ld ) (5.16)
 LF 

Now, using eq. (5.16) in eq. (5.15) we get,
 M dF 
Ä Ä Ä  R 
Ld ( sTdo + 1) + ( Ld − Ld ) sTdo
y d ( s) = Ä
id ( s) +  Ä F  v F ( s)
sTdo + 1  sTdo + 1
 

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 203 6/2/2018 3:55:56 PM

 204  |  Power System Analysis

 M dF 
=
( Ld + LÄd sTdo
Ä
)  R 
id ( s) +  Ä F  v F ( s) (5.17)
Ä
sTdo +1  sTdo + 1
 

LÄ F M2 Ä L
Note that with TdÄ = , where LÄ F = LF − dF , and Tdo = F , we have
RF Ld R F

Ä
Tdo LF (5.18)
=
TdÄ LÄ F

2
M dF M2
Since LÄd = Ld − , and LÄ F = LF − dF we have
LF Ld

LÄd LF = LÄ F Ld = ( Ld LF − M dF
2
)

Hence,
LF Ld (5.19)
Ä
=
LF LÄd

Using eq. (5.19) in eq. (5.18) we get

Ä
Tdo Ld
= (5.20)
TdÄ LÄd

Using eq. (5.20) in eq. (5.17) we can write

 M dF 
 sTdÄ + 1   RF 
y d ( s) = Ld  Ä  id ( s) +  Ä  v F ( s) (5.21)
 sTdo + 1  sTdo + 1
 

or
M dF
y d ( s) = Ld ( s) id ( s) + G( s) v F ( s) (5.22)
RF

where
 1 + sTdÄ 
Ld ( s) = Ld  Ä  (5.23)
 1 + sTdo 

1
G ( s) = Ä
(5.24)
1 + sTdo

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 204 6/2/2018 3:56:02 PM

 The Standard Parameters of Synchronous Machine  |  205 

From eq. (5.22) we have

y d ( s) G( s) M dF
id ( s) = − v F ( s) (5.25)
Ld ( s) Ld ( s) RF

The above equation represents the dynamics in a two-port network as shown in Figure 5.1.
id(s) iF (s)

Two
port
yd(s) vF (s)
network

Figure 5.1  Two-port Network Representation of d-axis-related Dynamics.

Note: Ld (s) given in eq. (5.23) clearly shows that in different time-frames, the machine offers varied
inductance. For example, in the time period immediately after the occurrence of a disturbance, along
the d-axis, the machine inductance attains a value equal to LÄd which is simply obtained as the high
frequency gain of the system function (5.23). However, when the machine reaches a steady-state, the
inductance of the machine changes to Ld and remains at this value. This value can be determined as
the low frequency gain of the system function.
In the following section, an example is presented to show that if short-circuit analysis is carried
out with the operational inductance function, instead of carrying with the basic parameters, it leads
to the same results.

5.2.1  N
 atural Response Under Short-circuit Condition
Using OI Model
Neglecting the armature resistance, i.e., Ra = 0 in eq. (5.1) and eq. (5.2), we have the stator voltage
equations as
dy d
vd = −w oy q − (5.26)
dt
dy q
vq = w oy d − (5.27)
dt

Also, we know that under the steady-state open-circuit condition prior to the occurrence short-circuit,
vd = 0 and vq = vqo . Hence, the above equations in Laplace-domain appear as
0 = −w oy q ( s) − sy d ( s) (5.28)

vq ( s) = w oy d ( s) − sy q ( s) (5.29)

From eq. (5.28), we have
sy d ( s)
y q ( s) = − (5.30)
wo

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 205 6/2/2018 3:56:06 PM

 206  |  Power System Analysis

Substituting eq. (5.30) in eq. (5.29), we have

s2y d ( s) (5.31)
vq ( s) = w oy d ( s) +
wo
or
w o vq ( s )
y d ( s) = (5.32)
( s2 + w o2 )

Substituting eqs (5.23), (5.24) and (5.32) in eq. (5.25), we get id (s) as
Ä
w o vq ( s)(1 + sTdo ) 1 1 M dF
id ( s) = − v F ( s) (5.33)
(s 2
+ w o2 ) Ld (1 + sTdÄ ) Ld (1 + sTdÄ ) RF

Ä
(1 + sTdo )
Setting RF = 0 can be achieved by replacing by its high frequency gain which is equal to
Ä (1 + sTdÄ )
Tdo 1
and replacing the second term by zero, i.e., the high frequency gain of . In addition,
TdÄ (1 + sTdÄ )
since the field circuit is assumed to be loss-less, there is no requirement of v F to maintain the current
at ioF. Hence, the field circuit is shorted without any source. From eq. (5.20) we have,
Ä
Tdo Ld
= (5.34)
TdÄ LÄd
Furthermore, note that in the Laplace-domain, applying a short-circuit at the terminals is equivalent
to applying a step-voltage of the following form at t = 0

vqo
vq ( s ) = − (5.35)
s
where
vqo = E ofd = w o M dF iFo

Now from eq. (5.33) we get,

−vqow o Ld 1
id ( s) = − (5.36)
sLd LÄd ( s + w o2 )
2

vqow o 1
=− (5.37)
LÄd s( s + w o2 )
2

Taking the Laplace inverse of the above equation and using the value of vqo , id(t) can be obtained as

M dF iFo
id (t ) = −
(1 − cos w ot ) (5.38)
LÄd
The above expression is identical to that obtained using the state-space model with basic parameters.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 206 6/2/2018 3:56:11 PM

 The Standard Parameters of Synchronous Machine  |  207 

5.2.1.1  Alternate Approach for Determining Ld

The expression for LÄd in terms of the basic parameters can also be obtained as follows:
Consider,

y d = Ld id + M dF iF (5.39)

y F = M dF id + LF iF (5.40)

From eq. (5.40) an expression for iF is written as
1 M
iF = y F − dF id (5.41)
LF LF

Using the above equation in eq. (5.39), we get

 1 M 
y d = Ld id + M dF  y F − dF id 
 LF LF 

 M2  M
=  Ld − dF  id + dF y F (5.42)
 LF  LF

From the above expression, if we consider a small deviation in the variables immediately after the
disturbance, we have

 M2  M
∆y d =  Ld − dF  ∆id + dF ∆y F
 LF  LF

Assuming DyF is negligible immediately after the disturbance, we have

 M2 
∆y d =  Ld − dF  ∆id = LÄd ∆id (5.43)
 LF 

This implies that

 M2 
LÄd =  Ld − dF 
 LF 

5.3  PER-UNIT REPRESENTATION OF GENERATOR QUANTITIES

In power system analysis, it is usually convenient to use a per-unit system to normalise generator quanti-
ties. Compared to the use of physical units (A, V, W, Wb, H, etc.), the per-unit system offers computational
simplicity by eliminating units and expressing system quantities as dimensionless ratios. Therefore,
Actual quantity
Quantity in per-unit =
Base value of quantity

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 207 6/2/2018 3:56:15 PM

 208  |  Power System Analysis

Some base quantities may be chosen independently and quite arbitrarily, while others follow automat-
ically, depending on the fundamental relationships between the system variables. Normally, machine
ratings are chosen as base values so that principal variables will be equal to one per-unit under rated
condition.
Representation of generator quantities in per-unit on an appropriate base values facilitates the
following [1, 2]:
1. It removes arbitrary constants and simplifies mathematical equations so that they may be expressed
in terms of equivalent circuits.
2. The parameter values of machines of a given rating generally lie in a close range; a safe assump-
tion can be made in case the exact values are not known.
3. It permits easy understanding of system characteristics.
For synchronous machines, the choice of base quantities for the stator is straightforward, whereas, it
requires a careful consideration for the rotor [10]. However, it is noted that if the numerical values of
the rotor variables are not required at actuals, there is no need to explicitly define the base quantities
for the rotor circuits. The effects of the rotor circuits are accounted in the model by simply referring
the rotor quantities to the stator circuits.
In this section, a superbar will be used to identify per-unit quantities. However, the superbar will be
dropped once all quantities are expressed in per-unit in the later part.

5.3.1  Base Quantities

The universal practice is to use machine ratings as the base values for the stator quantities. It is to
noted that in per-unit representation, an equivalent starr-connected machine is assumed. All param-
eters correspond to the per-phase values. Let us choose the following base quantities for the stator:

VB = Voltage base ( RMS value of rated line voltage), V

S B = Power base ( Apparent power rating), VA
f B = Frequency base ( rated frequency ), Hz

The base values of the remaining quantities are automatically set and depend on the above as follows:

SB
Base current ( A ) I B =
VB
VB
Per phase impedance base (Ω) Z B =
IB
ZB
Inductance base ( H) LB =
wB
Angular frequency base (elec. rad/s) w B = 2p f B
VB
Flux linkage base ( Wb - turns) y B =
wB

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 208 6/2/2018 3:56:18 PM

 The Standard Parameters of Synchronous Machine  |  209 

5.3.2  Per-unit Stator Voltage Equations

Reproducing the stator voltage equations from eqs (5.1) and (5.2), we have,

dy d
vd = −id Ra − wy q −
dt
dy q
vq = −iq Ra + wy d −
dt

Dividing throughout by VB, and noting that VB = IB Z B = w By B , we get

vd i R wy q 1 dy d
=− d a − −
VB I B Z B w By B y Bw B dt

Expressing in per-unit notation, we have,

w 1 dy d
vd = − id Ra − yq − (5.44)
wB w B dt

Similarly,

w 1 dy q
vq = − iq Ra + yd − (5.45)
wB w B dt

Per-unit Reactances:
d-axis synchronous reactance xd = w B Ld. In per unit notation,

xd w B Ld
xd = = = Ld
Z B w B LB

d-axis transient reactance xÄd = w B LÄ d . In per unit notation,

Ä
xÄ d = L d

Similarly, q-axis synchronous reactance xq in per unit notation,

xq = Lq

5.3.3  Flux-linkage Expressions in per-unit for 1.1 Model

Reproducing eq. (5.22), we have,
M dF
y d ( s) = Ld ( s) id ( s) + G( s) v F ( s)
RF

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 209 6/2/2018 3:56:24 PM

 210  |  Power System Analysis

VB
Dividing throughout by yB, and noting that y B = LB I B = , we get
wB

y d ( s) Ld ( s)id ( s) G( s) M dF v F ( s)
= + (5.46)
yB LB I B RF  VB 
 w 
B

Since,

M dF E fd ( s)
G ( s) v F ( s) G ( s)
RF wo
=
 VB   VB 
 w   w 
B B

E fd ( s)
G ( s)
VB
=
 wo 
 w 
B

w 
= G( s) E fd ( s) as  o  = 1
wB 

Hence, eq. (5.46) in per unit is given by

y d ( s) = X d ( s) i d ( s) + G( s) E fd ( s) (5.47)

with

Ld ( s)  1 + sTdÄ 
X d ( s) = = xd  Ä 
LB  1 + sTdo 

and
1
G ( s) = Ä
1 + sTdo

5.3.4 Derivation of a State-space Model for d-axis Rotor

Circuits without Damper Winding
From eq. (5.47), an equivalent circuit is derived for d-axis in two stages, first setting Efd (s) to zero and
then setting yd (s) to zero in the following expression:

y d ( s) G ( s)
id ( s) = − E fd ( s) (5.48)
X d ( s) X d ( s)

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 210 6/2/2018 3:56:28 PM

 The Standard Parameters of Synchronous Machine  |  211 

Note that all quantities are in per unit and the superbar on variables have been dropped for conveni-
ence. Since the equation is linear, the effects of yd (s) and Efd (s) on id (s) are considered separately as
shown in the following lines.
Setting Efd (s) = 0:
From eq. (5.48), we have
 1 
id ( s) |E fd ( s)= 0 = y d ( s)  (5.49)
 X d ( s) 

We know from eq. (5.23) that we have
Ä 
1 1  1 + sTdo
=  
X d ( s) xd  1 + sTdÄ 

or
Ä
1 1  1  1  sTdo 
=  Ä  +  Ä 
X d ( s) xd 1 + sTd  xd 1 + sTd 

 1  1  sTdo Ä 
1
=  Ä
+ 1 − 1 +  Ä 
xd 1 + sTd  xd 1 + sTd 

 Ä x 
 sTd  Äd  
1  sTdÄ
 1   xd  
= 1 − +
xd  1 + sTd  xd  1 + sTdÄ 
Ä 
 
 
1 1 1 1  sTdÄ
= + Ä −  (5.50)
X d ( s) xd  xd xd  1 + sTdÄ

1
Note that the above expression for is identical to that derived using the oscillogram of the
X d ( s)
machine per-phase current ia(t), in the previous chapter. This shows that the standard parameters can
be obtained for a machine (at least for d-axis), by employing the procedure given in [9]. However, to
determine q-axis parameters, the standstill frequency response tests have be carried out [10].
Let us denote xf l as

1 1
x fl = = (5.51)
 1 1  xd − xÄd 
 Ä −x   Ä 
 xd d  xd xd 

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 211 6/2/2018 3:56:32 PM

 212  |  Power System Analysis

Using eq. (5.51) in the second term in eq. (5.50), we get

1 1 1 sTdÄ
= +
X d ( s) xd x fl 1 + sTdÄ
1 1 (5.52)
= +
xd  x fl 
 Ä + x fl 
 sTd 

Using eq. (5.52) in eq. (5.49), we get
 
 
 1 1  (5.53)
id ( s) |E fd ( s)= 0 = y d ( s)  +
xd  x fl 
  Ä + x fl  
  sTd 

Note that the second term in the above equation represents an admittance function of a series RC cir-
TÄ
cuit with R = xf l and Cf = d . Therefore, an equivalent can be written as shown Figure 5.2.
x fl
1
id(s) xfl sCf

yd(s) xd

Figure 5.2  Partial d-axis Equivalent Circuit with Efd (s) = 0 for No Damper Winding Case.

Setting yd (s) = 0:

From eq. (5.48), we have

G ( s)
id ( s) |y d ( s) = 0 = − E fd ( s) (5.54)
X d ( s)

Substituting for G(s) and Xd (s) in the above equation we have
1
id ( s) |y d ( s) = 0 = − E fd ( s)
xd (1 + sTdÄ )

 E fd ( s) 
 x  x fl
d
=−
x fl (1 + sTdÄ )

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 212 6/2/2018 3:56:36 PM

 The Standard Parameters of Synchronous Machine  |  213 

 1   x fl 
E fd ( s)    Ä 
 xd   sTd 
id ( s) |y d ( s) = 0 = −
 x fl 
 x fl + Ä 
 sTd 
E
=−
 x fl  (5.55)
 x fl + Ä 
 sTd 

where
E fd ( s)  x fl 
E=   (5.56)
xd  sTdÄ 

It can be seen that for eq. (5.55), we can write an equivalent circuit as shown in Figure 5.3.

1
id(s) xfl sCf

yd(s) = 0 +
xd E
−

Figure 5.3  Partial d-axis Equivalent Circuit with yd (s) = 0 for No Damper Winding Case.

Combining the effects of yd (s) and Efd (s):

Combining the effect of yd (s) and Efd (s), we can write the complete equivalent circuit for eq. (5.48)
as shown in Figure 5.4.

1
id(s) xfl sCf

+
yd(s) xd E
−

Figure 5.4  d-axis Equivalent Circuit Considering yd and Efd for No Damper Winding Case.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 213 6/2/2018 3:56:39 PM

 214  |  Power System Analysis

5.3.4.1 Choice of a State-variable for d-axis Equivalent Circuit

without Damper Winding
Since there are infinite ways in which one can choose a state variable, it is always desirable to choose
such a state-variable which closely indicate the original variable yF. In this context, consider the orig-
inal field equation in per unit as follows:
1 dy F
v F = iF RF +
w B dt

Taking the Laplace-transform of the above equation and rewriting, we get,

RF v ( s)
y F ( s ) = −i F ( s ) wB +wB F (5.57)
s s

R
From the above equation, it can be seen that the term iF ( s) F w B denotes the equivalent voltage
s
1
drop across the operational reactance in Figure 5.5, and E maps to the second term in the above
sC f
equation. Hence, y ′ F ( s) represents the referred value of yF (s).

1
id(s) xfl sCf

i ′F (s)

+
yd(s) xd yf (s) x 1
E = Efd(s) xfl
d sT d′
−

Figure 5.5  d-axis Equivalent Circuit: Choosing a State-variable (no Damper Winding Case).

Therefore, from the circuit, we can write

y ′ F ( s) −y d ( s)
i ′ F ( s) = (5.58)
x fl

where
1
y ′ F ( s) = E − i ′ F ( s) (5.59)
sC f

or
x fl
y ′ F ( s) = E − i ′ F ( s) (5.60)
sTd′

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 214 6/2/2018 3:56:43 PM

 The Standard Parameters of Synchronous Machine  |  215 

Substituting eq. (5.56) and eq. (5.58) in the above equation and simplifying, we get

 1  y ( s) E fd ( s) x fl
y ′ F ( s)  1 + ′  = d ′ +
 sTd  sTd xd sTd′
( sTd′ + 1) y d ( s) E fd ( s) x fl
y ′ F ( s) = +
sTd′ sTd′ xd sTd′
x fl
sTd′y ′ F ( s) = −y ′ F ( s) + y d ( s) + E fd ( s)
xd

Therefore, the differential equation governing the dynamics of d-axis rotor field winding is given by

dy ′ F 1  xd′ 
= ′  −y ′ F + y d + ′
E fd  (5.61)
dt Td  xd − xd 

5.3.4.2  Algebraic Relationship between id and Flux-linkages

To account the effect of yd, y ′ F on id , consider the expression eq. (5.48) and write id as

id = id1 + id 2 (5.62)

where
id1 = id |y ′ =0 and id 2 = id |y d = 0
F

From Figure 5.5, the component id1 is given by

yd
id1 =
 xd x fl 
x +x 
 d fl 

 x xÄ 
y d  xd + d dÄ 
 xd − xd 
id1 =
 x xÄ 
xd  d dÄ 
 xd − xd 

y d {xd ( xd − xÄd ) + xd xÄd }

=
xd ( xd xÄd )

y d (5.63)
id1 =
xÄd

Also, id2 is given by

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 215 6/2/2018 3:56:48 PM

 216  |  Power System Analysis

y ′F
id 2 = −
x fl
 x − x′ 
id 2 = −y ′ F  d ′ d  (5.64)
 xd xd 

Substituting eqs (5.63) and (5.64) in eq. (5.62), we have

yd  xd − xd′ 
id = − y ′ F  (5.65)
xd′  xd xd′ 

or we can write
( xd − xd′ )
y d = xd′ id + y ′ F (5.66)
xd

5.3.5  Derivation of State-space Model for q-axis Rotor Circuit

In this case, to have symmetry, one damper winding, i.e., G-coil is considered along q-axis. Now, for
q-axis, the relationship between yq and iq is derived in the operation impedance form in a manner
similar to that of d-axis coils. The flux linkage-current expressions are given by

y q   Lq M qG   iq 
y  =  M LG  i 
 G   qG  G

Also,
dy G
0 = − RG iG − (5.67)
dt

In the per unit operational inductance form, we can write

y q ( s) = X q ( s) iq ( s) (5.68)

where
(1 + sTqÄ )
X q ( s ) = xq Ä
(5.69)
(1 + sTqo )

with
Ä
Tqo xq
= (5.70)
TqÄ xÄq

Ä
and Tqo represents q-axis open-circuit transient time-constant.
Tq′ denotes q-axis short-circuit transient time-constant.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 216 6/2/2018 3:56:53 PM

 The Standard Parameters of Synchronous Machine  |  217 

xÄq denotes q-axis transient reactance.

Similar to eq. (5.50), we have

Ä
1 1  1 1  sTq
= + Ä −  (5.71)
X q ( s) xq  xq xq  1 + sTqÄ

Also, define
xq xÄq
x gl = (5.72)
xq − xÄq

Using eq. (5.71) in eq. (5.68) and from the definition of xgl, we obtain an equivalent circuit for the
q-axis as shown in Figure 5.6.

iq(s) i′G(s)

xgl

yq(s) xq
1
sCg
y ′G(s)

Figure 5.6  q-axis Equivalent Circuit: Choosing a State-variable (G-damper Winding Case).

In Figure 5.6,
TqÄ
Cg = (5.73)
x gl

Writing eq. (5.67) in per unit, we have
1 dy G
= − RG iG
w B dt

and in Laplace-domain,
RG
y G ( s) = −iG ( s) wB
s

R
From the above equation, it can be seen that iG ( s) G w B denotes the equivalent voltage drop
s
across the operational reactance 1 in Figure 5.6. Therefore, choosing the state-variable y ÄG as in
sC g
Figure 5.6, we have

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 217 6/2/2018 3:56:57 PM

 218  |  Power System Analysis

y ′G ( s) − y q ( s)
iG′ ( s) =
x gl

Analogous to capacitor current, for the circuit in Figure 5.6, we can write,
1
y ′G ( s) = −iG′ ( s) × (5.74)
sC g

Rearranging the terms, taking Laplace-inverse and substituting for Cg from eq. (5.73), we get,

 Tq′  dy ′ −y ′G + y q
 
G
= −iÄG =
 x gl  dt x gl

Therefore, the differential equation governing the dynamics of the q-axis circuit is given by,

dy ′G 1
= ′  −y ′G + y q  (5.75)
dt Tq  

Similar to eq. (5.66), the corresponding q-axis algebraic equation is given by,

 xq − xq′ 
yq = xq′ iq +  y ′ (5.76)
 xq  G

5.4  SUMMARY OF EQUATIONS PERTAINING TO 1.1 MODEL

For 1.1 model of the synchronous machine, the complete set of electrical equations are summarised
below:
Stator Equations:

1 dy 0
v0 = −i0 R0 − (5.77)
w B dt

 w  1 dy d
vd = −id Ra −   yq − (5.78)
wB  w B dt

 w  1 dy q
vq = −iq Ra +   yd − (5.79)
wB  w B dt

where w represents the rotor speed in electrical rad/s.
Rotor equations:
dy ′ F 1  xd′ 
= Ä  −y ′ F + y d + ′
E fd  (5.80)
dt Td  xd − xd 

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 218 6/2/2018 3:57:01 PM

 The Standard Parameters of Synchronous Machine  |  219 

dy ′G 1
= ′  − y ′G + y q  (5.81)

dt Tq  

Algebraic equations:

 x − xÄd 
y d = xÄd id + y ′ F  d  (5.82)
 xd 

 xq − xÄq 
yq = xÄq iq +y ′G   (5.83)
 xq 

Ä
Define q-axis transient internal voltage, Eq as

( xd − xÄd )
EÄq = y ′F (5.84)
xd

and d-axis transient internal voltage, EÄd as

 xq − xÄq 
EÄd = −   y ′ (5.85)
 xq  G

In the above equation, note that a negative sign is introduced for the voltage as the resulting voltage
due to y ÄG is along the negative d-axis. Also, this facilitates easy interpretation of phasor voltages.
Therefore, from eqs (5.82) and (5.83) we have,

y d = xÄd id + EÄq (5.86)

y q = xÄq iq − EÄd (5.87)

Notes:
• The machine parameters such as TdÄ , TqÄ , xd, xq, xÄd and xÄq are generally obtained from stand-
still frequency response (SSFR) tests [6, 1]. In this method, test is conducted when the machine
is at standstill. The machine is energised using a variable frequency voltage source to obtain the
data points pertaining to the responses between voltage and stator currents, and the field voltage
changes for both d- and q-axes. These data points are related through a transfer function by a
curve-fitting process to provide the standard transfer functions of the machines. From these trans-
fer functions, the standard parameters are derived.
• Only the d-axis parameters may be determined by conducting a short-circuit test on an unloaded
synchronous machine [9].
• EÄq and EÄd notations facilitate us to develop an equivalent circuit for the generator which per-
mits easy interfacing of the machine to the network in a multi-machine environment.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 219 6/2/2018 3:57:07 PM

 220  |  Power System Analysis

5.5  A CASE STUDY: VOLTAGE BUILD-UP ON NO-LOAD

In the following lines, a time-domain simulation of voltage-build up case is carried out to demonstrate
the usage of 1.1 model:
1. The stator voltage expressions given in eqs (5.78) and (5.79) are rewritten as follows with
w = w B:

dy d
= − Raw B id − w By q − w B vd (5.88)
dt

dy q
= − Raw B iq + w By d − w B vq (5.89)
dt

Note: If mechanical equations are modelled, the equations are effectively disabled by setting the
inertia of the machine very large, (say 10,000 s) so that the speed deviation is negligible.
2. The rotor differential equations are identical to that given in eqs (5.80) and (5.81).
3. The initial condition on all states is set equal to zero to signify that the machine is initially
unenergised.
4. Having solved for y ÄF and y ÄG , we can determine EÄq and EÄd .
5. Now using eqs (5.86) and (5.87), the machine currents are obtained as:

(y d − EÄq )
id = (5.90)
xÄd

(y q + EÄd )
iq = (5.91)
xÄq

While solving the stator differential equations, set vd = RL id and vq = RL iq with RL = 100 pu.
Further, note that the machine currents are functions of the state variables when the stator tran-
sients are accounted, i.e., when yd and yq are treated as state variables.
6. At t = 0 s, the Efd is stepped up to 1.0 pu, signifying energisation of the field winding.

For a typical 1.1 model, parameters, xd = 1.790, xÄd = 0.169, Tdo Ä = 4.30 s, x = 1.710, Ä = 0.228, TÄ
q xq qo
= 0.850 s, Ra = 0.01, 60 Hz, the vq(t) and vd(t) plots are shown in Figure 5.7. From Figure 5.7, it can
be seen that vq(t) tends to 1.0 pu, whereas, vd (t) is very small (since RL is finite, otherwise it is zero) as
steady-state is reached. vq(t) builds up exponentially, with a settling time of the order of 20 s. This is
Ä .
approximately equal to 5 times Tdo
Using the Park-inverse transformation, the va(t) is obtained and is plotted in Figure 5.8. From the
2
plot, it is clear that the steady-state peak amplitude is equal to × E fd 0 = 0.81649 pu, and it corre-
3
sponds to 60 Hz sinusoidal wave.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 220 6/2/2018 3:57:12 PM

 The Standard Parameters of Synchronous Machine  |  221 

1
0.8

vq (pu)
0.6
0.4
0.2
0
0 5 10 15 20 25 30
0

−0.005
vd (pu)

−0.01

−0.015

−0.02
0 5 10 15 20 25 30
Time (s)

Figure 5.7  vq(t) and vd (t) Plots for Voltage Build up Case-1.1 Model.

Voltage build up for 1.1 model.

1

0.8

0.6

0.4

0.2
va(t) pu

−0.2

−0.4

−0.6

−0.8

−1
0 5 10 15 20 25 30
Time (s)

Figure 5.8  va(t) for Voltage Build up Case-1.1 Model.

5.6  OPERATIONAL IMPEDANCE FOR 2.2 MODEL

In this case, all damper windings are considered in the rotor circuits. Therefore, there are two wind-
ings, i.e., F-coil and H-coil, along the d-axis, and two windings, i.e., G-coil and K-coil, along the

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 221 6/2/2018 1:05:53 PM

 222  |  Power System Analysis

q-axis. Hence, this model is referred to as the 2.2 model. Here, operational inductance expressions are
separately derived for d- and q-axes in the following sections.

5.6.1  Operational Impedance for d-axis

The flux linkage equations are given by

y d   Ld M dF M dH  id 
      (5.92)
y F  =  M dF LF LFH  i F 
y H   M dH LFH LH  iH 

Also, the voltage equations for the d-axis rotor coils are given by

dy F
v F = iF RF +
dt
dy H
0 = −iH RH −
dt

From the above differential and algebraic equations, in the s-domain, we can write the following func-
tional relationships:

iF ( s) = F1 (id ( s), iH ( s), v F ( s) ) (5.93)

iH ( s) = F2 (id ( s), iF ( s) ) (5.94)

Using eq. (5.94) in eq. (5.93) in place of iH(s) and rearranging, we get

iF ( s) = F3 (id ( s), v F ( s) ) (5.95)

Using the above expression in eq. (5.94) in place of iF (s), we get

iH ( s) = F4 (id ( s), v F ( s) ) (5.96)

Now using eqs (5.95) and (5.96) in yd (s), we get
M dF v F ( s)
y d ( s) = Ld ( s)id ( s) + G( s) (5.97)
RF

where

(1 + sa1 + s2 a2 )
Ld ( s) = Ld (5.98)
(1 + sb1 + s2 b2 )
ê
(1 + sTdc )
G ( s) = (5.99)
(1 + sb1 + s2 b2 )

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 222 6/2/2018 1:05:57 PM

 The Standard Parameters of Synchronous Machine  |  223 

Writing eq. (5.97) in per-unit, we have

y d ( s) = X d ( s)id ( s) + G( s) E fd ( s) (5.100)

Similar to the expression of Ld (s), derived for 1.1 model, (see expression (5.23)), we can factorise the
numerator and denominator polynomials in eq. (5.98) as

(1 + sa1 + s2 a2 ) = (1 + sc1 )(1 + sc2 )

(1 + sb1 + s2 b2 ) = (1 + sc3 )(1 + sc4 )

Note that the coefficients c1 to c4 are obtained as the inverse of the negative of the roots of the respec-
tive polynomials. It is seen that the roots of the polynomials are real values and possess the dimension
of inverse of time, i.e., s–1. Further, the respective root-pairs are quite distantly placed in the com-
plex-plane. These observations permit us to approximate the response of a second order system by
a combination of two first-order systems since it is generally not straightforward to estimate c1 to c4
starting from the basic parameters. A frequency–response technique (such as SSFR test), is usually
employed to determine the time-constants. Therefore, Xd (s) and G(s) are normally written in terms of
short-circuit and open-circuit time-constants as follows:

(1 + sTdÄ )(1 + sTdê )

X d ( s ) = xd Ä ê
(5.101)
(1 + sTdo )(1 + sTdo )
ê
(1 + sTdc )
G ( s) = Ä ê
(5.102)
(1 + sTdo )(1 + sTdo )

From the literature, the following observations are made:

1. The Xd (s) and G(s) are referred to as the standard transfer function.
ê is not generally specified along with the parameter
2. For an industry-grade generator model, Tdc
set. Hence, in most of the standardised generator models (see section 5.1.2), no additional effort is
ê in the modelling process. Therefore, in the industry-grade power
made to accurately account Tdc
system analysis packages, the performance of generator models, to some extent, depends on the
degree of fitment of the standard transfer functions through its state-space model.
ê is of the same order as that of ê in the OI
3. Since it is found that the numerical value of Tdc Td
ê = ê Such an approximation enables us to
model presented in this chapter, it is assumed that Tdc Td .
choose state variables which can be directly correlated to the original machine variables [5].
Therefore, we have,
(1 + sTdê )
G ( s) = Ä
(5.103)
(1 + sTdo )(1 + sTdê 0 )

From eq. (5.100), it can be written that

y d ( s) G ( s)
id ( s) = − E fd ( s) (5.104)
X d ( s) X d ( s)

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 223 6/2/2018 1:06:01 PM

 224  |  Power System Analysis

1
Observing the structure of in eq. (5.50) for 1.1 model, and as per the IEEE standard 115
(2009) [3], it is written that X d ( s)

1 1  1 1  sTdÄ  1 1  sTdê
= + Ä −  +  −  (5.105)
X d ( s) xd  xd xd  1 + sTdÄ  xêd xÄd  1 + sTdê

1
Note: The above expression for is employed to fit the peak envelope of the short-circuit fun-
X d ( s)
damental current waveform of an unloaded generator [9].
Defining
xd xÄd xÄd xêd
x fl = and xhl = (5.106)
xd − xÄd xÄd − xêd
1
Rewriting for 2.2 model, we get
X d ( s)

1 1 sTdÄ sTdê
= + + (5.107)
X d ( s) xd x fl (1 + sTdÄ ) xhl (1 + sTdê )

where
ê
Tdo = d - axis subtransient open-circuit time-constant.
Tdê = d - axis subtransient short-circuit time-constant.
xêd = d - axis subtransient reactance.

5.6.1.1 Relationship between Open- and Short-circuit d-axis

Time-constants
From eq. (5.107), we can write that

xd x fl xhl (1 + sTdÄ )(1 + sTdê )

X d ( s) =
x fl xhl (1 + sTdÄ )(1 + sTdê ) + xd xhl (1 + sTdê ) sTdÄ + xd x fl (1 + sTdÄ ) sTdê

N ( s)
= (5.108)
D ( s)

Now consider the denominator polynomial D(s)

( ) (
D( s) = x fl xhl 1 + (TdÄ + Tdê ) s + TdÄ Tdê s2  + xd xhl sTdÄ + s2TdÄ Tdê + xd x fl sTdê + s2TdÄ Tdê
  )
= x fl xhl + x fl xhl (TdÄ + Tdê ) s + x fl xhl TdÄ Tdê s2 + xd xhl TdÄ s + xd xhl TdÄ Tdê s2 + xd x f l ( sTdê + s2TdÄ Tdê )

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 224 6/2/2018 1:06:06 PM

 The Standard Parameters of Synchronous Machine  |  225 

D( s) = s2  xd x fl TdÄ Tdê + xd xhl TdÄ Tdê + x fl xhl TdÄ Tdê 

 
+ s  x fl xhl (TdÄ + Tdê ) + xd xhl TdÄ + xd x fl Tdê  + [ x fl xhl ]
 

Further, we have
 x x 
D( s) =  x fl xhl   s2  d TdÄ Tdê + d TdÄ Tdê + TdÄ Tdê 
  xhl x fl 
 x x  
+ s (TdÄ + Tdê ) + d TdÄ + d Tdê  + 1 (5.109)
 x fl xhl  

Using the above equation in eq. (5.108) and comparing the resulting denominator with that in
eq. (5.101), we have

Ä ê
 x   xd 
Tdo + Tdo = TdÄ 1 + d  + Tdê 1 + x  (5.110)
 x fl  hl

Ä ê
x x 
TdoTdo = TdÄ Tdê  d + d + 1 (5.111)
 xhl x fl 

Substituting for xf l and xh l from eq. (5.106) in the above equations, we have the following relations:

xd  x x 
TdÄ + Tdê 1 − Äd + êd  = Tdo
Ä ê
+ Tdo (5.112)
xÄd  x d xd 

x 
TdÄ Tdê  êd  = Tdo
Ä ê
Tdo (5.113)
 xd 

Further, note that
Ä
Tdo > TdÄ > Tdo
ê
> Tdê (5.114)

Notes:
• Using standstill frequency response tests, Xd (s) (see eq. (5.101)) can be directly obtained by
employing curve fitting techniques. It provides xd , TdÄ , Tdê , Tdo
Ä ê
, and Tdo .
• For 2.2 model, xêd can be evaluated from eq. (5.113) as
 TÄ T ê 
xêd = xd  Äd dê 
 TdoTdo 
In other words, xêd is obtained as the high-frequency gain of the transfer function Xd(s).

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 225 6/2/2018 1:06:11 PM

 226  |  Power System Analysis

• Further, xêd can also be related to the basic parameters as shown below [8], though it is of little
importance:

 xdF
2 2
x H + xdH x F − 2 xdF xdH x FH 
xêd = xd −  
( 2
x F x H − x FH )
• For 2.2 model, there is no exact definition for xÄd in terms of the basic parameters like in 1.1
model. Its value can be evaluated only numerically from eq. (5.112).
ê
• The assumption Tdc = Tdê enables the state variable y ÄF to be directly mapped to the original
field flux-linkages. Similarly, we can relate y ÄH to the H-coil flux-linkages.

5.6.1.2  Derivation of State-equations for d-axis Rotor Circuits

From eq. (5.104), we have
1 G ( s)
id ( s) = y d ( s) − E fd ( s) (5.115)
X d ( s) X d ( s)

1
Substituting for and G(s) in the above equation, we get
X d ( s)

1 sTdê sTdÄ  E fd ( s)
id ( s) = y d ( s)  + + − (5.116)
 xd xhl (1 + sTd ) x fl (1 + sTd )  (1+ sTdÄ ) xd
ê Ä

Similar to the procedure given in Section 5.3.4.1 for 1.1 model, an equivalent circuit can be developed
for d-axis accounting the damper coil H as shown in Figure 5.9.
1
id(s) xfl sCf

i′F(s)
xhl +
yd(s) i′H(s)
xd y ′F(s) E = Efd (s) xfl 1
1 xd sT ′d
−
y ′H(s) sCh

Figure 5.9  d-axis Equivalent Circuit Including the H-damper Winding.

The parameters in Figure 5.9 are defined as follows:

x fl E fd ( s)
E= (5.117)
xd sTdÄ

TdÄ
Cf = (5.118)
x fl

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 226 6/2/2018 1:06:15 PM

 The Standard Parameters of Synchronous Machine  |  227 

Tdê
Ch = (5.119)
xhl
From the equivalent, choosing appropriate state variables, the differential equations are written in the
state-space form as follows:

dyÄ F 1  xÄd E fd 
= Ä  −yÄ F + y d +  (5.120)
dt Td  ( xd − xÄd ) 

dyÄ H 1
= ê [ −yÄ H + y d ] (5.121)
dt Td

5.6.1.3  Algebraic Equation for d-axis

In order to derive the algebraic relationship for the d-axis, the current id can be expressed as

id = F1 (y d , yÄ F , yÄ H ) (5.122)

The above equation can also be written as

id = id1 + id 2 + id 3 (5.123)

where
id1 = id |yÄ = 0, yÄ H = 0
F (5.124)
id 2 = id |y = 0, yÄ H = 0 (5.125)
d

id 3 = id |y = 0, yÄ F = 0 (5.126)
d

Following the similar steps as in 1.1 model and rearranging the terms, we get

xêd xê
y d = xêd id + yÄ H + d yÄ F (5.127)
xhl x fl

Substituting for xf l and xh l from eq. (5.106) in eq. (5.127)

( xÄd − xêd )  x − xÄd   xêd 

y d = xêd id + yÄ H +  d    yÄ F (5.128)
xÄd  xd   xÄd 

5.6.2  Operational Impedance for q-axis

Employing the procedure as in the d-axis case, we have the following results:

y q ( s) = X q ( s)iq ( s) (5.129)

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 227 6/2/2018 1:06:19 PM

 228  |  Power System Analysis

where

(1 + sTqÄ )(1 + sTqê )

X q ( s ) = xq Ä ê
(5.130)
(1 + sTqo )(1 + sTqo )

or
Ä ê
1 1  1 1  sTq  1 1  sTq
= + Ä −  +  −  (5.131)
X q ( s) xq  xq xq  1 + sTqÄ  xêq xÄq  1+ sTqê

Defining

xq xÄq xÄq xêq

x gl = and xkl = (5.132)
xq − xÄq xÄq − xêq

where
ê
Tqo = q - axis subtransient open-circuit time-constant.
Tqê = q - axis subtransient short-circuit time-constant.
xêq = q - axis subtransient reactance.

5.6.2.1 Relationship between Open- and Short-circuit q-axis

Time-constants
From eq. (5.131), we can write that

xq x gl xkl (1 + sTqÄ )(1 + sTqê )

X q ( s) = (5.133)
x gl xkl (1 + sTqÄ )(1 + sTqê ) + xq xkl (1 + sTqê ) sTqÄ + xq x gl (1 + sTqÄ ) sTqê

Having simplified and comparing the like coefficients of s2 and s in the denominators of the above
equation and eq. (5.130), we write that

xq  xq xq 
Ä ê
TqÄ + Tqê 1 − Ä + ê  = Tqo + Tqo (5.134)
xÄq  xq xq 

 xq 
Ä ê
TqÄ Tqê  ê  = TqoTqo (5.135)
 xq 

Further, note that
Ä
Tqo > TqÄ > Tqo
ê
> Tqê (5.136)

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 228 6/2/2018 1:06:23 PM

 The Standard Parameters of Synchronous Machine  |  229 

Notes:
• Only the d-axis parameters may be determined by conducting a short-circuit test on an unloaded
synchronous machine [9].
• Using standstill frequency response tests, Xq(s) (see eq. (5.130)) can be directly obtained by
employing curve fitting techniques. It provides xq , TqÄ , Tqê , Tqo
Ä ê
and Tqo .
• For 2.2 model, xêq can be evaluated from eq. (5.135) as

 TqÄ Tqê 
xêq = xq  Ä ê 
 TqoTqo 

In other words, xêq is obtained as the high frequency gain of the transfer function Xq(s).
• Further, xêq can also be related to the basic parameters as shown below [8], though it is of little
importance:

 xqG
2 2
x K + xqK xG − 2 xqG xqK xGK 
xêq = xq −  
2
( xG x K − xGK )

• For 2.2 model, there is no exact definition for xÄq in terms of the basic parameters like in 1.1
model. Its value can be evaluated only numerically from eq. (5.134).

5.6.2.2  Derivation of State-equations for q-axis Rotor Circuits

From eq. (5.129), we can write that
 1 
iq ( s) = y q ( s)   (5.137)
 X q ( s) 
1
Substituting for in the above equation we get
X q ( s)
1 sTqê sTqÄ 
iq ( s) = y q ( s)  + +  (5.138)
 xq xkl (1 + sTqê ) x gl (1 + sTqÄ ) 

We can obtain an equivalent circuit for q-axis similar to that for d-axis as shown in Figure 5.10.

iq(s) i′G(s)

xkl xgl
i′K(s)
yq(s) xq
1
1
sCg
sCk y ′K(s) y ′G(s)

Figure 5.10  q-axis Equivalent Circuit Including the K-damper Winding.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 229 6/2/2018 1:06:27 PM

 230  |  Power System Analysis

The parameters in Figure 5.10 are defined as follows:

TqÄ
Cg = (5.139)
x gl

Tqê
Ck = (5.140)
xkl

From the equivalent, choosing appropriate state variables, the differential equations are written in
the state-space form as follows:

dy ÄG 1
= Ä  −y ÄG + y q  (5.141)
dt Tq  

dy ÄK 1
= ê  −y ÄK + y q  (5.142)
dt Tq  

5.6.2.3  Algebraic Equation for q-axis

In order to derive the algebraic relationship for the q-axis, the current iq can be expressed as

iq = F2 (y q ,y ÄG ,y ÄK ) (5.143)

Following the similar steps as in the d-axis case, we get

xêq xêq
y q = xêq iq + y ÄK + y ÄG (5.144)
xkl x gl

Substituting for xgl and xkl from eq. (5.132) in the above equation, we get

( xÄq − xêq )  xq − xÄq   xêq 

y q = xêq iq + y ÄK +  Ä
   y (5.145)
xÄq  xq   xÄq  G

5.7  SUMMARY OF EQUATIONS PERTAINING TO 2.2 MODEL

For 2.2 model of the synchronous machine, the complete set of electrical equations are summa-­
rised below.
Stator Equations:

1 dy 0
v0 = −i0 R0 − (5.146)
w B dt
 w  1 dy d
vd = −id Ra −  yq − (5.147)
 w B  w B dt

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 230 6/2/2018 1:06:31 PM

 The Standard Parameters of Synchronous Machine  |  231 

 w  1 dy q
vq = −iq Ra +   yd − (5.148)
wB  w B dt

where w represents the rotor speed in electrical rad/s.
Rotor Equations:

dy ÄF 1  xÄd 
= Ä  −y ÄF + y d + E fd  (5.149)
dt Td  xd − xÄd 

dy ÄH 1
= ê  −y ÄH + y d  (5.150)

dt Td  

dy ÄG 1
= Ä  −y ÄG + y q  (5.151)
dt Tq  

dy ÄK 1
= ê  −y ÄK + y q  (5.152)
dt Tq  

Algebraic Equations:

( xÄd − xêd )  x − xÄd   xêd  Ä

y d = xêd id + y ÄH +  d    y F (5.153)
xÄd  xd   xÄd 

( xÄq − xêq )  xq − xÄq   xêq 
y q = xêq iq + y ÄK +  Ä
   y (5.154)
xÄq  xq   xÄq  G

Define q-axis subtransient internal voltage, Eêq as

 ( xÄ − xê )  x − xÄd   xêd  Ä 
Eêq =  d Ä d y ÄH +  d    y F  (5.155)
 xd  xd   xÄd  

and d-axis subtransient internal voltage, Eêd as

 ( xÄ − xê )  xq − xÄq   xêq  
q q
Eêd = −  y Ä
K +     y Ä 
G (5.156)

 xÄq  xq   xÄq  


Therefore, from eqs. (5.153) and (5.154) we have,
y d = xêd id + Eêq (5.157)

y q = xêq iq − Eêd (5.158)

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 231 6/2/2018 1:06:36 PM

 232  |  Power System Analysis

Notes:
• It is also noted that in general the following relationship is true among the reactances.

xd ≥ xq > xÄq ≥ xÄd > xêq ≥ xêd

• Eêq and Eêd notations facilitate us to develop an equivalent circuit for generator which permit
easy interfacing of the machine to the network.
• Stator voltage equations are used as they are given in eqs. (5.147) and (5.148) in such studies
where dynamics of interest involve frequency of oscillations in the range close to that of the net-
work dynamics. For example, in torsional oscillation studies where the frequency of oscillations
lies in the range of 10 to 40 Hz, the stator/network transients are retained.
• In electromechanical oscillation or power swing -related stability studies where the frequency of
oscillations lies in the range of 0.1 to 3 Hz, the stator transients can be neglected without appreci-
ably influencing the accuracy of the desired results. This simplification also removes the stiffness
of the system, enabling the choice of higher step-size for numerical integration of system of dif-
ferential equations. When such an assumption is made, it is also found that the generator currents
in abc-frame will have only fundamental component making the equations compatible with the
network equations where they are assumed to be in (quasi) sinusoidal steady-state. The resulting
stator voltage equations appear as follows:

 w 
vd = −id Ra −  y q (5.159)
 w B 

 w 
vq = −iq Ra +  y d (5.160)
 w B 

In the above equation, note that w is the actual rotor speed in elec-rad/s and is given by

dd
w = wo + = w o + Dw
dt
When stator transients are neglected, it is customary to neglect even the speed deviations, Dw in com-
parison with wo so that w ≈ wo. Assuming that wo = wB, the above voltage equations are modified as

vd = −id Ra - y q (5.161)

vq = −iq Ra + y d (5.162)

5.8  SWING EQUATION

The equations of central importance in power system stability analysis are the rotational inertia
equations describing the effect of imbalance between the electromagnetic torque and the mechan-
ical torque of the individual machines. In this section, we derive the swing equation in per unit
form and define parameters that are used to represent mechanical characteristics of synchronous
machines in stability studies.
From Newton’s second law of motion, we can write that

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 232 6/2/2018 1:06:39 PM

 The Standard Parameters of Synchronous Machine  |  233 

dw r
J = − Dw r + Tm − Teg (5.163)
dt
where
    J = combined moment of inertia of generator and turbine, kg - m2
   wr = angular velocity of the rotor, mech - rad/s
    D = damping torque coefficient, N−m
mech − rad / s
and
Tm = mechanical torque in applied to the shaft of the generator in N-m
Teg = actual electromagnetic torque developed in N-m
Multiplying both sides of eq. (5.163) by the base rotor angular speed, wrB, we get
dw r
J w rB = − Dw rBw r + Tmw rB − Teg w rB (5.164)
dt

5.8.1  Swing Equation in Form-1

Dividing the above equation by the base power SB, of the machine, we have,

dw r
J w rB
dt = − Dw rBw r + Tmw rB − Teg w rB (5.165)
SB SB SB SB

Further, we have
  wr  
 2 d w   2 
 Dw rB Teg
1 Jw  rB   wr T
2 ×  rB = −  + m − (5.166)
2  SB dt   S B  w rB SB SB
  w w
  rB rB

The above equation can be rewritten as
w 
d r 
 w rB  wr (5.167)
2H = −D + Tm − Teg
dt w rB

where H is known as inertia constant, defined as the kinetic energy stored in MJ at rated speed divided
by the MVA base.
2
1 J w rB
H= in MJ/MVA (5.168)
2 SB
SB
Tm and Teg are expressed in per-unit. TrB = = Base torque for the rotor.
w rB

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 233 6/2/2018 1:06:43 PM

 234  |  Power System Analysis

2
D w rB
D= is the per-unit damping torque/per-unit speed.
SB
Notes:
• The inertia constant H has the dimension of time expressed in seconds. H varies in a narrow range
(2–10) for most of the machines irrespective of their ratings.
• The rated angular velocity of the rotor in elec-rad/s, wB, is related to wrB by an expression:

 P
w B = w rB   (5.169)
 2

Similarly, the angular velocity of the rotor in elec-rad/s, w, is related to wr by an expression:

w = w r   (5.170)
P
 2

• Making use of the above equations, it can be shown that

wr w
= (5.171)
w rB w B

Using eq. (5.171) in eq. (5.167), we can also write that

 w 
d
 w B  w (5.172)
2H = −D + Tm − Teg
dt wB

To determine the rotor angle, consider

q = w ot + d (5.173)

Taking the time derivative, we have

dq dd (5.174)
= w = wo +
dt dt

Rewriting the above equation, we get

dd
= w − wo (5.175)
dt
or
dd  w 
= −1 w B (5.176)
dt  w B 

with w o = w B .

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 234 6/2/2018 1:06:46 PM

 The Standard Parameters of Synchronous Machine  |  235 

5.8.2  Swing Equation in Form-2

From eq. (5.176), we can write that
 w 
d
 w B  1 d 2d (5.177)
=
dt w B dt 2

Using the above result in eq. (5.172), we write
2 H d 2d  1 dd 
2
= −D  + 1 + Tm − Teg (5.178)
w B dt  w B dt 

5.8.3  Swing Equation in Form-3

Defining the slip speed of the machine in per unit as
 w 
Sm =  − 1 (5.179)
wB 

From eq. (5.176), we have

dd
= Smw B (5.180)
dt
From eq. (5.179), we can write that
 w 
d
dSm  w B  (5.181)
=
dt dt
Using the above result in eq. (5.172), we have
dSm
2H = − D (1 + Sm ) + Tm − Teg (5.182)
dt

or
dSm Ä
2H = − D Sm + T m − Teg (5.183)
dt
Ä
where T mm = Tm − D .

—
5.8.3.1  Expression for Air-gap Torque in Per-unit, Teg
The expression for actual torque developed corresponding to the rotor speed w r (mech-rad/s) is given by

Teg =
P
2
(
y d iq −y q id)
(5.184)

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 235 6/2/2018 1:06:50 PM

 236  |  Power System Analysis

We know that
Teg
Teg =
SB (5.185)

w rB

Using eq. (5.184) in the above equation, we get

Teg =
(y d iq − y qid )
2  (5.186)
 w B  y B I B
P
w rB

Now, making use of relation (5.169) in the above equation, we have
Teg = y d iq −y q id (5.187)

5.8.4  Summary of Different Forms of Swing Equations

Form-1:

dd  w 
= − 1 w B (5.188)
dt  w B 

 w 
d
 w B  w (5.189)
2H = −D + Tm − Teg
dt wB

Form-2:

dd  w 
= − 1 w B (5.190)
dt  w B 

2 H d 2d  1 dd 
= −D  + 1 + Tm − Teg (5.191)
w B dt 2  w B dt 

Form-3:
dd
= Sm w B (5.192)
dt
dSm Ä
2H = − D Sm + Tm − Teg (5.193)
dt

Notes:
• We know that the mechanical power input is given by
Pm = Tm w r

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 236 6/2/2018 1:06:54 PM

 The Standard Parameters of Synchronous Machine  |  237 

In per-unit,

Pm T  w w
= Pm =  m  r = Tm
SB  TrB  w rB wB

• The electrical power generated at the air-gap is given by

Peg = Teg w r

Further, under steady-state, Peg = PLe + I 2 Ra where PLe represents the real power output and I 2 Ra
denotes the total armature copper losses.
In per-unit
Peg  Teg  w r w
= Peg =   = Teg
SB  TrB  w rB wB

However, in most cases, it is customary to neglect the speed deviation in the stator voltage equa-
tions, see eqs (5.161) and (5.162). To be consistent with this, we set w = w B in the above expres-
sions to get,
Pm = Tm (5.194)

Peg = Teg (5.195)

• Using the results given in eqs (5.157) and (5.158) in eq. (5.187), the expression for per unit elec-
tric torque can also be written as

( ) (
Teg = xêd id + Eêq iq − xêq iq − Eêd id )

( )
= xêd − xêq id iq + Eêq iq + Eêd id
(5.196)

As id and iq are functions of state variables, Teg becomes a function of the product of state ­variables.
This, in turn, makes the swing equation non-linear.

5.9  STEADY-STATE OPERATION OF SYNCHRONOUS GENERATOR

The equations corresponding to steady-state operation can be derived by setting all the derivatives
equal to zero in the 2.2 model differential equations.
Consider,
dd  w w  (5.197)
= − o wB
dt  w B w B 

dd w w
Setting = 0, we have = o =1
dt wB wB

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 237 6/2/2018 1:06:59 PM

 238  |  Power System Analysis

dy d dy q
Setting the derivatives in eqs (5.147) and (5.148), i.e., and , to zero and using the
w dt dt
identity = 1 , we have,
wB
vd 0 = −id 0 Ra −y q0 (5.198)

vq0 = −iq0 Ra +y d 0 (5.199)

dyÄ F dyÄ H
By setting = 0 and = 0 , in eqs (5.149) and (5.150), respectively, we have
dt dt

xÄd
y F 0 = y d0 + E fd 0
xd − xÄd
x fl
= y d0 + E fd 0 (5.200)
xd

and

y H 0 = y d0 (5.201)

Substituting eqs. (5.200) and (5.201) in eq. (5.127), we get

xd″ x″  x fl 
y d 0 = xd″ id 0 + y d0 + d y d 0 + x E fd 0 
(5.202)
xhl x fl d

Writing the above equation after rearranging the terms and substituting for xfl and xhl from eq. (5.106),
we get

 x′ − x″  x − x′   x″   xd″
″
y d 0 1− d ′ d −  d d
 
d
 = x i
d d0 + E fd 0 (5.203)
 xd  xd   xd′   xd

Consider the LHS term,

 x′ − x″  x − x′   x″  
y d 0 1− d d
− d d d
xd xd′ − xd xd′ − xd″ − xd − xd′ xd″ ( ) ( )
    = y d0
 xd′  xd   xd′   xd xd′

xd″
= y d0 (5.204)
xd

Using the above result in eq. (5.203), we have

y d 0 = xd id 0 + E fd 0 (5.205)

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 238 6/2/2018 1:07:05 PM

 The Standard Parameters of Synchronous Machine  |  239 

Similarly, we can show that

y q0 = xq iq0 (5.206)

Substituting eqs (5.205) and (5.206) in eqs (5.198) and (5.199), we get

vd 0 = −id 0 Ra − xq iq0 (5.207)

vq 0 = −iq0 Ra + xd id 0 + E fd 0 (5.208)

For a compact notation, combining the above equations in a ‘vector’ form, we get

( )
vq0 + jvd 0 = − Ra iq0 + jid 0 + xd id 0 + E fd 0 − jxq iq0 (5.209)

We know that the rotor-based quantities are related to the synchronous-frame-based quantities as

( )
f Q + jf D = f q + jf d e jd (5.210)

Notes:
• In the above equation, fQ and fD represent a phasor quantity in the Kron’s or the synchro-
nous-frame-of-reference. It denotes a balanced three-phase system in the abc-frame. A constant
fQ and fD represents the fundamental sinor steady-state condition in the abc-frame. In rotor-angle
stability studies, the network is assumed to be at a quasi-sinor steady-state where the phasor quan-
tities are assumed to exhibit oscillations at a very low frequency in the range of 0.1 to 3 Hz. This
assumption is employed in power- (rotor) swing analysis.
• In a multi-machine environment, while interfacing a generator to the network, the generator quan-
tities are transformed to the synchronous-frame-of-reference using the corresponding rotor angle
d. This ensures that machine variables of all generators are represented in the common reference
frame, i.e., the synchronous-frame-of-reference, before the KVL and KCL laws are applied.
Using the above relationship, the quantities defined in the rotor-frame-of-reference in eq. (5.209) are
expressed in the synchronous-reference-frame as
Vt 0 = (vq0 + jvd 0 )e jd 0

= (vQ 0 + jvD 0 )

Vt 0 = − Ra (iq0 + jid 0 ) e jd 0 + xd id 0 e jd 0 + E fd 0 e jd 0 − jxq iq0 e jd 0 (5.211)

Defining
I 0 = (iq0 + jid 0 )e jd 0

= iq0 e jd 0 + jid 0 e jd 0 (5.212)

Further, denoting I aq = iq0 e jd 0 and I ad = jid 0 e jd 0 we have,

I 0 = I aq + I ad (5.213)

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 239 6/2/2018 1:07:11 PM

 240  |  Power System Analysis

Using the above definition in eq. (5.211), we get

Vt 0 = − Ra I 0 − jxd I ad − jxq I aq + E fd 0

where E fd 0 = E fd 0 e jd 0 .
The above equation can be rearranged to get

E fd 0 = Vt 0 + Ra I 0 + jxd I ad + jxq I aq (5.214)

which is applicable for salient-pole machines where xd ≠ xq .
However, for a round-rotor machine, where xd = xq = xs , the above equation simplifies as

( )
E fd 0 = Vt 0 + Ra I 0 + jxs I ad + I aq (5.215)

Using the definition (5.213), we have

E fd 0 = Vt 0 + ( Ra + jxs ) I 0 (5.216)

Efdo

Ra jXs
Io
jxsIo
do +
Fo Efdo∠d Vto∠0°
Vto
IoRa

Io

Figure 5.11  Phasor Diagram for Round- Figure 5.12  Equivalent Circuit Round-
rotor Synchronous Generator.      rotor Synchronous Generator.

The corresponding phasor diagram and the equivalent circuit are as shown in Figures 5.11 and 5.12,
respectively.

5.9.1 Graphical Approach to Determine the Rotor-angle

in Salient-pole Machines
From eq. (5.214), it can be seen that determining E fd0 is not straight-forward as it involves the
calculation of I aq and I ad , and in turn, they require the knowledge of the location of d- and q-axes
which are known only if E fd0 is available. In order to overcome this deadlock, the following graphical
procedure is employed:

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 240 6/2/2018 1:07:17 PM

 The Standard Parameters of Synchronous Machine  |  241 

Equation (5.209) is re-written by subtracting and adding the term xq id0 to the RHS,

(vq0 + jvd 0 ) = − Ra (iq0 + jid 0 ) + xd id 0 − xqid 0 + xqid 0 − jxqiq0 + E fd 0

= − Ra (iq0 + jid 0 ) + ( xd − xq ) id 0 − jxq jid 0 − jxq iq0 + E fd 0

= − Ra (iq0 + jid 0 ) + ( xd − xq ) id 0 − jxq (iq0 + jid 0 ) + E fd 0

Defining

( )
Eq0 = E fd 0 + xd − xq id 0 (5.217)

( )
and rewriting vq0 + jvd 0 equation, we have

( )(
vq0 + jvd 0 = − Ra + jxq iq0 + jid 0 + Eq0 (5.218) )
Now, writing the above equation in synchronous-frame, we have

( )
Vt 0 = − Ra + jxq I 0 + Eq0 Æd (5.219)

or

(
Eq0 = Eq0 ∠d = Vt 0 + Ra + jxq I 0 (5.220) )
Note that the above equation can be readily evaluated and Eq0 and E fd0 are collinear. Therefore,
Eq0 locates q- and d-axes. Once this is known, E fd0 can be easily obtained from eq. (5.214). The
corresponding phasor diagram is as shown in Figure 5.13. However, an equivalent circuit cannot be
drawn as that in the round rotor case.
d−axis
q−axis
Efdo
Eqo
jxqIaq

Iaq
jxqIo
do jxdIad
Fo Vto
IoRa

Io
Iad

Figure 5.13  Phasor Diagram for Salient-pole Synchronous Generator.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 241 6/2/2018 2:03:22 PM

 242  |  Power System Analysis

5.9.2 Example: Computation of Field Voltage Efd0 in a Salient-pole

Machine
Let Vt0 = 1.0 ∠0°, current delivered by the generator, I 0 = 1.0 ∠ − 45° with xd = 1.0 pu and xq = 0.6 pu.
Determine the field voltage, Efd0 using eq. (5.214).

1. Obtain Eq0 as

Eq0 = 1.0 ∠0° + j 0.6 × 1.0 ∠ − 45°

= 1.0 + 0.6∠45°
Eq0 ∠d 0 = 1.4243 + j 0.4243 = 1.486∠16.59° pu.
(5.221)
jd 0
2. Obtain I aq and I ad .=Tojiddetermine
0e this, first calculate iq0 and id0 as

iq0 + jid 0 = I 0 e − jd 0 = 1.0 ∠( −45° − 16.59°)

= 0.476 − j 0.880 (5.222)

Therefore, I aq = iq0 e jd 0 = 0.476 ∠16.59° and I ad = jid 0 e jd 0 = 0.880 ∠(-90° + 16.59°)
= 0.880 ∠-73.41°.
3. Now, using eq. (5.214), we get

E fd0 = 1.0 + j1.0 × 0.880 ∠ − 73.41° + j 0.6 × 0.476∠16.59°

= 1.7620 + j 0.5244 = 1.8384 ∠16.59°pu. (5.223)

5.9.3  Expression for Developed Torque Under Steady-state

Consider a case where a synchronous generator is connected to an infinite bus as shown in Figure 5.14.

E = E∠0
Vt∠q
jxL

Generator Peg

Figure 5.14  Single-machine Connected Infinite Bus System

Under steady-state, neglecting Ra, from eqs (5.207) and (5.208), we can write that

vd 0 = − xq iq0 (5.224)

vq0 = xd id 0 + E fd 0 (5.225)

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 242 6/2/2018 2:03:30 PM

 The Standard Parameters of Synchronous Machine  |  243 

− j (d 0 −q 0 )
Since we know that (vq0 + jvd 0 ) = Vt 0 e , or vq0 = Vt 0 cos(d 0 − q 0 ) and vd 0 = − Vt 0 sin(d 0 − q 0 ),
vd 0 = − Vt 0 sin(d 0 − q 0 ), we write an expression for iq0 and id0 as

Vt 0 sin(d 0 − q 0 )
iq0 = (5.226)
xq

Vt 0 cos(d 0 − q 0 ) − E fd 0
id 0 = (5.227)
xd

We know that
Teg = y d 0 iq0 −y q0 id 0

or

Teg = vq0 iq0 + vd 0 id 0 (5.228)

Using eqs (5.226) and (5.227) in the above equation in addition to the expression for vq0 and vd0,
we get

 V sin(d 0 − q 0 )   Vt 0 cos(d 0 − q 0 ) − E fd 0 
Teg = Vt 0 cos(d 0 − q 0 )  t 0  − Vt 0 sin(d 0 − q 0 )  
 xq   xd

Simplifying the above equation, we get

E fd 0 Vt 0 sin(d 0 − q 0 ) Vt20  1 1
Teg = +  −  sin 2 (d 0 − q 0 ) (5.229)
xd 2  xq xd 

Note:

• The second term in eq. (5.229) denotes the torque developed in the machine without the field
voltage and is generally referred to as the torque due to saliency.
• For a round rotor machine where xq = xd , eq. (5.229) simplifies as
E fd 0 Vt 0 sin(d 0 − q 0 )
Teg =
xd

• If the line reactance is absorbed into the machine reactances, we can define xd1 = xd + x L and
xq1 = xq + x L, and the infinite bus becomes the machine terminal. With these, eq. (5.229) modi-
fies as

E fd 0 E sind 0 E2  1 1 
Teg = +  −  sin 2d 0 (5.230)
xd 1 2  xq1 xd1 

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 243 6/2/2018 2:03:38 PM

 244  |  Power System Analysis

5.10  STEADY-STATE ANALYSIS: SPACE-PHASOR APPROACH

In this section, an alternate explanation is provided for a steady-state model of the synchronous
machine based on space-phasors [8].
Generally, it is assumed that the flux density B due to the field winding varies sinusoidally as
Bm sinw ot in the air-gap over a pole-pair. This sinusoidal space distribution is represented by a
Bm
space-phasor, Bave = , which rotates in the air-gap at speed w 0 , . Hence, the space-phasor
(p / 2)
flux-­linkage aligned along the d-axis has a form given by

l f = (faveT ) ∠0° (5.231)

where

fave = Bavet p L ,

T represents the number of turns per phase,

t p = pole-pitch, and L = active length of the rotor core.
Due to this space-phasor, the instantaneous flux-linkage of phase-a coil is given by

l f = (faveT ) cos q (5.232)

Note that q is the angular position of the d-axis with respect to the stationary magnetic axis of phase-a
coil andq = w ot at any time t.
Now, the instantaneous value of induced voltage in phase-a is given by,

dl f d
e=−
dt
=−
dt
{ }
(faveT ) cosw ot 
= (faveT w o ) sinw ot 

 p
e = Emax cos  w ot −  (5.233)
 2

where Emax = ( 2pf ofaveT ) .

The RMS value of voltage per phase is given by

Emax
Erms = = 4.44 f ofaveT (5.234)
2

p
Therefore, like a space-phasor, time-phasor can be written as E = Erms ∠ − taking flux-linkage
2
space-phasor as reference. Considering an unloaded generator, a hybrid phasor diagram can be drawn
as shown in Figure 5.15.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 244 6/2/2018 2:03:44 PM

 The Standard Parameters of Synchronous Machine  |  245 

d−axis

lf

E
q−axis

Figure 5.15  Hybrid Phasor Diagram Under No-load.

5.10.1  Space-phasor Due to Armature Currents-Revolving Field

Having made the machine to build voltage E at its terminals, a connection of a three-phase balanced
load circulates armature current iabc through its windings. Now, consider phase-a winding alone to
analyse for MMF due armature current ia. The MMF distribution in the air-gap due a concentric and
full-pitch winding, for a two-pole machine, is shown in Figure 5.16.
Now, the fundamental component of air-gap MMF per pole pertaining to phase-a is obtained as
 4Tia 
Fag1 =  cos( b )(5.235)
 p 2 
where b is the space-angle which varies between 0 and 2p elec. rad. over a pole pair.

Magnetic axis of phase−a coil

Fag1
Tia
2
−p p
0 2 2
0 p 2p
−Tia b
2

Rotor surface
Stator surface

Figure 5.16  MMF Distribution in the Air-gap Due Phase-a coil.

If phase-a winding is spatially spread over P poles, the MMF expression becomes

 4Tia 
Fag1 =  cos( b ) (5.236)
 pP 

where the phase-a current is assumed to be of the form, ia = I m cos(w ot + f ) .

Similarly, the MMF expression for phase-b, is given by,

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 245 6/2/2018 2:03:47 PM

 246  |  Power System Analysis

 4Tib   2p 
Fbg1 =  cos  b −  (5.237)
 pP  3 
and for phase-c, it is given by,

 4Ti   2p 
Fcg1 =  c  cos  b +  (5.238)
 pP   3 

 2p  2p
where, ib = I m cos  w ot − + f  and ic = I m cos  w ot + 
+f .
 3   3 

It can be seen from eq. (5.236) that for every time t, ia (t ) = ia1 , and b varies between 0 and 2p.
Therefore, Fag1 appears pulsating as shown in Figure 5.17. In the figure f is set to 0. This pulsating
behaviour is seen even with Fbg1 and Fcg1 .

1
wot1 = 0
0.8

0.6

0.4

0.2 wot2 = p /4
wot3 = p /2
Fag1

−0.2
Pulsates as time
−0.4 progresses

−0.6

−0.8

−1
0 1 2 3 4 5 6 7
b (elec. rad.)

Figure 5.17  Pulsating MMF Distribution Due to Phase-a.

Rewriting Fag1 , Fbg1 and Fcg1 we have

 4TI m 
Fag1 =  cos(w ot + f )cos( b ) (5.239)
 pP 

 4TI m   2p   2p 
Fbg1 =  cos  w ot − + f  cos  b −  (5.240)
 pP   3   3 
 4TI m   2p   2p 
Fcg1 =   cos  w ot + + f  cos  b +  (5.241)
 pP   3   3 

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 246 6/2/2018 2:03:53 PM

 The Standard Parameters of Synchronous Machine  |  247 

In a three-phase machine, MMF are distributed in space. Hence, at any time instant, the net MMF in
the air-gap over a pole pair is given by

Fg1R = Fag1 + Fbg1 + Fcg1

Using eqs. (5.239), (5.240), and (5.241) in the above expression, it can be shown that the resultant
MMF distribution in the air-gap will have a waveshape given by

 4TI m 
Fg1R = 1.5  cos(w ot + f − b ) (5.242)
 pP 

This denotes a traveling wave or drifting wave as shown in Figure 5.18 for different time instants,
p /4 p /2
t1 = 0 , t2 = and t3 = s. This is also referred to as revolving field.
wo wo
Notes:
1. The expression given in eq. (5.242) denotes a forward drifting field for a positive-sequence phase
currents.
2. For a negative-sequence phase currents, the resultant MMF distribution in the air-gap is given by 
 4TI m 
Fg1R = 1.5  cos(w ot + f + b ) (5.243)
 pP 

The above expression denotes a backward drifting field.

3. For triplen currents in the phase windings, the resultant MMF distribution in the air-gap is zero.

1.5
Wave moves
in space as time progresses
1
wot3 = p /2

0.5

0 wot1 = 0
Mag (pu)

−0.5
wot2 = p /4

−1

−1.5
0 1 2 3 4 5 6 7
b (elec. rad.)

Figure 5.18  Drifting MMF Distribution Due to Positive-sequence Phase Currents in abc-windings.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 247 6/2/2018 2:03:56 PM

 248  |  Power System Analysis

5.10.2 Space-phasor Diagram for Round Rotor

Synchronous Machine
Since the speed of the revolving field indicated in eq. (5.242) corresponds to synchronous speed w 0 ,
the flux-linkage of phase-a due to this field can also be represented by a space-phasor as shown below
similar to that established by the field winding:

l ar = K L I a (5.244)
where K L is a constant which has a property of inductance in henry.

Note that the space-phasor position of l ar depends on the time-phasor I a location which, in turn,
depends on the power factor of operation of the generator. While denoting the power factor, the E
phasor is taken as the reference. In Figure 5.19, the relative orientation of different phasors are demon-
strated for unity power factor, purely lagging, and leading loads. From Figure 5.19, it is clear that a
lagging load causes a demagnetising effect, whereas a leading type of load magnetises the machine.
It should be noted that when both field and armature currents are flowing in the machine, the phase-a
coil links with a net air-gap flux-linkage, l ar , which is the resultant of l f and l ar . Therefore, the
air-gap voltage is given by,
d lag
vag = − (5.245)
dt
or
 d l f d lar 
vag = −  + (5.246)
 dt dt 
d−axis d−axis
lf lf
upf

Lag

E E
q−axis
lar q−axis

d−axis lar
lf

lar
Lead

E
q−axis

Figure 5.19  Simplified Hybrid-phasor Diagram Under Loaded Condition.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 248 6/2/2018 2:04:00 PM

 The Standard Parameters of Synchronous Machine  |  249 

The phasor representation of the above equation is depicted in Figure 5.20 for a round-rotor synchronous
machine. Note that a lagging power factor angle, f, is considered and the phasor Vag is take as reference.
Now, eq. (5.246) can be rewritten as

d λ ar
vag = e − 
dt

dia
= e − Ls1 (5.247)
dt
where Ls1 replaces KL from eq. (5.244), which is referred to as synchronous armature reaction
inductance.
In eq. (5.247), since all quantities are sinors of single power frequency, under steady-state, the
phasor equation can be written as

Vag = E − jw o Ls1 I a (5.248)

d−axis

lf

lag q−axis

lar F

Vag
I
Figure 5.20  Phasor Representation of Air-gap Voltage for a Round-rotor Synchronous Machine.

where w o Ls1 is denoted as synchronous armature reaction reactance, xs1. This equation, along with
the phase-a armature winding leakage reactance, xl and resistance, Ra an equivalent-circuit, is drawn
as shown in Figure 5.21.
xs1 Ia Ra x1

+
E Vag Vt

Figure 5.21  Steady-state Equivalent-circuit for a Round-rotor Synchronous Machine.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 249 6/2/2018 2:04:03 PM

 250  |  Power System Analysis

From the equivalent-circuit, the terminal voltage of the machine can be obtained as

Vt = E − jxs1 I a − ( Ra + jxl ) I a 

= E − ( Ra + jxs ) I a (5.249)

where ( Ra + jxs ) is denoted as synchronous impedance, Zs, and xs = xs1 + xl is referred to as

synchronous reactance.

5.10.3  Space-phasor Diagram for Salient-pole Synchronous Machine

For a salient-pole machine, where the length of the air-gap is not uniform, i.e., the air-gap length is
minimum along the d-axis and is large along the q-axis, the armature current’s flux-linkage effects are
not identical along both the axes. Therefore, eq. (5.246) is modified as given below:

 d λ f d λ ad d λ aq 
vag = −  + + (5.250)
 dt dt dt 

The above equation in the phasor form is depicted in Figure 5.22. It is clear from Figure 5.22 that the
component of armature current, I aq is less effective in establishing the reaction than that established
by the component I ad due to the fact that the air-gap length is relatively more along the q-axis than
that along the d-axis. Hence, it can be said that

lad = Ld1iad

laq = Lq1iaq (5.251)

where Ld1 and Lq1, are referred to as d- and q-axes synchronous armature reaction inductances,
respectively. Further, Ld1 > Lq1 .
dl f
Now, using eq. (5.251) in eq. (5.250) and noting that − = e , we have
dt

diad diaq
vag = e − Ld1 − Lq1 (5.252)
dt dt
In the phasor-form, the above equation can be written as

Vag = E − jw o Ld1 I ad − jw o Lq1 I aq (5.253)

Since terminal voltage is related to the air-gap voltage by the expression,

Vt = Vag − ( Ra + jxl ) I a (5.254)

using eq. (5.253) in eq. (5.254), we get,

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 250 6/2/2018 2:04:09 PM

 The Standard Parameters of Synchronous Machine  |  251 

d−axis

lf
lar

lag
laq Iaq
q−axis
E

F
Iad

lad Ia Vag
lar

Figure 5.22  Phasor Representation of Air-gap Voltage for a Salient-pole Synchronous Machine.

Ia Ra x1

+ Armature

E reaction Vag Vt
effects

Figure 5.23  A Block Schematic Representation for a Salient-pole Synchronous Machine.

Vt = E0 − Ra I a − jxd I ad − jxq I aq (5.255)

Note that the above equation is identical to that given in eq. (5.214), E fd0 is same as E0. The suffix 0
denotes the steady-state values of quantities. The expression given in eq. (5.255) is represented by a
block schematic as shown in Figure 5.23. Compare Figure 5.23 with the equivalent circuit given in
Figure 5.21 for a non-salient pole machine.

5.11  MODIFICATION OF 2.2 MODEL AND EQUIVALENT CIRCUITS

Once a 2.2 model is implemented for a synchronous generator, other reduced order models can be
employed with ease without rebuilding the programme. The possible reduced order models that can
be obtained are tabulated in Table 5.1. The table denotes the parameter settings to be made in the 2.2
model while initialising the generator parameters [2].

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 251 6/2/2018 2:04:11 PM

 252  |  Power System Analysis

Table 5.1  Simplifications in 2.2 Model.

Model Basic modifications Settings for no dynamic saliency

2.2 - xq″ = xd″

2.1
xq″ = xq′ and Tqo
′′ ≠ 0 xq′ = xd″

xd″ = xd′ and Tdo

″
≠0
1.1 xq′ = xd′
xq″ = xq′ and Tqo
′′ ≠ 0

1.0
xd″ = xd′ and Tdo
″
≠0 xq = xd′

xq″ = xq′ = xq and Tqo

′′ ≠ 0

′ ≠0
Tqo

xd″ = xd′ and Tdo

″
≠0

0.0 ′ = 10000 (say)

Tdo
-
xq″ = xq′ = xq = xd′ and
(classical)

′′ ≠ 0 , Tqo
Tqo ′ ≠0

Note that all statements in Table 5.1 denote assignment statements. For example, xd″ = xd′ , implies
that while setting the value of xd″ , care is taken to see that it is assigned with the value of xd′ .

5.11.1  Observations with Reduced-order Models

1. Classical model: The classical model is generally used when detailed data is not available and
where a rough estimate of the first-swing stability behaviour of a system is desired. The first-swing
stability performance is related to a large-disturbance stability of power system and denotes the
initial period after the disturbance where the rotor angle swing is large enough to threaten the
synchronous operation of power systems.
Classical model can be used where the time-span of simulation is relatively small compared
′
to  Tdo .
For classical model, the q-axis transient voltage, Eq′ = Ed″ is assumed to be a constant. To
achieve this in 2.2 model, one may require to disable the exciter while carrying out time-domain
simulation. In addition, if xd value is not specified then we can set xd = 6 xd′ .
2. Generator model 2.1: In this case, approximations are made along the q-axis by neglecting
the dynamics of K -coil. As shown in Table 5.1, xq″ is set to have xq′ value. With this setting in

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 252 6/2/2018 2:04:23 PM

 The Standard Parameters of Synchronous Machine  |  253 

eq. (5.156), the flux-linkage y K′ , is not accounted in Ed″ , even if the respective differential equa-
tion involving Tq′′ (which depends on Tqo ′′ ), is solved. Therefore, Ed″ becomes identical to Ed′ . It
is noted that in some representations, G -coil dynamics are neglected instead of the K-coil. This
calls for a setting where xq′ = xq (in other words, xq′ is not specified) and xq″ is left unaltered.
This retains the influence of only the K -coil in Ed″ . Here, only Tqo ′′ is specified.
3. Generator model 1.1 or lower order: All standardised generator models (see section 5.1.2),
remain identical in terms of their performance.

5.11.2  Dynamic Equivalent Circuits for Generators

Neglecting stator transients and ignoring speed variations, the stator d- and q-axes voltage equations
are given by
vd = −id Ra −y q (5.256)

vq = −iq Ra +y d (5.257)

Using eqs (5.157) and (5.158) in eqs (5.256) and (5.257), we have

vd = −id Ra − xq′′iq + Ed′′(5.258)

vq = −iq Ra + xd′′id + Eq′′(5.259)

For a compact notation, combining the above equations in a ‘vector’ form we get

( ) ( )
vq + jvd = − Ra iq + jid + xd′′id − jxq′′iq + Eq′′ + jEd′′ (5.260)

Note that since xq′′ ≠ xd′′, there is no dynamic equivalent circuit representation for eq. (5.260).
Therefore, when this model is interfaced to the network in a multi-machine environment, where the
network is modelled in the synchronous-frame-of-reference (using the bus-admittance matrix), it
results in a time-varying algebraic equation. This difficulty is generally overcome by employing a
dummy-coil approach [2]. However, such a difficulty does not arise when a generator is interfaced to
an SMIB system. This is because, in the SMIB system, the network equations can be transformed to
the machine-frame and computations can be done in the machine-frame itself.
If xq′′ = xd′′ = x ′′, an equivalent circuit can be developed (even in the synchronous-frame), and it
facilitates easy interfacing of the machine to the network. Employing the result given in eq. (5.210),
the above equation can be written in synchronous-frame as

( Eq′′ + jEd′′ ) e jd = Vt + ( Ra + jx ′′ ) I (5.261)

The corresponding equivalent circuit is as shown in Figure 5.24.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 253 6/2/2018 2:04:30 PM

 254  |  Power System Analysis

Ra jX″
I
+
(E″q + j E″d)∠d Vt∠q

Figure 5.24  Dynamic Equivalent Circuit of the Machine Neglecting Dynamic Saliency.

5.11.2.1  Classical Model for a Generator

For classical model of a generator, we can rewrite eq. (5.260) with the 0.0 model settings (see Table 5.1)
as

( )
vq + jvd = − Ra iq + jid + xd′ id − jxd′ iq + Eq′ (5.262)

or

( )
vq + jvd = − Ra iq + jid − jxd′ iq + jid + Eq′ ( ) 

( )
= − ( Ra + jxd′ ) × iq + jid + Eq′ (5.263)

Note that using the setting xq′′ = xq′ = xq = xd′ , in eq. (5.156), we get Ed′′ = 0, and with set-
ting xd′′ = xd′ , in eq. (5.155), Eq″ becomes Eq′ . Further, xq = xd′ removes the dynamic saliency, and
dy ′ F
by setting Tdo ′ to a large value, Td′ turns out be large and hence, is negligible. This effectively
dt
removes the dynamics associated with the field winding and the flux-linkage remains constant at y F 0 .
The other time constants are set to a non-zero value so that ‘divided by a zero’ condition does not
occur while carrying out the time-domain simulation.
Representing eq. (5.263) in the synchronous-frame and rearranging, we get,

Eq′ e jd = Vt + ( Ra + jxd′ ) I (5.264)

The corresponding equivalent circuit is as shown in Figure 5.25.

Ra jX′d
I
+
E′q∠d Vt∠q

Figure 5.25  Dynamic Equivalent Circuit of the Machine for Classical Model.

Note that the magnitude of the internal voltage Eq′ , is assumed to be a constant (since Eq′ =
( xd − xd′ )
y F 0 ) and only the angle will change as per the solution of the swing equation. Further, in
xd
the swing equation, the mechanical damping is generally neglected.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 254 6/2/2018 2:04:38 PM

 The Standard Parameters of Synchronous Machine  |  255 

The expression for torque developed by the machine can be written from eq. (5.196) as

Teg = Eq′ iq (5.265)

For a single-machine infinite-bus system, as shown in Figure 5.14, from eq. (5.258) neglecting Ra,
and for classical model, we obtain,
iq xd′ = −vd (5.266)
We know that
V t = E ∠0° + j I x L (5.267)

Writing the above equation in the machine frame-of-reference, we get,

vq + jvd = Ee − jd + j (iq + jid ) x L (5.268)

Comparing imaginary parts on both sides in the above equation, we obtain,

vd = − Esinδ + iq x L (5.269)
Using eq. (5.269) in eq. (5.266) and solving for iq, we get
E sind
iq = 
xd′ 1
where xd′ 1 = xd′ + x L
Using the above equation in eq. (5.265), we get
Eq′ E sind
Teg = (5.270)
xd′ 1

Note that in the above expression, the I 2 Ra loss is neglected and the resulting power-angle curve
is generally referred to as transient power-angle curve.

5.12  INITIAL CONDITION CALCULATIONS

Initial condition calculation involves computation of the appropriate starting value of state variables
so that a long time-domain simulation required to arrive at an operating point (before a given distur-
bance scenario is studied) is avoided. For example, in section 5.5, where voltage build up condition
is presented, in order reach the nominal value of terminal voltage on open-circuit, it took nearly 20 s.
If a study requires to understand how the generator would behave after the application of a three-
phase short-circuit (when the generator is working on a steady-state open-circuit condition), it is
not necessary to show the voltage build-up phenomenon, i.e., one can completely avoid the initial
voltage build-up -related simulation time. The steady-state nominal terminal voltage condition (on
open-­circuit) can be obtained, to begin with itself, by simply calculating the appropriate initial values
on the states. With such a provision, the advantage is that on starting the time-domain simulation, the
generator directly acquires the steady-state operating condition. Following this, one can easily per-
form the time-domain simulation only related to the disturbance to be studied.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 255 6/2/2018 2:04:42 PM

 256  |  Power System Analysis

To establish the power system operating point, first of all, the load-flow analysis is carried out.
From the converged power flow results, the following end results are noted to compute the initial value
of states:
1. Real power output of generator, Pg0
2. Reactive power output of generator, Qg0
3. Terminal bus voltage, Vt0 ∠q 0
Using these values, the initial conditions of states variables are calculated as follows [2]:
1. Compute
Vt 0 = Vt 0 (cos q 0 + jsinq 0 ) (5.271)
*
 Pg 0 + jQg 0 
I0 =   = I 0 ∠f 0 (5.272)
 Vt 0

Eq0 = Vt 0 + jxq I 0 (5.273)

d 0 = ∠Eq0 (5.274)

2. Compute
iq0 + jid 0 = I 0 e − jd 0 

= I 0 ∠(f 0 − d 0 ) 

iq0 = I 0 cos(f 0 − d 0 ) (5.275)

id0 = I 0 sin(f 0 − d 0 ) (5.276)

3. Compute

vq0 + jvd 0 = Vt 0 e − jd 0 
= Vt0 ∠(q 0 − d 0 ) 

vq0 = Vt 0 cos(q 0 − d 0 ) (5.277)

vd 0 = Vt 0 sin(q 0 − d 0 ) (5.278)

4. Compute
E fd 0 = Eq0 − ( xd − xq )id 0 (5.279)

5. Compute

y d 0 = n q0 (5.280)

y q0 = −n d 0 (5.281)

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 256 6/2/2018 2:05:43 PM

 The Standard Parameters of Synchronous Machine  |  257 

6. Compute
y H 0 = y d 0 (5.282)

xd′
y F 0 = y d0 + E fd 0 (5.283)
xd − xd′

y K 0 = y q0 (5.284)

y G 0 = y q0 (5.285)

7. Compute
Tm0 = Pg 0 (5.286)

8. Compute
The generator field current,
(y F 0 −y d 0 )
if 0 = (5.287)
x fl

The exciter current,

I FD 0 = xd i f 0 (5.288)

Note: In the above calculations, the armature resistance, Ra and the mechanical damping coefficient
D have been neglected. For convenience, the superscript, ′ is dropped in the rotor variables.
If Ra and D are to be accounted in the initial condition calculation, the we can use

Eq0 = Vt 0 + ( Ra + jxq ) I 0 (5.289)

In order to compute the mechanical torque input to the generator, first compute,

y d 0 = xd id 0 + E fd 0 (5.290)

y q0 = xq iq0 (5.291)

Then, we have,
Teg 0 = y d 0 iq0 −y q0 id 0
(5.292)
Tm0 = Teg 0 + D
(5.293)

5.12.1  Example: Initial Condition Evaluation in an SMIB System

For a 60 Hz power system as shown in Figure 5.26, obtain the initial values of all variables for the
following power flow, Pb + jQb , at the infinite bus as follows:
1. Case-1: 0.9 + j 0.4
2. Case-2: 0.3 + j 0.1

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 257 6/2/2018 2:05:50 PM

 258  |  Power System Analysis

3. Case-3: 0.9 − j 0.165
4. Case-4: 0.9 − j 0.4
5. Case-5: 0.3 − j 0.3

Vt∠q Eb = Eb ∠0

jXL

Generator Sb = Pb + jQb

Figure 5.26  A Sample SMIB Power System for Initial Condition Calculations.

Use Eb = 1.0 ∠0 and x L = 0.2 pu. The standard parameters of the machine are given in Table 5.2.

Table 5.2  Standard Parameters of the Generator on 865 MVA, 22 kV Machine Base.

xd = 1.79 xd′ = 0.169 xd′′ = 0.135

xq = 1.71 xq′ = 0.228 xq′′ = 0.2

′ = 4.3 s
Tdo ″
Tdo = 0.032s
′ = 0.85 s
Tqo ′′ = 0.05 s
Tqo

Ra = 0.01 H = 2.86 MJ/MVA

Employing the expression listed in the earlier section, the steady-state values of the variables are
obtained as follows:
*
 P + jQb 
1. Compute line current: I g =  b
 Eb 
2. The generator terminal voltage, Vt = Eb + jx L I g

3. The generated power, S g = Vt I g*

The results obtained are tabulated in Table 5.3.

Table 5.3  Initial Condition of Variable for the SMIB Example.

Quantity Case-1 Case-2 Case-3 Case-4 Case-5

Vt0 1.0949 1.0218 0.9836 0.9374 0.9419

q0 (deg) 9.4623 3.3665 10.5445 11.0702 3.6522

pfg 0.8346 (lag) 0.9285 (lag) 1.0000 0.9748 (lead) 0.7507 (lead)

(Continued)

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 258 6/2/2018 2:05:57 PM

 The Standard Parameters of Synchronous Machine  |  259 

Table 5.3 (Continued)
Quantity Case-1 Case-2 Case-3 Case-4 Case-5
d0 (deg) 44.0473 25.5973 68.0383 81.9071 53.2576

vq0 0.9014 0.9458 0.5286 0.3077 0.6104

vd0 -0.6215 -0.3866 -0.8295 -0.8855 -0.7174

iq0 0.3688 0.2274 0.4896 0.5227 0.4199

id0 -0.9132 -0.2198 -0.7730 -0.8347 -0.0609

Efd0 2.5398 1.3415 1.9171 1.8071 0.7237

yd0 0.9051 0.9481 0.5335 0.3130 0.6146

yq0 0.6306 0.3888 0.8372 0.8938 0.7180

yF0 1.1699 1.0879 0.7334 0.5014 0.6901

yH0 0.9051 0.9481 0.5335 0.3130 0.6146

yG0 0.6306 0.3888 0.8372 0.8938 0.7180

yK0 0.6306 0.3888 0.8372 0.8938 0.7180

Ralosses 0.0097 0.0010 0.0084 0.0097 0.0018

Teg0 0.9097 0.3010 0.9084 0.9097 0.3018

IFDO 2.5398 1.3415 1.9171 1.8071 0.7237

The following observations are made:

1. For lagging power factor, Efd0 is quite large and the rotor angle is relatively low even at heavy
loading levels.
2. For leading power factor, the rotor angle is relatively large and whether id 0 is positive or negative
depends on the value of xd and xq and the loading level.
3. Only at light loads, for leading power factor, the Efd0 is small. It is noted that a large lagging
loading on the machine is constrained by armature winding and field structure heating limits,
whereas the leading load capability is decided by the end-ring heating and permissible low excita-
tion level, in turn, influencing the stability aspects [12].

5.13 EXAMPLE: EIGENVALUE ANALYSIS OF A GENERATOR

UNDER SHORT-CIRCUIT CONDITION
Consider a three-phase generator connected to an RL load as shown in Figure 5.27.
Assume that the rotor is driven at a constant speed equal to w B = 2p 60 rad/s. The field excitation
is on manual control and is adjusted so that the generator is delivering a pure resistive load of RL = 100

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 259 6/2/2018 2:05:57 PM

 260  |  Power System Analysis

Vt pu at Vt = 1.0 ∠0ο pu. Since RL is quite large, it approximately models an

open-circuit ­condition on the generator. Corresponding to this condition,
RL
the initial values of all states are calculated. When a time-domain simula-
tion is ­carried out with RL = 100 pu, it can be verified that the trajectory of
all variables remains undisturbed at the initial value denoting the operating
point. A three-phase symmetrical short-circuit at the generator terminals is
Figure 5.27  Generator ­simulated by switching the RL to zero at certain instant. In order to ­predict
Connected to Load. the frequency components of the resulting trajectories, an eigenvalue anal-
ysis is performed as follows:

5.13.1  Short-circuit Analysis Considering Stator Transients

For the generator, 2.2 model has been employed. The modelling equations are given as follows:
Stator Equations:

dy d
= − Raw B id − w By q - w B vd (5.294)
dt
dy q
= − Raw B iq + w By d - w B vq (5.295)
dt
Rotor Equations:
d-axis Equations:
dy F 1  xd′ 
= ′  −y F + y d + ′
E fd 
dt Td  ( xd − xd )  (5.296)

dy H 1
= ″ [ −y H + y d ] (5.297)
dt Td
and

y d = xd″ id +
( xd′ − xd″ )y +
( xd − xd′ ) xd″ y
H F (5.298)
xd′ xd xd′
q-axis Equations:
dy G 1
= ′  −y G + y q 
dt Tq (5.299)

dy K 1
= ″  −y K + y q 
dt Tq (5.300)

and

yq = xq″ iq +
( xq′ − xq″ )
y +
( xq − xq′ ) xq″
y (5.301)
K G
xq′ xq xq′

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 260 6/2/2018 2:06:01 PM

 The Standard Parameters of Synchronous Machine  |  261 

In addition, the terminal conditions are defined by

 RL 0 0 
 
vabc = 0 RL 0  iabc
0 0 RL 

Therefore, in the 0dq-frame the above equation is written as

v0 = RL i0 ; vd = RL id and vq = RL iq (5.302)

The terminal relationships given in eq. (5.302) are used only for the time-domain simulation. Further, for
a symmetrical system, io = 0 and id/q are calculated using the states, (y d ,y F and y H ) / (y q ,y G and y K ).
For eigenvalue analysis, we simply set v0 dq to zero.

5.13.1.1  Linearisation of Electrical Equations

Linearising eqs. (5.294) and (5.295) and expressing in matrix form, we get

 d ∆y d 
 dt   ∆id   ∆yd   ∆vd 
  = [ Rw ]( 2× 2)   + [ ABG ]( 2× 2)   + [Cw ]( 2× 2)  
(5.303)
 d ∆y q   ∆iq  ( 2×1)  ∆y q  ( 2×1)  ∆vq  ( 2×1)
 dt 
( 2 ×1)   
where
 −w B Ra 0  0 −w B 
   
[ Rw ](2×2) =  [ ABG ]( 2× 2) =  
0 −w B Ra  w B 0 

 −w B 0 
 
[Cw ](2×2) = 
0 −w B 

Linearising eq. (5.296) to eq. (5.300) and expressing in matrix form, we get
 d ∆y F 
 dt 
 
 d ∆y H 
 dt  (5.304)
 d ∆y  =  Ag 2  ∆xe (6×1) +  Fg1  ∆E fd (1×1)
G ( 4 × 6) ( 4 ×1)
 
 dt 
 d ∆y K 
 
 dt  ( 4 ×1)

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 261 6/2/2018 2:06:04 PM

 262  |  Power System Analysis

where
1 1 
 ′ 0 − 0 0 0 
 Td Td′ 
 
 
 1 0 0 −
1
0 0 
T ″ Td″ 
 d 
 Ag 2  = 
  ( 4 × 6)
 1 1 
0 0 0 − 0 
 Tq′ Tq′ 
 
 
 1 1 
0 0 0 0 − ″
 Tq″ Tq 

 1  x′  
d
 ′ 
 Td  xd − xd′  
 Fg1 
  ( 4 ×1) = 0

 
 0 
 
 0 

and
xe(6×1) = [y dy q y F y H y G y K ]T

Linearisation of id and iq:

Referring to eqs (5.298) and (5.301), id and iq can be expressed as

id =
1
yd −
( xd′ − xd″ )
y −
( xd − xd′ )
y (5.305)
H F
xd″ xd′ xd″ xd xd′

iq =
1
yq −
( xq′ − xq″ )
y −
( xq − xq′ )
y
K G (5.306)
xq″ xq′ xq″ xq xq′

Linearisation of above equations would result in

 ∆id  (5.307)
 ∆i  = [Ce1 ]( 2×6) ∆x e (6×1)
 q  ( 2×1)

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 262 6/2/2018 2:06:07 PM

 The Standard Parameters of Synchronous Machine  |  263 

where

1
0 −
( xd − xd′ )
−
( xd′ − xd″ )
0 0


 ″ 
 xd xd xd′ xd′ xd″ 
[Ce1 ](2×6) = 
 

0
1
0 0 −
( xq − xq′ ) ( −
′
xq − xq )
″ 


 xq″ xq xq′ xq′ xq″ 

Using eq. (5.307) in eq. (5.303), we get

 d ∆y d 
 dt 
  = [ Rw ]( 2× 2) [Ce1 ]( 2×6) ∆ x e(6×1) + [ ABG ]( 2× 2) [ Fe ]( 2×6) ∆ x e(6×1)
 d ∆y q 
 dt 
( 2 ×1)
 ∆vd 
+ [Cw ]( 2× 2)  
 ∆vq  ( 2×1)

with
1 0 0 0 0 0 
 
[ Fe ](2×6) = 
0 1 0 0 0 0 

or
 d ∆y d 
 dt   ∆vd 
  =  Ag1  ∆ x e(6×1) + [Cw ]( 2× 2)  
(5.308)
 d ∆y q  ( 2 × 6)
 ∆vq  ( 2×1)
 dt 
( 2 ×1)
where
 ( 2×6) = [ Rw ]( 2× 2) [Ce1 ]( 2×6) + [ ABG ]( 2× 2) [ Fe ]( 2×6)
 Ag1 

Combining eqs. (5.308) and (5.304), the overall linearised equations which describe the perfor-
mance of the synchronous machine are given by the following matrix equation:
 ∆vd 
∆ x e(6×1) = [ AG ](6×6) ∆ x e(6×1) + [CWm ](6× 2)   +  Fg  ∆E fd (1×1) (5.309)
 ∆vq  ( 2×1) (6 ×1)

where
 A   [Cw ]( 2× 2) 
  g1  ( 2 × 6 )   
[ AG ](6×6) =

 [C ] =
 Wm (6× 2) 

  Ag 2 
 
 

[0]( 4×2) 
 ( 4 × 6) 

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 263 6/2/2018 2:06:10 PM

 264  |  Power System Analysis

0 
[ ]( 2×1) 
 Fg 
  (6×1) = 
 
  Fg1  
   ( 4 ×1) 

For short-circuit condition of the machine, the terminal voltage is zero and in the current case, the
field controllers are not considered. Hence, the state-equations are simplified as
∆ x e(6×1) = [ AG ](6×6) ∆ x e(6×1) (5.310)

The eigenvalues of the state-matrix, [AG] provides the natural frequencies associated with the
short-circuited generator.
Note: The system equations remain linear when the exciter is disabled and the generator speed is
assumed constant.
The standard parameters of the machine are given in Table 5.2.
For the tabulated parameters, the short-circuit time-constants are obtained as shown in Table 5.4.
These are obtained by using eqs (5.112) and (5.113) for d-axis, and using eqs (5.134) and (5.135) for
q-axis. For both axes, a quadratic equation needs to be solved to obtain short-circuit time-constants
from the open-circuit time-constants.
Table 5.4  Short-circuit Time-constants.

T d′ = 0.4 s T d″ = 0.0259 s
T q′= 0.1073 s T q″ = 0.0463 s

The [AG] with Ra = 0.01 is given by,

 −27.925 −376.99 20.201 5.6181 0 0 
 
 376.99 − 18.85 0 0 14.33 2.3149
 2.4999 0 −2.4999 0 0 0 
 
 38.546 0 0 −38.546 0 0 
 0 9.316 0 0 −9.316 0 
 
 0 21.595 0 0 0 −21.595 
The corresponding eigenvalues are given in Table 5.5.

Table 5.5  Eigenvalues: Short-circuit Condition.

Ra = 0.01 Ra = 0

-23.385 ± j376.37 ± j376.99

-38.564 -38.546
-21.596 -21.595
-9.2985 -9.316
-2.502 -2.4999

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 264 6/2/2018 2:06:11 PM

 The Standard Parameters of Synchronous Machine  |  265 

It can be seen that the response will have decaying fundamental frequency component
(when Ra ≠ 0) in addition to other neper frequency components. It can be verified that 1/2.502 ≈ 0.4 s
(T ′d).

5.13.1.2  Time-domain Simulation Considering Stator Transients

The machine is initially under open-circuit condition with Vt = 1.0 ∠0o pu and RL = 100 pu. This is
obtained by appropriately calculating the initial values of the state variables. At t = 1 s, a short-­circuit
is applied by switching RL = 0. Further, to keep the speed unchanged through out the simulation, one
may set H = inf in the MATLAB command window if swing equation is implemented. The id(t), iq(t)
and ia(t) plots are shown in Figures 5.28, 5.29, and 5.30, respectively.

−2

−4

−6

−8
id(pu)

−10

−12

−14
1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2

Time (s)

Figure 5.28  The Plot of id During Short-circuit.

The following observations are made:

1. As predicted by the eigenvalues, a decaying fundamental frequency component is seen in id and iq plots.
2. In Figure 5.28, upon reaching the steady-state with the fault, id becomes approximately equal to
E fd 0 1.0
− , which is equal to − = −0.5587 pu. However, from Figure 5.29, it can be seen
xd 1.79
that iq becomes approximately equal to zero.
3. A plot for is ia(t) obtained by using the Park-inverse transformation and i0dq(t). Here, i0 = 0 with
q = w 0t + d 0 .

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 265 6/2/2018 2:06:13 PM

 266  |  Power System Analysis

0
iq(pu)

−1

−2

−3

−4
1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2
Time (s)

Figure 5.29  The Plot of iq During Short-circuit.

10

0
ia(pu)

−2

−4

−6

−8

−10
1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2
Time (s)

Figure 5.30  The Plot of ia During Short-circuit.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 266 6/2/2018 2:06:15 PM

 The Standard Parameters of Synchronous Machine  |  267 

5.13.2  Short-circuit Analysis Neglecting Stator Transients

dy q
Transients due to stator circuits can be neglected by setting dy d = 0 and = 0 in eqs (5.294) and
dt dt
(5.295). To obtain the corresponding state-matrix for eigenvalue analysis, we can rewrite eq. (5.310)
as
  ∆y     ∆y  
  dq  ( 2 ×1)  [ AG11 ]( 2× 2) [ AG12 ](2× 4)    dq  ( 2 ×1) 
     
  =    (5.311)
 A

  ∆y r 

 [ G 21 ]( 4 × 2) [ AG 22 ]( 4× 4) (6×6)   ∆y  
( 4 ×1)  r  ( 4 ×1) 
  (6×1)   (6×1)

T
where y dq = y dy q 
T
and y r = [y F y H y G y K ]

Since ∆y dq = 0 , we can write from eq. (5.311) that

 ∆y  = − [ A ]−1
G11 ( 2 × 2) [ AG12 ]( 2 × 4 )  ∆y r 
 
 dq  ( 4 ×1) (5.312)

From eq. (5.311), we also write that

 ∆y  = [ AG 21 ]( 4 × 2)  ∆y  + [ AG 22 ]( 4 × 4)  ∆y 
 r  ( 4 ×1)  dq  ( 2 ×1)  r  ( 4 ×1) (5.313)

Using eq. (5.312) in eq. (5.313), we have the reduced state-model given by

 ∆y  = [ AGR ]( 4 × 4)  ∆y r  (5.314)

 r  ( 4 ×1) ( 4 ×1)

where

[ AGR ] = − [ AG 21 ][ AG11 ]−1 [ AG12 ] + [ AG 22 ]

The [AGR] with Ra = 0.01 pu is given by,

 −2.4932 0.0018559 − 0.094675 − 0.015294 

 
 0.10289 −38.517 −1.4598 − 0..23581 
 0.49735 0.13832 − 9.2898 0.0042216
 
 1.1529 0.32063 0.06058 −21.585 

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 267 6/2/2018 2:06:19 PM

 268  |  Power System Analysis

The corresponding eigenvalues are given in Table 5.6. From the tabulated eigenvalues, it is clear
that the fundamental frequency component will be absent in the response. Further, the effect of stator
resistance is very minor.

Table 5.6  Eigenvalues: short-circuit Condition-reduced Model.

Ra = 0.01 Ra = 0

-38.5059 -38.5459
-21.5886 -21.5948
-9.2898 -9.3160
-2.5011 -2.4999

5.13.2.1  Time-domain Simulation Neglecting Stator Transients

From eqs (5.294) and (5.295), neglecting stator transients we have

vd = −id Ra −y q
(5.315)
vq = −iq Ra +y d
(5.316)
In terms of subtransient voltages, eqs (5.298) and (5.301) can be written as
y d = xd″ id + Eq″ (5.317)

y q = xq″ iq − Ed″ (5.318)

Using the above relationships, eqs (5.315) and (5.316) can be compactly written in matrix form as
 Ra xq″  id   Ed″ − vd 
   =   (5.319)
 − xd″ Ra  iq   Eq″ − vq 

where vd and va represent d- and q- axes generator terminal voltages given by eq. (5.302). To
simulate an open-circuit condition RL is set to 100 pu and a short-circuit is applied by resetting
RL to zero.
The id(t), iq(t) and ia(t) plots are shown in Figures 5.31, 5.32, and 5.33, respectively. From Figure 5.31,
it can be seen that the waveform is made up of two dominant time constant-based parts, where during the
initial part, i.e., upto 1.1 s, the subtransient time constant influence is strongly seen whereas, in the later
part, the trend changes to a relatively large time constant, transient period-guided response. These parts
are not dominantly observed in iq plot. Further, from Figure 5.33, it can be noted that in the abc-frame,
ia plot does not possess any DC offsets or second harmonic components. Figure 5.34 denotes the plot of
iq when Ra = 0. However, it is observed that there is no major change in id with Ra = 0.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 268 6/2/2018 2:06:21 PM

 The Standard Parameters of Synchronous Machine  |  269 

−1

−2

−3

−4
id(pu)

−5

−6

−7

−8
1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2

Time (s)
Figure 5.31  The Plot of id During Short-circuit, Stator Transients Ignored (Ra = 0.01).

0.4

0.35

0.3

0.25

0.2

0.15
iq(pu)

0.1

0.05

−0.05
1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2

Time (s)

Figure 5.32  The Plot of iq During Short-circuit, Stator Transients Ignored (Ra = 0.01).

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 269 6/2/2018 2:06:23 PM

 270  |  Power System Analysis

0
ia(pu)

−2

−4

−6
1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2
Time (s)

Figure 5.33  The Plot of ia During Short-circuit, Stator Transients Ignored (Ra = 0.01).

0.01

−0.01

−0.02

−0.03

−0.04
iq(pu)

−0.05

−0.06

−0.07

−0.08
1 1.2 1.4 1.6 1.8 2
Time (s)

Figure 5.34  The Plot of iq During Short-circuit, Stator Transients Ignored (Ra = 0).

In Figures 5.35 and 5.36, the currents id and iq, with and without considering the stator transients, are
re-plotted. From the plots, it can be seen that the currents without the stator transients depict the trend
of the currents when the stator transients are considered.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 270 6/2/2018 2:06:25 PM

 The Standard Parameters of Synchronous Machine  |  271 

−2

−4

−6
id

With stator transients

−8 Without str. transients

−10

−12

−14
0.9 1 1.1 1.2 1.3 1.4 1.5
Time (s)

Figure 5.35  The Plot of id with and without Stator Transients (Ra = 0.01).

5
With stator transients
Without str. transients
4

1
iq

−1

−2

−3

−4
0.9 1 1.1 1.2 1.3 1.4 1.5
Time (s)

Figure 5.36  The Plot of iq with and without Stator Transients (Ra = 0.01).

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 271 6/2/2018 2:06:27 PM

 272  |  Power System Analysis

10 With Stator Trans

Without Stator Trans
60 Hz component
9

6
iF (pu)

3 T′d is dominant

0
1 1.5 2 2.5
Time (s)

Figure 5.37  The Plot of iF with and without Stator Transients (Ra = 0.01).

In Figure 5.37, the equivalent field currents are plotted with and without considering the stator tran-
sients. From the plot, it is clearly observed that when the stator transients are considered, it leads to a
60 Hz fundamental frequency component in the filed current as in id and iq signals. It is also seen that
in the field current T ′d -related mode is dominant which decides the transient-duration of the distur-
bance. Here, the equivalent field current is obtained as

(y F −y d )
iF = (5.320)
x fl

xd xd′
where, x fl =
( xd − xd′ )

5.14 EXAMPLE: SYNCHRONISATION OF A STANDALONE

GENERATOR TO MAINS
Here, a standalone generator which is initially under open-circuit condition with Vt = 1.0 ∠0° pu (and
RL = 100 pu), is synchronised to the mains. The generator is on manual excitation. In this case, a
simplified primemover control is provided so that on synchronisation, the generator delivers certain
real power as decided by its initial reference speed. In this study, the influence of the stator transients
on the rotor variables is also illustrated through time-domain simulation.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 272 6/2/2018 2:06:30 PM

 The Standard Parameters of Synchronous Machine  |  273 

In addition to the earlier listed electrical circuit equations, the swing-equation is also used in the
following form:
dd
dt
(
= w y −1 w B ) (5.321)

dw y
2H = − D w y + Tm − Teg (5.322)
dt
The initial speed of the machine, w y0 , is set to 1.01 pu and D is set to zero. Further, Teg is given by

Teg = y d iq − y q id (5.323)

Note: Since the closing of the synchronisation switch is done based on the quantities in the synchro-
nous-frame-of-reference, the modelling is valid only for w y0 ≠ 1.0 pu, but close to 1.0 pu. This offers
a time variation of phase angle of Vt with respect to Eb .
The stator equations are rewritten considering the speed term as
dy d
= − Raw B id − w yw B y q − w B vd (5.324)
dt
dy q
= − Raw B iq + w yw B y d − w B vq (5.325)
dt
A simplified governor model of the following form is used to simulate the primemover system:

 1 

Tm =  s 
(1 + s0.01) 
( w ref − w y ) (5.326)
 

where w ref is the per unit reference speed and s is the per Vt Eb = 1.0 ∠0°
unit regulation of the governor and are set to 1.01 (= w y0 )
and 0.05, respectively. S
The synchronisation switch, S is closed at t = 0.01 s (see RL
Figure 5.38). The synchronisation process is realised by cal-
culating vd and vq as follows:
Figure 5.38  Synchronisation
vd = − Eb sind and vq = Eb cosd of Machine to Mains.

where Eb set to 1.0 pu.

Here, the synchronisation transients with and without stator transients are compared. When the sta-
tor transients are neglected, it is customary to neglect the speed deviations as well, in the stator voltage
equations. Here, the expressions from eq. (5.315) to eq. (5.319) are applicable. The rotor-angle, speed
and torque plots are shown in Figures 5.39, 5.40, and 5.41, respectively. From the figures, it can be
seen that the influence of the stator transients (in the form of high frequency oscillations) is dominant
on the torque than that on the rotor variables like rotor-angle and speed. This result justifies the sim-
plifying assumptions made in low frequency oscillations (i.e., electromechanical oscillations)-related
studies where stator transients are neglected.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 273 6/2/2018 2:06:37 PM

 274  |  Power System Analysis

0.3
with stator transients

0.25

0.2

Without stator transients

0.15
delta (rad)

0.1

0.05
WST WoST

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Time (s)

Figure 5.39  The Plot of Rotor-angle During Synchronisation to Mains.

1.015
WST
WoST

1.01
Wy(pu)

1.005

0.995
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (s)

Figure 5.40  The Plot of Speed During Synchronisation to Mains.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 274 6/2/2018 2:06:39 PM

 The Standard Parameters of Synchronous Machine  |  275 

0.9
WST WoST
0.8

0.7

0.6

0.5
Torque (pu)

0.4

0.3

0.2

0.1

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

Time (s)

Figure 5.41  The Plot of Torque During Synchronisation to Mains.

5.15 EXAMPLE: TRANSIENT STABILITY SIMULATION

OF AN SMIB SYSTEM
In this section, a large-signal stability analysis of a single-machine connected to infinite bus (SMIB)
system is carried out. As stated earlier, the stator transients are neglected while modelling the genera-
tor. A 60 Hz system as shown in Figure 5.42 is considered for the analysis. Note that Pg0 is the output
of the generator and is set to 0.5 pu.
Eb = Eb∠0 = 1.0∠0
Vt∠q = 1.0∠q
jxL = j0.7

Generator Pg

Figure 5.42  An SMIB Power System for Transient Stability Simulation.

Since Vt and Eb are specified, the following steps are used to compute the initial value of states:
1. Compute q0 using
Vt 0 Eb
Pg 0 = sinq 0 (5.327)
xL

2. Compute line current as

I0 =
(Vt 0 − Eb )
(5.328)
jx L

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 275 6/2/2018 2:06:41 PM

 276  |  Power System Analysis

I 02 x L
3. Also compute S g 0 = Pg 0 + jQg 0 = Vto ( I 0 )* . Note that Qg 0 = .
2
Note: If line impedance Z L = ( RL + jx L ) is specified, then to calculate the bus angle the following
expression can be used
Vt20 V E
Pg 0 = cosb − t 0 b cos (q 0 + b ) (5.329)
| ZL | | ZL |

where ZL is represented as Z L = | Z L | ∠b . Its per-unit value is given on the machine base.
For the example, the initial values are tabulated in Table 5.7 employing 2.2 model for the generator.
The standard parameters of the machine are given in Table 5.2.

Table 5.7  Steady-state Values for the System Shown in Figure 5.42.

Quantity Value
q0 (deg) 20.489
d0 (deg) 56.860
Efd0 1.4697
yd0 0.8086
yq0 0.5968
yF0 0.9619
yH0 0.8086
yG0 0.5968
yK0 0.5968
Ra losses 0.0026
Teg0 0.5026

5.15.1  SMIB System without AVR

The following equations are used to obtain generator currents. Neglecting stator transients and rotor
speed deviations, the stator voltage equations are given by
vd = −id Ra − xq″ iq + Ed″

vq = −iq Ra + xd″ id + Eq″ (5.330)

Neglecting network transients and assuming the line to be lossless, the network equation is given by
Vt = Eb ∠0 + jIx L (5.331)

Writing the above equation in the machine frame-of-reference and comparing the real and imaginary
parts on both sides, we obtain,
vq = Eb cosd − id x L
vd = − Eb sind + iq x L (5.332)

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 276 6/2/2018 2:28:32 PM

 The Standard Parameters of Synchronous Machine  |  277 

Substituting eq. (5.332) in eq. (5.330) and solving for the currents, id and iq we get,
−1
id   Ra ( xq″ + x L )   Ed″ + Eb sind 
i  =     (5.333)
″
 q   −( xd + x L ) Ra   Eq″ − Eb cosd 

Note: If line losses are to be accounted, in the above equation, Ra is replaced by (RL + Ra).
In eq. (5.333), if we assume that Ra is neglected, then it simplifies to
Eq″ − Eb cos d
id = − (5.334)
( xd″ + x L )
Ed″ + Eb sin d
iq = (5.335)
( xq″ + x L )

When a three-phase symmetrical fault is applied at the infinite bus, during the fault period, the gen-
erator currents are computed by setting Eb = 0 in the above equations. The corresponding machine
currents, id and iq can be obtained as,
−1
id   Ra ( xq″ + x L )   Ed″ 
i  =     (5.336)
″
 q   −( xd + x L ) Ra   Eq″ 

If the system is assumed to be lossless, then we have,
Eq″
id = − (5.337)
( xd″ + x L )
Ed″
iq = (5.338)
( xq″ + x L )

It is assumed that the fault is of self-clearing type and hence the post-fault network conditions are iden-
tical to the pre-fault system. For Pg0 = 0.5 pu, the fault is applied at t = 1 s, for period of Tclear seconds.
While implementing this scenario in a time-domain simulation, the following steps are employed:
1. For t ≤ 1 s, the generator current is calculated using eq. (5.333). This corresponds to pre-fault
system.
2. During the fault interval, i.e., time period Tclear, the generator current is switched to that obtained
using eq. (5.336). This corresponds to faulted system.
3. The fault clearance is achieved by switching the current back to the value given by eq. (5.333). This cor-
responds to post-fault system. This assumes that the post-fault system is identical to the pre-fault system.

5.15.1.1  First-swing Transient Stability

The electromagnetic torque developed is computed using the expression given by,
( )
Teg = xd″ − xq″ id iq + Eq″ iq + Ed″ id
This, in turn, is used in the swing-equation of the generator neglecting mechanical damping. Having
carried out the time-domain simulation, the rotor angle plots are obtained as shown in Figure 5.43 for
different values of Tclear.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 277 6/2/2018 2:28:35 PM

 278  |  Power System Analysis

Without AVR
5

Tclear = 0.226 s
4.5

3.5
Rotor Angle (rad)

Tclear = 0.225 s
2.5

1.5

0.5

Tclear = 0.2 s
0
0 1 2 3 4 5 6 7 8 9 10
Time (s)

Figure 5.43  Plot of d for Pg0 = 0.5 with Field on Manual Control Considering Ra.

From Figure 5.43, it can be seen that the rotor-angle of the generator increases without bound for a
fault clearing time equal to 0.226 s. However, for a clearing time less than or equal to 0.225 s, the
swing of the rotor reaches a maximum and then decreases in the following swings. This implies that
the system is large-signal stable in the first-swing and it is referred to as the first-swing (transient)
­stability. Based on this observation, a critical clearing time is defined as the fault clearing time for
which the system remains just stable. Figure 5.43 also shows that once the system is first-swing stable,
the oscillations are damped out in the later part of the cycles.

5.15.2  SMIB System with AVR

In this case, a single-time constant static exciter as shown in Figure 5.44 is used to regulate the termi-
nal voltage of the generator. For the regulator, KA = 200, TA = 0.025 s and EfdMax/Min are set to ±6 pu.
Vt
Efdmax
−
+ KA
Vref Σ Efd
1 + sTA
+ Efdmin

VS

Figure 5.44  A Single-time Constant Static Exciter Used with SMIB System
for a Transient Stability Run.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 278 6/2/2018 2:28:37 PM

 The Standard Parameters of Synchronous Machine  |  279 

For the loading condition considered, the initial values of the states are identical to that shown in Table 5.7.
In addition, VRef 0 is computed as

E fd 0
VRef 0 = Vt 0 +
KA

and is obtained as 1.0073. The dynamic equation for the exciter is given by,

dE fd
dt
=
1 
TA 
( )
− E fd + K A VRef 0 + Vs − Vt  (5.339)

Note that the power system stabiliser signal, Vs, is not used here. Further, the terminal voltage magni-
tude, Vt, is obtained by computing vd and vq from eq. (5.332), and we obtain,

Vt = vq2 + vd2

Now, for Pg0 = 0.5 pu, a 3-phase symmetrical fault is applied at the infinite bus at t = 1 s for period of
Tclear. The corresponding rotor angle plots are shown in Figure 5.45.

With AVR
5

Tclear = 0.252 s

Tclear = 0.251 s
3
Rotor angle (rad)

0 Tclear = 0.24 s

−1
0 1 2 3 4 5 6 7 8 9 10
Time (s)

Figure 5.45  Plot of d for Pg0 = 0.5 with Excitation Controller, i.e., with AVR.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 279 6/2/2018 2:28:39 PM

 280  |  Power System Analysis

By repeated time-domain simulations, the critical clearing time is determined for both the cases,
with and without AVR. It is noted that without AVR, the critical clearing time is 0.225 s, whereas,
when the AVR is enabled, the value increased to 0.251 s. Note that the maximum angle swing with
AVR is larger than that in a case where AVR is absent. When the generator bus voltage plots are com-
pared for the two cases, it can be seen from Figure 5.46 that the voltage profile is much better with the
AVR. This demonstrates that an AVR is needed not only to regulate the terminal voltage when load on
the generator changes, but also, is desired to improve the first-swing stability of a system.

1.15

1.1 Tclear = 0.2 s

With AVR

1.05

1
Magnitude of Vg (pu)

0.95

0.9
Without AVR

0.85

0.8

0.75

0.7

0.65
0 2 4 6 8 10 12 14 16 18 20
Time (s)

Figure 5.46  Plot of Vg for Pg0 = 0.5 with and without AVR Following Fault Clearing.

Note: In the above studies Ra is considered and the simulations are carried out using SIMULINK
choosing ode4, (RK 4th order) method with a step size, h = 0.001 s.

5.15.2.1  Results for Stable Case with AVR

In Figure 5.47, rotor angle, rotor speed, and generated torque are plotted when the system is critically
stable with AVR. Here, the fault clearing time is set to 0.251 s.
The corresponding generator current, ia(t), bus voltage, va(t), and the magnitude of the line-to-line
bus voltage are plotted in Figure 5.48. It is clear that the instantaneous values show large fluctuations
when rotor exhibits oscillations. However, these fluctuations reduce with time and eventually reach a
steady-state since the post-fault system is stable.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 280 6/2/2018 2:28:39 PM

 The Standard Parameters of Synchronous Machine  |  281 

Delta (rad.)
2
1
0
−1
0 2 4 6 8 10 12 14 16 18 20
1.05
Speed (pu)

0.95
0 2 4 6 8 10 12 14 16 18 20
2

1
Teg (pi)

−1
0 2 4 6 8 10 12 14 16 18 20
Time (s)

Figure 5.47  Plot of d, w, and Teg with AVR for Clearing Time Equal to 0.251 s.
2
1
0
ia

−1
−2
0 1 2 3 4 5 6
1
0.5
0
va

−0.5

0 1 2 3 4 5 6
1.2
1.1
Vg (mag)

1
0.9
0.8
0 1 2 3 4 5 6
Time (s)

Figure 5.48  Plot of ia(t), va(t), and Vt with AVR for Clearing Time Equal to 0.251 s.

5.15.2.2  Results for Unstable Case with AVR

Now, in Figure 5.49, again the rotor angle, rotor speed, and generated torque are plotted when the sys-
tem is critically unstable with AVR. Here, the fault clearing time is set to 0.252 s. From Figure 5.49, it
can be seen that the rotor angle and speed increase without any bound, whereas, the torque shows large

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 281 6/2/2018 2:28:41 PM

 282  |  Power System Analysis

pulsations. These pulsations are due to the phenomenon of pole slipping of the rotor poles relative to
the synchronously revolving stator poles.
30
Delta (rad.)
20

10

0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
1.3
Speed (pu)

1.2
1.1
1
0.9
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

1
Teg (pi)

0
−1

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Time (s)

Figure 5.49  Plot of d, w, and Teg with AVR for Clearing Time Equal to 0.252 s.

Further, in Figure 5.50, the generator current, ia(t), bus voltage, va(t), and the magnitude of the line-to-
line bus voltage are plotted for unstable case. In these plots too, the instantaneous values show large
fluctuations and they never reach a steady-state since the post-fault system is unstable.

2
1
0
ia

−1
−2
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
1
0.5
0
va

−0.5
−1
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
1.2
Vg (mag)

0.8

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Time (s)

Figure 5.50  Plot of ia(t), va(t), and Vt with AVR for Clearing Time Equal to 0.252 s.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 282 6/2/2018 2:28:42 PM

 The Standard Parameters of Synchronous Machine  |  283 

5.15.2.3  Loss-of-synchronism of a Lab Synchronous Generator

Here, a DC-motor driven synchronous generator is synchronised to another machine at a low terminal
voltage, without any field control. After some time, the mechanical input to the generator is gradually
increased using the DC-motor armature voltage control. This is evident in ia(t) and va(t) plots, and also
in Ia and Va plots, in Figure 5.51. At a certain instance, during the process, since the operating voltage is
low, the real power delivered exceeds its limiting value causing the generator to lose synchronism. This

Going out of Machine is

Synchronism Instantaneous Current ia in (A) tripped
5
3
1
−1
−3
−5
42.5 43 43.5 44 44.5 45 45.5 46 46.5 47 47.5
Instantaneous Voltage va in (V)
100
50
0
−50
−100
42.5 43 43.5 44 44.5 45 45.5 46 46.5 47 47.5
RMS Current Ia in (A), Scaled Field Current If , RMS Voltage Va in (p.u)
5
Ia
4 If
3 Va
2
1
0
42.5 43 43.5 44 44.5 45 45.5 46 46.5 47 47.5
Time (s)

Figure 5.51  Plot of ia(t), va(t), Ia, If , and Va for a Lab Machine When it Loses
Synchronism with Respect to another Machine.

is marked by collapsing of the terminal voltage, Va and a rise in Ia at around 43.8 s. Once the machine
loses synchronism, it exhibits a slipping of pole and it continues until the machine is finally tripped at
47.15 s. During this event, the current pulsations are quite high and the voltage dips to a value as low
as zero. The field current also shows fluctuations during the period of loss-of-synchronism.

References
 [1]  P. Kundur, Power System Stability and Control, McGraw-Hill Inc., NY, 1994.
 [2] K. R. Padiyar, Power System Dynamics - Stability and Control, BS Publications, Hyderabad,
India, 2002.
 [3]  IEEE Std 115–2009, IEEE Guide for Test Procedures for Synchronous Machines.
 [4] IEEE Std 1110–2002, IEEE Guide for Synchronous Generator Modeling Practices and
Applications in Power system Stability Analysis.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 283 6/2/2018 2:28:44 PM

 284 | Power System Analysis

[5] Shashidhara M. Kotian and K. N. Shubhanga, Performance of synchronous machine models in

a series-capacitor compensated system, The EEE Transaction on Power Systems, vol.29(3), pp.
1023–1032, May 2014.
[6] A. M. Kulkarni, Power System Dynamics and Control [Online]. Available at: http://nptel.iitm.
ac.in/courses/108101004/.
[7] I. M. Canay, Determination of model parameters of synchronous machines. Proc. of Institute of
Electrical Engg., vol. 130(2), pp. 86–94, 1983.
[8] A. R. Bergen and V. Vittal, Power System Analysis, Pearson Education Asia, India, 2001.
[9] E. W. Kimbark, Power System Stability- Vol-III: Synchronous Machines, Jhon Wiley, NY, 1956.
[10] P. M. Anderson and A. A. Fouad, Power System Control and Stability, Iowa State University
Press, Ames, Iowa, 1977.
[11] K.R. Padiyar, Analysis of Subsynchronous Resonance in Power Systems, Kluwer, Academic
Publishers, Boston, 1999.
[12] John J. Grainger and William D. Stevenson Jr, Power System Analysis, McGraw-Hill Inc,
New York, 1994.

Review Questions
1. Mention a type of power system dynamic study in which the generator stator transients are usu-
ally neglected. What are the reasons for neglecting the stator transient effects?
2. Enumerate different types of standardised synchronous machine models.
3. Derive an expression for Ld (s) and Lq(s) for a simplified synchronous machine model where
damper windings are neglected.
4. A three-phase star-connected synchronous generator has the following rating: 164 kVA, 2.2 kV,
60 Hz. From the open- and short-circuit test, it is noted that at a field current of 31 A, the open-
circuit terminal voltage is 1846 V/phase and the corresponding short-circuit current is 70 A. The
effective armature resistance is 0.6 Ω . Determine the following in per-unit using the machine
ratings as the base values:
(a) The synchronous impedance of the generator.
(b) The internal voltage of the machine when the machine is delivering the rated load at 0.8
power factor leading.
(c) The voltage phasor diagram of the machine.
5. Consider the system shown in Figure 5.52.
xd = xq = 0.9, x ′ d = 0.2, T ′ do = 2 s

Va V = 1.0 ∠0°

j0.1 p.u.

Figure 5.52 SMIB System for Steady-state Calculations.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 284 6/2/2018 2:28:45 PM

 The Standard Parameters of Synchronous Machine | 285

The generator is operating in steady state with Pg = 0.5 and E fd0 = 1.5 pu. Neglecting the arma-
ture resistance, determine the following:
(a) The power angle.
(b) The generator current in the machine frame.
(c) The terminal voltage of the generator, Va .
(d) The value of the power angle for which the power developed by the machine is maximum.
(e) The steady-state value of the flux linkage of the field coil.
(f) The steady-state short circuit current for a three-phase fault at the generator terminals.
6. Figure 5.53 shows an equivalent circuit for the d-axis rotor windings. The circuit parameters are
given in their usual notations and units. Estimate the standard machine parameters for the d-axis.
id(s) 0.36 Cf = 2.3147

+
yd(s) 1.8 pu E
−

Figure 5.53 d-Axis Equivalent Circuit for Determining the Standard Parameters.

7. Obtain an expression for the terminal voltage va (t ) on open circuit using the operational impedance
approach in per-unit if the field voltage v of is applied at t = 0 (see Figure 5.54). Assume that the
generator is driven at a constant speed w B. Neglect the stator transients and all damper windings.
t=0
f
(open circuit)

a
+
vfo Generator b
−
ff c

Main field

Figure 5.54 Open-circuit Voltage Build-up of a Generator.

8. Consider an unloaded generator driven at a constant rated speed of w 0. A three-phase symmet-
rical short circuit is applied at its terminals at t = 0 with iF (0) = iF0 . Obtain an expression for
id (t ) for t ≥ 0 in terms of the usual notations using the operational impedance approach in
per unit for the following cases:
(a) The generator is assumed to be lossless and the stator transients are accounted. This implies
that v Fo = 0
(b) Neglect only the stator transients and its resistance. Here, it is required that v Fo ≠ 0.
Assume 1.0 model for the generator.
9. Write an MATLAB programme to demonstrate a travelling field in each of the following cases
separately:
(a) When a symmetrical three-phase abc-windings are energised from a three-phase balanced
currents corresponding to 50 Hz (fundamental) and ABC-sequence components.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 285 6/2/2018 2:28:53 PM

 286 | Power System Analysis

(b) When a symmetrical three-phase abc-windings are energised from a three-phase balanced
third harmonic currents.
Assume the winding details appropriately.
10. State different time-constants defined for a 2.2 model of a synchronous machine.
11. In a 2.2 model of a synchronous machine what are the settings to be made so that the model
behaves like a 1.1 model?
12. Derive torque equation in per-unit and show that it is independent of number of poles.
13. For the system shown in Figure 5.52, if xd ≠ xq with xd = 1.0 pu and xq = 0.6 pu, and Pg = 0.5
and E fd0 = 1.0 pu. Determine the power angle of the machine neglecting the resistance.
14. For the system shown in Figure 5.52, the generator has xd = 1.0 pu and xq = 0.6 pu. Neglect
resistance. The generator has just been synchronised to the infinite bus. The excitation is left unal-
tered at a value that existed at the time of synchronisation. Then, the mechanical input is slowly
increased (from zero) until the power angle is 21°. Determine the following:
(a) The generator current.
(b) The terminal voltage of the generator, Va .
(c) The value of the power angle for which the power developed by the machine is maximum.
15. Consider a 60-Hz SMIB system shown in Figure 5.55. For a power level of Sb= (0.9 + j 0.4) pu
and x L = 0.2 pu, the initial value of some of the states is obtained as:

Eb = 1.0, Vt 0 = 1.0949, d 0 = 44.0470 , Teg 0 = 0.9097, y F 0 = 1.1699, y d 0 = 0.9051,

y G0 = 0.6306.
Vt∠q Eb = Eb ∠0
jxL

Generator Sb = Pb + jQb

Figure 5.55 SMIB System for Fault Current Calculations.

For a three-phase symmetrical fault at the infinite bus, determine the following accounting Ra :
(a) The magnitude of the short-circuit current immediately after the occurrence of the fault.
(b) The final value of the short-circuit current if the circuit breaker fails to remove the fault.
Employ 2.2 model for the generator with the following data (pu on machine base) and neglect
stator transients and speed deviations. The generator field excitation is on manual control.
----------------------------------------------------------------------- ----------------------------------------------
′ ″
xd = 1.79, x ′ d = 0.169, x ″ d = 0.135, xq = 1.71, x ′ q = 0.228, x ″ q = 0.2, Tdo = 4.3 s, Tdo = 0.032 s,
′ ″
Tqo = 0.85 s, Tqo = 0.05 s, Ra = 0.01, H = 2.86 s
--------------------------------------------------------------------------------------------------------------------

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 286 6/2/2018 2:29:03 PM

 The Standard Parameters of Synchronous Machine | 287

16. Obtain the short-circuit d-axis time constants, given that xd = 1.79, xd′ = 0.169, xd″ = 0.135,
′ ″
Tdo= 4.3 s, and Tdo = 0.032 s.
17. Derive an expression for the sub-transient inductance L″d of a synchronous machine in terms of
the primitive inductances of the machine.
18. A 50-Hz synchronous generator having H = 5 MJ/MVA and x ′ d = 0.3 pu is connected to an
infinite bus through a purely reactive circuit as shown in Figure 5.56. The generator is delivering
a real power equal to 0.6 pu (Peg0) to the infinite bus at E = 1.0 ∠0°. A three-phase fault occurs at
the high voltage bus of the generator at t = 1 s. The fault is cleared by tripping the faulted line L2.
Assume that the mechanical input to the machine is held constant. Neglect mechanical damping
of the machine.
Vt = 1.0
E
j0.2 L1 j0.3

j0.3
L2

P
eg0
All reactances are in per unit

Figure 5.56 SMIB System to Perform Transient Stability Analysis with 0.0 Model.
Write a SIMULINK-based programme to perform the following:
(a) Determine the critical clearing time for the system.
(b) Draw the speed and rotor angle of the machine and magnitude of the generator terminal
voltage for a fault clearing time of 0.25 s.

M05_POWER-SYSTEM-AN00_SE_XXXX_CH05.indd 287 6/2/2018 2:29:08 PM

 Numerical Integration
of ODEs 6
CHAPTER OUTLINE

6.1 System of Differential Equations: 6.3 Accuracy and Stability of Numerical

Some Observations Integration Methods
6.2 Classification of Numerical 6.4 Demonstration of Some Numerical
Integration Algorithms Integration Methods
6.5 Example: Solution of Swing Equation

In this chapter, numerical integration techniques to obtain time–domain solution of ordinary

differential equations (ODE) are reviewed from fundamentals. Accuracy and stability performances
of numerical integration methods are introduced by using typical methods such as Forward Euler,
Backward Euler, Trapezoidal, and Runge Kutta methods. In addition, some features of Dormand-Prince,
a variable-step method, are discussed.
In order to demonstrate the usage of the methods, a swing-equation (which is a non-linear ODE),
is numerically integrated to obtain the swing-curves. Further, energy-function-based stability analysis
is carried out to appreciate the swings of the rotor-angle.

6.1 SYSTEM OF DIFFERENTIAL EQUATIONS: SOME OBSERVATIONS

A reliable time-domain simulation tool which involves solution of differential equations is an indispensa-
ble part of any power system dynamics studies. The differential equations are obtained as a mathematical
representation of the physical behaviour of real-world systems or components. Such approaches involve
engineering assumptions to arrive at an acceptable mathematical model. The simulation tools help us to
analyse and design systems at the planning stage as an alternative to uneconomical and unsafe in-site meth-
ods. These simulation platforms provide a way to carry out the following typical power system analysis:

1. Electromagnetic transient analysis: An example in this regard is the travelling-wave phenomenon

in transmission lines. This involves distributed parameter representation of systems and results in
partial differential equations (PDEs). Here, time and space are the independent variables.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 288 6/2/2018 3:58:39 PM

 Numerical Integration of ODEs  |  289 

2. Electromechanical transient analysis: Power system stability studies fall under this category.
These studies generally consider lumped parameter representation and result in ODEs. Here,
only time is the independent variable.

In this chapter, numerical integration of ODEs is discussed. The following observations with respect
to ODEs have necessitated the use of numerical integration of ODEs:

1. Linear time-invariant (LTI) ODEs cannot be easily solved analytically if the order of the ODEs is
above 4.

2. Analytical solution of non-linear ODEs cannot be obtained. Only numerical solution

exists.
3. Numerical methods are best suited for off-line analysis of large systems (where there is no timing
constraints), while carrying out online simulations (where time required to obtain solution is not
very critical, however, the simulation should be over in a stipulated time), and in real-time simu-
lations (where timing constraints are very critical).

In order to solve differential equations using numerical integration methods, the ODEs are first
represented in a state-space form [1] where an nth order system is represented by n-first order
differential equations given by

x = F ( x , t ) (6.1)

with initial value x (t0 ) = x0 .

where, x denotes the state variable vector of the system containing n items, and F represents the
vector of n functions.

Note: Only the LTI system can be represented by a set of equations in the matrix-form as

x = Ax + Bu

where A - n × n: state-feedback matrix and B - n × p: input matrix.

In the numerical solution of eq. (6.1), a sequence of points x0 , x1 , x2 , …, are computed that approx-
imate the true solution at a set of time points t0, t1, t2,…,. The time interval between adjacent time
points is called the time-step or step-size, and an integration algorithm advances the numerical solu-
tion by one time-step with each application. The time-step, hn+1 = tn+1 − tn may be constant for
all time intervals over the entire integration interval t ∈ [t0, tN], or may vary at each step. If hn+1
remains constant at h, then such a numerical method is classified as fixed-step method and if hn+1
varies from step to step, such a method is categorised as variable-step method. An integration algo-
rithm advances the solution from tn to tn+1 with integration step-size, hn+1 based on a calculation
that involves previously computed values, xn , xn −1 , xn − 2 , …, and functions F ( xn , tn ), F ( xn −1 , tn −1 ),
and so on.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 289 6/2/2018 3:58:45 PM

 290  |  Power System Analysis

6.2 CLASSIFICATION OF NUMERICAL INTEGRATION ALGORITHMS

A broad classification of numerical integration methods is presented in Table 6.1.

Table 6.1  Classification of Numerical Methods.

Nature of solution Nature of step size Dependence on Type of starting Order

previous step
Implicit Fixed step Single-step Self-starting Number of terms in
Explicit Variable step Multi-step Non-self starting Taylor series

For example, Forward Euler method is an explicit, fixed step, single step, self-starting, and first order
method, whereas the trapezoidal method is an implicit, fixed step, single step, self-starting, and second
order method. The different types are explained in the following sections.

6.2.1  Taylor Series-based Methods

Consider an ODE of the form x = f ( x, t ) with x0 as the initial value of x. Let x (t ) denote the exact
solution. Expanding x (t ) in a Taylor series about t = tn and obtaining the value of the series at t = tn+1
yields the following series expansion for x (tn+1 ) [2]:

x (t( n+1) ) = x (tn ) + ∆x

or
1
= x (tn ) + x (tn ) [tn+1 − tn ] + x(tn ) [tn+1 − tn ] +  +
2

2
1 ( p)
x (tn ) [tn+1 − tn ] + h.o.t .
p
+
p!

where h.o.t. stands for higher order terms of the expansion. If the time-step, h = tn+1 − tn, then
h2 h p ( p)
x (tn+1 ) = x (tn ) + hx (tn ) + x (t n ) +  +
 x (tn ) + h.o.t (6.2)
2 p!
Since x = f ( x, t ), we have,
h2 h p ( p −1)
x (tn+1 ) − h.o.t = x (tn ) + hf ( x(tn ), tn ) + f ′ ( x(tn ), tn ) +  + f ( x(tn ), tn ) (6.3)
2 p!
If the higher order terms are small, then x (tn+1 ) − h.o.t . can be approximated as xn+1 and is given
by the right hand side of eq. (6.3). Thus, in general, the Taylor series-based integration methods can
be expressed as
xn+1 = xn + hTp(xn) (6.4)
where
h h p −1 p −1
T p ( xn ) = f ( x(tn ), tn ) + f ′( x(tn ), tn ) +  . + f ( x(tn ), tn ) (6.5)
2 p!

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 290 6/2/2018 3:58:56 PM

 Numerical Integration of ODEs  |  291 

and the integer p is called the order of the integration method. This method is very accurate for
large p, but is not computationally efficient for large p since it requires a large number of function-­
derivative evaluations. Examples of Taylor series-based integration methods are Forward Euler
method (for p = 1), Runge-Kutta methods (for p = 2, 3, 4,….) etc. The advantages of Taylor series-
based methods are that they are straightforward to program and only depend on the previous time-
step values. However, these methods (especially the Runge-Kutta methods) suffer from involved
error analysis, since the derivatives are approximated and not found analytically. Therefore, the inte-
gration step-size is typically chosen conservatively (on smaller side), and computational efficiency
may be lost.

6.2.2  Single-step and Multi-step Methods

• Single-step methods: An application of an integration algorithm advances the solution from xn
to xn+1. Several well-known methods such as the Euler method, Heuns method, Runge-Kutta
method, and so on, fall into this category where they use only the currently available value
of the states and their function values. For a large scale problem, these methods involve less
computations.
• Multi-step methods: A multi-step method is the one in which xn+1 uses many values of states, say,
xn, xn−1, … and the corresponding functions f (xn, tn), f(xn−1, tn−1), …unlike in one-step methods
which depend only on the information from the immediately previous step. In general, the equa-
tion for multi-step methods is given by

xn+1 = a0 xn + a1 xn −1 +  + a p xn − p + h[b−1 f ( xn+1 , tn+1 ) + b0 f ( xn , tn ) +

+ b1 f ( xn −1 , tn −1 ) +  + b p f ( xn − p , tn − p )] (6.6)

Examples of multi-step methods are non-self-starting Heuns method, Newton-cotes method, Adams
methods, Gears methods, etc.

6.2.3  Self Starting and Non-self Starting Methods

• Self starting methods: Self-starting methods use only the initial values of states to start the
solution process. Examples of self starting methods are Forward Euler method, Heuns method,
Runge-Kutta methods, etc.
• Non-self starting methods: Non-self starting methods cannot be started directly only with the
initial values of states. Such methods require previous values such as x−1, x−2, …, to begin
with. Generally, methods such as Runge-Kutta methods are used for evaluating the starting val-
ues before a non-self method is started. It can be said that all multi-step methods are non-self
starting.

6.2.3.1  Evaluation- and Solution-based Methods

• Evaluation-based methods: While advancing from xn to xn+1, these methods use calculations
which simply involve function evaluations at each time-step. Thus, these methods are also called
explicit methods. Explicit methods obtain the successive values of xn+1 in terms of given or pre-
viously computed quantities. It can be symbolically represented in the form,

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 291 6/2/2018 3:58:58 PM

 292  |  Power System Analysis

xn+1 = f (tn , h, xn ). (6.7)

where f represents the final function which is evaluated as a simple algebraic expression.
Examples of explicit methods are Forward Euler method, Heuns method, Runge-Kutta methods,
Dormand Prince, Adams-Bashforth methods, etc.
• Solution-based methods: While advancing from xn to xn+1, these methods use calculations which
involve solution of equations in each time-step. These methods are also called implicit methods.
A method is said to be implicit if xn+1 depends implicitly on itself through f. A one-step implicit
method can be written as

xn+1 = f (tn , h, xn , xn+1 ). (6.8)

From the above equation, it is evident that the value of xn+1 cannot be obtained as a simple function
evaluation; instead, we need to solve an equation to find xn+1. It should be noted that if the given ODEs
are linear, the application of an implicit integration method necessitates solution involving linear
algebraic equations (i.e., in A x = b form) and if the given ODEs are non-linear, it involves solution
of non-linear algebraic equations (i.e., in F ( x ) = 0 form) in each time-step. This, in turn, requires a
numerical iterative solution technique to solve non-linear algebraic equations. Thus, implicit integra-
tion methods are computationally much more involved than explicit methods. However, the implicit
methods, in general, are numerically stable than explicit methods with higher step-sizes. Examples
of implicit methods are Backward Euler method, trapezoidal method, Gears method, and Adams-
Moulton method.

6.3 ACCURACY AND STABILITY OF NUMERICAL

INTEGRATION METHODS
In this section, accuracy and stability of numerical integration methods are discussed.

6.3.1  Accuracy of Numerical Integration Methods

The expression (6.4) can be rewritten as

xn+1 − xn
= T p ( xn ) (6.9)
h
From the above equation, it can be seen that the accuracy of numerical integration methods depends
on the following factors:
1. By considering more terms on the RHS, the accuracy improves as it closely approximates the
derivative function f (x,t). This is equivalent to increasing the order p of a numerical method.
The resulting error is commonly known as error due to truncation. Thus, higher order methods
are more accurate than lower order methods. However, higher order methods involve more func-
tion evaluations in each time-step, leading to more computational efforts.
2. Now, considering the LHS of the above equation, we can state that smaller the step-size, h, more
accurately the above difference equation approaches numerical differentiation. This may improve

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 292 6/2/2018 3:59:03 PM

 Numerical Integration of ODEs  |  293 

the accuracy of computation much more effectively than the order selection. However, a too small
value for h increases the number of time-steps for a given time interval of simulation, in turn,
affecting the overall computation time. Thus, it is always desired to run a simulation with the
highest possible h without compromising the error limits. This is the driving concept of varia-
ble-step methods. In case of fixed-step methods, especially with explicit type, choice of h is the
constraining factor to prevent a method from becoming numerically unstable. In this regard, most
implicit methods remain numerically stable even with higher time-step (or in some cases inde-
pendent of h). In such situations, it may be required to reduce h so that the solutions are accurate
to adequate level.
A detailed accuracy or error analysis is presented in [2].

6.3.2  S
 tability of Numerical Integration Methods
Through Eigenvalue Analysis
It is seen that due to unbounded accumulation of error, sometimes, numerical methods may exhibit
numerical instability. We know that an application of a numerical integration method discretises the
differential equation. Thus, similar to eigenvalue-based stability analysis of continuous-time systems,
eigenvalue analysis of discrete-time systems can be carried out to predict the numerical stability of
an integration algorithm. The relation between continuous-time system eigenvalues (lc) and dis-
crete-time system eigenvalues (ld) is given by

ld = e hlc (6.10)
We know that, a continuous-time system is considered to be stable if all eigenvalues have a negative
real part. A linear discrete-time system is asymptotically stable if and only if all eigenvalues have
magnitude less than one [3]. The procedure for finding the discrete eigenvalues is as follows:

1. A numerical method is chosen to discretise the differential equations.

2. The discrete equations are written in the form of
xn+1 = Ad xn + Bd un (6.11)

3. The roots of the characteristic equation | ld I − Ad | = 0 are nothing but the discrete eigenvalues
of the system. In this equation, I denotes the identity matrix.
4. From the discrete-time system eigenvalues, stability analysis can be done by computing the contin-
uous-time eigenvalues using eq. (6.10). The MATLAB/OCTAVE instruction is log(lambda_d)/h.

In the following section, some examples of numerical integration of ODEs are discussed.

6.4 DEMONSTRATION OF SOME NUMERICAL

INTEGRATION METHODS
6.4.1  Forward Euler Method
Euler method is a simple explicit, single-step, self starting, first-order method. The Euler method is the
most straightforward method to integrate a differential equation. From the category of Taylor series-
based methods, it is the simplest method with p = 1. For a general ODE of the form

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 293 6/2/2018 3:59:06 PM

 294  |  Power System Analysis

x = F ( x , t )

with initial value x0 , the integration algorithm is given by:

xn+1 = xn + h F ( xn , tn ) (6.12)

In the following section, the application of the method is illustrated for a series connected resistor-
inductor (RL) circuit and a series connected resistor-inductor-capacitor (RLC) circuit.

6.4.1.1  Forward Euler Method Applied to RL Circuit

The connection diagram of RL series circuit is shown in Figure 6.1.

iL
R

v L

Figure 6.1  Connection Diagram of Series RL Circuit.

The governing equation is obtained by applying KVL to the loop and is given by

diL
L + R iL = v
dt

The state-space equation is given by

diL R v
= − iL +
dt L L

with iL(0) = i0 which is set equal to zero.

Note that above equation is in the form,
x = f ( x, t )

Now applying the Forward Euler method, we have for the nth interval,

in+1 = in + Kh (6.13)

where
 R  1
K =  −  in +   vn
 L  L

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 294 6/2/2018 3:59:15 PM

 Numerical Integration of ODEs  |  295 

Table 6.2  Sample Values—Forward Euler Method Applied to Series RL Circuit.

n t=nh in K

0 0 0 10

1 0.01 0.1 9

2 0.02 0.19 8.1

For R = 1Ω, L = 0.1H, h = 0.01 s, i(0) = 0, and v(t) = u(t), a unit-step function, the results are tabulated
in Table 6.2.
From eq. (6.13), we can write that

 R   1 
in+1 = in +  −  in +   vn  h
  L   L 

 R h  h
= 1 −  in +   vn (6.14)
 L   L

 R h
In the above equation, Ad = 1 − and for the parameter specified, Ad = [1 − 10 h] . The eigenval-
 L 
ues of Ad are given by ld = [1 − 10 h] . For different values of step-size, h, both ld and lc are given in
Table 6.3.

Table 6.3  System Eigenvalues for Different h- Forward Euler Method Series RL Circuit.

h (s) 0.01 0.1 0.15 0.2 0.205

ld 0.9 0 −0.5 −1 −1.05

lc −10.536 − −4.6210 ± j20.9440 ± j15.708 0.2380 ± j15.3248

L
From Table 6.3, it can be seen that for h < 2t, where t = , which is the time-constant of the
R
circuit (equal to 0.1 s), the discrete-time system represented by eq. (6.14) remains stable. However,
for h = 2t = 0.2s, the system is marginally stable (see Figure 6.2). For h > 2t, i.e., for h = 0.205 s, as
shown in Figure 6.3, the response of the discrete-time system is unstable. It should be noted that the
response of a passive RL series circuit is always stable. However, when its behaviour is analysed using
the Forward Euler-based discretisation process with step-size, h > 2t, the method provides an unsta-
ble response. This nature of the Forward Euler algorithm is more critical when the original system is
purely oscillatory.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 295 6/2/2018 3:59:21 PM

 296  |  Power System Analysis

2.5

h = 0.01s h = 0.1s h = 0.15s h = 0.2s

1.5

0.5

0
1 1.5 2 2.5 3

Figure 6.2  Forward Euler Response for RL Series Circuit—Effect of Step-size.

2.5

1.5

0.5

−0.5
h = 0.01s h = 0.2s h = 0.205s

−1
1 1.5 2 2.5 3

Figure 6.3  Forward Euler Response for RL Series Circuit—Effect of Step-size-unstable Case.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 296 6/2/2018 3:59:23 PM

 Numerical Integration of ODEs  |  297 

6.4.1.2  Forward Euler Method Applied to RLC Series Circuit

The connection diagram of an RLC series circuit is shown in Figure 6.4.

R(Ω) L(H)

iL C(F)
v vc

Figure 6.4  Connection Diagram of Series RLC Circuit.

The state-equations are given by

 R 1
− −  i 1
iL   L L  L  +  v
  1=   L
vC   v  
0   C  0 
 C 
Using the forward Euler method, we get
 R 1
− −  i 1
 K1   L L  Ln  +   v
=
K   1  L
 2  v    n
0   C n  0 
 C 

iLn+1  iLn   K1  (6.15)

v = +  h
 C n+1  vC n   K 2 

Assume R = 1 Ω, L = 0.1 H, C = 0.5 F, h = 0.05 s, iL0 = 0, vC0 = 0, and v(t) = u(t), a unit-step function,
For n = 0
 K1   −10 −10  0  10  10 
K  =  2    +   1.0 =  
0  0 0 
 2  0 
We get,

iL1  0.5
v  = 0 
 C1   
For n = 1
 K1   −10 −10  0.5 10   5
K  =  2    +   1.0 =  
0  0  0 
 2  1 
We get,
iL 2  0.75
v  = 0.05
 C2   
Note that this algorithm involves only evaluation, hence referred to as an explicit algorithm.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 297 6/2/2018 3:59:30 PM

 298  |  Power System Analysis

Table 6.4  System Eigenvalues for Different h- Forward Euler Method Series RLC Circuit.

h (s) 0.01 0.001

ld 1.00000 ± j0.04472 1.00000 ± j0.00447
lc 0.0994 ± j4.4692 0.0095 ± j 4.4721

In the following case, the resistance R is set to 0.0001 Ω (with L = 0.1 H, C = 0.5 F, i(0) = 0,
vc(0) = 0, and v(t) = u(t), a unit-step function) so that the response of i(t) is close to undamped sinor,
i.e., purely oscillatory. To understand the numerical stability performance, from eq. (6.15) the Ad
matrix is obtained as
 R h
1 − L h − L 
Ad =   (6.16)
 h 1
 C 

For h = 0.01 s and 0.001 s, the eigenvalues are given in Table 6.4. It can be seen that the magnitude
of the eigenvalues ld are greater than 1, indicating that the method is unstable (Figure 6.5). Further,
from the continuous-time system eigenvalues, lc, it can be seen that the real part is positive and the
4.469
imaginary part denotes the frequency of the signal, i.e., = 0.7112 Hz. The period is 1.406 s.
2p
1
The radian frequency 4.469 rad/s is also approximately equal to .
LC

6
h = 0.001s h = 0.01s

2
Current

−2

−4

−6
0 1 2 3 4 5 6 7 8 9 10
Time (s)

Figure 6.5  Forward Euler Response for RLC Series Circuit-effect of Step-size-unstable Cases.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 298 6/2/2018 3:59:34 PM

 Numerical Integration of ODEs  |  299 

Thus, it is noted that in order to obtain a correct oscillatory response using the method, it may be
required to choose an extremely small step-size, which will increase the computation.
Note: A variant of the Forward Euler method is discussed in [4] to improve the stability performance
of the method. This method is known as Euler-Cromer algorithm. This method has been used in the
real-time simulation of systems.

6.4.2  Backward Euler Method

Backward Euler method is a single-step, implicit, self-starting, and first order method. The algorithm
of the Backward Euler method is given by

xn+1 = xn + h F ( xn+1 , tn+1 ) (6.17)

In the following section, the application of the method is illustrated for a series RL and RLC
circuits.

6.4.2.1  Backward Euler Method Applied to RL Circuit

Using this algorithm when the state-equation pertaining to the series RL circuit is discretised,
we get

 R   1 
in+1 = in +  −  in+1 +   vn+1  h (6.18)
  L   L  

Now rearranging the terms, we get,

 R h  h
in+1 1 + = in +   vn+1 (6.19)
 L   L

or

−1 −1
 R h  R h  h
in+1 = 1 + in + 1 +   vn+1 (6.20)
 L   L  L

−1
 R h −1
For the circuit, Ad = 1 + and for the parameter specified, Ad = [1 + 10 h] . The eigenvalue is
 L 
1
given by ld = . This implies that for any realizable step-size, h, the magnitude |ld| is always
[1 + 10 h]
less than 1. This shows that the Backward Euler method remains the numerical stability for any value
of step-size. However, for large h, the values of the estimates are quite inaccurate (Figure 6.6).

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 299 6/2/2018 3:59:41 PM

 300  |  Power System Analysis

1.5
h = 0.205s
h = 0.01s
h = 0.3s

1
Current

0.5

0
−0.5 0 0.5 1 1.5 2 2.5 3
Time (s)

Figure 6.6  Backward Euler Response for RL Series Circuit-effect of Step-size.

6.4.2.2  Backward Euler Method Applied to RLC Series Circuit

The state-equations are given by

 R 1
− −  i 1
iL   L L  L  +  v
  = 1  
v
L
 
vC   0   C  0 
 C 
or

iL  i L 
  = [ A] v  + [ B]v
v
 C  C

Now, applying the algorithm, we have,

iLn+1  iLn   iLn+1  

v  = v  +  [ A] v  + [ B]vn+1  h
 C n+1   C n    C n+1  
or

i  i 
( I − [ A]h)  Ln+1  =  Ln  + h[ B ]vn+1
vC n+1  vC n 

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 300 6/2/2018 3:59:46 PM

 Numerical Integration of ODEs  |  301 

Solving for (iLn + 1, vCn + 1), we get,

iLn+1  −1  i Ln  
v  = ( I − [ A]h)  v  + h[ B]vn+1  (6.21)
 C n+1    Cn  

Assume R = 1 Ω, L = 0.1 H, C = 0.5 F, h = 0.05 s, iL0 = 0, vC0 = 0, and v(t) = u(t), a unit-step function,

1.5 0.5
( I − Ah) =  
 −0.1 1 

0.6452 −0.3226
( I − Ah)−1 =  
0.0645 0.9677 
For n = 0

iL1  0.3226
v  = 0.0323
 C1   
For n = 1

iL 2  0.5203
v  = 0.0843
 C2   

Note: This algorithm involves the solution of equations, hence referred to as an implicit method.
In the following case, the resistance R is set to 0.0001 Ω (with L = 0.1 H, C = 0.5 F, i(0) = 0, vc0 = 0,
and v(t) = u(t) a unit-step function) so that the response of i(t) is close to undamped sinor, i.e.,
purely oscillatory. To understand the numerical stability performance, from eq. (6.21) the Ad matrix
is obtained as
 h 
1 − L 
1   (6.22)
Ad = 2  
hR h 
1+ + h R 
L LC  1 + h
C L 

For h = 0.01 s and 0.001 s, the eigenvalues are given in Table 6.5. It can be seen that the magnitude of
the eigenvalues are less than 1, indicating that the method is stable.

Table 6.5  Eigenvalue Magnitude for Different h- Backward Euler Method Series RLC Circuit.

h (s) 0.01 0.001

ld 0.99800 ± j0.04463 0.99998 ± j0.00447
| ld | 0.9990 0.9999
lc −0.1004 ± j4.4691 −0.0105 ± j4.4721

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 301 6/2/2018 3:59:51 PM

 302  |  Power System Analysis

These performances are plotted in Figure 6.7. It can be observed from Figure 6.7 that the Backward
Euler method adds fictitious damping at large step-sizes although the method is numerically stable.
This fact is further justified by the continuous-time system eigenvalues where the real part is negative.
The imaginary part denotes the radian frequency of the signal.

2.5
h = 0.001s h = 0.01s

1.5

0.5
Current (A)

−0.5

−1

−1.5

−2

−2.5
0 1 2 3 4 5 6 7 8 9 10
Time (s)

Figure 6.7  Backward Euler Response for RLC Series Circuit-effect of Step-size.

6.4.2.3  Backward Euler Method Applied to a Non-linear System

Consider the Van der Pol equation given by

y + ( y 2 − 1) y + y = 0


Let us choose the state variables as: x1 = y and x2 = y . Now we can write the state equations as

x1 = x2

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 302 6/2/2018 3:59:56 PM

 Numerical Integration of ODEs  |  303 

and
x2 = −( x12 − 1) x2 − x1

with initial values of states as x1 (0) = 1 and x2 (0) = 1.

The application of the Backward Euler method leads to the following equations:

x1( k +1) = x1( k ) + x2( k +1) h (6.23)

x2( k +1) = x2( k ) +  −( x12( k +1) − 1) x2( k +1) − x1( k +1)  h (6.24)
 

Note that the above system is non-linear algebraic in nature and requires iterative solution at each
time-step. One can apply the fixed-point iteration technique to solve the equation at each step. This
method is discussed in Chapter 7.

6.4.3  Trapezoidal Method

Trapezoidal method is a single-step, implicit, self starting, and second order method. The algorithm
is given by
h
xn+1 = xn +
2
[ F ( xn , tn ) + F ( xn+1, tn+1 )] (6.25)
For the RLC series circuit mentioned above, after applying the algorithm, we get

iLn+1   h  −1  h  iLn  h 

v =
  I − [ A]   I + [ A]   v  + [ B] ( vn + vn+1 ) (6.26)
 C n+1  2  2  Cn  2 

Assume R = 1 Ω, L = 0.1 H, C = 0.5F, h = 0.05 s, iL0 = 0, vC0 = 0, and v(t) = u(t), a unit-step function.
Then
For n = 0

iL1  0.39604 
v  = 0.01980 
 C1   
For n = 1

iL 2  0.61955
v  = 0.07058
 C2   
The current plot for R = 0.0001 Ω is shown in Figure 6.8. It can be noted from Figure 6.8 that the accu-
racy is much superior compared to the Backward Euler method and like the Backward Euler method,
this method is also numerically stable for all step-sizes.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 303 6/2/2018 4:00:04 PM

 304  |  Power System Analysis

Trapezoidal Method (RLC)

2.5
h = 0.001s
h = 0.01s
2 h = 0.1s

1.5

0.5
Current (A)

−0.5

−1

−1.5

−2

−2.5
0 1 2 3 4 5 6 7 8 9 10
Time (s)

Figure 6.8  Trapezoidal Response for RLC Series Circuit-effect of Step-size.

6.4.4  Runge-Kutta (RK) Fourth Order Method

Runge-Kutta fourth order method is a single-step, explicit, self-starting, and fourth order method.
Consider an ODE of the form x = f ( x, t )
The algorithm is given by

h
xn+1 = xn + [ K1 + 2 K 2 + 2 K3 + K 4 ]
6

where
K1 = f ( xn , tn )

 h h
K 2 = f  xn + K1 , tn + 
 2 2

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 304 6/2/2018 4:00:09 PM

 Numerical Integration of ODEs  |  305 

 h h
K 3 = f  xn + K 2 , t n + 
 2 2
K 4 = f ( x n + hK3 , tn + h)

6.4.4.1  Some Clarification about the Order of the Method

Consider the following equation,
dy
= −2 t 3 + 12 t 2 − 20 t + 8.5
dt

with y(0) = 1
It is required to apply the RK-method with h = 0.5 to calculate y(0.5).
 h
Compute: K1 = f (t = 0) = 8.5, K2 = f  t + 2  = 4.21875

 h
K3 = f  t + 2  = 4.21875 and K4 = f (t + h) = 1.25

Using the algorithm, we get

y(0.5) = 3.21875
By actual integration, we get

y(t ) = −0.5 t 4 + 4 t 3 − 10 t 2 + 8.5 t + 1

For t = 0.5, we get,
y(0.5) = 3.21875

It can be seen that the solution is exact for a single variable function y(t) of the same order as that of
the algorithm. This clarifies the meaning of the order of the method.

6.4.4.2  RK Method Applied to RLC Series Circuit

The application of the method is illustrated for the RLC series circuit mentioned above. After applying
the algorithm we get for the nth interval,
 R 1
− −  i 1
 11   L
K L  Ln  +   v
K  =   L
 21   1   vC n    n
0 0
 
 C 

 R 1 h 
− −  iLn + K11   1 
 K12   L L 2
K  =    +  L  vn
 22   1 h  
0  vC n + K 21  0 
 C   2 

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 305 6/2/2018 4:00:18 PM

 306  |  Power System Analysis

 R 1 h 
− −  iLn + K12   1 
 13   L
K L 2
=
K   1   +  L  vn
 23     h  
0 v + K  0
 C   C n 2 22   

 R 1
− −  i + hK 1
 K14   L L  Ln 13   
K  =  1  + L v
v + hK 23    n
 24   0   Cn
 C  0 

We evaluate iLn+1 and vCn+1 as

iLn+1  iLn  h  K11 + 2 K12 + 2 K13 + K14 

v  = v  +  K + 2 K + 2 K + K 
.  C n+1   C n  6  21 22 23 24 

Assume R = 1 Ω, L = 0.1 H, C = 0.5 F, h = 0.05 s, iL0 = 0, vC0 = 0, and v(t) = u(t), a unit-step function,
For n = 0
 K11  10 
 K  = 0 
 21   

 K12  7.5
 K  = 0.5
 22   

 K13  8 
 K  = 0.375
 23   

 K14  5.8125
 K  = 0.8 
 24   
We get,
iL1  0.3901
v  = 0.0213
 C1   

For n = 1

 K11  5.886 
 K  = 0.7802
 21   

 K12   4.219 
 K  = 1.0746
 22   

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 306 6/2/2018 2:19:39 PM

 Numerical Integration of ODEs  |  307 

 K13   4.562 
 K  = 0.9912
 23   

 K14  3.109 
 K  = 1.2364 
 24   
We get,
iL 2  0.6114 
v  = 0.0725 
 C2   

The current plot for R = 0.0001 Ω is shown in Figure 6.9. From the plot, it can be seen that the method
remains stable even when h is set to 0.1 s. However, it decreases the accuracy of the estimates.

2.5

h = 0.001s h = 0.1s

1.5
Current

0.5

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (s)

Figure 6.9  RK Fourth Order Response for RLC Series Circuit-effect of Step-size.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 307 6/2/2018 2:19:40 PM

 308  |  Power System Analysis

6.4.5  Variable-step Methods

Variable-step implies adaptive feature of step-size while obtaining the solution of the differentiation
equations. In the following section, Dormand-Prince method, popularly known as ode45 in the litera-
ture [5]-[7], is discussed. Some features of the method are listed below:
1. The step-size is varied in order to obtain an estimate of states which satisfies a desired error toler-
ance level, eT. It is assumed that the error in the estimates is dominantly due to truncation error.
2. An approximate value of truncation error is determined at any time instant tk with step-size
hk, by solving the differential equation by two different order ode solvers, for example, Runge-
Kutta fourth order and fifthorder. The difference between these two solutions is considered as the
estimated error, eE.
3. If this error, eE ≤ eT, the current step-size is accepted as successful. Otherwise, the step-size
is reduced progressively in a guided fashion.
Considering familiarity with the fixed step method, one can easily appreciate adaptive or variable
step methods. Earlier, in fixed-step methods, the step-size h (chosen before the start of the integration
process), is held constant throughout the time-span of simulation. In the Dormand-Prince method,
the user is supposed to specify eT. Two solutions, fourth order and fifth order, are calculated using
eqs (6.27) and (6.28), respectively. They use step-size h along with slope factors, K obtained from
eq. (6.29) to eq. (6.35).

 5179 7571 393 92097 1877 1 

yk +1 = yk + h  K1 + K3 + K4 − K5 + K6 + K 7  (6.27)
 57600 16695 640 339200 2100 40 

 35 500 125 2187 11 

zk +1 = zk + h  K1 + K3 + K4 − K5 + K6  (6.28)
 384 1113 192 6784 84 
where

K1 = f (tk , yk ) (6.29)

 1 1 
K 2 = f  tk + h, yk + hK1  (6.30)
 5 5 

 3 3 1 3 
K3 = f  tk + h, yk + h  K1 + K 2  (6.31)
 10 10  4 4 

 4 4 11 14 40 
K 4 = f  tk + h, yk + h  K1 − K 2 + K3  (6.32)
 5 5 9 3 9 

 8 8  4843 3170 8056 53 

K5 = f  tk + h, yk + h  K1 − K2 + K3 − K 4  (6.33)
 9 9  1458 243 729 162 

  9017 355 46732 49 5103 

K6 = f  tk + h, yk + h  K1 − K2 + K3 + K4 − K5  (6.34)
  3168 33 5247 176 18656 

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 308 6/2/2018 2:19:43 PM

 Numerical Integration of ODEs  |  309 

  35 500 125 2187 11 

K 7 = f  tk + h, yk + h  K1 + K3 + K4 − K5 + K6  (6.35)
  384 1113 192 6784 84 

If the truncation error, eE with current h is very less in comparison to eT, then there is some margin
to increase step-size using a scaling factor. Similarly, if the truncation error with current h is more
than eT, the algorithm reduces the step-size. This is brought about by scaling factor Sc such that

hnew = Sc *h (6.36)

The scaling factor signifies the magnitude by which the step-size has to increase or decrease, and is
calculated by using eq. (6.37). It can be clearly stated that the scaling factor depends on the tolerance
value imposed by the user and the truncation error. This factor ‘Sc’ may be any number between zero
and infinity.

eT
Sc = 0.9 4 (6.37)
eE
where,
eT : Tolerance error specified.
y(k + 1): Fourth-order evaluation of ode.
z(k + 1): Fifth-order evaluation of ode.
eE = | y(k + 1) – z(k + 1) |: Truncation error collectively.

Consider an ODE evaluation using Dormand-Prince method. Assume that tfinal = 10 s, hk = 0.1 s. Now,
for a given tolerance, assume Sc value comes out to be 100, then according to the procedure indicated
above, hnew = 100*0.1 = 10 s. This is not a faithful solution. In order to restrict this chaotic behaviour,
one has to restrict the value of Sc to an acceptable level. One can specify the maximum and minimum
bound for the scaling factor, depending on the nature of the differential equations as they decide the
number of iterations before a successful h is reached. Further, it should be noted that the number of
iterations also depends on the initial value of step-size, hinit, chosen. In the current implementation,
the bounds on the scaling factor are chosen as 0.01 (lower bound) and 1.5 (upper bound). It is also
assumed that hinit = hmax.

The algorithm is summarized in the following lines:

1. Initialise tfinal, eT, and hmax in addition to initial values of states. Further, set k = 0, t(k) = 0, and
h = hmax. Note that the initial value of step-size is set to hmax.
2. Using eqs (6.27) and (6.28), calculate the truncation error, eE employing h.
3. Calculate Sc using eq. (6.37). Apply the maximum and minimum bound on Sc.
4. Calculate hnew = Sc * h, set hnew = hmax if hnew > hmax.
5. Again calculate the truncation error, eE with h = hnew.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 309 6/2/2018 2:19:45 PM

 310  |  Power System Analysis

6. If eE ≤ eT, evaluate y(k+1) and set the successful step-size as hks = h. Increment t(k+1) = t(k) + hks. If
t(k +1) ≥ tfinal Stop. Otherwise, go to step [8].
7. If eE > eT, then go to step [3] with h = hnew.
8. Set k = k + 1, and repeat the process from step [2] with h = hmax.

Notes:
• At kth step, if the used hnew does not successfully reduce the error, eE below the specified eT in
Itmax number of iterations, then the loop breaks. One has to retry the simulation by changing
eT to a higher value, and hmax to a lower value. In the programme, Itmax has been set to 100.
• In the above steps, before a new time interval is started, the step-size h is initialised to hmax. In
other words, as stated earlier, hinit is set to hmax. In such a case, the maximum bound on Sc is not
effective. If one specifies hinit in addition to hmax such that hinit < hmax, then the maximum bound
will come into force. This is not considered in the above algorithm.
• The weights used along with slope-coefficients in fourth and fifth order (see eqs (6.27) and
(6.28)) evaluations sum, respectively, to 1.

6.4.5.1  Dormand-Prince Method Applied to RLC Series Circuit

Here, it is assumed that R = 1 Ω, L = 0.1 H, C = 0.5 F, iL0 = 0, vC0 = 0, and v(t) = u(t), a unit-step
function. For the algorithm, hmax is set to 2 s. The plots of iL for two values of error specifications,
eT = 0.01 and 0.001 are obtained and are shown in Figure 6.10. Here, the response is compared with
RK fourth order fixed step method with h fixed at 0.01 s.

0.8
Error: 1e−2
Error:1e−3
0.7 RK4: h = 0.01s

0.6

0.5

0.4
iL(t)

0.3

0.2

0.1

−0.1
0 1 2 3 4 5 6
Time in seconds

Figure 6.10  Response of iL for Dormand-Prince Method with R = 1 Ω .

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 310 6/2/2018 2:19:45 PM

 Numerical Integration of ODEs  |  311 

The variation in the step-size from one time interval to another for the specified errors is also shown
in Figure 6.11. It can be seen from Figure 6.11 that upto 2 s until which there is a continuous change
in the signal value (see Figure 6.10), the step-size takes smaller values and then onwards, it adopts a
higher value. Further, it can be noted that during this time interval, the h value for eT = 0.001 is lower
than that for eT = 0.01.

1
Error: 1e−3
Error:1e−2
0.9

0.8

0.7
Step value h (s)

0.6

0.5

0.4

0.3

0.2

0.1

0
0 1 2 3 4 5 6 7 8 9 10
Time in seconds

Figure 6.11  Variation in h for the Specified Errors: 0.01 and 0.001 with R =1 Ω.

The current plot for R = 0.0001 Ω is shown in Figure 6.12 for eT = 0.01 and 0.001. The correspond-
ing variation in the step-size over the complete time-span is depicted in Figure 6.13. It can be clearly
seen from Figure 6.12 that for eT = 0.001, the Dormand-Prince method provides iL plot almost close
to that obtained with RK fourth order method. However, it terms of step-size, the RK fourth order
method runs with a constant h of 0.01 s, whereas with the Dormand-Prince method, the step-size
varies between 0.065 s and 0.085 s.

6.5 EXAMPLE: SOLUTION OF SWING EQUATION

To illustrate the application of numerical methods to solve a swing equation, a single-machine con-
nected to an infinite bus system is considered. The one-line diagram of the sample power system is
shown in Figure 6.14.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 311 6/2/2018 2:19:46 PM

 312  |  Power System Analysis

2.5
Error: 1e−2
Error:1e−3
2 RK4: h = 0.01s

1.5

0.5
iL(t)

−0.5

−1

−1.5

−2

−2.5
0 0.5 1 1.5 2 2.5 3
Time in seconds

Figure 6.12  Response of iL for Dormand-Prince method with R = 0.0001 Ω.

0.16
Error: 1e−3
Error:1e−2
0.15

0.14

0.13
Step value h (s)

0.12

0.11

0.1

0.09

0.08

0.07

0.06
0 1 2 3 4 5 6 7 8 9 10
Time in seconds

Figure 6.13  Variation in h for the Specified Errors: 0.01 and 0.001 with R = 0.0001 Ω.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 312 6/2/2018 2:19:47 PM

 Numerical Integration of ODEs  |  313 

1 j0.3 2 S = (0.8 + j0.074) pu

j0.2

x′d = 0.3 pu j0.15 j0.15 E = 1.0∠0° pu

H = 5 MJ/MVA

Figure 6.14  The One-line Diagram of the SMIB System (Pre-fault Condition).

For this 60 Hz system, the pre-fault details are given in Figure 6.14. At t = Tfault = 0.1 s, a symmetrical
three-phase to ground fault is applied at the middle of a line as shown in Figure 6.15.

1 2
j0.3
j0.2

x′d = 0.3 pu j0.15 j0.15 E = 1.0∠0° pu

H = 5 MJ/MVA 3−phase fault

jxi2 = 1.8 pu
i 2
j0.5 1 j0.3
j0.15 j0.15

Equvalent circuit during the fault

Figure 6.15  The System Structure During the Faulted Condition.

At t = Tfault + Tclear, the faulted line is tripped in order to clear the fault. This implies that the fault exists
for a duration of Tclear, and the post-fault system contains only one line as depicted in Figure 6.16.

1 j0.3 2 P = 0.8 pu
j0.2

x’d = 0.3 pu E = 1.0 0° pu

H = 5 MJ/MVA

Figure 6.16  The System Structure in the Post-fault Condition.

In order to understand the large-signal stability behaviour of the generator for different fault durations,
the swing equation of the generator is solved employing a simplified model, referred to as classical
model, for the generator.
The swing equation is given by
dd
= w − w0
dt

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 313 6/2/2018 2:19:49 PM

 314  |  Power System Analysis

2 H dw
= Pm − Pmax sin d (6.38)
w 0 dt

Where w0 = 2p 60 rad/s (electrical), Pm denotes the mechanical input (in pu) which is assumed to be
a constant and Pe = Pmax sin d represents the electrical output (in pu). These notations assume that the
per unit torque values are identical to the per unit power values (neglecting speed deviations). Further,
it should be noted that the above system of differential equations is non-linear in nature since the
developed power Pe is related to d by a trigonometric function. This makes the swing equation solu-
tion feasible only by numerical integration methods. Also, note that in the above swing equation, the
mechanical damping component is neglected.

6.5.1  Initial Condition Calculation for the Machine

Variables and System Equations
For obtaining the solution of the swing equations, the initial value of the states [d,  w]T has to be deter-
mined for each condition of the system. The system conditions are categorised as shown in Figure 6.17
pertaining to the network structure.
x (0)F Faulted x (0)P
Pre−fault Post−fault
Network Network system

At t = Tfault At t = (Tfault + Tclear)

a 3−phase fault the fault is cleared.
is applied

Figure 6.17  The Transition of the System Conditions.

From Figure 6.17, it can be seen that as per the network structure, the system conditions are catego-
rised as follows:

1. Pre-fault system (see Figure 6.14): For this condition, the swing equation appears as
dd
= w − w 0 (6.39)
dt
2 H dw
= Pm − Pmax1 sin d (6.40)
w 0 dt
Here, the initial value of the states is obtained by setting the derivative equal to zero. This results in
w = w0 (6.41)
Pmax1 sind = Pm
or
 P 
d 0 = sin −1  m  (6.42)
 Pmax1 
For the system details given, the steps employed are

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 314 6/2/2018 2:19:52 PM

 Numerical Integration of ODEs  |  315 

(a) Compute the current entering Source-2:

(0.8 − j 0.074)
I = = 0.8034Æ − 5.28ο pu
1.0Æ 0ο
(b) Compute the internal voltage of the machine employing x′d = 0.3 pu as
E ′ = 1.0 ∠0ο + 0.8034 ∠ − 5.28ο × j 0.65 = 1.1698∠26.388ο
(c) Determine Pmax1 as
1.1698 × 1
Pmax1 = = 1.8 pu
0.65
(d) Thus, the initial value of the states is
[d , w ]T0 = [26.388ο (0.46055 rad ), 376.99 rad / s]T
For the pre-fault condition of the system, the swing equations need not be solved.
2. Faulted system (see Figure 6.15): For this condition, the swing equation is written with Pmax2 as
follows:
dd
= w − w 0 (6.43)
dt
2 H dw
= Pm − Pmax 2 sin d (6.44)
w 0 dt

where Pmax2 is obtained for the faulted condition as

1.1698 × 1 1.1698 × 1
Pmax 2 = = = 0.65 pu
xi 2 1.8
with xi2 obtained as the transfer reactance between the internal bus of the generator and bus-2.
Note the following:
(a) To solve the above swing equation, the operating value of the states obtained for the pre-fault
condition, [d , w ]T0 , is used as the initial value, x ( 0) F , and the equations are numerically inte-
grated between t = Tfault and t = Tfault + Tclear.
(b) The value of the states determined at the end of the fault period, i.e., at t = Tfault + Tclear is
noted as x ( 0) P .
3. Post-fault system (see Figure 6.16): For this condition, the swing equation is written with Pmax3
as follows:
dd
= w − w 0 (6.45)
dt
2 H dw
= Pm − Pmax 3 sin d (6.46)
w 0 dt
where Pmax3 is obtained for the post-fault condition as
1.1698 × 1
Pmax3 = = 1.4623pu
0.8

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 315 6/2/2018 2:19:57 PM

 316  |  Power System Analysis

Note the following:

(a) To solve the above swing equation, the value of the states obtained at the end of the faulted
condition, i.e., x ( 0) P , serves as the initial value.
(b) The equations are numerically integrated between t = Tfault + Tclear and t = Tend. Here, Tend
should be chosen such that stable/unstable nature of the swings is adequately captured.
In the above implementation, it is assumed that the event occurs immediately after the time instants
marked. For example, if a fault is initiated at time t = Tfault, then the faulted system details are accounted
in the next time step. Similarly, when the fault is cleared at time t = Tfault + Tclear, the post-fault Pmax3
is used at t = Tfault + Tclear + h.

6.5.2 Solution of Swing Equations by Using

Forward Euler Technique
Employing the Forward Euler algorithm, the swing equation is discretised with a fixed step-size,
h = 0.001s. Using the value of states at the kth step, the value of the states at (k + 1)th step is obtained as:

d ( k +1)  d ( k ) 
w =
 ( k +1)  w ( k ) 
 + F x ( k ) × h (6.47) ( )
where
 w (k ) − w 0 ( ) 
F x ( ) 
= w
 (6.48)

( )
(k ) 0
 2 H 0.8 − Pmax sin d ( k ) 

For the above system, between t = 0 and t = Tfault , the following F-function is used:

 w (k ) − w 0 ( ) 
F1 x ( ) 
= w
 (6.49)

( )
(k ) 0
 2 H 0.8 − 1.8 sin d ( k ) 

between t = Tfault and t = Tfault + Tclear, the following F-function is used:

 w (k ) − w 0 ( ) 
( 
F2 x ( k ) =  w )  (6.50)

0
(
 2 H 0.8 − 0.65 sin d ( k ) ) 

Between t = Tfault + Tclear and t = Tend, the following F-function is used:

 w (k ) − w 0 ( ) 
( )
F3 x( k )

= w
 (6.51)

0
(
 2 H 0.8 − 1.4623 sin d ( k ) ) 

It should be noted that eq. (6.47) need not be evaluated with F1 and [d , w ]T0 since it pertains to
pre-fault system, and even if it is evaluated the state values remain at x ( 0) F . For the faulted system,

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 316 6/2/2018 2:20:01 PM

 Numerical Integration of ODEs  |  317 

eq. (6.47) is employed with F2 and x ( 0) F . These evaluations provides x ( 0) P at time t = Tfault + Tclear.
With this initial value for the states, eq. (6.47) is used with F3 until t = Tend is reached.

6.5.2.1  Stability Performance for Different Fault Clearing Time

Here, Tfault is set to 0.1 s and Tend is taken as 2 s. To begin with, Tclear is set to 0.1 s and eq. (6.47) is
evaluated employing F1, F2, and F3in order. The states are noted as in Table 6.6.

Table 6.6  States of the SMIB System-Forward Euler Method.

Tclear s 0.1 0.410 0.411

x ( 0) F 0.46055, 376.99 0.46055, 376.99 0.46055, 376.99

x ( 0) P 0.55430, 378.85 1.7187, 381.97 1.7237, 381.98

For Tclear = 0.1 s, the rotor-angle and rotor-speed plots are obtained as in Figure 6.18. It can be seen
from Figure 6.18 that the rotor swings are purely oscillatory as the system possesses zero damping.
While producing such a behaviour using the Forward Euler method, it is known that the response
grows in amplitude as the method is numerically unstable. However, with a small-enough step size,
the method offers an easy solution procedure when a short time span (say, 1–2 s), of simulation is
desired. Thus, for t = Tclear = 0.1 s, the system can be claimed to be stable, with an understanding that
if damping were to be present the rotor oscillations would eventually damp out.

Forward Euler (Tclear = 0.1 s)

1

0.8
Delta (rad)

0.6

0.4

0.2
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (s)
379

378
Omega (rad/s)

377

376

375
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (s )

Figure 6.18  The Rotor-angle and Rotor-speed Plots for Tclear = 0.1 s with Forward Euler Method.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 317 6/2/2018 2:20:03 PM

 318  |  Power System Analysis

For Tclear = 0.410 s, the rotor-angle and speed plots are drawn against time as shown in Figure 6.19.
Here, some transition points are marked as A, B, C, and D. These events are also indicated in
Figure 6.20, where the rotor-speed variable is plotted against the rotor-angle variable. These plots
are referred to as phase-plane trajectories. Of these, points B and D denote a condition where
the rotor speed is again equal to the pre-fault value, whereas the rotor-angle shows a maximum
deviation. At point A, the fault is cleared and the rotor-angle continuous to increase and the speed
decreases until point B; beyond this, both angle and speed decrease to reach C. At this point, the
rotor-angle is again equal to the pre-fault value and the speed reaches a minimum. These points
denote energy transition events.

Forward Euler (Tclear = 0.410 s)

3
X: 0.51
Y: 1.719
2
Delta (rad)

2.4913 rad (B) (D)

1

−1
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (s )

385

380
Omega (rad/s)

(C)
375

370 t = (0.1 + 0.41) = 0.51 s(A)

365
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (s )

Figure 6.19  The Rotor-angle and Rotor-speed Plots for Tclear = 0.410 s with Forward Euler Method.

For a fault-clearing time equal to 0.411 s, the rotor-angle and speed trajectories are obtained as shown
in Figure 6.21. A comparison of these plots with those shown in Figure 6.19 clearly demonstrates
that when the fault-clearing time is 0.410 s, the rotor-angle having reached point B, correspond-
ing to a maximum angle-excursion, decreases below this angle, whereas when the clearing time is
increased to 0.411 s, the rotor-angle trajectory does not turn back, leading to a monotonic increase
of angle as well as the speed. This condition is attributed as the unstable condition of the system.
Therefore, for a clearing time equal to 0.410 s, the system is said to be critically stable and the
corresponding angle at point A is referred to as the critical clearing angle, and the clearing time
is known as the critical clearing time. From Table 6.6, the critical clearing angle can be noted
as, dcr = 1.7187 rad.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 318 6/2/2018 2:20:04 PM

 Numerical Integration of ODEs  |  319 

Phase plot for Tclear = 0.410s

384
A
382

380

D
378 X: 0.4606
Y: 377
Omega (rad/s)

B
376
Pre−fault

374

372
C

370

368
−1 −0.5 0 0.5 1 1.5 2 2.5
d (rad)

Figure 6.20  Phase-plane Trajectory of Rotor-angle and Rotor-speed for Tclear = 0.410 s with
Forward Euler Method.
Forward Euler, Tclear = 0.411s
10

8
Delta (rad)

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (s )

400
Omega (rad/s)

390

380

370
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (s )

Figure 6.21  The Rotor-angle and Rotor-speed Plots for Tclear= 0.411 s with Forward Euler Method.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 319 6/2/2018 2:20:05 PM

 320  |  Power System Analysis

6.5.3 Solution of Swing Equations by Backward

Euler and RK-4 Techniques
In the following section, the solution of the swing equation is determined by using Backward Euler
method and Runge-Kutta fourth order methods.

6.5.3.1  Solution of Swing Equations by Backward Euler Method

This method is a first-order implicit method. The application of this method leads to the following
solution process: Using the value of the states at the kth step, the value of the states at (k + 1)th step
is obtained as follows:

d ( k +1)  d ( k ) 
w = (
 + F x ( k +1) × h (6.52) )
 ( k +1)  w ( k ) 
where
 (
w ( k +1) − w 0 ) 
(
F x( k +1) ) 
= w

 (6.53)
0
(
 2 H 0.8 − Pmax sin d ( k +1) ) 

( )
As in the Forward Euler method the function, F x( k +1) , is appropriately changed depending on the
condition of the system, i.e., pre-fault, faulted, and post-fault conditions. Further, it should be noted
that while discretising the swing equation, it involves the solution of a non-linear algebraic equation
since the F-function evaluation cannot be done until we obtain the states. To solve the equation at the
(k + 1)th step, a fixed-point iteration scheme has been described below:
1. Step 1: The function F is evaluated as
 i
w ( k +1) − w 0( ) 
F ( x(ik +1) ) 
=

 (6.54)
w
(
 0 0.8 − Pmax sin d
 2H
i
( k +1) ) 

(i = 0)
where x( k +1) is taken as the state of the system at the previous, kth time-step.

2. Step 2: From eq. (6.52), an updated state value is obtained as

(i +1)
d ( k+1)  d ( k ) 
w 
 ( k +1) 
=
w ( k ) 
i
( )
 + F x( k +1) × h (6.55)

3. Step 3: Now compute

( i +1) (i )
d ( k +1)  d ( k +1) 
max   −  = error (6.56)
w ( k +1)  w ( k +1) 
If error ≤ ¨, then stop the process of iteration, set
x( k +1) = x((ki++11))

and move to the next time-step. Otherwise, set i = i + 1 and go to Step-1. The ¨ is set to 0.0001
and the maximum number of iterations is taken as 10.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 320 6/2/2018 2:20:08 PM

 Numerical Integration of ODEs  |  321 

6.5.3.2  Solution of Swing Equations by RK-fourth-order Method

This method is a fourth-order explicit method. Using the value of the states at the kth step, the value
of the states at the (k + 1)th step is obtained as follows:

d ( k +1)  d ( k )  h
w  = w  + F × (6.57)
 ( k +1)   ( k )  6
where
( ) ( ) ( )
F = F A1 x( k ) + 2 × F A2 x((k1)) + 2 × F A3 x((k2)) + F A4 x((k3)) (6.58) ( )
where
 (
w (k ) − w 0 ) 
F A1
( x(k ) ) 
= w


0
(
 2 H 0.8 − Pmax sin d ( k ) ) 


(
w ((1k)) − w 0 ) 
( )
F A2 x((1k))

= w



0
 2H
(
0.8 − Pmax sin d (1)
(k ) ) 


with
1 ()
d ( k )  d ( k ) 
w  = w  + F
A1
x( k ) ×
h
2
( )
 (k )   (k ) 

(
w ((k2)) − w 0 ) 
( )
F A3 x (( 2k ))

= w



0
 2H
(
0.8 − Pmax sin d ( 2)
(k ) ) 


with
( 2)
d ( k )  d ( k ) 
w 
 (k ) 
= 
w ( k ) 
( )
+ F A2 x((1k)) ×
h
2


(
w ((k3)) − w 0 ) 
F A4
( )
x((k3))

= w



0
 2H
(
0.8 − Pmax sin d ( 3)
(k ) ) 


with
( 3)
d ( k )  d ( k ) 
w 
 (k ) 
=
w ( k )
+F

( )
A3 ( 2)
x( k ) × h

It should be noted that F A1, F A2, F A3, and F A4 involve simple function evaluation of F for different
state values. As in the Forward Euler method the function, F, is appropriately changed depending on
the condition of the system, i.e., pre-fault, faulted and post-fault conditions.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 321 6/2/2018 2:20:12 PM

 322  |  Power System Analysis

Table 6.7  States of the SMIB System-Backward Euler Method.

Tclear s 0.1 0.411 0.412

x( 0) F 0.46055, 376.99 0.46055, 376.99 0.46055, 376.99

x( 0) P 0.55603, 378.85 1.7235, 381.96 1.7285, 381.97

Table 6.8  States of the SMIB System-RK-fourth-Order Method.

Tclear s 0.1 0.411 0.412

x( 0) F 0.46055, 376.99 0.46055, 376.99 0.46055, 376.99

x( 0) P 0.55516, 378.85 1.7236, 381.97 1.7286, 381.98

For different fault-clearing time, the initial values of states for the faulted system and the post-fault sys-
tem are tabulated in Tables 6.7 and 6.8, for Backward Euler and RK- fourth order methods, respectively.
For a fault-clearing time equal to 0.411 s, the rotor-angle and speed plots are shown in Figures 6.22
and 6.23, respectively, for RK method as well as Backward Euler method. These figures show that the
trajectories remain stable.
Tclear = 0.411 s
3

2.5

1.5
Delta (rad)

0.5

0
RK−4 (h = 0.001 s)
BEuler (h = 0.001 s)
−0.5
BEuler (h = 0.0001 s)

−1
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (s )

Figure 6.22  The Rotor-angle Plot for Tclear = 0.411 s with Backward Euler
and RK Method.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 322 6/2/2018 2:20:14 PM

 Numerical Integration of ODEs  |  323 

Tclear = 0.411s
384
RK−4 (h = 0.001 s)
BEuler (h = 0.001s)
382 BEuler (h = 0.0001s)

380

378
Omega (rad/s)

376

374

372

370

368
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (s)

Figure 6.23  The Rotor-speed Plot for Tclear = 0.411 s with Backward Euler and RK Method.

However, when the fault-clearing time is increased to 0.412 s, with both these methods the system
exhibits unstable behaviour (see Figures 6.24 and 6.25).
Tclear = 0.412s
14
RK−4
Backword Euler
12

10

8
Delta (rad)

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (s)
Figure 6.24  The Rotor-angle Plot for Tclear = 0.412 s with Backward Euler and RK Method.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 323 6/2/2018 2:20:15 PM

 324  |  Power System Analysis

Tclear = 0.412s
410
RK−4
Backword Euler
405

400
Omega (rad/s)

395

390

385

380

375
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (s )

Figure 6.25  The Rotor-speed Plot for Tclear = 0.412 s with Backward Euler
and RK Method.

From Figures 6.22 and 6.23, we can also note that for a step-size, h = 0.001 s, the estimates obtained
using the Backward Euler method are less accurate than the RK fourth-order method, as it is, just a
first-order method. This is evident, when the solution with the Backward Euler method is repeated
with a lower step-size, h = 0.0001 s.

6.5.4 Evaluation of Critical Clearing-angle

Using Equal-area Criteria
For the purpose of evaluating the critical clearing-angle, a well-known equal-area criterion is employed
in the following lines. In this analysis, the mechanical damping is neglected and the system losses are
assumed to be zero. Note that this analysis does not provide any information about the critical clearing
time and it uses the (transient) power-angle curves in an SMIB-kind of system, employing classical
model for the generators. This method is evolved based on the idea that for the system to be critically
stable, the accelerating area under the power-angle curve must balance the decelerating area. For the
system considered, the three power-angle curves are as shown in Figure 6.26.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 324 6/2/2018 2:20:17 PM

 Numerical Integration of ODEs  |  325 

1.8
Pre−fault
1.6
Post−fault
1.4

1.2 A3 A2
d0
1
Pm
Pe (pu)

0.8
During Fault
0.6

0.4 A1

0.2 ds dcr duep

0
0 0.5 1 1.5 2 2.5 3 3.5
δ (rad)

Figure 6.26  Three Power-angle Curves for the SMIB System - Equal-area Criteria.

From Figure 6.26, the accelerating-area, A1 is compared with the the decelerating area, A2. For
critically stable system, we have

A1 = A2
where,
d cr
A1 = ∫ ( Pm − Pmax2 sin d )dd
d0

with Pmax2 denoting the maximum value of the power during the fault and
d uep
A2 = ∫ ( Pmax3 sin d − Pm )dd
d cr

with Pmax3 denoting the maximum value of the power in the post-fault system.
Thus we have,
d cr d uep

∫ ( Pm − Pmax 2 sin d )dd = ∫ ( Pmax3 sin d − Pm )dd (6.59)

d0 d cr

Evaluating the integration within the limits we get,

   Pm (d cr − d 0 ) + Pmax 2 (cos d cr − cos d 0 ) = Pmax 3 (cos d cr − cos d uep ) − Pm (d uep −d cr ) (6.60)

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 325 6/2/2018 2:20:19 PM

 326  |  Power System Analysis

Rearranging the terms, and writing an expression for dcr, we get,

− Pmax 3 cos d uep + Pmax 2 cos d 0 + Pm (d 0 − d uep )

cos d cr =
( Pmax 2 − Pmax 3 )

Now, substituting the values and solving for the critical clearing-angle, we get,

−1.4623 cos(2.5627) + 0.65 cos(0.46055) + 0.8 (0.46055 − 2.5627)

cos d cr =
(0.65 − 1.4623)

d cr = 1.7248 rad
Note that the above value is close to that indicated in Table 6.8 for Tclear = 0.411 s

6.5.5 Large-signal Stability Evaluation Using Energy

Function Method
In this method, the transient stability performance of a system is analysed in terms of energy associ-
ated with the post-fault condition of the system. In order to determine whether a system remains stable
or unstable following a fault clearing, the following steps are employed [8]:
1. An energy function, W is constructed using the network conditions of the post-fault system.
2. The maximum energy the post-fault system can absorb, Wcr, is computed using the constructed W.
3. By numerically solving only the faulted system and substituting the corresponding state values of
the generators in the function W, the energy Wcl, at the time of fault clearing, is noted.
4. If Wcl ≤ Wcr, then the post-fault system is said to be stable, otherwise, declared as unstable.
Although the energy function method avoids complete time-domain simulation and even pro-
vides a quantitative measure about the degree of system stability, it is not easy to derive an energy
function for a typical multi-machine power system. Having constructed one such function which
accounts the system’s energy content, it is not straightforward to estimate the critical energy accu-
rately. In following section, the method has been demonstrated for the SMIB system discussed
earlier.

6.5.5.1  Construction of an Energy Function for an SMIB System

The swing equation shown in eq. (6.38) is rewritten substituting wm = w − w0 as
dd
= wm
dt

dw m
M = Pm − Pmax sin d = f (d )
dt
2H
where M = and wm is the relative speed rad/s.
w0

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 326 6/2/2018 2:20:21 PM

 Numerical Integration of ODEs  |  327 

To construct an energy function, consider

dd dw m
dt = =M
wm f (d )

f (d )dd = M w m dw m

Integrating the pair on both sides, we get,

d wm

∫ f (d )dd = M ∫ w m dw m
ds 0

substituting for f (d  ) and evaluating the terms, we get,

1 2
Pm (d − d s ) + Pmax (cos d − cos d s ) = Mw m
2

Now, define an energy function for the SMIB system as

1 2
W (d , w m ) = Mw m − Pm (d − d s ) − Pmax 3 (cos d − cos d s ) (6.61)
2

= WKE + WPE

Note that energy function is defined for the post-fault system and the energy components are
identified as

1
Kinetic Energy = WKE = M w 2m (6.62)
2
Potential Energy = WPE = − Pm (d − d s ) − Pmax 3 (cosd − cosd s ) (6.63)

Critical Energy:
In order to obtain the critical energy, which is the maximum energy the post-fault system can absorb,
only the potential energy expression is used. Therefore, set d = duep in the WPE-expression, (6.63),
to get,
Wcr = − Pm (d uep − d s ) − Pmax 3 (cos d uep − cos d s ) (6.64)

Using the post-fault system details, we have,

 P 
d uep = (p − d s ) = p − sin −1  m 
 Pmax 3 

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 327 6/2/2018 2:20:25 PM

 328  |  Power System Analysis

we obtain, ds = 0.57888 and duep = 2.5627. Therefore, we have,

Wcr = −0.8 × (2.5627 − 0.57888) − 1.4623 × (cos(2.5627) − cos(0.57888)) = 0.86105

From the equal-area criterion as well, (see Figure 6.26), it can be verified that

d uep
Wcr = A2 + A3 = ∫ ( Pmax3 sin d − Pm )dd
ds

6.5.5.2 Components of Energy Associated

with the Post-fault System
Here, the fault is cleared with Tclear= 0.410 s. Using the Forward Euler method it is noted earlier that
for this fault-clearing time, the post-fault system remains stable. To analyse the energy components in
the post-fault system, the evaluated states d and w, are substituted in eqs (6.62) and (6.63) to obtain
the kinetic and potential energy components as

1  2H  2 (6.65)
WKE = w
2  w 0 
m

WPE = −0.8 × (d − 0.57888) − 1.4623 × (cos d − cos(0.57888)) (6.66)

where, wm = (w − w0).
Also, the total energy is computed as

WT = WKE + WPE

For this case, the rotor-angle and rotor-speed plots shown in Figure 6.19 are replotted along with
kinetic energy and potential energy components, and the total energy in Figure 6.27.
In Figure 6.27, subplot-3 denotes the kinetic energy (KE) component and subplot-4 represents
the potential energy (PE) variation. In each of these plots the total energy associated with the post-
fault system is also plotted. It is evident that once the fault is cleared the WT remains constant
at 0.8579 pu denoting that there is no damping in the system. Further, the WT value is less than
Wcr (=0.86105 pu) showing that the system is stable. Figure 6.27 also depicts four salient points, A,
B, C, and D, where the maximum energy changes from one form to another. Until point A (where
the fault is cleared), the energy is acquired both in the form of KE and PE. At point B, the acquired
KE completely gets converted into PE, leading to a maximum angle excursion, close to duep.
At point C, this PE, is released to make KE maximum. Thus, this pattern repeats so that WT =
WKE + WPE, is a constant. These observations can also be verified in the phase-plane trajectory
drawn in Figure 6.20.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 328 6/2/2018 2:20:27 PM

 Numerical Integration of ODEs  |  329 

Delta (rad) 2

1 B D
0

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Time (s )
Omega (rad/s)

380

375 C
A (t = 0.51 s)
370
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (s )
1
WT (0.8576 pu)
KE (pu)

0.5

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (s )
1
PE (pu)

WT
0.5

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (s )

Figure 6.27  The Rotor-angle and Rotor-speed Plots for Tclear = 0.410 s
with Energy Components.

6.5.5.3  Determination of Critical Clearing Time

In this case, the differential equations pertaining to the faulted system alone are numerically solved
using the Forward Euler method for a longer duration starting from Tfault. These equations are
given by

dd
= w − w 0 (6.67)
dt

2 H dw
= Pm − Pmax 2 sin d (6.68)
w 0 dt

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 329 6/2/2018 2:20:28 PM

 330  |  Power System Analysis

with [d , w ]T0 = [0.46055 rad , 376.99 rad/s]T . The differential equations related to the per-fault systems
are not integrated since the states value anyway remain at [d , w ]T0 .
The evaluated states, d and w for the faulted system, are substituted in eqs (6.65) and (6.66) to
obtain the total energy WT. This energy is plotted in Figure 6.28 with respect to time.

1.6

1.4

WT
1.2
Wcr = 0.86105 pu

1
Total Energy (pu)

0.8

0.6

0.4

Tfault = 0.1s
0.2 Tcr = 0.512 − 0.1 = 0.412 s

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Time (s)

Figure 6.28  The total energy plot for the faulted system.

From the plot the time instant at which the value of WT is just greater than or equal to Wcr = 0.86105,
is noted as t = 0.512 s. From this, the critical clearing time is calculated as tcr = 0.512 – 0.1= 0.412 s.

6.5.5.4  Determination of Critical Energy

In this method of stability analysis of a power system, it is important to get an accurate estimate of the
critical energy of the post-fault system. In the SMIB system the determination of the critical energy
using the WPE expression, is straightforward because, this requires the calculation of an unique angle,
duep, which can be obtained very easily from the Pm and Pmax3. However, for a multi-machine power
system, there will be a large number of unstable equilibrium points (uep), of which only one set
decides the critical energy. This set is referred to as the controlling uep. It is noted that [8, 9] obtain-
ing this controlling uep may be computationally involved. Therefore, a simulation-based procedure
called exit point method, is generally employed to obtain the critical energy directly using the WPE
expression.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 330 6/2/2018 2:20:30 PM

 Numerical Integration of ODEs  |  331 

Fault−on trajectory
4
duep
3

Delta (rad) 2

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Time (s)

390
Omega (rad/s)

385

380

375
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Time (s)

0.5
Wcr = max(WPE)
PE (pu)

−0.5
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Time (s)

Figure 6.29  The Fault-on Trajectory to Compute the Critical Energy. 

In the previous section (see Figure 6.27), it is seen that in a critical cleared system, when WPE
function is plotted, it reaches a value very close to the critical value of the energy. This is because,
the d-trajectory goes very near to the controlling uep. Using this observation, the exit point method is
devised as follows:
1. Numerically integrate only the faulted power system for a prolonged time. The d values obtained
are substituted in the WPE expression.
2. Note the maximum value of the potential energy. This value is denoted as the critical energy of
the post-fault system (see Figure 6.29).
3. In the SMIB system, the d-trajectory goes through the duep value.

References
[1] Katsuhiko Ogata, Modern Control Engineering, Prentice Hall, PTR, New Jersey, Fourth Edition,
2002.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 331 6/2/2018 2:20:31 PM

 332 | Power System Analysis

[2] Mariesa Crow, Computational Methods for Electric Power Systems, CRC Press, New York, 2010.
[3] Ton van den Boom, Discrete-time systems analysis, 2 October, 2006.
[4] Adharapurapu Hema Latha, Shashidhar M.K., and Shubhanga K.N, ‘A toy model to under-
stand subsynchronous resonance and Real-time simulation of the model using RTAI-Linux’,
Proceedings of the IEEE International conference on Control, Instrumentation, Communication
and Computational Technologies, ICCICCT-2015, India, December, 2015.
[5] Dormand J.R. and Prince P. J., ‘A family of embedded Runge-Kutta formulae,’ Journal of
Computational and Applied Mathematics, vol. 6, pp. 19–26, 1980.
[6] http://blogs.mathworks.com/cleve/2014/05/26/ordinary-differential-equation-solvers-ode23-
and-ode45/
[7] https://help.scilab.org/docs/5.5.0/en_US/DoPri.html
[8] M.A. Pai, Energy Function Analysis for Power System Stability, Kluwer Academic Publishers,
Boston, 1989.
[9] K.R. Padiyar, Structure Preserving Energy Functions in Power System, CRC Press, London, 2013.

Review Questions
1. Give a typical classification of different numerical integration methods.
2. What is an explicit numerical integration technique? How is it different from implicit numerical
integration technique? Explain with an example.
3. Obtain the numerical solution using Forward Euler integration technique for the following tasks:
(a) Obtain the current iL (t ) through an RL-series circuit for a sinusoidal input of the form
v(t ) = Vm sin(w t + a ) . Determine the value of a for which the iL (t ) reaches steady state with-
out any dynamics, for iL (0) = 0. Here, iL (0) = 0 denotes the initial condition on the state, iL.
Consider w = 2p 50 rad/s. Verify the steady-state solution by carrying out the phasor analysis.
(b) Obtain the solution of
⋅⋅
x = −x

Choose x1 = x and x2 = x with x1 (0) = 0 and x2 (0) = 1 as the initial conditions.

(c) Obtain the plot of d (t ) and w (t ) for 3 s with d (t= 0) = 30° and w (t= 0) = 2p 50 * 1.005 rad/s.
dd
= w − 2p 50
dt
dw 2p 50
= [1 − 2sin d ]
dt 2H
Take H = 5 MJ/MVA.
4. Obtain the voltage waveform associated with the capacitor in the following RC-series circuit, see
Figure 6.30. Assume that the capacitor is initially charged to 0.5 V and R = 1 Ω and C = 0.5 F.
Use Backward Euler numerical integration technique with h = 0.01 s.
R

v(t) C

Figure 6.30 RC Series Circuit.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 332 6/2/2018 2:20:41 PM

 Numerical Integration of ODEs | 333

Simulate the circuit for a period of 10 s for the following input signal v(t ), applied at t = 0.
(a) When v(t ) is a unit step.
(b) When v(t ) = 1 + 1.0 sin(2p *5*t ) V.
5. From the literature state the algorithm for Modified Euler method (which is also referred to as
Heun’s method).
6. Using Heun’s method available in SIMULINK, determine the critical clearing time for the
Example given in Section 6.5.
7. Discuss the significance of energy function method of determining the critical clearing time for a
power system.

M06_POWER-SYSTEM-AN00_SE_XXXX_CH06.indd 333 6/2/2018 2:20:42 PM

 Numerical Iterative Methods 7
CHAPTER OUTLINE

7.1 Features of Non-linear Algebraic 7.3 Gauss Seidel Iteration Method

System of Equations 7.4 Newton-Raphson Iteration
7.2 Fixed-Point Iteration Method Method

In this chapter, numerical iterative methods to solve non-linear algebraic system of equations are
discussed. In these methods, the solution process starts from an initial guess, and the guessed value is
updated by applying a correction, whereas in the numerical integration techniques, the time-domain
trajectory of states begins from a specified initial value and the flow progresses with respect to time.
Another important observation is that the process of obtaining a solution through a numerical iterative
method may not lead to convergence of a solution. This is not similar to a numerical integration tech-
nique becoming unstable as it is a time-domain performance.
The simplicity of a fixed-point iterative technique has been discussed through illustrative exam-
ples. Further, its limitation in terms of non-convergence behaviour is also explained. Using typical
power flow problems, Newton Raphson methods have been demonstrated.

7.1 FEATURES OF NON-LINEAR ALGEBRAIC

SYSTEM OF EQUATIONS
Numerical iterative techniques are used to solve non-linear algebraic equations. The need arises when
we analyse several power system problems. For example, while performing power flow solution, when
implicit numerical integration techniques are used to solve non-linear differential equations, and so
on. Some characteristics of non-linear algebraic systems are as follows:
1. Consider a non-linear algebraic function of the form F(x), where x is a vector of variables and F
represents a vector of functions. The requirement is to find x = x * such that F ( x * ) = 0 , then x *
is said to be the solution of F ( x ).

M07_POWER-SYSTEM-AN00_SE_XXXX_CH07.indd 334 9/12/2017 8:05:43 AM

 Numerical Iterative Methods  |  335 

2. For such systems, a closed form of solution is not possible. Solution can be obtained only through
iterative techniques.
3. By any iterative technique, only an approximate solution of x *, denoted as x ′, is possible so that
F ( x ′ ) ≈ 0.
Some important issues with an iterative procedure are as follows:
1. Selection of a correct initial guess closer to x *.
2. Construction of an iterative function.
3. Obtaining updates for variables.
4. Checking whether the sequence of updates converge to the desired solution.
5. Number of iterations to achieve x ′.
6. The choice of error tolerance to stop the iterative process.
In literature, many simple iteration schemes such as bisection method, fixed-point iteration method
(also known as Gauss method), modified Gauss method, referred to as Gauss Seidel method, and
many Newton-based methods are suggested for solving non-linear algebraic equations [2]. In this
chapter, we discuss some of these methods in brief.

7.2  FIXED-POINT ITERATION METHOD

Although many Newton-based efficient iteration schemes are documented in literature [1], fixed-point
iteration technique offers an easy solution scheme to solve simple non-linear algebraic equations. The
steps involved in fixed-point iteration scheme for a scalar function are as follows:
1. The given equation,  f (x) = 0, is written in an alternate form
x = g(x)
2. Make an initial guess, x( 0).
3. Compute x(1) = g(x(0)).
4. Thus, a sequence of updates are obtained using the previous updates.
x(k + 1) = g(xk)
5. The iteration is ended when
| x( k +1) − xk | ≤ e

Note that e is chosen based on the accuracy of the solution desired.

7.2.1  Example: A Scalar Function

Consider f ( x ) = e − x − x = 0. The requirement is to solve for x′ which satisfies f ( x ′ ) = 0 . The itera-
tive function is given by:
− x( k )
x( k +1) = e
with initial guess x(0) = 1.

M07_POWER-SYSTEM-AN00_SE_XXXX_CH07.indd 335 9/12/2017 8:05:47 AM

 336  |  Power System Analysis

Table 7.1  A Scalar Function—Fixed-point Iteration Scheme.

k x(k) | x( k +1) − x( k ) |

0 1 0.6321
1 0.3679 0.3243
2 0.6922 0.1917
3 0.5005 0.1058
4 0.6062 0.0608
5 0.5454 0.03422
6 0.5796 0.0195
7 0.5601 0.0110
8 0.5711 0.00626
9 0.5649 0.0035
10 0.5684 0.0020
11 0.5664 0.0011
12 0.5675 0.00064
13 0.5669 0.00036

The sequence of updates is listed in Table 7.1. From the tabulated results, it can be seen that the
error | x( k +1) − x( k ) | continuously reduces, denoting that the updates are converging. If the error toler-
ance is set to 0.0005, the iterative process can be stopped at k = 13 and the solution for x is x′ = 0.5669.
The process of convergence to solution, x′ is graphically shown in Figure 7.1.

0.9

0.8

0.7 y1 = e−x

0.6
y1 and y2

0.5

0.4

0.3 y2 = x

0.2 X*

0.1

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
x

Figure 7.1  Graphical Representation of Scalar Function and Fixed-point Iteration.

M07_POWER-SYSTEM-AN00_SE_XXXX_CH07.indd 336 9/12/2017 8:05:49 AM

 Numerical Iterative Methods  |  337 

7.2.2  Example: Power Flow in a Simple Power System-1

For the system shown in Figure 7.2, V1 = 1.0 ∠0° and the load power, S L = 1.0 + j 0.5 p.u. Determine
V2 using fixed-point iteration method assuming V2( 0) = 1.0 ∠0°.

RL = 0.05 pu XL= 0.2 pu

V2
V1
+ I

SL (Complex power)

Figure 7.2  Power Flow in a Simple Power System-1 (Fixed-point Iteration).

By applying Kirchhoff’s voltage law, we obtain

V2 = V1 − I ( RL + jX L )
*
(7.1)
S 
where I =  L  and an iterative function can be written as
V  2
S L*
V2( k +1) = V1 − ( RL + jX L )
V2*( k )

The sequence of updates is listed in Table 7.2.

Table 7.2 Power Flow in a Simple Power System-1

(Fixed-point Iteration Scheme).

k V2( k ) pu

0 1.0 ∠0°

1 0.86783∠ − 11.634°

2 0.80661∠ − 11.634°

3 0.79364 ∠ − 12.739°

4 0.78638∠ − 12.739°

5 0.78464 ∠ − 12.887°

 
11 0.78320 ∠ − 12.911°

M07_POWER-SYSTEM-AN00_SE_XXXX_CH07.indd 337 9/12/2017 8:06:00 AM

 338  |  Power System Analysis

Notes:
1. The following criterion is used to stop the iteration process:

V2( k +1) − V2( k ) ≤¨

With e = 0.00001, the solution converged in 11 iterations.

2. The complex power supplied by source-1 is computed as

S1 = V1 ( I ) = 1.10189 + j 0.90756 pu
*

Note that the current is calculated using the converged value of V2 and is equal to
I = 1.4275∠ − 39.476ο p.u. The line losses are noted as 0.10189 + j0.40757 p.u.
3. Consider an iteration function obtained as follows:
(V1 − V2 )
I =
( RL + jX L )
We know that S L = V2 I *. Using the above expression for I in SL and rearranging for V2 , we get,

S L ( RL + jX L )
*
V2( k +1) = (7.2)
(V1 − V2( k ) )*

With V2( 0) set to 0.99 (any value other than 1.0), the solution will converge to a very low value
(= 0.29429 ∠ −36.488° ), which is not acceptable. Similarly, if other iterative functions are tried,
it may even diverge.
4. The example given above can be solved by approaching it as the solution of quadratic equation
representing V2 in rectangular form. However, here, it is desired to demonstrate using the fixed-
point iteration technique.

7.2.3  Example: Power Flow in a Simple Power System-2

In the above example, a capacitor of reactance, Xc = 1.2 p.u., is connected across the complex power
load as shown in Figure 7.3.

RL = 0.05 pu XL= 0.2 pu V2

V1
+ I

SL (Complex power)
Xc = 1.2 pu

Figure 7.3  Power Flow in a Simple Power System-2 (Fixed-point Iteration).

M07_POWER-SYSTEM-AN00_SE_XXXX_CH07.indd 338 9/12/2017 8:06:09 AM

 Numerical Iterative Methods  |  339 

Table 7.3 Power Flow in a Simple Power System-2

(Fixed Point Iteration Scheme).

k V2( k ) pu
0 1.0 ∠0°

1 1.0395∠ − 12.031°

2 1.0070 ∠ − 12.214°

3 0.99624 ∠ − 12.454°

4 0.99194 ∠ − 12.541°

5 0.99025∠ − 12.577°

 

11 0.98913∠ − 12.600°

Here, the expression given in eq. (7.1) is used with a current given by

V2( k ) S L*
I = +
− jX c V2*( k )

The sequence of updates is listed in Table 7.3.

Notes:
1. With e = 0.00001, the solution converged in 11 iterations.
2. The complex power supplied by source-1 is computed as

S1 = V1 ( I ) = 1.056183 − j 0.090564 pu
*

3. The line current is I = 1.0601∠ 4.90° . The line losses are noted as 0.056186 + j 0.224745 pu.
4. The reactive power generated by the capacitor is Qc = 0.81531 p.u. It can be verified that
Qc − (0.5 + 0.224745) = 0.090564 p.u. This implies that source-1 is absorbing the additional
reactive power.
5. From the previous case study, it can be seen that with the introduction of a capacitor, the line
current reduces appreciably and the line losses decrease by nearly 50%.

M07_POWER-SYSTEM-AN00_SE_XXXX_CH07.indd 339 9/12/2017 8:06:14 AM

 340  |  Power System Analysis

7.2.4 Example: Simultaneous Non-linear Equation

Consider

f1 ( x, y ) : x 2 + xy − 10 = 0

f 2 ( x, y ) : y + 3 xy 2 − 57 = 0
with x0 = 1.5 and y0 = 3.5
For such problems, the iterative function is given by:
x( k +1) = g1 ( x( k ) , y( k ) )

y( k +1) = g2 ( x( k ) , y( k ) )
and the convergence criterion is

(
max x( k +1) − x( k ) , y( k +1) − y( k ) ≤ e )
Employing the above scheme, the first equation can be rewritten as

x( x + y ) = 10
10
x=
( x + y)
Hence, the first iterative function is given by
10
x( k +1) =

( x(k ) + y(k ) ) (7.3)
Similarly, for the second equation, the iterative function is given by
57
y( k +1) =

(1 + 3 x( k ) y( k ) ) (7.4)
Using eqs (7.3) and (7.4), the updates are obtained as shown in Table 7.4.

Table 7.4  Simultaneous Non-linear Equation-fixed Point Iteration Scheme.

k x(k) y(k)
0 1.5 3.5
1 2 3.4030
2 1.8508 2.6613
3 2.2162 3.6129
4 1.7155 2.2781
  

M07_POWER-SYSTEM-AN00_SE_XXXX_CH07.indd 340 9/12/2017 8:06:19 AM

 Numerical Iterative Methods  |  341 

It should be noted that these values fail to converge. Hence, one may require to choose alternate iter-
ative functions. It is observed that in this example, even with other choices for the iterative functions,
the solution tends to diverge. Hence, a variation in the fixed-point method is employed in the following
lines, while constructing an iterative function. This modification leads to the Guass Seidel method.

7.3  GAUSS SEIDEL ITERATION METHOD

Here, new iterative functions are constructed as follows [3]:
10
x( k +1) =

( )
x( k ) + y( k ) (7.5)

57
y( k +1) =

(1 + 3x(k +1) y(k ) ) (7.6)
It is found that the above iterative functions converge to solution, x = 2 and y = 3. The intermediate
updates are tabulated in Table 7.5.

Table 7.5  Simultaneous Non-linear Equation-Gauss Seidel Iteration Scheme.

k x(k) y(k)
0 1.5 3.5
1 2 2.5909
2 2.1782 3.1789
3 1.8667 3.0316
4 2.0415 2.9130
  
12 1.9995 2.9980
13 2.0010 3.0005

This method is generally used for a problem involving more than one variable. The application of
this method to power-flow problems is found to take many iterations. In this context, Newton-based
methods are generally preferred.

7.4  NEWTON-RAPHSON ITERATION METHOD

This method is noted to take less number of iterations compared to Gauss-based method due to its
quadratic convergence property [2]. Here, the method has been demonstrated for a scalar case and for
a system with n variables.

7.4.1  Iterative Function for a Scalar System

Consider a non-linear function of the form
f(x) = 0 with x(0) as the initial guess.

M07_POWER-SYSTEM-AN00_SE_XXXX_CH07.indd 341 9/12/2017 8:06:21 AM

 342  |  Power System Analysis

Here, it is desired to obtain x^  so that f(x^ ) = 0. In this process, we start with an initial guess x(0) for
the solution and obtain an update x(1) = x( 0) + Dx( 0) . Thus, it is required to determine this correction
quantity Dx( 0) such that

( )
f x( 0) + Dx( 0) ⇒ 0

To evaluate Dx( 0) , the above function is expanded employing the Taylor series as

∂f

( ) ( )
f x( 0) + Dx( 0) = f x( 0) + |x Dx( 0) + higher order terms = 0 (7.7)
∂x ( 0 )

Assuming that the error in the initial guess is relatively small and hence neglecting the higher order
terms, we have,
∂f

0 = f x( 0) + ( )|x Dx( 0) (7.8)
∂x ( 0 )

∂f ∂f
It is to be noted that the notation |x( 0 ) denotes that is evaluated at x(0). Thus, the correction
∂x ∂x
quantity, Dx( 0) is calculated as
−1
 ∂f
 ∂x

Dx( 0) = −  |x( 0 ) 

( )
f x( 0) (7.9)

Now, the new update, x(1) is obtained as,
−1
 ∂f
 ∂ x

x(1) = x( 0) −  |x( 0 ) 

( )
f x( 0) (7.10)

The algorithm can be summarised as follows:

1. Start with an initial guess x(0).

2. Obtain an update x(k + 1) using x(k), employing an iterative function given by

−1
 ∂f
 ∂ x

x( k +1) = x( k ) −  |x( k ) 

( )
f x( k ) (7.11)

3. If the following convergence criterion is satisfied,

( x(k +1) − x(k ) ) ≤ e

stop the iteration process.

7.4.2  Iterative Function for a System with n Variables

Consider a system with n variable as F ( x ) = 0. The function, F, in turn, consists of n sub-functions
given by

M07_POWER-SYSTEM-AN00_SE_XXXX_CH07.indd 342 9/12/2017 8:06:27 AM

 Numerical Iterative Methods  |  343 

 f1 ( x )   x1 
  x 
f 2 ( x )
F (x) =  x= 
2
with
    
   
 f n ( x )  xn 

As in the scalar case, we can obtain the correction vector as

( ) ( )
−1
Dx( k ) = −  J x( k )  F x( k ) (7.12)

where the Jacobian is given by

 ∂f1 ∂f1 ∂f1 

 ∂x 
∂x2 ∂ xn 
 1 
 
 ∂f ∂f 2 ∂f 2 
 ( )
 J x( k )  =  2
  ∂x1 ∂x2

∂ xn 
 
   
 ∂f n ∂f n ∂f n 
  
 ∂x1 ∂x2 ∂ xn  x
(k )

Now, the new update x( k +1) is obtained as

x( k +1) = x( k ) + Dx( k )

The iteration process is terminated if

(
max x( k +1) − x( k ) ≤ e )
Note: To avoid explicit inverse calculation of the Jacobian, eq. (7.12) is rewritten as follows and solved
for the unknown vector, Dx( k ) ,

 ( )  ( )
 J x( k )  Dx( k ) = − F x( k ) (7.13)

7.4.3 Example: Simultaneous Non-linear Equation

with Newton-Raphson Method
Consider the system given by

f1 ( x, y ) : x 2 + xy − 10 = 0

M07_POWER-SYSTEM-AN00_SE_XXXX_CH07.indd 343 9/12/2017 8:06:30 AM

 344  |  Power System Analysis

f 2 ( x, y ) : y + 3 xy 2 − 57 = 0
The iterative function is given by

( ) ( )
−1
z( k +1) = z( k ) −  J z( k )  F z( k ) (7.14)

T
where z = [ x, y ] and
2 x( k ) + y( k ) x( k ) 
 ( )
 J z( k )  = 
 2
 3 y( k )

1 + 6 x( k ) y( k ) 

Iteration 1: The update is calculated using z ( 0) = [1.5, 3.5]T .

 6.5 1.5   −2.5 

 ( )
 J z( 0)  = 

 36 . 75 32 .5
and  ( )
 F z( 0)  = 
 1.625
 
Therefore, we can write,

1.5  1  32.5 −1.5   −2.5 

z(1) =   −   
3.5 156.12  −36.75 6.5  1.625

Thus,
1.5  0.53603  2.0360 
z(1) =   +  = 
3.5  −0.65612 2.8439 
T
Continuing with the above update, we get z( 2) = [1.9987, 3.0023] . At the end of the third iteration,
T
the solution converges to the solution, z(3) = [2, 3] .

7.4.4 Example: Power Flow in a Simple Power System-1

(Newton-Raphson Method)
For a power system shown in Figure 7.2, by applying Kirchhoff ’s current law (KCL) at bus-2, we get,

 PL + jQL 
*
=I =
(V1 − V2 ) (7.15)
 V2 
 ZL

where Z L = ( RL + jX L ) =| Z L | ∠q L .

Now, we can write that

*
V −V 
PL + jQL = V2  1 2  (7.16)
 ZL 

M07_POWER-SYSTEM-AN00_SE_XXXX_CH07.indd 344 9/12/2017 8:06:35 AM

 Numerical Iterative Methods  |  345 

Substituting for V2 = V2 ∠d and V1 = 1.0∠0°, we get

 1 − V2 ∠ − d 
PL + jQL = V2 ∠d   (7.17)
 ZL ∠ −q L 

The above equation can be rewritten as
1 V 2 ∠q L
PL + jQL = V2 ∠(d + q L ) − 2 (7.18)
ZL ZL

Comparing real and imaginary parts on both sides, we get the power balance equations as
1 
PL = P (V2 , d ) = V2 cos (d + q L ) − V22 cos (q L ) (7.19)
| ZL |  

1 
QL = Q (V2 , d ) = V2sin (d + q L ) − V22sin (q L ) (7.20)
| ZL |  

Identifying the states of the system as z = [V2 , d ]T and using the Taylor series expansion as in
eq. (7.8), we get an iterative function for the first iteration as

(
PL = P V2( 0) , d ( 0) + ) 1 
| ZL | 
( )
cos d ( 0) + q L − 2V2( 0) cos (q L ) DV2( 0) +

+
1 
ZL 
( )
−V2( 0) sin d ( 0) + q L  Dd ( 0) (7.21)

(
QL = Q V2( 0) ,d ( 0) + ) 1 
ZL 
( )
sin d ( 0) + q L − 2V2( 0) sin (q L ) DV2( 0) +

+
1 
ZL 
( )
V2( 0) cos d ( 0) + q L  Dd ( 0) (7.22)

The above function can be written in matrix form as


(
 PL − P V2( 0) , d ( 0) )  =  J11 J12   DV2( 0) 
 Dd 

 (
QL − Q V2( 0) , d ( 0) )  J 21 J 22 
( 0)  ( 0 )  (7.23)

In general, at kth iterative step, we have,

J11 =
1 
| ZL |  ( )
cos d ( k ) + q L − 2V2( k ) cos (q L ) ; J12 =
1 
| ZL | 
−V2( k ) sin d ( k ) + q L  ( )

J 21 =
1 
| ZL | 
( )
sin d ( k ) + q L − 2V2( k ) sin (q L ) ; J 22 =
1 
| ZL | 
( )
V2( k ) cos d ( k ) + q L 

and the function is given by

M07_POWER-SYSTEM-AN00_SE_XXXX_CH07.indd 345 9/12/2017 8:06:40 AM

 346  |  Power System Analysis


(
 PL − P V2( k ) , d ( k ) )  =  J11 J12   DV2( k ) 
 Dd  (7.24)

(
QL − Q V2( k ) , d ( k )
 )  J 21 J 22 
(k )  (k ) 

Note that the matrix

(
 PL − P V2( k ) , d ( k )

)  =  DP 
(
QL − Q V2( k ) , d ( k )
 )  DQ 
is referred to as the residue of the system, and to stop the iteration process, the following criterion
is used:

( )
max DP ≤ e
DQ

The states are updated as

V2( k +1)  V2( k )   DV2( k ) 
d = +  (7.25)
 ( k +1)  d ( k )   Dd ( k ) 

With Z L = 0.2062∠75.96° and V2( 0) = 1.0 ∠0°, for k = 0, we can write eq. (7.23) as

1.0 − 0   −1.1765 −4.7059  DV2( 0) 

0.5 − 0  =  −4.7059 1.1765   Dd  (7.26)
    ( 0)  (0) 

Solving the above linear system of equations the correction vector is obtained as

 DV2( 0)   −0.15 
 Dd = 
 ( 0)   −0.175
and the first update is determined as
V2(1)   0.85 
d  =  −10.027°
 (1)   

The iterations are continued until the maximum value of the residue falls below 0.0001.
For this tolerance, the solution converges in five iterations and the bus-2 voltage is given by
V2 = 0.78319∠ − 12.911° p.u.

7.4.5 Example: Power Flow in a Simple Power System-2

(Newton-Raphson Method)
For a power system shown in Figure 7.3, by applying KCL at bus-2 we get,
*
 PL + jQL  V V −V
  + 2 = I = 1 2 (7.27)
V2  − jX c ZL

M07_POWER-SYSTEM-AN00_SE_XXXX_CH07.indd 346 9/12/2017 8:06:47 AM

 Numerical Iterative Methods  |  347 

Rewriting the above expression, we get

* *
V −V   V 
PL + jQL = V2  1 2  − V2  2  (7.28)
 ZL   − jX c 

Substituting V2 = V2 ∠d and V1 = 1.0 ∠0°, and rearranging, we get,

 (1 − V2 ∠ − d )  V22
PL + jQL = V2 ∠d   + j (7.29)
 ZL ∠ −q L  Xc

The above equation can be further simplified as

1 V 2 ∠q L V2
PL + jQL = V2 ∠(d + q L ) − 2 + j 2 (7.30)
ZL ZL Xc

Comparing real and imaginary parts on both sides, we get the power balance equations as

1 
PL = P1 (V2 , d ) = V2 cos (d + q L ) − V22 cos (q L ) (7.31)
| ZL |  

1  V2
QL = Q1 (V2 ,d ) = V2sin (d + q L ) − V22sin (q L ) + 2 (7.32)
ZL   X
c

By similarity, we get an iterative function for the first iteration as


(
 PL − P1 V2( 0) , d ( 0) )  =  J11 J12   DV2( 0)  (7.33)
 Dd 
 (
QL − Q1 V2( 0) , d ( 0) )  J 21m J 22 
( 0)  ( 0) 

For kth iteration, we have

J11 =
1 
( )
1 cos d ( k ) + q L − 2V2( k ) cos (q L ) ;
 J12 =
1 
( 
1  −V2( k ) sin d ( k ) + q L  )
( )
J11 =Z L cos d ( k ) + q L − 2V2( k ) cos (q L ) ;
ZL
(
J12 =Z L  −V2( k ) sin d ( k ) + q L 
ZL
)
2V
J 21m =
1 
sin d ( + q − 2V ) sin (q ) (
 + 22V( k ) ; J 22 = 1 V2( k ) cos d ( k ) + q L  )
1 
( )  1
(
)
( k ) L 2 ( k ) L
J 21m =Z L sin d ( k ) + q L − 2V2( k ) sin (q L ) + X c
2( k )
; J 22 =Z L V2( k ) cos d ( k ) + q L 
ZL Xc ZL
In eq. (7.33) the initial values of the states are substituted to obtain

 1.0 − 0   −1.1765 −4.7059  DV2( 0) 

0.5 − 0.83333 =  −3.0392 1.1765   Dd  (7.34)
   ( 0)  ( 0) 
Solving the above linear system of equations, the correction vector is calculated as

 DV2( 0)   0.025 
 Dd = 
 ( 0)   −0.218750 

M07_POWER-SYSTEM-AN00_SE_XXXX_CH07.indd 347 9/12/2017 8:06:51 AM

 348 | Power System Analysis

and the first update is obtained as

V2(1)  1.0250 
d  =  ο
 (1)   −12.533 

The iterations are continued until the maximum value of the residue falls below 0.0001. For this
tolerance, the solution converges in four iterations and the bus-2 voltage is given by V2 = 0.98912∠ − 12.600ο p.u.
V2 = 0.98912∠ − 12.600ο p.u. These results are very close to those obtained with fixed-point iteration technique.
Here, although the number of iterations are very few compared to the fixed-point method, the compu-
tations involved are relatively more. However, in most cases, a reliable convergence is not guaranteed
with fixed point iteration method as it strongly depends on the type of iterative function employed.
Note: It should be noted that while handling non-linear algebraic equations using fixed-point iteration
technique, the variable can be either real or complex, whereas with Newton-Raphson method, (which
involves evaluation of Jacobian) the variable should comprise real values. If they are complex varia-
bles, they must be rewritten in terms of rectangular or polar forms.
The solution of a conventional load-flow problem offers bus voltage magnitudes and angles which
are valid under sinusoidal steady-state condition of a power system. This load-flow problem differs
from a linear circuit analysis due to the fact that the power-flow equations are written in terms of
complex power specifications for system loads and generators. In power system analysis, the load-
flow-like calculations are carried out in it following cases [4]:
1. In future-expansion planning analysis, to understand bus voltage profile and to assess transmission
adequacy.
2. In operation and control, for example, to determine economic seheduling of generation.
3. In security analysis, to obtain utilization of line capacity within loading limits.
4. To establish the initial operating point for stability studies.
5. As a tool in state-estimation to get best idea about line flows and bus voltages.

References
[1] Mariesa Crow, Computational Methods for Electric Power Systems, CRC Press, New York, 2010.
[2] Steven C. Chapra and Raymond P. Canale, Numerical Methods for Engineers, Tata McGraw-Hill
Publishing Co. Ltd., New Delhi, 2000.
[3] Richard L. Burden and J. Douglas Faires, Numerical Analysis, Thomson Learning, USA, 2001.
[4] Kusic, G.L., Computer Aided Power System Analysis, Prentice-Hall, Englewood Cliffs, NewJersey,
1986.

Review Questions
1. What is the basic difference between a numerical integration method and numerical iteration
method?
2. Solve the following non-linear system of equations employing fixed-point iteration technique:
(a) sin( x ) = 0.5 with appropriate initial value of x.
(b) 0.9091 sin(d ) + 0.2597 sin(2d ) = 0.5 with appropriate initial value of d .
(c) −0.9 x 2 + 1.7 x + 25 = 0 with x0 = 5.

M07_POWER-SYSTEM-AN00_SE_XXXX_CH07.indd 348 9/12/2017 8:06:54 AM

 Numerical Iterative Methods | 349

3. Consider a generator connected to an infinite bus as shown in the following single-line diagram
(see Figure 7.4).
VSSSC
I + −
G
R jx
E V

Figure 7.4 Series Static-synchronous Compensator-connected SMIB System.

A voltage source VSSSC is injected in series with line so that

VSSSC = − j 0.2e jf t ,

where f t is the angle of the line current I = I ∠f t .

Note that VSSSC is such that it tracks the current phasor and it remains quadrature to the current.
This implies that the injected voltage behaves like a capacitor. Calculate I , employing fixed-point
iteration technique.
Given that E = 1.0 ∠30°, V = 1.0 ∠0°, and R + jx = 0.01 + j0.5 . All quantities are in per unit.
Use f t( 0) = 0.
4. For the system shown in Figure 7.5, V1 = 1.0 ∠0° .
RL = 0.05 pu XL = 0.2 pu V
2
V1
+ I
SL (complex power)

Figure 7.5 A Sample Power System with a Voltage-dependent Load.

The load power, which is a function of the bus voltage magnitude, is given by,

(
S L = 1.0 × | V2 |4 + j 0.5 × | V2 |4 ) pu
Determine V2 using fixed-point iteration and Newton–Raphson methods. Verify the power
balance.
5. Solve the equation:
x2 − x + 1 = 0

employing Fixed point iteration and Newton–Raphson methods.

M07_POWER-SYSTEM-AN00_SE_XXXX_CH07.indd 349 9/12/2017 8:07:00 AM

 Fault Analysis of Power
Systems 8
CHAPTER OUTLINE

8.1 Introduction to Fault Analysis 8.3 Analysis of Unsymmetrical Faults

8.2 Symmetrical Three-phase
Short-Circuit Analysis

In this chapter, short-circuit fault analysis of a power system is discussed. Both symmetrical three-
phase and unsymmetrical faults are analysed to perform fault current calculations and power system
stability studies.
Symmetrical fault analysis is carried out using bus admittance matrix, employing efficient solution
techniques in MATLAB. For analysing unsymmetrical faults, symmetrical component transformation
has been introduced while comparing it with Park transformations. Further, sequence impedances
of power system components are detailed so as to build a sequence network and hence construct the
corresponding bus admittance matrix instead of building a bus impedance matrix using connectivity
details of the components.
Different types of unsymmetrical shunt faults are analysed, computing various system quantities,
taking into account the phase-shifting effects of transformers. Employing the derived fault equations,
transient stability analysis has been carried out with interconnected generators.

8.1 INTRODUCTION TO FAULT ANALYSIS

Occurrence of a fault in a normally operating power system is treated as a disturbance since it causes
severe magnitude of fault currents and upsets the load-generation balance. High fault currents not
only damage costly electrical equipment due to their limited short-time over-load ratings, but also
may cause danger to human life as they produce excessive electromagnetic forces and high ground
voltages. Since faults are generally accompanied by a large bus voltage dip, they lead to tripping of
loads and a reduction in the electrical output of the generators. Such a condition causes the generator
to accelerate, affecting the stability performance of a power system.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 350 1/29/2018 1:56:41 PM

 Fault Analysis of Power Systems  |  351 

Therefore, it is necessary to sense the occurrence of a fault reliably through a proper relaying
scheme and to trip the faulty element/section selectively, using adequately rated protective gears. To
obtain the data required to coordinate the relaying systems and to choose an appropriate rating for
protective gears (such as circuit breakers), a fault analysis is generally carried out. Here, the main
objective is to obtain an approximate estimate of the fault currents in different parts of the network
unlike in a power system stability study involving faults.
In fault analysis, the synchronous machine is a major electrical component which influences the
fault current in the system. Therefore, an accurate modelling of a synchronous generator is very
essential. In order to simplify the analysis, while using a synchronous machine model, the stator tran-
sients are generally neglected in addition to the effect of speed deviation. This leads to a fundamental
frequency model for a synchronous machine. For such a model of a synchronous machine, we can see
that following the application of a fault, the machine offers different reactances as a function of time.
These reactances are generally stated as follows:

1. Subtransient reactance: This is valid in the first few cycles of the fault condition, where field
winding and all damper windings actively take part in machine dynamics.
2. Transient reactance: This is considered to be effective in the later cycles before a steady state is
reached. During this period, the transients in damper windings are mostly decayed and only the
field winding dynamics are effective.
3. Synchronous reactance: This corresponds to steady-state condition of the machine.

In a power system stability study, a generator is implemented using a set of differential equations to
completely capture the transition from the initial part of the disturbance (where subtransient condi-
tions are effective) to a condition where the system reaches a steady state. However, in fault analysis,
only a ‘snap-shot’ of the system condition is considered following a fault, involving only linear alge-
braic equations in terms of phasor quantities. Thus, fault analysis ‘appears’ like a steady-state analysis.
In most cases, simplified dynamic models of synchronous machines are employed which are applica-
ble in different time frame of study. In addition, other rotating machines such as induction machines
are also considered with their transient or subtransient models.

8.2  SYMMETRICAL THREE-PHASE SHORT-CIRCUIT ANALYSIS

In this analysis, a sinusoidal steady-state condition is assumed for the entire system with fault even if
subtransient duration of the synchronous machine is considered. This leads to a linear system analysis,
permitting phasor-based computations. Since network is assumed to be balanced (even with a fault),
the complete system is symmetrical with respect to all three phases. Hence, a single-phase equivalent
circuit with positive sequence impedances is employed for the analysis. In addition, the following
simplifying assumption are made:

1. Generator and network elements are lossless.

2. Pre-fault load currents are negligible.
3. The controllers on the generator are disabled.
4. Only rotating type of loads such as synchronous machines, induction machines, etc., are consid-
ered as they contribute to fault as a voltage source.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 351 1/29/2018 1:56:41 PM

 352  |  Power System Analysis

5. Dynamic saliency of synchronous machines is neglected as it offers a single-phase equivalent

circuit for the machine. Thus, a voltage behind a subtransient or transient reactance can be
employed.
6. At a time, one fault occurs at a given bus, i.e., simultaneous-type of faults are not considered.

In most cases, the pre-fault bus voltages are set to 1.0 ∠ 0° pu for ease of analysis. This assumption
removes the requirement of running a full-fledged power flow and it corresponds to ‘just synchronised
condition’ of the synchronous machine. This is identical to the statement that the pre-fault currents
in the network are neglected. Such a fault analysis is generally done to get a rough estimate of the
magnitudes of the fault currents. These details are used:

1. To select circuit breakers of suitable ratings.

2. To decide the rating of measuring instruments.
3. To set the operating point of relays for co-ordination purpose.
4. To carry out network planning.

8.2.1  Two-machine, Five-bus Power System

In order to perform symmetrical three-phase fault analysis, consider a power system as shown in
Figure 8.1.

PL3
T1 Y Y
1 2 3 4
G4
G1

Y Y T2

PL5

Figure 8.1  Single-line Diagram of Two-machine, Five-bus Power System.

The positive-sequence impedances of the components are given below on 865 MVA base:
Generators G1 and G4: 865 MVA, 22 kV, 60 Hz, x″d = 0.135 pu, x′d = 0.169 pu, and xd = 1.79 pu.
Transformers T1 and T2: 22/500 kV, 60 Hz with tap setting=1.0 pu and Zeq = j0.1 pu.
Transmission lines: 500 kV, 60 Hz.
Line between buses 2 and 3: xL = 0.5 pu, Line between buses 3 and 5: xL = 0.1 pu.
To determine the fault current for a three-phase symmetrical fault at bus 2, first construct the
positive-sequence impedance diagram as shown in Figure 8.2.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 352 1/29/2018 1:56:41 PM

 Fault Analysis of Power Systems  |  353 

j0.1 j0.5 j0.1 4

1 2 3

j0.1 j0.135
j0.135

+ 5 +
E1 =1.0 0° E4 =1.0 0°

Figure 8.2  The Positive-sequence Impedance Diagram of Two-machine,

Five-bus Power System.

8.2.1.1  Symmetrical Three-phase Fault at Bus 2

Now, connect bus 2 to ground as shown in Figure 8.3. The pre-fault voltage at all buses is set equal to
1.0 ∠ 0° pu. Since pre-fault load currents are neglected, E1 = E4 = 1.0 ∠ 0o.

j0.1 j0.5 j0.1 4

1 2 3

j0.1 j0.135
j0.135 IfG4

+ 5 +
E1 =1.0 0° IfG1 0°
If2 E4 =1.0

Figure 8.3  Symmetrical Three-phase Fault at Bus 2 in Two-machine, Five-bus Power System.

From Figure 8.3, it is clear that the fault current, I f 2 is given by

I f 2 = I fG1 + I fG 4

E1 E4
= +
j (0.135 + 0.1) j (0.135 + 0.1 + 0.5)

 1 1 
= 1.0  +
 j 0.235 j 0.735 

= − j 4.255319 − j1.360544

= − j5.615863 pu
(8.1)
Note that the fault current is dominantly inductive in nature.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 353 1/29/2018 1:56:46 PM

 354  |  Power System Analysis

8.2.1.2  Symmetrical Three-phase Fault at Bus 5

Now consider a three-phase symmetrical fault at bus 5 and the corresponding equivalent circuit with
fault at bus 5 is shown in Figure 8.4.

j0.1 j0.5 j0.1 4

1 2 3

j0.1 j0.135
j0.135
IfG1
+ 5 IfG4 +
E1 =1.0 0° E4 =1.0 0°

If5

Figure 8.4  Symmetrical Three-phase Fault at Bus-5 in Two-machine,

Five-bus Power System.

Writing KVL for each loop, we get,

j (0.135 + 0.1 + 0.5 + 0.1) I fG1 + j 0.1I fG 4 = E1

j 0.1I fG1 + j (0.135 + 0.1 + 0.1) I fG 4 = E4

Writing the above equations in matrix form, we have

 j 0.835 j 0.1   I fG1  1.0 ∠ 0o 

 j 0.1  = 
 j 0.335  I fG 4  1.0 ∠ 0o 
 

Solving for I fG1 and I fG4 , we get

0.335 −0.1  1.0 ∠ 0 

o
 I fG1  1
  = − j  
 I fG 4  0.269725  −0.1 0.835 1.00 ∠ 0o 
 

0.871258
= −j 
2.724998

Therefore, the fault current at bus 5 is

I f 5 = I fG1 + I fG 4 = − j 3.596256 pu

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 354 1/29/2018 1:56:50 PM

 Fault Analysis of Power Systems  |  355 

8.2.2 Analysis of Fault Current Using Thevenin’s Theorem

Here, fault current calculations are done in three steps:
1. Obtain the Thevenin’s voltage, VThf at bus f, where a dead short-circuit connection is to be
applied to determine the fault current through the short. This simply denotes the bus voltage,
VPRf prior to the application of the short.
2. Compute the Thevenin’s impedance, ZThf , as ‘seen’ from bus f ‘looking into’ the system. This
involves passivation of all sources before measuring the impedance. This means that all sources
of the rotating machine-type, are replaced by their appropriate internal impedances.
3. The fault current is given by
VThf
If = (8.2)
ZThf

The corresponding equivalent circuit is denoted in Figure 8.5.

ZThf bus f

VPRf
I
f

Figure 8.5:  Thevenin’s Equivalent Circuit of Fault at Bus f.

Now, for a fault at bus 2, VPR2 = 1.0 ∠ 0° and the passivated equivalent circuit appears as shown in
Figure 8.6.

j0.1 j0.5 j0.1 4

1 2 3

j0.1 j0.135
j0.135
ZTh2
5

Figure 8.6  Passivated Equivalent Circuit for Fault at Bus 2.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 355 1/29/2018 1:56:51 PM

 356  |  Power System Analysis

The ZTh2 is obtained by connecting (j 0.135 + j 0.1) in parallel with (j 0.135 + j 0.1 + j 0.5), and is
given by
( j 0.235)( j 0.735)
ZTh2 = = j 0.178067 pu (8.3)
j (0.235 + 0.735)

Therefore, the fault current is given by

1.0 ∠0o
If2 = = − j 5.615863 pu
j 0.178067

o
For a fault at bus 5, the VPR5 = 1.0 ∠ 0 and the passivated equivalent circuit appears as shown in
Figure 8.7.

j0.1 j0.5 j0.1 4

1 2 3

j0.1 j0.135
j0.135
ZTh5 5

Figure 8.7  Passivated Equivalent Circuit for Fault at Bus 5.

The ZTh5 is obtained by connecting (j 0.135 + j 0.1 + j 0.5) in parallel with (j 0.135 + j 0.1) and in turn
in series with j 0.1 and is given by :

( j 0.735)( j 0.235)
ZTh5 = + j 0.1 = j 0.278067 pu (8.4)
j (0.735 + 0.235)

The fault current is given by

1.0 ∠0o
If5 = = − j 3.596255 pu
j 0.278067

8.2.2.1 Procedure to Determine Thevenin Impedance at a Bus

From the previous analysis, it is evident that in order to perform fault analysis, it becomes simpler if we
use Thevenin-based calculations. However, for a large network, it is not straightforward to determine
the ZThf at a bus. Therefore, a systematic procedure is presented below employing circuit-­analysis
based on the bus-frame-of-reference.
First all voltage sources are converted into current source as

E1 E4
I GS1 = and I GS 4 = (8.5)
jxd′′1 jxd′′4

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 356 1/29/2018 1:56:55 PM

 Fault Analysis of Power Systems  |  357 

where I GS1 and I GS 4 represent current injections into the buses as shown in Figure 8.8.

j0.1 j0.5 j0.1 4

1 2 3

+ +
j0.1
IGS1 j0.135
IGS 4
5 j0.135

Figure 8.8  Impedance Diagram with Current Injections.

By applying KCL at each node, node-equations are established as follows:

 1 1   1 
V1  + − V2  = I GS1
 j 0.135 j 0.1  j 0.1

 1   1 1   1 
−V1   + V2  +  − V3  =0
 j 0.1  j 0.1 j 0.5   j 0.5 

 1   1 1 1   1   1 
−V2   + V3  + + − V4  − V5  =0
 j 0.5  
 j 0.5 j 0.1 j 0.1 
 j 0.1  j 0.1

 1   1 1 
−V3  + V4  + = I GS 4
 j 0.1  j 0.1 j 0.135 

 1   1 
−V3   + V5  =0
 j 0.1  j 0.1

The above set of equations can be assembled in a matrix form as follows:

 1 1  1 
 +  − 0 0 0 
  j 0.135 j 0.1 j 0.1 
 1  1 1  1 
−  j 0.1 + j 0.5  − 0 0  V1   I GS1 
 j 0.1 j 0.5    
  V2  0 
0 1  1 1 1  1 1     (8.6)
− + + − − V3 = 0
 j 0.5  j 0.5 j 0.1 j 0.1 j 0.1 j 0.1    
  V4   I GS 4 
0 1  1 1   V  0 
 0 −  j 0.1 + j 0.135  0  5  
j 0.1
 
 1 1 
0 0 − 0
 j 0.1 j 0.1 

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 357 1/29/2018 1:56:59 PM

 358  |  Power System Analysis

The above equation can be compactly written as

[YBUS ]( n × n ) V ( n ×1) =  I ( n ×1) (8.7)

b b b b

where
[YBUS] represents the bus admittance matrix;
[V] represents the vector of bus voltage;
[I] represents the vector of bus current injections;
nb is number of buses excluding the ground reference bus.
From the circuit theory, we know that the bus impedance matrix [ZBUS] is obtained as

[ Z BUS ] = [YBUS ]−1 (8.8)

The [ZBUS] elements are given by

 Z11 Z12 Z13 Z14 Z15 

Z Z22    
 21 
[ Z BUS ] = . 
 
. 
 Z51 Z52   Z55 

The diagonal elements of [ZBUS], Zii with (i = 1, 2, 3, 4, 5), denotes the Thevenin’s impedance, ZThf
at the fth bus.
From eq. (8.8), we have,

[YBUS] [ZBUS] = [U](n × nb) (8.9)

b

where [U(n × nb)] denotes the identity matrix and is given by

b

[U ]( nb × nb ) =  e1 , e2 , …, e f , …, e nb  (8.10)

and e f is a column vector of zeros with 1 at the fth row position, i.e.,

e f = [0, 0, …, 1, , …, 0]T (8.11)

Notes:
1. The impedance matrix [ZBUS] is generally a full matrix, i.e, in a total of (nb × nb) elements, most
elements are non-zeros. However, [YBUS] is generally sparse. This implies that in this matrix,
most entries are zeros since in a typical power system network, a bus is connected to a few num-
ber of other buses.
2. From the above analysis, it can be seen that the construction of [YBUS] is straightforward and can
be built directly from the connection details of the network [5].

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 358 1/29/2018 1:57:02 PM

 Fault Analysis of Power Systems  |  359 

3. Since [YBUS] contains a few non-zero elements, by declaring the matrix variable as sparse (in
MATLAB [12]), it enables less memory storage requirement and usage of sparsity-based solution
techniques [7]. This results in reduced number of floating-point operations.
4. [YBUS] and [ZBUS] are symmetric matrices (without phase-shifting transformers [5]) and corre-
spond to positive-sequence network of the power system.
Therefore, while computing [ZBUS] elements, explicit inverse of [YBUS] is not carried out. In most
fault analysis, the entire [ZBUS] determination is not required as fault bus is chosen at a time. To get
the f th column of [ZBUS], the expression given in eq. (8.9) is rewritten as

[YBUS ]( n × n )  Z jf ( n ×1) = e f ( n ×1) (8.12)
b b b
b

The above linear equation is in the form [A]x = b, and we can solve for (x = [Zjf ]) by simply executing
the following command in MATLAB as follows:

x = A\b

where, x and b are column vectors of dimension (nb × 1), and A is a matrix of dimension (nb × nb).
The above command corresponds to back-slash function of the MATLAB and it solves the unknown
vector by using efficient LU-decomposition techniques.
Now, from eq. (8.12), [Zjf ] is obtained as

 Z1 f 
Z 
 2f 
 
 
Zif
 Z jf  =  
   
 
 Z ff 
 
 
 Z nb f 
( nb ×1)

and Zff is the desired ZThf.

From eq. (8.6), for the two-machine, five-bus power system, the positive sequence [YBUS] is
obtained as,

 − j17.40741 j10 0 0 0 
 j10 − j12 j 2 0 0 
 
[YBUS ] = 0 j2 − j 22 j10 j10  (8.13)
 
0 0 j10 − j17.40741 0 
0 0 j10 0 − j10 
( 5 × 5)

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 359 1/29/2018 1:57:03 PM

 360  |  Power System Analysis

For a fault at bus 2, e2 = [0, 1, 0, 0, 0]T and from (8.12) [Zj2] ( j = 1, 2, 3, 4 and 5) is determined as

 Z12   j 0.10229
Z   j 0.17807
 22   
 Z j2  =  Z32  =  j 0 . 05693 
     
Z
 42   j 0 . 03271 
 Z52   j 0.05693
( 5×1)

and, we can note ZTh2 = Z22 = j0.17807 pu and is identical to that given in (8.3). Similarly, for a fault
at bus 5, e5 = [0, 0, 0, 0, 1]T and [Zj5] ( j = 1, 2, 3, 4 and 5) can be obtained as,

 Z15   j 0.03271
Z   j 0.05693
 25   
 Z j5  =  Z35  =  j 0.17807
     
 Z 45   j 0.10229
 Z55   j 0.27807
(5×1)

and, we can observe that ZTh5 = Z55 = j0.27807 pu and is identical to that shown in (8.4). If all ele-
ments of [ZBUS](5 × 5) are to be obtained, it can be carried in MATLAB using a single-instruction as

Z_BUS = Y_BUS\U
where Y_BUS is of dimension (5 × 5) and U is constructed in MATLAB as :

U = eye(5)

where ‘eye(5)’ creates an identity matrix of dimension (5 × 5).

Thus, [ZBUS](5 × 5) is determined as,

 j 0.116211 j 0.102294 j 0.032706 j 0.018789 j 0.032706 

 j 0.102294 j 0.178067 j 0.056933 j 0.032706 j 0.056933 
 
[ Z BUS ](5×5) =  j 0.032706 j 0.056933 j 0.178067 j 0.102294 j 0.178067  (8.14)
 
 j 0.018789 j 0.032706 j 0.102294 j 0.116211 j 0.102294 
 j 0.032706 j 0.056933 j 0.178067 j 0.102294 j 0.278067 
( 5 × 5)

8.2.3 Calculation of Bus Voltages in a Faulted System

In order to determine the fault currents through elements connected between various buses and to
know the swing performances of synchronous machines in power system stability studies, it is neces-
sary to compute the bus voltages during the faulted condition of the system.
The fault current, If , computed using (8.2) simply denotes the current, which the network drives at
bus f, ‘into the ground’, while forcing the pre-fault bus voltage VPRf  to zero. This implies that if (-If) is
‘injected’ into the network at bus f, with respect to ground as in Figure 8.9, it forces the bus voltage to zero.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 360 1/29/2018 1:57:05 PM

 Fault Analysis of Power Systems  |  361 

j0.1 j0.5 j0.1 4

1 2 3

+ +
j0.1
IGS1 j0.135
IGS4
( − If2 ) 5 j0.135

Figure 8.9  Fault Current Injection in the Impedance Diagram Along with Source Current Injections.

Now using eq. (8.7) we have,

 I GS1   − j 7.407407
   j 5.615863 
− I f 2   
[YBUS ] VPF  = 0  = 0  (8.15)
 
 I GS 4   − j 7.407407
0  0 
  (5×1)

Solving for [VPF], we get the bus voltage during the fault at bus 2 as,

0.42553
0 
 
VPF  = 0.68027 (8.16)
( 5×1)  
0.81633
0.68027

8.2.3.1  Calculation of Bus Voltages Using [ ZBUS  ]

The bus voltage can also be determined using [ZBUS] as follows:
We have,

[ Z BUS ]( n × n )  I ( n ×1) = V ( n ×1)

b b b b
(8.17)

Now, knowing [ZBUS] elements from eq. (8.14) and substituting the bus-current injection-vector, we
get,

 I GS1 
 
− I f 2 
[ Z BUS ](5×5) 0  = VPF 
( 5×1)
 I GS 4 
0 
  ( 5×1)

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 361 1/29/2018 1:57:07 PM

 362  |  Power System Analysis

Since the network is linear, using the principle of superposition, we can write the current-injection
vector [I ] as a combination of pre-fault currents and fault current to get,

 I GS1  0 
  − I 
 0   f 2
[ Z BUS ](5×5) 0  + [ Z BUS ](5×5) 0  = VPF (5×1) (8.18)
 
 I GS 4  0 
0  0 
  (5×1) (5×1)

In the above equation, we can clearly see that the first term on the left-hand side represents the pre-
fault bus voltage vector, VPR, and these values are already known to us. Note that in this vector,
VPRf  denotes the fth-bus voltage prior to the application of the fault. The second term in eq. (8.18), on
the left-hand side, shows the change in bus voltages ∆V caused by a fault current injection. Pertaining
to bus 2, we can rewrite the second term as

0   Z12 
1  Z 
   22 
( )
 ∆V  = [ Z BUS ](5×5) 0  − I f 2 =  Z32  − I f 2 (8.19)
   
( )
0   Z 42 
0   Z52 

With If 2 = VPR2 the above equation modifies as

Z22

 Z12  
 
 Z22  
1 
 
 Z32  
 
 ∆V  = −  Z22   VPR2 (8.20)
 Z  
 42  
 Z22  
 
 Z52  
 Z22  
 

Using eq. (8.20) in eq. (8.18), we obtain the bus voltage during the fault as,

VPF  = VPR  +  ∆V  (8.21)

( 5×1) ( 5×1) ( 5×1)

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 362 1/29/2018 1:57:10 PM

 Fault Analysis of Power Systems  |  363 

or

 −0.57447
 −1.0 
 
VPF  = VPR  +  −0.31973
( 5×1) ( 5×1)  
 −0.18367
 −0.311973

Since VPR is set equal to 1.0 ∠ 0° pu at all buses, the bus voltages during fault are obtained as

0.42553
0 
 
VPF  = 0.68027
( 5×1)  
0.81633
0.68027

The above results are identical to that listed in eq. (8.16).

8.2.3.2 Calculation of Bus Voltage for a Fault at Bus 5

The bus voltage during the fault is obtained employing the following steps:

• Step1: Note the pre- fault voltage [VPR].

• Step2: Compute [Zj5] using (8.12).
• Step3: Noting [VPR5] from [VPR]-vector, compute If 5 as

VPR5
If5 = (8.22)
Z55

• Step4: Calculate the change in bus voltage [∆V ], due to the fault as

 ∆V  =  Z j 5  ( − I f 5 ) (8.23)
• Step5: Compute

0.88238
0.79525
 
VPF  = VPR  +  ∆V  = 0.359633
( 5×1) ( 5×1)  ( 5×1)  
0.63213
0 
( 5×1)

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 363 1/29/2018 1:57:12 PM

 364  |  Power System Analysis

From the above calculations, we can see that if bus voltage during a fault is to be determined, it is
easier to obtain using [Zjf] calculations, rather than employing [YBUS] and [I ] due to the following
reasons:
1. If and ZThf have to be determined through [Zjf ].
2. Once [Zjf ] is calculated, [∆V ] can be readily obtained.

8.2.4 Bus Voltage Calculation without the Knowledge

of the Fault Current
The following procedure can be employed, if required, to determine the bus voltage during the fault
condition without computing the fault current values:
We know that,
[YBUS ]( n × n ) VPF ( n ×1) =  I ( n ×1) (8.24)
b b b b

As indicated in eq. (8.15), to obtain [VPF], the bus current injection vector should be modified with
(-If) at the f th location. On computing [VPF], the fth bus voltage, VPFf , will be equal to zero. Therefore,
in the above [YBUS] matrix, the fth row and the fth column are removed to obtain YBUS f 
of dimen-
 
sion (nb – 1) × ( nb – 1). Similarly, in [I ] vector, (-If) substitution is skipped by removing the f th row
element to reduce its dimension to (nb – 1) × 1. Thus, the bus voltage calculation equation becomes,

Y f  V f  = I f  (8.25)
 BUS  (( nb −1) ×( nb −1))  PF  (( nb −1) ×1)   (( nb −1) ×1))

Therefore, for a fault at bus 2, the expression, given in eq. (8.15) becomes,

 − j17.40741 0 0 0   − j 7.407407
0 − j 22 j10 j10  0 
  V f  = 
0 j10 − j17.40741 0   PF  ( 4 ×1)  − j 7.407407
   
 0 j10 0 − j10  0  ( 4 ×1)
( 4 × 4)

Solving for bus voltages, we get,

0.42553
0.68027
V f  = 
 PF  ( 4 ×1) 0.81633
 
0.68027 ( 4 ×1)

This is identical to that obtained earlier in eq. (8.16), except that there exits VPF2 which is zero. It is to
be noted that while computing the bus voltages without computing the fault current If , the dimension
of the [YBUS] is altered. This may impose difficulties when fault analysis is carried out in a large system.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 364 1/29/2018 1:57:14 PM

 Fault Analysis of Power Systems  |  365 

For example, when power system stability analysis is done for a symmetrical fault in the network, it
requires additional programming to reduce the size of the matrix while running time–domain simula-
tion. To simplify this, the following procedure is employed:
In the pre-fault [YBUS], at a fault bus, a large conductance, say 106, is added to introduce a short-
circuit. Thus, for a symmetrical three-phase fault at bus 2, the faulted YBUS
f 
appears as,
 

 − j17.40741 j10 0 0 0   − j 7.407407

  0 
 j10 106 − j12 j 2 0 0   
0 j2 − j 22 j10 j10  VPF  = 0 
  ( 5 ×1)  
0 0 j10 − j17.40741 0   − j7.407407
0 0 j10 0 − j10  0 
 ( 5 × 5) ( 5×1)

Now, solving for the bus voltages, we get,

0.42553
0 
 
VPF  = 0.68027
( 5×1)  
0.81633
0.68027
( 5×1)
Notes:
1. In most low-frequency power system stability analysis programmes, the bus-current injection
models are employed for generators including network devices. In such cases, it is easy to model
the power system network in terms of the bus-admittance matrix instead of the bus-impedance
matrix.
2. As stated earlier, the bus-admittance matrix is highly sparse and hence, it involves a smaller
number of floating point operations when bus voltages are computed using the sparse-enabled
LU-decomposition-based solution techniques provided by the back-slash command.

8.2.5 Calculation of System Quantities During a Fault

For example, the machines power base, SB = 865 MVA has been chosen as the system MVA base. The
base current in the transmission section is given by

865 × 106
IB = = 1.73 kA
500 × 103

1. For the fault at bus 2, the fault current in phase-a, in amperes, is obtained as

1.73
I fa = − j 5.615863 × = − j 5.6092 kA
3

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 365 1/29/2018 1:57:17 PM

 366  |  Power System Analysis

For the other two phases, the currents are given by

I fb = − j 5.6092∠ − 120o = 5.6092∠ − 210o kA

I fc = − j 5.6092∠120o kA = 5.6092∠ 30o kA

The above angles are with respect to the fault bus voltage, where all pre-fault bus voltages are set
to 1.0 ∠ 0° pu. The above currents are shown in Figure 8.10.

c
Ifc Ifb Ifa

Zf Zf Zf

Figure 8.10  Three-phase Symmetrical Fault at a Bus.

In Figure 8.10, Zf  denotes the fault impedance. With this, the expression for the fault current
given in eq. (8.2) modifies as,
VThf
If = (8.26)
( ZThf + Z f )
In this example, Zf is set to zero and such a fault is known as bolted-type three-phase faults.
2. The fault current through a line connected between buses 2 and 3 is obtained as,

VPF 3 − VPF 2 0.68027 − 0

I f 32 = = = − j1.3605 pu
Zl j 0.5

Note that this line current is equal to IfG4 in Figure 8.3. The line current in phase-a, in amperes,
is given by
1.73
I f 32a = − j1.3605 × = − j1.3589 kA
3
For the other two phases, the currents are given by

I f 32b = 1.3589∠ − 210o kA; I f 32c = 1.3589∠30o kA

3. The generator terminal voltage are computed as follows:

Neglecting the phase-shift introduced by the ∆ - Y transformer T1, we get,
VPF1 = 0.42553 × 22 = 9.3617 kV

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 366 1/29/2018 1:57:20 PM

 Fault Analysis of Power Systems  |  367 

Further, the phase-a voltage is given by

22
VPF1a = 0.42553 = 5.4050 kV
3
4. The generator-1 current is obtained as follows:
E1 − VPF1 1.0 − 0.42553
I fG1 = =
jxd′′1 j 0.135
= − j4.25533pu

The current base in the generator 1 circuit is,

865 × 106
IB = = 39.318 kA
22 × 103
Therefore, the generator phase-a current is given by,
39.318
I fG1a = − j 4.25533 × = − j 96.5976 kA
Notes: 3

1. The generator voltage and current calculations do not take into account the 30° phase-shift intro-
duced by ∆-Y transformer. In power system stability calculations, where real and reactive powers
shared by a generator are of importance, the phase shift introduced by a ∆-Y transformer is gener-
ally neglected. This is because, in a ∆-Y transformer, both voltage and current undergo the same
phase shift, without affecting the power computations. A discussion to this effect will be taken up
in the later part of the chapter.
2. In power system studies, a generator is treated as a source of only positive-sequence voltages and
hence it contributes to average power only by driving a positive-sequence currents.
3. The phase shift introduced by a ∆-Y transformer is of significance if voltage and current levels are
to be computed under unbalance system conditions.

8.2.6 Effect of Synchronous Machine Models on the Fault Current

In order to understand the effect of the synchronous machine models on the fault current at different
time frames, in the earlier example, the following machine models (see Figure 8.11) are employed
under no-load condition of the machine at buses 1 and 4.

jx''d jx'd jxd

+ + +
E E E

Subtransient Transient Steady−state

Figure 8.11  Simplified Generator Models for Fault Studies.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 367 1/29/2018 1:57:23 PM

 368  |  Power System Analysis

In a lossless machine, if a fault occurs under no-load condition, the q-axis current remains zero. This
is because, following the fault, the q-axis circuits are not energised at all, due to zero initial value of
the d-axis internal voltage. This implies that the fault current is decided only by the d-axis parameters.
Further, note that all bus voltages are set to 1.0 ∠ 0° at the pre-fault condition and the internal voltages
behind the reactances E, are also equal to 1.0 ∠ 0° pu since there is no current in the pre-fault network.
For a fault at bus 2, the fault current and bus voltages are tabulated in Table 8.1 for different syn-
chronous machine models.

Table 8.1  Fault Current and Bus Voltages for Shunt-fault at Bus 2.

Model x″d = 0.135 x′d = 0.169 xd = 1.79

If 2 −j5.615863 −j5.01786 −j0.94751

VPF1 0.42553 0.37175 0.05291

VPF2 0 0 0

VPF3 0.68027 0.65020 0.20921

VPF4 0.81633 0.78023 0.25105

VPF5 0.68027 0.65020 0.20921

From Table 8.1, it can be seen that the fault current magnitude is the highest during subtransient
period and lowest at the steady-state. Also, it is clear that the bus voltages are quite low in the steady-
state condition. The relaying systems are expected to sense the fault during the subtransient period
and isolate the faulty section through circuit-breaker operation. Such a scheme is desired to protect
equipment from damage.

8.2.7  S
 ymmetrical Fault Calculations Accounting
Pre-fault Load Currents
In order to perform the fault calculations with pre-fault load currents, the following procedure is
employed:
1. Using the bus-specifications and network-connectivity details, a conventional power-flow analysis
is carried out. The results are tabulated in Table 8.2.

Table 8.2  Power Table Flow Results - Two-machine, Five-bus Power Systems.

Bus no. V q (deg) Pg0 Qg0 PL0 QL0

1 1.000000 0 0.8 0.248341 – –
2 0.978442 −4.689 – – – –
3 0.976987 −29.425 – – 1.7 –
4 1.00000 −23.551 1.0 0.281446 – –
5 0.976933 −30.026 – – 0.1 –

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 368 1/29/2018 1:57:23 PM

 Fault Analysis of Power Systems  |  369 

From the converged load flow results, voltages and bus power injections of the generators are
noted.
2. Employing a detailed generator model and through a systematic initial condition calculation
procedure, the subtransient/transient internal voltage behind a subtransient/transient reactance is
computed. This method is readily usable in a normal power system stability calculations. Since
this process involves many calculations, to simplify the fault analysis where an estimate of fault
current alone is desired, the internal voltages are computed using simplified dynamic models (see
Figure 8.11) as follows:
For subtransient condition,

E = Vto + jxd′′ I o (8.27)

For transient condition,

E = Vto + jxd′ I o (8.28)

where
*
 Pgo + jQgo 
Io =  
 Vto

and (Pgo + jQgo) is the complex bus power injection of the generator.
In the above calculations, the dynamic saliency of the machine is neglected. For the example,
I1o = 0.8377∠ − 17.246o and I 4o = 1.0389∠ − 39.269o and the internal voltages of the genera-
tors are given by

E1 = 1.0392 ∠ 5.965o and E4 = 1.0467 ∠ − 16.140o pu

3. The loads are represented by a constant admittance and are given by

P   QLo 
YLo =  Lo
2  − j  2 
 VLo   VLo 

where PLo and QLo represent the real and reactive powers drawn by a load at bus voltage VLo.
For example, YLo3 = 1.78103 pu, YLo5= 0.10478 pu, and the [YBUS] given earlier in eq. (8.13), is
augmented with YLo at the load buses. This implies that in the [YBUS], only the following elements
are updated:
old
Y33 = Y33 + YLo3 = 1.78103 − j 22

old
Y55 = Y55 + YLo5 = 0.10478 − j10

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 369 1/29/2018 1:57:27 PM

 370  |  Power System Analysis

4. Setting e2 = [0,1,0,0,0]T in (8.12), the new [Z j2] is obtained as

0.0032 + j 0.1012 
0.0055 + j 0.1762 
 
 Z j2  = 0.0172 + j 0.0512 
   
0.0099 + j 0.0298 
0.0177 + j 0.0510 

Note that Z22 with load is equal to (0.0055 + j0.1762), and the value of Z22 without considering
the load is j0.17807. This shows that the change in Z22 is negligible with the inclusion of the
loads.
5. Compute the fault current If 2 as

VPR2 0.978442∠ − 4.689o

If2 = = = 5.5497∠ − 92.90o puu (8.29)
Z22 0.0055 + j 0.1762

Note that If 2 without accounting pre-fault load is 5.6158 ∠ - 90°.

6. Using the change in bus voltages, [∆V] = [Zj2](-If 2), the bus voltages during the fault are obtained
as [VPF] = [VPR] + [∆V], and are given by,

0.44219∠ 5.96o 
 
0 
 o
VPF  = 0.68166∠ − 32.91  (8.30)
 o
0.82810 ∠ − 23.98 
0.68163∠ − 33.51o 
 

Comparing the above [VPF] with those without the loads, we can see that the effect of load is
insignificant with regard to the amplitude of the voltages.
7. The fault current contributed by the generators is determined as,

E1 − VPF1
I fG1 = = 0.4596 − j 4.3980 = 4.4219∠ − 84.03o pu
jxd′′1

E4 − VPF 4
I fG 4 = = 0.3380 − j1.8435 = 1.8743∠ − 79.61o pu
jxd′′4

It is observed that in the unloaded case, I fG1 = 4.2553∠ − 90o pu and I fG 4 = 1.3605∠ − 90o pu.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 370 1/29/2018 1:57:30 PM

 Fault Analysis of Power Systems  |  371 

8. The load currents during the fault are given by

I Lf 3 = VPF 3YLo3 = 1.0192 − j 0.6597 pu

I Lf 5 = VPF 5YLo5 = 0.0595 − j 0.0394 pu

9. The line current, I32, is obtained as,

VPF 3 − VPF 2
I 32 = = ( −0.7408 − j1.1445) pu
jx L

We can verify that I fG1 + I 32 = I f 2 and I Lf 3 + I Lf 5 + I 32 = I fG 4 .

10. The complex power outputs of generators 1 and 4 are given by
*
S g1 = VPF1 I fG1 = 0 + j1.9553 pu

*
S g 4 = VPF 4 I fG 4 = 0.87625 + j1.28106 pu

Note that G1 does not supply any real power due to the fault, whereas G4 supplies a real power
load equal to
Pf 3 + Pf 5 = V2PF3 YL03 + V2PF5 YL05

= 0.82757 + 0.048682 = 0.87625 pu

From the above calculations, the following observations are made:

1. A symmetrical, three-phase fault seriously influences the real power balance, thus affecting the
stability of a power system.
2. The fault-current contribution of a generator is decided by the location of the fault relative to the
generator and load buses.
3. The fault current amplitude does not change appreciably whether the pre-fault load is considered
or not.
4. The slack bus voltage is taken as the reference for all angle measurements.
Therefore, pre-fault loading levels are important for power system stability analysis against fault.
However, if an estimate of fault current is desired for the purpose designing/coordinating the protec-
tive gears/relaying systems, pre-fault loads can be neglected.

8.2.8  Short Circuit Capacity at a Bus

From the previous example, neglecting pre-fault load currents, consider that a line between buses
2 and 3 interconnects Area-2 (where generator 4 exists) to Area-1 (where generator-1 is present), at
bus-2 as shown in Figure 8.12.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 371 1/29/2018 1:57:34 PM

 372  |  Power System Analysis

Area−2
Area−1
3

Figure 8.12  Interconnection of Areas to Have Two-machine Five-bus Power System.

From the fault calculations, for a fault at bus 2, we can write the Thevenin equivalent circuit with
respect to bus 2 as shown in Figure 8.5 and the fault current is obtained as

VTh2 VPR2
If2 = = (8.31)
ZTh2 Z22

If we compute the complex power using the above quantities, we get,

S f 2 = VPR2 I f*2 = 1.0 ∠ 0o × 5.615863∠ 90o = 5.615863∠ 90o

This denotes the capacity of the system to drive a fault current at bus 2 with a solid short-circuit,
i.e., with fault impedance, Zf = 0. This value is attributed as a short-circuit capacity (SCC) of the
system at bus 2. This value is generally calculated neglecting the pre-fault load currents and assuming
VPRf = 1.0 ∠ 0°. Therefore, the power associated with fault is given by,
*
 VPRf  1.0
S ff = VPRf   = * (8.32)
Z
 ff  Z ff

The following observations are made:

1. In a fault analysis, since network is assumed to be lossless, and pre-fault load currents
are neglected, the Zff is assumed to be purely reactive and the magnitude of SCC in pu is
specified as,
1
SCC f = (8.33)
x ff

2. The SCCf in per-unit, when multiplied by the base MVA, SB, indicates the ‘fault MVA’ at the fth
bus. This implies that if fault MVA at a bus is high then, following a fault (with Zf = 0), it drives
a large fault current.
3. If SCC (in pu) at bus k, is given, the impedance (in pu) as viewed into the network from the kth-
bus can be readily determined using eq. (8.33). With this, the model at bus k can be written as in
Figure 8.13.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 372 1/29/2018 1:57:36 PM

 Fault Analysis of Power Systems  |  373 

jx bus k
kk

+
o
1.0 0

Figure 8.13  Thevenin’s Equivalent Circuit at Bus k.

Also, note that xkk gives an indication about the source impedance. If a bus has infinite SCC, it
implies zero source impedance. Such a bus would be strong enough to maintain a constant voltage
under normal operating condition (except, for a short circuit at that bus itself). Thus, SCC decides
the voltage regulation of a bus.
4. The SCC value can be used to specify the interruption capacity of a circuit breaker. In terms
of SCC, say, with respect to bus 2, before the interconnection of the areas, we can write the
following:
(a) SCC of area 1 is 4.255319 pu and that of area 2 is 1.360544 pu.
(b) Using the SCC, the source impedance can be computed as
1 1
x22(1) = = 0.235 and x22( 2) = = 0.735
4.255319 1.360544

(c)
With interconnection, the system appears as in Figure 8.14.

j0.235 pu bus 2 j0.735 pu

+ +
o
1.0 0
o 1.0 0

Figure 8.14  Interconnection of Equivalent Circuits of Two Areas at Bus 2.

The above example clearly shows that with interconnection of sources, the fault level at a bus increases.

8.3  ANALYSIS of UNSYMMETRICAL FAULTS

In addition to performing analysis of symmetrical faults, it is desirable to study the unsymmetrical
faults, in view of the practical importance of faults not involving all three phases. Although the major
effect of unsymmetrical faults is to increase the apparent fault impedance, it is required to analyse a
power system behaviour under unbalanced fault conditions to design and monitor system protection
schemes as 95% of the faults that occur on power system are unsymmetrical faults.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 373 1/29/2018 1:57:37 PM

 374  |  Power System Analysis

In the previous sections, the fault analysis of a three-phase balanced system is carried out on sin-
gle-phase basis since the system remains balanced even in the presence of a fault. Therefore, once
voltage and current are computed in the positive sequence equivalent circuit (pertaining to phase-a),
the voltages and currents in phase-b and phase-c are simply written by introducing an appropriate
phase-shift of 120°, assuming abc-phase sequence. Thus, in these analysis, it should be noted that both
the network and the terminal operating condition due to a fault are balanced. However, in a symmetric
network, even if the terminal condition alone is unbalanced (due to a fault or load), it is not straight-
forward to carry out the analysis in the abc-frame. In order to simplify the analysis, a transformation
procedure, known as symmetrical components, is applied to the original system [1, 2].
The symmetrical component transformation is similar to the Park transformation. The following
observations are made with regard to these transformations:
1. The Park transformation is a time-varying and real-valued transformation which transforms the
instantaneous three-phase quantities from the abc-frame to the machine or 0dq-frame, whereas
the symmetrical component transformation is a constant and complex-valued transformation
which transforms three-phase phasor quantities in the abc-frame to three sets in the sequence
domain, referred to as positive-, negative-, and zero-sequence.
2. The zero component in the machine frame is identical to the zero-sequence quantity, and both
d- and q-components for the fundamental frequency synchronous machine model are treated as
the positive-sequence quantities.
3. The 0dq-components are real-valued quantities, whereas the quantities resulting from the sym-
metrical component transformations are the fundamental frequency phasors pertaining to phase-a
of the respective three-phase sequence networks which are balanced.
4. Similar to the concept of reactances in the 0dq-frame, impedances are derived for each of the
linear sequence networks.
5. The Park transformation can be viewed as if it produces a two-phase equivalent from a three-
phase quantity, whereas the symmetrical component transformation produces residues in the
form of sequence components if three-phase phasor quantities are not balanced.
As per the theory of symmetrical components [3, 4], it is assumed that the entire network is balanced,
except at the fault point. Therefore, a balanced network with an unsymmetrical fault at a bus is ana-
lysed by transforming the unbalanced system from the abc-frame to the sequence domain. In the
sequence domain, the following steps are employed:
1. The original unbalanced system (having the a-b-c phase sequence) is decomposed into three bal-
anced sequence networks given by the following:
(a) Positive-sequence network: This is a three-phase symmetrical network made up of posi-
tive-sequence impedances and has the same phase-sequence as that of the original system.
Since a generator is treated as a source of only positive-sequence voltage, this network alone
has pre-fault bus voltages which have fault-driving ability. For this network, the positive-­
sequence bus admittance matrix, [YBUS(1)] is constructed directly using network connectivity
details.
(b) N
 egative-sequence network: This is a three-phase symmetrical network made up of nega-
tive-sequence impedances and has a phase-sequence, a-c-b which is opposite to that of the
original system. For this network, the negative-sequence bus admittance matrix, [YBUS(2)] is

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 374 1/29/2018 1:57:37 PM

 Fault Analysis of Power Systems  |  375 

constructed similar to that of positive-sequence network. It is noted that this matrix differs
from [YBUS(1)] only with respect to generators.
(c) Zero-sequence network: This is a three-phase balanced network made up of zero-sequence
impedances. In this network, voltages/currents in each phase are equal in magnitude and are
in-phase with respect to one another. They are referred to as co-phasors.
It is noted that both positive- and negative-sequence networks have no mutual coupling among the
elements in their respective networks. However, in the zero-sequence network, the flow of ground
currents can cause mutual coupling between transmission line elements (each of three-phase)
which share a common tower/corridor. In such cases, a bus admittance matrix cannot be directly
constructed. In literature [2, 5], the fault analysis is carried out using bus impedance matrix
employing [ZBUS] -building algorithm [6]. This method also handles mutual coupling between
elements. However, it is advantageous to perform fault calculations following bus admittance
approach as it enables the usage of sparsity-based solution techniques [7]. A procedure to handle
mutuals in [YBUS(0)] construction, is presented in the later part of the text.
2. Since, the negative- and zero-sequence currents that flow during unbalanced faults are driven only
by the positive-sequence voltage source at a fault bus, the effects of a fault are represented by
their Thevenin equivalents at the fault bus. Therefore, Thevenin impedance is computed for each
sequence network with respect to fault bus, on a single-phase basis.
3. The terminal unsymmetric fault condition at a bus in the abc-frame is also transformed into the
sequence domain and is represented by an appropriate combination of sequence voltages and
currents. To suit this combination, the Thevenin equivalent for each network is interconnected to
calculate the fault current.
4. Treating the sequence components of the fault current as the negative current injection at the fault
bus in the corresponding sequence networks, the bus voltage deviations are obtained as indicated
earlier for the symmetrical fault case. Here, it should be noted that only the positive-sequence
network carries the per-fault bus voltage specifications and in all other sequence networks, the
pre-fault bus voltages are zero.
5. Once all sequence components of voltage and current are estimated they are transformed back to
the abc-frame. In the transient stability studies, only positive-sequence bus voltages are computed
as they are the only components needed to compute the electromagnetic torques in generators
[8, 9]. The effect of negative-sequence torques are generally neglected.

8.3.1 Calculation of Sequence Voltages and Currents

_ _ _ _
Consider a set of three-phase unbalanced voltage phasors given by Va, Vb, and Vc. Assuming that Va
is taken as the reference for all angle measurements, the positive-sequence voltage component is
obtained as [6]
o o
Va(1) = Va + e j120 Vb + e − j120 Vc
(8.34)

Denoting e120° as a unit vector, a, we have,

o

a = e j120 (8.35)

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 375 1/29/2018 1:57:38 PM

 376  |  Power System Analysis

o o o

a 2 = a × a = e j120 × e j120 = e − j120 (8.36)

1 + a + a 2 = 0 (8.37)

and
o o

a 3 = a 2 × a = e − j120 × e j120 = 1 (8.38)

Thus, a is treated as a unit vector which when ‘operated’ on a phasor rotates it by 120° electrical in
the counter clockwise direction. Using the above notations, the positive-sequence phase-a voltage is
computed as

Va(1) = Va + aVb + a 2Vc

(8.39)
For an a-b-c phase sequence, if it is given that

Va = V ∠0o ; Vb = V ∠ − 120o ; Vc = V ∠120o

then, from eq. (8.39), we have, Va(1) = 3(V ∠ 0°). This shows that the positive sequence per-phase
voltage is only due to Va and any deviation in Vb and Vc from the balance case results in Va(1) which is
different from 3(V ∠ 0°) .
Similarly, the negative sequence per-phase voltage can be determined as

Va( 2) = Va + a 2Vb + aVc (8.40)

The applicability of the above expression can be observed as follows:

1. Let us assume a balanced three-phase phasors of a-c-b phase sequence such that

Va = V ∠0o ; Vb = V ∠120o ; Vc = V ∠ − 120o

Using eq. (8.40), we get,

o o
Va( 2) = V ∠ 0o + e − j120 V ∠120o + e j120 V ∠ − 120o

= 3(V ∠ 0°)

This shows that Va(2) is entirely due to phase-a voltage pertaining to a-c-b phase sequence.
2. With the above set (i.e., with a-c-b phase sequence), if we use eq. (8.39) to obtain Va(1), we get,
o o
Va(1) = V ∠ 0o + e j120 V ∠120o + e − j120 V ∠ − 120o

= V ∠ 0o + V ∠ − 120o + V ∠120o

=0

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 376 1/29/2018 1:57:44 PM

 Fault Analysis of Power Systems  |  377 

The above result indicates that for a-c-b phase sequence (which can be regarded as negative-
sequence three-phase quantities), the positive-sequence voltage does not exist.
3. If we start with a-b-c phase sequence, from eq. (8.40), we obtain zero negative-sequence voltage,
Va(2).
In addition to positive- and negative-sequence voltages, a resultant of 3-phase quantities is deter-
mined to give zero-sequence voltage for phase-a as

Va( 0) = Va + Vb + Vc (8.41)

Thus, from eqs (8.39), (8.40), and (8.41), a matrix representation is obtained with a scalar factor
1
of  as,
3
Va( 0)  1 1 1  V 
a
  1 2    (8.42)
V
 a(1)  3=  1 a a  V
 b
Va( 2)   2  V 
  1 a a   c 
or

 (3×1) = [Ts ](3×3) Vabc  (3×1)

Va( 012) 


where [Ts] is given by

1 1 1 
1 
[Ts ] = 3 1 a a 2  (8.43)
 2 
1 a a 

In the above expression, premultiplying by [Ts]−1 on both sides, we get,

−1
Vabc  = [Ts ] Va( 012)  (8.44)
 

In the literature, [Ts]−1 is regarded as the transformation matrix and is given by

1 1 1 
−1  
[Ts ] = 1 a 2 a  (8.45)
 
1 a a 2 
Thus, from eq. (8.44), we have,

Va = Va( 0) + Va(1) + Va( 2)

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 377 1/29/2018 1:57:48 PM

 378  |  Power System Analysis

Vb = Va( 0) + a 2Va(1) + aVa( 2)

= Vb( 0) + Vb(1) + Vb( 2)

Vc = Va( 0) + aVa(1) + a 2Va( 2)

= Vc( 0) + Vc(1) + Vc( 2)

These sequence components are shown in Figure 8.15.

Vc(1)
Vb(2)
o
Va(2)
120
o 120
Va(1)
Va as
Ref

acb Va(0)
abc Vb(0)
Vc(0)
Vc(2)
Vb(1)

Positive−sequence Negative−sequence Zero−sequence

Figure 8.15  Sequence Components of Voltage.

Similar to eqs (8.42) and (8.44), when sequence component transformation is applied to phase
currents, we get,

 (3×1) = [Ts ](3×3)  I abc  (3×1) (8.46)

 I a( 012) 

and
−1
 I abc  = [Ts ]  I a( 012)  (8.47)
 
or

I a = I a( 0) + I a(1) + I a( 2)

I b = I b( 0) + I b(1) + I b( 2)

I c = I c( 0) + I c(1) + I c( 2)

Therefore, once sequence voltages and currents are determined in the sequence networks, on a single-­
phase basis, the phase quantities in the abc-frame are computed using the principle of superposition
as per eqs (8.44) and (8.47).

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 378 1/29/2018 1:57:52 PM

 Fault Analysis of Power Systems  |  379 

Notes:
1. The main advantage of the symmetrical component-based fault analysis is that the sequence
networks are completely decoupled from one another. This implies that the current of a given
sequence produces voltage drop of the same sequence. This is due to the fact that while design-
ing three-phase electrical equipment such as synchronous machines, transformers, transmission
lines, loads, etc., care is taken to see that the elements are symmetric for three-phase operation.
For example, transmission lines are made symmetric by completely transposing them all along
their length. This further ensures that sequence networks are balanced.
2. In the fault analysis, the sequence networks are assumed to be connected in Y and balanced. The
∆-connected elements are converted into an equivalent Y.
3. The per-phase voltages in the sequence networks are measured from line-to-ground. Since the
positive- and negative-sequence networks are balanced, line-to-neutral voltage is identical to the
line-to-ground voltage. However, in a three-phase Y-connected network, the neutral terminal may
be connected to the ground through a finite grounding-impedance. In such cases, only in the
zero-sequence network the per-phase zero-sequence voltage is given by the summation of the
line-to-neutral and neutral-to-ground voltages.
4. In a three-phase power system, the presence of neutral current (at fundamental frequency) rep-
resents the zero-sequence currents. However, for ease of operation and to facilitate detection of
abnormal conditions, the neutral terminal is grounded selectively, instead of running the neutral
wire all along the network. Therefore, the existence of ground current denotes the zero-sequence
current. If the grounding-impedance is infinite (which represents no connection from the neutral
point to the ground), then zero-sequence current does not exist in the network. This also depends
on the transformer winding connection. The zero-sequence currents considered in the fault cal-
culation are only of fundamental frequency. While carrying out harmonic analysis, the triplen
currents which remain co-phasors, are also treated as zero-sequence currents.

8.3.2  S
 equence Impedances of Y- and ∆-connected
Passive Elements
Let us consider a symmetrical Y-connected impedance system with impedance Z in each phase and
impedance Zn in the neutral as shown in Figure 8.16.

Ia
A

Va Z
Zn In = 3Ia0
N n
Z
Vc Ic
Z
C
Vb

B
Ib

Figure 8.16  Starr-connected Three-phase Passive Element.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 379 1/29/2018 1:57:53 PM

 380  |  Power System Analysis

Applying KVL in phase-a loop, we have,

Va = I a Z + I n Z n

= I a Z + ( I a + I b + I c ) Zn

= I a ( Z + Zn ) + I b Zn + I c Zn

Similarly, we can write,

Vb = I a Z n + I b ( Z + Z n ) + I c Z n
and
Vc = I a Z n + I b Z n + I c ( Z + Z n )

Writing the above phase voltages in matrix form, we get,

Va   Z + Z n Zn Zn   Ia 
    
Vb  =  Z n Z + Zn Zn   Ib 
Vc   Z n Zn Z + Z n   I c 

or
Vabc  = [ Z abc ]  I abc 

Using eqs (8.44) and (8.47) in the above expression, we get,

[Ts ]−1 Va(012)  = [ Zabc ][Ts ]−1  I a(012) 

Premultiplying the above equation by [Ts] on both sides, we get,

Va( 012)  = [Ts ][ Z abc ][Ts ]−1  I a( 012) 

   

By evaluating[Ts] [Zabc] [Ts]−1, we obtain the sequence impedances, [Z012] of the Y-connected ele-
ment. From eqs (8.43) and (8.45), [Ts] [Zabc] [Ts]−1 is evaluated as follows:

[Z012] = [Ts] [Zabc] [Ts]−1 (8.48)

1 1 1  Z + Z Zn Zn  1 1 1 
n
1   
= 1 a a 2   Zn Z + Zn Zn  1 a
2
a 
3 
1 a
2
a   Z n Zn Z + Z n  1 a 
a 2 


M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 380 1/29/2018 1:57:57 PM

 Fault Analysis of Power Systems  |  381 

1 1 1   Z + 3Z n Z Z 
1  
= 1 a a 2   Z + 3Z n a Z 2
aZ 
3  
2
1 a a   Z + 3Z n aZ a 2 Z 

 Z + 3Z n 0 0
 
= 0 Z 0  (8.49)
0 0 Z 

From the above expression, we can see that the zero-sequence impedance, Z0 = Z + 3Zn, positive-­
sequence impedance, Z1= Z, and negative-sequence impedance, Z2= Z. Thus, for the above Y-connected
impedances with neutral connection to ground through a grounding-impedance, Zn, the sequence net-
works on single-phase basis are as follows:

Va( 0) = Z0 I a( 0) Va(1) = Z1 I a(1) Va( 2) = Z2 I a( 2)

The sequence networks for phase-a are shown in Figure 8.17.

Z1 Z2 Z
a n a n a n
I a (1) I a (2) I a (0)

Va (1) Va 2 Va (0) 3Z n

Positive Negative Zero

Figure 8.17  The Sequence Networks for Phase-a for Starr-connected Three-phase Passive Element.

Notes:
1. For the above system, the zero-sequence current can be obtained as

Va( 0)
I a( 0 ) = (8.50)
( Z + 3Z n )

From eq. (8.50), it can be seen that if the neutral of the Y-connected circuit is solidly grounded, we
set Zn= 0, and the system becomes three-phase, four-wire configuration, and the zero-sequence
network is as shown in Figure 8.18(a).

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 381 1/29/2018 12:20:34 PM

 382  |  Power System Analysis

a Z a Ia (0) = 0 Z
n n
Ia (0)

Va (0) Va (0)

(a) With neutral (b) Without neutral

Figure 8.18  Zero-sequence Network for Phase-a for Starr-connected Three-phase Passive Element
With and Without Neutral.

If there is no connection between the neutral and the ground, then Zn = ∞ and as per eq. (8.50),
there cannot be any zero-sequence current flow in the system. This corresponds to three-phase,
three-wire configuration with zero-sequence network as shown in Figure 8.18(b).
2. In power system protection schemes (used for protecting transmission lines/ transformers/­rotating
machines) the occurrence of ground fault is detected by the presence of zero sequence currents.
Prior to the fault, In = Ia + Ib + Ic = 0 and once the fault occurs involving the ground, a return path
for zero-sequence current exists and In ≠ 0 as it is equal to 3Ia(0).
  Let us now consider a symmetrical ∆-connected impedance system with impedance Z∆ in each
phase as shown in Figure 8.19.

Ia a Ia
a
+

Iab I ca ZY Van
Vab
z∆ z∆ Vab
−
Ib z∆ n
b Ib
ZY ZY
b
Vbc Ibc
Vca Vca Vbc

c c
Ic Ic

∆-connected load Y-equivalent load

Figure 8.19  ∆-connected Three-phase Passive Element.

It is to be noted before analysing a ∆-connected three-phase element, it should be converted into an

Z
equivalent Y-connected three-phase element with ZY = ∆ as in Figure 8.19. Now for the Y element,
3

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 382 1/29/2018 12:20:36 PM

 Fault Analysis of Power Systems  |  383 

the sequence networks can be derived as described earlier. Hence, for the ∆-element, the positive- and
Z∆
negative-sequence networks are identical to that given in Figure 8.17, except that Z1 = Z2 = . As
3
far as the zero-sequence network is concerned, the circuit given in Figure 8.18(b) is applicable with
Z∆
Z= . This clarifies that in a ∆-connected element, zero-sequence current does not flow in the line;
3
in other words, the line currents Ia (or Ib or Ic) cannot possess zero-sequence components.

Notes:
1. Although zero-sequence current cannot flow in the line wires, the closed path within the ∆-loop
can carry zero-sequence currents in case of rotating machines or transformer windings. This is
possible if there is a residual zero-sequence voltage in the loop or in case of transformers where
the loop can provide a path for zero-sequence mmf-balance at fundamental frequency or in case
of non-sinusoidal magnetising current where a triplen current flows in the closed loop. The cor-
responding equivalent circuit is shown in Figure 8.20.

z∆
a Ia (0) = 0

Va (0)

Figure 8.20  An Equivalent Zero-sequence Circuit for ∆-connected Element.

2. The zero-sequence components of line-to-line voltages is always zero. This is because in terms of
sequence components, Vab can be written as

Vab = Va – Va
= (Va(0) + Va(1) + Va(2)) – (Vb(0) + Vb(1) + Vb(2))

Since Va(0) = Vb(0), we can see that Vab possesses only positive- and negative-sequence compo-
nents as,
Vab = (Va(1) + Va(2)) – (Vb(1) + Vb(2)) (8.51)

From the above result, we can also write that

Vbc = (Vb(1) + Vb(2)) – (Vc(1) + Vc(2)) (8.52)

Vca = (Vc(1) + Vc(2)) – (Va(1) + Va(2)) (8.53)

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 383 1/29/2018 12:20:39 PM

 384  |  Power System Analysis

From eqs (8.51), (8.52), and (8.53), we note that

Vab + Vbc + Vca = 0 (8.54)

The above result implies that the sum of line-to-line voltages is always zero.

8.3.3 Sequence Impedances of Synchronous Machines

Sequence impedances of rotating machines are different from those defined for static elements like
passive loads, transformers and transmission lines [3].

8.3.3.1 Positive- and Negative-Sequence Impedances

For synchronous machines, the positive- and negative-sequence impedances are not the same. The
d- and q-axes subtransient, transient, and synchronous reactances (such as x″d, x″q, x′d, x′q, xd, and xq)
are treated as the positive-sequence reactances, Zlg.
Although a synchronous machine is a source of only positive-sequence voltage, it carries a
negative-sequence current due to unbalanced loading conditions and during unsymmetrical faults.
This produces a rotating field in a direction opposite to that of the rotor main field. Hence, the neg-
ative-sequence reactance corresponds to a flux which rotates at twice the synchronous speed with
respect to the rotor. This reactance is mainly due to the combined effects of the damper windings and
field winding in a manner similar to the subtransient positive-sequence values. The negative-­sequence
impedance is 70%–95% of the subtransient reactance. However, in most of the cases it is taken as,

xd′′ + xq′′
Z2 g = (8.55)
2

for 2.2 model and for 1.1 model, it is assumed to be equal to

xd′ + xq′
Z2 g = (8.56)
2

8.3.3.2 Zero-Sequence Impedance
It is known that in a symmetrical synchronous machine, if the three-phase windings are energised
by a zero-sequence current, it produces a negligible resultant field in the air gap, except due to
leakage flux and end turns. Therefore, the zero-sequence impedance of a synchronous machine
is taken as a small percentage of the d-axis positive-sequence subtransient reactance. Thus, x0 =
(0.1 – 0.7) × x″d.
If the neutral point of the generator is grounded through an impedance Zng, then as per eq. (8.49),
3Zng is added to the zero-sequence impedance of generator before incorporating it in the zero-
sequence network. Therefore, the zero-sequence impedance of the machine is given by

Z0g = x0 + 3Zng (8.57)

A generator can carry zero-sequence current depending on the terminal conditions if Zng is finite.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 384 1/29/2018 12:20:40 PM

 Fault Analysis of Power Systems  |  385 

Notes:
1. For rotating machines such as synchronous generators, synchronous motors, and induction
motors, only the positive-sequence circuit of the machines possesses a voltage source behind Z1g.
2. Synchronous motors are treated similar to that of generators.
3. For induction motors, Z1g = Z2g and is set to either x′′ for double-cage rotors or x′ for single-cage
rotors. The zero-sequence impedance x0 is set to (0.1 – 0.7) × Z1g.
4. For rotating machines, the sequence reactances appear as shunts in the respective bus admittance
matrices.
5. The zero-sequence impedance, x0 is identical to that obtained using Park transformation.
6. For simplicity, the generator neutrals are assumed to be grounded through reactances. If it is
required to simulate an open neutral, set the grounding reactance to a very high value, say 106 in
a simulation programme.

8.3.4 Sequence Impedance of Transmission Lines

Consider a symmetrical three-phase transmission line connected between nodes 1 and 2 as shown in
Figure 8.21. A remotely placed ground return is assumed for zero sequence currents. In Figure 8.21,
Zs denotes the self-impedance of each line, and Zm represents the mutual impedance between lines.

Zs
Ia

Zm Zs
Ib

Ic Zm Zs Zm

V1a V2a
V1b V2b
V1c V2c

Figure 8.21  A Simplified Symmetrical Transmission Line with Remotely

Placed Ground Return.

Using KVL, we can write in matrix form the voltage drop across the line as

V(1abc )  − V( 2abc )  =  Z( abc )   I ( abc )  (8.58)

      

Using eqs (8.44) and (8.47) in the above expression and rearranging, we get,

V1a( 012)  − V2a( 012)  = [ Z012 ]  I a( 012) 

     

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 385 1/29/2018 12:20:43 PM

 386  |  Power System Analysis

where

[ Z012 ] = [Ts ][ Zabc ][Ts ]−1 (8.59)

Now substituting for [Ts] and [Ts]–1, and simplifying the expressions, we get [Z012] as,

 Z s + 2Zm 0 0 
[ Z012 ] = 0 Z s − Zm 0

 (8.60)
0 0 Z s − Z m 

Thus, we can observe that the series sequence-impedances, Z1 = Zs – Zm, Z2 = Zs – Zm, and
Z0 = Zs + 2Zm.

8.3.4.1 Positive- and Negative-Sequence Impedances

A transmission line is made symmetric by transposing the phase wires with respect to one another
all along its length [10]. Hence, for a symmetrical three-phase transmission line, the positive- and
negative-­sequence networks are uncoupled and balanced. Further, it should be noted that the positive-
and negative-sequence series impedances and line charging are identical because the transmission line
circuit remains independent of the phase-sequence of the applied voltages. For a line, the positive-­
sequence series impedance, Z1 and line charging, B1 are included in [YBUS(1)] in the usual way.
Similarly, the negative-sequence series impedance, Z2 and line charging, B2 are included in [YBUS(2)].

8.3.4.2  Zero-Sequence Impedance

The zero sequence impedance of a transmission line depends on the tower configuration, the return
paths through earth and ground wires, tower footing resistances, soil conditions, etc. The co-phasor
nature of zero-sequence currents causes a transmission line to offer an impedance, Z0 which is quite
large compared to either positive- or negative-sequence impedance. Typically, Z0 is taken as three
times that of the positive-sequence impedance, Z1. However, the zero-sequence line charging, B0 is
smaller than the positive-sequence line charging. Therefore, B0 is set to 0.7 times B1. Note that these
parameters pertain to only fundamental frequency zero-sequence currents. For harmonic frequencies,
the earth path may offer varied impedances.

8.3.4.3 Handling of Mutuals in Zero Sequence Modelling

of Transmission Lines
In case of parallel or nearby transmission lines on the same tower, the flow of zero-sequence currents
in one circuit leads to mutual effects in the neighbouring circuits. This introduces mutual coupling
among the line elements in the zero-sequence network. It produces significant effects on fault-current
levels and hence, on the operation of protective relaying systems. Since such mutual coupling effects
resulting from positive and negative sequence currents are negligible in their respective networks,
[YBUS(1)] and [YBUS(2)] can be constructed directly from the connectivity details of each element.
In the following lines, a procedure to handle mutual coupling in the zero-sequence network while
constructing [YBUS(0)] from connection details, is presented [7]:
Consider an nb-bus zero-sequence network, where there are m mutually coupled lines.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 386 1/29/2018 12:20:45 PM

 Fault Analysis of Power Systems  |  387 

1. Initialise YBUS ( 0)  with zeros.

( nb × nb )

2. For those lines, which are not mutually coupled to any other line(s), update the elements of
YBUS ( 0)  , in the usual way.
  ( nb × nb )
3. For m lines, obtain the voltage drop and branch current deviation matrix as
I b 
 ∆Vb1   Z11 Z12 . . Z1m   p1q1 
    b 
 ∆Vb2   Z21 Z22 . . Z2m   I p2 q2 

.  . . . . .  . 
 =  b 
 ∆Vbk   Z k1 Z k 2 . . Z km   I
pk qk
 (8.61)
.  . . . . .   
    . 
∆
 bm   m1 Z m2
V Z . . Z mm   b 
 I pm qm 

where
∆Vbk = is the voltage drop in the kth line.

I pb q = is the branch current through the kth line in the direction of voltage drop.
k k

pk and qk denote ‘from’ and ‘to’ bus of the kth line.

Zkk = (rkk + jxkk) -is the self impedance (zero-sequence) of the kth line.
and Zij = (jxij) is the mutual reactance between the coupled lines i and j.
Rewriting eq. (8.61), we have,
 ∆Vb   b
  ( m ×1) = [ Z m ]( m × m)  I  ( m ×1) (8.62)
If the inverse of matrix [Zm] exists, the branch currents can be computed as follows:
−1
  ( m ×1) [ m ]( m × m)  b  ( m ×1)
I b  = Z  ∆V  (8.63)

4. Obtain [PI] matrix such that the voltage drop across a line is expressed in terms of the vector
of bus voltages, [V] as

 ( m ×1) = [ PI ]( m × nb ) V  ( nb ×1) (8.64)

 ∆Vb 


where [PI] is such that for the kth line, connected between nodes pk and qk, we have P(k, pk) = 1
and P(k, qk) = −1. The remaining elements in the k th row are set to zero.

5. Therefore, from eqs (8.63) and (8.64), we can express the branch currents as a function of
bus ­voltages as follows:

−1
  ( m ×1) [ m ] [ I ]   [ m ]( m × nb )   ( nb ×1)
I b  = Z P V  = Y V  (8.65)

where [Ym] = [Zm]−1 [PI] and is an admittance matrix.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 387 1/29/2018 12:20:53 PM

 388  |  Power System Analysis

6. After constructing [Ym](m × nb), note that the kth row of this matrix represents the branch current
contribution, I pb q of the kth line due to bus voltages. This is shown in Figure 8.22(a). This cur-
k k
th th
rent can be viewed as a bus current ‘entering’ the pk bus and ‘leaving’ the qk bus. Hence, as a
bus current injection (with respect to ground), this is modelled as in Figure 8.22(b).

7. Thus, to account the kth line element (along with its mutual coupling) in [YBUS(0)], perform the
following:
th
(a) Add the kth row of [Ym] matrix to the pk row of [YBUS(0)].
th
Subtract the kth- row of [Ym] matrix from the qk -row of [YBUS(0)].
(b)
(c)
The line charging, B0 of the kth line is added to [YBUS(0)] in a usual manner.
Repeat the above process for all mlines to obtain the final [YBUS(0)].

b
pk I pq qk
k k

kth line

(a)
pk qk

b b
I Ip q
pkqk k k

(b)

Figure 8.22  Branch Current in a Zero-sequence Network Due to Mutuals.

Consider a four-bus zero-sequence network as shown in Figure 8.23 to illustrate the above algorithm.
A line between buses 2 and 4 is mutually coupled to a line between buses 2 and 3. Thus, there
are two lines which are mutually coupled and hence, m = 2. Since nb is 4, [YBUS(0)] is of dimension
(4 × 4). The steps are as follows:
1. Construct [YBUS(0)] with a line between buses 1 and 3 alone, as it has no mutual coupling. We have,

− j2 0 j2 0
0 0 0 0
YBUS( 0)  =  
   j2 0 − j2 0
 
0 0 0 0
( 4 × 4)

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 388 1/29/2018 12:20:59 PM

 Fault Analysis of Power Systems  |  389 

2 4
j 0.2
j 0.1
j 0.4

j 0.5
Zero−sequence network

Figure 8.23  Four-bus Power System for Obtaining [YBUS(0)] with Mutual Couplings.

2. Form the [Zm] matrix, according to eq. (8.62) for lines 2–4 and 2–3. Therefore,

 ∆V24   j 0.2 j 0.1   I 24

b
( k =1)

 
 ∆V  =  j 0.1 j 0.4   I b
 23    23( k = 2) 
3. Compute [Zm]−1 as

1  j 0.4 − j 0.1  − j 5.71428 j1.42857 

[ Zm ]−1 = − 0.07  − j 0.1 =
j 0.2   j1.442857 − j 2.85714 

4. Construct [PI] matrix of dimension (2 × 4) for all mutually coupled lines as

0 1 0 −1
[ PI ] = 0 
 1 −1 0 
5. Obtain [Ym] = [Zm]−1 [PI] as

2 − 4( k = 1) 0 − j 4.28571 − j1.42857 j 5.71428 

[Ym ] = 2 − 3(k = 2) 0 
 − j1.42857 j 2.85714 − j1.42857
6. Now update [YBUS(0)] to account for lines, k = 1 and k = 2 as
(a)
F
 or k = 1, add the first row of [Ym] to the second row of [YBUS(0)] and subtract the first row of
[Ym] from the fourthrow of [YBUS(0)]. Thus, the new [YBUS(0)] is:

− j2 0 j2 0 
0 − j 4.28571 − j1.42857 j 5.71428 
YBUS( 0)  =  
   j2 0 − j2 0 
 
0 j 4.28571 j1.42857 − j 5.71428

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 389 1/29/2018 12:21:04 PM

 390  |  Power System Analysis

(b)
For k = 2, add the second row of [Ym] to the second row of the updated [YBUS(0)] and subtract
the second row of [Ym] from the third row of [YBUS(0)]. Thus, the final [YBUS(0)] is given by

− j2 0 j2 0 
0 − j 5.71428 j1.42857 j 4.28571 
YBUS( 0)  =  
   j2 j1.42857 − j 4.85714 j1.42857 
 
0 j 4.28571 j1.42857 − j 5.71428
Notes:
The ordering of nodes pk and qk must be consistent with the convention of voltage drops. For the
p­ urpose of understanding the method, having separated the mutually coupled line, if the mutual
impedance is set to zero, we get,

2 − 4( k = 1) 0 − j 5 0 j 5
[Ym ] = 2 − 3(k = 2) 0 
 − j 2.5 j 2.5 0 
Now the final [YBUS(0)] is given by
− j2 0 j2 0 
0 − j 7.5 j 2.5 j5 
YBUS( 0)  =  
   j2 j 2.5 − j 4.5 0 
 
0 j5 0 − j 5
This is identical to the admittance matrix that results by direct construction from connectivity
details of elements.

8.3.5 Sequence Impedance of Transformers

In the literature [10, 11], commonly used connections for a two-winding, three-phase transformer are
as follows:
1. Y - Y and ∆ - ∆ connections: In these transformers, there is no phase-shift between primary and
secondary equivalent phase-to-neutral voltages/currents. These details are depicted in Figure 8.24.
2. ∆ - Y and Y - ∆ connections: In these transformers, there exists a phase-shift of 30° between
primary and secondary equivalent phase-to-neutral voltages/currents depending on the phase-­
sequence of the quantities (see Figure 8.25).

8.3.5.1 Positive- and Negative-Sequence Impedances

For a symmetrical two-winding, three-phase transformers, the sequence networks are uncoupled and
balanced. Since the fundamental frequency equivalent circuit is independent of the phase sequence of
the energising quantities, the positive- and negative-sequence leakage impedances of the transformers
are identical as shown in Figure 8.26. In Figure 8.26(c), a denotes the off-nominal tap setting of the
transformer and q represents the phase-shift.
Although the positive- and negative-sequence impedances are identical, the phase-voltages/cur-
rents experience a phase-shift while referring them from one side to the other side of the windings,

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 390 1/29/2018 12:21:07 PM

 Fault Analysis of Power Systems  |  391 

Y Y ∆ ∆

p s p s

cp(1) cp(1) cs(1)

cs(1)

ap(1) as(1)
bp(1)
ap(1) bs(1) as(1)
Vpn Vsn Vpn Vsn

bp(1) bs(1)
Positive−sequence (a)

bp(2)
bs(2)
bp(2) Vpn ap(2) bs(2) Vsn as(2)

ap(2) as(2)
V pn V sn
cp(2) cs(2)

cp(2) cs(2)
Negative−sequence (b)

Figure 8.24  Phase-shift Related Details in Y - Y and ∆ - ∆ Three-phase Transformers.

depending on their sequence and the transformer winding connections. These details briefed in the
following lines:

1. Y - Y and ∆ - ∆ transformers: From Figure 8.24, we can see that the transformers provide no
phase-shift for both sequences. Hence, in Figure 8.26, we have,

a^1 = a^2 = a ∠ 0o

2. ∆ - Y and Y - ∆ transformers: From Figure 8.25, it can be noted that in these transformers, the
phase-voltages/currents undergo a phase-shift depending on their sequence. To handle this, the
following convention is generally employed [6]:
(a) The high-voltage windings of a transformer are generally connected in Y and the low-voltage
side is connected in ∆.
(b) While referring to the phase-voltage/current from low-voltage side to high-voltage-side, it
is assumed that the positive-sequence quantities acquire a phase-lead of 30° and the nega-
tive-sequence voltage/current experience a phase-lag of 30°.
As per the above convention, the q is set in the following tap setting:
a^1/ 2 = a e jq

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 391 1/29/2018 12:21:11 PM

 392  |  Power System Analysis

∆ Y ∆
Y
p s p s

cp(1) cp(1) cs(1)

o cs(1)
30

as(1) ap(1)
bp(1)
ap(1) bs(1) as(1)
Vpn V sn Vpn V sn

o o
e −j30 : 1 o bs(1) bp(1) e j30 : 1 o
−j30 j30
Vpn = V sn e V pn = V sn e
Positive−sequence (a)
bp(2)
bs(2)

ap(2) bs(2) Vsn as(2)

bp(2) Vpn ap(2) as(2)
Vsn V pn

cp(2) cs(2)
cp(2) cs(2)
o o
e j30 : 1 o e −j30: 1 o
j30 −j30
Vpn = Vsn e Negative−sequence (b) Vpn = Vsn e

Figure 8.25  Phase-shifts in ∆ - Y and Y - ∆ Three-phase Transformers.

T
In stability analysis, the phase-shifts are generally neglected by considering only upto V pa in Figure
8.26(c)), and the off-nominal tap ratio, a is modelled as an equivalent π circuit [5]. However, in a
c­ onventional fault analysis programme, even this off-nominal tap setting is neglected to obtain an
estimate of fault currents. If phase quantities are to be obtained in the abc-frame, then only the phase-
shifts for sequence components are accounted.
Further, it should be noted that the positive- and negative-sequence impedances of transformers are
independent of the status of the neutral-point with respect to the ground.

8.3.5.2  Zero-Sequence Networks of Two-winding Transformers

The zero-sequence network of a transformer is decided by the availability of a path for the flow zero-se-
quence (or ground) currents through the transformer windings. This depends on the following factors [7]:

1. The type of the transformer construction, i.e., core-type or shell-type or five-limbed core or a
bank of three-single phase transformers. For a core-type construction, a path for co-phasor flux
is not available through the limbs, except through insulating medium and the transformer tank.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 392 1/29/2018 12:21:15 PM

 Fault Analysis of Power Systems  |  393 

Z1 = Zeq
Vpa(1)
p s
Vsa(1)
Isa(1)
^
a1 1
Positive-sequence (a)

Vpa(2) Z2 = Zeq
p s
Vsa(2)
Isa(2)
^
a2 1
Negative-sequence (b)

Vpa T Zeq
Vpa s
p Vsa
Isa

e jθ 1 a 1
(c)

Figure 8.26  The Single-phase Equivalent Circuit of a Three-phase Transformer

for Positive- and Negative-sequences.

In such cases, the zero-sequence leakage impedance, Z0L may be taken as 85%–90% of the
p­ ositive-sequence impedance, Zeq.
2. The connection employed for the transformer windings, i.e., Y – Y, ∆ - ∆, ∆ - Y and Y - ∆.
3. The nature of the connection between the neutral-point and the earth.
4. The availability of path for current flow in the coupled winding, when the other side carries
­current so as to provide mmf-balance.

The zero-sequence networks for a Y - Y transformer are shown in Figure 8.27 for different conditions
of the neutral-point connection to the ground.
From Figure 8.27, it is clear that the zero-sequence currents flow through the windings from one
side of the transformer into the line wires on the other side only when both the neutral points are
grounded. If the neutral point on any one side is ungrounded, the zero-sequence currents cannot flow
in the windings on either side as the zero-sequence impedance viewed from either the primary or
secondary side is infinite.
The zero-sequence networks for a ∆-Y or a Y-∆ transformer are shown in Figure 8.28 for different
conditions of the neutral-point connection to the ground.
From Figure 8.28, it can be noted that the zero-sequence currents do not flow through the windings
from one side of the transformer into the line wires on the other side. This feature of the transformer
is used to isolate the ground-related faults in a section of the network from the other.
The above types of zero-sequence networks are summarised in Table 8.3. Here, the grounding
impedance Zn can take any value between zero and infinity. If Zn is zero, the transformer winding

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 393 1/29/2018 12:21:15 PM

 394  |  Power System Analysis

P S

Neutral connection to ground

Zero−sequence network
P S Y Y
P P0 S

Z np Z ns
Z0 = Z0L + 3ZnP + 3Zns
Zero−sequence current path

P S
Zero−sequence network

P S
P S
Z np

P S

Z ns

Figure 8.27  The Zero-sequence Networks for a Y - Y Transformer.

is solidly grounded and if it is infinity, the transformer is ungrounded. The resulting zero sequence
impedances are used in Figure 8.29 to construct the zero-sequence network.
Notes:
1. It is common to see that the transformer configurations given in Table 8.3 are used for the follow-
ing applications:
(a) Generator transformers: Configurations type-1 and type-2 are used for interfacing the gener-
ators to the network. It is assumed that the generator is always connected to the ∆-side of the
transformer.
(b) Interconnecting transformers: Configuration type-3 is used as inter-connecting transformers
in the network.
2. In a simulation programme, the infinite impedance may be approximated as 106 pu. For simplic-
ity, the zero-sequence impedance and grounding-impedance are assumed to be of pure reactive.

8.3.6 Example: Sequence Networks for a 2-machine 5-bus

Power Systems
For a lossless power system shown in Figure 8.1, the components details are given below:
Generators G1 and G4: 865 MVA, 22 kV, 60 Hz, H = 2.8941 MJ/MVA.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 394 1/29/2018 12:21:16 PM

 Fault Analysis of Power Systems  |  395 

Transformer connections Zero-sequence networks

P S
P S

Zns Z0
Z0 = Z0L + 3Zns
Zero−sequence current paths

P S
P S

Znp
Z0 = Z0L + 3Znp

Zero-sequence current paths

P S

P S
P S

Figure 8.28  The Zero-sequence Networks for a ∆-Y or a Y-∆ Transformer.

P Z0PS S

Z0P Z0S

Figure 8.29  The π Circuit Model of Zero-sequence Networks for Transformers.

Table 8.3  A Summary of Zero-sequence Model Settings for Transformers.

Type no. Transformer type Z0P Z0PS Z0S

1 ∆-Y ∞ ∞ Z0L + 3Zns

2 Y-∆ Z0L + 3Znp ∞ ∞

3 Y-Y ∞ Z0L + 3Znp + 3Zns ∞

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 395 1/29/2018 12:21:18 PM

 396  |  Power System Analysis

x″d= 0.135, x″q= 0.135, x0 = 0.0675, and Zng= 0

Transformers T1 and T2: 22/500 kV, 60 Hz with tap setting = 1.0 pu.
Zeq= j0.1, Z0L= j0.085, and Zns= Znp = 0.
Transmission lines: 500 kV, 60 Hz.
Line between buses 2 and 3: xL = 0.5 pu and Z0= j1.5 pu.
Line between buses 3 and 5: xL= 0.1 pu and Z0= j0.3 pu,
Note that all parameters have been given on 865 MVA base. Using these parameters, the sequence
networks are drawn on the single-phase basis as shown below:
The positive-sequence network is given in Figure 8.2. Here, the pre-fault load currents are neglected.
The negative-sequence network is as shown in Figure 8.30. This circuit is identical to the positive-­
sequence network (except for the positive-sequence voltage sources) since

xd′′ + xq′′
Z2 g = j = Z1g ( = jxd′′ )
2
for the generators. This also implies that [YBUS(2)] is identical to [YBUS(1)].

j0.1 j0.5 j0.1 4

1 2 3

j0.1 j0.135
j0.135
5

Figure 8.30  The Negative-sequence Network for Two-machine, Five-bus Power Systems.

Following the equivalent circuits shown in Figure 8.28, the zero-sequence network for the power sys-
tem is drawn as depicted in Figure 8.31.

j1.5
1 2 3 4

j0.3
j0.0675 j0.085 j0.0675
j0.085
5

Figure 8.31  The Zero-sequence Network for Two-machine, Five-bus Power Systems.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 396 1/29/2018 12:21:20 PM

 Fault Analysis of Power Systems  |  397 

For the network, the [YBUS(0)] can be constructed directly from the connectivity details as follows:

 − j14.81482 0 0 0 0 
0 − j12.43137 j 0.66667 0 0 
 
YBUS( 0)  = 0 j 0.66667 − j15.76471 0 j 3.33333 
   
0 0 0 − j14.81482 0 
0 0 j 3.33333 0 − j 3.33333
( 5 × 5)

Hence, the [ZBUS(0)] is obtained by executing the following command in MATLAB as

Z_BUS0=Y_BUS0\eye(5)
and is given by

 j 0.06750 0 0 0 0 
0 j 0.080674 j 0.00433 0 j 0.00433 
 
 Z BUS( 0)  = 0 j 0.00433 j 0.080674 0 j 0.080674 
   
 0 0 0 j 0.06750 0 
0 j 0.00433 j 0.080674 0 j 0.380674 
( 5 × 5)

In the following sections, various types of shunt faults are discussed [6, 3].

8.3.7 Analysis of Unsymmetrical Short-circuit Faults

Shunt faults such as single line-to-ground (LG), line-to-line (LL), and double line-to-ground (LLG)
are discussed in the following section.

8.3.7.1 Analysis of Single Line-to-ground Fault

For a single line-to-ground fault at bus f on phase-a, through a fault impedance Zf (see Figure 8.32(a)),
we can state the following conditions in the abc-frame with respect to the fault point:
Ifa(1) f
+
a .f VPRf
ZThf(1)
Vfa(1)
Ifa −
Zf Ifa(2)
b .f +
3Zf
Ifb ZThf(2) Vfa(2)

Ifa(0)
c .f +
Ifc ZThf(0) Vfa(0)

Reference
(a) (b)

Figure 8.32  Single Line-to-ground Fault: (a) Fault Schematic (b) Sequence Networks Connection.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 397 1/29/2018 12:21:22 PM

 398  |  Power System Analysis

1. The fault currents are given by

I fb = I fc = 0 (8.66)

2. The phase-a voltage with respect to ground is given by,

V fa = V fa( 0) + V fa(1) + V fa( 2) = I fa Z f (8.67)

T
Setting  I ( fabc )  =  I fa 0 0  and using eq. (8.46), the symmetrical components of the currents are
obtained as
I fa
I fa( 0) = I fa(1) = I fa( 2) = (8.68)
3
Therefore, the fault current can be determined as,
I fa = I fa( 0) + I fa(1) + I fa( 2) = 3 I fa(1) (8.69)

Writing the voltage equations for the Thevenin equivalent circuit of the sequence networks with
respect to the fault bus, we have,

V fa(1) = VPRf − I fa(1) ZThf (1) (8.70)

V fa( 2) = − I fa( 2) ZThf ( 2) (8.71)

V fa( 0) = − I fa( 0) ZThf ( 0) (8.72)

Substituting the above sequence components of voltage V fa in eq. (8.67) and applying the conditions
given in eqs (8.68) and (8.69), we get,
V fa = − I fa(1) ZThf ( 0) + VPRf − I fa(1) ZThf (1) − I fa(1) ZThf ( 2) = 3 I fa(1) Z f (8.73)

where VPRf is the pre-fault (positive-sequence) voltage at fault bus f pertaining to phase-a and in all
angle measurements, the slack-bus voltage is taken as the reference.
Solving for I fa(1) , we obtain,

VPRf
I fa(1) = (8.74)
Z LG − eq

Z LG − eq = ZThf ( 0) + ZThf (1) + ZThf ( 2) + 3Z f

As per the conditions given in eqs (8.67) and (8.68), we can conclude that to simulate an LG-fault, the
Thevenin equivalents of the sequence networks - 012, at the fault bus f should be connected in series
along with Zf as shown in Figure 8.32(b). For the bolted fault, set Zf= 0. Also note that the condition
given in eq. (8.67) simply denotes a mathematical summation of sequence components.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 398 1/29/2018 12:21:36 PM

 Fault Analysis of Power Systems  |  399 

8.3.7.2 Example: Single Line-to-ground Fault at Bus 2

For an LG fault at bus 2, the sequence networks are connected as shown in Figure 8.33 to simulate
the fault.
From the bus-admittance matrices for the sequence networks, the second column elements of the
corresponding bus impedance matrices are exclusively obtained as

Z_i2=Ybus\e2
where e2 = [0,1,0,0,0]T.

j0.1 j0.5 j0.1 4

1 2 3

j0.1 j0.135
j0.135
o + 5 + o
E1 =1.0 0 E4 =1.0 0

j0.1 j0.5 j0.1 4

1 2 3 3 Zf

j0.1 j0.135 2
j0.135 I fa (1)
5
+
j0.178067
VPR2 V 2a (1)
−
=1.0 0o
I fa (2)
j1.5 +
1 2 3 4 j0.178067 V 2a (2) 3Z f

j0.3 j0.0675 I fa (0)

j0.0675 j0.085
j0.085
5 +
j0.080674 V 2a(0)

Reference
(a) Reference
(b)

Figure 8.33  Sequence Networks Connection for an LG Fault at Bus 2.

The column elements of the positive-, negative- and zero-sequence bus impedance matrices are
given by
 j 0.10229 0 
 j 0.17807  j 0.080674 
   
 Z j 2(1)  =  Z j 2( 2)  =  j 0.05693 and  Z j 2( 0)  =  j 0.00433  (8.75)
         
 j 0.03271  0 
 j 0.05693  j 0.00433 
( 5×1) ( 5×1)

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 399 1/29/2018 12:21:38 PM

 400  |  Power System Analysis

Note that ZTh2(0) = j0.080674, which can also be verified as

j 0.085 × j1.585
ZTh2( 0) =
j (0.085 + 1.585)

from Figure 8.31, and ZTh2(1) = ZTh2(2) = j0.17807.

Assuming Zf = 0 and using eq. (8.74), I fa(1) can be computed as

1.0 ∠0o
I fa(1) = = − j 2.289335 pu (8.76)
( j 0.080674 + j 0.17807 + j 0.17807)
From eq. (8.68) we can see that
I fa( 0) = I fa( 2) = I fa(1) = − j 2.289335 pu
Hence, the fault current can be obtained as

I fa = 3 × I fa(1) = 3 × I fa( 0) = − j 6.86801 pu

and it denotes the circulation of the ground currents.
Bus Voltage Computations:
To compute element currents during the fault, perform the following steps:
1. Obtain the sequence components of the bus voltage deviations due to the fault. A negative of the
sequence components of the fault currents are injected into their corresponding sequence net-
works at the fault bus to get,

 ∆Va(1)   
  ( nb ×1) =  Z jf (1)  ( nb ×1) ( − I fa(1) )

 ∆Va( 2)   
  ( nb ×1) =  Z jf ( 2)  ( nb ×1) ( − I fa( 2) )

 ∆Va( 0)   
 ( nb ×1) =  Z jf ( 0)  ( nb ×1) ( − I fa( 0) )
(8.77)


where  Z jf ( 012)  denotes the fth -column of the corresponding bus impedance matrices
  ( nb ×1)
of the sequence networks.
2. The sequence components of the bus voltages during the fault are obtained as

Va(1)   
  ( nb ×1) = VPR  ( nb ×1) +  ∆Va(1) 

 ( nb ×1) = [ 0 ]( nb ×1) +  ∆Va( 2) 

Va( 2)   


 ( nb ×1) = [ 0 ]( nb ×1) +  ∆Va( 0) 

Va( 0)    (8.78)


M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 400 1/29/2018 12:21:50 PM

 Fault Analysis of Power Systems  |  401 

Notes:
1. The VPR  denotes the vector of pre-fault bus voltages in the positive-sequence network. If
( nb ×1)
the pre-fault loading condition is to be considered, then these bus voltages are determined from
a power flow analysis. If pre-fault load currents are to be neglected, then set all elements in the
vector VPR  to 1.0 ∠ 0°. In per-unit, this also indicates the per-phase voltage with respect
( nb ×1)
to ground.
2. The vector [ 0 ]( n ×1) represents a zero-vector. This denotes the absence of negative- and zero-­
b

sequence bus voltages in the pre-fault condition.

3. The procedure indicated above is similar to that described for symmetrical three-phase faults, where
eq. (8.78) is used only for positive-sequence voltages and are the phase-a voltages in per-unit.
4. In the above calculations, the phase-shifts introduced by ∆ - Y or Y- ∆ transformers are neglected.
For example, we obtain the sequence components of the bus voltages as

0.76581   −0.23419  0 
0.59234   −0.40766   −0.18469 
     
Va(1)  = 0.86966  Va( 2)  =  −0.13034  and Va( 0)  =  −0.00990 
           

0.92512   −0.07488  0 
0.86966   −0.13034   −0.00990 

5. The phase voltages at bus 2 are now computed as

V2a = V2a( 0) + V2a(1) + V2a( 2)

= −0.18469 + 0.59234 − 0.40766 = 0 As Z f is set to zero.

V2b = V2a( 0) + a 2V2a(1) + aV2a( 2)

= 0.90926∠ − 107.74o

V2c = V2a( 0) + aV2a(1) + a 2V2a( 2)

= 0.90926∠107.74o

6. The phase voltages at the terminals of generator 1, are obtained by accounting the phase shifts
offered by the ∆ - Y transformer, T1 as,
p p
−j j
V1a = V1a( 0) + V1a(1) e 6 + V1a( 2) e 6

= 0.67969∠ − 47.36o

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 401 1/29/2018 12:22:01 PM

 402  |  Power System Analysis

p p
−j j
V1b = V1a( 0) + a 2V1a(1) e 6 + aV1a( 2) e 6

= 0.67969∠ − 132.64o
p p
−j j
V1c = V1a( 0) + aV1a(1) e 6 + a 2V1a( 2) e 6

= 1.0 ∠ 90o
Since the fault is on the HV-side of T1, i.e., at bus 2, appropriate 30° phase shifts are provided,
depending on the sequence component as per Figure 8.25.
7. The sequence components of the generator 1 currents are first computed by neglecting the phase
shifts offered by the ∆ - Y transformer, T1:

E − V1a(1)
I g1a(1) =
jxd′′
−V1a( 2)
I g1a( 2) =
Z2 g

−V1a( 0)
I g1a( 0) =
Z0 g

where E represents the voltage behind the subtransient reactance x″d in a simplified dynamic
model. This is obtained by using the complex power supplied by the generator. If the pre-fault
load currents are to be neglected, then set E = 1.0 ∠ 0°.
For example, the sequence currents of generator 1 are determined as

1.0 ∠ 0o − V1a(1)
I g1a(1) = = − j1.73470
j 0.135
−V1a( 2)
I g1a( 2) = = − j1.73470
j 0.135
I g1a( 0) = 0

If pre-fault currents are not considered, from Figure 8.2, I g1a(1) can also be obtained by current
division principle as,
j 0.735
I g1a(1) = ( − j 2.289335) = − j1.73470
j (0.235 + 0.735)

8. To obtain the phase currents the phase-shifts offered by the - Y transformer, T1, are now accounted as
p p
−j j
I g1a = I g1a( 0) + I g1a(1) e 6 + I g1a( 2) e 6

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 402 1/29/2018 12:22:10 PM

 Fault Analysis of Power Systems  |  403 

= 3.0046∠ − 90o
p p
−j j
2
I g1b = I g1a( 0) + a I g1a(1) e 6 + aI g1a( 2) e 6

= 3.0046∠ 90o
p p
−j j
2
I g1c = I g1a( 0) + aI g1a(1) e 6 + a I g1a( 2) e 6

=0

The above calculations clearly show that due to a ∆ - Y transformer, the zero-sequence component
of the generator current, I g1a( 0) = 0. Further, I g1a = − I g1b .

8.3.7.3 Analysis of Line-to-line Fault

Consider a line-to-line fault at bus f between phases b and c through a fault impedance Zf as shown
in Figure 8.34(a).

a
I fa
.f I fa (1)
f f
I fa (2)

+ +
b .
f
+
Z Thf (1) Zf Z Thf (2)

V PRf V fa (1) V fa (2)

I fb −
Zf
Reference
c
.
f

I fc

(a) (b)

Figure 8.34  Line-to-Line fault: (a) Fault Schematic (b) Sequence Networks Connection.

The following relations are satisfied at the fault point in the abc-frame:
1. The fault currents are given by
I fa = 0 and I fb = − I fc (8.79)

2. The phase voltages are related as,

V fb − V fc = Z f I fb (8.80)

where, the phase voltages are measured with respect to ground.

T
Setting  I ( fabc )  = 0 I fb − I fb  and using eq. (8.46), the symmetrical components of the cur-
rents are obtained as
Ifa(0) = 0
1
I fa(1) = ( a − a 2 ) I fb
3

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 403 1/29/2018 12:22:15 PM

 404  |  Power System Analysis

1
I fa( 2) = ( a 2 − a ) I fb
3
This implies that,
Ifa(0) = 0
Ifa(1) = – Ifa(2) (8.81)

Now, rewriting eq. (8.80) in terms of sequence components, we get,

(V fa(0) + a 2V fa(1) + aV fa(2) ) − (V fa(0) + aV fa(1) + a 2V fa(2) ) = Z f ( I fa(0) + a 2 I fa(1) + aI fa(2) )

After simplifying and using eq. (8.81) we get,

(a 2 − a )V fa(1) − (a 2 − a )V fa(2) = Z f (a 2 − a ) I fa(1)

Thus, we have
V fa(1) − V fa( 2) = Z f I fa(1) (8.82)

The sequence networks connection satisfying eqs (8.81) and (8.82) is shown in Figure 8.34(b).
Since the fault does not involve ground, the zero-sequence network is absent. To obtain the posi-
tive-sequence component of the fault current, use eqs (8.70) and (8.71) in eq. (8.82) to get,

VPRf − I fa(1) ZThf (1) + I fa( 2) ZThf ( 2) = Z f I fa(1)

Using the result given in eq. (8.81) and simplifying the above expression, we get,
VPRf
I fa(1) = (8.83)
Z LL − eq
where,
ZLL – eq = ZThf (1) + ZThf (2) + Zf

The above expression for Ifa(1) can also be verified from Figure 8.34(b).

8.3.7.4 Example: Line-to-Line Fault at Bus 2

For an LL fault at bus 2, the sequence networks are connected as shown in Figure 8.35 to simulate
the fault.
From eq. (8.83), the positive-sequence component of the fault current is obtained as,

1.0 ∠0o
I fa(1) = = − j 2.80793 pu (8.84)
( j 0.17807 + j 0.17807)
Hence,
I fa( 2) = − I fa(1) = j 2.80793 pu

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 404 1/29/2018 12:22:19 PM

 Fault Analysis of Power Systems  |  405 

Zf

j0.1 j0.5 j0.1 4

1 2 3
j0.1 j0.5 j0.1 4
1 2 3 j0.1
j0.135 j0.135
5
j0.135 j0.1 j0.135
o + 5 + o
E1 =1.0 0 E4 =1.0 0

Reference

Figure 8.35  Sequence Networks Connection for an LL Fault at Bus 2.

and Ifa(0) = 0. From these, the following calculations are done:

1. The fault currents can be computed as

I fb = I fa( 0) + a 2 I fa(1) + aI fa( 2)

= −4.8635 pu

I fc = I fa( 0) + aI fa(1) + a 2 I fa( 2)

= 4.8635 pu

2. Using Ifa(1), Ifa(2) and Ifa(0) in eqs (8.77) and (8.78), the phase voltages at bus 2 are computed as

V2a = 1.0, V2b = −0.5, V2c = −0.5 pu

3. The phase voltages at the terminals of generator-1 are,

V1a = 0.89178∠ − 13.80o , V1b = 0.89178∠ − 166.2o , V1c = 0.42553∠90o

4. The sequence components of the generator-1 currents neglecting the phase-shifts offered by the
∆-Y transformer, T1, are

I g1a(1) = − j 2.12766, I g1a( 2) = j 2.12766, I g1a( 0) = 0 (8.85)

5. Now, the phase currents are obtained as,

I g1a = 2.12766∠180o , I g1b = 2.12766∠180o , I g1c = 4.2553

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 405 1/29/2018 12:22:24 PM

 406  |  Power System Analysis

Note that I g1a + I g1b = − I g1c . Also, note that if Y- Y connection is employed for T1, instead of
∆ - Y, then from eq. (8.85), we have I g1a =0, I g1b = −3.6852 and I g1c =3.6852.

8.3.7.5 Analysis of Double Line-to-ground Fault

Consider a double line-to-ground fault at bus f between phases b and c through a fault impedance Zf
as shown in Figure 8.36(a).

a
I fa
.
f
f

.
f I fa (1) I fa (0)
b I fa(2)
Z Thf (0) V fa (0)
Ifb Z Thf (1)
V fa (1) V fa (2)
c . f +
I fc .( I VPRf
−
Z Thf (2)
3Z f
fb + I fc )

Zf Reference

(a) (b)

Figure 8.36  Double Line-to-ground Fault: (a) Fault Schematic (b) Sequence Network Connection.

The following relations are satisfied at the fault point in the abc-frame:
1. The phase-a fault current is given by
Ifa = 0 (8.86)

2. The phases b and c voltages are related as,

Vfb = Vfc (8.87)

and
Vfb = Vfc = (Ifb + Ifc) Zf (8.88)

where the phase voltages are measured with respect to ground.

In terms of the symmetrical components (8.86) is written as

Ifa(0) + Ifa(1) + Ifa(2) = 0 (8.89)

Similarly, rewriting eq. (8.87) in terms of sequence components, we get,

(V fa(0) + a 2V fa(1) + aV fa(2) ) = (V fa(0) + aV fa(1) + a 2V fa(2) )

or
(a 2 − a )V fa(1) = (a 2 − a )V fa(2)

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 406 1/29/2018 12:22:27 PM

 Fault Analysis of Power Systems  |  407 

Vfa(1) = Vfa(2) (8.90)

From eq. (8.88), we have,

(V fa(0) + a 2V fa(1) + aV fa(2) ) = Z f [( I fa(0) + a 2 I fa(1) + aI fa(2) ) +

+ ( I fa( 0) + aI fa(1) + a 2 I fa( 2) )]

Using eq. (8.90) in the above expression and noting that a + a 2 = −1 , we get,

(V fa(0) − V fa(1) ) = Z f 2 I fa(0) + (a 2 + a ) I fa(1) + (a 2 + a ) I fa(2) 

= Z f 2 I fa( 0) − I fa(1) − I fa( 2) 

Now, using the result given in eq. (8.89), we obtain,

(V fa(0) − V fa(1) ) = 3Z f I fa(0)

or
V fa(1) = V fa( 0) − 3Z f I fa( 0) (8.91)

The sequence networks connection satisfying eqs (8.89), (8.90), and (8.91) is shown in
Figure 8.36(b). In order to determine an expression for Ifa(1), the following steps are employed:
Using eqs (8.70) and (8.72) in eq (8.91) and simplifying, we obtain,

(
I fa(1) ZThf (1) = VPRf + I fa( 0) ZThf ( 0) + 3Z f )
From (8.89), we can rewrite the above expression as

( ) (
I fa(1) ZThf (1) + ZThf ( 0) + 3Z f = VPRf − I fa( 2) ZThf ( 0) + 3Z f (8.92) )
Using eqs (8.70) and (8.71) in eq. (8.90), we get,

VPRf − I fa(1) ZThf (1)

I fa( 2) = − (8.93)
ZThf ( 2)

Now, substituting eq. (8.93) in eq. (8.92) and simplifying for Ifa(1), we obtain,

VPRf
I fa(1) = (8.94)
Z LLG − eq
where,

Z LLG − eq = ZThf (1) +

(
ZThf ( 2) ZThf ( 0) + 3Z f )
ZThf ( 2) + ZThf ( 0) + 3Z f

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 407 1/29/2018 12:22:33 PM

 408  |  Power System Analysis

The above expression for Ifa(1) can also be verified from Figure 8.36(b). From the circuit, Ifa(2)
and Ifa(0) can be determined as,

I fa( 2) = − I fa(1)
( ZThf (0) + 3Z f ) (8.95)
ZThf ( 2) + ZThf ( 0) + 3Z f
and
ZThf ( 2)
I fa( 0) = − I fa(1) (8.96)
ZThf ( 2) + ZThf ( 0) + 3Z f

Also, note that the fault current, (Ifb + Ifc) is given by

( ) ( )
( I fb + I fc ) =  I fa( 0) + a 2 I fa(1) + aI fa( 2) + I fa( 0) + aI fa(1) + a 2 I fa( 2))  (8.97)
 
= 2 I fa( 0) − I fa(1) − I fa( 2)  (8.98)

= 3 I fa( 0) (8.99)

As in the LG case, here also, the fault current represents the flow of the ground currents.

8.3.7.6 Example: Double Line-to-ground Fault at Bus 2

For an LLG fault at bus 2, the sequence networks are connected as shown in Figure 8.37 to simulate
the fault.

j0.1 j0.5 j0.1 4

1 2 3

j0.135 j0.1
j0.1 j0.5 j0.1 4
1 2 3 5 j0.135

j0.135 j0.1 j0.135

o + 5 + o 3 Zf
E1 =1.0 0 E4 =1.0 0

Reference
j1.5
1 2 3 4

j0.0675 j0.3 j0.085

j0.085
5
j0.0675

Figure 8.37  Sequence Networks Connection for an LLG Fault at Bus 2.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 408 1/29/2018 12:22:36 PM

 Fault Analysis of Power Systems  |  409 

From eq. (8.94), the positive-sequence component of the fault current is obtained as,

1.0 ∠0o
I fa(1) = = − j 4.281055 pu (8.100)
( j 0.178067 + j 0.055520)
Hence,
I fa( 2) = j1.334809 and I fa( 0) = j 2.946246 pu

From these, the following calculations are done:

1. The fault currents can be computed as

I fb + I fc = 3 I fa( 0) = j 8.83875 pu

2. Using Ifa(1), Ifa(2) and Ifa(0) in eqs. (8.77) and (8.78), the phase voltages at bus 2 are computed as

V2a = 0.71305, V2b = 0, V2c = 0 pu

3. The phase voltages at the terminals of generator-1 are,

V1a = 0.64134 ∠ − 19.38o , V1b = 0.64134 ∠ − 160.62o , V1c = 0.42553∠90o

4. The sequence components of the generator-1 currents neglecting the phase-shifts offered by
the ∆ - Y transformer, T1, are

I g1a(1) = − j 3.24389, I g1a( 2) = j1.01143, I g1a( 0) = 0 (8.101)

5. Now, the phase currents are obtained as,

I g1a = 2.8749∠ − 137.74o , I g1b = 2.8749∠137.74o , I g1c = 4.2553 pu

Note that I g1a + I g1b = − I g1c .

6. If Ia(0) is required to be determined for a line element which has a mutual coupling then, [Va(0)]
is used in association with [Ym](m × nb). In other words, Ia(0) is obtained by calculating the corre-
b
sponding I pk qk as per eq. (8.65).

8.3.7.7 Summary of Shunt-fault Currents

Table 8.4 presents a comparison between the magnitude of fault currents with different types of faults.
From the tabulated results, it is clear that the magnitude of the fault current for LG fault is more
than that for the symmetrical three-phase (LLL) fault. If the transformer, T1-neutral alone is grounded
through a reactance of j0.05 pu, the fault current for LG fault at bus 2 reduces to –j5.34959 pu.
However, the voltage magnitude of the healthy phases increases from 0.90926 pu (with solidly
grounded neutral) to 1.0245 pu. If neutrals of both transformers T1 and T2 are ungrounded, from

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 409 1/29/2018 12:22:40 PM

 410  |  Power System Analysis

Table 8.4  Summary of Fault Currents With Different Types of Faults.

Type of fault Magnitude of fault current (pu)

LG 6.86801
LL 4.8635
LLG 8.83875
LLL 5.6159

the zero-sequence network, it can be seen that ZTh2(0) becomes infinity and leads to a negligible
LG-fault current. However, the healthy phase voltages attain line-to-line value with respect to ground
and causes increased voltage stress on the insulation of the high-voltage windings.
For a given type of shunt fault, it is sufficient to determine the positive-sequence component of the
fault current, Ifa(1), from the following expression:
VPRf
I fa(1) = (8.102)
Z shf − eq
where Zshf − eq is the equivalent impedance of the sequence networks for shunt faults. Its expression
for different types of faults with and without fault impedance Zf is tabulated in Table 8.5.

Table 8.5  Equivalent Impedance of Sequence Network for Different Short-circuit Faults.

Equivalent Impedance of sequence network, Zshf − eq

Fault Type Without fault impedance With fault impedance Zf

LG ZThf(0) + ZThf(1) + ZThf(2) ZThf(0) + ZThf(1) + ZThf(2) + 3Zf

LL ZThf(1) + ZThf(2) ZThf(1) + ZThf(2) + Zf

LLG ZThf (1) +

ZThf ( 2) ZThf ( 0)
ZThf (1) +
(
ZThf ( 2) ZThf ( 0) + 3Z f )
ZThf ( 2) + ZThf ( 0) ZThf ( 2) + ZThf ( 0) + 3Z f

LLL ZThf(1) ZThf(1) + Zf

From this, all other quantities can be computed depending on the type of fault. This way of handling
fault analysis is helpful while performing stability studies for unsymmetrical faults.

8.3.8  Transient Stability Analysis for Unsymmetrical Shunt Faults

In order to illustrate the transient stability analysis of power systems for unsymmetrical shunt faults, a
simplified dynamic model of synchronous machines is employed where the generator is modelled in
the synchronous-frame of reference. Here, a voltage behind a subtransient reactance-model is devel-
oped based on the complex power supplied by the generator.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 410 1/29/2018 12:22:42 PM

 Fault Analysis of Power Systems  |  411 

Since a generator is treated as a source of positive-sequence voltage, to compute the electromagnetic

torque generated, only positive-sequence bus voltages, [Va(1)], are estimated during an unsymmetrical fault.
The stability analysis is carried out employing the following steps:
1. From the system details, the sequence-networks are obtained as explained earlier. This facilitates
easy construction of the bus admittance matrices such as [YBUS(1)], [YBUS(2)], and [YBUS(0)]. Here,
the line charging, bus shunts, and off-nominal tap ratios of transformers are also accounted.
2. The complex loads are modelled as constant admittance loads and are augmented to the positive-
and negative-sequence bus admittance matrices.
3. Using these matrices, the Thevenin impedances, ZThf (1), ZThf (2) , and ZThf (0) are obtained, at the
fault bus f.
4. A power flow analysis is performed in the usual way to obtain the pre-fault bus voltages, [VPR],
considering the bus, load and network specifications.
5. Using the complex powers delivered by the generators the internal voltages behind the sub-
transient reactance of the machines, are computed. For the ith generator, the internal voltage is
denoted as Ei = E0i ∠ d0i and the complex power is given by Pg0i + jQg0i. It should be noted that
in these simplified models for the generator, it is assumed that E0i remains constant and only the
angle di, changes as per the swing-equation given by
dd i
= Smi w B
dt

dSmi
dt
=
1
2 Hi
( )
Tm0i − Tegi (8.103)

where Tm0i = Pg0i.

The initial values of di and Smi are d0i and zero. The related quantities are computed as follows:
(a)
For a lossless generator, the electromagnetic torque developed is given by

Tegi = Real  Ei I Gi
* 
 

where Ei = E0i ∠ di and di is obtained by solving (8.103).

(b)
The generator current, IGi, is obtained as
c
Ei − VPRi (8.104)
I Gi =
′′
jxdi
c
where VPRi is the terminal voltage of the machine, which is the ith element in the bus voltage

vector VPR
c 
.
 
The vector VPR
(c) c  , which represents the bus voltage of the system, is determined by solving
 
the following linear system of equations:

YBUS (1)   c 
  ( nb × nb ) VPR  ( nb ×1) =  I GS  ( nb ×1) (8.105)

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 411 1/29/2018 12:22:45 PM

 412  |  Power System Analysis

where [IGS] denotes the vector of bus current injections due to the internal voltages of the
machines. The ith element in this vector can be obtained as

Ei
I GSi =
′′
jxdi

In the pre-fault system, the bus voltage VPR

c 
remains at the pre-fault value, [VPR] and
 
hence, the rotor angle di stays at d0i.
6. Once a fault occurs, the bus voltage changes from the pre-fault value. To estimate the voltage
during the fault, perform the following steps:
_
(a) From the computed VPR c 
(see eq. (8.105)), the fth element V cPRf, is picked. This gives
  ( nb ×1)
 

the pre-fault positive-sequence voltage which drives the fault current, Ifa(1), through the fault.
This is calculated using eq. (8.102), where, the expression for Zshf − eq is chosen from Table 8.5,
depending on the type of fault.
(b) N
 ow, from eq. (8.77), the bus voltage deviation pertaining to the positive-sequence compo-
nent,  ∆Va(1)  is computed. This value is used in eq. (8.78) to determine the bus volt-
  ( nb ×1)
ages, [Va(1)], during the fault interval.
(c)
D
 etermine the current IGi, for the ith generator during the fault as

Ei − Via(1)
I Gi = (8.106)
′′
jxdi

where, Via(1) denotes the ith element in the vector Va(1)  .

  ( nb ×1)

(d)
Using this current (instead of that determined using eq. (8.104)), Tegi, is recomputed and used
in the swing equation to solve for the new value of di. From this, Ei is re-constructed. Noting
the variation of the rotor-angle with respect to time, for all generators, provides the swing
curves.
7. If fault is to be removed, use the generator current which is obtained from expression (8.104)
instead of that determined using eq. (8.106).
8. If fault removal involves line tripping, then in eq. (8.105), a new YBUS (1)  is used for the
  ( nb × nb )
c
post-fault system. Further, in eq. (8.104), the corresponding VPRi is employed.
The above procedure is also depicted in Figure 8.38.

For the power system shown in Figure 8.1, the rotor-angle stability analysis is carried out for various
shunt-faults at bus 2. The fault is applied at t = 0.5s and is maintained for a duration of 0.1 s. For
LG-fault, the rotor-angles and Teg are plotted as shown in Figure 8.39.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 412 1/29/2018 12:22:48 PM

 Fault Analysis of Power Systems  |  413 

Pre- and post-fault

Generator
current IGi
c
[ VPR] calculation
[ I GS] + [ V a(1) ]
[YBUS(1) ]
+ During fault

c
V PRf Short−circuit faults
c
I V PRf
fa(1) = [ Va(1)]
Zshf−eq
Fault Type
[ V a(1)] = [ Z jf(1) ] (− Ifa(1) )
Fault Simulation Logic

Figure 8.38  Calculation of Positive-sequence Bus Voltages for Stability Studies.

2−machine 5−bus power system (LG fault)

0.3

0.2
Rotor angle wrt COI (rad)

0.1
G−1
0
G−4

−0.1

−0.2

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Time (s)

1.1
Torque developed (pu)

0.9

0.8

0.7

0.6

0.5

0.4
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (s)

Figure 8.39  Rotor-angle and Torque Plots for LG-fault at Bus 2.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 413 1/29/2018 12:22:49 PM

 414  |  Power System Analysis

Note that the rotor-angles are plotted relative to the COI-frame and are obtained as

d COIi = d i − d COI

H1d1 + H 4d 4
where d COI = and significance of this is explained in the latter Chapters.
( H1 + H 4 )
Similar plots are shown for LLG- and symmetrical three-phase (LLL)-faults in Figures 8.40 and
8.41, respectively.
Comparing the torque plots in Figures 8.40 and 8.41, it can be seen that in terms of severity of
faults, the symmetrical three-phase fault is more severe than other types of faults. In other words,
during a symmetrical LLL-fault, the torque developed by machine-1 becomes zero thus, creating a
large accelerating torque unlike during LLG or LL or LG-faults. It is noted that LL-fault has an effect
similar to that of an LG-fault. The critical clearing time, TCr, for these faults is tabulated in Table 8.6.

2−machine 5−bus power system (LLG fault)

0.4

0.3
Rotor angle wrt COI (rad)

0.2

0.1

−0.1

−0.2

−0.3

−0.4
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (s)

1.2
Torque developed (pu)

0.8

0.6

0.4
G−1
0.2
G−4
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (s)

Figure 8.40  Rotor-angle and Torque Plots for LLG-fault at Bus 2.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 414 1/29/2018 12:22:50 PM

 Fault Analysis of Power Systems  |  415 

2−machine 5−bus power system (LLL fault)

0.5
Rotor angle wrt COI (rad)

−0.5
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (s)

1.5
Torque developed (pu)

0.5

G−1
G−4
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (s)

Figure 8.41  Rotor-angle and Torque Plots for Symmetrical Three-phase Fault at Bus 2.

Table 8.6  Critical Clearing Time for Shunt-faults at Bus 2.

Type of fault TCr (s)

LG More than 1 s
LL More than 1 s
LLG 0.339
LLL 0.240

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 415 1/29/2018 12:22:51 PM

 416  |  Power System Analysis

Notes:
1. The model used for the generators in the above example can be treated as a classical model in
terms of subtransient reactance.
2. The rotor swings clearly show that the system possesses negligible damping.
3. In the bus voltage computations, it can be seen that the voltages become a function of the
state-variable d and hence avoid algebraic loops [12] in the computation process. This makes the
implementation simple.
4. To carry out transient stability analysis with detailed model for generators, the generators are
modelled in the usual way in the 0dq-frame employing positive-sequence reactances. The nega-
tive-sequence reactance is obtained as per eq. (8.55) or (8.56) depending on the model. The net-
work is modelled in the 0DQ-frame. Hence, all machine currents are referred from one frame to
the other using the respective rotor angles. These details are given in the following chapters. The
machine current, IGi, is determined using only the positive-sequence component of the generator
terminal voltage in the 0DQ-frame.

References
[1]  Fortescue C. L., Method of Symmetrical Co-ordinates Applied to the Solution of Polyphase
Networks, AIEE Transactions, vol. 37, pp. 1027–1140, 1918.
[2]  Glenn W. Stagg and Ahmed H. El-Abiad, Computer Methods in Power System Analysis, McGraw-
Hill Inc., New York, 1968.
[3]  Blackburn J. L., Symmetrical Components for Power Systems Engineering, Marcel Dekker,
New York, 1993.
[4]  J. Arrillaga and C. P. Arnold, Computer Analysis of Power Systems, John Wiley & Sons Ltd,
England, 1990.
[5]  M. A. Pai, Computer Techniques in Power System Analysis, Tata McGraw-Hill Inc., Delhi, 1979.
[6]  John J. Grainger and William D. Stevenson, Jr. Power System Analysis, McGraw-Hill Inc.,
New York, 1994.
[7]  S. A. Soman, S. A. Khaparde and Shubha Pandit, Computational Methods for Large Sparse
Power System Analysis, Kluwer, 2002.
[8]  P. Kundur, Power System Stability and Control, McGraw-Hill Inc., New York, 1994.
[9]  K. R. Padiyar, Power system dynamics-stability and control, BS Publications, Hyderabad, India,
2002.
[10]  Olle I. Elgerd, Basic Electric Power Engineering, Addison - Wesley Publishing Co., London,
1977.
[11]  J. Duncan Glover and Mulukutla S. Sarma, Power System - Analysis and Design, Thomson
Learning-Brooks Cole Pvt. Ltd, Asia, 2002.
[12]  Using MATLAB, Version 5.3, Release 11, The Math Works Inc.

Review Questions
1. Enumerate the major differences between the symmetrical component transformation matrix and
the Park transformation matrix.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 416 1/29/2018 12:22:51 PM

 Fault Analysis of Power Systems  |  417 

2. Given that T
  2p   2p  
vabc = 10 sin(w 0 t ) 9 sin  w 0 t −  10 sin  w 0 t + 
  3  3  

determine the sequence components, Va(012).

3. For a power system shown in Figure 8.42, obtain the positive-sequence bus admittance matrix.
The line data, transformer data and generator data are given in Exercise section of Chapter 14.

1 2

3 4

Figure 8.42  Two-machine 6-bus Power System.

4. For the system shown in Figure 8.42, neglecting the pre-fault load currents and dynamic saliency
of the generators, determine the subtransient fault current for a symmetrical 3-phase fault at bus 1.
5. Consider the single-line diagram of a 60-Hz power system shown in Figure 8.43.
PL3 Y ∆ Y
T1
1 2 3 4
G1 G4

Y ∆ Y T2

5
PL5

Figure 8.43  A Two-machine Bus 5 Power System.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 417 1/29/2018 12:22:52 PM

 418 | Power System Analysis

The component details are given in the text. By performing the fault calculations for LG, LL, and
LLG fault at bus 5, determine the following:
(a) The fault currents at actuals.
(b) The currents delivered by generators G1 and G4 during the faults in abc-frame.
(c) The terminal voltages of the generators in abc-frame at actuals.
6. In Figure 8.43, if the neutrals of both transformers T1 and T2 are un-grounded, determine the
phase-b and phase-c voltages in kV for an LG phase–a fault at bus 2.
7. Perform the transient stability analysis of the power system shown in Figure 8.43, for LG, LL, and
LLG fault at bus 5 and estimate the critical clearing time.

M08_POWER-SYSTEM-AN00_SE_XXXX_CH08.indd 418 1/29/2018 12:22:54 PM

 Introduction to
Sub-synchronous Resonance 9
CHAPTER OUTLINE

9.1 Sub-synchronous Resonance 9.2 SSR Study of a Simplified System

In this chapter, the subsynchronous resonance (SSR) that occurs in a fixed series capacitor (FSC)-
compensated transmission line connected to a steam turbine-generator power system is illustrated by
means of a tutorial example to enhance the understanding of SSR. When such an event takes place,
it triggers large amplitude oscillations of the rotor shaft system below the fundamental frequency. If
this occurs repeatedly, the shaft system is subject to severe strain and fatigue, leading to damage of
the mechanical systems.
Here, the occurrence of such a phenomenon is explained in steps, considering a simplified model
for the generator. The discussion helps the reader to appreciate the need for involved computations to
study SSR.

9.1 SUB-SYNCHRONOUS RESONANCE

It is well known that the application of FSC compensation is an economical method to enhance the
power transfer capability and stability of long-distance transmission systems [1]. However, SSR is a
potential threat to steam turbine generators connected to FSC-compensated transmission lines, lead-
ing to adverse torsional interactions which may result in shaft failure of mechanical systems [2, 3,
4, 5, 6]. To analyse SSR, a detailed representation needs to be employed for the mechanical system,
network, and generators. This makes the analysis of SSR, a complex task. The IEEE first-benchmark
model (FBM) [7], which consists of a synchronous generator connected to an infinite bus through an
FSC-compensated transmission line, is a frequently used system for SSR studies. However, under-
standing of SSR even in such a simple radial system is not easy as it involves many modal interactions.
In order to introduce the SSR in a tutorial manner, a two-mass mechanical system is designed in
such a way that it possesses only one torsional mode interaction. A simplified generator model is
employed which acts as a medium for exchange of energy between electrical and mechanical systems.

M09_POWER-SYSTEM-AN00_SE_XXXX_CH09.indd 419 9/12/2017 5:50:04 PM

 420  |  Power System Analysis

Each system is studied individually to clearly bring out its natural behaviour. Finally, the systems are
interconnected to demonstrate the mutual excitation of the system modes depending on the level of
compensation of the transmission system. This demonstration model is expected to give better insight
into the concepts of SSR.

9.2  SSR STUDY OF A SIMPLIFIED SYSTEM

A steam turbine generator system connected to an FSC-compensated transmission line is shown in
Figure 9.1. A simplified voltage source-behind a sub-transient reactance model has been employed for
the generator electrical circuit. The generator voltage Vg is held constant at the specified value Vg0 and
only dg is assumed to change as constrained by the mechanical system. The generator sub-transient
reactance is absorbed in the line reactance for ease of analysis.

Turb Gen

Pg RL xFC Eb = Eb∠0°
xL

Vg = Vg ∠d g
Generator Infinite bus

Figure 9.1  A Simple Power System Model for SSR Analysis.

9.2.1  Turbine-generator Mechanical Systems

The mechanical system is represented by a two mass-spring-dashpot system as shown in Figure 9.2.

Tmt −Teg

Ktg

Turbine Generator
(Ht) (Hg)
Dtg

Dt Dg

Tmt
Mechanical system dg
Teg

Figure 9.2  Model of the Turbine-generator Mechanical System.

The equations of the two rotor-mass system are summarised as follows [4]:

M09_POWER-SYSTEM-AN00_SE_XXXX_CH09.indd 420 9/12/2017 5:50:20 PM

 Introduction to Sub-synchronous Resonance  |  421 

Turbine:

dd t  w 
= wt − 0  w B
dt  wB 

dw t
2H t = Tmt − Ktg (d t − d g ) − Dtg (w t − w g ) − Dt w t (9.1)
dt

Generator:

dd g  w 
= w g − 0  w B
dt  wB 

dw g
2H g = −Teg − Ktg (d g − d t ) − Dtg (w g − w t ) − Dg w g (9.2)
dt

The above equations are now written in terms of slip speed. We know that

 w 
Si =  w i − 0 
 wB 

Neglecting damping and setting the forcing functions, Tm = 0 and Teg = 0, we have,
dd t
= St w B
dt

dSt
2H t = − Ktg (d t − d g ) (9.3)
dt
dd g
= S gw B
dt
dS g
2H g = − Ktg (d g − d t ) (9.4)
dt
Since the above system of equations represent a linear time-invariant (LTI) system, we can write the
above equations in matrix form to get,

   0 0 wB 0  
  
d t   0 0 0 w B  d t 
 
d   K Ktg  d 
 g  =  − tg 0 0  g
 St   2 H t 2 Ht   St 
    S 
 Sg   Ktg −
Ktg
0 0  g
   2H 2H g   
   g

M09_POWER-SYSTEM-AN00_SE_XXXX_CH09.indd 421 9/12/2017 5:50:24 PM

 422  |  Power System Analysis

To determine the modal (natural) frequencies of the spring-mass system, the eigenvalues of the above
state matrix are determined for Ktg = 41.88, pu/rad., Ht = 0.72727 s, Hg = 1.6 s, and wB = 2π 60 rad/s,
using the eig function in MATLAB. The eigenvalues are given by

0, 0, + j125.66, − j125.66 rad/s

In the above list, the two zero eigenvalues denote the redundancy associated with state variables,
125.66
and the natural frequency corresponds to ±  j125.66 rad/s and f m = = 20 Hz. This denotes the
2p
oscillatory torsional frequency.

9.2.2  Electrical Systems

The electrical circuit for the network is shown in Figure 9.3. Here, RL denotes the combined resist-
ance of the line and generator and xL represents the combined inductive line reactance and the sub-
transient reactance of the generator at fundamental frequency. xFC denotes the reactance of the FSC
at fundamental frequency. Neglecting resistance, the natural frequency, fer, of the electrical circuit is
given by
1
2p f er = (9.5)
LC
where L and C denote the inductance (in henry) and capacitance (in farad), respectively.
Now using the fundamental frequency reactances, xL and xFC, we can write the above equation as

1
2p f er = (9.6)
xL 1
w 0 x FC w 0
or
x FC
f er = f 0 (9.7)
xL

where f0 is the nominal system frequency in Hz.

Vmsin(w 0t + dg) Emsin(w 0t)

RL xL xFC

Pg
Generator Infinite bus

dg Electrical system Teg

Figure 9.3  An FSC Compensated Electrical Network for Phase-a.

Since normally xFC < xL, this implies that fer < f0, signifying that the resonance phenomenon occurs at
a frequency below the nominal frequency.

M09_POWER-SYSTEM-AN00_SE_XXXX_CH09.indd 422 9/12/2017 5:50:28 PM

 Introduction to Sub-synchronous Resonance  |  423 

The differential equation in per-unit (pu) for the electrical system for a-phase is written as in
eq. (9.8), where ia and vca denote the line current and the capacitor voltage for a-phase, respectively.

L dia
= va − RL ia − vCa − Eba
w B dt

C FC dvC a
= ia (9.8)
w B dt
where

va = Vm sin(w 0 t + d g ) and Eba = Em sin(w 0 t )

(9.9)

2 2
with Vm = Vg and Em = Eb .
3 3

Also, note that

1
L( pu ) = x L ( pu ) and C FC ( pu ) =
x FC ( pu)

and the base frequency wB = w0.

The above equations are written similarly for other two-phases. After solving the above equations
in the abc-frame, the generator torque, Teg, is computed as

Teg = vaia + vb ib + vc ic (9.10)

In the above expression, it is assumed that the speed deviation is negligible.

9.2.2.1 Equations for Electrical Systems in Synchronous-frame

Instead of solving the electrical equations in the abc-frame, we can also write the equations in the
synchronous-frame (i.e., DQ-frame) [8] for a balanced system as follows:

dvCD w
= B iD − w B vCQ
dt C FC

dvCQ wB
= iQ + w B vCD
dt C FC
diD w B w R w w
= vD − B L iD − w B iQ − B vCD − B ED
dt L L L L

diQ wB w R w w
= vQ − B L iQ + w B iD − B vCQ − B EQ (9.11)
dt L L L L
where
vQ = Vg 0 cos d g and vD = Vg 0 sin d g (9.12)

M09_POWER-SYSTEM-AN00_SE_XXXX_CH09.indd 423 9/12/2017 5:50:33 PM

 424  |  Power System Analysis

It is to be noted that such a system is less stiff compared to that given in eq. (9.8) since an unstable
frequency fer in abc-frame appears as (f0 − fer) in the synchronous-frame.
The generator torque can be calculated as

Teg = vQ iQ + vD iD (9.13)

9.2.3  System Parameters and Operating Conditions

The mechanical and electrical system parameters (in pu) are given in Tables 9.1 and 9.2, respectively.

Table 9.1  Mechanical System Parameters.

Ktg Dtg Dt Dg Ht Hg

41.88 0.8 0.6 0.9 0.72727 1.6

Table 9.2  Electrical System Parameters.

RL xL xFC f0

0.002 0.5 kcxL 60 Hz

where, kc denotes the degree of compensation.

The system operating conditions (in pu) are listed in Table 9.3.

Table 9.3  System Operating Conditions.

Vg0 Pg Eb

1.0 1.0 1.0∠0°

The SSR analysis has been carried out in the following order employing two values for kc:
1. Analysis of only the electrical system.
2. Analysis of the electrical system which is interfaced to the mechanical system without connecting
the mechanical variable back to the electrical system—partial system.
3. Analysis of the combined system where the electrical system is connected to the mechanical sys-
tem with the mechanical variable connected back to the electrical system—complete system.

9.2.4  Analysis of Only Electrical Systems

To start with, the electrical system alone is analysed using eq. (9.11). The magnitude of the terminal
voltage, Vg, is assumed to be constant at the specified value. The angle dg is held fixed at the nominal
value, implying that the mechanical system dynamics are not considered. For kc = 0.45, the resonance
frequency of the electrical system is obtained as fer = 40.25 Hz, using eq. (9.7). Without any distur-
bance (up to t = 0.5 s), Teg is constant as shown in Figure 9.4. This corresponds to only the fundamental

M09_POWER-SYSTEM-AN00_SE_XXXX_CH09.indd 424 9/12/2017 5:50:33 PM

 Introduction to Sub-synchronous Resonance  |  425 

frequency component in generator currents. The system is perturbed by 0.01 pu step reduction in
the infinite bus voltage at t = 0.5 s which lasts for 0.01 s. This results in a frequency component of
fer = 40.25 Hz in generator current in addition to the fundamental frequency component. This gives
rise to an oscillatory component of torque having frequency (  f0 − fer) = 19.75 Hz, which is referred to
as sub-synchronous frequency (see the expanded view in Figure 9.4).

1.04

1.02

Teg
1
(pu)

0.98

0.96
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

1.04
Expanded view
1.02

Teg
1
(pu)

0.98 1 = 19.76 Hz
f=
(t2 −t1)
t1 = 1.0849 t2 = 1.1355
0.96
1 1.02 1.04 1.06 1.08 1.1 1.12 1.14 1.16 1.18 1.2
Time (s)

Figure 9.4  Only Electrical System: Teg Plot for kc = 0.45.

9.2.5  Analysis of Partial Systems

In this case, the electrical system is connected to the mechanical system without using the mechan-
ical variable dg in the electrical system. This has been achieved with a configuration as shown in
Figure 9.5. The mechanical system is implemented using eqs (9.1) and (9.2).

Tmt
dg Teg
Mechanical system Electrical system

Figure 9.5  Partial System Inter-connection.

In this system, the following case studies are carried out.

M09_POWER-SYSTEM-AN00_SE_XXXX_CH09.indd 425 9/12/2017 5:50:34 PM

 426  |  Power System Analysis

9.2.5.1  Case-A (kc = 0.45)

When the electrical system is perturbed as mentioned above, the torque contains a component of
frequency 19.75 Hz, which is close to the torsional frequency, fm = 20 Hz. This excites the torsional
mode as shown in Figure 9.6. However, note that the oscillations are not sustained as the interaction
is only in one way.

1.04

1.02

Teg
1
(pu)

0.98

0.96
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

1.002

1.001

wt
1
(pu)

0.999

0.998
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (s)

Figure 9.6  Partial System: kc = 0.45.

9.2.5.2  Case-B (kc = 0.25)

The above case study has been repeated with kc = 0.25, which results in fer = 30 Hz. Since the fre-
quency of the sub-synchronous component in the torque is ( f0 − fer ) = 30 Hz, which is not in the
vicinity of the torsional frequency ( fm = 20 Hz), the torsional mode does not get excited. This can be
inferred from wt plot as shown in Figure 9.7.

9.2.6  Analysis of the Complete Systems

Having considered the partial interaction among the systems, in order to study the SSR phenomenon,
the block diagram is completed as shown in Figure 9.8. Here, the mechanical variable dg obtained
from eq. (9.2) is used in eq. (9.12) of the electrical system.
The time-domain simulation of the complete system is carried out by using the following set of equations:
1. The turbine mechanical systems are given by eq. (9.1).
2. The generator mechanical systems are given by eq. (9.2).

M09_POWER-SYSTEM-AN00_SE_XXXX_CH09.indd 426 9/12/2017 5:50:34 PM

 Introduction to Sub-synchronous Resonance  |  427 

1.04

1.02

Teg
1
(pu)

0.98

0.96
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

1.002

1.001
wt
1
(pu)

0.999

0.998
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (s)

Figure 9.7  Partial System: kc = 0.25.

Tmt
dg Teg
Mechanical system Electrical system

Figure 9.8  Complete Block Diagram for SSR Study.

3. The electrical system equations in the synchronous frame are given by eqs (9.11) and (9.12).
4. The electromagnetic torque generated is given by eq. (9.13). This value is, in turn, used in
eq. (9.2).
5. The rotor-angle obtained from eq. (9.2) is substituted in eq. (9.12).
The system is perturbed by 0.01 pu step reduction in the infinite bus voltage at t = 0.5 s, which lasts
for 0.01 s. In this system, the following cases are considered.

9.2.6.1  Unstable Torsional Mode (kc = 0.45)

Here, the parameters used in Case-A are adopted. This excites the torsional mode and the electri-
cal circuit oscillations in a cumulative manner as the interactions work both ways. The resulting
20 Hz oscillation in dg variable modulates the generator voltage at two dominant frequencies, one at

M09_POWER-SYSTEM-AN00_SE_XXXX_CH09.indd 427 9/12/2017 5:50:35 PM

 428  |  Power System Analysis

0.8

0.7

0.6

0.5
va - FFT

0.4

0.3

0.2

0.1

0
0 1 2 3 4 5 6 7 8 9 10
Order of Harmonic

Figure 9.9  Fast Fourier Transform (FFT) of va Waveform for kc = 0.45 with 10 Hz
as the Base for FFT.
(  f0 − fm) = 40 Hz and another at (  f0 + fm) = 80 Hz [4]. This has been demonstrated in Figure 9.9 by
extracting the frequency components in va.
Since the sub-synchronous frequency component in voltage is close to fer, it leads to a large sub-­
synchronous current in ia in addition to the fundamental frequency. This fact is substantiated by deter-
mining the root mean square (RMS) value of the line current (see Figure 9.10).
Until 0.5 s, a constant RMS line current denotes only the fundamental frequency component in ia,
and beyond 0.5 s, the RMS current contains a dominant frequency (  f0 − fm) = 20 Hz. This, in turn,
produces sub-synchronous torque component which reinforces the rotor oscillations at frequency, fm.
Such a cumulative process causes the coupled electromechanical system to experience oscillations of
large magnitude (Figure 9.11).

9.2.6.2  Stable Torsional Mode (kc = 0.25)

In this case, the parameters selected in Case-B are used. Here, the system remains stable. The corre-
sponding time-domain simulation is shown in Figure 9.12.
Notes:
1. The vabc and iabc are obtained by transforming the quantities in the synchronous frame to the abc-
frame by using KRON’s inverse transformation. Here, v0 and i0 are set to zero. The transformation
is given by
−1
f abc = [ PK ] f 0 DQ

2. The RMS current is obtained as

I RMS = (iQ2 + iD2 )

M09_POWER-SYSTEM-AN00_SE_XXXX_CH09.indd 428 9/12/2017 5:50:36 PM

 Introduction to Sub-synchronous Resonance  |  429 

0.5
ia
0
(pu)

−0.5

−1
0.4 0.5 0.6 0.7 0.8 0.9 1

1.15

1.05
Irms
(pu)
0.95

0.85
0.4 0.5 0.6 0.7 0.8 0.9 1
Time (s)

Figure 9.10  Complete System: ia and IRMS Waveforms for kc= 0.45.

1.05

Teg
1
(pu)

0.95

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

1.002

1.001

wt
1
(pu)
0.999

0.998
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (s)

Figure 9.11  Complete System: kc = 0.45.

M09_POWER-SYSTEM-AN00_SE_XXXX_CH09.indd 429 9/12/2017 5:50:37 PM

 430  |  Power System Analysis

1.04

1.02
Teg
(pu) 1

0.98

0.96
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

1.002

1.001

wt
1
(pu)

0.999

0.998
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (s)

Figure 9.12  Complete System: kc= 0.25.

9.2.7  Eigenvalue Analysis of the Complete System

In order to support the stability information obtained from the time-domain simulation, an eigenvalue
analysis of the complete system is carried out by linearising the equations in synchronous reference
frame [4].

9.2.7.1  Linearising the Mechanical Equations

T
Choosing the state vector as x t = [d t , w t , d g , w g ] and from eqs (9.1) and (9.2), we have,

∆ x t = [ At ]( 4 × 4) ∆ x t + [ Bt ]( 4 ×1) ∆Teg (9.14)

where
 0 wB 0 0 
 K   0 
Dtg + Dt Ktg Dtg  0 
 − tg − 
 2 Ht 2 Ht 2 Ht 2 Ht   
[ At ]( 4 × 4) =  and [ Bt ]( 4 ×1) =  0 
 0 0 0 wB   
 Ktg Dtg Ktg Dtg + Dg  − 1 
 − −   2H g 
 
 2 H g 2H g 2H g 2 H g 

M09_POWER-SYSTEM-AN00_SE_XXXX_CH09.indd 430 9/12/2017 5:50:39 PM

 Introduction to Sub-synchronous Resonance  |  431 

T
For the electrical circuit, choose the state vector as xN = [vcD , vcQ , iD , iQ ] . Now, linearizing the
generator torque expression from eq. (9.13) we get,

∆Teg = vQ 0 ∆iQ + iQ 0 ∆vQ + vD 0 ∆iD + iD 0 ∆vD (9.15)

From eq. (9.12), we can see that ∆vQ = −Vg 0 sind g 0 ∆d g and ∆vD = Vg 0 cosd g 0 ∆d g . Therefore,
the above equation can be rewritten as

∆Teg = [Sr 2 ](1× 4) ∆x t + [Sr1 ](1× 4) ∆x N (9.16)

where
(
[Sr 2 ](1× 4) = 0, 0, iD 0Vg 0 cosd g 0 − iQ 0Vg 0 sin d g 0 , 0  )
and
[Sr1 ](1× 4) = 0, 0, Vg 0 sind g 0 , Vg 0 cos d g 0 

Using the above matrices, expression in (9.14) can be written as

( )
∆x t = [ At ]( 4 × 4) + [ Bt ]( 4 ×1) [Sr 2 ](1× 4) ∆x t + [ Bt ]( 4 ×1) [Sr1 ](1× 4)) ∆x N (9.17)

9.2.7.2  Linearising the Electrical Equations

The electrical equations given in eqs. (9.11) and (9.12) are linearised to obtain,

∆x N = [ AN ]( 4 × 4) ∆x N + [ S r 3 ]( 4 ×1) ∆d g
(9.18)
where
 wB 
 0 −w B 0 
C FC  0 
   
 wB  0
 
 wB 0 0
C FC   w 
[ AN ]( 4 × 4) =  and [Sr 3 ]( 4 ×1) =  B
Vg 0 cos d g 0 
 wB w B RL   L 
− L 0 −
L
−w B 
 − w B V sind 
   L g 0 g0 
 0 wB w R 
− wB − B L
 L L 

with ∆d g = [ I 3 ](1× 4) ∆x t and [I3] = [0,0,1,0].

Using the above matrices, we can write eq. (9.18) as

∆x N = [ AN ]( 4 × 4) ∆x N + [ BN ]( 4 × 4) ∆x t (9.19)

where [ BN ]( 4 × 4) = [Sr 3 ]( 4 ×1) [ I 3 ](1× 4) .

M09_POWER-SYSTEM-AN00_SE_XXXX_CH09.indd 431 9/12/2017 5:50:47 PM

 432  |  Power System Analysis

Using eqs. (9.17) and (9.19), the complete state matrix can be obtained as
∆x T = [ A](8×8) ∆x T (9.20)
where


(
 [ At ]( 4 × 4) + [ Bt ]( 4 ×1) [Sr 2 ](1× 4) ) [ Bt ]( 4 ×1) [Sr1 ](1× 4) 

[ A](8×8) = 
 [ BN ]( 4 × 4) [ AN ]( 4 × 4) 
 

and xT = [ xt , xN ] . .
T

The eigenvalue of the [A] matrix is determined to analyse the possibility of torsional interaction.
Table 9.4 shows the initial value of variables for the two compensation levels considered.

Table 9.4  Initial Values of Variables for Two Values of kc.

Variable kc = 0.45 kc = 0.25

dg0 15.95° 22°
iD0 0.1472 0.1995
iQ0 0.9979 0.9979
vD0 0.2747 0.3746
vQ0 0.9615 0.9272

The eigenvalues for the above cases are tabulated in Table 9.5. It can be seen from Table 9.5 that for
kc = 0.45, since the network mode-2 frequency is close to the torsional mode, the torsional mode gets
destabilised. For kc = 0.25, the eigenvalues are tabulated in the second column of Table 9.5. Here, since
the network mode-2 frequency (equal to 29.87 Hz) is not in the vicinity of the torsional frequency
(which is 20.09 Hz), the torsional interaction is not sustained.

Table 9.5  Eigenvalues for the Two Cases—Complete System.

k kc = 0.45 kc = 0.25 Comments

l12 −0.7539 ± j 629.83 −0.7538 ± j 565.41 Network mode-1
l34 −5.8357 ± j 124.52 −0.7615 ± j 187.7 Network mode-2
l56 4.4840 ± j 124.54 −0.5821 ± j 126.28 Torsional mode
l78 −0.1492 ± j 16.912 −0.1573 ± j 14.157 Swing mode

9.2.7.3  Mode Identification

After obtaining the eigenvalues, in order to identify the nature of a mode one can use the right-­
eigenvectors. For kc = 0.25, the magnitude and angle of the right-eigenvectors are listed in Tables 9.6
and 9.7, respectively.

M09_POWER-SYSTEM-AN00_SE_XXXX_CH09.indd 432 9/12/2017 5:50:48 PM

 Introduction to Sub-synchronous Resonance  |  433 

Table 9.6  Magnitude of Right-eigenvectors for kc = 0.25.

State k12 k34 k56 k78

dt 0 0.0013 0.3742 0.3331
wt 0 0 0.1253 0.0125
dg 0 0.0028 0.1755 0.3269
wg 0 0.0014 0.0588 0.0122
vcD 0.1715 0.1709 0.1247 0.1017
vcQ 0.1715 0.1707 0.1043 0.0428
iD 0.6859 0.6845 0.5091 0.3356
iQ 0.6860 0.6877 0.7241 0.8106

Table 9.7  The Angle (in deg.) of Right-eigenvectors for kc = 0.25.

State k12 k34 k56 k78

dt −156.52 −158.14 13.89 −178.52
wt −66.440 −67.90 104.16 −87.88
dg 21.89 21.56 −166.19 −178.51
wg 111.97 111.97 −75.92 −87.87
vcD −0.23 −179.85 −3.77 −0.87
vcQ −90.23 90.06 −103.86 −164.8
iD 89.98 89.75 −110.63 −169.83
iQ 0 0 180 180

Using the magnitude of right-eigenvectors, it is not easy to identify the nature of a mode. Hence, in
Table 9.8, the magnitude of the participation factors are listed. From the table it can be seen that in
modes l12 and l34, the participation of mechanical variables is negligible, whereas the network vari-
ables have the highest participation. Hence, these modes are referred to as network modes. In modes
l56 and l78, the mechanical variables have the highest participation. Therefore, these modes are asso-
ciated with the rotor oscillations.

Table 9.8  Magnitude of the Participation Factors for kc = 0.25.

State k12 k34 k56 k78

dt 0 0 0.3394 0.1610
wt 0 0 0.3394 0.1610
dg 0 0.0054 0.1642 0.3413
wg 0 0.0054 0.1642 0.3413
vcD 0.2501 0.2518 0.0014 0.0024
vcQ 0.2501 0.2518 0.0014 0.0024
iD 0.2499 0.2540 0.0024 0.0096
iQ 0.2499 0.2540 0.0024 0.0093

M09_POWER-SYSTEM-AN00_SE_XXXX_CH09.indd 433 9/12/2017 5:50:48 PM

 434  |  Power System Analysis

To further classify the mechanical modes, the right-eigenvectors pertaining to the speed variables are
drawn using compass plot corresponding to each of the modes. For mode l56, the compass plot is shown
in Figure 9.13. Here, it can be noted that the vectors are 180° apart (as can be verified from Table 9.7,
where the angles are 104.16° and −75.92°) and they constitute a relative motion with respect to each other.
This mode is referred to as torsional mode in which two masses oscillate one against the other. This can be
verified in the time-domain plots of turbine-rotor speed, wt and the generator-rotor speed, wg (see Figure
9.14). Due to these oscillations, while the rotors are running at the synchronous speed, the shaft connecting
these two rotor masses experiences a torsional force. Hence, this mode is attributed as a torsional mode.
90 0.15
120 60

0.1
150 30
wt 0.05

180 0
wg

210 330

240 300
270
Figure 9.13  Compass Plot of Speed Right-eigenvectors for Mode l56.
kc = 0.25

1.0001 wt
wg
1.0001

1.0001

1.0001

1
pu Speed

0.9999

0.9999

0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8

Time (s)
Figure 9.14  The Plots of wt and wg for kc = 0.25.

M09_POWER-SYSTEM-AN00_SE_XXXX_CH09.indd 434 9/12/2017 5:50:51 PM

 Introduction to Sub-synchronous Resonance  |  435 

For mode l78, the compass plot is shown in Figure 9.15. Here, it can be noted that the vectors are
collinear and they constitute a common motion (as can be verified from Table 9.7, where the angles
are −87.88° and −87.87°). In this mode, the two rotor masses swing together (or in-phase) against the
infinite bus. This mode is referred to as swing mode in which the two masses acts as one rigid body.

90 0.015
120 60

0.01
150 30
0.005

180 0

w t and w g

210 330

240 300
270

Figure 9.15  Compass Plot of Speed Right-eigenvectors for Mode l78.

The above example illustrates the following important concepts:

1. Occurrence of SSR in an electromechanical system denotes a state where electrical systems
exchange energy with the mechanical systems through the synchronously revolving rotor systems.
2. Torsional frequencies are generally in the sub-synchronous frequency range. Depending on the
degree of line compensation, the sub-synchronous torque components may or may not excite a
particular torsional mode. Thus, SSR is a discrete event. Each torsional mode gets tuned only for
a certain level of series compensation.
3. Since the magnitude of the disturbance chosen is small, the super-synchronous frequency compo-
nent in the torque is not excited appreciably and it damps out quickly.
4. The example considers only the torsional interaction effects. In general, SSR is triggered due to
the combined effects of torsional interaction and induction generator effects.

References
[1]  E. W. Kimbark, ‘Improvement of system stability by switched series capacitors.’ IEEE Transaction
on Power Apparatus and Systems, vol. 85(2), pp. 180–188, 1966.
[2]  P. M. Anderson, B. L. Agrawal, and J. E. Van Ness, Subsynchronous resonance in power systems,
IEEE Press, Piscataway, New Jersey, 1990.
[3]  IEEE Committee Report ‘Reader’s guide to subsynchronous resonance.’ IEEE Transaction on
Power Systems, vol. 7(1), pp. 150–157, 1992.

M09_POWER-SYSTEM-AN00_SE_XXXX_CH09.indd 435 9/12/2017 5:50:52 PM

 436  |  Power System Analysis

[4]  
K. R. Padiyar, Analysis of subsynchronous resonance in power systems, Kluwer Academic
Publishers, Norwell, MA, USA, 1999.
[5]  K. Clark, ‘Overview of subsynchronous resonance related phenomena.’ Proceedings of IEEE/PES
Transmission, Distribution Conference and Exposition, Orlando, Florida, pp. 1–3, 2012.
[6]  C. E. Ugalde-Loo, J. B. Ekanayake, and N. Jenkins, ‘Subsynchronous resonance in a series-
compensated Great Britain transmission network.’ IET Generation, Transmission and Distribution,
vol. 7(3), pp. 209–217, 2013.
[7]  IEEE SSR Task Force, ‘First benchmark model for computer simulation of subsynchronous res-
onance.’ IEEE Transaction on Power Apparatus and Systems, vol. 96(5), pp. 1565–1572, 1977.
[8]  K. R. Padiyar, Power system dynamics - stability and control, BS Publications, Hyderabad,
India, 2002.

Review Questions
1. How to identify the torsional modes and swing modes in a system?
2. For the spring-mass turbine-generator system shown in Figure 9.16, write a state-space model and
determine the torsional frequency of the (unconnected) system. Given that Hg = 0.3003 s, Ht = 4.5 s
and Ktg = 0.3 pu. All dampings are set to zero.

K tg
Tw Wind Induction
Te
turbine generator
D tg
Dt Dg

Figure 9.16  Two-mass Spring System.

3. Derive an expression for natural frequency of the electrical system.

4. For a balanced system, show that in per-unit, Teg = vQiQ + vDiD. State the assumption made in this
derivation.
5. Obtain the initial values of variables as in Table 9.4, for kc = 0.15.

M09_POWER-SYSTEM-AN00_SE_XXXX_CH09.indd 436 9/12/2017 5:50:52 PM

 SSR Analysis of the IEEE
First Benchmark Model 10
CHAPTER OUTLINE

10.1 Turbine-Generator Mechanical 10.4 Transmission Network Modelling

System Equations 10.5 Listing of Linearised State Equations
10.2 Generator Modelling 10.6 Machine-frame-based Time-
10.3 Exciter Modelling domain Simulation of SSR

In this chapter, sub-synchronous resonance (SSR) analysis of the IEEE first benchmark model (FBM)
is studied in detail. Since SSR involves exchange of energy between the electrical systems and the
turbine mechanical systems through the generator rotor, the analysis requires the representation of
both the electromechanical dynamics of the generating unit and the electromagnetic dynamics of the
transmission network.
We are aware that the differential and algebraic equations (DAEs) which describe the dynamic
performance of the synchronous machine and the transmission network are, in general, non-linear.
Therefore, to carry out the small-signal stability analysis, these equations must be linearised with
respect to an operating point. First, individual components such as synchronous generator, turbine-
generator mechanical system, and electric network are linearised. Then, all the equations are combined
in a systematic fashion to evolve the system state matrix, using which eigenvalues are computed.
Further, the time-domain simulation of the complete system is presented to validate the eigen-
predictions. To get better insight into the system behaviour at different fixed series capacitor compen-
sation levels, modal speeds are computed. In addition to torsional interaction-related performances,
induction generator effects are also simulated.

10.1 TURBINE-GENERATOR MECHANICAL SYSTEM EQUATIONS

Since the problem due to interaction between the electrical systems and the rotor mechanical systems
is principally in the sub-synchronous frequency range, a simple lumped multi-mass model for rotor
mechanical system is employed [3].

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 437 9/12/2017 5:31:44 PM

 438  |  Power System Analysis

Figure 10.1 shows an example of the shaft model of the turbine generator system pertaining to the
IEEE FBM [2]. The model consists of six turbine sections modelled separately: a high-pressure stage
(denoted by HP), an intermediate-pressure stage (IP), two low-pressure stages (LPA and LPB), the
generator (G), and the exciter (E)-rotor masses. Each rotating mass is connected to the previous and
the following mass by elastic shaft sections, represented by a spring coefficient in pu torque/rad angle
(Ki, i − 1 and Ki, i + 1, respectively) and a mutual damping term (Di, i − 1 and Di, i + 1, respectively). The
term Di represents the self-damping term in pu damping torque/pu speed. TmH, TmI, TmLA, TLB, and
−Te are the forcing functions acting on the masses. wi denotes the per-unit speed with respect to the
absolute reference frame for the ith mass. w0 (elec. rad/s) represents the nominal speed and wB (elec.
rad/s) represents the base speed.

TmH TmI TmLA TmLB −Te

KHI KILA KLALB KLBG KGE

HP IP LPA LPB G E

DHI DILA DLALB DLBG DGE

DH DI DLA DLB DG DE

Figure 10.1  Mechanical Model of a Rotor-shaft Torsional System with Six Masses.

The equations of the complete rotor-mass system shown in Figure 10.1 may be summarised as follows:
HP:
dd H  w 
=  w H − o  w B (10.1)
dt  wB 
dw H
2H H = [TmH − K HI (d H − d I ) − DHI (w H − w I ) − DH w H ] (10.2)
dt
IP:
dd I  w 
=  w I − o  w B (10.3)
dt  wB 
dw I
2H I = [TmI − K HI (d I − d H ) − K ILA (d I − d LA ) − DHI (w I − w H )
dt
− DILA (w I − w LA ) − DI w I ] (10.4)
LPA:
dd LA  w 
=  w LA − o  w B (10.5)
dt  wB 
dw LA
2H LA = [TmLA − K ILA (d LA − d I ) − K LALB (d LA − d LB )
dt
− DILA (w LA − w I ) − DLALB (w LA − w LB ) − DLAw LA ] (10.6)

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 438 9/12/2017 5:32:08 PM

 SSR Analysis of the IEEE First Benchmark Model  |  439 

LPB:
dd LB  w 
=  w LB − o  w B (10.7)
dt  wB 
dw LB
2H LB = [TmLB − K LALB (d LB − d LA ) − K LBG (d LB − d G )
dt
− DLALB (w LB − w LA ) − DLBG (w LB − w G ) − DLBw LB ] (10.8)
GEN:
dd G  w 
=  w G − o  w B (10.9)
dt  wB 
dw G
2H G = [ −Te − K LBG (d G − d LB ) − KGE (d G − d E )
dt
− DLBG (w G − w LB ) − DGE (w G − w E ) − DGw G ] (10.10)

EXC:
dd E  w 
=  w E − o  w B (10.11)
dt  wB 
dw E
2H E = [ − KGE (d E − d G ) − DGE (w E − w G ) − DE w E ] (10.12)
dt
The rotor mechanical system data for the IEEE First Benchmark Model [2] is given in Table 10.1.

Table 10.1  Shaft Inertias and Spring Constants.

Mass Inertia H(s) Shaft section Spring constant K (pu torque/rad)
HP 0.092870 HP-IP 19.303
IP 0.155589 IP-LPA 34.929
LPA 0.858670 LPA-LPB 52.038
LPB 0.884215 LPB-GEN 70.858
GEN 0.868495 GEN-EXC 2.822
EXC 0.0342165

The damping data is not provided as a part of the FBM data. However, for case studies, the mechanical
damping (both self and mutual) data are selected as given in [1], which is listed below:

Self-damping: DHP = DIP = DLA = DLB = 0.2, DE = 0 DG = 0

Mutual damping: DHI = DILA = DLALB = DLBG = 0.3, DGE = 0.005

10.1.1 The IEEE First Benchmark System and Modal Frequencies

The single line diagram of the IEEE FBM is shown in Figure 10.2. In this system, a turbo-generator
is radially connected to a fixed series capacitor (FSC)-compensated network.

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 439 9/12/2017 5:32:12 PM

 440  |  Power System Analysis

Turbine RL = 0.02 RL0 = 0.02

xL = 0.50 xL0 = 0.56
EXC GEN LPB LPA IP HP
xSYS = 0.06
Infinite
Pg xT bus
2 3 4
Vg = 1.0 1 RL xL xFC xSYS

Eb = 1.0∠0°

xT = 0.14 Eb
Vg ∆

865.0 MVA
All data are in pu on generator MVA base

Figure 10.2  IEEE First Benchmark Model.

Following the modal transformation procedure indicated in [1], the modal frequencies are obtained
as shown in Table 10.2.

Table 10.2  Modal Frequencies for the IEEE FBM.

Mode 0 Mode 1 Mode 2 Mode 3 Mode 4 Mode 5
Frequency(rad/s) 0.0000 98.723 126.99 160.52 202.85 298.18
Frequency(Hz) 0.0000 15.712 20.211 25.547 32.285 47.456

The above frequencies are the natural frequencies of the isolated mechanical system neglecting damp-
ing. These frequencies are also obtained as the eigenvalues of the state-matrix [Asm](12 × 12), given in
eq. (10.13) with all damping set to zero.

10.1.2  Linearisation of Mechanical System of Equations

After linearising the mechanical system of eqs (10.1) to (10.12), the overall shaft equations are given
by the following matrix equation
∆xsm(12×1) = [ Asm ](12×12) ∆xsm(12×1) + [ ∆Tm′ ](12×1) − [ ∆Te′](12×1) (10.13)
where
x sm(12×1) = [d H w H d I w I d LA w LA d LB w LB d G w G d E w E ]T

The non-zero elements of [ ∆Tm′ ](12×1) are given by

∆TmH
[ ∆Tm′ ] (2,1) = 2H H
∆TmI
[ ∆Tm′ ] (4,1) = 2H I
∆TmLA
[ ∆Tm′ ] (6,1) = 2 H LA

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 440 9/12/2017 5:32:28 PM

 SSR Analysis of the IEEE First Benchmark Model  |  441 

∆TmLB
[ ∆Tm′ ] (8,1) = 2 H LB
The non-zero element of [ ∆Te′](12×1) is given by

∆T
[ ∆Te′] (10,1) = 2 H e
G

Notes:

1. [ ∆Tm′ ](12×1) is set equal to [ DTm1 ](12×6) ∆x t (6×1) to account for steam turbine and speed gov-
ernor system, if any. ∆xt (6×1) represents the states related to the turbine and governor models.
[ ∆Tm′ ](12×1) is set equal to zeros if turbine and governor systems are not modelled. This implies
that ∆TmH , ∆TmI , ∆TmLA , and ∆TmLB are zeros.

2. [ ∆Te′](12×1) in eq. (10.13) can be written as

[ ∆Te′](12×1) = [ DTe ](12×1) [ ∆Te ](1×1)(10.14)

where
1
[ DTe ](12×1) (10,1) =
2 HG

and rest of the elements is zero.

Also, note that ∆Te is the electromagnetic torque deviation. When this is expressed in terms of the
state variables pertaining to the generator, ∆xe (6×1) , we get,

[ ∆Te ](1×1) = [Ce ](1×6) ∆xe (6×1)(10.15)

This has been derived in Section (10.2.2). Substituting this in eq. (10.14), we get,
[ ∆Te′](12×1) = [ DTe1 ](12×6) ∆xe (6×1)(10.16)
where
[ DTe1 ](12×6) = [ DTe ](12×1) [Ce ](1×6)
Now, eq. (10.13) can be written as,
∆xsm (12×1) = [ Asm ](12×12) ∆xsm (12×1) + [ DTm1 ](12×6) ∆ xt (6×1) − [ DTe1 ](12×6) ∆ xe (6×1) (10.17)

Notes:
1. For jth mass, its slip Sj and its pu speed wj are related by
 w 
S j =  w j − o  (10.18)
 w B

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 441 9/12/2017 5:32:37 PM

 442  |  Power System Analysis

Linearising the above equation, we get,

∆S j = ∆w j
(10.19)
The non-zero elements of the matrix [Asm](12 ×12) are given by
1. 1st row:
Asm (1, 2) = w B

2. 2nd row:

K HI D + DH
Asm (2,1) = − ; Asm (2, 2) = − HI
2H H 2H H

K HI DHI
Asm (2, 3) = ; Asm (2, 4) =
2H H 2H H

3. 3rd row:

Asm (3, 4) = w B

4. 4th row:

K HI DHI
Asm ( 4,1) = ; Asm ( 4, 2) =
2H I 2H I

K HI + K ILA D + DILA + DI
Asm ( 4, 3) = − ; Asm ( 4, 4) = − HI
2H I 2H I

K ILA DILA
Asm ( 4, 5) = ; Asm ( 4, 6) =
2H I 2H I

5. 5th row:

Asm (5, 6) = w B

6. 6th row:

K ILA DILA
Asm (6, 3) = ; Asm (6, 4) =
2 H LA 2 H LA

K ILA + K LALB D + DLALB + DLA

Asm (6, 5) = − ; Asm (6, 6) = − ILA
2 H LA 2 H LA

K LALB DLALB
Asm (6, 7) = ; Asm (6, 8) =
2 H LA 2 H LA

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 442 9/12/2017 5:32:39 PM

 K ILA DILA
Asm (6, 3) = ; Asm (6, 4) =
2 H LA 2 H LA

K ILA + K LALB DILA + DLALB + DLA

Asm (6, 5) = − SSR ;Analysis
Asm (6,of = −IEEE
6) the First Benchmark Model  |  443 
2 H LA 2 H LA

K LALB DLALB
Asm (6, 7) = ; Asm (6, 8) =
2 H LA 2 H LA
7. 7th row:
Asm (7, 8) = w B

8. 8th row:
K LALB DLALB
Asm (8, 5) = ; Asm (8, 6) =
2 H LB 2 H LB

K LALB + K LBG D + DLBG + DLB

Asm (8, 7) = − ; Asm (8, 8) = − LALB
2 H LB 2 H LB

K LBG DLBG
Asm (8, 9) = ; Asm (8,10) =
2 H LB 2 H LB

9. 9th row:
Asm (9,10) = w B

10. 10th row:

K LBG DLBG
Asm (10, 7) = ; Asm (10, 8) =
2 HG 2 HG

K LBG + KGE D + DGE + DG

Asm (10, 9) = − ; Asm (10,10) = − LBG
2 HG 2 HG

KGE DGE
Asm (10,11) = ; Asm (10,12) =
2 HG 2 HG

11. 11th row:

Asm (11,12) = w B

12. 12th row:

KGE DGE
Asm (12, 9) = ; Asm (12,10) =
2H E 2H E

KGE D + DE
Asm (12,11) = − ; Asm (12,12) = − GE
2H E 2H E

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 443 9/12/2017 5:33:04 PM

 444  |  Power System Analysis

10.2  GENERATOR MODELLING

The generator is modelled using a 2.2 model. The generator parameters are given in Table 10.3.

Table 10.3  Generator Standard Parameters (on Machine Base 865 MVA, 22 kV), IEEE FBM.

xd = 1.79 x′d = 0.169 x″d = 0.135 Td0 = 4.3 s T ″d0 = 0.032 s

xq = 1.71 x′q = 0.228 x″q = 0.2 Tq0 = 0.85 s T ″q0 = 0.05 s
xl = 0.13 Ra = 0 HG = 0.868495 s DG = 0 fB = 60 Hz

10.2.1 Stator and Rotor Equations

Stator Equations:
dy d
= − Raw B id − w Gw B y q − w B vdg (10.20)
dt
dy q
= − Raw B iq + w Gw B y d − w B vqg (10.21)
dt
where wG is the per-unit speed of the generator.

Rotor d-axis Equations:

dy F 1  xd′ 
=  −y F + y d + E fd  (10.22)
dt Td′  ( xd − xd′ ) 
dy H 1

dt
=
Td′′
[ −y H + y d ] (10.23)
and

y d = xd′′id +
( xd′ − xd′′ ) y +
( xd − xd′ ) xd′′ y (10.24)
H F
xd′ xd xd′

Rotor q-axis Equations:

dy G 1
=  −y G + y q  (10.25)
dt Tq′  

dy K 1
=  −y K + y q  (10.26)
dt Tq′′  
and

y q = xq′′iq +
( xq′ − xq′′) y +
( xq − xq′ ) xq′′ y (10.27)
K G
xq′ xq xq′

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 444 9/12/2017 5:33:13 PM

 SSR Analysis of the IEEE First Benchmark Model  |  445 

10.2.2  Linearisation of Te
The per unit electrical torque acting on the generator rotor is given by

Te = y d iq − y q id (10.28)

Referring to eqs (10.24) and (10.27), id and iq can be expressed as

id =
1
yd −
( xd′ − xd′′ ) y − ( xd − xd′ ) y
H F (10.29)
xd′′ xd′ xd′′ xd xd′

iq =
1
yq −
(
xq′ − xq′′
yK −
)
xq − xq′ (
y G (10.30)
)
xq′′ xq′ xq′′ xq xq′

Linearisation of the above equations would result in

 ∆id 
 ∆i  = [Ce1 ]( 2×6) ∆xe (6×1) (10.31)
 q  ( 2×1)
where

 
1 0 −
( xd − xd′ ) −
( xd′ − xd′′ ) 0 0 
 xd′′ xd xd′ xd′ xd′′ 
 
[Ce1 ](2×6) = 

 1
−
( xq − xq′ ) (
−
)
xq′ − xq′′ 

0 xq′′
0 0
xq xq′ xq′ xq′′ 


Linearising eq. (10.28), we get

 ∆y d   ∆id 
∆Te = iq0 −id 0    +  −y q0 y d 0   
 ∆y q   ∆iq 

∆ Te(1×1) = [Ce 2 ](1×6) ∆ xe (6×1) + [Ce3 ](1× 2) [Ce1 ]( 2×6) ∆ xe(6×1)

or
∆Te(1×1) = [Ce ](1×6) ∆xe (6×1)(10.32)
where

[Ce 2 ](1×6) = iq0 −id 0 0 0 0 0 

[Ce3 ](1× 2) =  −y q0 y d 0 

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 445 9/12/2017 5:33:23 PM

 446  |  Power System Analysis

[Ce ](1×6) = [Ce 2 ](1×6) + [Ce3 ](1× 2) [Ce1 ]( 2×6)

The above results are used in eq. (10.17).

10.2.3  Linearisation of Electrical Equations

Before taking up the linearisation of the stator differential equations, first consider the linearisation
of the stator voltage components vd and vq.

10.2.3.1  Linearisation of vd and vq

We know that the stator terminal voltage components in the machine frame and the voltage
components in the synchronous frame are related by
vdg  cos d G −sin d G   vDg 
v  = sin d cos d G  v  (10.33)
 qg   G  Qg 
Linearising the above equation we get,

 ∆vdg   ∆vDg 
 ∆v  = [TP 0 ]( 2× 2)   + [ M ]( 2×1) ∆d G(1×1) (10.34)
 qg  ( 2×1)  ∆vQg  ( 2×1)

Notes:
1. Generator slip Sm and its pu speed wG are related by an expression

 w 
Sm =  w G − o  (10.35)
 wB 
Linearising eq. (10.35), we get,
∆Sm = ∆w G (10.36)

2. ∆w G and ∆d G can be expressed as

∆w G (1×1) = [ R10 ](1×12) ∆xsm (12×1) (10.37)

∆d G (1×1) = [ R9 ](1×12) ∆xsm (12×1) (10.38)

where

[ R10 ](1×12) = 0 0 0 0 0 0 0 0 0 1 0 0 

[ R9 ](1×12) = 0 0 0 0 0 0 0 0 1 0 0 0 

After substituting eq. (10.38), (10.34) can be written as

 ∆vdg   ∆vDg 
 ∆v  = [TP 0 ]( 2× 2)   + [ Av ]( 2×122) ∆xsm (12×1) (10.39)
 qg  ( 2×1)  ∆vQg  ( 2×1)

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 446 9/12/2017 5:33:32 PM

 SSR Analysis of the IEEE First Benchmark Model  |  447 

where [TP0](2 × 2) is given by

cos d G 0 − sin d G 0 
 
[TP 0 ]( 2× 2) =  
sin d G 0 cos d G 0 

 −vqg 0 
 
[ M ]( 2×1) =  and [ Av ]( 2×12) = [ M ]( 2×1) [ R9 ](1×12)
vdg 0 
 

10.2.3.2  Linearisation of Stator Equations

Now, linearising eqs (10.20) and (10.21) and expressing in matrix form, we get

 d ∆y d 
 dt   ∆id   ∆y d 
  = [ Rw ]( 2× 2)   + [ ABG ]( 2× 2)   + [ Dw ]( 2×1) ∆wG (1×1)
 d ∆y q   ∆iq  ( 2×1)  ∆y q  ( 2×1)
 dt 
( 2 ×1)

 ∆vdg 
+ [Cw ]( 2× 2)  
 ∆vqg  ( 2×1)
Using eqs (10.31), (10.39), and (10.37), the above equation can be rewritten as
 d ∆y d 
 dt 
  = [ Rw ]( 2× 2) [Ce1 ]( 2×6) ∆xe (6×1) + [ ABG ]( 2× 2) [ Fe ]( 2×6) ∆xe (6×1)
 d ∆y q 
 dt 
( 2 ×1)
  ∆vDg  
+ [Cw ]( 2× 2) [TP 0 ]( 2× 2)   + [ Av ]( 2×12) ∆xsm (12×1) 

  ∆vQg  ( 2×1) 

+ [ Dw ]( 2×1) [ R10 ](1×12) ∆xsm (12×1)

or
 d ∆y d 
 dt   ∆vDg 
  =  Ag1  ∆xe (6×1) + [ Aw ]( 2×12) ∆xsm (12×1) + [TP1 ]( 2× 2)  
 d ∆y q ( 2 × 6 )
 ∆vQg  ( 2×1) (10.40)
 dt 
( 2×1)
where
 −w B Ra 0   0 −w Bw G 0 
   
[ Rw ](2×2) =  [ ABG ](2×2) = 
 0 −w B Ra  w Bw G 0 0 

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 447 9/12/2017 5:33:39 PM

 448  |  Power System Analysis

 −w By q0   −w B 0 
   
[ Dw ](2×1) =  [Cw ](2×2) = 
 w By d 0   0 −w B 

1 0 0 0 0 0 
 
[ Fe ](2×6) = 
0 1 0 0 0 0 

and
 ( 2×6) = [ Rw ]( 2× 2) [Ce1 ]( 2×6) + [ ABG ]( 2× 2) [ Fe ]( 2×6)
 Ag1 

[ Aw ](2×12) = [ Dw ](2×1) [ R10 ](1×12) + [Cw ](2×2) [ Av ](2×12)
[TP1 ](2×2) = [Cw ](2×2) [TP 0 ](2×2)

10.2.3.3  Linearisation of Rotor Equations

Linearising eq. (10.22) to eq. (10.26) and expressing in matrix form, we get
 d ∆y F 
 dt 
 
 d ∆y H 
 dt 
 d ∆y  =  Ag 2  ∆xe (6×1) +  Fg1  ∆E fd (1×1) (10.41)
G ( 4 × 6) ( 4 ×1)
 
 dt 
 d ∆y K 
 
 dt  ( 4 ×1)
where
1 1 
T ′ 0 − T ′ 0 0 0 
 d d 
 
1 1 
 0 0 − 0 0 
 Td′′ Td′′ 
 
 Ag 2  = 
  ( 4 × 6)
 1 1 
0 0 0 − 0 
 Tq′ Tq′ 
 
 
0 1 1
 0 0 0 − 
 Tq′′ Tq′′

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 448 9/12/2017 5:33:40 PM

 SSR Analysis of the IEEE First Benchmark Model  |  449 

 1  xd′  
  
 Td′  xd − xd′  
 
 
 0 
 Fg1  =
  ( 4 ×1) 
 
 0 
 
 
 0 

Combining eqs (10.40) and (10.41), the overall linearised flux equation for the generator is given by
the following matrix equation:

 ∆vDg 
∆xe (6×1) = [ AG ](6×6) ∆xe (6×1) + [ Aws ](6×12) ∆xsm (12×1) + [TPg ](6× 2)   +
 ∆vQg  ( 2×1)

 Fg 
  (6×1) ∆E fd (1×1) (10.42)

where

xe (6×1) = [y d yq yF yH yG y K ]T

 A   [ Aw ]( 2×12) 
  g1  ( 2 × 6 )   
[ AG ](6×6) =



[ Aws ](6×12)) = 
0 
  Ag 2   [ ]
 ( 4 ×12) 
   ( 4 ×6) 

[TP1 ]( 2× 2)  0 
  [ ]( 2×1) 
TPg 
  ( 6 × 2) =   Fg 
  (6×1) = 
[ 0 ]   
 ( 4 × 2)    Fg1  
   ( 4 ×1) 

10.3  EXCITER MODELLING

A single-time constant static excitation system is considered for the generator with KA = 200,
TA = 0.025 s, Efdmax = 6.0, and Efdmin = −6.0 pu.

10.3.1  Single-time Constant Static Exciter

The block diagram representation of the exciter is shown in Figure 10.3.

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 449 9/12/2017 5:33:42 PM

 450  |  Power System Analysis

Vg
Efdmax
−
+ KA
Vref Σ Efdst
1 + sTA
+ Efdmin

VS

Figure 10.3  Single-time Constant Static Excitation System.

The time delay associated with the bus voltage measuring transducer is neglected. The differential
equation is given by
dE fdst

dt
=
1 
TA 
( )
− E fdst + K A Vref + Vs − Vg  (10.43)

After linearising, the above equation can be expressed in matrix form as

(
∆xst (1×1) = [ Ast ](1×1) ∆xst (1×1) + [ Est ](1×1) ∆Vref + ∆Vs )(1×1) + [ Bst ](1×1) ∆Vg (1×1) (10.44)
where
xst (1×1) = E fdst = E fd

and
 1 
[ Ast ](1×1) =  − T 
 A

K 
[ Est ](1×1) =  T A 
 A 

 K 
[ Bst ](1×1) =  − T A 
 A 

10.3.2 Derivation of ∆Vg in terms of vQg and vDg Components

We know that
Vg2 = vQg
2 2 (10.45)
+ vDg

Linearising the above expression, we get

2Vg 0 ∆Vg = 2vQg 0 ∆vQg + 2vDg 0 ∆vDg (10.46)

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 450 9/12/2017 5:33:47 PM

 SSR Analysis of the IEEE First Benchmark Model  |  451 

In matrix form we have

 ∆vDg 
∆Vg (1×1) = [Vtr ](1× 2)   (10.47)
 ∆vQg  ( 2×1)
where

v vQg 0 
[Vtr ](1×2) =  VDg 0 
Vg 0 
 g0

Using eq. (10.47) in eq. (10.44), we get,

 ∆v 
(
∆xst (1×1) = [ Ast ](1×1) ∆xst (1×1) + [ Est ](1×1) ∆Vref + ∆Vs )(1×1) + [ Bst1 ](1×2)  ∆vQg
Dg
 (10.48)
  ( 2×1)

where
[ Bst1 ](1×2) = [ Bst ](1×1) [Vtr ](1×2)
Now, using the exciter state, eq. (10.42) can be rewritten as

 ∆vDg 
∆xe (6×1) = [ AG ](6×6) ∆xe (6×1) + [ Aws ](6×12) ∆xsm (12×1) + [TPg ](6× 2)  
 ∆vQg  ( 2×1)

+  Fg  ∆xst (1×1) (10.49)

( 6 ×1)

10.4 TRANSMISSION NETWORK MODELLING

Figure 10.4 shows the single-line diagram of the series-compensated transmission line.

xE = xT + xL + xSYS Infinite
Pg xT bus
1 2 RL xL xFC 3 xSYS 4

Eb
Vg Eb =1.0∠0°

Figure 10.4  Transmission Line and the Rest of the IEEE FBM System.

The equations for the three-phase AC network are linear and time-invariant. However, to interface
with the generator models, it is desirable to express the equations in the KRON’s (synchronously
rotating) reference frame. These equations can be conveniently derived by first writing the state
equations in the stationary reference-frame, i.e., a–b reference frame. This is obtained by applying
Clarke’s transformation on the set of equations written in the abc-reference-frame. Although the ‘a’

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 451 9/12/2017 5:33:50 PM

 452  |  Power System Analysis

and ‘b ’ networks are identical, it is to be noted that the currents and voltages are different in both
the networks.

10.4.1  Derivation of Capacitor Voltage Equations

Consider the a-sequence network of series-compensated transmission line shown in Figure 10.5. ia
is the a-sequence component of the generator armature current and eba is the a-sequence component
of the infinite bus voltage.

R L C
+
ia vCa
vga eba

Figure 10.5  a-sequence Network.

Referring to Figure 10.5, the only state variable is vCa and the state equation is

dvCa 1
= ia (10.50)
dt C
Similarly, in the b-sequence network,

dvCb 1
= ib (10.51)
dt C
Writing eqs (10.50) and (10.51) in matrix form, we get

vCa  ia 
v  = [C1 ]( 2× 2)   (10.52)
 C b  ( 2×1) ib  ( 2×1)
where

1 
C 0
  1
[C1 ](2×2) =  and C =
 x FC w B
1
0 
 C

10.4.1.1  Derivation in DQ-Reference-Frame

We know that,
 fD   fa 
 f  = [TK ]  f  (10.53)
 Q  b

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 452 9/12/2017 5:33:54 PM

 SSR Analysis of the IEEE First Benchmark Model  |  453 

or
 fa   fD 
 f  = [TK ]
T
 f  (10.54)
 b  Q
where
cos w t −sin w 0 t 
[TK ] = sin w 0t cos w 0 t 
 0

Using the relation (10.54), in eq. (10.52), we get

d vCD   iD 
 [TK ] ( 2× 2)   = [C1 ]( 2× 2) [TK ] ( 2× 2)  
T T
v  (10.55)
dt   CQ  ( 2×1)  iQ  ( 2×1)

v  d vCD  iD 

[TK ]T (2×2) vCD  + [TK ] ( 2× 2) 
T
 = [C1 ]( 2× 2) [TK ] ( 2× 2)  
T
 CQ  ( 2 ×1) dt vCQ ( 2×1) iQ  ( 2×1) (10.56)

Pre-multiplying both sides of the above equation by [TK](2 × 2) and making use of the identities,

[TK ][TK ]T = [ I ](10.57)

 0 w0 
d  
[TK ] dt [TK ] = 
T
 (10.58)
 −w 0 0 

[TK ][C1 ][TK ]T = [C1 ](10.59)

we then have,
0 w0 
vCD    vCD  iD 
v  +  v  = [C1 ]( 2× 2)   (10.60)
 CQ  ( 2×1)  −w  CQ  ( 2×1) i
  ( 2×1)
Q
or  0 0  ( 2× 2)

0 −w 0 
vCD    vCD  iD 
v  =  v  + [C1 ]( 2× 2)   (10.61)
 CQ  ( 2×1) w  CQ  ( 2 ×1)  iQ 
 0 0   ( 2 ×1)
( 2 × 2)

Let xN ( 2 ×1) = [vCD vCQ ]T. The above equation can be rewritten as

0 −w 0 
  i 

x N ( 2×1) =   xN ( 2 ×1) + [C1 ](2×2) iD 
w 0 0   Q  ( 2×1) (10.62)
( 2 × 2)
This result is identical to that obtained in chapter 3.

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 453 9/12/2017 5:34:02 PM

 454  |  Power System Analysis

10.4.1.2  Linearisation of iD and iQ

It can be seen from eqs (10.29) and (10.30) that the generator currents are expressed as a function
of state variables. Since this current flows through the line inductor (see eq. (10.62)), the inductor
current cannot be considered as additional state variables. Even if expressed, it leads to redundant
states. As per their representation, the only difference is that the generator currents, id and iq, are in
the ‘machine-­reference-frame’, whereas the inductor currents, iD and iQ are in the ‘synchronous-ref-
erence-frame’. In eq. (10.31), the generator currents in the ‘machine-reference-frame’ are linearised.
Now, we must linearise iD and iQ. This is done as follows:
id  cos d G − sin d G  iD 
i  = sin d cos d G  i  (10.63)
 q  G Q
or

iD   cos d G sin d G  id 

i  =  − sin d cos d G  i  (10.64)
Q  G  q
Linearising eq. (10.64), we get
 ∆iD   ∆id 
 ∆i  = [TP 0 ]T ( 2× 2)   + [ N ]( 2×1) ∆d G (1×1) (10.65)
 Q  ( 2×1)  ∆iq  ( 2×1)
Substituting eqs (10.31) and (10.38) in eq. (10.65), we then have,
 ∆iD 
 ∆i  = [TP 0 ]T ( 2× 2) [Ce1 ]( 2×6) ∆x e(6×1) + [ N1 ]( 2×12) ∆xsm (12×1) (10.66)
 Q  ( 2×1)
where
cos d G 0 − sin d G 0 
 
[TP 0 ]( 2× 2) = 
sin d G 0 cos d G 0 
 iQ 0 
[ N ]( 2×1) =  
 −iD 0 
and
[ N1 ]( 2×12) = [ N ]( 2×1) [ R9 ](1×12)

10.4.1.3  Linearisation of Capacitor Voltage Equations

Linearising eq. (10.62) and using eq. (10.66), we get,

∆x N ( 2×1) = [ AN ]( 2× 2) ∆xN ( 2 ×1) + [C2 ](2×6) ∆xe (6×1) + [C3 ](2×12) ∆xsm (12×1) (10.67)
where
0 −w 0 
 
[ AN ](2×2) = 
w 0 0 

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 454 9/12/2017 5:34:09 PM

 SSR Analysis of the IEEE First Benchmark Model  |  455 

and

[C2 ](2×6) = [C1 ](2×2) [TP 0 ]T (2×2) [Ce1 ](2×6)

[C3 ](2×12) = [C1 ](2×2) [ N1 ](2×12)

10.4.2  Derivation of Line Voltage Equations

Applying KVL to the system depicted in Figure 10.5, the following equation can be written
dia
v ga = Ria + L + vCa + eba (10.68)
dt
Similarly, in the b-sequence network, we have
dib
v g b = Rib + L
+ vC b + ebb (10.69)
dt
The above equations can be expressed in matrix form as

v ga  ia  d ia  vCa  eba 

v  = [ R ]( 2× 2)   + [ L ]( 2× 2)   +  + 
 g b  ( 2×1) ib  ( 2×1) dt ib  ( 2×1) vC b  ( 2×1) ebb  ( 2×1) (10.70)

where

R 0  L 0 
[ R](2×2) = 0 R 
[ L ](2×2) = 0L 
 

and
xE
L= ; x E = xT + x L + xSYS
wB

10.4.2.1  Derivation in DQ-Reference-Frame

Using the relation (10.54), in eq. (10.70), we get,

v   iD  
[TK ]T (2×2) vDg  = [ R ]( 2× 2)  [TK ] ( 2× 2)  
T

 Qg  ( 2 ×1)  iQ  ( 2××1) 

d iD  
 [TK ] ( 2× 2)  
T
+ [ L ]( 2× 2) 
dt  iQ  ( 2×1) 

vCD   EbD 
+ [TK ] + [TK ] ( 2× 2) 
T T
( 2 × 2) v  
 CQ  ( 2×1)  EbQ  ( 2×1) (10.71)

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 455 9/12/2017 5:34:31 PM

 456  |  Power System Analysis

or
v  iD  d iD 
[TK ]T vDg  = [ R ]( 2× 2) [TK ]
T
( 2 × 2) i  + [ L ]( 2× 2) [TK ] ( 2× 2)  
T
 Qg  ( 2×1)  Q  ( 2×1) dt iQ  ( 2×1)

d iD 
+ [ L ]( 2× 2) [TK ] ( 2× 2)  
T
dt iQ  ( 2×1)

vCD   EbD 
+ [TK ] + [TK ] ( 2× 2) 
T T
( 2 × 2) v   (10.72)
 CQ  ( 2×1)  EbQ  ( 2×1)

Pre-multiplying both sides of the above equation by [TK](2 × 2) and making use of the identities,

[TK ][TK ]T = [ I ](10.73)

[TK ][ R][TK ]T = [ R](10.74)
[TK ][ L ][TK ]T = [ L ](10.75)
we then have,
vDg  iD  d iD 
v  = [ R ]( 2× 2)   + [ L ]( 2× 2)  
 Qg  ( 2×1) i
 Q  ( 2×1) dt iQ  ( 2×1)

d iD 
+ [TK ]( 2× 2) [ L ]( 2× 2) [TK ] ( 2× 2)  
T
dt iQ  ( 2×1)

vCD   EbD 
+  +  (10.76)
vCQ  ( 2×1)  EbQ  ( 2×1)

d
Consider the term [TK ]( 2× 2) [ L ]( 2× 2) [TK ]T (2×2) :
dt

d d dq 0
[TK ](2×2) [ L ](2×2) dt [TK ]T (2×2) = [TK ](2×2) [ L ](2×2) dq [TK ]T (2×2) dt (10.77)
0

where q 0 = w 0 t.
Hence, we have

d  − sin w t coosw 0 t 
[TK ](2×2) [ L ](2×2) dt [TK ]T (2×2) = [TK ](2×2) [ L ](2×2)  − cos w0 t w
− sinw 0 t  0
 0

 0 w 0 L
=

 −w 0 L 0  (10.78)

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 456 9/12/2017 5:34:40 PM

 SSR Analysis of the IEEE First Benchmark Model  |  457 

Using eq. (10.78) in eq. (10.76), we get

vDg  iD  d iD  vCD   EbD 

v  = [ F1 ]( 2× 2)   + [ F2 ]( 2× 2)   +  + 
 Qg  ( 2×1) i
 Q  ( 2×1) dt  Q  ( 2×1)  CQ  ( 2×1)  EbQ  ( 2×1) (10.79)
i v

where

 R w 0 L L 0 
[ F1 ](2×2) =  −w R 
[ F2 ](2×2) = 0 L 
 0L 

Note that eq. (10.79) is identical to that derived in chapter 3 starting from abc-frame.

10.4.2.2  Linearisation of the Line Voltage Equations

Before linearising eq. (10.79), it is to be noted that the infinite bus voltage Eb can be expressed as

Eb = EbQ + jEbD

and since the infinite bus voltage is taken as the reference ( i.e., Eb = Eb ∠0 ), we have, EbD = 0 and
EbQ = Eb which is assumed to be a constant.
Now, linearising eq. (10.79), we get
 ∆vDg   ∆iD  d  ∆iD   ∆vCD 
 ∆v  = [ F1 ]( 2× 2)   + [ F2 ]( 2× 2)   + 
 Qg  ( 2×1)  ∆iQ  ( 2×1) dt  ∆iQ  ( 2×1)  ∆vCQ  ( 2×1) (10.80)

10.5  LISTING OF LINEARISED STATE EQUATIONS

In the following lines, all state equations are listed. The state vector is given by
T
xnp ( 21×1) = [ xsm (12×1) xe (6×1) xst (1×1) xN ( 2 ×1) ]

Note that in the above set, the turbine and speed governor are not modelled.
Mechanical system of equations:
From eq. (10.17), we have,

∆xsm (12×1) = [ Asm ](12×12) ∆xsm (12×1) − [ DTe1 ](12×6) ∆xe (6×1) (10.81)

Generator electrical equations:

From eq. (10.49), we have,
 ∆vDg 
∆xe (6×1) = [ AG ](6×6) ∆xe (6×1) + [ Aws ](6×12) ∆xsm (12×1) + [TPg ](6× 2)  
 ∆vQg  ( 2×1)

+  Fg  ∆xst (1×1) (10.82)

(6 ×1)

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 457 9/12/2017 5:34:43 PM

 458  |  Power System Analysis

Exciter equations:
From eq. (10.48), we have,
 ∆v 
(
∆xst (1×1) = [ Ast ](1×1) ∆xst (1×1) + [ Est ](1×1) ∆Vref + ∆Vs )(1×1) + [ Bst1 ](1×2)  ∆vQg
Dg
 (10.83)
  ( 2×1)

Line capacitor equations:

From eq. (10.67), we have,

∆x N ( 2 ×1) = [ AN ](2×2) ∆xN (2×1) + [C2 ](2×6) ∆xe (6×1) + [C3 ](2×12) ∆xsm (12×1) (10.84)
The overall linearised system model in the state-space form is obtained by combining the above listed
equations to get

∆xnp ( 21×1) =  Anp 

( 21× 21)
∆xnp ( 21×1) +  Enp 
( 21×1)
( ∆Vref + ∆Vs )(1×1)
 ∆vDg 
+[ Bnp ]( 21× 2)   (10.85)
 ∆vQg  ( 2×1)
where

[ Asm ](12×12) − [ DTe1 ](12×6) [0](12×1) [0](12× 2) 

 
[ Aws ](6×12) [ AG ](6×6) [ Fg ](6×1) [0](6× 2) 
 Anp 
  ( 21× 21) = 
[0 ] [0](1×6) [ Ast ](1×1) [0](1× 2) 
 (1×12) 
[C3 ]( 2×12) [C2 ]( 2×6) [0]( 2×1) [ AN ]( 2× 2) 

[ 0 ](12× 2)  [ 0 ](12×1) 
   
   
  [ 0 ] 
 T 
  Pg  (6× 2)   (6×1) 
[ Bnp ]( 21× 2) =   [E ]
 np ( 21×1) =
 
 
[ B ]
st1 (1× 2)
 [ Est ](1×1) 
   
   
 
[ 0 ]( 2× 2) 
[ 0 ]
 ( 2×1) 


10.5.1  Interfacing of the Network to the Generator

Rewriting eq. (10.80), we have

 ∆vDg   ∆iD  d  ∆iD   ∆vCD 

 ∆v  = [ F1 ]( 2× 2)   + [ F2 ]( 2× 2)   + 
 Qg  ( 2×1)  ∆iQ  ( 2×1) dt  ∆iQ  ( 2×1)  ∆vCQ  ( 2×1) (10.86)

In order to remove the redundant states associated with the line inductor, a procedure described below
is employed:

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 458 9/12/2017 5:34:50 PM

 SSR Analysis of the IEEE First Benchmark Model  |  459 

Define the output variable vector as,

 ∆iD 
 ∆i 
 Q 
 ∆iD  (10.87)
y(6×1) =  
 ∆iQ 
 ∆vCD 
 
 ∆vCQ 

The above variable set is chosen so that the generator voltage components given in eq. (10.86) are
expressed only in terms of the state variables.
Using eq. (10.87) in eq. (10.86), we can write,

 ∆vDg 
 ∆v  = [ F ]( 2×6) y(6×1) (10.88)

 Qg  ( 2×1)
where

[ F ](2×6) = [[ F1 ](2×2) [ F2 ]( 2× 2) [ I 2 ]( 2× 2) ]

Now, let us attempt to express each of the variables in y(6×1) in terms of the state vector, x np( 21×1).

10.5.1.1  Variables ∆iD and ∆iQ

Reproducing eq. (10.66), we have,
 ∆iD 
 ∆i  = [TP 0 ]T ( 2× 2) [Ce1 ]( 2×6) ∆xe (6×1) + [ N1 ]( 2×12) ∆xsm (12×1) (10.89)
 Q  ( 2×1)
or

 ∆iD 
 ∆i  = [TP 0 ]T ( 2× 2) [Ce1 ]( 2×6) [ F3 ](6× 21) ∆xnp ( 21×1)) + [ N1 ]( 2×12) [ F4 ](12× 21) ∆xnp ( 21×1)
 Q  ( 2×1)

or
 ∆iD 
 ∆i  = C DQi  ∆xnp ( 21×1) (10.90)
( 2 × 21)
 Q  ( 2×1)
where

 ( 2× 21) = [TP 0 ] ( 2× 2) [Ce1 ]( 2×6) [ F3 ](6× 21) + [ N1 ]( 2×12) [ F4 ](12× 21)

C DQi  T


M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 459 9/12/2017 5:34:53 PM

 460  |  Power System Analysis

and

[ F3 ](6×21) = [[0](6×12) [ I 6 ](6×6) [0](6×1) [0](6× 2) ]

[ F4 ](12×21) = [[ I12 ](12×12) [0](12×6) [0](12×1) [0](12× 2) ]

10.5.1.2  Derivative of Variables ∆iD and ∆iQ

From eq. (10.90), we have,

d  ∆iD 
  = C DQi  ∆ x np( 21×1) (10.91)
dt  ∆iQ  ( 2×1)  ( 2 × 21)

Using eq. (10.85) in the above equation, we get,

d  ∆iD 
  = C DQi  [A ] ∆x np( 21×1) + C DQi 
dt  ∆iQ  ( 2×1)  ( 2 × 21) np ( 21× 21) ( 2 × 21)

 ∆vDg  (10.92)
[ Bnp ]( 21× 2)  
 ∆vQg  ( 2×1)
Note that ∆Vref and ∆Vs are set to zero for ease of analysis.

10.5.1.3  Variables ∆tCD and ∆tCQ

Since xN ( 2 ×1) = [vCD vCQ ]T , we have,

 ∆vCD 
 ∆v  = C DQv  ∆xnp ( 21×1) (10.93)
( 2 × 21)
 CQ  ( 2×1)
where
C DQv   [ I 2 ]( 2× 2) 
  ( 2× 21) = [0]( 2×12) [0]( 2×6) [0]( 2×1)
( 2 × 21) (10.94)

Now, using eqs (10.90), (10.92), and (10.93) in eq. (10.87), the y(6×1) matrix can written as

 ∆vDg 
y(6×1) = [C y ](6× 21) ∆xnp ( 21×1) + [ D y ](6× 2)  
(10.95)
 ∆vQg  ( 2×1)
where
 C DQi  
   ( 2× 21) 
 [0](2×2) 
   
   
 
C y  =  C DQi ( 2× 21) [ Anp ]( 21× 21)   Dy  =  C DQi  [B ]
  (6× 21)     ( 6 × 2) ( 2 × 21) np ( 21× 2) 
   
   
 C DQv    [ 0 ]( 2× 2) 
   ( 2× 21)   

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 460 9/12/2017 5:34:57 PM

 SSR Analysis of the IEEE First Benchmark Model  |  461 

Now, using eq. (10.88) in eq. (10.95), we get,

y(6×1) = [C y ](6× 21) ∆xnp ( 21×1) + [ D y ](6× 2) [ F ]( 2×6) y(6×1) (10.96)

or
y(6×1) = [ FT ](6× 21) ∆xnp ( 21×1) (10.97)
where
(
[ FT ](6× 21) = [ I 6 ](6×6) − [ D y ](6× 2) [ F ]( 2×6) )−1 [C y ](6×21)
Substituting eqs (10.88) and (10.97) in eq. (10.85), we get,

∆ x np( 21×1) = [ AT ]( 21× 21) ∆xnp ( 21×1) +  Enp 

( 21×1)
( ∆Vref + ∆Vs )(1×1) (10.98)
where
[ AT ]( 21× 21) = [ Anp ]( 21× 21) + [ Bnp ]( 21× 2) [ F ]( 2×6) [ FT ](6× 21)

The eigenvalues of [AT] are used to determine the stability of the system.

Note: If a power system stabiliser is to be enabled, a state model is derived first for the chosen PSS.
This model is interfaced to eq. (10.85) in the usual manner as a ∆Vs signal. The network is then
interfaced to the model as described above.

10.6 MACHINE-FRAME-BASED TIME-DOMAIN SIMULATION OF SSR

Time–domain tools basically solve the linear or non-linear system of equations numerically, employ-
ing numerical techniques to provide solution to each variable at regular intervals. This method of
analysis has exceptional ability to handle mathematical models of various components of a power sys-
tem to any desired degree of detail. Time-domain simulation techniques are also helpful in validating
the inferences made out from eigenvalue analysis. In this method, simulation is carried out using the
equation expressed in the machine-reference-frame. The set of equations to be solved is listed below.
1. Mechanical systems: Here, the set of differential eqs (10.1) to (10.12) are solved. TmH, TmI, TmLA,
and TmLB are set to constant, signifying disabling of prime-mover controllers. The electromag-
netic torque Te is computed using eq. (10.28).
2. Generator electrical systems: Here, the complete set from eq. (10.20) to eq. (10.27) is used.
3. Exciter system: The differential eq. (10.43) is solved using the generator terminal voltage Vg.
4. Capacitor voltage equation: Here, eq. (10.62) is solved using iD and iQ.
5. The line voltage which involves the computation of vgd and vgq are handled as mentioned below.

10.6.1  Computation of vgd and vgq in Simulation

While solving for yd and yq using eqs (10.20) and (10.21), vgd and vgq values are required. These are
obtained using the network-interfacing equations, given by,
d
vabcg = [ R]iabc + [ L] iabc + vCabc + eabc (10.99)
dt

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 461 9/12/2017 5:35:04 PM

 462  |  Power System Analysis

Writing the above equation in 0dq-coordinate system, i.e., machine-reference-frame, and neglecting
0-sequence quantities, we get,
 xE 
0
vdg   R w G x E  id   w B  d id 
v  =  −w x + 
 qg   G E R  iq   x E  dt iq 
0 w B 


cos d G −sin d G   vCD  0 

+ +

sin d G cos d G   vCQ   Eb  (10.100)

where wG is the per-unit speed of the generator.

Expressing the above equation in terms of known variables, we get,

vdg  id   0  
v  = [ N s1 ]( 2× 2)   + [TP ]( 2× 2)  xN ( 2×1) +   
 qg  ( 2×1) iq  ( 2×1)   Eb  
( 2×1) 

d id 
+ [ N s2 ]( 2× 2)   (10.101)
dt iq  ( 2×1)
where

 xE 
w 0 
 R w G xE 
 B 
[ s1 ](2×2) 
N =

 [ N s2 ](2×2) = 
 −w G x E  
R  0 xE 
 w B 

cos d G − sin d G 
 
[TP ](2×2) = 
sin d G cos d G 

Note that vCD and vCQ are obtained by solving x N . From eq. (10.62), we have,

1 
−w 0  vCD   C 0 i
vCD   0  D
v  = w +
0  vCQ  
   (10.102)
1  iQ 
 CQ   0 0
 C 
where

iD   cos d G sin d G  id 

i  =  − sin d cos d G  i  (10.103)
Q  G  q

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 462 9/12/2017 5:35:08 PM

 SSR Analysis of the IEEE First Benchmark Model  |  463 

di di
Also, note that, the interfacing equation needs derivation of d and q . From eqs (10.29) and
(10.30), we have dt dt

did 1 dy d ( xd′ − xd′′ ) dy H ( xd − xd′ ) dy F (10.104)

= − −
dt xd′′ dt xd′ xd′′ dt xd xd′ dt

diq
=
1 dy q
−
( )
xq′ − xq′′ dy K
−
(
xq − xq′ dy G )
(10.105)
dt xq′′ dt xq′ xq′′ dt xq xq′ dt

did diq
10.6.1.1 Derivation of Expressions for and
dt dt
Substituting for y d , y F , y H from eqs (10.20), (10.22), and (10.23) in eq. (10.104), we get

1 ( x ′ − xd′′ ) −y + y
id =
xd′′
 − Raw B id − wG w B y q − w B vdg  − d
  Td′′xd′ xd′′
[ H d]

−
( xd − xd′ )  −y +yd +
xd′ 
E fd 
 F (10.106)
Td′ xd xd′  ( xd − xd′ ) 

Substituting for y q , y G , y K from eqs (10.21), (10.25) and (10.26) in eq. (10.105)

iq =
1
 − Raw B iq + wG w By d − w B vqg  −
xq′ − xq′′

( 
)
xq′′   T ′′x ′ x ′′  −y K + y q 
q q q

−
( xq − xq′ )  −y + y q 
Tq′ xq xq′ 
G (10.107)

The above equations are expressed in matrix form as

d id  id  vdg 

i  = [ M s1 ]( 2× 2)   + [ M s2 ]( 2× 2)   + [ M s3 ]( 2×6) xe (6×1)
dt  q  ( 2×1) iq  ( 2×1) vqg  ( 2×1)

+ [ M s 4 ]( 2×1) E fd (1×1) (10.108)

where
 − Raw B   −w B   −1 
 0   0  T ′ x 
 xd′′   xd′′   d d
[ M s1 ](2×2) =



[ M s2 ](2×2) =



[ M s4 ](2×1) = 
− Raw B  −w B   0 
 0  0  
 xq′′   xq′′   
 

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 463 9/12/2017 5:35:29 PM

 464  |  Power System Analysis

[Ms3](2 × 6) is given by

 −( xd′ − xd′′ ) ( xd − xd′ ) −wG w B ( xd − xd′ ) ( xd′ − xd′′ ) 

 − 0 0 
 Td′′xd′ xd′′ Td′ xd xd′ xd′′ Td′ xd xd′ Td′′xd′ xd′′ 
[ M s3 ] = 
 wGw B −
(
xq′ − xq′′ ) (
−
xq − xq′ ) 0 0
( xq − xq′ ) ( )
xq′ − xq′′ 

 xq′′ Tq′′xq′ xq′′ Tq′ xq xq′ Tq′ xq xq′ Tq′′xq′ xq′′ 


 did 
 dt 
Substituting eq. (10.101) in eq. (10.108) and solving for   , we get
 diq 
 dt 

id   i 
  (
= [ I 2 ]( 2× 2) − [ M s6 ]( 2× 2) )−1  [ M s5 ](2×2) idq  + [ M s3 ]( 2×6) xe (6×1)
iq  ( 2×1)    ( 2×1)

 0  
+[ M s 4 ]( 2×1) E fd (1×1) + [ M s7 ]( 2× 2)  xN ( 2 ×1) +   
  Eb   (10.109)
( 2 ×1)  

where

[ M s5 ](2×2) = [ M s1 ](2×2) + [ M s2 ](2×2) [ N s1 ](2×2)

[ M s6 ](2×2) = [ M s2 ](2×2) [ N s2 ](2×2)
[ M s7 ](2×2) = [ M s2 ](2×2) [TP ](2×2)
d id 
Having evaluated   using eq. (10.109), in terms of state-variables, the generator terminal
dt iq 
voltage is obtained from eq. (10.101).

10.6.2  Initial Condition Calculations

The following data is given:
1. Real power output of the generator, Pg0.
2. Generator terminal voltage, Vg0 = 1.0 pu.
3. Infinite bus voltage Eb = 1.0 pu and is taken as reference.
4. For the turbine FHP = 0.3, FIP = 0.3, FLPA = 0.2, and FLPB = 0.2.
5. Base speed, wB = 1.0 pu and nominal speed, w0 = 1.0 pu.
6. The series capacitive reactance xFC in per-unit.

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 464 9/12/2017 5:35:33 PM

 SSR Analysis of the IEEE First Benchmark Model  |  465 

Using these values, the initial conditions of states variables are calculated as follows [1]:
1. Compute

Z n = RL + j ( x E − x FC ) (10.110)

= Z n ∠b (10.111)

Vg 0 cos b Pg 0 Z n 
q g 0 = cos −1  −  − b (10.112)
 Eb Vg 0 Eb 
2. Compute

  Vg 0 = Vg 0 (cos q g 0 + j sin q g 0 ) (10.113)

 Vg 0 − Eb 
  I g 0 =  Z  = I g 0 ∠f 0 (10.114)
 n 
( )
Eq0 = Vg 0 + Ra + jxq I g 0 (10.115)

d 0 = ∠Eq0 = d G 0 (10.116)
3. Compute

iq0 + jid 0 = I g 0 e − jd 0

= I g0 ∠(f 0 − d 0 )

iq0 = I g 0 cos(f 0 − d 0 ) (10.117)

id 0 = I g 0 sin(f 0 − d 0 ) (10.118)

4. Compute
vq0 + jvd 0 = Vg 0 e − jd 0

= Vg 0 ∠(q g 0 − d 0 )

vq0 = Vg 0 cos(q g 0 − d 0 ) (10.119)

vd 0 = Vg 0 sin(q g 0 − d 0 ) (10.120)

5. Compute
E fd 0 = Eq0 − ( xd − xq )id 0 (10.121)

6. Compute
y d 0 = xd id 0 + E fd 0 (10.122)

y q0 = xq iq0 (10.123)

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 465 9/12/2017 5:35:45 PM

 466  |  Power System Analysis

7. Compute

y H 0 = y d 0 (10.124)

xd′
y F0 = y d0 + E fd 0 (10.125)
xd − xd′

y K 0 = y q0 (10.126)

y G 0 = y q0 (10.127)

8. Compute
w0
wG 0 = (10.128)
wB
w E0 = wG 0 (10.129)

w LB 0 = wG 0 (10.130)

w LA0 = wG 0 (10.131)

w I 0 = wG 0 (10.132)

w H 0 = wG 0 (10.133)

9. Compute

Te 0 = y d 0 iq0 −y q0 id 0 (10.134)

10. Compute

Tm0 = Te 0 + DH w H 0 + DI w I 0 + DLAw LA0 + DLBw LB 0 + DG wG 0 + DE w E 0 (10.135)

TmH 0 = FHP Tm0

TmI 0 = FIP Tm0

TmLA0 = FLPATm0

TmLB 0 = FLPBTm0

11. Compute
DE
d E0 = d 0 − w E0
KGE

1
d LB 0 = d 0 + Te 0 + KGE (d 0 − d E 0 ) + DG wG 0 
K LBG 

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 466 9/12/2017 5:35:53 PM

 SSR Analysis of the IEEE First Benchmark Model  |  467 

1
d LA0 = d LB 0 +  −TmLB 0 + K LBG (d LB 0 − d 0 ) + DLBw LB 0 
K LALB 

1
d I 0 = d LA0 +  −TmLA0 + K LALB (d LA0 − d LB 0 ) + DLAw LA0 
K ILA 

1
d H0 = d I0 +  −TmI 0 + K ILA (d I 0 − d LA0 ) + DI w I 0 
K HI 

12. Calculate Vref

E fd 0
Vref = + Vg 0 (10.136)
KA

13. For the network,

VC 0 = − jxC I g 0 = VC 0 ∠f C

vCQ 0 = VC 0 cosf C (10.137)

vCD 0 = VC 0 sinf C (10.138)

10.6.3  Case Studies and Modal-speed Calculations

Since there are six rotor masses, the IEEE FBM has five torsional modes. Each of the torsional modes
has its largest SSR interaction at a certain value of the series compensation level xFC. This means that
torsional interaction is a discrete event. Therefore, any compensation level can be used unless the net-
work sub-synchronous mode does not coincide with any of the torsional mode frequency. By repeated
runs of the eigenvalue programme with different xFC, the level of series compensation associated
with the maximum torsional interaction for a given mode is determined. The results are tabulated in
Table 10.4 for all torsional modes.

Table 10.4 Frequency of Torsional Modes and the Capacitive Reactance for the Maximum
Torsional Interaction.
Mode Frequency (Hz) xFC (pu)
Torsional mode-4 32.285 0.18
Torsional mode-3 25.54 0.29
Torsional mode-2 20.21 0.38
Torsional mode-1 15.74 0.44

At very low compensation level, say xFC = 0.038, even though the frequency of sub-synchronous
mode coincides with that of the torsional mode-5; mode-5 damping remains unaltered due to its high
value of modal inertia. This implies that mode-5 cannot be controlled by any means.

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 467 9/12/2017 5:35:58 PM

 468  |  Power System Analysis

An operating condition is chosen with Pg0 = 0.5 pu, and FSC compensation set to xFC = 0.3.
Eigenvalues for this case are listed in column-1 of Table 10.5 where all oscillatory modes are listed.
As the frequency of the sub-synchronous electrical mode (also known as sub-synchronous network
mode) does not lie in the vicinity of any of the torsional mode frequency, none of the torsional modes
are destabilised.

Table 10.5  Eigenvalues for the IEEE FBM system with Pg0 = 0.5 pu.

xFC = 0.3 xFC = 0 .29 Comments

−4.693 ± j598.87 −4.6896 ± j595.14 Super-synchronous mode

−3.2554 ± j156 −4.18 ± j159.34 Sub-synchronous mode

−1.8504 ± j298.17 −1.8504 ± j298.17 Torsional mode-5

−0.3588 ± j202.82 −0.35759 ± j202.8 Torsional mode-4

−0.13808 ± j159.7 0.71369 ± j160.11 Torsional mode-3

−0.071561 ± j127.08 −0.072362 ± j127.08 Torsional mode-2

−0.2341 ± j99.522 −0.23436 ± j99.489 Torsional mode-1

−0.90836 ± j10.352 −0.88913 ± j10.27 Swing mode (mode-0)

−19.156 ± j23.702 −19.104 ± j23.848 Exciter mode

For xFC = 0.29, the eigenvalues are listed in column-2 of Table 10.5. It can be clearly seen that as the
sub-synchronous electrical mode frequency coincides with that of torsional mode-3, the torsional
mode becomes unstable due to torsional interaction (TI). For this condition, all other torsional modes
are stable.

10.6.3.1  Modal-speed Calculations

While running the corresponding time-domain simulation, it is found that the following difficulties
are generally faced:
1. The stability inferences made out of the eigenvalue analysis about any particular mode is very
difficult to infer from the time-domain simulation as a time-domain response is generally made
up of many modes associated with the system.
2. As an extension to the determination of eigenvalues, the participation factor -based analysis pro-
vides information about a dominant state variable with respect to a mode and by observing such a
state variable in the time-domain simulation, the stability information of that particular mode can
be inferred. However, such an analysis is found to be effective in the power swing (i.e., mode-0)
oscillation studies, and they fail in the analysis of the stiff systems such as SSR studies, especially
when multiple modes are unstable.
In this connection, the mode identification using modal-speed calculation [4] is found to be
very effective in time-domain simulations. Further, the modal-speed deviation can also be used as
a control signal for the supplementary controllers to damp the torsional oscillations [5, 6].

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 468 9/12/2017 5:35:58 PM

 SSR Analysis of the IEEE First Benchmark Model  |  469 

In the time-domain simulation process, the modal-speed deviation ∆w Ml , corresponding to

the lth mode, is approximately obtained as follows:
∆w Ml = wl × [ ∆w H , ∆w I , ∆w LA , ∆w LB , ∆w G , ∆w E ]T
where,
• wl is a row vector containing the left-eigenvectors corresponding to the lth mode and is obtained
from eq. (10.146).
• ∆w H , ∆w I , ∆w LA , ∆w LB , ∆w G , and ∆w E represent the individual angular speed devia-
tions of the rotor masses of the turbine-generator system and these are obtained from simulation.

Such computation of the modal speed is done based on the observation that when state values are
weighted with the left eigenvectors, it provides the modal contribution due to states. It is to be noted
that these eigenvectors are obtained corresponding to the unconnected mechanical system, neglecting
damping. This is based on the fact that the torsional mode frequencies of the entire connected system
do not significantly change from that of the unconnected mechanical system. Using eq. (10.1) to
eq. (10.12), the speed equations are rewritten such that speed variables are replaced by angle variables,
and in addition all damping components are neglected. Then we get,
For HP turbine:
 2 H H d 2d H 
 2
+ K HI d H − K HI d I  = TmH (10.139)
 w B dt 
For IP turbine:
 2 H I d 2d I 
 2
− K HI d H + ( K HI + K ILA )d I − K ILAd LA  = TmI (10.140)
 w B dt 
For LPA turbine:
 2 H LA d 2d LA 
 2
− K ILAd I + ( K ILA + K LALB )d LA − K LALBd LB  = TmLA (10.141)
 w B dt 
For LPB turbine:
 2 H LB d 2d LB 
 2
− K LALBd LA + ( K LALB + K LBG )d LB − K LBGd G  = TmLB (10.142)
 w B dt 
For generator:
 2 H G d 2d G 
 2
− K LBGd LB + ( K LBG + KGE )d G − KGEd E  = −Te (10.143)
 w B dt 
and for exciter:
 2 H E d 2d E 
 2
− KGEd G + KGEd E  = 0 (10.144)
 w B dt 

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 469 9/12/2017 5:36:02 PM

 470  |  Power System Analysis

The eqs. (10.139) to (10.144) are combined and written in matrix form as,

[ M ] p2d + [ K ]d = [Tm ] − [Te ] = [T ] (10.145)

where
d = [d H , d I , d LA , d LB , d G , d E ]T

From the knowledge of a single spring-mass system, it can be stated that the eigenvalue of the [M]−1[K]
matrix provides the square value of the natural frequencies of the unconnected mechanical system.
The left eigenvector matrix pertaining to this matrix is used for the estimation of the modal speed. The
left eigenvector matrix for the IEEE FBM is given below:

 −0.41619 0.88526 −0.55343 0.10622 −0.022043 0.00018403 Mode − 5

 −0.084486 0.0071565 0.45448 −0.93096 0.56738 −0.013573  Mode − 4
 
 0.54936 0.31482 −11.1665 −0.49902 0.85243 −0.051087  Mode − 3 
W = 
 −0.078626 −0.13169 −0.72676 −0.74838 −0.73508 −0.02896  Mode − 0
 0.28569 0.35943 1.1636 −0.39084 −1.2825 −0.13543  Mode − 1
 
0.26583 0.26191 0.33547 −0.90943 −0.8448 0.89102  Mode − 2
(10.146)

For xFC = 0.3 pu and xFC = 0.29 pu, the modal speeds are plotted in Figures 10.6 and 10.7 respectively,
that validate the eigen-inferences. Here, the time-domain simulation is carried out for a reduction in
the torque input of all turbines by 0.01 pu, at t = 0.3 s, lasting for 0.01 s.
0.01
Mode−5

−0.01
0 0.5 1 1.5 2 2.5 3
0.05
Mode−4

−0.05
0 0.5 1 1.5 2 2.5 3
0.2
Mode−3

−0.2
0 0.5 1 1.5 2 2.5 3
0.1
Mode−2

−0.1
0 0.5 1 1.5 2 2.5 3
0.2
Mode−1

−0.2
0 0.5 1 1.5 2 2.5 3
0.1
Mode−0

−0.1
0 0.5 1 1.5 2 2.5 3
Time (s )

Figure 10.6  The IEEE FBM: Modal-speeds for xFC = 0.3 pu.

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 470 9/12/2017 5:36:04 PM

 SSR Analysis of the IEEE First Benchmark Model  |  471 

0.01
Mode−5

−0.01
0 0.5 1 1.5 2 2.5 3
0.05
Mode−4

−0.05
0 0.5 1 1.5 2 2.5 3
1
Mode−3

−1
0 0.5 1 1.5 2 2.5 3
0.1
Mode−2

−0.1
0 0.5 1 1.5 2 2.5 3
0.2
Mode−1

−0.2
0 0.5 1 1.5 2 2.5 3
0.1
Mode−0

−0.1
0 0.5 1 1.5 2 2.5 3
Time (s )

Figure 10.7  The IEEE FBM: Modal-speeds for xFC = 0.29 pu.
The expanded plot of mode-3 speed is also given in Figure 10.8 for xFC = 0.29 pu, where the fre-
quency of the signal is estimated to be equal to 25.64 Hz approximately. For this case, the capacitor
voltage, vCD component is plotted in Figure 10.9. Since the system is unstable, even the capacitor
voltage grows slowly.

0.4

0.2
Mode−3 Speed deviation

−0.2

−0.4

−0.6
f = 1/0.039 = 25.64 Hz

−0.8
2.5 2.55 2.6 2.65 2.7 2.75 2.8 2.85 2.9 2.95 3
Time (s )

Figure 10.8  The IEEE FBM: Mode-3 Speed for xFC = 0.29 pu.

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 471 9/12/2017 5:36:05 PM

 472  |  Power System Analysis

−0.138

−0.139

−0.14

−0.141
Capacitor voltage, VcD (pu)

−0.142

−0.143

−0.144

−0.145

−0.146

−0.147

−0.148
0 0.5 1 1.5 2 2.5 3
Time (s )

Figure 10.9  The IEEE FBM: the Capacitor Voltage, vCD for xFC = 0.29 pu.

10.6.3.2  Induction Generator Effects

It is generally said that SSR is due to the effect of torsional interaction and induction-generator.
However, it is not easy to separately deal with each of these effects. In the above analysis, it can be
seen that the torsional interaction effects are dominant. The induction generator effects are felt when
the rotor effective resistance at slip frequency pertaining to the network resonance frequency is higher
than the overall circuit resistance. This leads to negative damping of electrical modes. In this case, the
oscillations grow much faster than those with TI effects. In order to illustrate this, in the above case
of XFC = 0.3 pu, where all torsional modes are stable, the network resistance is set to zero instead of
0.02 pu. The eigenvalues are listed in Table 10.6.

Table 10.6  Eigenvalues for the IEEE FBM System with Pg0 = 0.5 pu and RL = 0.

xFC = 0.3 Comments

−0.33744 ± j 598.92 Super-synchronous mode

2.1418 ± j156.07 Sub-synchronous mode

−1.8504 ± j298.17 Torsional mode-5

−0.36622 ± j 202.81 Torsional mode-4

−1.1608 ± j159.7 Torsional mode-3

(Continued)

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 472 9/12/2017 5:36:06 PM

 SSR Analysis of the IEEE First Benchmark Model  |  473 

Table 10.6 (Continued)

xFC = 0.3 Comments

−0.083138 ± j 127.08 Torsional mode-2

−0.27346 ± j 99.515 Torsional mode-1

−0.86152 ± j10.369 Swing mode (mode-0)

−19.145 ± j 23.792 Exciter mode

The corresponding modal-speeds and capacitor voltage vCD component are shown in Figures 10.10
and 10.11, respectively. The frequency of vCD (t) is around 24.39 Hz. It should be noted the network
resonance frequency is given by

x FC
f er = f 0
xeff (10.147)
xd′′ + xq′′
where xeff = x E + = 0.7 + 0.1675 = 0.8675 pu.
2
With xFC = 0.3 pu, we obtain fer as 35.284 Hz. In the synchronous-reference-frame-based
quantities, this appears as f 0 − f er = 24.716 Hz.

0.01
Mode−5

−0.01
0 0.5 1 1.5 2 2.5 3
0.5
Mode−4

−0.5
0 0.5 1 1.5 2 2.5 3
10
Mode−3

−10
0 0.5 1 1.5 2 2.5 3
2
Mode−2

−2
0 0.5 1 1.5 2 2.5 3
1
Mode−1

0
−1
0 0.5 1 1.5 2 2.5 3
0.5
Mode−0

−0.5
0 0.5 1 1.5 2 2.5 3
Time (s )

Figure 10.10  The IEEE FBM: Modal Speeds for xFC = 0.3 pu and RL = 0.

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 473 9/12/2017 5:36:07 PM

 474  |  Power System Analysis

0.1

0.05

−0.05
Capacitor voltage, VcD (pu)

−0.1

−0.15

−0.2
f = 24.39 Hz

−0.25

−0.3

−0.35

−0.4
0 0.5 1 1.5 2 2.5 3
Time (s)

Figure 10.11  The IEEE FBM: The Capacitor Voltage, vCD for xFC = 0.3 pu and RL = 0.

Notes:
1. In the above time-domain simulation, since the system is symmetric, the zero-sequence network
is not considered. Hence, RL0 and xL0 of the line is neglected.
2. If the network is unbalanced due to unsymmetrical faults or modelling of a thyristor-controlled
series capacitor (TCSC) in the abc-frame, the zero-sequence network is explicitly considered in
the time–domain simulation.

References
[1]  K. R. Padiyar, Analysis of subsynchronous resonance in power systems, Kluwer Academic
Publishers, Norwell, MA, USA, 1999.
[2]  IEEE SSR Task Force, ‘First benchmark model for computer simulation of subsynchronous resonance.’
IEEE Transaction on power Apparatus and systems, vol. PAS–96(5), pp. 1565–1572, 1977.
[3]  P. Kundur, Power System Stability and Control, McGraw-Hill Inc., New York, 1994.
[4]  G. D. Jennings, R. G. Harley, and D. C. Levy, ‘Sensitivity of subsynchronous resonance predictions
to turbo-generator modal parameter values and to omitting certain active subsynchronous modes,’
IEEE Transaction on Energy conversion, vol. 2(3), pp. 470–479, September 1987.
[5]  Wasynczuk O., ‘Damping subsynchronous resonance using reactive power control,’ IEEE Transaction
on power Apparatus and systems, vol. PAS-100 (3), pp. 1096–1104, March 1981.
[6]  F. C. Jusan, S. Gomes, and G. N. Taranto, ‘SSR results obtained with a dynamic phasor model
of SVC using modal analysis,’ International Journal on Electrical Power and Energy Systems,
vol. 32(6), pp. 571–582, July 2010.

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 474 9/12/2017 5:36:08 PM

 SSR Analysis of the IEEE First Benchmark Model  |  475 

Review Questions
1. For the IEEE FBM, determine the natural frequencies of the unconnected spring-mass system.
Use the parameters given in Table 10.1, neglecting damping.
2. Briefly discuss the phenomena of subsynchronous resonance in power system.
3. How does SSR manifest in a steam-turbine-generator system as torsional interaction and induc-
tion generator effects?
4. In SSR analysis it is essential to consider the generator stator transients along with detailed model
for the rotor circuits. Comment on this statement.
5. In the FBM, if classical model is employed for the generator, then induction generator effects are
absent. How do you analyse this behaviour qualitatively?
6. Indicate the properties of the ([M]−1[K]) matrix.
7. How do you approximately calculate the modal speed variables from the actual speed variables?
8. In section 10.6, the time-domain simulation of SSR is carried out in the machine-frame by
deriving appropriate interfacing matrices to avoid redundant state variables. Instead, perform this
task by terminating the generator terminals by a large resistance.
9. Explain, why a typical SSR simulation is considered as a stiff-system analysis.

M10_POWER-SYSTEM-AN00_SE_XXXX_CH10.indd 475 9/12/2017 5:36:08 PM

 Controllers for Synchronous
Generator 11
CHAPTER OUTLINE

11.1 Real and Reactive Power Controllers 11.3 Prime mover Controllers
for a Synchronous Generator 11.4 Windup and Non-windup Type
11.2 Functions and Types of Excitation Limiters on Integrator Blocks
Systems

In this chapter, typical types of controllers used for a synchronous generator in stability studies are
presented. The IEEE-type excitation and prime mover systems are discussed with their structure and
typical values of parameters.
For ease of implementation of models, state models of some system functions are also given. In
addition, the processes of computing initial value of states are briefed.

11.1 REAL AND REACTIVE POWER CONTROLLERS

FOR A SYNCHRONOUS GENERATOR
Consider the steady-state equivalent circuit of a lossless non-salient pole synchronous generator con-
nected to mains as shown in Figure 11.1.
Peg + j Qeg
j Xs
Io

+
Vto ∠0°
Efdo ∠d

Figure 11.1 Simplified Equivalent Circuit for a Round-rotor Generator.

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 476 9/12/2017 8:19:30 AM

 Controllers for Synchronous Generator  |  477 

At the generator terminals, an expression for the delivered complex power in per-unit can be written as
– –
Peg0 + jQeg0 = Vt0 I *0
where
E fd 0 − Vt 0
I0 =
jxs

After simplification and equating the real and imaginary parts on both sides, we get,

E fd 0 Vt 0 sind
Peg 0 = (11.1)
xs

E fd 0 Vt 0 cosd − Vt20
Qeg 0 = (11.2)
xs

Though both Peg0 and Qeg0 are functions of field voltage and rotor angle, their steady-state values are solely
decided by the attainment of equilibrium in the mechanical system and electrical systems, respectively. For
example, when the generator is delivering Peg0, if Efd is increased manually, although Peg may increase
temporarily; under a steady state, it settles at Peg0 itself with a reduction in the value of d for a given Vt0.
This happens because under steady state, the real power Peg0 must be equal to the mechanical input, Tm0.
However, Qeg increases to a new value since the term ( E fdVt0 cosd ) in eq. (11.2) increases. Thus, it can
be said that since reactive power is a quantity which is associated with the field energy, by controlling
the field flux via the field voltage adjustment, we can regulate the reactive power delivered by the
generator.
With regard to the real power delivered by the generator, it can be seen from the swing equation
that if real power equilibrium is perturbed, it alters the speed of the rotor mass. Thus, by employing the
speed as the actuating signal we can regulate the mechanical power input to the generator and hence,
the real power output of the generator.
Further, it is to be noted that the reactive power control is associated with electrical systems whose
response time is much smaller as against mechanical systems with which real power regulation is
connected. This difference in response time naturally provides a way to achieve decoupling between
the reactive power and real power control loops. Based on these observations, the following two major
types of controllers are employed for a generator:
1. Excitation controller
2. Prime mover controller
In the following sections, these controllers are briefly discussed.

11.2 FUNCTIONS AND TYPES OF EXCITATION SYSTEMS

The basic function of an excitation system is to provide DC power to the field winding of a synchro-
nous generator. In addition, the excitation system performs control functions such as control of voltage

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 477 9/12/2017 8:19:32 AM

 478  |  Power System Analysis

and reactive power shared by the generator, and provides means to improve small-signal and transient
stability performances of the power systems by controlling the field voltage and thereby, the terminal
voltage. Such a control of the terminal voltage of a generator is essential for the following reasons:
1. Since the synchronous reactance is very large, the generator exhibits large voltage regulation in
the range of 70%–150%. This leads to wild fluctuations of terminal voltage beyond the rating of
loads following load changes. Thus, for a desirable performance of loads, the terminal voltage
should be regulated. An exciter should have high gain (with proportional-type controllers) to keep
the terminal voltage very close to the reference value.
2. Following the occurrence of a large disturbance (in the form of a fault), the terminal voltage
should be brought back to the nominal value as quickly as possible. This facilitates the generator
to maintain its real power output at a possible high value so that the net kinetic energy imparted
to the rotor mass does not threaten the synchronous operation of the generator. To achieve this, an
exciter should be fast-acting and should have high field forcing levels which are indicated by high
ceiling voltage (3–10 pu) and current levels (2–3 pu). Such field forcing involves an application of
field voltage three to 10 times the nominal voltage so as to increase the rate of rise of field current.
3. It is generally seen that a fast-acting high-gain exciter with a high ceiling voltage levels is found
to improve the first-swing stability of system. This first-swing stability denotes large distur-
bance performance of a system and represents the critical condition of the system at the verge of
loss-of-synchronism.

Notes: Normally, excitation controllers employ proportional controllers instead of integral con-­
trollers. This is due to the following reasons:
1. Realisation of a proportional controller is easy for meeting given performance specifications.
2. Rigid control of voltage with zero steady-state error is not very essential for voltage control.
3. With analog implementation, realisation of an integral controller is not straightforward due to its
offset voltage problem. However, on a digital platform, such a restriction does not exist.
The block schematic of a typical excitation control system is shown in Figure 11.2. The functions of
each of the blocks are highlighted below.

Terminal voltage
measurement

Load
compensator

Vc
− PT
CT
+
Σ Regulator Exciter
VRef. To
+
power system
Main field Generator
Vs Speed/Power signal
Power System Stabilizer (PSS)

Figure 11.2  Functional Block Diagram of an Excitation System.

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 478 9/12/2017 8:19:33 AM

 Controllers for Synchronous Generator  |  479 

The exciter provides DC power to the synchronous machine field winding. This exciter may be in
the form of a thyristor-based bridge rectifier or in the form of an auxiliary AC generator followed
by an uncontrolled rectifier. The terminal voltage block measures the generator terminal voltage,
and rectifies and filters it to DC quantity. In addition, load compensation may be provided, which
accounts for the current carried by the generator. Such compensators are desired if generators are
connected to a common bus without any transformer, and the voltage is measured at the common
bus for the controllers. These signals are compared with a reference signal to generate error signal.
The regulator processes the error signal and develops a control signal to drive the exciter output to
a desired value so that error is minimised. The control signal is designed either to adjust the firing
angle of the controlled rectifier or to vary the field setting of the auxiliary generator. This system
is also provided with a power system stabiliser (PSS) which inputs an auxiliary signal to the regu-
lator to damp power system oscillations. The excitation system includes a wide variety of control
and protective functions which ensure that the capability limits of the exciter and synchronous
generator are not exceeded. Some of the commonly used functions are the field-current limiter,
maximum excitation limiter, terminal voltage limiter, volts-per-hertz regulator and protection, and
under excitation limiter [1].

Types of Excitation Systems

Excitation systems are classified into three broad categories based on the excitation power source
[1, 2]:
1. DC excitation system
2. AC excitation system
3. Static excitation system

11.2.1  DC Excitation Systems

The block diagram of a typical DC type excitation system is shown in Figure 11.3.

DC exciter Main generator

Amplifier

Slip ring
Field Field Armature PT
Armature

Regulator

Figure 11.3  Field Controlled DC Excitation System.

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 479 9/12/2017 8:19:33 AM

 480  |  Power System Analysis

Excitation systems of this type utilise DC generators as sources of excitation power to the field of
the synchronous machine through slip rings. The exciter may be driven by a motor or the shaft of the
generator. It may be either self-excited or separately excited. When separately excited, the exciter field
is supplied by a pilot exciter comprising a permanent magnet generator.
Type DC1A exciter model represents field-controlled DC commutator exciters, with continu-
ously acting voltage regulators. The structure of IEEE-type DC1A excitation model is shown in
Figure 11.4.

VS VRMAX

+
VC − 1 + sTC KA VR + 1 EFD
Σ Σ
1 + sTB 1 + sTA sTE
+
− −
TGR REG VFE
VF
VREF VRMIN +
Σ KE

VX
VX = EFD SE (EFD)

sKF
1 + sTF
ESS

Figure 11.4  IEEE-type DC1A Excitation System.

In Figure 11.4,
• TGR represents transient gain reduction block
• ESS represents excitation system stabiliser block
• SE represents exciter saturation function given by A.eB.E, where A and B are constants.
• E is set to Efd in Vx calculations.
The typical values of the parameters of DC1A -excitation system are tabulated in Table 11.1.

Table 11.1  Typical Values of DC1A Excitation System Parameters.

KA = 200 TA = 0.02 s TC = 1.0 TB = 10.0 s KE = -0.0485

 TE = 0.25 s KF = 0.04 TF = 0.56 s VRMAX = 6.0 VRMIN = -6.0
E1 = 3.5461 SE1 = 0.08 E2 = 4.7281 SE2 = 0.260

The time constant of the bus voltage measuring transducer can be taken as 0.02 s.

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 480 9/12/2017 8:19:34 AM

 Controllers for Synchronous Generator  |  481 

11.2.2  AC Excitation Systems

The excitation systems of this type utilise alternators (AC machines) as sources of the main generator
excitation power. Usually, the exciter is on the same shaft as the turbine generator. The AC output of
the exciter is rectified by either controlled or non-controlled rectifiers to produce the DC needed for
the generator field. AC exciters can be classified as follows:

1. Field-controlled alternator with non-controlled rectifiers:

(a) Rotating rectifier arrangements—Brushless excitation system
(b) Stationary rectifier arrangements
2. Alternator-supplied controlled-rectifier excitation systems:

Two typical types of AC excitation systems are discussed in the following sections.

11.2.2.1 Field-controlled Alternator with Non-controlled Rectifiers:

Brushless Excitation Systems
With rotating rectifiers, the need for slip rings and brushes is eliminated, and the DC output is directly
fed to the main generator field. The armature of the AC exciter and the diode rectifiers rotate with the
main generator field. A small AC pilot exciter with a rotating permanent magnet rotor mounted on
the main shaft energises the stationary field of the AC exciter. The voltage regulator controls the AC
exciter field which, in turn, controls the field of the main generator.
The functional block schematic of a typical brushless excitation system is shown in Figure 11.5.

Rotor of pilot exciter

Rotating structure Main generator

Pilot exciter AC Exciter Field

N
Armature Armature Armature
S
PT

Field of
AC exciter

Three phase AC

Regulator

Figure 11.5  Brushless Excitation System.

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 481 9/12/2017 8:19:35 AM

 482  |  Power System Analysis

The structure of IEEE-type AC1A excitation model is shown in Figure 11.6.

VS
VAMAX VRMAX
+
VC − 1 + sTC KA VR 1 VE + EFD
Σ
1 + sTB 1 + sTA + Σ sTE Π
+ − − +F
EX
VF 0
VREF VAMIN VRMIN
VX = VE SE (VE) FEX = f [IN]

VX +
Σ KE
+ KCIFD
IN =
VE
sKF VFE +
1 + sTF Σ KD
+ IFD

Figure 11.6  IEEE-type AC1A Excitation System.

This exciter also includes the rectifier regulation characteristics which is given by the following set
of equations:
f ( I N ) = 1.0 − 0.577 I N if I N ≤ 0.433 (11.3)

f ( I N ) = 0.75 − I N2 if 0.433 < I N < 0.75 (11.4)

f ( I N ) = 1.732 (1.0 − I N ) if 0.75 ≤ I N ≤ 1.0 (11.5)

f (IN ) = 0 if I N > 1.0 (11.6)

The above set of equations introduces non-linearity and hence, it requires iterative steps to evaluate the
initial conditions of the states pertaining to the exciter.
Typical values of the parameters of AC1A -excitation system are tabulated in Table 11.2.

Table 11.2  Typical Values of AC1A Excitation System Parameters.

KA = 100 TA = 0.02 s TC = 1.0 s TB = 1.0 s VAMAX = 14.5
VAMIN = −14.5 VRMAX = 6.03 VRMIN = −5.43 KE = 1.0 TE = 0.8 s
KF = 0.03 TF = 1.0 s KD = 0.38 KC = 0.2 E1 = 3.14
SE1 = 0.03 E2 = 4.18 SE2 = 0.1

The time constant of the bus voltage measuring transducer can be taken as 0.02 s.

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 482 9/12/2017 8:19:38 AM

 Controllers for Synchronous Generator  |  483 

11.2.2.2  Alternator-supplied Controlled-rectifier Excitation Systems

In this case, the DC output is fed to the field winding of the main generator through brushes and
slip-ring arrangements. The AC output of the exciter is rectified using controlled rectifiers. The
regulator directly controls the DC input to the main field winding by adjusting the firing angle
of the controlled rectifiers. Hence, this system inherently provides high initial response. The AC
exciter is a self-excited type and uses an independent static voltage regulator to maintain its output
voltage.
The functional block schematic of a typical alternator-supplied controlled-rectifier excitation
system is shown in Figure 11.7.

Main generator
AC Exciter Stationary Field
Controlled rectifier
Field Slip ring
Armature Armature

PT

Exc.
reg.
Regulator

Figure 11.7  Alternator-supplied Controlled-rectifier Excitation System.

The IEEE-type AC4A excitation model is an example to this kind of excitation systems. The transfer
function level block schematic of this model is shown in Figure 11.8.

Vs
V1MAX (VRMAX − KCIFD)

+
VC − V1 1 + sTC KA EFD
Σ 1 + sTA
1 + sTB
+
VRMIN
VREF V1MIN

Figure 11.8  IEEE-type AC4A Excitation System.

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 483 9/12/2017 8:19:41 AM

 484  |  Power System Analysis

Typical values of the parameters of IEEE-type AC4A excitation system are tabulated in Table 11.3.

Table 11.3  Typical Values of AC4A Excitation System Parameters.

KA = 200 TA = 0.015 s TC = 1.0 s

TB = 12.0 s V1MAX = 10.0 V1MIN = −10.0
VRMAX = 5.64 VRMIN = −4.53 KC = 0

The time constant of the bus voltage measuring transducer can be taken as 0.02 s.

11.2.3  Static Excitation Systems

All components in these systems are static or stationary. Static rectifiers, controlled or uncontrolled,
supply the excitation current directly to the field of the main synchronous generator through slip rings.
The main generator (or the station auxiliary bus) supplies power to the rectifiers through a transformer
to step down the voltage to an appropriate level; in some cases, power is supplied from auxiliary
windings in the generator.
The two major types of static excitation systems are as follows:
1. Potential-source controlled rectifier systems
2. Compound-source rectifier systems

11.2.3.1  Potential-Source Controlled Rectifier Systems

In this system, the excitation power is supplied through a transformer from the generator terminals
or the station auxiliary bus, and is regulated by a controlled rectifier. This type of excitation system
is also called a bus-fed or transformer-fed static system. A functional block schematic of this type of
exciter is shown in Figure 11.9.

Exciter Main generator

transformer Controlled rectifier Field
Slip ring

Armature
Three
phase AC*
PT

* Alternatively, from
auxiliary bus

Regulator

Figure 11.9  Potential-source Controlled Rectifier Excitation System.

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 484 9/12/2017 8:19:42 AM

 Controllers for Synchronous Generator  |  485 

The IEEE-type ST1A exciter model (see Figure 11.10) represents a potential-source controlled recti-
fier systems.

Vs (Vg VRMAX − KCIFD)

+ EFD
VC − V1 (1 + sTC) KA VA
Σ Σ
+ (1 + sTB) 1 + sTA +
− −
VREF VF TGR
Vg VRMIN

sKF
1 + sTF
KLR IFD
ESS Σ
0
−

ILR

Figure 11.10  IEEE-type ST1A Excitation System.

Notes:
1. KLR and ILR represent the field current limiter parameters. These are used to protect the exciter
and field circuit as the exciter ceiling voltage tends to be high in static exciters.
2. Since the excitation power is supplied through a transformer from the generator terminals, the
exciter ceiling voltage is directly proportional to the generator terminal voltage. This is accounted
by Vg (which is same as VC), in the limiter.
3. The effect of rectifier regulation on ceiling voltage is represented by KC. For a transformer-fed
system, KC is usually small.
Typical values of the parameters of IEEE-type ST1A excitation system are tabulated in Table 11.4.

Table 11.4  Typical Values of IEEE-type ST1A Excitation System Parameters.

KA = 200 TA = 0.02 s TC = 1.0 s TB = 1.0 s

KF = 0 TF = 1.0 s KC = 0.04 KLR = 4.54
ILR = 5 VRMAX = 7.0 VRMIN = –6.4

The time constant of the bus voltage measuring transducer can be taken as 0.02 s.

Note: If TGR, ESS, the effect of terminal voltage and rectifier regulation on the ceiling voltage, and
the field current limiters are neglected, it results in a simple single-time constant static exciter [3]. The
block schematic of such an exciter is shown in Figure 11.11.

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 485 9/12/2017 8:19:43 AM

 486  |  Power System Analysis

VC
Efdmax
−
+ KA
VREF Σ Efd
1 + sTA
+
Efdmin

VS

Figure 11.11  Single-time Constant Static Excitation System.

The time delay associated with the bus voltage measuring transducer is neglected.

11.2.3.2  Compound-Source Rectifier Systems

In this case, the power to the excitation system is formed by utilising the current as well as the voltage
of the main generator. This may be achieved by means of a power potential transformer and a saturable
current transformer as depicted in Figure 11.12.

Main generator
Saturable
current
transformer Armature

CT
Current PT
source Field Power
transformer
Source
regulator
Slip ring

Voltage source

Regulator

Figure 11.12  Compound-source Rectifier Excitation System.

The IEEE-type ST2A exciter model (see Figure 11.13) represents a compound-source rectifier
excitation system.

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 486 9/12/2017 8:19:44 AM

 Controllers for Synchronous Generator  |  487 

VS VRMAX VFDMAX

+
EFD
VC − KA VR + + 1
Σ Π Σ
+ 1 + sTA sTE
− + −
VB 0
VF
VREF VRMIN
KE

sKF
1 + sTF

Vg
VE +
Ig VE = KpVg + j KIIg Π
+

KCIFD
IFD IN = FEX = f [IN]
VE IN

Figure 11.13  IEEE-type ST2A Excitation System.

This exciter also includes the rectifier regulation characteristics as given in eq. (11.3) – eq. (11.6).
Typical values of the parameters of IEEE-type ST2A excitation system are tabulated in Table 11.5.

Table 11.5  Typical Values of IEEE-type ST2A Excitation System Parameters.

KA = 120 TA = 0.15 s KE = 1.0 TE = 0.5 s

KC = 0.65 KF = 0.05 TF = 1.0 s VRMAX = 4.20

VRMIN = −4.20 KP = 1.19 K1 = 1.62 E FDMAX = 3.55

The time delay associated with the bus voltage measuring transducer is neglected.

11.2.4  State-space Model of Some System Functions

Example 1

Y ( s) sK F
= (11.7)
U ( s) 1 + sTF

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 487 9/12/2017 8:19:46 AM

 488  |  Power System Analysis

The above transfer function can also be written as

 
Y ( s) K F  s 
=  + 1 − 1 (11.8)
U ( s) TF  1 + s 
 TF 

 1 
Y ( s) K F  TF 
= 1 − 
U ( s) TF  1
+ s
 TF 

KF  1 
= 1 − 
TF  1 + sTF 

KF
KF TF
= −
TF 1 + sTF

Therefore, we have
KF
K TF
Y ( s) = F U ( s) − U ( s) (11.9)
TF 1 + sTF
Let us define
KF
TF
x ( s) = U ( s) (11.10)
1 + sTF

From the above equation, we can write a state model as,

1 K
x = − x + F2 u
TF TF
(11.11)
and from eq. (11.9), the output equation is given by
KF
y= u − x (11.12)
TF

Example 2

Y ( s)  1 + sTC 
= (11.13)
U ( s)  1 + sTB 

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 488 9/12/2017 8:19:49 AM

 Controllers for Synchronous Generator  |  489 

 1  
  + s 
T  TC 
= C + 1 − 1 (11.14)
TB   1  
 + s 
  TB  

  1 1 
  − 
T  TC TB 
= C 1 + 
TB   1   (11.15)
  T + s 
 B 

  TB  
−1
T   TC  
= C 1 + (11.16)
TB  (1 + sTB ) 
 
 

 T  
T 1−  C  
 TB 
= C +  (11.17)
 TB (1 + sTB ) 
 
 

Considering the second term in the above equation and writing the corresponding state-space
equation, we get
1 K
x = − x + c1 u (11.18)
TB TB
and
y = x + K c 2 u (11.19)
where
T  TC
K c1 = 1 −  C  and Kc2 =
 TB  TB

Example 3
Y ( s) s
= (11.20)
U ( s) 1 + sT

The above function is rewritten as

sU ( s) (11.21)
Y ( s) =
1 + sT

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 489 9/12/2017 8:19:52 AM

 490  |  Power System Analysis

Let us set
U ( s)
Z ( s) = (11.22)
1 + sT

and with this, eq. (11.21) is restated as Y(s) = sZ(s). A state model of eq. (11.22) is given by

1 1
z = − z+ u (11.23)
T T
Now, the output equation is given by
y(t ) = z (11.24)
By adjusting the time-constant, (11.20) is used as an approximate differentiator.

11.2.5  Initial Condition Calculations for Exciters

The main purpose of initial condition calculation is to establish the operating point. In the time-
domain simulation, if the state variables are set with this initial values, it avoids the simulation time
to reach a steady-state. This facilitates that the time-domain simulation be restricted for carrying out
disturbance-related studies.

11.2.5.1  Single-time Constant Static Exciter

From Figure 11.11, with Vs set to zero, we have

E fd ( s) KA
= (11.25)
[VREF − Vc ]( s) 1 + sTA

Writing a state model for the above transfer function, we get,

dE fd 1 K
=− E fd + A [VREF − Vc ] (11.26)
dt TA TA

dE fd
Under steady state, we know that = 0 and the initial value of Efd is Efd0. Using this, we can
dt
calculate the value of the reference voltage as
E fd 0
VREF 0 = + Vc 0 (11.27)
KA

where Vc0 represents the steady-state value of the generator terminal voltage in per-unit and is obtained
from the load flow result.
Using the above value for VREF and with Efd0 as the initial value on the integrator (see Figure 11.14)
dE fd
the input to the integrator at point z (which represents signal) remains at zero when the system
dt
is in steady-state. This is because under steady-state, Vc remains at Vc0. This, in turn, ensures that the
output of the integrator remains at Efd0 as long as there is no disturbance.

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 490 9/12/2017 8:19:54 AM

 Controllers for Synchronous Generator  |  491 

VC

−
+ + 1 z 1 Efd
Σ KA Σ TA s
VREF0
−

Figure 11.14  Building Blocks of Single-time Constant Static Exciter.

11.2.5.2  IEEE-type AC4A Exciter

Using eqs. (11.18) and (11.19), Figure (11.8) can be redrawn as in Figure 11.15.

VS
KC2
+
+ H
VC V1 KA EFD
− Σ
Σ 1 + sTA

+ +

VREF
KC1
1 + sTB
V2

Figure 11.15  IEEE-type AC4A Excitation System-simplified.

From Figure 11.15, we have,

E fd ( s) KA H ( s)  1 + sTC 
= and = (11.28)
H ( s) 1 + sTA V1 ( s)  1 + sTB 

E fd ( s)
The state-model of is given by
H ( s)
dE fd 1 K 
=− E fd +  A  h (11.29)
dt TA  TA 

dE fd
Since under steady state, = 0 and the initial value of E fd is E fd0 we have
dt
E fd 0
h0 = (11.30)
KA

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 491 9/12/2017 8:19:57 AM

 492  |  Power System Analysis

From Figure 11.15, we have

V2 ( s) K C1
=
V1 ( s) 1 + sTB

Under steady-state, using eq. (11.18), we have

V20 = V10 KC1 (11.31)

Further, we have
h0 = V20 + KC 2V10

Using eq. (11.31) in the above equation and rewriting for V10, we get
V10 = h 0 (11.32)

Now, substituting for V10 in eq. (11.31) and from eq. (11.30) we get
E fd 0
V20 = K C1
KA
(11.33)

Since V10 = VREF0 – Vc0 we get VREF0 as

E fd 0
VREF 0 = + Vc 0
KA

where Vc0 is the terminal voltage of the machine under steady state.

11.3  PRIME MOVER CONTROLLERS

Consider a case where load-generation balance exists at rated frequency. Following a real power (load)
change, the field energy responds immediately and quickly reaches a new steady-state. This makes the
mechanical system to respond meeting the load change through either release or absorption of energy
in the rotating masses until the mechanical input power is adjusted by the governors’ controllers. This,
in turn, causes the speed to deviate from the nominal value. Having sensed this speed deviation, the
governor system initiates the valve control mechanism to effect the mechanical input change. Thus, a
prime mover controller performs the following functions:
1. It controls real power shared by a generator.
2. It provides facilities to control system frequency.

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 492 9/12/2017 8:19:59 AM

 Controllers for Synchronous Generator  |  493 

A typical block schematic of a prime mover system is shown in Figure 11.16.

PRef Speed ref.

changer
(Load ref.)

Σ Valve control PGV PM

LFC Governor
mechanism Turbine
controlled
using
gates Gate
servomotors
position

Speed deviation
PM = Mechanical power
LFC = Load frequency control
Speed Speed deviation
governor

Figure 11.16  Block Schematic of a Prime Mover System.

A prime mover generally provides the following control functions:

1. Primary control: It is in the form of a speed governor and comes into action following speed
deviation from the nominal value. Its primary function is to change the input proportional to
the frequency deviation so as to prevent large frequency excursions. This type of control is
generally enabled on all generators in a power system and permits load share among the gen-
erators as per their droop settings. This control is also referred to as free-governor mode of
operation.
2. Secondary control: It is a slow acting control function and comes into action after the primary con-
troller settles with a steady state frequency error through appropriate load share among generators.
It employs an integral controller to modify the load reference in such a way that there is no steady-
state frequency error. This function is also referred to as load frequency control (LFC). This con-
trol is enabled on any one selected unit. If present on more than one unit, it must be coordinated.
3. Tertiary control: It sets PRef, the economic dispatch of committed generators as per the generation
cost.

Notes:
1. In stability studies, only speed-governor (primary) control systems, including turbine systems,
are represented.
2. In turbine systems, the output is generally assumed to be in the form of torque. If it is given as
power output, it should be converted into torque by dividing it by pu speed before using it in the
swing equation.

11.3.1  Influence of Prime Mover Controllers on Load Sharing

The following two-machine analysis demonstrates the importance of enabling LFC on selected units.
Consider two generators working in parallel sharing a common load at rated speed, w in per-unit. Each
generator has its own turbine system with the following types of prime mover controllers.

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 493 9/12/2017 8:19:59 AM

 494  |  Power System Analysis

11.3.1.1  With Primary Controllers

In this case it is assumed that only speed governor controller is enabled on both machines with their
respective droop settings (see Figure 11.17).

w
Tm10

− +
+ w
Turbine
Σ Kp1 Σ
w Ref1 generator−1
+ ∆Tm1

w Tm20

− +
+ Turbine
Kp2
w Ref2 Σ Σ
+ ∆Tm2 generator−2

Figure 11.17  Machines With Only Speed Governor Blocks.

Under steady state, we have,

∆Tm1 = (w Ref 1 − w ) K p1(11.34)

∆Tm2 = (w Ref 2 − w ) K p2 (11.35)

Note that due to a proportional controller, there exists a steady-state frequency error. Hence, we can
see that machines share a common load distinctly depending on K p1 and K p2 . These gains represent
the speed regulation, given by,
1
Kp =
s

∆f R
f
where s = 0 in pu.
∆PL
P0

11.3.1.2  With Integral (or Secondary) Controllers

In this case, it is assumed that only the integral controller is enabled on both the machines with their
respective gain settings (see Figure 11.18).

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 494 9/12/2017 8:20:02 AM

 Controllers for Synchronous Generator  |  495 

w Tm10

− +
KI1 + Turbine w
w Ref1 Σ Σ
+ s ∆Tm1 generator−1

w Tm20

− +
+ Turbine
KI2
w Ref2 Σ Σ
+ s ∆Tm2 generator−2

Figure 11.18  Machines with Only LFC Blocks.

Consider the following cases:

1. Let wRef1 = wRef2 = wRef
Due to an integral controller, the steady-state frequency error is zero. Hence, we have,

(wRef − w) = 0
We also have,

d ∆Tm1
= (w Ref − w ) K I 1 = 0
dt

d ∆Tm2
= (w Ref − w ) K I 2 = 0
dt

d ∆Tm1 d ∆Tm2
Thus, now, the load share is constrained by the relationships, = = 0. This indi-
dt dt
cates that there is no unique sharing of load between the units.

2. If w Ref 1 ≠ w Ref 2 , then each controller tends to drive the system speed to its reference setting.
Therefore, a steady-state cannot be reached, as the units should have a common frequency w
under steady-state.
The above example clearly demonstrates that LFC cannot be provided on more than one
machine.
As per the IEEE committee report [4], the following are the typical types of turbines employed
in stability studies.

11.3.2  Model of Hydraulic Turbines

The hydraulic turbine model suited for stability studies is approximately represented as shown in
Figure 11.19.

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 495 9/12/2017 8:20:04 AM

 496  |  Power System Analysis

PGV (1 − sTW) PM

(1 + 0.5 sTW)

Figure 11.19  Hydraulic Turbine Model.

The input PGV for the turbine comes from the speed-governor. It represents gate opening
expressed in per unit. The time constant TW is called water starting time or water time constant.
Values for TW lie in the range of 0.5–5 s with the typical value around 1.0 s. It is to be noted that,
since hydraulic turbine has non-minimum phase characteristic, (i.e., it has a zero on the RHS of
the complex plane s), it exhibits sluggish behaviour. This requires some dashpot arrangements in
the speed-governor systems to improve its response. Let us obtain the unit-step response of the
system function.
 1 − sTw  1
Y ( s) = 
 1 + 0.5sTw  s

1 − sTw
=
 2  Tw
 T + s 2 s
w

A B
= +
s 2
+s
Tw

Determining the residues, we have,

A = Y ( s ) s |s = 0 = 1

 2
B = Y (s) 

+ s =
(1 − sTw )
2 Tw
 Tw  s =−
s
Tw 2
2 s =−
Tw
1+ 2
=
Tw  2 
−
2  Tw 

= −3

Thus, Y(s) can be written in partial fraction form as

1 −3
Y ( s) = +
s  2 
 T + s
w

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 496 9/12/2017 8:20:06 AM

 Controllers for Synchronous Generator  |  497 

1 3
= −
s  2 
 T + s
w

Taking the Laplace inverse, we get the time-domain response as

 −t 
2

y(t ) = 1 − 3e Tw 
 
 

It can be seen that at t = 0, y = −2, whereas the input is 1.0. This is the characteristic of a non-­minimum
phase system.

11.3.3  Model of Steam Turbines

Figure 11.20 shows the configuration of a typical steam system.

Reheater Crossover

Valve Control Shaft

HP IP LP LP
position valve

Figure 11.20  Steam System Configuration: Tandem Compound Type.

The low frequency power-swing stability models of steam turbines is as follows:

1. Tandem compounded, single reheat type (see Figure 11. 21):

+ +
Σ Σ
PM
+ +

FHP FIP FLP

PGV
1 1 1
1 + sTCH 1 + sTRH 1 + sTCO

Figure 11.21  Tandem Compounded, Single Reheat Type Steam Turbine Model.

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 497 9/12/2017 8:20:08 AM

 498  |  Power System Analysis

The typical values of the parameters are tabulated in Table 11.6.

Table 11.6  Typical Values of Tandem Compounded, Single Reheat Type Steam Turbine Parameters.

TCH = 0.1- 0.4 s TRH = 4–11 s TCO = 0.3–0.5 s

FHP  =  0.3 FIP   = 0.3 FLP  = 0.4

where TCH, TRH, and TCO represent steam chest, reheater, and crossover piping delays, respectively.
2. Non-reheat Type (see Figure 11.22):

1 PM

PGV 1 + sTCH

Figure 11.22  Non-reheat Type Steam Turbine.

The typical value of TCH is 0.1-0.4 s.

Note: For studying power-swing dynamics (in range of 0.5–3 Hz), it is assumed that inertia constants
of turbine systems are aggregated into the generator inertia, treating them as a single rigid mass.

11.3.4  Modelling of Speed-Governing Systems

Typically, there are two types of speed-governing systems [3] as follows:
1. Speed-governing systems for hydraulic turbines
2. Speed-governing system for steam turbines

11.3.4.1  Speed-Governing Systems for Hydraulic Turbines

An approximate non-linear model for the hydro-speed-governing system is shown in Figure 11.23.
Speed Rate limiter
reference in p.u.
+ + 1 1 1 Nonlinear
Σ Σ
1 + sTP TG s function Gate
− − position
Pilot valve Distributor valve
and and gate
Actual servomotor servomotor
speed
in p.u. + b sTR
Σ
1 + sTR
+
Transient droop Comp.

Permanent droop comp.

Figure 11.23  Speed-governing System for Hydro Turbines.

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 498 9/12/2017 8:20:09 AM

 Controllers for Synchronous Generator  |  499 

In the above figure, TR and b are calculated as

1.25TW
TR = 5TW , b =
H
where
TW = water time constant in seconds.
H = inertia constant of a machine in MJ/MVA.
A simplified block schematic that can be employed in stability studies which is derived from
Figure 11.23 is shown in Figure 11.24.

Po
Pmax

+
K (1 + sT2)
Σ
p.u. slip (1 + sT1)(1 + sT3) ∆Pe − PGV

Pmin

Figure 11.24  General Model for Speed-governor for Hydro Turbines.

The various parameters and time constants shown in the simplified block can be obtained using
the following expressions.

TB T2
T1 , T3 = ± B − TA (11.36)
2 4
where
 1  1
TA =   TRTG , TB =   (s + b ) TR + TG 
s  s 

and
1
K=
s

Typical values of parameters for speed-governor of hydro turbines are tabulated in Table 11.7.

Table 11.7  Typical Values of Parameters for Speed-governor of Hydro Turbines.

TW = 1.0 s TG = 0.2 s T2 = 0
s = 0.05

In Figure 11.24, Po represents the nominal value of the mechanical input PM. Limits on PGV can be
selected as Pmax = 1.1 Po, and Pmin = 0.1 Po.

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 499 9/12/2017 8:20:11 AM

 500  |  Power System Analysis

11.3.4.2  Speed-Governing System for Steam Turbines

A simplified, general model for the speed-governing system for steam turbine is shown in
Figure 11.25.

PO Pmax
+
K (1 + sT2) − 1 1 PGV
Σ
p.u slip 1 + sT1 T3 s
−
Pmin

Figure 11.25  General Model for Speed-governor for Steam Turbines.

Typical values of parameters for speed-governor of steam turbines are tabulated in Table 11.8.

Table 11.8  Typical Values of Parameters for Speed-governor of Steam Turbines.

T1 = 0.2 s T2 = 0 T3 = 0.1 s
Pmax = 1.1 Po Pmin = 0.1 Po

11.4  W
 INDUP AND NON-WINDUP TYPE LIMITERS
ON INTEGRATOR BLOCKS
While implementing integrators in time-domain simulation, it is necessary to limit the signal levels to
realise limiters on controller blocks. There are two types of limiters as follows:
1. Windup limiters
2. Non-windup limiters
The block schematics of these limiters are shown in Figure 11.26.

3
3
y1 y h
u 1 u 1
s s
−1 h(0) = 0
y1(0) = 0
−1
(a) Windup (soft) limiter (b) Non−windup (hard) limiter

Figure 11.26  Windup and Non-windup Type Limiters on Integrator Blocks.

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 500 9/12/2017 8:20:12 AM

 Controllers for Synchronous Generator  |  501 

11.4.1  Windup Type Limiter on Integrator Blocks

This type of limiter is also known as soft limiter since the signal levels are amplitude limited outside
the integrator blocks using a separate block referred to as saturation block (see Figure  11.26(a)).
This implies that limits are imposed after carrying out the integration process in the usual manner. To
demonstrate this, a symmetrical square wave of amplitude 5 and - 5 V is integrated, and the interme-
diate results are as shown in Figure 11.27. Here, the limits are chosen as 3 V and - 1 V.

Windup limiters (Soft limiter)

8
Input−u
out−y1
out−y(Ltd)
6
5V

4
3V

2
Signals (V )

−2

−4

−6

−8
0 1 2 3 4 5 6
Time (s )

Figure 11.27  Windup Type Limiter on the Integrator Block.

In the figure the signal y1 shows the output of the integrator block whose maximum amplitude is 5 V,
whereas the amplitude of y-signal is limited to 3 V. Since the minimum amplitude of the signal is zero,
the lower limit of - 1 V is ignored. This type of limiter is easy to implement. However, it may take long
time to respond if the peak amplitude of y1-signal is relatively larger than the limiter levels.

11.4.2  Non-windup Type Limiter on Integrator Blocks

This type of limiter is also known as hard limiter because if the limits are hit then, the process of integration
itself, is stopped. Here, the limit values are set inside an integrator block as it is denoted symbolically in

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 501 9/12/2017 8:20:13 AM

 502  |  Power System Analysis

Figure 11.26 (b). To demonstrate the limiting action, a symmetrical square wave of amplitude 5 and −5 V
is considered as the input signal for the integrator. The output signal is obtained as in Figure 11.28. In this
Figure the signal h1 denotes the output of the integrator if there were no limits.

8
Input−u
out−h1
out−h(Ltd)
6
5V

4
3V

2
Signal (V )

0
(D) 2.8 s
−1 V
(A)1.6 s
−2
(C) 2.2 s
(B)1.8 s
−4

−6

−8
0 1 2 3 4 5 6
Time (s )

Figure 11.28  Non-windup Type Limiter on the Integrator Block.

The following time-instants are highlighted:

1. At t = 1 s, the signal h comes out of the limit (3 V) since the input makes a transition to −5V.
2. At point-A, h becomes zero at t = (1.0 + 0.6) s, since, 3 – 5 × 0.6 = 0.
3. At point-B, h becomes −1 V at t = (1.6 + 0.2) s, since, −5 × 0.2 = −1.
4. At point-C, h becomes zero again at t = (2.0 + 0.2) s, since, −1 + 5 × 0.2 = 0.
5. At point-D, h becomes 3 V at t = (2.0 + 0.8) s, since, −1 + 5 × 0.8 = 3.
This type limiter offers much better response time than the windup type limiter.

References
[1]  P. Kundur, Power System Stability and Control, McGraw-Hill Inc., NY, 1994.
[2]  IEEE Recommended Practice for Excitation Systems Model for Power System Stability Studies,
IEEE Standard 421. 5-1992.

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 502 9/12/2017 8:20:13 AM

 Controllers for Synchronous Generator | 503

[3] K. R. Padiyar, Analysis of Subsynchronous Resonance in Power Systems, Kluwer, Academic

Publishers, Boston, 1999.
[4] IEEE Committee Report, ‘Dynamic Models for Steam and Hydro Turbines in Power System
Studies,’ IEEE Trans. on Power Apparatus and Systems, Vol. PAS-92, pp. 1904–1915, November–
December 1973.

Review Questions
1. Enumerate different functions of an exciter.
2. Categorize different types of excitation systems and indicate which of them offer a fast response.
3. Sketch the response of the following system functions for a unit step input.
(1 + sT1 )
(a) Lead compensator: with T1 > T2.
(1 + sT2 )

(b) TGR: (1 + sT1 ) with T1 < T2.

(1 + sT2 )
1
(c) Delay circuit: .
(1 + sT )
sTw
(d) Wash-out circuit: .
(1+ sTw )
sT1
(e) High pass circuit: with T1 < T2.
(1 + sT2 )
4. Write a state-space model for the exciter system shown in Figure 11.29. Determine the stability
of the equilibrium point given that
VR (0) = 1.0 pu. and E FD (0) = 2.0 pu.
K A = 25, TA = 0.09 s, K E = 0.8, TE = 0.8 s and S E ( E FD ) = 0.003 e 0.98 EFD
VS VRMAX

+ VR
VC KA + 1 EFD
− Σ
+ Σ 1 + sTA sTE
− −
REG VFE

VREF VRMIN +
Σ KE

VX
VX = EFD SE(EFD)

Figure 11.29 DC-type Exciter Model for Operating Point Stability Analysis.

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 503 9/12/2017 8:20:17 AM

 504 | Power System Analysis

5. For the system shown in Figure 11.30, with (1.1) model for the generator, it is found that E fdo = 2 pu.
Obtain a SIMULINK model for the following exciter (see Figure 11.31) with an appropriate ini-
tial value for the state variables and VREF so that the simulation starts from steady state.
V = 1.0∠0°

j0.11

S = (0.6667 + j0.0677) p.u.

Figure 11.30 SMIB System for Initial Condition Calculations.

VS V1MAX (VRMAX − KCIFD)

+
VC − V1 1 + sTC KA EFD
Σ 1 + sTA
1 + sTB
+
VRMIN
VREF V1MIN

Figure 11.31 AC4A Exciter for Initial Condition Calculations.

Consider TC = 1 s, TB = 10 s, K A = 200, and TA = 0.015 s. Vc represents the magnitude of the

generator terminal voltage. Further, ignore Vs signal.
6. A 210-MW unit (247 MVA) carries full load at 3000 rpm. The generator circuit breaker is sud-
denly tripped because of a short circuit at its terminals. What is the new speed after 0.5 s if the
time delay for a noticeable governor action is 0.5 s or longer? For the generator, H = 4.22 s on the
machine base.
7. List the important functions of a prime mover system.
8. For a 50-Hz system shown in Figure 11.32, obtain the structure of the small-signal state-space
model in the usual notations.
a
+ b RL
Generator
− c
RL
RL

Figure 11.32 Figure for Analysing the Small-signal Stability of the Operating Point
without Speed Governor.

Assume that the state vector is given by x = [w , y d , y q , y F , y H , y G , y K ]T . Neglect armature

resistance. Assume that the mechanical input and field voltage are held constant. Employing the

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 504 9/12/2017 8:20:23 AM

 Controllers for Synchronous Generator | 505

IEEE FBM generator data, determine RL for which the system becomes unstable with Vg0 = 1.0 pu
and mechanical damping = 0.005 pu.
Show that with the above determined value of RL , if a speed governor with the following structure
∆Tm = − K sg ∆w

is used, then the system becomes stable for K sg = 20. Perform a time-domain simulation to verify
the inferences.
9. Compare the responses of windup and non-windup type limiters and discuss their applications.

M11_POWER-SYSTEM-AN00_SE_XXXX_CH11.indd 505 9/12/2017 8:20:24 AM

 Power System
Angle Stability 12
CHAPTER OUTLINE

12.1 Relative-Angular Stability 12.3 Effect of a Fast-acting High-gain

Analysis Static Exciter on SMIB System
12.2 Synchronising and Damping Torque 12.4 Small-signal Stability Analysis of an
Analysis SMIB System

In this chapter, an effort is made to clarify the difference between small- and large-signal rotor-angle
stability analysis of power systems. A single spring-mass mechanical analogy is described to under-
stand the well-known single machine infinite-bus (SMIB) system.
The small-signal analysis of an SMIB system is presented showing the influence of a fast-acting
and high-gain static exciter. Synchronising and damping torque analysis are also carried out to explain
the design of a power system stabiliser (PSS) which is used to overcome the negative damping effects
of a static exciter.
Eigenvalue-based analyses are included to demonstrate the performance of different models for the
generator with and without exciters. A slip-signal PSS is also designed to show its influence on the
operating point stability.

12.1 RELATIVE-ANGULAR STABILITY ANALYSIS

In this analysis, angular displacement between machines caused by a disturbance is studied. Based on
the magnitude of disturbance, the following types of stability analysis are carried out:
1. Small-signal stability (operating point stability)
2. Large disturbance stability (transient stability)
Small-signal stability denotes the stability of an equilibrium point of a system when the system is per-
turbed by an infinitesimally small disturbance. An equilibrium point is said to be small-signal stable if
the resulting trajectory of the state variables eventually converges to the initial equilibrium point. On
the other hand, for an unstable equilibrium point, the flow of state variables never comes back to the

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 506 9/12/2017 3:24:02 PM

 Power System Angle Stability  |  507 

initial equilibrium point. However, it should be noted that a system operating at a stable equilibrium
point may lose synchronism if it is subject to a large disturbance.
To introduce the concept of small-signal and transient stability, consider a system shown in
Figure 12.1 which shows two possible equilibria for a ball. In position A of the ball, if it is perturbed
by a small magnitude of force, the ball gets displaced from its equilibrium and acquires energy due to
displacement. This may lead to up-and-down movement of the ball, exhibiting oscillations due to the
restraining gravitational force. If the system possesses some positive damping (in the form of friction),
the ball may eventually settle back at position A. This implies that position A is a stable equilibrium
point. However, if the ball is carefully made to occupy position B, even an infinitesimally small force
can displace it permanently from its position B. Thus, we can conclude that position B represents an
unstable equilibrium point. However, it should be noted that if the ball is kicked by a ‘large’ force
when it is at position A, it may roll down the hill, crossing position B. This indicates that position A
is only a small-signal stable operating point; however, it could be large-signal (transient) unstable.
With respect to position B, it can be seen that it is not only a small-signal unstable operating point,
but also transient unstable. These observations highlight that small-signal stability is a fundamental
requirement of a stable system.

A−stable equilibrium

B−unstable equilibrium
FL
B

FR

FR = Restrainig force due to gravity

FL = Applied disturbance

Figure 12.1  Ball in a Valley Example System.

12.1.1  Small-signal Stability Analysis of SMIB System

Consider a lossless SMIB system as shown in Figure 12.2. For ease of analysis, the classical model
has been employed for the generator. With this, the equivalent circuit of the system appears as shown
in Figure 12.3.

Vt∠q
E = E∠0
jXL

Generator Peg

Figure 12.2  Single-machine Connected Infinite Bus System: Small-signal Analysis.

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 507 9/12/2017 3:24:03 PM

 508  |  Power System Analysis

I jX′d jXL

E′q∠d E∠0

Figure 12.3  Equivalent Circuit of SMIB System Employing Classical Model for Generator.

Neglecting the stator and rotor winding, and the network dynamics, the only dynamic equations are
due to the mechanical system. Writing the swing equation in slip form, we have,

dd
= Smw B (12.1)
dt
dSm
2H = − D Sm + Tm − Teg (12.2)
dt

Note that in the above equation, all overbars have been removed for ease of representation and Tm is
set to Tm′ , neglecting the steady-state damping component. The state-vector is given by x = [d Sm ]t .
Further,
Eq′ E sind
Teg = (12.3)
x ′ d1
with xd′1 = x ′ d + x L
The operating point, xo is evaluated by setting x = 0 . Hence, we get,

Sm = 0
and
T 
Teg = Tmax sin d = Tm0 ; This implies d os = sin −1  m0  (12.4)
 Tmax 
Eq′ E
where Tmax =
xd′1

Thus, we have two possible equilibrium points (see Figure 12.4). They are given by

xoA = [d 0s 0]t ; xoB = [d 0u 0]t

u s
Note that d 0 = p − d 0.

Now, linearising the swing equation around the operating points, we have,

d ∆d
= ∆Smw B (12.5)
dt

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 508 9/12/2017 3:24:06 PM

 Power System Angle Stability  |  509 

1.4
Teg

1.2

1 A B

Tm0
0.8

0.6
Teg

0.4

d s0 d u0
0.2

0
0 20 40 60 80 100 120 140 160 180
d in (deg)

Figure 12.4  Transient Power Angle Curve.

d ∆Sm
2H = − D ∆Sm + ∆Tm − ∆Teg (12.6)
dt

Assuming Tm to be a constant, we have

d ∆d
= ∆Smw B (12.7)
dt

d ∆Sm
2H = − D ∆Sm − ∆Teg (12.8)
dt
From eq. (12.3), we have

∆Teg = K S ∆d (12.9)
where

∂Teg Eq′ E cosd 0

KS = |d 0 = (12.10)
∂d xd′ 1

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 509 9/12/2017 3:24:09 PM

 510  |  Power System Analysis

Note that Ks is referred to as synchronising torque coefficient and is evaluated at two operating points,
d 0 = d 0s or d 0 = d u0 .
Writing the linearised swing equation in state-matrix form, we have,

 ∆d   0 w B
 ∆d 
  =  KS D  (12.11)
 m   −
∆S −   ∆Sm 
 2H 2H 
To determine the operating point stability, the eigenvalues of the state matrix are evaluated as,

D K  D 
2
1
l1,2 = − ± j  S −   (12.12)
2H  M  2H  4
2H
with M =
wB
Note that for the system to be stable at the operating point, xo , we know that all Re( li ) < 0. This
implies that
D > 0 and Ks > 0
where D represents the mechanical damping due to viscous friction which is generally small and is
always greater than zero.
Therefore, we can conclude that
• At xoA, Ks > 0 and it is a stable equilibrium point.
• At xoB , Ks < 0 and it represents an unstable equilibrium point.
T
When the system is at xoA , the deviations [ ∆d (t ) ∆Sm (t )] exhibit pure oscillations (from their ini-
0 0 T
tial values [ ∆d ∆Sm ] ) with D = 0. The angle deviations result in torque deviations which are purely
synchronising-kind. A positive Ks represents a restraining force which tries to keep the machine in
synchronism. However, at xoB , the restraining force is negative, resulting in eigenvalues which are
purely real, with one them having a positive value due to a small D. This results in monotonic increase
of angle deviation ending up with loss-of-synchronism.

12.1.2  A Mechanical Analogy

Consider a linear spring-mass system as shown in Figure 12.5.

x
K
F M

Friction is zero

Figure 12.5  A Single Spring-mass System.

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 510 9/12/2017 3:24:13 PM

 Power System Angle Stability  |  511 

Writing the differential equation governing the natural response of system, we have,

d2x
M + K x = 0 (12.13)
dt 2
where M is the mass of the system in kg and K represents the stiffness coefficient in N/m.
To obtain the natural frequency of oscillation of the system, we can write the characteristic
­equation as
 2 K
 l +  = 0
M

Thus, the natural frequency of oscillation in rad/s is given by

K
l1,2 = ± j  
M

The structure of the above equation is identical to that given in eq. (12.12) with D = 0. In this respect,
the following points can be noted:
1. Since the spring-mass system is inherently linear, it does not involve linearisation of equations.
However, to predict the operating point behaviour of a power system through the eigenvalue anal-
ysis, the swing-equations should be linearised around the operating point since the equations are
non-linear in nature. Such a process establishes a state matrix using which the behaviour of the
state trajectories emerging from the operating point (for a small perturbation) can be predicted
without obtaining the time-domain solutions.
2. With D = 0 and DTeg = KSDd, substituting eq. (12.7) in eq. (12.8), we have

2 H d 2 ∆d
= − K S ∆d (12.14)
w B dt 2

Note that the above expression is identical to that given in eq. (12.13). Thus, the natural frequency
of oscillation of the rotor in rad/s is given by

K w 
l1,2 = ± j  S B  (12.15)
 2H 

This represents the frequency of the swing-mode, Ω.

3. In the spring-mass system, K represents a linear spring which offers restraining force whenever
there is displacement, x. In the linearised model of the power system, Ks also offers a restraining
force proportional to the angular displacement, Dd.
4. In the spring-mass system, since the system is linear, the stiffness coefficient K remains constant,
independent of the initial displacement, if any. However, in the power system example, the value
of Ks depends on the initial ‘stretch’ d0.
5. In the spring-mass system, the force due to the spring is always restraining type (even for large
p
displacements) as the system is linear. In power system case, if 0 ≤ d 0 < , the force due to the
2

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 511 9/12/2017 3:24:15 PM

 512  |  Power System Analysis

p
equivalent spring is restraining, whereas when d 0 > , the restraining force is lost or can be
2
treated as negative and causes a monotonic increase of angle displacement.
6. It should be noted that when a disturbance is large enough to cause a large angle deviation, the
torque produced is no longer linearly related to angle deviation. In such cases, the restraining
torque should be treated as a non-linear spring. This is mainly due to the fact that a power system
is inherently non-linear.

12.2  SYNCHRONISING AND DAMPING TORQUE ANALYSIS

Assuming that the rotor angle and the speed deviations oscillate sinusoidally, the phasor notations can
be used to analyse the stability performance of power systems at a given swing-mode frequency Ω.
We know that,
dd
= Smw B (12.16)
dt
and
d ∆d
= ∆Smw B (12.17)
dt
For sinusoidal oscillations,
jWDd (jW) = DSm ( jW)wB

This shows that the speed deviation phasor leads the angle deviation phasor by 90°. These phasors
are shown in Figure 12.6.

∆Sm( jΩ)

∆TegD( jΩ) ∆Teg( jΩ)

FL
∆d ( jΩ)
∆TegS( jΩ)

Figure 12.6  Phasor Representation of Sinusoidally Varying Angle, Speed and Torque Deviations.

In Figure 12.6, the phasor DTeg ( jW) denotes a sinusoidally varying DTeg. If fL is the angle between
this torque phasor and the speed phasor, the component of torque in phase with DSm ( jW) is given by

DTegD (  jW) = DTeg ( jW)cosfL (12.18)

and the component of torque in phase with Dd  ( jW) is given by

DTegS ( jW) = DTeg ( jW)sinfL (12.19)

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 512 9/12/2017 3:24:16 PM

 Power System Angle Stability  |  513 

Notes:
1. The component of torque in phase with the speed deviation, DTegD (jW) is referred to as the
damping torque component.
2. The component of torque change in phase with the rotor angle perturbation, DTegS (jW) is referred
to as the synchronising torque component.
3. In the SMIB system with classical model for the generator, DTeg is purely of synchronising type
p
and does not possess any damping component. Also, the angle fL = .
2
From eqs. (12.18) and (12.19), it can be seen that if DTegMax is the amplitude of the DTeg (jW)-phasor,
then the damping torque phasor will have an amplitude equal to DTegMaxcosfL and the synchronising
torque phasor will have an amplitude DTegMaxsinfL. Further, if DdMax is the amplitude of the angle
deviation and DSm is the amplitude of the speed deviation, we can write the unit amplitudes of the
Max
in-phase torque components in terms of the angle and speed deviations as

∆TegMax cosfL
KD = (12.20)
∆SmMax

∆TegMax sinfL
KS = (12.21)
∆d Max

Using the above coefficients, the instantaneous value of torque deviation can be obtained as

DTeg = KSDd + KD DSm (12.22)

where
• KSDd is the component of torque change in phase with the rotor-angle deviation, and Ks is referred
to as the synchronising torque coefficient.
• KD DSm is the component of torque in phase with the speed deviation, and KD is referred to as the
damping torque coefficient.
As there is only one frequency which corresponds to the swing-mode, in the SMIB system with classical
model for the generator, neglecting mechanical damping and using eq. (12.22) in eq. (12.8), we get

d ∆Sm
2H = − K S ∆d − K D ∆Sm (12.23)
dt

The following observations are made:

1. We can see that for the system to be small-signal stable Ks > 0 and KD > 0. This implies that if
either or both damping and synchronising torques, i.e., if DTegD (jW) and/or DTegS (jW), are in
phase opposition to the corresponding angle/speed deviation phasors, the system is unstable. A
negative damping torque implies that the response will be in the form of growing oscillations, and
a negative synchronising torque implies monotonic instability.

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 513 9/12/2017 3:24:17 PM

 514  |  Power System Analysis

2. A higher value of the synchronising torque coefficient denotes that the generator is strongly
coupled to the system. A small angle deviation produces enough restraining force to keep the
generator in synchronism. In other words, if Ks is large, the angle deviations are small.
3. From eq. (12.10), it can be seen that the Ks increases if the initial power-angle, d0, is smaller. Also,
note that, for a given xd′, the value of Ks varies inversely with the line reactance, xL.
4. A higher value of the damping torque coefficient implies that the resulting oscillations damp out
faster as it acts similar to D.

Note: In this analysis, the angle fL is measured taking DSm ( jW) phasor as the reference and is treated
as positive for lagging angle.

12.3 EFFECT OF A FAST-ACTING HIGH-GAIN STATIC

EXCITER ON SMIB SYSTEM
It is generally observed that when a fast-acting, high-gain static exciter is used for a generator, it
may destabilise the swing mode, depending on the loading level and the strength of the supporting
system. If the reactance of the transmission line connecting the generator to the system is very
large, the generator is said to be weakly coupled to the system. In such cases, if the generator is
delivering a large power, an exciter which is provided to regulate the terminal voltage very rigidly
and quickly, may tend to establish a negative damping torque which makes the swing mode unsta-
ble. Using the damping and synchronising torque analysis, we can easily depict this nature of the
static exciter (see Figure 12.7).

∆Sm( jΩ)
∆TegD( jΩ)

∆Teg1( jΩ)

∆TegS( jΩ)

∆d ( jΩ)
∆TegE( jΩ)

Figure 12.7  Torques Deviation Phasors with and without the Static Exciter.

In Figure 12.7, DTeg1 (jW) denotes the torque deviation phasor without the exciter and DTegE (jW)
represents the torque deviation phasor with a fast-acting and high-gain static exciter. It is clear that

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 514 9/12/2017 3:24:18 PM

 Power System Angle Stability  |  515 

without the exciter, both synchronising and damping torque components are positive. When the
static exciter is enabled, although the synchronising torque component is positive, the damping
torque become negative. The amplitude of this component is large enough to nullify the inherent
positive damping associated with the system without the exciter. This causes the swing mode to
become unstable.
Since fast-acting and high-gain static exciters with high field-forcing levels are required from the
point of view of better steady-state behaviour and large disturbance performance of a system, as a
remedy, it is necessary to introduce some damping controllers into the system to improve the small-
signal performance of the system. In this context, network controllers such as FACTS systems, HVDC
controllers, and local controllers such as power system stabilisers (PSS) are the possible options. Of
these, PSS is a generally preferred option as it provides a cost effective way to improve the damping
of electromechanical oscillations.

12.3.1  PSS Location and its Design

PSS is an auxiliary controller which achieves the damping job by simply modulating the VRef of
the excitation controller (see Figure 12.8).

PT
Measurement
block

Main th
Vg Field i machine
−
VRef Power
Σ AVR Exciter system
+
+/−
VS

PSS
PSS input signal(s)

Figure 12.8  Location of PSS in a Power System.

It can be seen from Figure 12.8 that the necessary power amplification of the signal is provided by
the existing exciter circuitry itself. It is well known that a modulation in the terminal voltage of the
generator inherently causes a modulation in the developed torque. This, in turn, influences the rotor
swings, and hence provides good controllability of the swing modes. Apart from this, for an effective
control of a mode, such a mode must be observable to an acceptable level in the input signal chosen
for the PSS. In this regard, the generator slip, power output of the generator, and the bus frequency
are the candidate PSS input signals. In the following lines, the slip signal is used for demonstrating
the design of a PSS.
To have sufficient level of damping torque component in the total torque deviation, it is necessary to
p
see that the DTeg (jW)-phasor lags or leads the DSm (jW)-phasor by an angle fL < -see Figure 12.6.
2
To ensure this, it is required to know the inherent phase angle (lag) introduced by the signal path, i.e.,

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 515 9/12/2017 3:24:19 PM

 516  |  Power System Analysis

exciter-generator-power systems, for a signal DVs(t), of known frequency, Ωm. The resulting torque
deviation, DTeg(t) will also have the same frequency variation as that of DVs(t), but with a phase angle
lag bm (see Figure 12.9). This is mainly due to the predominantly inductive nature of the components,
such as exciter, generator field circuits, and so on. Further, it should be noted that while evaluating the
DTeg(t)-waveform, care should be taken to see that the influence of rotor-angle and speed deviations
on the torque deviation is not accounted as we desire to obtain the variation of DTeg(t) only due to
DVs(t)-signal.

PT
Measurement
block

Main
Vg Field ith machine
−
VRef
Power
Σ AVR Exciter system
+
+/−
with ∆d and ∆w = 0

∆V S(t) = Vsmsin(Ωmt) ∆Teg(t) = Nmsin(Ωmt − bm)

Figure 12.9  Determination of DTeg(t) due to DVs(t).

Notes:

1. If such a DTeg(t) is determined by a time-domain simulation run, the effect of rotor-angle and
speed deviations on torque variations can be removed by setting the inertia of the machine(s) to a
very large value.
2. The angle information about b is generally obtained analytically by evaluating a transfer function
∆Teg ( s)
of a system consisting of exciter-generator-power system with Dd  (s) = DSm(s) = 0. This
∆Vs ( s)
transfer function is referred to as GEP(s) [1, 2]. A typical plot of the angle b = ∠GEP ( jw) is as
shown in Figure 12.10. Here, the frequency, w, is varied between 0.63 and 22 rad/s (i.e., 0.1–3.5 Hz)
to include the possible swing-mode frequency.

If the torque deviation produced due to DVs-signal is required to have an adequate level of in-phase
component with the speed deviation (see Figure 12.6), it is necessary to compensate the angle bm by a
lead angle fm by using a lead compensator Gc(s) in the PSS. For simplicity, the compensator transfer
function is assumed to be of the form given by

Gc ( s) =
(1 + sT1 ) (12.24)
(1 + sT2 )

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 516 9/12/2017 3:24:20 PM

 Power System Angle Stability  |  517 

Phase angle of GEP

0

−10

−20

−30

−40
Phase−deg

−50

−60

−70

−80

−90

−100
0 0.5 1 1.5 2 2.5 3 3.5 4
Frequency in Hz

Figure 12.10  Phase angle b of GEP ( jw ).

12.3.1.1  Design of a Lead Compensator

The phase angle lead fm to be provided by the compensator is related to T1 and T2 as [3],

1−a
sinf m = (12.25)
1+ a
where

T2
a = with 0 < a < 1, (12.26)
T1

Further, the centre frequency at which it offers a phase lead fm, is given by

1
Wm= (12.27)
a T1

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 517 9/12/2017 3:24:22 PM

 518  |  Power System Analysis

Choosing fm = 20° and fm = 3 Hz, (with Ωm = 2πfm), and using eqs (12.25), (12.26) and (12.27), we
T
get, T1 = 0.07577 s and T2 = 0.03715 s with 1 = 2.0396 . The phase angle of the compensator is
T2
shown in Figure 12.11.

Phase angle of compensator

20

18

16

14

12
Phase−deg

10

0
0 0.5 1 1.5 2 2.5 3 3.5 4
Frequency in Hz

Figure 12.11  Phase Angle of the Compensator GC ( jw ).

From Figure 12.11, it can be seen that although the lead angle fm = 20° occurs at Ωm, the angle
value is considerable in the neighbourhood of Ωm. Assuming that the change in the swing-mode
frequency with the introduction of a PSS is small, if we select Ωm = Ω, i.e., the swing-mode fre-
quency itself, and if fm = bm, then fL = 0 and the torque deviation produced is purely of damping
kind without contributing anything to the synchronising torque. In order to see that a damping
controller not only introduces a damping torque component but also causes an improvement in the
inherent synchronising torque component, it is ensured that the lag angle fL is greater than zero and
is below 45°. In this regard, the following guidelines are generally employed while choosing the
compensator parameters [1]:

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 518 9/12/2017 3:24:22 PM

 Power System Angle Stability  |  519 

1. The centre frequency fm is chosen in a range (2.5–3.5 Hz) which is higher than the desired swing-
mode frequency (which typically lies in the range of 0.2–2 Hz).
2. The phase angle lead fm is so chosen that the maximum compensated lag angle, (b - fm) remains
p
less than at around 3 Hz. This ensures that other swing-modes, if any, are not destabilised.
2
T
3. Typically 1 must be less than 10 to prevent noise amplification. If a large phase angle lead is
T2
required, then compensators may be used in cascade.

In addition, it is required that a damping controller does not interfere with the steady state and regular
functions of the exciter controller. In this context, the following blocks are added to the compensator
transfer function indicated in eq. (12.24).

1. A wash-out circuit which is a high pass filter. Its output is zero under steady state and this block
removes the DC-offset in the input signal, passing only the transient variation in the input signal.
2. A limiter block which limits the level of modulation of the VRef.

Thus, a typical PSS block appears as shown in Figure 12.12. The gain block denotes the PSS gain
which can be varied for the desired level of damping of the swing-mode [2]. However, the gain should
not be too large as it may destabilise some other mode of oscillation [1]. Note that the output of
the PSS, i.e., the Vs-signal, is connected to the exciter VRef -summer block with a positive sign for a
slip-input PSS. This is because for a positive value of PSS gain, the damping torque component of
DTeg will be in-phase with the DSm-signal. However, from the swing-equation, it is clear that DTeg acts
oppositely with respect to speed-deviation. Therefore, with a positive Vs-signal, if speed-deviation
increases, the DTeg causes it to decrease, and vice-verse.

sTw 1 + sT1 Vs
Slip KPSS
input 1 + sTw 1 + sT2

Wash-out Phase−lead Gain

circuit Limiter
compensator

Figure 12.12  Transfer Function Representation of a PSS Block.

For the PSS block shown in Figure 12.12, a typical curve for the compensated angle, is as shown in
Figure 12.13. For the wash-out circuit, Tw is set at 10 s. From Figure 12.12, it can be seen that for a
frequency below 2 Hz, the compensated angle is below 40°.
Employing the torque deviation phasor diagram, we can show how a properly tuned PSS overcomes
the negative damping torque introduced by a fast-acting and high-gain static exciter (see Figure 12.14).
In Figure 12.14, DTeg3 ( jW) denotes the torque deviation produced by a PSS and DTegE ( jW) represents
the torque deviation phasor with the static exciter. Also, note that the resulting torque deviation phasor
DTegR ( jW) now possesses a positive damping torque component.

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 519 9/12/2017 3:24:23 PM

 520  |  Power System Analysis

Phase angle response

80

60

40
PSS

20
Angle (deg.)

−20
Compensated GEP

−40

−60

GEP
−80

−100
0 0.5 1 1.5 2 2.5 3 3.5 4
Frequency (Hz)

Figure 12.13  Compensated Angle of the GEP( jw ).

∆Sm( jΩ)

∆Teg3( jΩ)
∆TegD( jΩ)

FL
∆Teg1( jΩ)

∆TegR( jΩ) ∆TegS( jΩ)

∆d ( jΩ)

∆TegE( jΩ)

Figure 12.14  Torques Deviation Phasors with PSS.

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 520 9/12/2017 3:24:25 PM

 Power System Angle Stability  |  521 

12.4  SMALL-SIGNAL STABILITY ANALYSIS OF AN SMIB SYSTEM

In the following section, the small-signal stability of the operating point is studied for different models
for the generator.

12.4.1 Example: Small-signal Stability of SMIB System

with 0.0 Model for Generator
Consider an SMIB system as shown in Figure 12.15. The generator has x′d = 0.32 pu, H = 5 MJ/MVA,
D = 0, and wB = 2π60 rad/s.

Vt = 1.0∠q Eb = 1.0∠0 pu
jXL = 0.4 pu

Generator Pg

Figure 12.15  SMIB System Used for Analysing Small-signal Stability.

The following steps are used to calculate the initial operating point:
1. Compute q0 using

Vt 0 Eb
Pg 0 = sinq 0 (12.28)
xL
2. Compute line current as

I0 =
(Vt 0 − Eb )
(12.29)
jx L
3. Compute Eq′ 0 as

Eq′ 0 = Eq′ 0 ∠d 0 = Vt 0 + jxd′ I 0 (12.30)

For Pg0 = 1.0, we obtain,

Vt 0 = 1.0 ∠ 23.57o pu , I 0 = 1.0215∠ 11.78o pu and Eq′ 0 = 1.1137∠40.276o pu

The differential equations are given by,

dd
= Smw B (12.31)
dt
dSm
2H = Tm − Teg (12.32)
dt

Eq′ 0 Eb
where Teg = sin d and Tm is assumed to be constant at Tm0 = Pg0, neglecting losses.
( x′d + xL )

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 521 9/12/2017 3:24:28 PM

 522  |  Power System Analysis

The linearised state equations are given by

 ∆d   0 wB 
∆d 
  =  Ks  (12.33)
 
 ∆Sm   − 0   ∆Sm 
 2H 
E q′ 0 Eb
where K s = cos d 0 .
( xd′ + x L )

At the operating point, the state-matrix is given by

 0 376.9911
A( 2× 2) =   (12.34)
 −0.1180 0 
The eigenvalues of the system at the operating point are
0 ± j 6.6702
In order to understand the existence of this oscillatory mode, for Pg0 = 1.0 pu, a three-phase fault is
applied at t = 1 s for a small duration of 0.01 s at the infinite bus. The corresponding rotor angle plot
is shown in Figure 12.16. It is clear from Figure 12.16 that the swings are purely oscillatory and the
classical model for the generator does not offer any damping of electrical origin. Figure 12.16 shows
even the frequency of the oscillation which is verified by noting the time period of the waveform.

0.8
∆ t = (3.128−2.1855) = 0.9425 s
0.78 w = 2*p * (1/∆ t) = 6.6665 rad/s

0.76

0.74

0.72
d (rad)

0.7

0.68

0.66

0.64

0.62

0.6
0 2 4 6 8 10 12 14 16 18 20
Time (s)

Figure 12.16  The Plot of d for Pg0 = 1.0 pu with 0.0 Model for the Generator
with a Three-phase Fault.

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 522 9/12/2017 3:24:31 PM

 Power System Angle Stability  |  523 

Notes:
1. While verifying the modal performance of a system through time-domain simulations, care
should be taken to see that the perturbation applied is small enough to cause a non-linear system
to produce deviations in the state trajectories which are smaller in amplitude. This ensures, to
some extent, the linear approximation of the system around the operating point.
2. If an operating point is unstable, such a time-domain simulation, if run for longer time, may trig-
ger non-linear behaviour. This is especially true when control limits are hit.

12.4.1.1  Small-signal Oscillation for Initial Condition Perturbation

Here, the swing mode performance of the above system predicted using the eigenvalue is checked
by perturbing the initial value of the d -state instead by applying a small duration fault. Here, before
starting the time-domain simulation, the d0 value is set as d0 + 0.01 = 0.71284 rad. With this perturbed
value of the initial state, the trajectory of the rotor angle deviation is plotted in Figure 12.17. Similar
observations can be made as noted with a three-phase fault.

0.02

0.015
(2.826 − 1.884) = 0.942 s

0.01
∆d (rad)

0.005

−0.005

−0.01
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (s )

Figure 12.17  The plot of Dd for Pg0 = 1.0 pu with Initial Condition Perturbation.

12.4.1.2  Small-signal Oscillation for Mechanical Input Perturbation

Here, instead of perturbing the initial condition of a state, the mechanical torque input to the generator
is perturbed by a small amount equal to DTm = 0.05 pu. The following two operating points are con-
sidered for analysis (see Table 12.1):

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 523 9/12/2017 3:24:32 PM

 524  |  Power System Analysis

Table 12.1  Operating Conditions Considered for the 0.0 Model.

Pg0 E′q0 d0 (deg.) Ks k (rad/s)

1.0 1.1137 40.276 1.1802 6.6702
0.5 1.0287 20.485 1.3384 7.1032

In the first case, T ′m0 (=Tm0 + DTm ) is set to 1.05 pu and for the second case, it is set to 0.55 pu.
When a time-domain simulation is carried out, in each case, the Dd  (t) plot is obtained as shown in
Figure 12.18.

0.1
Tm0 = 1.0
0.09 Tm0 = 0.5

0.08

0.07

0.06
∆ d (rad)

0.05

0.04

0.03

0.02

0.01

0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (s )

Figure 12.18  The Plot of Dd with Tm0 Perturbation.

From Figure 12.18, it can be seen that when the power level is halved, the maximum deviation in the
rotor angle is lower than that observed with Pg0 = 1.0 pu for a given magnitude of perturbation. This
clearly demonstrates the improvement in the synchronising torque coefficient as tabulated in the table.
Also, note that at the lower power level, the swing-mode frequency is higher than that at Pg0 = 1.0 pu.

12.4.2 Example: Small-signal Stability of SMIB

System with 1.0 Model for Generator
The generator parameters are given below and the system is identical to that given in Figure 12.15.
xd = 1.6, xq = 1.55, xÄd = 0.32, Tdo
′ = 6 s, Ra = 0, H = 5 MJ/MVA, D = 0, and wB = 2p 60 rad/s.

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 524 9/12/2017 3:24:33 PM

 Power System Angle Stability  |  525 

The system equations are as follows:

1. The stator voltage equations are given by

nd = - yq

nq = yd(12.35)

where,
yq = xqiq

y d = x ′d id + Eq′ (12.36)

( xd − xd′ )
with Eq′ = C1YF and C1 =
xd

2. The network voltage equation in the machine frame is given by

vq = Eb cosd − id x L

vd = − Eb sind + iq x L (12.37)

Using eq. (12.37) in eq. (12.35) and solving for the machine currents, we get,

Eb cos d − Eq′
id = (12.38)
( xd′ + x L )

Eb sin d
iq = (12.39)
( xq + x L )
3. The differential equations are

dd
= Smw B (12.40)
dt

dSm
2H = Tm − Teg (12.41)
dt

dy F 1  xd′ 
=  −y F + y d + E fd  (12.42)
dt T d′  ( xd − xd′ ) 

where Teg = (ydiq - yqid) and Tm is assumed to be constant.

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 525 9/12/2017 3:24:36 PM

 526  |  Power System Analysis

T
The state vector is x = d Sm y F  . Hence, the deviation in the state vector is given by ∆x =  ∆d ∆Sm ∆y
T
∆x =  ∆d ∆Sm ∆y F  . Linearising the above set of equations, we obtain,

1. From eqs. (12.38) and (12.39), we get,

∆id = B1∆x

∆iq = B3 ∆x (12.43)

 − E sin d 0 −C1   E cos d 0 

B1 =  b , 0,  and B3 =  b , 0, 0 
 ( xd′ + x L ) ( xd′ + x L )   ( xq + x L ) 

2. ∆Eq′ = B2 ∆x with B2 = 0, 0, C1  .

3. From eq. (12.36), we get,

∆y d = B4 ∆x

∆y q = xq B3 ∆x
(12.44)
with B4 = ( xd′ B1 + B2 )

4. Linearising the torque equation, we get,

∆Teg = y d 0 ∆iq + iq0 ∆y d −y q0 ∆id − id 0 ∆y q

Now using eqs (12.43) and (12.44) we have,

∆Teg = y d 0 B3 ∆x + iq0 B4 ∆x −y q0 B1∆x − id 0 xq B3 ∆x

= B5 ∆x (12.45)

(
where, B5 = y d 0 B3 + iq0 B4 −y q0 B1 − id 0 xq B3 )
5. Now linearising the differential equations, we get,

d ∆d
= A1 ∆x (12.46)
dt

d ∆Sm
= A2 ∆x (12.47)
dt

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 526 9/12/2017 3:24:41 PM

 Power System Angle Stability  |  527 

d ∆y F
= A3 ∆x (12.48)
dt
with

−1  1 
A1 = 0, w B , 0  ; A2 = B5 and A3 =  B6 + B4 
2H  Td′ 

 1 x′
′ d
where B6 = 0, 0, −  and Td′ = Tdo
 Td′  xd

The final state matrix is given by

∆x = A(3×3) ∆x

or

 ∆d   A   ∆d 
   1(1×3)    (12.49)

 ∆Sm  =  A2(1×3)   ∆Sm 
 ∆y   A 
 F   3(1×3)   ∆y F 

Note that in the above case the generator is assumed to be on manual field control. The eigenvalue of
A(3×3) will provide the small-signal stability performance of the system at Pg0 = 1.0. The initial value
of some of the quantities is indicated in Table 12.2.

Table 12.2  Initial Value of Variables: 1.0 Model.

p0 c0 xd0 xq0 id0 iq0

23.57° 73.085° 0.6494 0.7605 −0.8960 0.4906

The state matrix is obtained as

 0 376.9911 0 
 
A(3×3) =  −0.1106 0 −0.1063  (12.50)
 −0.3543 0 −0.4630 

The eigenvalues of the system at the operating point are

−0.1702 ± j 6.4518 and − 0.1225

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 527 9/12/2017 3:24:44 PM

 528  |  Power System Analysis

1.34

∆t = (3.207 − 2.2315) = 0.9755 s

1.32
w = 2*p * (1/∆t) = 6.4410 rad/s

1.3
d (rad)

1.28

1.26

1.24

1.22
0 2 4 6 8 10 12 14 16 18 20

Time (s)

Figure 12.19  The Plot of d for Pg0 = 1.0 with 1.0 Model for the Generator.

For Pg0 = 1.0 pu, a three-phase fault is applied at t = 1 s for a period of 0.01 s at the infinite bus. The
corresponding rotor angle plot is shown in Figure 12.19. It is clear from Figure 12.19 that the rotor
oscillations damp out as predicted by the eigenvalues. This is mainly due to the fact that the field
dynamics produce torque deviations which possess damping torque components.

12.4.3 Example: Small-signal Stability of SMIB System

with 2.2 Model for Generator
Consider an SMIB system as shown in Figure 12.20.

V = 1.0∠q E = 1.0∠0pu
jxL = 0.7 pu

Generator Pg

Figure 12.20  Generator with Detailed Model Connected to Infinite Bus.

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 528 9/12/2017 3:24:45 PM

 Power System Angle Stability  |  529 

The standard parameters of the machine are given in Table 12.3.

Table 12.3  Standard Parameters for a 865 MVA, 22 kV Generator in pu on the Machine Base.

xd = 1.79 x ′d = 0.169 x ″q = 0.135

xq = 1.71 x ′q = 0.228 x ″q = 0.2
T ′do = 4.3 s Têdo = 0.032 s
T ′qo = 0.85 s Têqo = 0.05 s
Ra = 0.01 H = 2.86 s 60 Hz

12.4.3.1  Modal Performance without AVR

For the generator, 2.2 model has been employed. Linearising the stator and rotor equations, we get,

 ∆vd 
∆xe (6×1) =  AG  ∆xe (6×1) + [CWm ](6× 2)   +  Fg  ∆E fd (1×1) (12.51)
( 6 × 6)  ∆vq  ( 2×1) (6 ×1)

where
xe (6×1) = [y d y q y F y H y G y K ]T

The above matrices are defined in the generator short-circuit example presented in chapter 5.

Swing-equation:
The swing-equation is used in the following form:

dd

dt
( )
= w y − 1 w B (12.52)

dw y
2H = − D w y + Tm − Teg (12.53)
dt
where Teg = (ydiq - yqid) and Tm is assumed to be constant and D is set to zero.

Linearising Teg, we get,

 ∆y d   ∆id 
∆Teg = iq0 −id 0    +  −y q0 y d 0   ∆i 
 ∆y q  q

∆Teg (1×1) = [Ce 2 ](1×6) ∆xe (6×1) + [Ce3 ](1× 2) [Ce1 ]( 2×6) ∆xe (6×1)

where
[Ce 2 ](1×6) = iq0 −id 0 0 0 0 0 

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 529 9/12/2017 3:24:47 PM

 530  |  Power System Analysis

and

[Ce3 ](1× 2) =  −y q0 y d 0 

with yq0,  yd0,  iq0, and id0 known from the initial condition calculations for a given Pg0. Note that
[Ce1] is defined earlier in the generator short-circuit example (chapter 5).
Now, we have,

∆Teg (1×1) = [Ce ](1×6) ∆ xe (6×1) (12.54)

where
[Ce ](1×6) = [Ce 2 ](1×6) + [Ce3 ](1× 2) [Ce1 ]( 2×6)

T
Now, linearising the swing equation with xs =  ∆d ∆w y  , we get,

∆xs ( 2×1) = [ Ams ]( 2× 2) ∆xs ( 2×1) + [ Ame ]( 2×6) ∆xe (6×1) (12.55)
where
   
0 w B  [ 0 ](1×6) 
   
[ Ams ](2×2) =  [ Ame ](2×6) = 
 D   −1 
0 −


2H 

 2H
[ Ce ](1×6) 


Network-equation:
Neglecting transients, the network equation is given by

V = E ∠0 + jIx L (12.56)

Writing the above equation in machine frame-of-reference, we get,

vq + jvd = Ee − jd + j (iq + jid ) x L (12.57)

Comparing real and imaginary parts on both sides in the above equation, we obtain,

vq = E cosd − id x L (12.58)

vd = − E sind + iq x L (12.59)
Linearising vq and vd , we get

 ∆vd 
 ∆v  = [ Avs ]( 2× 2) ∆xs ( 2×1) + [ Ave ]( 2×6) ∆xe (6×1) (12.60)
 q  ( 2×1)

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 530 9/12/2017 3:24:51 PM

 Power System Angle Stability  |  531 

where

 − E cos d 0 0  0 xL 
   
[ Avs ](2×2) =  [ Ave ](2×6) =  [Ce1 ]
 − E sin d 0 0   − x L 0 

Now, using eq. (12.60) in eq. (12.51), we get

∆xe (6×1) = [ AG ](6×6) ∆xe (6×1) + [CWm ](6× 2) [ Avs ]( 2× 2) ∆xs ( 2×1)

+ [CWm ](6× 2) [ Ave ]( 2×6) ∆xe (6×1) +  Fg  ∆E fd (1×1) (12.61)

(6 ×1)

Now, from eqs (12.61) and (12.55), we can combine the state equations as

∆ x es(8×1) = [ AES ](8×8) ∆xes (8×1) +  FgES  ∆E fd (1×1)

(8×1) (12.62)

where

 ∆xe (6×1) 
 
∆xes (8×1) =  
 ∆xs ( 2×1) 
 

and

  Fg  
[ AG ] + [CWm ][ Ave ] [CWm ][ Avs ]    (6×1) 
[ AES ] =  
  FgES  = 
  


 [ Ame ] [ Ams ]   [ 0 ]( 2×1) 
 

Since the field controller is not considered in the current case, ∆Efd = 0. Further, stator transients are
dy d dy q
neglected by setting = 0 and = 0 . To obtain the corresponding state-matrix for eigenvalue
dt dt
analysis, we can rewrite eq. (12.62) using only [AES] as

  ∆y dq   [ AES11 ]( 2× 2) [ AES12 ](2×6)    ∆y dq  

 ( 2×1) 
    ( 2×1) 
  =    (12.63)
   
 [ ∆xrs ](6×1) 
 A
[ ES 21 ](6×2) [ AES 22 ](6×6) (8×8)  [ ∆xrs ](6×1) 
  (8×1)    (8×1)

T
where y dq = y d y q  and xrs = [y F y H yG y K d w y ]T

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 531 9/12/2017 3:24:54 PM

 532  |  Power System Analysis

Since ∆y = 0 , we can write from eq. (12.63) that

dq

 ∆y dq  = − [ AES11 ]−1
( 2 × 2) [ ES12 ]( 2 ×6) [ rs ]( 6 ×1) (12.64)
  A ∆x

From eq. (12.63), we can also write that

 ∆ x rs  = [ AES 21 ](6× 2)  ∆y dq  + [ AES 22 ](6×6) [ ∆xrs ](6×1) (12.65)

(6 ×1) ( 2 ×1)

Using eq. (12.64) in eq. (12.65), we have the reduced state-model given by

( 6 ×1) [ ESR ]( 6 ×6) [ rs ]( 6 ×1)

 ∆ x rs  = A ∆x (12.66)
where

[ AESR ] = − [ AES 21 ][ AES11 ]−1 [ AES12 ] + [ AES 22 ]

The corresponding eigenvalues are given in Table 12.4. It is clear from Table 12.4 that as the load-
ing level increases, the damping for the swing-mode (p23) reduces. This observation is true for the
frequency of oscillation as well.

Table 12.4  Eigenvalues for Different Loading Levels without Static Exciter.
Number Pg0 = 0.1 Pg0 = 0.5 Pg0 = 0.9
p1 −32.354 −32.303 −32.278
p23 −0.92723 ± j 8.3376 −0.4869 ± j 8.2855 −0.35148 ± j 7.513
p4 −0.57906 −0.2423 −0.07081
p5 −1.3911 −2.6364 −3.0925
p6 −20.189 −20.213 −20.224

To know the association of a mode to a state, the participation factors are obtained for Pg0 = 0.9 pu and
are given below. From the list, it can be seen that the states d and wy have the highest participation in
mode p23. Hence, this mode is referred to as swing-mode.

p1 p2 p3 p4 p5 p6     State

0.0102 0.0383 0.0383 0.9826 0.0016 0     

–   F
0.9944 0.0066 0.0066 0.0167 0 0     
–  H
0 0.0017 0.0017 0.0019 0.9594 0.0404 –  G
0 0.0001 0.0001 0.0001 0.0406 0.9595 –  K
0.0023 0.5023 0.5023 0.0007 0.0008 0     
–  delta
0.0023 0.5023 0.5023 0.0007 0.0008 –  omega
0     

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 532 9/12/2017 3:24:56 PM

 Power System Angle Stability  |  533 

Time-domain Simulation without AVR:

Neglecting stator transients and rotor speed deviations, the stator voltage equations are compactly
written in matrix form as
 Ra xq′′  id   Ed′′ − vd 
= (12.67)
 − x ′′
 d Ra  iq   Eq′′ − vq 
In the above equations, using vd and vq from eqs (12.59) and (12.58), we get

 Ra ( xq′′ + x L )  id   Ed′′ + E sin d  (12.68)

=
 −( x ′′ + x )
 d L Ra  iq   Eq′′ − E cos d 
where d is obtained by solving eq. (12.52).
A three-phase fault at the infinite bus is simulated by setting E to zero temporarily in eq. (12.68).
For Pg0 = 0.9 pu, the fault is applied at t = 1 s for a period of 0.01 s. The corresponding rotor angle
plot is shown in Figure 12.21 which shows that rotor oscillations damp out, and eventually, settle to
the same initial angle, d0, validating the eigen-predictions.

1.56
0.841s
1.54

1.52 w = 2*pi/0.841
=7.47 rad/s
1.5
d (rad)

1.48

1.46

1.44

1.42

1.4
0 2 4 6 8 10 12 14 16 18 20
Time (s )

Figure 12.21  The Plot of d for Pg0 = 0.9 with Field on Manual Control.

12.4.3.2  Modal Performance of 1.1 Model without AVR

To understand the influence of the reduced order model, the 2.2 model is modified so that it becomes
applicable to 1.1 model. To affect this, set x″d = x′d and x″q = x′q, in the 2.2 model equations. For this model,
the eigenvalues are given in Table 12.5 for different loading conditions without AVR.

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 533 9/12/2017 3:25:01 PM

 534  |  Power System Analysis

Table 12.5  Eigevalues for Different Loading Levels without Static Exciter for 1.1 Model.
Pg0 = 0.1 Pg0 = 0.5 Pg0 = 0.9
−0.8787 ± j 8.2586 −0.4414 ± j 8.2392 −0.3085 ± j 7.4778
DF: 0.1058[0.1105] 0.0535 [0.0587] 0.0412 [0.0467]
−0.5772 −0.2424 −0.0710
−1.3868 −2.5963 −3.0334

From Table 12.5, it is clear that the damping factor (DF) for the swing-mode is smaller for 1.1 model than
that for the 2.2 model (which is given in brackets [ ] using the swing mode in Table 12.4). This is because
of the presence of the additional damper windings, H and K coils in the 2.2 model which are neglected
in the 1.1 model. This observation is verified from a time–domain simulation by plotting the rotor-angle
deviations for a small duration (0.01 s) fault at the infinite bus for Pg0= 0.9 pu (see Figure 12.22).

Without AVR case Pg = 0.9 pu

1.56
1.1 model
2.2 model

1.54

1.52

1.5
Angle (rad)

1.48

1.46

1.44

1.42

1.4
0 1 2 3 4 5 6 7 8 9 10
Time (s)

Figure 12.22  The Plot of d for Pg0 = 0.9 without AVR for 2.2 and 1.1 Model.

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 534 9/12/2017 3:25:05 PM

 Power System Angle Stability  |  535 

12.4.3.3  P
 ower-ramping Exercises to Show the Importance
of Excitation Controller
To illustrate the importance of an automatic voltage regulator (AVR) for the generator, the fol-
lowing case studies are carried out for the system shown in Figure 12.20 with 2.2 model for the
generator.
Initially, the generator is made to deliver a power, Pg0 pu. The mechanical input to the generator
is ramped up and down to different power levels with a ramp rate of Pg0 per second. The mechanical
input torque is varied as per the procedure shown in Figure 12.23, where Tm0 denotes the initial steady
state mechanical torque input.

Tm0 Tm0

+ +

RampLevel*Tm0 Σ RampLevel*Tm0 Σ
Tm Tm
+ −
1 1
s s

0 0
Ramp up Ramp down

Figure 12.23  The Mechanical Input Torque Ramping Procedure.

In each case, the terminal voltage is observed with and without AVR to understand the influence of
an AVR. A single time–constant static exciter, shown in Figure 12.24, is employed for the generator.
The time delay associated with the bus voltage measuring transducer is neglected with KA = 200 and
TA= 0.025 s.

V
Efdmax

−
+ KA
Vref Σ Efd
1 + sTA
+
Efdmin

VS

Figure 12.24  Single Time–constant Static Excitation System.

The following cases are considered:

1. Initial Pg0 = 0.1 pu and the power is ramped up to 0.3 pu.
2. Initial Pg0 = 0.1 pu and the power is ramped up to 0.5 pu.
3. Initial Pg0 = 0.5 pu and the power is ramped down to 0.25 pu.

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 535 9/12/2017 3:25:06 PM

 536  |  Power System Analysis

Power Ramping-up to 0.3 pu:

In this case, when the generator is delivering a power of 0.1 pu under steady state, the mechanical
torque is ramped up to 0.3 pu. without the AVR. To achieve this, the RampLevel in Figure 12.23 is
set to 2. Without the AVR, the generator field voltage is left unaltered at Efd0= 1.023 pu. The various
variables such as Tm, torque output, rotor angle, and generator terminal voltage are plotted as shown
in Figure 12.25.

0.3
Tm (pu)

0.2

0.1

0 5 10 15 20

0.3
Teout (pu)

0.2

0.1
0 5 10 15 20

0.8
d (rad)

0.6

0.4

0.2
0 5 10 15 20
Terminal voltage (pu)

0.98

0.96

0.94
0 5 10 15 20
Time (s )

Figure 12.25  The Mechanical Input Torque Ramping from 0.1 pu to 0.3 pu
without AVR.

From Figure 12.25, it can be seen that the rotor angle increases to 0.8 rad as the terminal voltage of
the generator reduces to 0.94 pu. When this case is repeated with an AVR, the results are obtained as
shown in Figure 12.26. Here, for the controller, the VRef0 is set to 1.005115 pu.

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 536 9/12/2017 3:25:07 PM

 Power System Angle Stability  |  537 

0.3

Tm (pu)
0.2

0.1
0 5 10 15 20

0.3
Teout (pu)

0.2

0.1
0 5 10 15 20

0.6
d (rad)

0.4

0.2
0 5 10 15 20
Terminal voltage (pu)

0.999

0.998
0 5 10 15 20
1.2
Efd (pu)

1.1

1
0 5 10 15 20
Time (s )

Figure 12.26  The Mechanical Input Torque Ramping from 0.1 pu to 0.3 pu
with AVR.

Now, with the AVR, the terminal voltage is just above 0.9992 pu. Under steady-state, it can be seen that Efd
= (VRef0 − 0.99992) × 200 = 1.1830 pu. The rotor-angle becomes 0.65 rad instead 0.8 rad.

Power Ramping-up to 0.5 pu.:

In this case, when the generator is delivering a power of 0.1 pu under steady state, the mechanical
torque is ramped up to 0.5 pu. without the AVR. In this case, RampLevel in Figure 12.23 is set
to 4. Here also, the generator field voltage is left unaltered at Efd0 = 1.023 pu. The bus voltage plot
shown in Figure 12.27 demonstrates that when an effort is made to deliver a power of 0.5 pu, the bus
voltage decreases monotonically, failing to reach a steady-state. Simultaneously, the rotor angle also

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 537 9/12/2017 3:25:08 PM

 538  |  Power System Analysis

increases, eventually leading to loss of synchronism. A huge oscillation in the voltage and developed
torque denotes the loss of synchronism condition. This is mainly due to the fact that as and when the
real power output is raised, the machine fails to deliver adequate reactive power to maintain the nec-
essary bus voltage since the field voltage is held fixed. This also implies that the machine develops
inadequate synchronising torque component.

0.5
0.4
Tm (pu)

0.3
0.2
0.1
0 5 10 15 20

0.4
Teout (pu)

0.2
0
−0.2
−0.4
0 5 10 15 20

10
8
6
d (rad)

4
2

0 5 10 15 20
Terminal voltage (pu)

0.5

0
0 5 10 15 20
Time (s )

Figure 12.27  The Mechanical Input Torque Ramping from 0.1 pu to 0.5 pu
without AVR.

When the case is repeated with an AVR, the machine successfully delivers the increased power with-
out losing synchronism. The associated plots are shown in Figure 12.28.

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 538 9/12/2017 3:25:09 PM

 Power System Angle Stability  |  539 

0.5
0.4
Tm (pu) 0.3
0.2
0.1
0 2 4 6 8 10 12 14 16 18 20

0.5
0.4
Teout (pu)

0.3
0.2
0.1
0 2 4 6 8 10 12 14 16 18 20
1
0.8
d (rad)

0.6
0.4
0.2
0 2 4 6 8 10 12 14 16 18 20
Terminal voltage (pu)

0.998

0.996
0 2 4 6 8 10 12 14 16 18 20

1.4
Efd (pu)

1.2

1
0 2 4 6 8 10 12 14 16 18 20
Time (s )

Figure 12.28  The Mechanical Input Torque Ramping from 0.1 pu to 0.5 pu
with AVR.

Now, with the AVR, the terminal voltage is around 0.9978 pu. Under steady-state, it can be seen
that Efd = (VRef 0 – 0.9978) × 200 = 1.4630 pu. Note that in this case, the bus voltage magnitude is
lower than that existed in the previous case. This is due to the nature of the proportional controller
employed for the AVR. For the same VRef 0, the steady-state error should be more to establish a
higher Efd.

Ramping down of power from 0.5 pu to 0.25 pu:

In this case, when the generator is delivering a power of 0.5 pu under a steady state, the mechanical
torque is ramped down to 0.25 pu, without the AVR. Here, the RampLevel is set to 0.5 and the
generator field voltage is left unaltered at Efd0= 1.4697 pu. Note that this value of Efd is almost equal

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 539 9/12/2017 3:25:10 PM

 540  |  Power System Analysis

to the final value of the Efd that is reached due to AVR action in the previous case, since it also corre-
sponds to the same operating point. This observation illustrates the significance of the initial condition
calculations. The various plots obtained are shown in Figure 12.29. It can be seen from Figure 12.29
that without an AVR, it results in an overvoltage at the generator terminals reaching a magnitude of
more than 1.1 pu.

0.5
Tm (pu)

0.4

0.3

0 5 10 15 20

0.5
0.4
Teout (pu)

0.3
0.2
0.1
0 5 10 15 20

0.8
d (rad)

0.6

0.4
0 5 10 15 20
Terminal voltage (pu)

1.1

1.05

1
0 5 10 15 20
Time (s )

Figure 12.29  Ramping Down of the Mechanical Input Torque from 0.5 pu to 0.25 pu
without AVR.

Now, with the AVR, the terminal voltage is very well regulated around 1.0015 pu (see Figure 12.30).
Here, for the controller, the VRef0 is set to 1.0073485 pu. Under steady-state, it can be seen that
Efd = (1.0073485 − 1.0015) × 200 = 1.1697 pu. Note that in this case, the bus voltage magnitude is
higher than that existed in the previous cases (with AVR).

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 540 9/12/2017 3:25:11 PM

 Power System Angle Stability  |  541 

0.5

Tm (pu)
0.4

0.3

0 5 10 15 20

0.5
Teout (pu)

0.4
0.3
0.2
0.1
0 5 10 15 20

1
d (rad)

0.8

0.6
0 5 10 15 20
Terminal voltage (pu)

1.002
1.001
1
0.999
0 5 10 15 20

1.6
Efd (pu)

1.4

1.2
0 5 10 15 20
Time (s )

Figure 12.30  Ramping Down of the Mechanical Input Torque from 0.5 pu to 0.25 pu
with AVR.

12.4.3.4  Modal Performance with AVR

The dynamic equation for the exciter is given by,
dE fd

dt
=
1 
TA 
( )
− E fd + K A Vref + Vs − V  (12.69)

where, V denotes the magnitude of the generator voltage and Vs represents the input from an auxiliary
controller.
After linearising, the above equation can be expressed in matrix form as

∆E fd (1×1) =  Atf 

(1×1)
∆E fd +  Btf 
(1×1)
∆V(1×1) +  Etf 
(1×1)
(
∆Vref + ∆Vs )(1×1) (12.70)

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 541 9/12/2017 3:25:14 PM

 542  |  Power System Analysis

where

 1 
 Atf 
  (1×1) =  − T 
 A

 KA 
 Btf 
  (1×1) =  − T 
 A

 KA 
 Etf 
  (1×1) =  T 
 A

The terminal voltage magnitude V is expressed in terms of vd and vq as

V = vd2 + vq2

or

V 2 = vd2 + vq2 (12.71)

Linearising the above equation at the operating point, we get

2V0 ∆V = 2vd 0 ∆vd + 2vq0 ∆vq

or

 ∆vd 
∆V = [Vv ]   (12.72)
 ∆vq 
where

v vq 0 
[Vv ] =  Vd 0 V0 

 0

Using eq. (12.60) in eq. (12.72), we get,

∆V = [Vvsi ](1× 2) ∆xs ( 2×1) + [Vvei ](1×6) ∆xe (6×1) (12.73)

where

[Vvsi ] = [Vv ](1×2) [ Avs ](2×2)

[Vvei ] = [Vv ](1×2) [ Ave ](2×6)

Substituting eq. (12.73) in eq. (12.70), we get,

∆E fd (1×1) =  Atf 

(1×1)
∆E fd + [Vvs ](1× 2) ∆xs + [Vve ](1×6) ∆xe +  Etf  (
(1×1)
∆Vref + ∆Vs )(1×1) (12.74)

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 542 9/12/2017 3:25:18 PM

 Power System Angle Stability  |  543 

where

[Vvs ] =  Btf  [Vvsi ]

[Vve ] =  Btf  [Vvei ]
Now from eqs. (12.62) and (12.74), we combine the state equations as

∆xesf (9 ×1) = [ AFD ](9×9) ∆xesf (9×1) + [ EC ](9×1) ( ∆Vref + ∆Vs )(1×1) (12.75)
where

 ∆xe (6×1) 
 
 
∆xesf ( 9 ×1) =  ∆xs ( 2 ×1) 
 
 
 ∆E fd 
 
and


[ AES ](8×8)  FgES 


 (8×1) 
[ AFD ] =  

 [Vve ] [Vvs ]  Atf  
   (1×8)   (1×1) 

 0 
 [ ](8×1) 
[ EC ] =  

  Etf  
   (1×1) 

Since auxiliary controller (say PSS) is not considered in the current case, DVs = 0 and assume DVref = 0.
dy d dy q
Further, stator transients are neglected by setting = 0 and = 0. To obtain the corresponding
dt dt
reduced state-matrix for eigenvalue analysis, we can re-write eq. (12.75) using only [AFD] as

  ∆y     ∆y  
  dq  ( 2 ×1)  [ AFD11 ]( 2× 2) [ AFD12 ](2×7) 
     dq  ( 2 ×1)

  =    (12.76)
   A   
  ∆xrsf  ( 7×1)  [ FD 21 ]( 7× 2) [ AFD 22 ](7×7) (9×9)   ∆xrsf  
 ( 7 ×1) 
  (9×1) ( 9 ×1)

T
where y dq = y d y q  and xrsf = [y F y H yG y K d wy E fd ]T .

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 543 9/12/2017 3:25:21 PM

 544  |  Power System Analysis

Since ∆y dq = 0 , we can write from eq. (12.76) that

 ∆y dq  = − [ AFD11 ]−1
( 2 × 2) [ FD12 ]( 2 × 7)  rsf  ( 7 ×1) (12.77)
A  ∆x 
 

From eq. (12.76), we can also write that

 rsf  ( 7×1) = [ AFD 21 ]( 7× 2)  ∆y dq  ( 2×1) + [ AFD 22 ]( 7× 7)  ∆xrsf  ( 7×1)

 ∆ x      (12.78)

Using eq. (12.77) in eq. (12.78), we have the reduced state-model given by

 rsf  ( 7×1) = [ AFDR ]( 7× 7)  ∆xrsf  ( 7×1) (12.79)

 ∆ x   

where

[ AFDR ] = − [ AFD 21 ][ AFD11 ]−1 [ AFD12 ] + [ AFD 22 ]

The exciter data is given in Table 12.6.

Table 12.6  Single Time–constant Static Exciter Data.

KA = 200 TA = 0.025 s

Efdmax= 6 Efdmin = –6

Using the system data and for different loading conditions, the eigenvalues are determined and are
given in Table 12.7. From Table 12.7, it is clear that as the loading level increases, the damping for
the swing-mode (p45) reduces. When the generated power is raised to Pg0 = 0.9 pu, the swing mode
becomes unstable, denoted by a positive real part of p45.

Table 12.7  Eigenvalues for Different Loading Levels with AVR.

Number Pg0 = 0.1 Pg0 = 0.5 Pg0 = 0.9

p12 −17.1244 ± j 31.0705 −17.1959 ± j 27.0252 −17.3820 ± j 24.718
p3 −38.7851 −38.8324 −38.8635
p45 −0.9482 ± j 8.3599 −0.5044 ± j 8.5986 0.1807 ± j 8.4061
p6 −1.2502 −1.9368 −2.8808
p7 −20.1874 −20.1980 −20.2210

To verify the association of p45 mode with rotor variables, the participation factors are obtained for
Pg0 = 0.9 pu, and are given below. From the listing, it is clear that the states, d and wy have the highest
participation in this mode.

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 544 9/12/2017 3:25:23 PM

 Power System Angle Stability  |  545 

p1 p2 p3 p4 p5 p6 p7     
State
0.5882 0.5882 0.0071 0.0341 0.0341 0.0061 0.0001 –  F
0.1177 0.1177 0.7786 0.0005 0.0005 0.0000 0.0000 –  H
0.0006 0.0006 0.0000 0.0129 0.0129 0.9699 0.0407 –  G
0.0002 0.0002 0.0000 0.0014 0.0014 0.0429 0.9599 –  K
0.0105 0.0105 0.0000 0.4994 0.4994 0.0092 0.0003 –  delta
0.0105 0.0105 0.0000 0.4994 0.4994 0.0092 0.0003 –  omega_y
0.5263 0.5263 0.2285 0.0078 0.0078 0.0005 0.0001 –  Efd

For Pg0 = 0.9 pu, a three-phase fault is applied at t = 1 s for a period of 0.01 s at the infinite bus. The
corresponding rotor angle plot is shown in Figure 12.31. It is clear from Figure 12.31 that rotor angle
amplitude increases from cycle to cycle, exhibiting oscillatory instability. This is what the eigenvalue
pertaining to the swing-mode depicts with a positive real part, which denotes a negative damping
effect of the exciter. This case should be compared with a rotor-angle plot shown in Figure 12.21,
where, for the same power level, the system is stable without the exciter. From these, the following
observations can be made:

2.5 Pg0 = 0.9 pu

1.5
Delta (rad)

0.5

0
0 2 4 6 8 10 12 14 16 18 20
Time (s)

Figure 12.31  The Plot of d for Pg0 = 0.9 pu with AVR.

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 545 9/12/2017 3:25:24 PM

 546  |  Power System Analysis

1. From Table 12.4, we can note the swing-mode frequency for Pg0= 0.9 pu, as 7.513 rad/s (imagi-
nary part of p23). However, with the exciter, the swing-mode frequency increases to 8.4061 rad/s.
This clearly denotes an enhancement of the synchronising torque with the exciter (see eq. (12.15)).
2. With a fast-acting and high-gain static exciter, a system may offer poor damping for the swing-
mode when the loading level is higher and the connecting line has a large reactance. Such a sit-
uation may arise during operation due to changes in network conditions. A system operator can
detect such a condition by observing growing oscillations in line flows as rotor-angle deviations
directly influence the power developed by a generator.
3. These conditions can be handled either by reducing the loading level on the generator or by
appropriately tuning a power system stabiliser (PSS) on a generator. It should be noted that all
generators in a power system will have a PSS block as a part of its excitation controller unit.
However, to improve the system damping performance, only a few PSS units on specific genera-
tors are enabled in a coordinated manner.

12.4.4  Modal Performance with AVR and PSS

In order to establish a positive damping with the exciter, a slip-signal input power system stabiliser
is designed in this section. To design a power system stabiliser, we first obtain the angle of a system
∆Teg ( s)
function referred to as GEP ( s) = as stated earlier. This is derived in the following lines.
∆Vs ( s)

From eq. (12.54), we have

∆Teg (1×1) = [Cet ](1×9) ∆xesf ( 9 ×1) (12.80)

where

[Cet ] = [[Ce ] 0 0 0 ]

Further, the above equation can be rewritten as

∆Teg (1×1) = [Cet11 ](1× 2) ∆y dq ( 2×1) + [Cet12 ](1× 7) ∆xrsf ( 7 ×1) (12.81)

Now, since stator transients are neglected, using eq. (12.77), we have

∆Teg (1×1) = [Cetr ](1× 7) ∆xrsf ( 7 ×1) (12.82)

where
−1
[Cetr ] = −[Cet11 ][ AFD11 ] ( 2 × 2) [ AFD12 ](2×7) + [Cet12 ]
As stated earlier, while obtaining the GEP(s), it is required to set the Dd = 0 and Dwy = 0. To effect this
change, we modify the matrices as follows:
1. In the [Cetr] matrix, the fifth and sixth elements are removed so that the resulting [Cetr] is of size
(1 × 5).

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 546 9/12/2017 3:25:27 PM

 Power System Angle Stability  |  547 

2. In the [AFDR](7 × 7) matrix, the fifth and sixth rows and columns are removed so that it results in
a matrix [AFDRg](5 × 5).
3. In the [EC](9 × 1) matrix, the first, second, seventh, and eighth elements are removed so that it
results in a matrix [ECg](5 × 1).

Thus, we have,

∆Teg (1×1) = [Cetr ](1×5) ∆xrf ( 5×1) (12.83)

 ∆xrf 

      (
 (5×1) =  AFDRg  (5×5)  ∆xrf  (5×1) +  ECg  (5×1) ∆Vref + ∆Vs )(1×1) (12.84)
Considering only DVs in eq. (12.84) and rearranging the terms in s domain, we get

−1
∆xrf ( s) =  sI − AFDRg   ECg  ∆Vs ( s) (12.85)
 

Using eq. (12.85) in eq. (12.83), we have,

−1
∆Teg ( s) = [Cetr ]  sI − AFDRg   ECg  ∆Vs ( s)
 

∆Teg ( s) −1
GEP ( s) = = [Cetr ]  sI − AFDRg   ECg 
 
∆Vs ( s)

The frequency response is obtained by letting s = jw and spanning w in the desired range, i.e.,

∆Teg ( jw ) −1
GEP ( jw ) = = [Cetr ]  jwI − AFDRg   ECg  (12.86)
 
∆Vs ( jw )

The angle of the uncompensated GEP is shown in Figure 12.32.

Compensator Design:
A lead circuit is designed with a center frequency, fm = 3 Hz and phase angle lead, fm = 10°. The cor-
responding parameters of the lead circuit are T1 = 0.0632 s and T2 = 0.0445 s. Thus, the overall PSS
transfer function is given by

sTw (1 + sT1 )
PSS ( s) = K pss (12.87)
(1 + sTw ) (1 + sT2 )

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 547 9/12/2017 3:25:30 PM

 548  |  Power System Analysis

80

60

PSS
40

20
Angle in deg

0
Compensated GEP

−20

−40
Uncompensated GEP

−60

−80
0 0.5 1 1.5 2 2.5 3 3.5 4
Frequency (Hz)

Figure 12.32  The Plot of Angles of GEP for Pg0 = 0.9 pu with AVR.

For the washout circuit, Tw is set as 10 s. With the above PSS(s), the compensated GEP is also shown
in Figure 12.32.

12.4.4.1  Eigenvalue Analysis with PSS

To construct a state matrix of the power system with PSS, a state-model of the PSS should be
V ( s)
properly interfaced to the system equations. To achieve this, a state-model of the PSS ( s) = s
u( s )
is first obtained in the following lines:
Consider the washout circuit,

Yw ( s)  sTw 
= + 1 − 1 (12.88)
U ( s)  (1 + sTw ) 

 1 
= 1 − (12.89)
 (1 + sTw ) 

Writing the state–space equation for the above function, we get,

1 1
xw = − xw + u (12.90)
Tw Tw

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 548 9/12/2017 3:25:32 PM

 Power System Angle Stability  |  549 

and

yw = u − xw (12.91)

Now consider the compensator,

Vs′( s)  1 + sT1  (12.92)

=
Yw ( s)  1 + sT2 

 1  
  + s 
T T 
= 1 1 + 1 − 1 (12.93)
T2   1  
  T + s 
 2  

  1 1 
−
T   T T2  
= 1 1 + 1 (12.94)
T2   1  
  T + s 
 2 

  T2  
−1
T1   T1  
= 1+ (12.95)
T2  (1 + sT2 ) 
 
 

  T1  
 T 1 −  T  
2  (12.96)
= 1 +
 T2 (1 + sT2 ) 
 
 
The corresponding state–space equation is given by

1 K
xc = − xc + c1 yw (12.97)
T2 T2
and

Vs′ = xc + K c 2 yw (12.98)

where

T  T1
K c1 = 1 −  1  and Kc2 =
 T2  T2

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 549 9/12/2017 3:25:35 PM

 550  |  Power System Analysis

Using eq. (12.91) in eqs. (12.97) and (12.98) we get

K c1 1 K
xc = − xw − xc + c1 u (12.99)
T2 T2 T2

Further, we have Vs = KpssVÄs, therefore, we obtain

Vs = K pss xc − K pss K c 2 xw + K pss K c 2 u (12.100)

Writing the PSS equations in matrix form in terms of deviation, we get

∆ x pss( 2×1) = [ Apss ]( 2× 2) ∆x pss ( 2×1) + [ B pss ]( 2×1) ∆u(1×1) (12.101)

where

x pss = [ xw xc ]T

and

 1   1 
− T 0  T 
 w   w 
[ Apss ] =    B pss  =  
 
   
 − K c1 1
−   K c1 
 T2 T2   T2 

and

∆Vs(1×1) = [C pss ](1× 2) ∆x pss ( 2×1) + [ D pss ](1×1) ∆u(1×1)(12.102)

[C pss ] =  − K pss K c 2 K pss  [ D pss ] =  K pss K c 2 

Rewriting eq. (12.79) with the auxiliary input DVs, we get

  (
 rsf  ( 7×1) = [ AFDR ]( 7× 7)  ∆xrsf  ( 7×1) + [ EC1 ]( 7×1) ∆Vref + ∆Vs
 ∆ x  )(1×1(12.103)
)

where [EC1] = [EC] with zeros corresponding to yd and yq states removed.

Further, we can note that for a slip-input PSS, Du = Dwy and hence we have,

∆u = [ I 6 ]∆xrsf (12.104)

where [I6] = [0 0 0 0 0 1 0].

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 550 9/12/2017 3:25:38 PM

 Power System Angle Stability  |  551 

Using eq. (12.104) in eqs (12.101) and (12.102) we get,

∆ x pss( 2×1) = [ Apss ]( 2× 2) ∆x pss ( 2×1) + [ B pss ]( 2×1) [ I 6 ](1× 7) ∆xrsf (12.105)

∆Vs(1×1) = [C pss ](1× 2) ∆x pss ( 2×1) + [ D pss ](1×1) [ I 6 ](1× 7) ∆xrsf (12.106)

Setting DVref = 0 and using eq. (12.106) in eq. (12.103) and writing the complete state-matrix with
PSS, we have

 (9×1) = [ AFP ](9×9)  ∆xrsfp  (9×1) (12.107)

 ∆xrsfp   

where

[ A ] + [ E ][ D pss ][ I 6 ](7×7)
 FDR ( 7× 7)  C1
[ EC1 ][C pss ]


 ( 7 × 2) 
[ AFP ] =  

 [ B pss ][ I 6 ] [ Apss ] 
   ( 2 × 7) 

T
with xrsfp =  xrsf x pss 

For the designed lead circuit and with Kpss = 5.0, the eigenvalues obtained are as shown in Table 12.8.

Table 12.8  Eigenvalues with PSS for Pg0 = 0.9 pu.

Number Pg0 = 0.9 Damping factor

p12 −15.9177 ± j 24.1465 0.5504
p3 −38.8881 1.0000
p45 −0.7954 ± j 8.5756 0.0924
p6 −23.4315 1.0000
p7 −20.2216 1.0000
p8 −2.8641 1.0000
p9 −0.1003 1.0000

From the tabulated eigenvalues, it can be seen that the real parts of all eigenvalues are negative. This
indicates that with a properly tuned PSS, the system becomes small-signal stable even when it is
delivering Pg0 = 0.9 pu with a fast-acting and high-gain static exciter. To validate this, a three-phase
fault is applied at t = 1 s for a period of 0.01 s at the infinite bus. The corresponding rotor angle plot is
shown in Figure 12.33. It can be concluded from Figure 12.33 that the system is small-signal stable.

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 551 9/12/2017 3:25:40 PM

 552  |  Power System Analysis

1.52
Pg0 = 0.9 pu

1.5

1.48

1.46
Delta (rad)

1.44

1.42

1.4
0 2 4 6 8 10 12 14 16 18 20
Time (s)

Figure 12.33  The Plot of d for Pg0 = 0.9 pu with PSS.

For this case, other variables such as slip-signal, output of the PSS, magnitude of the generator termi-
nal voltage are also plotted in Figure 12.34. From the figure it can be seen that the slip reaches zero
under steady-state due to the presence of the infinite bus. A limit of +/− 0.1 is used on the Vs-signal.

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 552 9/12/2017 3:25:41 PM

 Power System Angle Stability  |  553 

1.5
d (rad)

1.45

1.4
0 1 2 3 4 5 6 7 8 9 10

x 10−3

1
Slip (pu)

−1

0 1 2 3 4 5 6 7 8 9 10

x 10−3
10
5
Vs (pu)

0
−5

0 1 2 3 4 5 6 7 8 9 10
Terminal voltage (pu)

0.95

0.9

0.85
0 1 2 3 4 5 6 7 8 9 10
Time (s )

Figure 12.34  The Plot of d, Slip, Vs and Terminal Voltage for Pg0 = 0.9 pu with PSS.

References
[1]  K. R. Padiyar, Power System Dynamics - Stability and Control, BS Publications, Hyderabad,
India, 2002.
[2]  Yadi Anantholla, ‘A Multi-machine Small-signal Stability Programme,’ M. Tech., Thesis submit-
ted to NITK, India, 2008.
[3]  K. Ogata, ‘Modern Control Engineering,’ Prentice-Hall, N.J, July 2001. EE Dept, NITK, Surathkal,
India, 2008.

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 553 9/12/2017 3:25:41 PM

 554 | Power System Analysis

Review Questions
1. What is the significance of small-signal stability analysis? How does it differ from transient sta-
bility analysis?
2. Briefly discuss the tools and techniques used to perform small- and large-signal stability analysis.
3. Explain, how the inferences made in the rotor-angle stability analysis are used to improve the
dynamic performance of power systems.
4. For an SMIB system with classical model for the generator, derive an expression for swing mode
frequency, if mechanical damping D is considered.
5. For the system shown in Figure 12.35, the classical model has been used for the generator with
xd′ = 0.33 pu., H = 5 MJ/MVA, and f o = 50 Hz. Neglect mechanical damping.
V = 1.0∠0°

j0.11

S = (0.6667 + j0.0677) p.u.

Figure 12.35 SMIB System for Torque Perturbation Analysis.

Employing a linearized model, obtain an expression for d (t ) and Sm (t ) for a step increase in the
mechanical input given by ∆Tm = 0.1 pu. Assume that [ ∆d o , ∆Smo ] = [0, 0]T .
T

6. Briefly state the importance of damping and synchronizing torque components with regard to a
swing mode. How do you identify their inadequacies in a power system?
7. What are the influences of a fast-acting high-gain exciter on the power system stability?
8. What are the functions of a power system stabilizer? Is it designed to augment transient stability?
9. What is GEP(s)? How do you design a PSS using this transfer function?
10. Design a lead circuit compensator with a centre frequency of 3.5 Hz and a lead angle of 30°. This
compensator is used in a PSS along with a wash-out circuit of time constant 2 s. Determine the
overall phase angle of the PSS at a swing-mode frequency of 1.5 Hz.
11. Assume that a generator is connected to a bus whose voltage magnitude is held constant at V and
is taken as the reference. The quiescent power angle is d o and xq = xd . Without changing the tur-
bine power, if the field voltage is given a small perturbation, prove that the reactive power received
at the bus under steady-state changes by the amount
V ∆E fd 0
∆Q =
xd cosd o
12. Name some other devices which can be used to achieve similar functions as PSS.
13. For the system shown in Figure 12.36, classical model has been used for the generator with
xd′ = 0.33 pu, H = 5 MJ/MVA and f o = 50 Hz. Given that Seg = 0.6667 + j 0.0677 pu, x L = 0.11
pu and E = 1.0 ∠0 pu. Assume that the mechanical input is constant and mechanical damping
is negligible. Employing a linearized model of the power system, determine the stability of the
operating point when the line reactance x L is varied as per the following strategy:
∆x L = −0.2 ∆Sm

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 554 9/12/2017 3:25:49 PM

 Power System Angle Stability | 555

where Sm denotes the slip of the machine. Verify the analysis by a SIMULINK-based time-
domain simulation.
Vt ∠q E = E∠0
jxL

Generator Seg

Figure 12.36 SMIB System for Line Reactance Modulation.

14. For a lossless system shown in Figure 12.37, the generator is modelled with 1.0 model with AVR.
Some of the model equations are as follows:

vd = −y q
vq = y d ,

where

y q = xq iq

y d = xd′ id + Eq′
Eb cos d − Eq′
id =
( xd′ + x L )
Eb sin d
iq =
( xq + x L )

Vt ∠q = 1.0∠q Eb = 1∠0
jxL

Generator Pg

Figure 12.37 SMIB System to Show AVR Effects with 1.0 Model.

The dynamic equation for a single-time constant exciter is given as:

dE fd
dt
1 
TA 
(
− E fd + K A VRef 0 − Vt 
= )
where the terminal voltage magnitude Vt is obtained by using vd and vq as

Vt = vq2 + vd2 .

Linearize the system of equations to determine the operating point stability noting the state vector
as x = [d , Sm , y F , E fd ]T . Perform the analysis when

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 555 9/12/2017 3:25:54 PM

 556 | Power System Analysis

(a) Pg0 = 0.5 pu

(b) Pg0 = 0.9 pu
Design a slip-signal PSS to stabilise the operating point at Pg0 = 0.9 pu. Given that xd = 1.79, xq =
1.71, xd′ = 0.169, Tdo
′
= 4.3 s, Ra = 0, H = 2.86 MJ/MVA, D = 0, x L = 0.7, and w B = 2p 60 rad/s.
For the PSS, take the centre frequency = 3 Hz, phase-lead angle =10°, wash-out circuit time con-
stant = 10 s, and gain of the PSS = 5.

M12_POWER-SYSTEM-AN00_SE_XXXX_CH12.indd 556 9/12/2017 3:25:57 PM

 Modal Analysis of Power
Systems with Interconnected
Generators 13
CHAPTER OUTLINE

13.1 Importance of Modal Analysis 13.5 Participation Matrix

of Power Systems 13.6 Modal Analysis of a Two-machine
13.2 Power System Oscillations Power System
13.3 Spring-Mass System Example 13.7 Modal Analysis of Four-machine
13.4 Linearisation of Power System Modified Power System
Modelling Equations

In this chapter, small-signal stability analysis of interconnected generators in a power system is

presented. To understand the performance of such a complex non-linear power system, a multi-
mass spring system is analysed for its different natural modes by performing eigenvalue analysis and
is further validated by a time–domain simulation.
For the power system, a detailed procedure is outlined to linearise the performance equations of
different components such as generator, network, and loads. These models are interfaced to obtain
the final state model of the system around an operating point. Two-machine and four-machine power
systems are used to illustrate the small-signal analysis.
For the four-machine power system, different operating scenarios are discussed to illustrate the
effect of loading level, presence of exciters, and influence of a power system stabiliser.

13.1 IMPORTANCE OF MODAL ANALYSIS OF POWER SYSTEMS

Small-signal rotor-angle stability analysis mainly deals with the study of electromechanical oscilla-
tions-related performance of a power system about an operating point. Thus, this study is concerned
with the ability of the power system to maintain synchronism against small perturbations. Here, the
analysis assumes that a disturbance is sufficiently small enough to permit linearisation of system
equations around an operating point. This enables the linear system theory to be applied for system
analysis even though the system is inherently non-linear.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 557 9/12/2017 4:04:58 PM

 558  |  Power System Analysis

If a power system, at a given operating condition, is large-disturbance unstable, such a system can
still be operated, although it is insecure. However, if the system is small-signal unstable at a given
operating condition, it cannot be operated. Therefore, small-signal stability is a fundamental require-
ment for the satisfactory operation of power systems. Such a study mainly involves the verification of
sufficiency of damping of all modes associated with a system so that power transfer is not constrained.
It is known that when a dynamic system such as a power system is perturbed from its steady state
condition, the system variables trace out a flow, which is referred to as trajectories. These trajectories
may exhibit oscillatory or monotonic behaviour. For the system to be stable, these trajectories must
remain bounded and converge to an acceptable operating point.

13.2  POWER SYSTEM OSCILLATIONS

A study of power system oscillations is of interest in a system where more than one generator works
in parallel to deliver a common load. In small systems, there may be only a few generators; in large
systems, there may be thousands of generators working in parallel. In such a situation, synchronous
machines produce torques that depend on the relative angular displacement of their rotors. These
torques act to keep the generators in synchronism and are referred to as synchronising torques. Thus,
if the angular displacement between generators increases, an electrical torque (i.e., a synchronising
torque) is produced that tries to reduce the angular displacement. It acts as though the generators were
connected by springs, just as in a spring-mass system where a restraining force due to the spring acts
against the movement of masses, resulting in oscillations. Similarly, the moment of inertia of rotors
and the synchronising torques cause the angular displacement of the generators to oscillate following
the occurrence of a disturbance when it is operating under steady state. Under these conditions, the
generators behave as rigid bodies and oscillate with respect to one another using the electrical trans-
mission path between them to exchange energy. If the system is small-signal unstable at the operating
point due to inadequate synchronising torque, the magnitude of the relative angular displacement can
grow monotonically over the span of many seconds and, can eventually result in outages of major
portions of the power system due to loss-of-synchronism.
Further, a power system is continuously subjected to random disturbances in the form of load
or generation changes, or changes in controller settings. Hence, a power system never settles to a
specified steady-state condition at any given point of time. Therefore, as and when an operating point
is reached, the system should possess adequate damping for all system modes (in addition to good-
enough synchronising torque component for all electromechanical modes). If an operating point offers
inadequate damping for a mode, it takes longer time to settle. Since the mechanical damping is quite
small, the damping of electrical origin should be positive for a given mode. If it is negative, the ampli-
tude of oscillations grow from cycle to cycle, exhibiting oscillatory instability, thus constraining the
power transfer. This is critical for system stability and therefore, to system security and reliability.
In a well-designed and operated system, the oscillations of the rotor angle displacement decay and set-
tle to a value that will not constraint power flow through the transmission network. Such a system is said to
be small-signal stable. In the following circumstances, the system may be small-signal unstable [1, 2, 3]:
1. Use of fast-acting and high-gain exciters.
2. Heavy power transfer over long transmission lines from remote generating plants.
3. Power transfer over weak ties between systems which may result due to line outages.
4. Inadequate tuning of controllers of equipment such as generator excitation systems, high-voltage,
direct current (HVDC) converters, static var compensators, etc.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 558 9/12/2017 4:04:58 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  559 

5. Adverse interaction of electrical and mechanical systems, causing instabilities of torsional

mode oscillations.
In an over-stressed system, a relatively low inherent damping and a small magnitude of synchronising
torque coefficient may constrain the system operation by limiting power transfer. Further, in such
cases, predicting oscillation boundaries and therefore, to manage them, becomes increasingly difficult.
While analysing the small-signal stability of an operating point, x0, the following points must be noted:
1. The state of the system after the occurrence of a disturbance should be x0.
2. If a perturbation occurs while operating at x0, the resulting trajectories should eventually settle
at the same operating point.

13.2.1 Classification of Power System Oscillation

Power system oscillations are mainly concerned with small excursions of the system conditions about
a steady state operating point, following a small disturbance. For convenience of analysis, the oscilla-
tions associated with a power system are classified as follows [1, 2].
1. Swing-mode oscillations
2. Control mode oscillations
3. Torsional mode oscillations

13.2.1.1  Swing-mode Oscillations

This mode is also referred to as electromechanical oscillations. For an ng generator system, there are
(ng – 1) swing modes associated with the generator rotors. A swing-mode oscillation is characterised
by a high association of the generator rotor in that mode, where generator(s) in two coherent groups
swinging against each other with an approximate phase difference of 180° among the groups. It is
shown later that in the eigenvalue analysis, a high association is denoted by participation factors and
formation of the coherent groups is identified by right eigenvectors associated with rotor slip. In addi-
tion, there is a rigid body mode or zero mode in which all generator rotor take part as a single rigid
rotor. This mode is generally associated with the movement of the centre of inertia which corresponds
to the dynamics of the average frequency. Not all generators are involved in all modes. Typically, each
mode is associated with a group of generators swinging against another group. The location of gener-
ators in the system determines the type of swing mode.
Swing-mode oscillations can be further grouped into four broad categories:
1. Local machine-system oscillations: These oscillations generally involve one or more synchronous
machines at a power station swinging together against a comparatively large power system or load
centre at a frequency in the range of 0.7–2 Hz. These oscillations become particularly troublesome
when the plant is at high load with a high reactance transmission system. The term ‘local’ is used
because the oscillations are localised at one station or a small part of the power system.
2. Interunit (Intra-plant) mode oscillations: These oscillations typically involve two or more syn-
chronous machines at a power plant swing against each other, usually at a frequency between
1.5–3 Hz.
3. Local mode oscillations: These oscillations generally involve nearby power plants in which
coherent groups of machines within an area swing against each other. The frequency of oscilla-
tions is in the range of 0.8–1.8 Hz.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 559 9/12/2017 4:04:58 PM

 560  |  Power System Analysis

4. Inter-area mode oscillations: These oscillations usually involve combination of many synchro-
nous machines on one part of a power system swinging against machines on another part of
the system. Inter-area oscillations are normally of a much lower frequency than local machine-­
system oscillations in the range of 0.1–0.5 Hz. These modes normally have widespread effects
and are difficult to control.

13.2.1.2 Control Mode Oscillations

Control modes are associated with generating units and other controls. Poorly tuned exciters, speed gov-
ernors, HVDC converters, and static var compensators are the usual causes of instability of these modes.

13.2.1.3 Torsional Mode Oscillations

These oscillations involve relative angular motion between the rotating elements (synchronous
machine, turbine, and exciter) of a unit, with frequencies ranging from 4 Hz and above. This mechani-
cal system has very little inherent natural damping. The source of torque for inducing torsional oscilla-
tions comes from a modulation of excitation system output and dynamic mechanisms associated with
DC lines, static converters, series-capacitor-compensated lines, and other devices.
A wide bandwidth excitation system may have the capability to provide sufficient negative damp-
ing at any of these torsional natural frequencies to destabilise one or more of these torsional modes,
particularly with the application of a power system stabiliser.
Of these oscillations, local machine-system mode, local mode, intra-plant mode, control mode, and
torsional mode are generally categorised as local problems as it involves a small part of the system.
Further, inter-area mode oscillations are categorised as global small-signal stability problems and are
caused by interactions among large groups of generators and have widespread effects.

13.2.2 Analysis of Small-signal Stability

In addition to eigenvalue analysis, other methods discussed in the literature are synchronising and damp-
ing torque analysis [1, 4], frequency response- and residue-based analysis [5, 6], time-domain solu-
tion-based analysis [2, 1, 7, 8], etc. Of all these methods, the eigenvalue or modal analysis is widely used
for analysing the small-signal stability of power system [6, 9, 10] as it has the following advantages:
1. With eigenvalue techniques, oscillations can be characterised easily, quickly, and accurately.
Different modes, which are mixed with each other in curves of time–domain simulation, are
identified separately. Root loci plotted with variations in system parameters or operating condi-
tions provide valuable insight into the dynamic characteristics of the system. Using eigenvectors,
coherent groups of generators, which participate in a given swing mode, can be identified. In
addition, linear models can be used to design controllers that damp oscillations. Further, informa-
tion regarding the most effective site of controller, tuning of the existing one, and installation of a
new controller can be decided.
2. From the eigenvalue-based analysis, time responses to any chosen small disturbance can be gen-
erated for comparison with field test results. In addition, frequency response characteristics of
the model can be generated. This is useful for comparison with system models developed from
frequency response measurements.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 560 9/12/2017 4:04:58 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  561 

3. Eigenvalue or modal analysis describes the small-signal behaviour of the system about an oper-
ating point, and does not take into account the non-linear behaviour of components such as con-
troller’s limits at large system perturbations. Further, design and analysis carried out using various
indices such as participation factors, residues, etc., may lead to many alternate options. These
options need to be verified for their effectiveness using system responses for small/large distur-
bances. In such cases, time–domain simulations are very essential. In this context, time–domain
simulation and modal analysis in the frequency domain should be used in a complement manner
in analysing small-signal stability of power systems [2, 11].
The complete eigenvalue analysis of a dynamic system is demonstrated below through a linear
spring-mass system.

13.3  SPRING-MASS SYSTEM EXAMPLE

Let us consider a multi-mass, multi-spring system as shown in Figure 13.1. There are two small
masses connected by a very stiff spring which, in turn, are connected to a relatively larger mass
through a much less stiff spring.
x
1

x
2

F1
x
3
k k
M 12 23
1

M M F3
2 3

Damping B1

Figure 13.1  Multi-mass Multi-spring System.

An equivalent circuit for the system is shown in Figure 13.2.

x k x k x
1 12 2 23 3

F1 F3
M M M
1 2 3
B1

Figure 13.2  Equivalent Circuit for Multi-mass Multi-spring System.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 561 9/12/2017 4:05:03 PM

 562  |  Power System Analysis

Choosing the state variable vector as x = [ x1, x2, x3, v1, v2, v3 ]T, we can write the dynamic equation
in state-space as

dx1
= v1 (13.1)
dt

dx2
= v2 (13.2)
dt

dx3
= v3 (13.3)
dt

dv1 k Bv F
= − 12 ( x1 − x2 ) − 1 1 + 1 (13.4)
dt M1 M1 M1

dv2 k12 k k  k
= x1 −  12 + 23  x2 + 23 x3 (13.5)
dt M2  M2 M2  M2

dv3 k23 F

dt
=
M3
( x2 − x3 ) + M3 (13.6)
3

Writing the above equations in matrix form, we get,

0 0 0 1 0 0
0 0 
0 0 0 1 0
 x1     x1  0 
 x  0 0 0 0 0 1    
 2   −kk12   x2  0 
k12 B1
 x3   0 − 0 0   x3   F1  (13.7)
  =  M1 M1 M1  + 
v1   k  v1   M1 
k k  k23  
v2   12 −  12 + 23  0

0 0  v2  0 
   M2  M2 M2  M2  
v3    v3   F3 
 k23 k23   
0 − 0 0 0  M3 
 M3 M3 

Choose the parameters as follows:

k12 = 2 N/m; k23 = 20 N / m; B1 = 0 N / m / s

M1 = 10 kg; M 2 = 1 kg; M 3 = 1 kg

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 562 9/12/2017 4:05:07 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  563 

We can observe that k23 >> k12. This means that masses M2 and M3 are more rigidly coupled than
the group M2 and M3 with M1. Also observe that M1 >> M2 and M3. From the knowledge of a simple
spring-mass system, it can be predicted that low frequency oscillation is mainly due to mass M1 and
high frequency oscillation is predominantly associated with M2 and M3. Also, it is assumed that the
external forces, F1 and F3 are 0.1 s duration pulses. These forces are applied at t = 1 s individually
with an amplitude equal to 5 N. Note that no damping is considered. Substituting the above said
parameters we get the state-feedback matrix A as

0 0 0 1 0 0
0 0 0 0 1 0
 
0 0 0 0 0 1  (13.8)
A=  
 −0.2 0.2 0 0 0 0
2 −22 20 0 0 0
 
0 20 −20 0 0 0 

The eigenvalues of the above matrix are

li with i =1,2,3, 4,5,6 = ± j 6.4055, ± j1.0816, 0, 0

Note:
1. Two zero eigenvalues represent the non-uniqueness of state variables. One zero eigenvalue is
due to the displacement variable x, and the other is due to the velocity variable v. A zero eigen-
value implies that if states x1, x2, and x3 are changed by a given amount, it still does not alter
the relative displacement between the masses. Similar inferences can be made about the states
v1, v2, and v3.
2. The eigenvalues for the above matrix A are determined using the MATLAB command [U, D] =
eig(A).
where U is the matrix of right eigenvectors and D is the diagonal matrix having the eigenvalues as the
diagonal elements.
The matrix of right eigenvectors is given by,
l1 l2 l3 l4 l5 l6
0.0005∠0° 0.0005∠0° 0.0950 ∠0° 0.0950 ∠0° 0.5774 ∠180° 0.5774 ∠180°   x1 
0.1118∠180° 0.1118∠ − 180° 0.4608∠ − 180° 0.4608∠180° 0.5774 ∠180° 0.5774 ∠180°   x2 
  
0.1063∠0° 0.1063∠0° 0.4894 ∠ − 180° 0.4894 ∠180° 0.5774 ∠180° 0.5774 ∠180°  x3 
U=  
0.0035∠90° 0.0035∠ − 90° 0.1028∠90° 0.1028∠ − 90° 0 0   v1 
0.7159∠ − 90° 0.7159∠90° 0.4984 ∠ − 90° 0.4984 ∠90° 0 0   v2 
  
0.6809∠90° 0.6809∠ − 90° 0.5293∠ − 90° 0.5293∠90° 0 0   v3 

Note: From the right eigenvectors listing, it is not easy to associate any zero eigenvalue with any
states, say x-vector or v-vector. Since there are two zero eigenvalues in a repeated fashion, the time–
domain response will have one term te0t-response, i.e., a ramp, and a e0t-response, i.e., a constant
DC-offset. However, from the right-eigenvector matrix, we can see that both the zero-eigenvalues, i.e.,
l5 and l6 show the association with x-states. This implies that a ramp response is dominant in x-states.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 563 9/12/2017 4:05:09 PM

 564  |  Power System Analysis

The association of one zero-eigenvalue representing a constant DC-offset (in no damping case) with
v-states will be clear when redundancy with x-state is removed. This is shown in the following section.

13.3.1 Removal of Redundancy of a State

To eliminate one-zero eigenvalue due to the variable x, the state variables are redefined as
xn = [ p, q, v1, v2, v3]T, where,

p = ( x2 − x1 ) and q = ( x3 − x1 )

Using this new state vector, the state-space equations from (13.1) to (13.6) are rewritten without
the forcing functions as follows:

dp
= v2 − v1 (13.9)
dt

dq
= v3 − v1 (13.10)
dt

dv1 k12
= p (13.11)
dt M1

dv2 k k
= − 12 p + 23 ( q − p ) (13.12)
dt M2 M2

dv3 k
= − 23 ( q − p ) (13.13)
dt M3

Writing the above equation in matrix form, we have,

0 0 −1 1 0 
0 0 −1 0 1 
 p    p 
 q   k12 0 0 0 0
 q 
   M1  
v1  =   v1  = Am xn
   k12 k23  k23   
v2   − M − M  M2
0 0 0 v2 
v3   2 2  v3 
k k23 
 23 − 0 0 0
 M 3 M3 

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 564 9/12/2017 4:05:11 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  565 

Using the numerical values of the parameters, we obtain,

0 0 −1 1 0
0 0 −1 0 1
 
Am =  0 .2 0 0 0 0
 
 −22 20 0 0 0
20 −20 0 0 0 

The eigenvalues of Am are

l1 = + j 6.4055, l2 = − j 6.4055,

l3 = + j1.0816, l4 = − j1.0816,

l5 = 0
The matrix of right eigenvectors is given by,

l1 l2 l3 l4 l5

0.1123∠ − 90 0.1123∠90 0.5096∠0 0.5096∠0 0 p

 
0.1057∠90 0.1057∠ − 90 
0.5358∠0 
0.5358∠0 
0 q
 
U = 0.0035∠ − 180 0.0035∠180 
0.0942∠ − 90 
0.0942∠90 0.5774 ∠0  v1
  
0.7160 ∠0 0.7160 ∠0 0.4569∠90 0.4569∠ − 90 0.5774 ∠0  v2
  
0.6809∠ − 180 0.6809∠180 0.4853∠90 0.4853∠ − 90 0.5774 ∠0  v3

Further, as shown in the later part of the section, where a decaying DC component denoted by a pure
negative real-eigenvalue with v-states will be clear when damping is made non-zero.
In order to understand the time–domain performance of the spring-mass system, the following
case studies are carried out. In the first three cases, an external force in the form of a short duration
pulse and in the remaining two cases, an external force in the form of a step signal is applied at t = 1s.
The state-space equations are numerically solved by using RK-4th order method with a step size of
0.001 s in SIMULINK platform. The initial value of state vector, xn0 = [0]T.

13.3.2 Case-1: With External F1 Without Any Damping

In this case, an external force F1 alone is applied at 1 s for a duration of 0.1 s. The time–domain plots
of all states are shown in Figure 13.3.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 565 9/12/2017 4:05:16 PM

 566  |  Power System Analysis

Response due to a short pulse at F1 (Case-1)

4

x1 2

0
0 5 10 15 20 25 30 35 40 45 50
4

2
x2

0
0 5 10 15 20 25 30 35 40 45 50
4

2
x3

0
0 5 10 15 20 25 30 35 40 45 50
0.05
0.04
V1

0.03 5.82 s
5 10 15 20 25 30 35 40 45
0.1
0.05
V2

0
0 5 10 15 20 25 30 35 40 45 50
0.1
0.05
V3

0
0 5 10 15 20 25 30 35 40 45 50
Time (s )

Figure 13.3  Response to a Short Duration F1 without Any Damping.

From Figure 13.3, the following observations can be made:

1. The states x1, x2, and x3 show a dominant ramping response, whereas the states v1, v2, and v3,
demonstrate a prominent DC-offset. These responses are generally referred to as common-mode
or zero-mode behaviour.
2. The states v1, v2, and v3 also show a sinusoidal oscillation of frequency close to 1.0816 rad/s. This
clearly shows that this disturbance does not excite the high frequency oscillations.
3. From the right eigenvector matrix (six-state case), we can see that the magnitudes of the eigen-
vectors associated with all velocity-states, corresponding to 1.0816 rad/s, are appreciable and the
angles of the eigenvectors pertaining to v2 and v3 are in phase, and 180° out-off phase with v1.
From this, we can conclude that M1, M2, and M3 participate in this mode and masses M2 and M3
together swing against M1. From the waveforms, it can be verified that at t = 10 s, v2 and v3 reach
their positive peak together, whereas v1 attains a minimum peak at the same instant. Further, from
1
the time-period measurement, it can be confirmed that the frequency of oscillation is 2p =
1.0796 rad/s. 5.82
4. In common-mode, as predicted by the right eigenvectors, all masses move together as a single
rigid body.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 566 9/12/2017 4:05:17 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  567 

13.3.3 Case-2: With External F3 Without Any Damping

In this case, an external force F3 alone is applied at 1 s for a duration of 0.1 s. The time–domain plots
of all states are shown in Figure 13.4.

Response for short pulse at F3 (Case-2)

1

0.5
x1

0
0 5 10 15 20 25
1
x2

0.5

0
0 5 10 15 20 25
1
x3

0.5

0
0 5 10 15 20 25
0.1
0.05
v1

0
9.82 s 5.84 s 15.66 s
−0.05
0 5 10 15 20 25
0.5

0
v2

0.998 s
−0.5
0 5 10 15 20 25
0.5

0
v3

−0.5
0 5 10 15 20 25
Time (s )

Figure 13.4  Response to a Short Duration F3 without Any Damping.

From Figure 13.4, the following observations can be made:

1. As noted earlier, ramp response is dominant in x-states and only DC-offset is present in v-states. This
is also consistent due to the fact that x-states are obtained by integrating the corresponding v-states.
2. This disturbance triggers low and high frequencies in the responses.
3. For the high frequency oscillation of 6.4055 rad/s, the corresponding right-eigenvectors denote
that the velocities (v2 and v3) have high participation in this mode than the velocity v1. The angle
of right eigenvectors also show that v2 and v3 are 180° out-off phase with respect to each other.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 567 9/12/2017 4:05:23 PM

 568  |  Power System Analysis

It is further evident in Figure 13.4 where neither x1 nor v1 shows any traces of the high frequency
component. This mode is visible only in x2, x3, v2, and v3 states in addition to the low frequency
1.0816 rad/s. From the plotted signals of v2 and v3, if we measure the period by tracking the
1
consecutive peaks, we get the approximate frequency as 2p = 6.3 rad/s. Further, by noting
0.998
the signal peaks at t = 5 s, we can conclude that the mass M2 swings against the mass M3 in this
frequency. However, in the low frequency mode masses, M2 and M3 together swing against the
mass M1.

13.3.4 Case-3: With External F3 With Damping B1

In this case, an external force F3 is applied at 1 s for a duration of 0.1 s considering viscous type
damping for mass M1. Here, B1 is set to 5 N/m/s. The corresponding A is given by

 0 0 0 1 0 0
 0 0 0 0 1 0
 
 0 0 0 0 0 1  (13.14)
A=  
 −0 . 2 0 . 2 0 −0.5 0 0
 2 −22 20 0 0 0
 
 0 20 −20 0 0 0 

The eigenvalues of A are

l1 = 0.0000 + j 6.4055, l2 = 0.0000 − j 6.4055,

l3 = −0.036 + j1.0667, l4 = −0.036 − j1.0667,

l5 = 0 l6 = −0.4279

( l1 and l2 have negligible damping)

The matrix of right eigenvectors is given by,

l1 l2 l3 l4 l5 l6

0.0005∠164.36 0.0005∠ − 164.36 0.0879∠ − 162.22 0.0879∠162.22 0.5774 ∠180 0.5912∠180  x1

 
0.1118∠ − 20.12 0.1118∠20.12 0.4652∠ − 7.78 0.4652∠7.78 0.5774 ∠180 0.5000 ∠180  x2
 
0.1063∠159.87 0.1063∠ − 159.87 0.4933∠ − 7.55 0.4933∠7.55 0.5774 ∠180 0.4955∠180  x3
U= 
   
0.0035∠ − 105.64 0.0035∠105.64 0.0939∠ − 70.28 0.0939∠ + 70.28 0 0.2530  v1
     
0.7160 ∠69.88 0.7160 ∠ − 69.88 0.4965∠84.15 0.4965∠ − 84.15 0 0.214  v2
     v
0.6809∠ − 110.12 0.6809∠110.12 0.5265∠84.39 0.5265∠ − 84.39 0 0.2121  3

From the list of eigenvalues, it can be concluded that the natural response, i.e., zero-input response,
will have only one zero-eigenvalue signifying a constant DC-offset. It should be noted in this case,

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 568 9/12/2017 4:05:29 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  569 

the force F3 is in the form of a short-duration pulse and it just perturbs the operating point. The time–
domain plots of all states are shown in Figure 13.5.

Response for a short pulse at F3 with damping (Case-3)

0.2

0.1
x1

0
0 5 10 15 20 25 30 35 40 45 50
0.4
0.2
x2

0 5 10 15 20 25 30 35 40 45 50
0.4
0.2
x3

0 5 10 15 20 25 30 35 40 45 50
0.05

0
v1

−0.05
0 5 10 15 20 25 30 35 40 45 50
0.5
v2

−0.5
0 5 10 15 20 25 30 35 40 45 50
0.5

0
v3

−0.5
0 5 10 15 20 25 30 35 40 45 50
Time (s )

Figure 13.5  Response to a Short Duration F3 with Damping.

From Figure 13.5, the following observations can be made:

1. As per the right eigenvector matrix, it can be seen that both l5 and l6 are seen as a dominant
constant DC-offset in x-states. The high frequency component is not observable in x1, whereas it
is present in x2 and x3. All x-states show the low frequency component significantly.
2. A decaying DC component denoted by l6= −0.4279 is dominantly seen in v-states. This associa-
tion can be demonstrated by removing redundancy in x-states.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 569 9/12/2017 4:05:31 PM

 570  |  Power System Analysis

3. l3 and l4 demonstrate a decaying low frequency component. This performance is seen in almost
all states. Thus, as time progresses, the state v1 tends to zero, whereas the states v2 and v3 contain
only the high frequency component as this mode represented by (l1 and l2) does not receive sig-
nificant damping due to B1.

13.3.5 Case-4: With External F3 as a Step-signal Without

Any Damping
In this case, an external force F3 alone is applied at 1 s in the form of a step-signal and is maintained.
The forced response of the system is shown in Figure 13.6.

Response for F3 as step (5 N) with no damping (Case-4)

100
x1

50

0
0 5 10 15 20 25
100
x2

50

0
0 5 10 15 20 25
100
x3

50

0
0 5 10 15 20 25
10

5
v1

0
0 5 10 15 20 25
10
5
v2

0
0 5 10 15 20 25
10
v3

5
0
0 5 10 15 20 25
Time (s )

Figure 13.6  Response to a Step-change in F3 without Any Damping.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 570 9/12/2017 4:05:31 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  571 

From Figure 13.6, the following observations can be made:

1. The x-states show a t2-response as against ramp in Case-2, whereas v states depict ramp
response.
2. The v2 and v3 states also show the high frequency components. As shown by the right eigenvec-
tors, low frequency is visible in all v-states.

13.3.6 Case-5: With External F3 as a Step-signal With Damping B1

In this case, an external force F3 alone is applied at 1 s in the form of a step-signal and B1 is made
non-zero. The forced response of the system is shown in Figure 13.7.

Response for F3 as Step with damping (Case-5)

50
x1

0
0 5 10 15 20 25 30 35 40 45 50
50
x2

0
0 5 10 15 20 25 30 35 40 45 50
50
x3

0
0 5 10 15 20 25 30 35 40 45 50
1.5
1
v1

0.5
0
0 5 10 15 20 25 30 35 40 45 50
3
2
1
v2

0
−1
0 5 10 15 20 25 30 35 40 45 50
3
2
1
v3

0
−1
0 5 10 15 20 25 30 35 40 45 50
Time (s )

Figure 13.7  Response to a Step-change in F3 with Damping.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 571 9/12/2017 4:05:36 PM

 572  |  Power System Analysis

From Figure 13.7, the following observations can be made:

1. Unlike in Case-3 due to forced step-response, the x-states show a ramp response, whereas a con-
stant DC-offset response is observed in v-states.
2. In addition, as predicted by the right-eigenvectors, v2 and v3 contain the high frequency compo-
nent as its damping is very poor. However, the low frequency component is seen in all v-states and
is getting damped out.

13.3.7 Performance Analysis in the COI-reference-frame

Since it is not easy to analyse the response due to the zero mode and its association with x- and
v-states, in this section, the response of the system (discussed in the above cases) is analysed using
the centre-of-inertia (COI)-reference-frame. To facilitate this, the COI-based variables are defined as
follows:
1. The COI-displacement variable, xCOI is defined as

M1 x1 + M 2 x2 + M 3 x3
xCOI = (13.15)
M1 + M 2 + M3

2. The COI-velocity variable, vCOI is defined as

M1v1 + M 2 v2 + M 3v3
vCOI = (13.16)
M1 + M 2 + M 3

Using the COI variables, we can compute the COI-responses as

x1Co = ( x1 − xCOI ) , x2Co = ( x2 − xCOI ) , x3Co = ( x3 − xCOI )

and

v1Co = ( v1 − vCOI ) , v2Co = ( v2 − vCOI ) , v3Co = ( v3 − vCOI )

13.3.7.1 Response with Respect to COI: Case-1

The time–domain response obtained earlier is redrawn in terms of the COI variables. The plots are
shown in Figure 13.8.
From Figure 13.8, the following observations can be made:

1. By computing xCOI, it can be noted that the zero-mode effect in the form of ramp is clearly
seen, whereas in vCOI, the zero-mode effect in the form of a constant DC level is evident.
Further, by obtaining, xiCo or viCo, the zero-mode influences are completely removed from
the states.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 572 9/12/2017 4:05:37 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  573 

2. Now, unlike in Figure 13.3, xiCo and viCo contain only the low frequency component.

Response wrt COI (CASE−1)

xCOI 2

20 25 30 35 40
0.05
0.04
vCOI

0.03

0 5 10 15 20 25 30 35 40 45 50
0.01
x1Coi

−0.01
0 5 10 15 20 25 30 35 40 45 50
0.05
x2Coi

−0.05
0 5 10 15 20 25 30 35 40 45 50
0.05
x3Coi

−0.05
0 5 10 15 20 25 30 35 40 45 50
0.01
v1Coi

−0.01
0 5 10 15 20 25 30 35 40 45 50
0.05
v2Coi

−0.05
0 5 10 15 20 25 30 35 40 45 50
0.05
v3Coi

−0.05
0 5 10 15 20 25 30 35 40 45 50
Time (s )

Figure 13.8  Response to a Short Duration F1 without Any Damping

wrt COI Variables.

13.3.7.2 Response with Respect to COI: Case-2

Here, an external force F3 alone is applied at 1 s for a duration of 0.1 s and no damping is considered.
The time–domain responses obtained earlier are redrawn in terms of the COI variables. The plots are
shown in Figure 13.9.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 573 9/12/2017 4:05:38 PM

 574  |  Power System Analysis

Response wrt COI (CASE−2)

4

xCOI 2

0
0 5 10 15 20 25 30 35 40 45 50
0.04
vCOI

0.02

0 5 10 15 20 25 30 35 40 45 50
0.05
x1Coi

−0.05
0 5 10 15 20 25 30 35 40 45 50
0.5
x2Coi

−0.5
0 5 10 15 20 25 30 35 40 45 50
0.5
x3Coi

−0.5
0 5 10 15 20 25 30 35 40 45 50
0.05
v1Coi

−0.05
0 5 10 15 20 25 30 35 40 45 50
0.5
v2Coi

−0.5
0 5 10 15 20 25 30 35 40 45 50
0.5
v3Coi

−0.5
0 5 10 15 20 25 30 35 40 45 50
Time (s )

Figure 13.9  Response to a Short Duration F3 without Any Damping

wrt COI Variables.

From Figure 13.9, the following observations can be made:

1. As in the previous case, the xCOI and vCOI variable clearly show the zero-mode effects.
2. Subtracting xCOI from x-states removes the ramp effect and similarly, subtracting vCOI from
v-states removes constant DC.
3. A major difference between the present and the previous case is that the disturbance F3 triggers
the high-frequency mode in addition to the low-frequency component. Having removed the zero-
mode responses from the states, now the oscillatory modes are clearly observable.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 574 9/12/2017 4:05:39 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  575 

13.3.7.3 Response with Respect to COI: Case-3

Here, an external force F3 alone is applied at 1 s for a duration of 0.1 s considering damping. The
time–domain responses obtained earlier are redrawn in terms of the COI variables in Figure 13.10.

Response wrt COI (Case-3)

0.2
xCOI

0.1

0
0 5 10 15 20 25 30 35 40 45 50
0.05
vCOI

−0.05
0 5 10 15 20 25 30 35 40 45 50
0.05
x1Coi

−0.05
0 5 10 15 20 25 30 35 40 45 50
0.5
x2Coi

−0.5
0 5 10 15 20 25 30 35 40 45 50
0.5
x3Coi

−0.5
0 5 10 15 20 25 30 35 40 45 50
0.05
v1Coi

−0.05
0 5 10 15 20 25 30 35 40 45 50
0.5
v2Coi

−0.5
0 5 10 15 20 25 30 35 40 45 50
0.5
v3Coi

−0.5 50
0 5 10 15 20 25 30 35 40 45
Time (s )

Figure 13.10  Response to a Short Duration F3 with Damping wrt COI Variables.

From Figure 13.10, the following observations can be made:

1. The xCOI variable clearly shows the DC-offset response, whereas vCOI depicts the decaying DC
component in addition to the traces of the low-frequency component.
2. In the v1Co-state, only the low-frequency component is evident which is damping out.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 575 9/12/2017 4:05:40 PM

 576  |  Power System Analysis

13.3.7.4 Response with Respect to COI: Case-4

Here, an external force F3 is applied at 1 s in the form of a step-signal without damping. The response
of the states in terms of the COI variables is plotted in Figure 13.11.

Response wrt COI (Case-4)

100
xCOI

50
0
0 5 10 15 20 25
10
vCOI

5
0
0 5 10 15 20 25
0
x1coi

−0.5
−1
0 5 10 15 20 25
4
x2coi

2
0
0 5 10 15 20 25
5
x3coi

0
0 5 10 15 20 25
0.5
v1coi

−0.5
0 5 10 15 20 25
2
v2coi

0
−2
0 5 10 15 20 25
2
v3coi

0
−2
0 5 10 15 20 25
Time (s )

Figure 13.11  Response to a Step-signal F3 without Damping wrt COI Variables.

From Figure 13.11, the following observations can be made:

1. The xCOI shows a pure t2-response, whereas vCOI shows a ramp response.
2. Having removed one form of the common-mode, i.e., t2-response, the xiCo-states now demon-
strate another form of the zero-mode, i.e., a constant DC. In addition, it shows a dominant low
frequency component which is not clearly seen in Figure 13.6.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 576 9/12/2017 4:05:41 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  577 

3. Similar observations can be made with v-state where ramp response is dominant in Figure 13.6.
Now, in Figure 13.11, oscillatory components are clearly brought out.

13.3.7.5 Response with Respect to COI: Case-5

Here, an external force F3 is applied at 1 s in the form of a step-signal with damping. The response of
the states in terms of the COI variables is plotted in Figure 13.12.

Response wrt COI (Case-5)

50
xCOI

0
0 5 10 15 20 25 30 35 40 45 50
2
vCOI

0
0 5 10 15 20 25 30 35 40 45 50
0
x1coi

−0.5

−1
0 5 10 15 20 25 30 35 40 45 50
5
x2coi

−5
0 5 10 15 20 25 30 35 40 45 50
4
x3coi

0
0 5 10 15 20 25 30 35 40 45 50
0.5
v1coi

−0.5
0 5 10 15 20 25 30 35 40 45 50
5
v2coi

−5
0 5 10 15 20 25 30 35 40 45 50
5
v3coi

−5
0 5 10 15 20 25 30 35 40 45 50
Time (s )

Figure 13.12  Response to a Step-signal F3 with Damping wrt COI Variables.

From Figure 13.12, the following observations can be made:

1. From Figure 13.7, it is not easy to recognise the oscillatory modes hidden in the states. However,
in Figure 13.12, where states are replotted with respect to COI variables, it can be seen that

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 577 9/12/2017 4:05:41 PM

 578  |  Power System Analysis

the common-mode, i.e., zero-mode in the form of a ramp response is evident in the xCOI-state,
whereas the common-mode in the form of DC-offset is observed in vCOI-state.
2. The remaining common-mode response in the form DC-offset is now visible in xiCo-states.
This component cannot be easily identified from x-states as shown in Figure 13.7.
3. Subtracting vCOI from v-states completely removes the zero-mode response. Now viCo-states con-
tain only oscillatory modes.
Thus, the representation of states with respect to COI makes mode identification easier.

13.4  LINEARISATION OF POWER SYSTEM MODELLING EQUATIONS

The behaviour of a power system is described by a set of n first order non-linear ordinary differential
equations of the following form:

( )
x = F x , u , y (13.17)

The column vector x is referred to as the state vector, and its entries xi as state variables. The column
vector u is the vector of inputs to the system. These are the external signals that influence the perfor-
mance of the system. The column vector y is the vector of algebraic variables, say bus voltages. These
algebraic variables may be expressed in terms of the state variables in the following form:

( )
0 = g x, y (13.18)

where g is a vector of k non-linear functions. The set of equations (13.17) and (13.18) together consti-
tute the differential algebraic equations (DAEs) for the system.
Let x0 be the initial value of the state vector, u0 be the initial value of input vector, and y0 be the
initial value of output vector corresponding to the equilibrium point about which the small-signal
performance is to be investigated. Since x0, u0, and y0 satisfy (13.17), we have

( )
x0 = F x0 , u0 , y0 = 0 (13.19)

Let us perturb the system from the above equilibrium point, by letting

x = x0 + ∆ x , u = u0 + ∆u , y = y0 + ∆ y

where the prefix ∆ denotes a small deviation.

The new state must satisfy (13.17). Hence,

x = x0 + ∆x (13.20)

xi

( )
= F ( x0 + ∆x ) , ( u0 + ∆u ) , y0 + ∆y 

( )
As the perturbations are assumed to be small, the non-linear function F x , u , y can be expressed in
terms of Taylor’s series expansion. Neglecting the terms involving second and higher order powers of
Δx, Δu, and Δy, we can write,

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 578 9/12/2017 4:05:48 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  579 

 0 (
xi = xi 0 + ∆xi = fi ( x0 + ∆x ) , ( u0 + ∆u ) , y + ∆y 
 )
∂f i ∂f ∂f ∂f
(
= fi x0 , u0 , y0 + ) ∂x1
∆x1 +  + i ∆xn + i ∆u1 +  + i ∆ur +
∂ xn ∂u1 ∂ur

∂f i ∂f
+ ∆y1 +  + i ∆ym
∂y1 ∂ ym

( )
Since xi 0 = fi x0 , u0 , y0 = 0 , we obtain

∂f i ∂f ∂f ∂f
∆xi = ∆x1 +  + i ∆xn + i ∆u1 +  + i ∆ur +
∂x1 ∂ xn ∂u1 ∂ur

∂f i ∂f
+ ∆y1 +  + i ∆ym
∂y1 ∂ ym

for i = 1, 2, …, n.
Similarly, from (13.18), we have

∂g j ∂g j ∂g j ∂g j
0j = ∆x1 +  + ∆xn + ∆y1 +  + ∆ym
∂x1 ∂xn ∂y1 ∂ym

for j = 1, 2, …, k.
Therefore, the linearised forms of (13.17) and (13.18) in matrix notation can be written as

∆x = A1∆x + B1∆y + E ∆u (13.21)

0 = C ∆x + J ∆y (13.22)
where
 ∂f1 ∂f1   ∂f1 ∂f1 
     
 1∂ x ∂ x n  1∂ y ∂ ym 
A1 =     B1 =    
   
 ∂f n  ∂f n   ∂f n  ∂f n 
 ∂x1 ∂xn  ( n× n)  ∂y1 ∂ym  ( n× m)

 ∂f1 ∂f1 
 ∂u  ∂u 
 1 r
E =    
 
 ∂f n  ∂f n 
 ∂u1 ∂ur  ( n× r )

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 579 9/12/2017 4:05:57 PM

 580  |  Power System Analysis

 ∂g1 ∂g1   ∂g1 ∂g1 

   ∂  
 ∂x1 ∂ xn   y1 ∂ ym 
C =     J =    
   
 ∂g k 
∂g k   ∂g k 
∂g k 
 ∂x1 ∂xn  ( k × n)  ∂y1 ∂ym  ( k × m)

The above partial derivatives are evaluated at the equilibrium point about which the small perturbation
is being analysed.
From (13.22), writing ∆y = − J −1C ∆x and substituting it in (13.21), we get

∆x = Asys ∆x + E ∆u (13.23)

−1
where Asys = ( A1 − B1 J C ).
In general, to determine the small-signal stability behaviour of a non-linear dynamic system, it is
sufficient to obtain the eigenvalues of Asys matrix indicated above. However, for power system applica-
tions, determination of Asys matrix may be more involved because of the intricate relationship between
the state variables and the algebraic variables. In practice, the following are the two important methods
used for obtaining the state matrix:

1. Numerical approach
2. Analytical approach

Numerical approach:
In this approach, the state matrix is obtained using numerical differentiation. Here, starting from a
valid equilibrium condition, x0, a second state vector is created xi, in which the ith component of x0 is
x = xand
perturbed by a very small amount 0 +∆ x xi is computed using the F. This provides an intermediate
the
state matrix in which only the ith column is non-zero. This process is repeated until all columns of
the state matrix are obtained by sequentially perturbing all entries of x0. After constructing the state
matrix, eigenvalues can be obtained in the usual manner [12]. In SIMULINK toolbox [13], a function
known as linmod is available for numerical linearisation of systems.

Analytical approach:
In this approach, analytical expressions are obtained for all partial derivatives of variables. These
expressions are assembled in such a way that all elements are written only in terms of state variables.
In the literature, the following two basic approaches are employed:

1. Load flow Jacobian-based approach [7]

2. Current injection-based approach [1, 2]

13.4.1  Current-Injection Approach

The signal flow diagram for performing small-signal analysis of power system is shown in Figure 13.13.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 580 9/12/2017 4:05:58 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  581 

ith generator

V ith generator bus

gi

Stator Ckts. Network

Vj
Igi
Excitation Efdi Rotor Elec. Ckts. YBUS Load
Controller (4 states)
(ne states) I
Rotor Mech. Sys. Lj

(2 states) j th load bus

Speed
Speed−governor Tmi
(n states)
s

Figure 13.13  Signal Flow Diagram for Current-injection Approach.

The steps involved are enumerated below:

1. The generator differential equations are linearised to obtain

∆x g =  Ag  ∆xg +  Bg  ∆Vg +  E g  ∆uc

(13.24)

Note that the generator stator transients and network transients are neglected. In addition, the
loads are modelled as static loads. Now, the objective is to write ∆Vg in terms of the state
vector xg.
2. The generator and load are interfaced to the network as current injections as shown in
Figure 13.13. This leads to the following linearised interfacing equation in terms of DQ variables:

[YBUS ] ∆VQD = ∆I g − ∆I L (13.25)

where Ig and IL are vectorised currents with appropriate zeros. Note that the load current is treated
as a negative current injection.
3. The currents ∆Ig and ∆IL are expressed as

∆I g = C g  ∆xg +  Dg  ∆Vg , and ∆I L = [YL ] ∆VL (13.26)

Further, we have,

T T
∆Vg =  P g  ∆VQD , and ∆VL =  P l  ∆VQD
   

where [P g] and [P l] are interfacing matrices.

  Thus, we have,

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 581 9/12/2017 4:06:00 PM

 582  |  Power System Analysis

T T
∆I g = C g  ∆xg +  Dg   P g  ∆VQD , and ∆I L = [YL ]  P l  ∆VQD (13.27)
   

Using the above equations in (13.25), we can solve for ∆VQD as

∆VQD = [ N g ]∆xg (13.28)

Again, using the interfacing matrix, ∆Vg can be obtained as

T
∆Vg =  P g   N g  ∆xg (13.29)
 

Substituting the above equation in (13.24), the final state-feedback matrix can be determined as
T
[ At ] =  Ag  +  Bg   P g  Ng 
 

Here, the notations are used to symbolically represent the method.

13.4.1.1 Linearisation of Generator Equations

The generator equations are enumerated below for ease of reference.

d-axis Equations:

For ith generator, the differential equations are written in the state-space form as follows:

dy H 1

dt
=
Td′′
[ −y H +y d ] (13.30)

dy F 1  xd′ E fd 
=  −y F + y d +  (13.31)
dt Td′  ( xd − xd′ ) 
where

y d = xd′′id + Eq′′ (13.32)

 ( xd′ − xd′′ )y H ( xd − xd′ ) 

Eq′′ =  + xd′′y F  (13.33)
 xd′ xd xd′ 
q-axis Equations:

The differential equations are written in the state-space form as follows:

dy G 1
=  −y G + y q  (13.34)
dt Tq′  

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 582 9/12/2017 4:06:04 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  583 

dy K 1
=  −y K + y q  (13.35)
dt Tq′′  
where

y q = xq′′iq − Ed′′ (13.36)

 ( xq′ − xq′′)y K ( xq − xq′ ) 

Ed′′ = −  + xq′′y G  (13.37)
 xq′ xq xq′ 
Stator Equations:

Neglecting stator transients and ignoring speed variations, the stator d- and q-axes voltage equations
are given by

vd = −id Ra −y q (13.38)

vq = −iq Ra +y d (13.39)

where vd and vq represent d- and q- axis generator terminal voltages, respectively.

Using (13.32) and (13.36) in (13.38) and (13.39), we have

vd = −id Ra − xq′′iq + Ed′′(13.40)

vq = −iq Ra + xd′′id + Eq′′(13.41)

Neglecting Ra, we have from (13.38) and (13.39),

vq = y d (13.42)

vd = −y q (13.43)

     vq + jvd = Vg e
(
j q g −d ) (13.44)

vq = Vg cos (d − q g ) (13.45)

vd = −Vg sin (d − q g )
(13.46)

y d = Vg cos (d − q g )
(13.47)

y q = Vg sin (d − q g ) (13.48)

Using (13.32), (13.36), (13.47), and (13.48) we have,

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 583 9/12/2017 4:06:12 PM

 584  |  Power System Analysis

vq − Eq′′ Vg cos(d − q g ) − Eq′′

id = = (13.49)
xd′′ xd′′

Ed′′ − vd Ed′′ + Vg sin (d − q g )

iq = = (13.50)
xq′′ xq′′

From (13.33) and (13.37), we have,

Eq′′ = C1y H + C2y F (13.51)

Ed′′ = −C3y K − C4y G (13.52)

where

( xd′ − xd′′ ) (13.53)

C1 =
xd′

( xd − xd′ ) xd′′
C2 = (13.54)
xd xd′

( xq′ − xq′′)
C3 = (13.55)
xq′

( xq − xq′ ) x ′′q
C4 = (13.56)
xq xq′

Swing Equations:
dd
= Smw B (13.57)
dt

dSm 1
= [ − DSm + Tm − Teg ] (13.58)
dt 2H

In terms of the flux-linkages and the generator winding currents, the electromagnetic torque is
given by

Teg = (y d iq −y q id ) (13.59)

Using eqs. (13.47) - (13.50), (13.51) and eq. (13.52), the above torque expression is modified as

C C C
Teg =  2 (Vgy F sin(d − q g )) + 1 (Vgy H sin(d − q g )) − 4 (Vgy G cos(d − q g ))
x
 d′′ x ′′
d xq′′

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 584 9/12/2017 4:06:20 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  585 

C3  Vg2sin(2(d − q g ))  
− (Vgy K cos(d − q g )) + C5   (13.60)
xq′′  2  

where

 xd′′ − xq′′ 
C5 =   (13.61)
 xd′′ xq′′ 

Modification of Rotor Differential Equations:

Using (13.47) and (13.48) in (13.30), (13.31), (13.34), and (13.35) we get

dy F 1   xd′  
=  −y F + Vg cos(d − q g ) +   E fd  (13.62)
dt Td′   d
x − x ′
d 

dy H 1
= [ −y H + Vg cos(d − q g )] (13.63)
dt Td′′

dy G 1
= [ −y G + Vg sin(d − q g )] (13.64)
dt Tq′

dy K 1
= [ −y K + Vg sin(d − q g )] (13.65)
dt Tq′′

Derivation of IDg and IQg:

From (13.49) and (13.50), we can write,

iq + jid =
(
Ed′′ + Vg sin d − q g ) + j Vg cos (d − q g ) − Eq′′ (13.66)
xq′′ xd′′

(
I Qg + jI Dg = iq + jid e jd ) 
(13.67)

(
 Ed′′ + Vg sin d − q g
I Dg = 
)  sind + Vg cos (d − q g ) − Eq′′  cosd (13.68)
 xq′′   xd′′ 
   

(
 Ed′′ + Vg sin d − q g
I Qg = 
)  cosd − Vg cos (d − q g ) − Eq′′  sind (13.69)
 xq′′   xd′′ 
   

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 585 9/12/2017 4:06:24 PM

 586  |  Power System Analysis

Now, using (13.51) and (13.52), we have,

I Dg =
1
xq′′
( )
[ −C3y K − C4y G + Vg sin d − q g ]sind +

+
1
xd′′
( )
[Vg cos d − q g − C1y H − C2y F ]cosd (13.70)

I Qg =
1
xq′′
( )
[ −C3y K − C4y G + Vg sin d − q g ]cosd −

−
1
xd′′
( )
[Vg cos d − q g − C1y H − C2y F ]sind (13.71)

For the ith generator, the differential equations are linearised as

∆x g =  Ag  ∆xg +  Bgp  ∆V p +  E  ∆u (13.72)

( 6 × 6)   (6× 2) g  g  (6×1) c

where ∆uc is the vector of small perturbation in the reference input variables of the generator
( )
T
controllers given by ∆uc =  ∆Vref + ∆Vs , ∆w nL  . With ∆wnL assumed to be zero, we have
∆uc =  ∆Vref + ∆Vs  . Where wnL denotes load reference setting, Tm0 (or Po).
p
Note that ∆V  g is a small deviation in the generator terminal voltage expressed in polar coordinates
given by
Vg 0 ∆q g 
∆Vgp =  
 ∆Vg 

Further, the state-vector xg is given by

xg = [d Sm y F y H y G y K ]T

Generator current in the KRON’s reference frame is used as the output variable. Linearising (13.70)
and (13.71), we have,

 ∆I Dg  p (13.73)
 = C g  ∆xg +  Dg  ∆Vg
p
∆I g = 
∆I
 Qg 

The non-zero elements of  Ag  ,  Bgp  , C g  and  Dgp  matrices are given by

       

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 586 9/12/2017 4:06:37 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  587 

Ag (1, 2) = w B

1  C2Vg 0y F 0 cos(d 0 − q g 0 ) C1Vg 0y H 0 cos(d 0 − q g 0 ) C4Vg 0y G 0 sin(d 0 − q g 0 )

Ag (2,1) = −  + +
2 H  xd′′ xd′′ xq′′

C3Vg 0y K 0 sin(d 0 − q g 0 ) 
+ + C5Vg20 cos(2(d 0 − q g 0 )) 
xq′′ 

D
Ag (2, 2) = −
2H

1  C2Vg 0 sin(d 0 − q g 0 ) 
Ag (2, 3) = −  
2H  xd′′ 

1  C1Vg 0 sin(d 0 − q g 0 ) 
Ag (2, 4) = −  
2H  xd′′ 

1  C4Vg 0 cos(d 0 − q g 0 ) 
Ag (2, 5) =  
2H  xq′′ 

1  C3Vg 0 cos(d 0 − q g 0 )  Vg 0 sin(d 0 − q g 0 )

Ag (2, 6) =   Ag (3,1) = −
2 H  xq′′  Td′

1
Ag (3, 3) = −
Td′

1  xd′ 
Ag (3, 7) =   for E fd -state
Td′  xd − xd′ 

Vg 0 sin(d 0 − q g 0 )
Ag ( 4,1) = −
Td′′

1
Ag ( 4, 4) = −
Td′′

Vg 0 cos(d 0 − q g 0 )
Ag (5,1) =
Tq′

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 587 9/12/2017 4:06:42 PM

 588  |  Power System Analysis

1
Ag (5, 5) = −
Tq′

Vg 0 cos(d 0 − q g 0 )
Ag (6,1) =
Tq′′

1
Ag (6, 6) = −
Tq′′

1  C2y F 0 cos(d 0 − q g 0 ) C1y H 0 cos(d 0 − q g 0 ) C4y G 0 sin(d 0 − q g 0 )

Bgp (2,1) =  + +
2 H  xd′′ xd′′ xq′′

C3y K 0 sin(d 0 − q g 0 ) 
+ + C5Vg 0 cos(2(d 0 − q g 0 )) 
xq′′ 

1  C2y F 0 sin(d 0 − q g 0 ) C1y H 0 sin(d 0 − q g 0 ) C4y G 0 cos(d 0 − q g 0 )

Bgp (2, 2) = −  + −
2 H  xd′′ xd′′ xq′′

C3y K 0 cos(d 0 − q g 0 )  sin(d 0 − q g 0 )

− + C5Vg 0 sin(2(d 0 − q g 0 ))  , Bgp (3,1) =
xq′′  Td′

cos(d 0 − q g 0 )
Bgp (3, 2) =
Td′

sin(d 0 − q g 0 )
Bgp ( 4,1) =
Td′′

cos(d 0 − q g 0 )
Bgp ( 4, 2) =
Td′′

cos(d 0 − q g 0 )
Bgp (5,1) = −
Tq′

sin(d 0 − q g 0 )
Bgp (5, 2) =
Tq′

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 588 9/12/2017 4:06:56 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  589 

cos(d 0 − q g 0 )
Bgp (6,1) = −
Tq′′

sin(d 0 − q g 0 )
Bgp (6, 2) =
Tq′′

Vg 0 sind 0 cos(d 0 − q g 0 ) Vg 0 cosd 0 .sin(d 0 − q g 0 )

C g (1,1) = − + I Qg0
xq′′ xd′′

C2 cosd 0
C g (1, 3) = −
xd′′

C1 cosd 0
C g (1, 4) = −
xd′′

C4 sind 0
C g (1, 5) = −
xq′′

C3 sind 0
C g (1, 6) = −
xq′′

Vg 0 sind 0 sin(d 0 − q g 0 ) Vg 0 cosd 0 cos(d 0 − q g 0 )

C g (2,1) = + − I Dg0
xd′′ xq′′

C2 sind 0
C g (2, 3) =
xd′′

C1 sind 0
C g (2, 4) =
xd′′

C4 cosd 0
C g (2, 5) = −
xq′′

C3 cosd 0
C g (2, 6) = −
xq′′

cosd 0 sin(d 0 − q g 0 ) sind 0 cos(d 0 − q g 0 )

Dgp (1,1) = −
xd′′ xq′′

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 589 9/12/2017 4:07:06 PM

 590  |  Power System Analysis

cosd 0 cos(d 0 − q g 0 ) sind 0 sin(d 0 − q g 0 )

Dgp (1, 2) = +
xd′′ xq′′

p sind 0 sin(d 0 − q g 0 ) cosd 0 cos(d 0 − q g 0 )

Dg (2,1) = − −
xd′′ xq′′

sind 0 cos(d 0 − q g 0 ) cosd 0 sin(d 0 − q g 0 )

Dgp (2, 2) = − +
xd′′ xq′′

13.4.1.2 Linearisation of Single Time-constant Static Exciter

A single time–constant static exciter as shown in Figure 13.14 is used.
Vc
E fdmax
−
+ KA
V REF E
fd
1 + s TA
+
E fdmin

VS

Figure 13.14  Single Time-constant Static Excitation System.

dE fd

dt
=
1 
TA 
( )
− E fd + K A Vref + Vs − Vc  (13.74)

The updated state-vector xg is given by

xg = [d Sm y F y H y G y K E fd ]T( 7×1)

After linearising (13.74), the non-zero elements of the matrices are given by,
1
Ag (7, 7) = −
TA

KA
Bgp (7, 2) = −
TA

KA
E g (7,1) =
TA

Note: In addition, two more the IEEE-type exciters such as DC1A and AC4A exciters are imple-
mented [14]. This adds new state variables xE given by

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 590 9/12/2017 4:07:09 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  591 

xE = [ vR x B x F ](3×1)
T

Thus, the new state-vector xg is

T
x g = d Sm y F y H y G y K E fd xET 
  (10 ×1)

13.4.1.3 Linearisation of Reheat Type Steam Turbine and its

Speed-governor Model
A reheat-type steam turbine and its speed-governor model are shown in Figures 13.15 and 13.16.

+ +

PM
+ +

F F F
HP IP LP

P
GV 1 1 1
1 + sT 1 + sTRH 1 + sT
CH CO

Figure 13.15  Tandem Compounded, Single-reheat-type Steam Turbine Model.

P P
O max

+
K(1+sT2) − 1 1 PGV

p.u slip 1+sT1 T3 s

Q
−

Pmin

Figure 13.16  Model for Speed-governor for Steam Turbines.

The above model is modified as shown in Figure 13.17 for the purpose of linearisation.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 591 9/12/2017 4:07:12 PM

 592  |  Power System Analysis

KT2

T1

T y1 +
pu slip K 1− 2
T1
+
Q
1+ sT1

Figure 13.17  Modified Model for Speed-governor for Steam Turbines.

The differential equations for the steam turbine and the associated speed-governor are given by,
dx1 1
=
dt TCH
( PGV − x1 ) (13.75)

dx2 1
= ( x1 − x2 ) (13.76)
dt TRH

dx3 1

dt
=
TCO
( x2 − x3 ) (13.77)

dy1 1   T2  
=  K 1 −  Sm − y1  (13.78)
dt T1   T1  

dPGV 1  T2 
=  P0 − K Sm − y1 − PGV  (13.79)
dt T3  T1 

PM = FHP x1 + FIP x2 + FLP x3 (13.80)

Notes:

1. P0 is a constant.
2. (13.80) is used in (13.58) in place of Tm.
The state-vector xg is now appended with xT to get
T
xg = d Sm y F y H y G y K E fd xET xT T 
  (15×1)

where
xT = [ x1 x2 x3 y1 PGV ](5×1)
T

After linearisation of the above equations, the non-zero elements of [Ag] are given by

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 592 9/12/2017 4:07:24 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  593 

FHP FIP FLP

Ag (2,11) = ; Ag (2,12) = ; Ag (2,13) =
2H 2H 2H

1 1
Ag (11,11) = − ; Ag (11,15) =
TCH TCH

1 1
Ag (12,11) = ; Ag (12,12) = −
TRH TRH

1 1
Ag (13,12) = ; Ag (13,13) = −
TCO TC 0

K  T2   1
Ag (14, 2) =  1− ; Ag (14,14) = −
T1  T1   T1

K T2 1 1
Ag (15, 2) = − ; Ag (15,14) = − ; Ag (15,15) = −
T1 T3 T3 T3

Note: In addition, one more hydro-type turbine is implemented [14]. This adds a new state variable z
to xg so that the new state-vector xg is given by

xg = [d Sm y F y H y G y K E fd xET xT T z ]T(16×1)

13.4.1.4 Transformation of Voltage Matrices From Polar

to Rectangular Forms
In the above linearisation process, it can be observed that  Bgp  and  Dgp  matrices are derived,
   
employing voltage quantities in the polar form. However, while performing interfacing of generators
to the network, it is necessary to write the equation in rectangular form. To perform this interfacing, a
transformation matrix is derived in the following lines:
The generator terminal voltage is given by,

(
Vg ∠q g = Vg cosq g + jsinq g )
Now consider,
∂Vg ∠q g

∂q g
( )
∆q g = Vg 0 −sinq g 0 + jcosq g 0 ∆q g (13.81)

 j  p +q  
 g0 
= Vg 0 ∆q g  e  2 

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 593 9/12/2017 4:07:35 PM

 594  |  Power System Analysis

Let
p 
j  +q g 0 
2 
Vg 0 ∆q g e = ∆VQg + j ∆VDg (13.82)

 − j  +q g 0  
p 
 (
Vg 0 ∆q g = Re ∆VQg + j ∆VDg e

2 
 )
 

 ( −Vg 0sinq g 0 − jVg 0 cosq g 0 ) 


(
Vg 0 ∆q g = Re  ∆VQg + j ∆VDg ) Vg 0 
 

1
Vg 0 ∆q g =  −VDg 0 ∆VQg + VQg 0 ∆VDg  (13.83)
Vg 0  

where, VQg 0 = Vg 0 cosq g 0 and VDg 0 = Vg 0 sinq g 0 .

We know that

Vg2 = VQg
2 2
+ VDg

Now consider,

2Vg 0 ∆Vg = 2VQg 0 ∆VQg + 2VDg 0 ∆VDg

VQg 0 VDg 0
∆Vg = ∆VQg + ∆VDg (13.84)
Vg 0 Vg 0

From (13.83) and (13.84), we can write

Vg 0 ∆q g  1  −VDg 0 VQg 0   ∆VQg   ∆VQg 

 ∆V = V    = [P]  
 g  Vg 0  Qg 0 VDg 0   ∆VDg   ∆VDg 

Denoting

Vg 0 ∆q g   ∆VQg 
∆Vgp =   and ∆Vgr =  
 ∆Vg   ∆VDg 

Thus,

∆Vgp = [ P ] ∆Vgr

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 594 9/12/2017 4:07:43 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  595 

where

1  −VDg 0 VQg 0 
[P] = V VDg 0 
Vg 0  Qg 0

Further, it can be seen that

[ P ]−1 = [ P ]
The matrices  Bgr  is now obtained as
 
 Bgr  =  Bgp  [ P ]
   
Similarly,  Dgr  can be obtained as
 
 Dgr  =  Dgp  [ P ]
   

The superscript r indicates the representation of matrices in rectangular coordinates.

13.4.1.5  Linearisation of Network Equations

Neglecting the transient, the network equation is written using the admittance matrix as

[YBUS ] V  =  I  (13.85)
where each element Yij = Gij + jBij. Note that transmission lines are modelled as a nominal π-circuit
and transformers with off-nominal-turns-ratio are modelled as equivalent π- circuit [15].
Now consider writing the net current injection at bus i = 1, i.e., IQ1 + jID1 in the expanded
form as

(G11 + jB11 ) (VQ1 + jVD1 ) + …+ (G1 j + jB1 j ) (VQj + jVDj ) + … = I Q1 + jI D1

or

(G11VQ1 − B11VD1 ) + j (G11VD1 + B11VQ1 ) + …+ (G1 jVQj − B1 jVDj ) +

( )
+ j G1 jVDj + B1 jVQj + … = I Q1 + jI D1

Comparing real and imaginary parts on both sides, we have,

nb
∑ (G1 jVDj + B1 jVQj ) = I D1 (13.86)
and j =1

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 595 9/12/2017 4:07:48 PM

 596  |  Power System Analysis

nb
∑ (G1 jVQj − B1 jVDj ) = I Q1 (13.87)
j =1

Thus, the jth term can be represented in the matrix form as,

 B1 j G1 j  VQj   I D1 j 
G =
 1j − B1 j  VDj   I Q1 j 

or
YDQ (1, j )VQDj = I DQ1 j

Using the above result in (13.86) and (13.87), we get,

nb
∑YDQ (1, j )VQDj = I DQ1
j =1

For nb-bus system, the bus current injections is written in the compact form as

YDQ 
 ( 2nb × 2nb ) ∆VQD ( 2nb ×1) = ∆I DQ ( 2nb ×1)
(13.88)


where each element of YDQ  is a 2 × 2 matrix. For example,

 
 Bij Gij 
YDQ (i , j ) = 
Gij − Bij 

The variables ∆VQDi and ∆IDQi are vectors with elements,

 ∆VQi   ∆I Di 
∆VQDi =   and ∆I DQi =  
 ∆VDi   ∆I Qi 

Note that the voltages are expressed with ∆VQi preceding ∆VDi. On the other hand, the currents are
expressed with ∆IDi preceding ∆IQi. This is deliberately done so that the matrix YDQ  is a real sym-
 
metric matrix (if phase-shifting transformers are not considered). Also note that the admittance matrix
representation is independent of the operating point.

13.4.1.6 Linearisation of Load Currents

The static loads are modelled employing polynomial representation given by

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 596 9/12/2017 4:07:54 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  597 

 V  mp
V 
mi
V 
mz 
PL = PL 0   p1 +   p2 +   p3  (13.89)
 V0   V0   V0  

 V  n p  V  z 
n n
V  i
QL = QL 0   r1 +   r2 +   r3  (13.90)
 V0   V0   V0 


where PL0 is the initial value of the active component of load, QL0 is the initial value of the reactive
component of load, and V0 represents the initial value of the bus voltage magnitude at load bus. Also
note that for real power, mp= 0.0 represents constant power, mi = 1.0, represents constant current and
mz = 2.0 represents constant impedance components; similarly, for reactive power, np = 0.0 repre-
sents constant power, ni = 1.0 represents constant current and nz = 2.0 represents constant impedance
components.
In order to linearise the load currents, consider,
*
 P + jQL  (13.91)
I Q + jI D =  L 
 V 
 PL − jQL  (13.92)
=
 V2 
 
 V 

 P − jQ 
=  L 2 L  V (13.93)
 V 

 P − jQ 
=  L 2 L  (VQ + jVD ) (13.94)
 V 

P Q  P Q 
=  L2 VQ + L2 VD  + j  L2 VD − L2 VQ  (13.95)
V V  V V 

Comparing the like terms, we get the Q and D components of the load current for jth load bus as

 VQ  V 
I Q = PL  2  + QL  D2  (13.96)
V  V 

V   VQ 
I D = PL  D2  − QL  2  (13.97)
V  V 

where the load bus voltage magnitude V is given by

V = VQ2 + VD2

Linearising (13.96) and (13.97), we get

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 597 9/12/2017 4:08:01 PM

 598  |  Power System Analysis

VQ 0 V P Q
∆I Q =  2 ∆PL + D20 ∆QL + L20 ∆VQ + L20 ∆VD
 V0 V0 V0 V0

 −2  
(  V0 
)
+ PL 0VQ 0 + QL 0VD 0  3  ∆V  (13.98)


V VQ 0 P Q
∆I D =  D20 ∆PL − 2 ∆QL + L20 ∆VD − L20 ∆VQ
 V0 V0 V0 V0

 −2  
(  V0 
)
+ PL 0VD 0 − QL 0VQ 0  3  ∆V  (13.99)


Considering only the first term in (13.89), one component of ∆PL can be obtained as follows:
mp
 1 ( m p −1)
= PL 0   p1 m p V0 ∆V
 V0 

P 
= p1 m p  L 0  ∆V
 V0 

Following the above procedure for the remaining two terms in (13.89), we have

P 
∆PL = mk  L 0  ∆V (13.100)
 V0 
where

mk = m p p1 + mi p2 + mz p3

Similarly, ∆QL can be obtained as

Q 
∆QL = nk  L 0  ∆V (13.101)
 V0 
where

nk = n p r1 + ni r2 + nz r3

Using the result given in (13.84), we can write

VQ 0 VD 0
∆V = ∆VQ + ∆VD (13.102)
V0 V0

Substitution of (13.102), (13.101), and (13.100) in (13.98) and (13.99) yields

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 598 9/12/2017 4:08:10 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  599 

∆I Q = GQQ ∆VQ + BQD ∆VD

∆I D = − BDQ ∆VQ + GDD ∆VD

or
 ∆I D   − BDQ GDD   ∆VQ 
 ∆I  = G BQD   ∆VD 
(13.103)
 Q   QQ
where

Q  VQ20  PL 0  VQ 0VD 0  
BDQ =  L20  ( nk − 2) 2 + 1 − 2  ( mk − 2)   (13.104)
 V0  V0  V0  V02  


Q  V2  P  VQ 0VD 0  
BQD =  L20  ( nk − 2) D20 + 1 + L20  ( mk − 2)   (13.105)
 V0  V0  V0  V02  

P  VQ20  QL 0  VQ 0VD 0  
GQQ =  L20  ( mk − 2) 2 + 1 + 2  ( nk − 2)   (13.106)
 V0  V0  V0  V02  


P  V2  Q  VQ 0VD 0  
GDD =  L20  ( mk − 2) D20 + 1 − L20  ( nk − 2)   (13.107)
 V0  V0  V0  V02  

Writing (13.103) in a compact form for jth load bus, we get,

∆I Lj = [YL ] j ∆VLj (13.108)

where
T
∆I Lj =  ∆I DLj ∆I QLj 

T
∆VLj =  ∆VQLj ∆VDLj 

and

 − BDQ GDD 
[YL ] j = G BQD 
 QQ

Thus, for ml load buses, the current deviation vector can be written as

∆I L( 2ml ×1) = [YL ] ∆VL( 2ml ×1)(13.109)

where [YL ](2m × 2m ) is a block diagonal matrix.

l l

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 599 9/12/2017 4:08:18 PM

 600  |  Power System Analysis

13.4.1.7 Interfacing of Generator and Loads to Network

Let the number of generators in the system be ng and the number of loads be ml. Rewriting (13.88),
we have,

YDQ  ∆VQD = [ PG ] ∆I G − [ PL ] ∆I L (13.110)

 
where,
• ∆IG(2ng × 1) is the complete vector of ng-generator currents as per (13.73).
• [PG] is a (2nb × 2ng) matrix with elements

1 0 
PG (i , j ) =  
0 1 

if generator j is connected to bus i.

Otherwise,

0 0 
PG (i , j ) =  
0 0

• ∆IL(2ml×1) is the complete vector of ml-load currents.

• [PL] is (2nb × 2ml) matrix. PL (i, j) is a unit matrix of dimension 2 × 2 if load j is connected to
bus i; otherwise, PL(i, j) is a null matrix.
• The sign associated with IL is taken as negative. This is because a load current flowing away from
the bus (with load convention) is treated as a negative current injection.
Also, note that the vector of load bus voltage is given by,

∆VL( 2ml ×1) = [ PL ] ∆VQD ( 2nb ×1)(13.111)

T

and for generator buses, the vector of DQ variables are,

∆VG ( 2ng ×1) = [ PG ] ∆VQD ( 2nb ×1)(13.112)

T

Using (13.111) in (13.109), we get,

∆I L = [YL ][ PL ] ∆VQD s(13.113)

T

The generator current vector, ∆IG is a collection of the quantities ∆Ig1, ∆Ig2, ∆Ig3, …, ∆Igng and using
(13.73), ∆Ig can be expressed as

∆I G = [CG ] ∆X G − [YG ] ∆VG (13.114)

where

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 600 9/12/2017 4:08:24 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  601 

T
∆X G =  ∆xTg1 ∆xTg 2  ∆xTgng 
 

∆VGT =  ∆VgT1r T r
∆VgT2r  ∆Vgn
 g 

T
∆I G =  ∆I Tg1 ∆I Tg 2  ∆I Tgng 
 

[CG ] = BlkDiag C g1 C g 2  C gng 



[YG ] = BlkDiag  − Dgr1 r 

− Dgr 2  − Dgng 

The superscript r indicates the representation of matrices in rectangular coordinates.

Now, using (13.112) in (13.114), we get

∆I G = [CG ] ∆X G − [YG ][ PG ] ∆VQD (13.115)

T

Substituting (13.113) and (13.115) in (13.110) we get

YDQ  ∆VQD = [ PG ][CG ] ∆X G − [ PG ][YG ][ PG ]T ∆VQD − [ PL ][YL ][ PL ]T ∆VQD

 

Rearranging the terms associated with ∆VQD , we get

 ′  ∆VQD = [ PG ][CG ] ∆X G (13.116)

YDQ 
where

 ′  = YDQ  + [ PG ][YG ][ PG ]T + [ PL ][YL ][ PL ]T

YDQ   

Solving for ∆VQD from (13.116) and using it in (13.112), we get ∆VG as,
−1
∆VG = [ PG ] YDQ [ PG ][CG ] ∆X G (13.117)
T
′ 

From (13.72), the collection of all the generator equations is expressed by

∆X G = [ AG ] ∆X G + [ BG ] ∆VG + [ EG ] ∆U c (13.118)

where

[ AG ] = BlkDiag  Ag1 Ag 2  Agng 



[ BG ] = BlkDiag  Bgr1 r 
Bgr 2  Bgng 

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 601 9/12/2017 4:08:32 PM

 602  |  Power System Analysis

[ EG ] = BlkDiag  Eg1 E g 2  E gng 



∆U cT =  ∆ucT1 , ∆ucT2  ∆ucn

T 
 g 

Substituting (13.117) in (13.118) we get,

∆X G = [ AT ] ∆X G + [ EG ] ∆U c (13.119)
where
−1
[ AT ] = [ AG ] + [ BG ][ PG ]T YDQ
′  [ PG ][CG ](13.120)

Since the analytical approach provides better insight into linearisation of system of equations and is
more accurate compared to the numerical approach (which may suffer from the problem of inaccurate
estimates depending upon the amount of perturbation chosen), the current injection-based analytical
method is employed and the system matrix [AT] given in (13.120) is used to perform the eigenvalue
analysis.

13.5  PARTICIPATION MATRIX

In the analysis of large power systems, it is desirable to know the level of impact that a set of state
variables has on a given mode so that suitable methods can be devised to control those modes.
In this regard, eigenvalue sensitivity analysis-participation factor provides a tool to identify the
nature of modes. In the following lines, a derivation has been presented to obtain the participation
matrix [4].
Assuming that the eigenvalues are distinct, we have

Au j = l j u j (13.121)
Now consider,
 ∂A   ∂ u j   ∂l j   ∂u j 
 ∂a  u j + A  ∂a  =  ∂a  u j + l j  ∂a 
rs rs rs rs

where ars is an element in the A matrix in the rth row and sth column position.
Simplifying the above expression, we get,

 ∂A   ∂u j   ∂ u j   ∂l j 
 ∂a  u j + A  ∂a  − l j  ∂a  =  ∂a  u j
rs rs rs rs

 ∂A   ∂ u j   ∂l j 
(13.122)
 ∂a  u j + ( A − l j I )  ∂a  =  ∂a  u j
rs rs rs

Pre-multiplying the above equation by the left eigenvector wj, we get,

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 602 9/12/2017 4:08:38 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  603 

 ∂A   ∂u j   ∂l j 
wj  u + w ( A − l I )   = w  ∂a  u j (13.123)
 ∂ars 
j j j j
 ∂ars  rs

Since we know that w j ( A − l j I ) = 0 (from the definition of the left eigenvectors) we have,

 ∂l j   ∂A 
wj   uj = w j  uj
∂
 rs 
a  ∂ars 
∂A
Since is a scalar, we can write
∂ars
 ∂A 
wj  uj
∂l j  ∂ars 
=
∂ars w j uj
∂A
Note that in all elements are zero except (r, s)th element, which is 1. Therefore,
∂ars
∂l j w jr usj
= (13.124)
∂ars w j uj

where w jr = w j ( r ) and usj = u j ( s ) , rth element and sth element in the vectors wj and uj,
respectively.
Participation matrix is obtained when r = s = k in (13.124), (i.e., when eigenvalue sensitivity
is obtained corresponding to the diagonal element, akk of A). With this substitution for r and s,
we get,

∂l j w jk ukj
Pjk = = (13.125)
∂akk w j uj
Notes:

1. In the above expression for the participation factor, a division by a scalar wj  uj normalises the
eigenvectors.
2. In MATLAB, if the eig function is used, the eigenvectors are inherently normalised. If eigs are
used, the normalisation should be carried out using the above expression.

Note that the value of Pjk is decided based on the value of ukj for a given wjk. From this, it can be
said that ukj measures the activity of the kth state variable in the jth mode, whereas wjk weighs the
contribution of this activity to the mode. Thus, Pjk can be used as a relative measure to indicate the net
participation of the kth state variable in building the time response of the jth mode [16, 17].
The following observations are made:
1. The components obtained in (13.124) are referred to as dimensional ‘generalised participation’.
As a special case, when we set r = s = k, the components constitute a P matrix. The entries, Pjk
with j, k = 1, 2, …, n, of the P matrix are termed as the participation factors (PF) of the system.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 603 9/12/2017 4:08:44 PM

 604  |  Power System Analysis

2. wjk and ukj , when taken separately, are unit-dependent. However, Pjk s are dimensionless, i.e.,
independent of the units used for the state variable. This provides a straightforward measure of
relative participation of states in a mode.
3. The sum of the values of all the entries of the jth row or column of P is always equal to 1.0, i.e.,
n n
∑ Pjk = 1.0 and ∑Pjk = 1.0 (13.126)
k =1 j =1
4. Even if Pjk is a complex number, the condition given by (13.126) is satisfied. However, the
relative participation is measured by computing the absolute value of Pjk.
5. The participation factor, Pjk represents the sensitivity of the jth eigenvalue to the variations in
the kth diagonal element, (akk) of the A matrix. For example, a positive real participation factor
denotes that an introduction of a damping coefficient usually shifts l to the left.
6. A large Pjk indicates that the jth eigenvalue is very sensitive to a local feedback around the kth
state variable.
7. The participation factor (or residue)-based analysis is valid only if the eigenvalues are distinct. If
the eigenvalues are nearly identical, the mode shapes given by the right-eigenvectors are physi-
cally meaningless and the participation factors do not give the correct sensitivity information. It
is observed in [4] that a situation of eigenvalues with degree of multiplicity greater than 1 rarely
arises in power systems. Even in such cases, frequency response or linear response calculated
using these eigenvalues/eigenvectors is correct.

13.5.1 Determination of Nature of Oscillatory Modes

To identify the nature of an oscillator mode, the following procedure is employed:
1. Compute the normalised slip participation factors (SPF) for the chosen mode.
2. Identify an actively participating generator if the amplitude of its SPF is greater than a certain
value. Thus, a set of candidate generators is formed.
3. If the sum of the amplitudes of the normalised participation factors pertaining to slip of the candi-
date generators is very low, the mode is declared a non-swing mode with very low slip participation.
4. For the mode, if the slip participation is relatively high, the feasibility of formation of two coher-
ent groups of generators among the generators in the set formed in item-2, is checked. The coher-
ency of generators is verified by using the phase angle of the right eigenvector associated with slip
[18]. If all modes are purely imaginary except the zero-eigenvalue(s), the phase angle difference
between the coherent group is 180°. Otherwise, this angle difference may be less than 180°.
5. If the set formed in item-2 can be divided into two groups of coherent generators, then the mode
is declared as a swing mode; otherwise, it is a non-swing mode.
6. If the mode has been classified as either non-swing mode or non-swing mode with very low slip
participation, such a mode’s association with the state variables is declared using the highest
magnitude of state participation factor.

The following difficulties may be encountered when identifying the nature of a mode:
1. Depending on the level of the damping factor, the angle of the slip-right-eigenvector pertaining to
a swing-mode may largely deviate from 180°.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 604 9/12/2017 4:08:44 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  605 

2. The association of state variables to a mode is not straightforward when the operating point is
small-signal unstable and the mode has a high negative damping factor. In such cases, time–
domain simulation-based mode verification from the plots is also difficult.

Using the above procedure, the oscillatory modes are characterised. For each mode, state variables
which have a normalised participation factor amplitude greater than 0.1 are listed. For swing modes,
the formation of coherent groups of generators is displayed by plotting the corresponding right eigen-
vectors associated with slip.

13.6 MODAL ANALYSIS OF A TWO-MACHINE POWER SYSTEM

For a 60 Hz, two-machine four-bus power system as shown in Figure 13.18, consider the following
cases to demonstrate the eigenvalue analysis:
1. Base case where no turbines are enabled.
2. A reheat-type turbine is enabled on generator 1 alone.

Pg0 = 1.0 pu
V1 =1.0 0 V3 =1.0

M/c−3

j0.35 j0.35
PL3 = 1.7 pu
M/c−1 1 2 3
j0.01

4 PL4 = 0.1 pu

Figure 13.18  Two-machine Power System.

Here, loads are modelled as constant impedance type by setting p1 = 0, p2 = 0, and p3 = 1 in mk

of (13.100) and r1 = 0, r2 = 0, and r3 = 1 in nk of (13.101). The total load on the system is 1.8 pu.
Generator-3 is delivering 1.0 pu power and the remaining 0.8 per-unit is delivered by generator-1
through the transmission line. Both generators are modelled with 2.2 model employing the IEEE
FBM data with H = 2.8941 MJ/MVA. Here, the mechanical damping is set to zero. The generators are
provided with static exciters having KA = 200 TA1/2 = 0.025/0.02 s, and Efd max/min = ±6 pu.

13.6.1 Base Case-Without Turbine

Here, eigenvalue evaluations are carried out for two cases:
• Unreduced system matrix
• Reduced system matrix

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 605 9/12/2017 4:08:45 PM

 606  |  Power System Analysis

13.6.1.1 Eigenvalues with Unreduced System Matrix

At the operating point, the eigenvalues of the final state matrix are obtained as shown in Table 13.1.
Since there are seven states (including the exciter) for each generator, there are 14 state variables for
the system.

Table 13.1  Eigenvalue Listing for Unreduced State Matrix, Two-machine System.

Sl no. Eigenvalue States

1,2 −18.619 ± j 29.829 Fld-3
3,4 −18.348 ± j 24.043 Fld-1
5 −41.024 DampH-3
6 −38.861 DampH-1
7,8 −1.0318 ± j 10.319 Slip of m/c-1 and m/c-3
9 −3.286 DampG-3
10 −2.653 DampG-1
11 −20.48 DampK-3
12 −20.206 DampK-1
13,14 −0.00020536 ± j 0.14251 2-Zero mode

As noted earlier, the spring-mass linear system has three masses, M1, M2, and M3. Therefore, the
spring-mass system possesses two oscillatory modes—one low frequency mode which is observable
in all three masses, and another high frequency mode which is dominantly seen only in masses of
smaller weight, M2 and M3. It is to be noted that in these oscillatory modes, the masses exhibit relative
movements or swings, i.e., if one mass shows an increase in amplitude, the other depicts a decreas-
ing trend. Thus, such modes are generally referred to as swing-modes. In addition, the spring-mass
system shows a common mode (represented by two zeros) in which all masses move together without
any relative swings. In the power system example, the generators are analogous to masses. The rotor-­
angle is identical to displacement, whereas the slip is analogous to velocity. Therefore, if there are
ng generators, there will be (ng – 1) swing-modes and two zero eigenvalues if there is no damping as
in the spring-mass system. This represents redundancy of state variables associated with rotor-angle
and rotor-speed. Due to errors in the load flow (mismatch in power) and other numerical errors in
the computations, the two eigenvalues which should have been zero are calculated as a complex pair
–0.00020536 ± j 0.14251.
The above problem with zero eigenvalues evaluation can be avoided if the state variables are rede-
fined to remove redundancy with the rotor-angle. This leads to evaluation of only one zero-eigenvalue
which is associated with the slip variable making the other zero-eigenvalue to vanish in the evalua-
tion. For example, in an ng-machine system, d 1 , d 2 , , d ng are the rotor-angles. The rotor variables

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 606 9/12/2017 4:08:45 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  607 

(d 2 − d 1 ),(d 3 − d 1 ) ,(d ng − d 1 ) are the new rotor-angle variables. In terms of these new variables,
the linearised differential equations now appear as

( ∆d2 − ∆d1 ) = −w B ∆Sm1 + 0 + w B ∆Sm2 +  + 0

    

    

( ∆d ng )
− ∆d1 = −w B ∆Sm1 + 0  + w B ∆Sng +  + 0

Thus, this process of matrix reduction removes one row and one column in the original state matrix and
hence removes the redundancy associated with the rotor-angle. However, the redundancy associated
with the rotor-speed (or slip) continues to exist, which is indicated by one zero-­eigenvalue. This
zero-eigenvalue also vanishes in the eigenvalue evaluation when the following cases are set:

1. Mechanical damping is accounted on any machine.

2. Speed-governor model is considered.
3. Frequency-dependent load models are included.

Note: The time–domain simulations are generally carried out using the original unreduced variables
which preserves the zero-mode effects.

13.6.1.2  Eigenvalues with Reduced System Matrix

Using the reduced state matrix, the eigenvalues are obtained as shown in Table 13.2. Note that now
effectively there are only 13 state variables for the system.

Table 13.2  Eigenvalue Listing for Reduced State Matrix, Two-machine System.

Sl no. Eigenvalue DF Freq. (Hz) States

1,2 −18.619 ± j 29.8286 0.5295 4.7474 Fld-3

3,4 −18.3475 ± j 24.0433 0.6066 3.8266 Fld-1

5 −41.0237 1 0 DampH-3

6 −38.8606 1 0 DampH-1

7,8 −1.0320 ± j 10.3192 0.0995 1.6424 Slip of m/c-1 and m/c-3

9 −3.2860 1 0 DampG-3

10 −2.6531 1 0 DampG-1

(Continued)

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 607 9/12/2017 4:08:49 PM

 608  |  Power System Analysis

Table 13.2 (Continued)

Sl no. Eigenvalue DF Freq. (Hz) States

11 −20.4799 1 0 DampK-3

12 −20.2055 1 0 DampK-1

13 0.000 – – 1-Zero mode

From Table 13.2, it is clear that one zero-eigenvalue has emerged out indicating that there is no exclu-
sive damping in the system. As shown above, the slip participations of machines 1 and 3 are dominant
in oscillatory eigenvalues −7 and 8. Further, when the slip-right eigenvectors pertaining to machines 1
and 3 are plotted (see Figure 13.19), for this mode, these machines swing against one another in this
oscillatory mode. Due to this relative motions of machines, this mode is referred to as swing-mode.
The modes represented by (1, 2) and (3, 4) are generally attributed as exciter-modes since either Fld or
Efd state has the highest participation in them. These modes are not easily observable in time–domain
plots of states as they possess high damping.

90
0.0015
120 60

0.001

150 30

0.0005

SG−3

180 0

SG−1

210 330

240 300

270

Figure 13.19  Plot of Slip-right Eigenvectors for Machines 1 and 3 in Swing-mode

(Two-machine System).

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 608 9/12/2017 4:08:50 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  609 

The inference made from the eigenvalue analysis is verified from a time–domain simulation
by applying a self-clearing fault at bus 4 at 0.5 s for a duration of 0.01 s. Here, no line is tripped
so that the post-fault system is identical to the pre-fault system. The plots are shown in Figure 13.20.
The following observations can be made:
1. The ð-plot has a dominant ramp-response, whereas the slip plot consists of a constant DC. This
shows that the original system exhibits two zero-eigenvalues effects. This is clearly evident in the
respective COI variables. The COI related to rotor-angle, is obtained as

d COI =
(d 1 H1 + d 3 H 3 ) (13.127)
( H1 + H 3 )
and the COI related to slip, is obtained as

SCOI =
( Sm1 H1 + Sm3 H 3 ) (13.128)
( H1 + H 3 )
2. Once the COI behaviour is removed from the actual trajectories by obtaining delta_COIi=
d i − d COI and slip_COIi = Smi −SCOI, we see that the ramp-response is removed in delta_COIi
and DC component is removed from slip_COIi. In Figure 13.20 SCOI is denoted by SysCOI.

10
δ (rad)

0
0 1 2 3 4 5 6 7 8 9 10
10
Del COI

0
0 1 2 3 4 5 6 7 8 9 10
0.5
δ wrt COI

−0.5 −3
0 x 10 1 2 3 4 5 6 7 8 9 10
4
Slip

0 −3
0 x 10 1 2 3 4 5 6 7 8 9 10
4
SysCOI

0 −3
0 x 10 1 2 3 4 5 6 7 8 9 10
2
Slip wrt COI

Mch−1
0
Mch−3
−2
0 1 2 3 4 5 6 7 8 9 10
Time (s )

Figure 13.20  Response for a Short Duration Fault at Bus 4 - No Turbine Case
(2-machine System).

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 609 9/12/2017 4:08:51 PM

 610  |  Power System Analysis

−3
x 10
1.5
Mch−1
(1.36 − 0.75) = 0.61 s
Mch−3
1

f = 1/0.61 = 1.639 Hz

0.5
Slip wrt COI (pu)

−0.5

−1

−1.5
0 1 2 3 4 5 6 7 8 9 10
Time (s )

Figure 13.21  Slip_COI Response for a Short Duration Fault at Bus 4 -No Turbine Case
(2-machine System).

For clarity, the slip_COI variables are again plotted in Figure 13.21. The figure also shows the time
period of the oscillations validating the swing-mode frequency.

13.6.2 A Reheat Turbine Enabled on Machine-1

Here, a reheat type of turbine is enabled only on machine-1. The block parameters of the reheat-type
turbine with the associated speed-governor system are given below:

T1 = 0.02 s T2 = 0 s T3 = 0.01 s s = 0.05 Pgmax = 1.1

Pgmin = 0.1 TCH = 2.0 s TRH = 10 s TCO = 0.4 s

FHP = 1.0 FIP = 0 FLP = 0

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 610 9/12/2017 4:08:52 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  611 

Table 13.3 Eigenvalue Listing for Reduced State Matrix, Two-machine System with Turbine on M1.

Sl no. Eigenvalue DF Freq (Hz) States

1 −2.5000 1 0 Turbx3-1

2 −0.1000 1 0 Turbx2-1

3 −99.9827 1 0 PG-1

4,5 −18.6188 ± j 29.8289 0.5295 4.7474 Fld-3

6 −50.0682 1 0 y1gv-1

7,8 −18.3477 ± j 24.0434 0.6066 3.8266 Fld-1

9 −38.8606 1 0 DampH-1

10 −41.0237 1 0 DampH-3

11,12 −1.0104 ± j 10.3565 0.0971 1.6483 Slip of m/c-1 and m/c-3 (swing)

13 −20.2056 1 0 DampK-1

14 −20.4801 1 0 DampK-3

15 −3.2831 1 0 DampG-3

16,17 −0.2448 ± j 0.8897 0.2653 0.1416 Slip of m/c-1, m/c-3 and Turbx1-1

18 −2.6585 1 0 DampG-1

At the operating point, the eigenvalues of the final reduced state matrix is obtained as shown in Table
13.3. There are 13 states for two generators and five states for the turbine. Effectively, there are 18 state
variables for the system.
From Table 13.3, it can be seen that in addition to the swing-mode, there is one more oscillatory mode
represented by mode (16, 17) in which slip variables have a high participation. However, when the slip-
right eigenvectors pertaining to machines 1 and 3 are plotted pertaining to this mode (see Figure 13.22),
these machines do not exhibit relative swings in this oscillatory mode. Thus, this mode is a ­common-mode.
It should be noted that one pure zero-eigenvalue which is ‘visible’ in the previous case (see Table 13.2)
has appeared as an oscillatory mode (with a negative real part) when the turbine is enabled.
The time–domain plots are shown in Figure 13.23. The following observations can be made:

1. The d-plot depicts a constant DC and the common-mode, whereas the slip-plot has both the
swing-mode and the common-mode which decays to zero.
2. Once the COI behaviour is removed from the actual ‘flows’, the remaining response is mainly due
to relative swings.
3. The frequency of oscillation of the COI variables is noted as 0.1416 Hz.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 611 9/12/2017 4:08:52 PM

 612  |  Power System Analysis

90
0.03
120 60

SG−3 0.02

150 30
SG−1
0.01

180 0

210 330

240 300

270

Figure 13.22  Plot of Slip-right Eigenvector for Machines 1 and 3 (2-machine System
with Turbine) for Mode (16, 17).

A separate plot of the slip variables in Figure 13.24 shows that the rotors exhibit both relative
oscillations and common-mode oscillations. In this case, since all eigenvalues have a positive damp-
ing factor, the operating point is small-signal stable and there is no drift in the steady-state system
frequency. However, in the previous case, though the swing-mode is stable, there is a constant drift in
the system frequency, SysCOI (see Figure 13.20).

13.7 MODAL ANALYSIS OF FOUR-MACHINE

MODIFIED POWER SYSTEM
In this section, a modified 60 Hz four-machine 10-bus power system is used (see Figure 13.25). In this
system, two of the three lines interconnecting the two areas have been assumed to be on outage. Thus,
this system has only one tie-line. To alter the network data for the modified system, in file busno.dat
the total number of lines is entered as nine (instead of 11) and removal of two lines in file nt.dat is
done by commenting out the respective row entries. The network details and load-generation profile
are assumed to be identical to those given in [1].

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 612 9/12/2017 4:08:54 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  613 

2
δ (rad)

0
0 2 4 6 8 10 12 14 16 18 20
2
fcoi = 0.1416 Hz
DelCOI

1
9.57−2.51=7.06 s
0
0 2 4 6 8 10 12 14 16 18 20
0.5
δ wrt COI

−0.5 −3
0 x 10 2 4 6 8 10 12 14 16 18 20
5
Slip

−5 −3
0 x 10 2 4 6 8 10 12 14 16 18 20
5
SysCOI

0
7.06 s
−5
−3
0 x 10 2 4 6 8 10 12 14 16 18 20
2
Slip wrt COI

Mch−1
0
Mch−3
−2
0 2 4 6 8 10 12 14 16 18 20
Time (s )

Figure 13.23  Response for a Short Duration Fault at Bus 4 - with Turbine
(Two-machine System).

The generator data is chosen to be pertaining to 2.2 model as shown in Table 13.4 and a power base
of 100 MVA is chosen.
The generators are provided with static exciters having KA= 200 TA = 0.02 s and Efd max/min =
± 6 pu. The constant impedance-type load models are employed.

13.7.1 Base Case Loading Condition

In the original system, with three tie-lines, the angle across the line is 8.87° and the total real power
flow from bus 9 to 10 is 1.9962 pu. In the modified system, when there is only one tie-line, the angle
across the line increases to 27.98° for a real power flow of 1.9874 pu. This operating condition is
assumed to be brought about by the line outages. The corresponding power flow results are listed in
Table 13.5.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 613 9/12/2017 4:08:55 PM

 614  |  Power System Analysis

−3
x 10
4

Mch−1

Relative oscillation of
frequency 1.648 Hz Mch−3

2
Slips of machines

Common oscillation of
1
frequency 0.142 Hz

−1

7.06 s

−2
0 2 4 6 8 10 12 14 16 18 20
Time (s )

Figure 13.24  Slip Response for a Short Duration Fault at Bus 4 - with Turbine
(Two-machine System).

1 5 6 9 10 8 7 3
6 2 4 8
1
1 3
7 3 5 9

Load A Load B
2 4

2 4

Figure 13.25  Four-machine 10 Bus Modified Two-area Power System.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 614 9/12/2017 4:08:56 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  615 

Table 13.4  Four-machine Power System Generator Standard Parameters.

Gen. no. xd xd xèd Td0 Tèd0 xq xq xèq Tq0 Tèq0 H D
1 0.2 0.033 0.0264 8.0 0.05 0.190 0.061 0.03 0.4 0.04 54 0
2 0.2 0.033 0.0264 8.0 0.05 0.190 0.061 0.03 0.4 0.04 54 0
3 0.2 0.033 0.0264 8.0 0.05 0.190 0.061 0.03 0.4 0.04 63 0
4 0.2 0.033 0.0264 8.0 0.05 0.190 0.061 0.03 0.4 0.04 63 0

Table 13.5  Base Case Power Flow Results-Modified Four-machine Power System.
Bus no. V q (deg) Pg0 Qg0 PL0 QL0
1 1.0300 27.016 7.0 1.5415 - -
2 1.0100 17.229 7.0 2.1980 - -
3 1.0300 0 7.31106 1.7134 - -
4 1.0100 –10.494 7.0 2.5559 - -
5 1.0084 22.463 - - - -
6 0.9803 12.490 - - - -
7 1.0064 –4.759 - - - -
8 0.9760 –15.234 - - - -
9 0.9612 4.174 - - 11.59 2.12
10 0.9536 –23.814 - - 15.75 2.88

At the new operating point, which is now considered as the base case, the eigenvalues of the final
reduced state matrix are obtained. There are totally 27 eigenvalues for the system. In Table 13.6 only
oscillatory modes are listed with their nature. Note that there are 3 swing-modes since there are
4 generators.

Table 13.6  Oscillatory Modes-Modified 4 Machine System (Base Case).

Sl no. Eigenvalue Nature of modes
1 −16.2759 ± j 16.9963 Exciter mode
2 −16.4400 ± j 15.0700 Exciter mode
3 −1.0909 ± j 7.3777 Swing-mode (m/cs. 1 and 2)
4 −1.0148 ± j 6.7475 Swing-mode (m/cs. 3 and 4)
5 0.0207 ± j 2.8918 Swing-mode (m/cs. (1,2) and (3,4))
6 −15.6662 ± j 1.3507 Exciter mode
7 −15.6941 ± j 1.7584 Exciter mode

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 615 9/12/2017 4:08:56 PM

 616  |  Power System Analysis

Further, from the tabulated values, it can be seen that a swing-mode in which machines 1 and 2 together
swing against machine group 3 and 4 is unstable, as its damping factor is negative (–0.007169). Its fre-
2.8918
quency is = 0.4602 Hz. The other two swing-modes are stable. This denotes that the operating
2p
point chosen is small-signal unstable. This performance is validated by applying a very short duration
(0.05 s) three-phase fault at bus 9 at t= 0.5 s, without any line clearing so that the post-fault system is
identical to the pre-fault. The resulting plots are shown in Figure 13.26.
From Figure 13.26, the following observations can be made:
1. The slip_COI and delta_COI variables bring out the unstable swing-mode in which all machines
participate. In this mode machines 1 and 2 together swing against the other group consisting of
machines 3 and 4. This mode is also referred to as inter-area mode.
2. Since a short duration disturbance is applied at bus 9, away from the generators, (i.e., in the net-
work), the other swing-modes which are local to each machine group (1 and 2) or (3 and 4) are not
excited sufficiently to be seen in slip-signals. Further, these modes have good enough damping
due to which they are not sustained for longer time.
3. Even the COI frequency does not stabilise at a new constant value.

1
0.5
δ wrt COI

2
0
3

−0.5 4
0 2 4 6 8 10 12 14 16 18 20
−3
x 10
8

6
SysCOI

0
0 2 4 6 8 10 12 14 16 18 20
−3
x 10
2
1
Slip wrt COI

1 2
0
−1 4

−2 3

0 2 4 6 8 10 12 14 16 18 20
Time (s )

Figure 13.26  Response of Slip_COI, Delta_COI and System Frequency (SysCOI)

for Base Case (4 Machine).

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 616 9/12/2017 4:08:57 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  617 

13.7.2 Base Case without Excitation Controllers

In this case, excitation controllers are disabled. Here, the only oscillatory modes are the swing-modes.
These are listed in Table 13.7. In addition, a pure real-eigenvalue which is unstable is also indicated
in the table.

Table 13.7  Oscillatory Modes-Modified 4 Machine System-without AVR in the Base Case.

Sl no. Eigenvalue Nature of modes

1 −0.7114 ± j 7.2373 Swing-mode (m/cs. 1 and 2)

2 −0.6600 ± j 6.6733 Swing-mode (m/cs. 3 and 4)

3 −0.1562 ± j 2.5449 Swing-mode (m/cs. (1,2) and (3,4))

4 0.0862 Slips of all m/cs

From the tabulated eigenvalues, the following observations are made:

1. The swing-mode (Sl No. 3) in which all machines take part in relative swings has become stable
(damping factor = 0.061264) when the exciters are disabled. This demonstrates the effect of a
fast-acting and high-gain static exciter.
2. Compared to the previous case where AVRs are enabled, the frequencies of the swing-modes have
2.5449
reduced. For example, the stabilised swing-mode now has a frequency =0.405 Hz. This
generally depicts a reduction in the synchronising torque component. 2p
3. In the unstable pure real mode, it is noted that the slips of all machines have the highest participa-
tion. This implies that although the relative swings are stable, due to lack of synchronising effects
across the tie-line, the system may be eventually unstable.
A time–domain simulation is carried out for a three-phase fault at bus 9 for a small duration as in the
previous case. The resulting plots are shown in Figure 13.27. From the plot, it can be seen that the
mode which exhibited growing oscillations in the previous case is now stable. However, the groups of
machines across the tie-line eventually separate from one another due to reduction in synchronising
torque. This causes the system to lose synchronism. This behaviour is also depicted by the common-­
mode frequency.

13.7.3 Base Case with Reduced Loading Condition

In this case, the load profile and the specified generation are reduced to 70% of its base case values. The
rest of the system details are identical to the base case. Thus, the new real power generation is 4.9 pu
instead of 7 pu for all generators. The new P/Q (old P/Q) loads are: 8.113/1.484 pu (11.59/2.12 pu)
at bus 9, and 11.025/2.016 pu (15.75/2.88 pu) at bus 10. These details are altered in pvpq.dat file
and the power flow programme is rerun. Thus, a new operating point is established. In this case the
angle across the tie-line is 18.45° for a real power flow of 1.49 pu. For running the eigenvalue and
time-domain simulation programmes, an additional file ld.dat has to be modified. In Table 13.8 only
oscillatory modes are listed with their nature.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 617 9/12/2017 4:08:58 PM

 618  |  Power System Analysis

2 (1, 2)
δ wrt COI

−2 (3, 4)

0 2 4 6 8 10 12 14 16 18 20

0.1
SysCOI

0.05

0
0 2 4 6 8 10 12 14 16 18 20
−3
x 10

5
1, 2
Slip wrt COI

−5 3, 4

0 2 4 6 8 10 12 14 16 18 20
Time (s )

Figure 13.27  Response of Slip_COI, delta_COI and System Frequency without AVR
in the Base Case (4-machine).

Table 13.8  Oscillatory Modes-Modified 4 Machine System (Reduced Load Condition).

Sl no. Eigenvalue Nature of modes

1 −16.4714 ± j 17.5902 Exciter mode

2 −16.7166 ± j 15.3698 Exciter mode

3 −1.6751 ± j 7.0265 Swing-mode (m/cs. 1 and 2)

4 −1.6293 ± j 6.3570 Swing-mode (m/cs. 3 and 4)

5 −0.1109 ± j 3.1151 Swing-mode (m/cs. (1,2) and (3,4))

6 −15.6273 ± j 1.0568 Exciter mode

7 −15.6765 ± j 1.9863 Exciter mode

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 618 9/12/2017 4:08:59 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  619 

From Table 13.8, it can be seen that the swing-mode (SL No. 5) which is unstable in the base-case
(with AVR) has become stable as its damping factor is now positive (0.035565) with a frequency equal
3.1151
to =0.496 Hz. Thus, the system has become stable by reducing the load level. This is generally
2p
referred to as preventive control. However, another way of stabilising the operating point in the base
case is by designing a power system stabiliser (PSS) as outlined in [14].
For a three-phase fault at bus 9, the slip_COI plots are obtained as shown in Figure 13.28. From
the plot, it can be verified that the perturbation excites the inter-area mode of frequency 0.497 Hz
dominantly. Other two swing-modes local to one side of the tie-line which involve relative oscil-
lations between machines 1 and 2 or machines 3 and 4, are not triggered. This is equivalent to the
application of force F1 in the earlier spring-mass system. Thus, for such a network disturbance,
inter-area modes are excited and the frequency component is seen in almost all variables. As
shown in Figure 13.28, the system frequency attains a constant DC shift as there is no exclusive
zero-mode damping.

1
0.2
δ wrt COI

0
3

−0.2 4

0 2 4 6 8 10 12 14 16 18 20
−4
x 10
15

10
SysCOI

−5
0 2 4 6 8 10 12 14 16 18 20
−3
x 10
1
(4.47−2.46) = 2.01 s
Slip wrt COI

0.5

−0.5
f = 1/2.01 = 0.497 Hz
−1
0 2 4 6 8 10 12 14 16 18 20
Time (s )

Figure 13.28  Response of Slip_COI, Delta_COI and System Frequency for the Reduced
Load Case (4-machine).

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 619 9/12/2017 4:09:02 PM

 620  |  Power System Analysis

13.7.3.1 Reference Voltage Perturbation of Exciter

to Trigger a Local Swing-mode
In order to excite a local swing-mode related to machines 1 and 2, the reference voltage, VRef,
setting of machine 1 exciter, is increased by a very small amount. This triggers the relative motion
in which machine 1 swings against machine 2 for a short period (see Figure 13.29). In later cycles,
the inter-area swing-mode dominates as the damping factor of the local mode is quite large com-
pared to that of the inter-area mode. Also note that a slight increase in VRef of machine 1 changes
its relative rotor-angle. However, this change is so small that it does not alter the operating-point
nature significantly.

0.2
1

2
0.1
δ wrt COI

−0.1

−0.2
0 1 2 3 4 5 6 7 8 9 10

−4
x 10
2
1.28−0.32 = 0.96 s
1 2
Slip wrt COI

−1

f = 1/0.96 = 1.042 Hz
−2 1

0 1 2 3 4 5 6 7 8 9 10
Time (s )

Figure 13.29  Response of Slip_COI and Delta_COI for the Reduced Load Case When Vref-1
is Perturbed (4-machine).

Thus, for a disturbance, all machines in a system (independent of their location) respond by showing
relative oscillations as decided by their inertias.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 620 9/12/2017 4:09:03 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  621 

13.7.4 Base Case with Power System Stabiliser

In this case, instead of reducing the load level to stabilise the system, a power system stabiliser
is employed. As per IEEE standards 421.5 - 1992 [19], the following are the two main categories
of PSS:

1. Single-input power system stabiliser: It is known that in order to modify a mode of oscillation by
feedback, the chosen input must excite the mode and the mode must be visible in the signal [4].
Thus, for this kind of PSS design, the commonly used input signals are shaft speed, terminal bus
frequency, and electrical power output.

In Figure 13.30, a typical structure of a slip-signal PSS is shown. In addition to other blocks, it has
a torsional filter to prevent adverse interaction of the PSS with torsional frequencies of the steam-­
turbine mechanical system [20] in the presence of series capacitor compensated lines.

VS
1 sTW 1 + sT1 a0
KS
1 + sTR 1 + sTW 1 + sT2 s2 + a1 s + a0
p.u.
Slip
Measurement Washout Compensator Gc(s) Filter Gain
Limiter
Delay circuit

Figure 13.30  Block Schematic of a Slip-signal PSS.

An electrical power-input PSS is shown in Figure 13.31. It is seen that power-input PSS provides
a high degree of attenuation to torsional modes unlike the slip-signal [21]. Note that an electrical
­power-input based PSS can be realised by observing the following relationship from swing equation
by neglecting the deviation in the mechanical power input and mechanical damping:

1
jw ∆Sm ( jw ) = − ∆Te ( jw ) (13.129)
2H

1 sT
∆T W K
e 1+sT 1+sT S V
R W S

Gain
Measurement Washout Limiter
Delay circuit

Figure 13.31  Block Schematic of a Power-signal PSS.

If the Vs-signal (see Figure 13.31) is fed to the exciter VRef - summing junction with a negative polar-
ity, then ∆Vs(jw) is equivalent to −∆Te(jw). Therefore, the above expression can be rewritten as

1
jw ∆Sm ( jw ) = ∆Vs ( jw ) (13.130)
2 HK s

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 621 9/12/2017 4:09:05 PM

 622  |  Power System Analysis

From the above expression, we can see that ∆Vs(jw) leads the slip-signal by 90°. This implies that
using electrical power signal is equivalent to using slip signal with 90° phase lead. In other words, to
get the phase angle of the compensated GEP, it is required to simply add 90° to the angle of GEP(jw).
2. Dual-input power system stabiliser: In this kind of PSS design, a combination of signals such as
speed and electrical power output are used. The objective of this PSS is to synthesise an equiva-
lent slip signal ∆Smeq so that it does not contain torsional modes [2, 14].

13.7.4.1  Slip-signal PSS Design

The steps involved in designing a PSS are as follows:
1. Location of PSS: Generally, it is not desirable to enable PSS block on all generators as it may lead
to adverse interactions and only a few generators have a dominant effect on the mode control.
Therefore, to decide the location of PSS, those machine(s) whose slip-participation is the highest
in a critical mode, is (are) chosen as the candidate generator(s). If other generators are chosen for
PSS placement, it may not be effective and would require higher PSS gain for the same level of
damping. For the base case, generator 3 has the highest slip participation in the unstable swing-
mode. Therefore, machine 3 is chosen to place the PSS.
2. Computation of GEP(jw): The GEP(jw) for machine 3 is obtained as follows:

∆Te ( jw )
GEP ( jw ) =
∆Vs ( jw )
= DT′ [ jwI − AT′ ]−1 [ EG′ ] (13.131)

where, [A′T] and [E′G] are the modified versions of [AT] and [EG] after making Dd = 0 and DSm = 0
for all machines. Further, [E′G] contains the third column of EG pertaining to generator 3.
  The matrix D′T relates DTe to DXT as

DTe = D′T DXT (13.132)

where DXT is the modified version of DXG after removing all Dd and DSm.
3. Design of compensator GC(S): Here, following the usual procedure, the compensator param-
eters are chosen as fm = 5° and fm= 3 Hz, and we get T1 = 0.05790 s and T2 = 0.04861 s with
T1
= 1.1910 .
T2
In Figure 13.32, the phase response of the PSS which is the combined phase response of GC(s)
and a washout circuit with TW = 10 s is depicted. The compensated phase response is also plot-
ted in the figure. The compensated phase angle is less than 20° at inter-area mode of 0.5 Hz.
4. Determination of the compensator gain: In order to decide the PSS gain, first, the PSS state-model
is interfaced to the rest of the system. By a repeated run of the eigenvalue analysis for different PSS
gains, a gain value is chosen which provides a damping factor of more than 0.05 for the critical
swing-mode. A procedure to interface the PSS model to the main state model is detailed below:
The linearised model of the slip-input PSS is given by

∆ x PSS = APSS ∆xPSS + BPSS ∆Sm (13.133)

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 622 9/12/2017 4:09:06 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  623 

80
Compensated GEP
Uncompen. GEP
60 Angle of PSS

40

20

0
Angle (deg)

−20

−40

−60

−80

−100
0 0.5 1 1.5 2 2.5 3 3.5 4
Frequency (Hz)

Figure 13.32  Phase Angle of GEP(jw), PSS(jw) and Compens. GEP(jw) for Machine 3.

∆Vs = C PSS ∆xPSS + DPSS ∆Sm (13.134)

where ∆Sm denotes the deviation in slip for the third generator.
Rewriting (13.119) considering only the change in VS, we have,

∆ X G = [ AT ] ∆X G + [ EG ] ∆Vs (13.135)

where [ EG ] = EG (:,3) , for 3rd generator.

Writing ∆Sm in terms of DXG, we have

∆Sm = e2T ∆X G (13.136)

where e2 = [ 0 1 0  0  0 ](1×16n ) , with 1 corresponding to slip state variable of third machine.
T

g

Now, using (13.136) in (13.133) and (13.134) we get,

∆ x PSS = APSS ∆xPSS + BPSS e2T ∆X G (13.137)

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 623 9/12/2017 4:09:11 PM

 624  |  Power System Analysis

T
∆Vs = C PSS ∆xPSS + DPSS e2 ∆X G (13.138)
Using (13.138) in (13.135) and rewriting the state model accounting PSS, we obtain,
∆ X = AN ∆X N
N
T
where ∆X N =  ∆X G
T T 
∆xPSS and


[ EG ] C PSS 
[ AT ] + [ EG ] DPSS e2T
AN =  
 BPSS e2T APSS 
The eigenvalues of AN are determined to analyse the small-signal stability of the system with PSS.

13.7.4.2 Eigenvalues with Slip-input PSS

For the PSS designed in the previous section, for machine-3, all oscillatory modes are listed in Table
13.9, for KS = 17. Note that in this implementation the torsional filter and the measurement delay
circuit are not considered.

Table 13.9  Oscillatory Modes for the Base Case With PSS on m/c-3.
Sl no. Eigenvalues Damping factor Freq. (Hz) Nature of the mode
1 −15.9429 ± j 16.8017 0.6883 2.67408 Exciter mode
2 −15.9766 ± j 15.0315 0.7283 2.3923 Exciter mode
3 −1.0909 ± j 7.3777 0.1463 1.1742 Swing mode (1 & 2)
4 −2.2002 ± j 7.9338 0.2672 1.2627 Swing mode (3 & 4)
5 −0.1554 ± j 2.8824 0.0538 0.4587 Swing mode ([1 2]&[3 4])
6 −15.6793 ± j 1.6468 0.9945 0.2621 Exciter mode

From Table 13.9, the following observations are made:

1. The inter-area mode which is unstable in the base case has become stable with the PSS for the
same loading condition.
2. The gain of the PSS is chosen so that the damping factor for the critical swing-mode is greater
than 0.05. There is a slight reduction in the frequency of the mode compared to the base case
(0.4602 Hz). This shows that the PSS has introduced a damping torque at the expense of the inher-
ent synchronising torque. This can be overcome by redesigning the compensator with a slightly
lower phase lead.
3. Since the PSS has been located on generator 3, the synchronising and damping components per-
taining to its local swing-mode have increased from the base case. In the base case the damping
factor is 0.1487 and frequency is 1.0739 Hz, whereas with PSS they have become, 0.2672 and
1.2627 Hz, respectively.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 624 9/12/2017 4:09:13 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  625 

13.7.4.3 Time-domain Verification of PSS Performance

A time–domain simulation is carried out by applying a three-phase fault at t = 0.5 s, at bus 9 with fault
clearing time of 0.05 s without any line tripping. In order to restrict the level of generator terminal
voltage fluctuations during transient conditions and to prevent the PSS action to counter-act the AVR,
limits are imposed on the PSS output. The positive output limit of the stabiliser is set in the range of
0.1 to 0.2 pu, whereas the negative limit is set on the lower side (–0.02 to –0.05 pu), to prevent the PSS
from reducing the generator’s terminal voltage excessively following a fault [2].
The rotor-angles and slips of the machines relative to COI-variables, are plotted in Figure 13.33.

1
2
0.5
δ wrt COI

3
0 4

−0.5
0 2 4 6 8 10 12 14 16 18 20
−3
x 10
3

2.5

2
SysCOI

1.5

0.5

0
0 2 4 6 8 10 12 14 16 18 20
−3
x 10
2

3&4 4.93 − 2.71 = 2.22 s

1
C
Slip wrt COI

A
B
0

−1
1&2
−2
0 2 4 6 8 10 12 14 16 18 20
Time (s )

Figure 13.33  Rotor-angle and SlipsCOIs of Machines in Base Case with PSS
on Machine 3.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 625 9/12/2017 4:09:16 PM

 626  |  Power System Analysis

The following observations are made.

• The rotor-angle oscillations damp out from cycle-to-cycle denoting that the operating point is
now stable with the introduction of the PSS.
• The system frequency, denoted by SysCOI, though decreases, it shows a constant drift under
steady-state.
• The slip_COI variables dominantly show the decaying inter-area swing-mode. The frequency
component can be obtained as follows:
At point A, tA = 2.71 s, and at point B, tB = 4.93 s. Therefore, the period of the waveform is

T = (t B − t A ) = 4.93 − 2.71 = 2.22s

1
Therefore, the frequency is given by f sw = = 0.45045 Hz.
T
• From the slip_COIi variable, the decrement factor (i.e., the real part of the eigenvalue) for the
inter-area mode can be obtained as follows:
At tA = 2.71 s, the magnitude of the slip is noted as yA = 1.162 × 10−3. At tB = 4.93 s, the magni-
tude of the slip is again noted as yB = 0.8184 × 10−3. Now, assuming that the variation of the signal
fits the following function, we have,

at
y = y Ae cos(2pf sw t ) for t ≥ t A

At the end of one cycle, starting from tA, we have

y B = y A ea T cos( 2p f swT )

Substituting the value of yA and yB and solving for a, we obtain, a = −0.1579, which is very close
that given in Table 13.9 for Sl No. 5.
In order to understand the behaviour of the PSS, the variables such as slip of machine 3 (which is used
as the input signal to the PSS), the output of the PSS, the field voltage applied to the generator and the
magnitude of the generator terminal voltage are plotted in Figure 13.34.
From Figure 13.34, it can be seen that the Vs-signal remains within the limit of ±0.1 pu and it
reduces to zero once the steady-state is reached though the input slip-signal has some DC offset,
due to the presence of the washout-circuit. Since the system is stable, all signal damp out as time
progresses.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 626 9/12/2017 4:09:18 PM

 Modal Analysis of Power Systems with Interconnected Generators  |  627 

−3
x 10
3
Slip m/c.3

0
0 5 10 15 20 25 30

0.05
Vs o/p

−0.05
0 5 10 15 20 25 30

4
Efd−3

0
0 5 10 15 20 25 30

1.1

1.05
Vbus−3

0.95
0 5 10 15 20 25 30
Time (s )

Figure 13.34  Slip, Vs, Efd, and Terminal Voltage for Machine 3 in Base Case with PSS.

References
[1]  K. R. Padiyar, Power System Dynamics - Stability and Control, BS Publications, Hyderabad,
India, 2002.
[2]  P. Kundur, Power System Stability and Control, McGraw-Hill Inc., NY, 1994.
[3]  Graham Rogers, ‘Demystifying Power System Oscillations’, IEEE Computer Application in
Power, 1996.
[4]  Graham Rogers, Power System Oscillations, Kluwer Academic Publishers, London, 2000.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 627 9/12/2017 4:09:22 PM

 628  |  Power System Analysis

[5]  Nelson Martins, ‘Efficient Eigenvalue and Frequency Response Methods Applied to Power
System Small-Signal Stability Studies,’ IEEE Transactions on Power Systems, vol. PWRS-1 (1),
pp. 217–224, February 1986.
[6]  M. Klein, G. J. Rogers, S. Moorthy, and P. Kundur, ‘Analytical Investigation of Factors
Influencing Power System Stabilizers Performance,’ IEEE Trans. on Energy Conversion, vol. 7(3),
pp. 382–390, September 1992.
[7]  P. W. Sauer and M. A. Pai, Power System Dynamics and Stability, Prentice Hall, Upper Saddle
River, New Jersey, 1998.
[8]  P. G. Murthy and M. Pavella, Transient Stability of Power Systems, Theory and Practice, John
Wiley and Sons Ltd., England, 1994.
[9]  P. Kundur, G. R. Rogers, D. Y. Wong, L. Wang, and M. G. Lauby, ‘A Comprehensive Computer
Program Package for Small-Signal Stability Analysis of Power System,’ IEEE Trans. on Power
Systems, vol. 5(4), November 1990.
[10]  P. Kundur, D. C. Lee and H. M. Zein El-Din, ‘Power System Stabilizers for Thermal Units:
Analytical Techniques and On-site Validation,’ IEEE Trans. on Power Apparatus and Systems,
vol. PAS-100, pp. 81–95, January 1981.
[11]  H. Breulmann, E. Grebe, W. Winter, R. Witzmann, P. Dupuis, M. P. Houry, T. Margotin, J. Zerenyi,
J. Dudzik, PSE S.A., J. Machowski, L. Martín, J. M. Rodríguez, and E. Urretavizcaya,
‘Analysis and Damping of Inter-Area Oscillations in the UCTE/CENTREL Power System,’
CIGRE, Germany, 2000.
[12]  J . G. Stoolweg, J. Person, A. M. van Voorden, G. C. Paap, W. L. Kling, ‘A Study of the Eigenvalue
Analysis Capabilities of Power System Dynamics Simulation Software,’ 14th PSCC, Sevilla,
24–28 June 2002.
[13]  U sing Simulink, Version 3, Release 11, The Math Works Inc.
[14]  K . N. Shubhanga and Yadi Anantholla, ‘Manual for A Multi-machine Small-signal Stability
Programme,’ EE Dept., NITK, Surathkal, India, 2008.
[15]  A. R. Bergen and V. Vittal, Power System Analysis, Pearson Education Asia, India, 2001.
[16]  F . L. Pagola, I. J. Perez-Arriaga, and G. C. Verghese, ‘On Sensitivities, Residues and Participations:
Applications to Oscillatory Stability Analysis and Control’, IEEE Trans. on power systems,
vol. PWRS-4 (1), pp. 278–285, February 1989.
[17]  G . C. Verghese, I. J. Perez-Arriage, and F. C. Schweppe, ‘Selective Model Analysis with
Application to Electric Power Systems, Part I: Heuristic Introduction, Part II: The Dynamic
Stability Problem,’ IEEE Trans. on power systems, vol. PAS-101(9), pp. 3117–3134, September
1982.
[18]  M . Klein, G. J. Rogers, and P. Kundur, ‘A Fundamental Study of Inter-area Oscillations in Power
System,’ IEEE Transaction on Power Systems, vol. 6(3), pp. 914–921, August 1991.
[19]  I EEE Recommended Practice for Excitation Systems Model for Power System Stability Studies,
IEEE Standard 421.5-1992.
[20]  S hashidhara M. Kotian and K. N. Shubhanga, ‘Performance of synchronous machine models in
a series-capacitor compensated system,’ The EEE Transaction on Power Systems, vol. 29 (3),
pp. 1023–1032, May 2014.
[21]  E . V. Larsen and D. A. Swann, ‘Applying Powers System Stabilizers, Part I; General Concepts,
Part II; Performance Objectives and Tuning Concepts, Part III; Practical considerations’, IEEE
Trans on power Apparatus and Systems, vol. PAS-100 (6), pp. 3017–3046, June 1981.

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 628 9/12/2017 4:09:22 PM

 Modal Analysis of Power Systems with Interconnected Generators | 629

Review Questions
1. Which of the following tabulated set of data (see Table 13.10) on machine MVA base is/are likely
to be incorrect?
Table 13.10 Generator Data.

Generator xd xd xdè 

Tdo s è
Tdo s xq xq xqè 
Tqo s è
Tqo s Hs
(A) 2.31 0.267 0.223 9 0.06 2.19 0.7 0.23 1 0.06 3.0
(B) 2.31 0.267 0.223 9 0.06 2.19 0.7 0.223 1 0.06 0.5
(C) 2.31 0.267 0.223 1 0.06 2.19 0.7 0.223 1 0.06 3.0
(D) 2.31 0.267 0.223 9 0.06 2.41 0.7 0.223 1 0.06 0.5
(E) 2.31 0.267 0.223 9 0.06 2.19 2.19 2.19 1 0.06 3.0

2. Write down the steps to identify swing modes from a list of system eigenvalues.
3. Give a typical classification of swing modes.
4. The matrix given in eqn (13.8) is rewritten as follows:

 [0](3×3) [ I ](3×3) 
A=  
[ Ak ](3×3) [0](3×3) 

If g is the eigenvalue of [Ak], then show that the eigenvalues of A matrix are given by l = ± g .
5. What is the significance of reduced system matrix in the modal analysis of power systems?
6. Enumerate the conditions during which zero-modes ‘acquire’ damping.
7. The two-machine power system given in Figure 13.18 is modified as shown in Figure 13.35,
employing classical model for the generators. Note that the bus numbers are re-named so that
generator internal nodes are designated first. Loads are modelled as constant impedance type and
are augmented to the bus admittance matrix. Use the load flow results given in Table 14.1.

Pg0 = 1.0 pu
M/c−1 V1 = 1.0 0 V 3 = 1.0
x′d
x′d

2 E′2
j0.35 j0.35
3 PL3 = 1.7 pu
E′1 1 5 6
j0.01 M/c−3

4 PL4 = 0.1pu

Figure 13.35 Two-machine Power System with Machines’ Internal Nodes.

Construct the bus admittance matrix accounting x′d of the machines. Show that the matrix can be
written including the internal nodes, as

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 629 9/12/2017 4:09:27 PM

 630 | Power System Analysis

[YGG ]( 2 × 2) [YGN ]( 2 × 4) 
YBus =  
[YNG ]( 4 × 2) [YNN ]( 4 × 4) 

Now, reduce the matrix to the internal nodes to obtain
[YGG Red ]( 2 × 2) = [YGG ]( 2 × 2) − [YGN ]( 2 × 4) [YNN ]−( 41× 4) [YNG ]( 4 × 2)

8. Using the above matrix, [YGG Red ]( 2 × 2) = Y ij = Yij ∠a ij , we have,

 E ′1   I ′1 
[YGG Red ]  = 
 E ′ 2   I ′ 2 
where I′1 and I′2 denote the currents delivered by the generators. Obtain an expression for real
power Pei, delivered by ith machine as
ng
Pei = ∑ Ei′ E ′jYij cos(a ij − d i + d j )
j =1
where ng represents the number of machines. Note that speed deviations are neglected.
9. Perform the linearized analysis of the swing equations and construct the state matrix for the sys-
tem shown in Figure 13.35. Determine the swing modes of the system at the operating point.
10. The single-line diagram of an SMIB power system is shown in Figure 13.36. The system details
are adopted from [1].
PL = 0.6

1 xt 2 3 4
L1
xsys

L2
Vg = 1.05 Eb = 1.0∠0°

Figure 13.36 SMIB System for Small-signal Analysis.

The system details on 1000 MVA base are as follows (all are in per-unit):
Generator: xd = 1.7572, xd′ = 0.4245, Tdo ′
= 6.66 s, xq = 1.5845, xq′ = 1.0400,
′
Tqo = 0.44 s, H = 3.542 MJ/MVA, f o = 50 Hz, D = 0.
Exciter: Single-time constant static exciter, K A = 400, TA = 0.025 s,
Efdlimits = ± 6
Transformer: xt = 0.1364
Lines (Each line): RL = 0.08593, x L = 0.81250 and Bc = 0.1184 (total),
xsys = 0.13636.
Load: Constant impedance type.
Carry out the analysis using the small-signal stability programme. The load PL is assumed to be
connected at the infinite bus, which is modelled as a PV-bus with no real power generation (see
pvpq.dat file).

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 630 9/12/2017 4:09:37 PM

 Modal Analysis of Power Systems with Interconnected Generators | 631

File name: pvpq.dat

--------------------------------------------------------------------------------------------------------
Bus No. Vg/PL0 Pg0/QL0 Remarks
-------------------------------------------------------------------------- -----------------------------
4 1.0000 0.0 ---> Generator (PV) -buses other than the slack bus
4 0.6000 0.0 ---> Load data (including loads at PV-bus)
--------------------------------------------------------------------------------------------------------
The infinite bus is modelled as a generator with classical model with the generator data given as:
File name: gen.dat
--------------------------------------------------------------------------------------------------------------------
Gen. No. xd xdd xddd Td0d Td0dd xq xqd xqdd Tq0d Tq0dd H D
-------------------------------------------------------------------------------------------------------------------
1 1.7572 0.4245 0.4245 6.66 0.05 1.5845 1.0400 1.0400 0.44 0.04 3.542 0
4 0.0006 0.0001 0.0001 10000 0.05 0.0001 0.0001 0.0001 0.44 0.04 10000 0
-------------------------------------------------------------------------------------------------------------------
The list of tasks are as follows:
(a) Perform the eigenvalue analysis and determine the stability of the operating point for the pre-
fault system.
(b) A three-phase fault is applied at bus 2 at t = 1 s for a period of 0.08 s and the fault is cleared
by tripping line L1. Determine the small-signal stability of the post-fault system by observing
the time-domain plots of rotor-angle and slip.
(c) If the post-fault system is small-signal unstable, then design a simplified slip-signal PSS so
that the post-fault operating point is small-signal stable. Verify this by an eigenvalue analysis
as well as through a time-domain simulation.
11. What is the function of a wash-out circuit in PSS block?
12. Determine the response of a wash-out circuit (with time-constant TW) for a ramp input of unit
slope. Mention a power system condition during which the PSS block receives slip signal of
ramp-kind.
13. What are preventive control measures?

M13_POWER-SYSTEM-AN00_SE_XXXX_CH13.indd 631 9/12/2017 4:09:40 PM

 Transient Stability Analysis
of Power Systems with
Interconnected Generators 14
CHAPTER OUTLINE

14.1 Interfacing Generator to Network 14.6 A Summary of the Implementation

14.2 Centre of Inertia Reference Procedure
14.3 Structure of Power System 14.7 Demonstration of System Frequency
Equations and its Solution with Two-machine Power System
14.4 Load Modelling 14.8 Large Disturbance Performance
14.5 Load Equivalent Circuit

In this chapter, implementation details to carry out time–domain simulation of power system stabil-
ity analysis with interconnected generators are presented. It is noted that the time–domain analysis
permits detailed modelling of power system components accounting all non-linearities of the system,
including the controllers. Thus, this analysis provides an accurate way to validate the inferences and
predictions made in linearised analysis where power swings due to small-signal disturbances are stud-
ied. While carrying out large-disturbance analysis, such as transient stability analysis, time–domain
analysis becomes a major tool.
The programme implementation not only requires a thorough insight into each of the component
models but also needs a proper interfacing among components. This aspect has been covered in detail.
To illustrate large-signal analysis, results of many simulation studies conducted on two machines,
four machines, and 50 machines power systems are included.

14.1 INTERFACING GENERATOR TO NETWORK

We note that the generator is modelled in the ‘rotor or machine-frame’ of reference and all gener-
ator variables are represented with respect to the ‘machine-frame’ of reference. For interfacing the
generator to the network and performing system studies, one has to transform these variables to the
‘synchronous-frame’ of reference, which is the only common frame for all machines. The transmis-
sion network mainly consists of transmission lines and transformers. Since the time constants of these
elements are relatively small compared to the mechanical time constants, the network transients are

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 632 9/12/2017 4:56:00 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  633 

generally neglected and the network is assumed to be in a quasi-sinusoidal steady state. Hence, the
network quantities are represented in the ‘synchronous-frame’ of reference.
Although the real and reactive power balance equations can be used for network solution, the
current balance formulation is preferred for time-domain simulation since the network admittance
matrix is constant (except when there is a network change). Therefore, the network is represented by
an algebraic equation as follows:

YBUS V = I
(14.1)
where, I is the vector of injected bus currents, V is the vector of complex bus voltages, and YBUS
is the bus admittance matrix.

14.1.1  Influence of Dynamic Saliency of Generator

Neglecting stator transients and ignoring speed variations, the stator d- and q-axes voltage equations
are given by

vd = −id Ra −y q (14.2)

vq = −iq Ra +y d (14.3)

where vd and vq represent d- and q- axes generator terminal voltages, respectively. Using E ″q and E ″d,
the above equations can be rewritten as
vd = −id Ra − xq″ iq + Ed″ (14.4)

vq = −iq Ra + xd″ id + Eq″ (14.5)

The relationship between the ‘machine-frame’ and the ‘synchronous-frame’ of references is shown in
Figure 14.1.

D−axis
s
axi
d−

q−axis

δ
Q−axis

dq− axes: Machine−frame of reference

DQ−axes: Synchronous−frame of reference

Figure 14.1  Relationship Between dq- and DQ- Reference Frames.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 633 9/12/2017 4:56:06 PM

 634  |  Power System Analysis

The relationship depicted in Figure 14.1 is mathematically expressed under sinusoidal steady-state
condition of the stator and for low frequency oscillations of the rotor systems, as [5]:

( Fqi + jFdi ) = ( FQi + jFDi )e − jd i

(14.6)
where di represents the rotor-angle of the ith machine.
The above equation can be written in matrix notation as:

 Fqi  cos d i sind i   FQi   FQi 

 F  =  −sind   
cosd i   FDi 
= [T ]   (14.7)
 di   i  FDi 
In the above expression, F represents voltages or currents and the matrix [T] is a function of the rotor
angle di.
From (14.4) and (14.5), it is clear that the generator winding currents are a function of generator
terminal voltage and in turn depend on the network conditions. Writing the stator voltage equations
in matrix form, we get,

 Ra − xd″  iq   Eq″ − vq 

   =  
 xq″ Ra  id   Ed″ − vd 
(14.8)
The currents iq and id can be obtained as
iq  1 xd″   Eq″ − vq 
 Ra
i  = 2   
″ ″  − xq″
 d  Ra + xd xq Ra   Ed″ − vd 
(14.9)
Note that the above expression has to be interfaced to the network to compute the generator terminal
voltage and hence its current. To perform this, (14.9) has to be transformed to the network side. In
case of single machine connected to infinite bus system (SMIB), due to its uniqueness, the bus voltage
itself is transformed to the machine-frame.
Now writing the above equation in the ‘synchronous-frame’ of reference, from (14.7) we have,

iQ  1  Ra xd″  E″ − v 

[T ]   = 2   [T ]  Q Q 
− vD  (14.10)
i ″ ″  − xq″ ″
 D  Ra + xd xq Ra   ED

Simplifying, we get iQ and iD as,

iQ   EQ″ − vQ 
i  = y gDQ ( t )  
″
 ED
 D − vD 

 EQ″  vQ 
= y gDQ (t )   − y gDQ (t )  
″
 ED  vD 

vQ 
= I DQS (t ) − y gDQ (t )   (14.11)
vD 

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 634 9/12/2017 4:56:10 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  635 

where

1 cosd i −sind i   Ra xd″  cosd i sind i 

y gDQ (t ) = 2  
″ ″ sind cosd i   − x ″ cosd i 
Ra + xd xq  i
 q Ra   −sind i

The above equation now shows that the generator is treated as a current source in parallel with an
admittance ygDQ(t). Thus, IDQS(t) represents Q and D components of the generator source current and
the second term implies a condition where the admittance is augmented to the bus admittance matrix,
YBUS along with the network elements. Note that ygDQ(t) denotes the machine admittance which is a
function of the rotor angle di. Since di varies with time, the inclusion of ygDQ(t) in the bus admittance
matrix makes the YBUS time-varying. Therefore, while solving for iQ and iD, one has to deal with a
matrix which is not constant and requires a repeated factorisation at every time step. This increases the
computational complexity. In other words, it leads to the following difficulties:
1. The evaluation of iQ and iD requires the solution of a time-varying algebraic equation.
2. The stator cannot be represented by an equivalent circuit.
″ ″
The above difficulties occur due to the fact that xd ≠ xq . Let us assume that this dynamic saliency
″ ″
is neglected by setting xq = xd . Now, in the machine frame, the stator voltage equations become

Ed″ − id Ra − xd″ iq = vd
(14.12)
Eq″ − iq Ra + xd″ id = vq
(14.13)
The above equations are written in the complex form as
( Eq″ + jEd″ ) − Ra (iq + jid ) − jxd″ (iq + jid ) = (vq + jvd )
(14.14)
or
( Eq″ + jEd″ ) − ( Ra + jxd″ )(iq + jid ) = (vq + jvd )
(14.15)
If the generator is to be represented by a current source in parallel with an impedance, then the
machine current is obtained as

( Eq″ + jEd″ ) (vq + jvd )

(iq + jid ) = − (14.16)
( Ra + jxd″ ) ( Ra + jxd″ )
(vq + jvd )
= (iqS + jidS ) −
( Ra + jxd″ ) (14.17)

While interfacing the generator to the network as a current source, the above equation is transformed
to the synchronous-frame as per (14.6) to get,

( EQ″ + jED″
) (vQ + jvD )
(iQ + jiD ) = − (14.18)
( Ra + jxd″ ) ( Ra + jxd″ )

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 635 9/12/2017 4:56:23 PM

 636  |  Power System Analysis

(vQ + jvD )
= (iQS + jiDS ) −
( Ra + jxd″ ) (14.19)
1
Note that now the machine offers a constant admittance, y g = even in the synchronous
frame, and the difficulties mentioned above disappear. ( Ra + jxd″ )
Since, in most cases, this condition of dynamic saliency cannot be neglected, in [5], a dummy coil
approach has been suggested to eliminate the solution of time-varying algebraic equations and to
develop an approximate equivalent circuit.

14.1.2  Dummy-coil Approach

From (14.4), we have
Ed″ − xq″ iq − id Ra = vd

Add and subtract xd″ iq from the LHS of the above equation, we get

Ed″ − xq″ iq + xd″ iq − xd″ iq − id Ra = vd

or
Ed″ − ( xq″ − xd″ )iq − xd″ iq − id Ra = vd

Now, using (14.5) and the above equation, we can write the stator voltage equations in a compact
form as,
( Eq″ + jEd″ ) − j ( xq″ − xd″ )iq − jxd″ (iq + jid ) − Ra (iq + jid ) = (vq + jvd ) (14.20)

Let E″dummy = −(x″q – x″d)iq and the stator voltage equations now appear as

Eq″ + j ( Ed″ + Edummy

″
) − ( Ra + jxd″ )(iq + jid ) = (vq + jvd ) (14.21)

Now, note that E″q + j(E″d + E″dummy) represents a sub-transient voltage behind a sub-transient imped-
ance which involves only x″d similar to that shown in (14.15).

14.1.2.1  Computation of Dummy-coil Voltage

Since E″dummy is a function of iq, it requires the solution of the network equations. This difficulty of
calculating E″dummy is handled in the following approximate way:
E″dummy is considered as a fictitious voltage source proportional to a flux linkage of a dummy
coil in the q-axis of the armature, which has no coupling with other coils. The differential equation,
considering E″dummy as a state variable, is given by
″
dEdummy 1 ″
=  − Edummy − ( xq″ − xd″ )iq  (14.22)
dt Tdummy  

where T dummy is the time constant of the dummy coil, usually set to 0.01 s.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 636 9/12/2017 4:56:30 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  637 

Thus, a time-varying algebraic system is represented by a fast-acting differential equation by choos-

ing a small time constant. From (14.22) it can be seen that for a sufficiently small Tdummy, we have,
″
dEdummy ″
Tdummy =  − Edummy − ( xq″ − xd″ )iq 
dt  

″
0 ≈  − Edummy − ( xq″ − xd″ )iq 
 

The above expression denotes the desired E ″dummy evaluation.

14.1.3  Generator Source Current Calculations

From (14.21), the generator current can be obtained in the ‘machine-frame’ of reference as,

Eq″ + j ( Ed″ + Edummy

″
) (vq + jvd )
(iq + jid ) = −
( Ra + jxd″ ) ( Ra + jxd″ ) (14.23)
(vq + jvd )
= (iqS + jidS ) −
( Ra + jxd″ )
(14.24)
where,

Eq″ + j ( Ed″ + Edummy

″
)
(iqS + jidS ) =
( Ra + jxd″ ) (14.25)
Rearranging the terms and representing them in the ‘synchronous-frame’ of reference, we have,

(vq + jvd )
(iqS + jidS )e jd − (iq + jid )e jd = e jd
( Ra + jxd″ )

(vQ + jvD )
(iQS + jiDS ) − (iQ + jiD ) =
( Ra + jxd″ ) (14.26)

The circuit representation of (14.26) is shown in Figure 14.2. In Figure 14.2, (Ra + jx″d) has been
absorbed into bus admittance matrix. Note that I g = (iQ + jiD ) is the generator winding current in
the ‘synchronous-frame’ of reference.
Since armature resistance, Ra is relatively small, it is generally neglected. Now, we can calculate
machine currents, iq and id in the following manner:
1. Step 1: From (14.25), comparing real and imaginary parts, we have

( Ed″ + Edummy
″
)
iqS =
xd″

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 637 9/12/2017 4:56:34 PM

 638  |  Power System Analysis

(iQ + j iD )

Ra

Vg = (vQ + j vD )
IDQS = (iQS + j iDS)

x´´
d

Augmented network

Figure 14.2  Current Source Representation of Generator.

Eq″
idS = −
xd″

The d-axis sub-transient internal voltage, E″d, is evaluated as

 ( x′ q − x ″ q )  xq − x ′ q   x ″ q  
E″d = −  yK +   yG 
 x′ q  xq   x ′ q  

and the q-axis sub-transient internal voltage, E″q is evaluated as

 ( x′ − x ″ d )  x − x′ d   x ″ d  
E″q =  d yH + d    yF
 x′ d  xd   x ′ d  

where y F , y H , y G , and y K are obtained by solving the following differential equations:

dy F 1  x′ d 
=  −y F + y d + E fd 
dt T ′d  xd − x ′ d 
dy H 1
= ″ [ −y H + y d ]
dt Td

dy G 1
=  −y G + y q 
dt T ′q  

dy K 1
= [ −y K + y q ]
dt T ″q

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 638 9/12/2017 4:56:38 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  639 

where,
y d = x ″ d id + E ″ q and y q = x ″ q iq − E ″ d

and E″dummy is obtained as a solution of (14.22).

  The above current components are transformed into the ‘synchronous-frame’ of reference as
shown in Figure 14.3.

Transformation matrix

i
cos δ −sin δ QS
δ
i
DS
sin δ cos δ

i
qS

i
dS

Figure 14.3  Calculation of iQS and iDS from iqS and idS for the ith machine using di.

Notes:

1. The iQS and iDS are functions of the state variables. Therefore, when (iQS + jiDS) is used in
(14.1) as known current injection (along with the load current injections), the bus voltage (vQ +
jvD) is obtained as a result.

2. Sometimes, even the injected load currents are made functions of fast-acting dummy states so
that the network solution is obtained without an algebraic loop. The algebraic loop denotes a
condition where inputs used to solve the equations become functions of the solution themselves.
2. Step 2: Now, from (14.26), we have,
(vQ + jvD )
(iQ + jiD ) = (iQS + jiDS ) − (14.27)
jxd″

Comparing the real and imaginary parts in the above equation, we compute the generator winding
currents, iQ and iD as

v
iQ = iQS − D″
xd

vQ
iD = iDS +
xd″

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 639 9/12/2017 4:56:44 PM

 640  |  Power System Analysis

Since machine quantities are defined in the ‘rotor-frame’ of reference, the above components
of generator winding currents are transformed into the ‘rotor-frame’ of reference as shown in
Figure 14.4, before they are used in the torque expression given below:

Teg = Eq″ iq + Ed″ id + ( xd″ − xq″ )id iq (14.28)

Transformation matrix

i
cos δ sin δ q
δ
i
d
−sin δ cos δ

i
Q

i
D

Figure 14.4  Calculation of iq and id from iQ and iD for ith Machine Using di.

This electromagnetic torque is substituted in the swing equation to understand the rotor mechanical
dynamics. Further, id is used in y d calculation, and iq is used in y q as well as E″dummy calculations.

14.1.4  Modelling of Network Elements

The network elements mainly consist of transmission lines and transformers. These are modelled
neglecting their transients.

14.1.4.1  Modelling of Transmission Lines

Transmission lines are modelled corresponding to fundamental frequency using a nominal π-circuit
as shown in Figure 14.5.
i j
From To
Node Z Node

y y
2 2

Z = (r + jx ) and y = line charging (jB)

L L

Figure 14.5  Nominal p-circuit Model of Transmission Line.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 640 9/12/2017 4:56:47 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  641 

In Figure 14.5, Z = (rL + jxL) denote the series line impedance and y = jB represents the total line
charging in pu. A transmission line is included in the YBUS as follows:

i th .. j th ..
 
 1 y   1 
+ ..  −  ..
i th  Z 2   Z
 
.. .. ..  (14.29)
th  
j  1   1 y
−  ..  +  ..
 Z   Z 2 
.. .. .. 
 

If a fault removal involves tripping of a line, then in the post-fault YBUS, these four elements are
removed from the pre-fault bus admittance matrix.

14.1.4.2  Modelling of Transformers

The transformers are generally used as inter-connecting transformers and generator transformers.
These transformers are usually with off-nominal turns ratio [3] and are modelled as an equivalent p
circuit as shown in Figure 14.6.
These components are accounted in the bus admittance matrix as follows:

 pth .. sth ..

 
pth ( y ps + y p ) .. ( − y ps ) ..
.. .. ..  (14.30)
 
sth ( − y ps ) .. ( y ps + ys ) ..
.. .. .. 
 

a:1 Zeq (pu)

From To

p s

y
y = t
p ps a s
From To
side side

(1 − a) ( a − 1)
y = 2
y ys = y
p t t
a a

1
yt =
zeq

Figure 14.6  p-circuit Model of an Off-nominal Tap for a Transformer.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 641 9/12/2017 4:56:49 PM

 642  |  Power System Analysis

Note: If a fixed shunt capacitor compensation is provided at the qth bus, it is included in the YBUS as
the element, Yqq = jBsh, where Bsh represents the shunt susceptance in pu.

14.2  CENTRE OF INERTIA REFERENCE

In order to understand the concept of centre of inertia (COI)-reference, consider a two-machine power
system as shown in Figure 14.7.

Pg0 = 1.0 pu
V1 =1.0 0 V3 =1.0

M/c−2

j0.35 j 0.35
PL3 = 1.7 pu
M/c−1 1 2 3
j0.01

4 PL4 = 0.1 pu

Figure 14.7  Two-machine Power System Example.

The swing equations of two machines are given by

dd 1
= w1 − w 0
dt
2 H1 dw1
= Pm1 − Peg1 (14.31)
w B dt

dd 2
= w2 − w0
dt
2 H 2 dw 2
= Pm2 − Peg 2 (14.32)
w B dt

Note that in the above equations, w1 and w2 denote the actual rotor speeds in rad/s. It is assumed that
the Teg and Tm are approximately equal to their respective powers, neglecting speed deviations.

14.2.1 Common-mode Frequency Calculation

(Frequency Stability Analysis)
Now, consider the following equation obtained from (14.31) and (14.32),
2 d ( H1w1 + H 2w 2 )
= ( Pm1 + Pm2 ) − ( Peg1 + Peg 2 ) (14.33)
wB dt

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 642 9/12/2017 4:56:52 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  643 

Similar to the principle of the centre of gravity in mechanics, consider,

( H1w1 + H 2w 2 )
w COI =
( H1 + H 2 )
( H1w1 + H 2w 2 )
= (14.34)
HT

or

( H1w1 + H 2w 2 ) = HT w COI (14.35)

where, HT = H1 + H2.
Now, using (14.35) in (14.33), we get,
2 2
2 HT dw COI
wB dt
= ∑ Pmi − ∑Pegi (14.36)
i =1 i =1

As per load-generation balance at any instant of time, we have,

2 4
∑ Pegi = ∑PLi + PLosses (14.37)
i =1 i =3

Hence, the swing-speed equation for the entire system becomes

2 4
2 HT dw COI
wB dt
= ∑ Pmi − ∑PLi − PLosses (14.38)
i =1 i =3

Note that the above equation is written in terms of COI speed or the ‘common-mode speed’ of the
system. It is felt that in comparison to the individual machine speeds (w1 and w 2) or the relative
machine speeds (w1 − w 2), the wCOI denotes more closely the common-mode speed or the system
speed or the average system frequency. This speed depends on the load-generation imbalance as
in (14.36). This is a long-term phenomenon (in comparison to the relative speed dynamics), as
its dynamics are decided by the cumulative inertias, HT. The common-mode frequency dynamics
dominates if the relative speed and angle dynamics (due to small- and large-signal disturbances) are
stable.
A stability study related to the common-mode frequency dynamics is generally referred to as
­frequency stability analysis. Since the frequency stability study depends on the cumulative load-­
generation imbalance, the location of the generators and loads are of little importance in the study.
Further, the frequency stability can be controlled only through generation and load-control measures
as the mechanical damping and frequency-dependency characteristics of loads have very little influ-
ence on the system frequency. The generation control is primarily in terms of prime-mover controller
and generator tripping, whereas the load control generally involves load tripping, based on frequency
signal or the rate of change of frequency signal. The frequency stability control may even include
controlled system separation measures.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 643 9/12/2017 4:57:08 PM

 644  |  Power System Analysis

14.2.2  Speed and Rotor-angle Calculations with respect to COI

Similar to wCOI, the common-mode rotor-angle is defined as
H1d 1 + H 2d 2
d COI = (14.39)
HT

From (14.31) and (14.32), we have

dH1d 1 dH 2d 2
= ( H1w1 − H1w 0 ) and = ( H 2w 2 − H 2w 0 ) (14.40)
dt dt

Summing the above two equations and using (14.39), we get,

dd COI
HT = H1w1 + H 2w 2 − ( H1 + H 2 )w 0 (14.41)
dt

Using (14.35) in the above equation, we get,

dd COI
HT = HT w COI − HT w 0 (14.42)
dt

or

dd COI w − w0 
= w COI − w 0 =  COI  w B (14.43)
dt  wB

Since we know that

w COI = (1 + SmCOI )w B

from (14.36), we can write in terms of slip, for an ng-machine system, as

ng ng
dSmCOI
2 HT
dt
= ∑ Tmi − ∑Tegi = TCOI (14.44)
i =1 i =1

From (14.43) with w0 = wB, we can write,

dd COI
= SmCOI w B
dt (14.45)
It is usually preferred to refer the rotor-angle and rotor-slip of a generator with respect to the above
said time-varying COI-reference to understand the stability of relative oscillations. This is because
such a referencing removes the common modes from the individual signals so that relative motions
can be clearly seen. This is achieved by defining the following variables.

di = d i − d COI (14.46)

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 644 9/12/2017 4:57:12 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  645 

Smi = Smi − SmCOI (14.47)

In time–domain simulation, instead of plotting d i and Smi with respect to any arbitrary, jth machine,
d i − d j and Smi − Smj, the relative angle-stability of a system is inferred by plotting di or Smi against
time.

14.2.2.1  Determination of COI-variables from Individual Machines

To obtain di and Smi , the required COI-angle, d COI and COI-slip, SmCOI, can be determined by
solving (14.44) and (14.45). Instead, these variables can be obtained directly from the individual
machine variables.
The COI-angle for an ng machine system can be computed using the individual machine angles as
ng
1
d COI =
HT ∑Hid i (14.48)
i =1

where
ng = number of generator
ng
HT = ∑H i
i =1
Hi = inertia constant (in MJ/MVA) of the ith generator.
d i = rotor angle of i th generator obtained by solving the ith swing equation.
Similarly, to obtain S , the required COI-slip can be computed as follows:
mi

ng
1
SmCOI =
HT ∑Hi Smi (14.49)
i =1

where, Smi = rotor-slip of the ith generator.

If the COI speed deviation in rad/s is required the slip quantities should be multiplied by wB.

14.2.3  Swing Equations in COI-reference

Instead of calculating Smi from the Smi, we can directly obtain Smi by the solving the differential
equation shown below:
Slip equation for the ith machine is given by
dSmi
2H i = Tmi − Tegi (14.50)
dt
Since Smi = Smi − SmCOI, we have from (14.44) and (14.50),

dSmi Tmi − Tegi 1

= − TCOI (14.51)
dt 2 Hi 2 HT

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 645 9/12/2017 4:57:19 PM

 646  |  Power System Analysis

dSmi H
2H i = Tmi − Tegi − i TCOI (14.52)
dt HT

Similarly, for rotor-angle, the corresponding equation can be obtained as follows:
Angle equation for the ith machine is given by
dd i
= Smiw B (14.53)
dt

Since di = d i − d COI , from (14.45) and (14.53), we have

ddi dd i dd COI
= − = ( Smi − SmCOI )w B (14.54)
dt dt dt

or
ddi
= Smiw B (14.55)
dt

14.3 STRUCTURE OF POWER SYSTEM EQUATIONS

AND ITS SOLUTION
The complete modelling equations of power system are in the form of a set of differential-algebraic
equations (DAEs) given by

x = f ( x ,V , u )
(14.56)

YBUS V = I ( x ,V , S L )
(14.57)
where x represents the state variables.
T
u represents input vector = [ u1T , …, unTg ]T with ui = [w nLi ,VREFi ]

In (14.57), the vector of injected bus currents, I in general, represent a combination of the generators
source currents, I DQS , and load currents, ( −I L ) . If dynamic saliency is neglected or if accounted
using dummy coil approach, I DQS is a function of only state variables, whereas I L is related to load
power, SL, and bus voltage, V , as given by the following expression.
*
S 
I L =  L  (14.58)
V 

Note that the load current is treated as negative injection at a bus. From (14.56) and (14.57), it is clear
that to understand the dynamic performance of a power system, it required to solve coupled non-lin-
ear differential and non-linear algebraic equations simultaneously. It is highly complex to solve such
a system of equations because if the solution for dynamic and algebraic states are not obtained at a

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 646 9/12/2017 4:57:35 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  647 

given instant of time, it will lead to interface error. In the literature, the following two approaches
have been suggested:
1. Simultaneous implicit solution
2. Partitioned solution

14.3.1  Simultaneous Implicit Solution

In the simultaneous implicit method, first, the differential equations are discretised using an implicit
method like the trapezoidal rule. These algebraised differential equations are then combined with the
algebraic equations of the system and are solved together, employing an iterative process such as the
Newton method. This method has been employed in [14] to solve the DAEs. This method has the
following features:
1. Free from interface errors.
2. Computationally more involved for large systems.
3. Does not have a modular approach.
4. Less flexible for modification.

14.3.2  Partitioned Solution

In the partitioned method of solution, the algebraic equations and differential equations are solved
alternately as a separate set. The differential equations are first integrated using a numerical
­integration technique. While doing this, the algebraic variables (e.g., bus voltages) which are
obtained from the previous time step, are used. Therefore, the method suffers from interface errors.
This method is further categorised depending on the type of numerical integration technique as
follows:
1. Partitioned implicit solution
2. Partitioned explicit solution
The features of the partitioned solution approach [2, 1] are as follows:
1. Interface error is present.
2. Computationally less involved if we follow explicit solution techniques.
3. Modular approach, hence suited for large systems.
4. Flexibility to modify the programme with less effort.
It is noted in [2] that for large systems, partitioned solution with explicit integration method is
normally employed. In SIMULINK, one can choose, fixed-step methods, which are of explicit
type. It is found that in this category, ODE-4 (RK 4th order) or ODE-5 (Dormand Prince) method
is best suited. However, the fixed-step methods have a major limitation of a long simulation time
with a small step-size. Hence, it is always recommended to use variable step methods of explicit
kind, for example, ODE-45, ODE-13, etc. By appropriately choosing the relative and absolute
error parameters, one can achieve the desired accuracy of the solution and improved speed of
simulation [16].

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 647 9/12/2017 4:57:35 PM

 648  |  Power System Analysis

14.4  LOAD MODELLING

In a conventional load flow, loads are generally modelled as a constant power type. Here, load details
are essentially treated as load specification. However, in dynamic analysis, load model is an important
issue as it strongly influences the stability behaviour. There is no standard rule for load model selec-
tion and there are only guidelines available in the literature [2]. Accurate models are not available for
loads in system studies mainly due to the following reasons:
1. Large number of diverse composition and class of loads.
(a) Load composition: For example, heating, lighting, air condition, and welding type of loads.
(b) Load class mix: For example, industrial, commercial, residential, and agricultural type of loads.
2. Changing load composition with time: day, week, seasons.
3. Uncertainties regarding the characteristics of loads with regard to their dependency on voltage
and frequency.
4. Wide variation in the location of loads: For example, loads at sub-transmission level, substation
level, and distribution level.
In system analysis, where analysis is carried out at extra high voltage (EHV) levels, a ‘system load’ or
‘bus load’ is represented as a single power consuming device connected to a bus in the system model
as shown in Figure 14.8.

Subtransmission
EHV bus level Large customers

Residential customres and

Medium customers
Distribution
transformer

their wiring
Distribution
Feeders
Substation level
PL , Q L
(System Load) HV customers

Figure 14.8  System Load Representation.

Since, in system analysis, bus loads are to be considered rather than individual load devices, it is
extremely difficult to suggest a single model for load representation. Therefore, approximate load
models are used based on some field study and heuristic approaches.

14.4.1  Load Model Classification

The following types of load models are generally employed in system studies:
1. Static load models: In this model, real and reactive powers at any point of time are expressed as
functions of the bus voltage magnitude and frequency at the same instant. They are assumed to
reach their new steady-state almost instantaneously, following deviations in voltage magnitude
and frequency. Examples of static loads are heating, lighting loads, etc. The variation of real and
reactive components of the loads are separately expressed using algebraic equations.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 648 9/12/2017 4:57:40 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  649 

2. Dynamic load models: These loads exhibit dynamics in reaching a steady state following changes
in bus voltage magnitude and frequency. Here, real and reactive load components are expressed
as functions of bus voltage magnitude and frequency and their derivatives, and are written using
differential equations. They contribute to system state conditions. For example, induction motor,
synchronous motor loads, and thermostat loads.
In literature, modelling of power system loads has received extensive attention with studies aimed at
incorporating load models in stability studies [9, 10, 11]. In most system studies, (unless it is very
crucial like in large disturbance and short-term voltage stability analysis), static load model approx-
imations are considered even for dynamic loads. This is because it is difficult to identify the exact
composition and contribution of dynamic nature of loads at the load bus.
The voltage-dependent static load models are represented either as polynomial or exponential load
models. Polynomial representation-based aggregate static load models are briefly explained in the
following section.

14.4.2  Polynomial Load Representation

The active and reactive components of load powers are represented separately as static voltage depend-
ent models, as given below [5]:
P  P  2
PL = a1 PLo + a2  Lo  V + a3  Lo V
 Vo   Vo2 

Q  Q  2
QL = b1 QLo + b2  Lo  V + b3  Lo  V (14.59)
 Vo   Vo2 

where PL0 and QL0 are nominal values of active and reactive components of load powers at nominal
voltage, Vo .
The coefficients a1, a2, and a3 are the fractions of the constant power, constant current, and con-
stant impedance components, respectively, in the active load powers. Similarly, the coefficients b1, b2,
and b3 are defined for reactive load powers. While selecting these fractions, it should be noted that
a1 + a2 + a3 = 1
b1 + b2 + b3 = 1

From (14.59), the load power absorbed is given by

S L = PL + jQL (14.60)

with QL is positive for inductive loads and negative for capacitive loads, i.e., with load convention.
Remarks:
1. Both active and reactive components are modelled as constant impedance type by setting
a1= a2 = 0 and a3 = 1, b1 = b2 = 0, and b3 = 1. The load admittance YL is obtained as
*
S Lo = PLo + jQLo = Vo I Lo
(14.61)

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 649 9/12/2017 4:57:42 PM

 650  |  Power System Analysis

Thus,
*
S 
I Lo =  Lo  (14.62)
 Vo 

PLo − jQLo
= (14.63)
Vo*

Load admittance, is given by,

I 
YL =  Lo  (14.64)
 Vo 

Using (14.63), we get
PLo − jQLo
YL = (14.65)
Vo2

Now, at any voltage V , current drawn by the load is

I L = YLV

Since I L is a linear function of bus voltage, the load admittance YL can be absorbed into YBUS, to
make the algebraic equation (14.57) linear.
2. Modelling of active and/or reactive components as constant power/current type. (i.e., by having
a1 and/or a2, and b1, and/or b2 non-zero), makes the algebraic equation (14.57) non-linear. This
calls for iterative solution within each time-step of numerical integration. This observation has
been demonstrated in the following lines:
  Constant Current-type loads:
  The load power is given by,

V  V 
S L = PLo   + jQLo   (14.66)
 Vo   Vo 

The load current is given by,

V  V 
PLo   − jQLo  
 Vo   Vo 
IL =
V*
( PLo − jQLo )V
=
Vo V *

( PLo − jQLo ) jq
= e with V = V ∠q (14.67)
Vo

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 650 9/12/2017 4:57:46 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  651 

Constant Power-type loads:

Using the load power, the current can be computed as,
( P − jQLo ) jq
I L = Lo e with V = V ∠ q (14.68)
V
In the above two cases, the load current I L is a function of q, thus making the algebraic equation
(14.57) non-linear.

14.4.3  Frequency-dependent Load Models

It is known that under steady-state conditions, the total real power generation in the system equals the
total real power demand plus real power losses. Any deviation in this real power balance is reflected
on the system frequency. Following a perturbation, during the transient period, the system absorbs the
change in the following ways [12]:
1. ‘Borrowed’ kinetic energy from the rotating system of machines.
2. Increase/Decrease of old customer loads due to frequency-dependency of loads.
3. Increase/Decrease of generation based on the governor control.
Initially, the last two components have negligible effects; however, as speed varies, they will account for
an increased contribution. Once a steady state is reached, the speed will level off at a new constant value.
At this point, the first component is zero and the change is absorbed only by the last two components.
Further, in a generation-deficient system, the increase in load demand is absorbed by a released power
from the existing loads. This is feasible mainly because of the presence of frequency-dependent loads.
In addition to voltage dependency, the effect of a small change in frequency on the active and reac-
tive components of loads is accounted as follows:
  V   
2
V  ∆f 
PL = PLo a1 + a2   + a3    1 + k pf 
  Vo   Vo    fo 

  V   
2
V  ∆f 
QL = QLo b1 + b2   + b3    1 − k qf  (14.69)
  o
V  o 
V fo 


where,

 P  Q
∂  ∂ 
 Po   Qo 
k pf = and k qf =
 f   f 
∂  ∂ 
 fo   fo 

represent frequency sensitivity coefficients and fo denotes the nominal frequency in Hz.
In modelling of ‘system loads’, the typical value of frequency sensitivity coefficients used are
k pf = 1.5 and k qf = 2.0

and ∆f = deviation in bus frequency in Hz and is treated as positive for a rise in frequency above nominal.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 651 9/12/2017 4:57:54 PM

 652  |  Power System Analysis

Note: Frequency deviation can be calculated using the rate of change of respective bus angles. The bus
frequency deviation in rad/s at bus i is given by

dqi d   v 
=  tan −1  Di  
dt dt   vQi  

 dvDi dvQi 
 vQi dt − vDi dt 
= 
 Vi2 
 
The derivatives of vQi and vDi are obtained approximately by using the following transfer function:
s
(1+ sT )
where T is set to 0.02 s.

14.5  LOAD EQUIVALENT CIRCUIT

The load at a bus is represented by an equivalent circuit as shown in Figure 14.9.

Augmented Network

YBUS

ILn YL

(Negative
Current injection)

Figure 14.9  Equivalent Circuit of Loads.

In Figure 14.9, the load admittance YL has been absorbed into YBUS. Therefore, the net equivalent load
current to be injected at a load bus is given by
*
S 
I Ln =  L  − (YLV )
V 
= g (V , f ) (14.70)

If both active and reactive load components are modelled as constant impedance type without any frequency
dependency, then I Ln is identically equal to zero. Otherwise, I Ln is zero only at the operating point.

14.5.1  Modification of Constant Power-type Load Characteristics

It is found that when the active load power component is represented as constant power type, the pro-
gramme encounters numerical convergence problem. This is true especially when there is a severe dip
in the bus voltage. This problem is overcome by adopting the following characteristic [13]:

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 652 9/12/2017 4:58:10 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  653 

1. When the voltage magnitude at the ith load bus drops below Vc (= 0.6), change the active load
power component at that bus as
2
 Vi 
PLi = PLoi  V  (14.71)
c
2. Otherwise, the active load power component is held at PLo . In general, the types of load models
i
used are summarised in Figure 14.10.

Load characteristics
1.2
Constant P characterstic

PLo 1
2
PLo( V i /V c)

0.8

0.6
Li
P

0.4 Constant Z
characteristic

0.2 V = 0.6
c

0
1 0.8 0.6 0.4 0.2 0
Magnitude of Bus Voltage V
i

Figure 14.10  Summary of Load Models.

14.5.2 An Approach to Avoid Iterative Solution

of Algebraic Equations
It is known that when loads are modelled as constant power/current type and/or frequency dependent,
it calls for an iterative solution within a time step. A procedure similar to dummy coil approach that is
used to account dynamic saliency [5], has been employed here also to avoid iterative solution.
The non-linear algebraic equations are converted into a combination of fast-acting differential
equations and linear algebraic equations. The algebraic equations are made functions of ‘dummy’
states of the fast-acting differential equations. Thus, the non-linear algebraic equation due to load
currents is modified as follows:
dI Ld 1
=  − I Ld + g (V , f ) 
dt TL 
I Ln = I Ld (14.72)

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 653 9/12/2017 4:58:12 PM

 654  |  Power System Analysis

The time–constant TL, is chosen to be small, which implies that I Ld ≈ g (V , f ) , except for a short while
after a disturbance. This is an approximate treatment, but the degree of accuracy can be controlled
directly by choosing TL appropriately. It is found that a reasonable accuracy can be obtained if TL is
about 0.01 s. The main advantage of this method is its simplicity and modularity. Since I Ld is a com-
plex number, the above strategy is realised for real and imaginary components separately as follows:
From (14.70), I L is written as
*
S 
IL =  L 
V 

P Q 
=  L − j L  e jq
V V 

= (iP − jiR ) (cosq + jsinq )

(iQL + jiDL ) = (iP cosq + iR sinq ) + j (iP sinq − iR cosq )

(14.73)
PL Q
where iP = and iR = L .
V V
The constant impedance part of the load current which is already accounted in the YBUS is given by

(iQYL + jiDYL ) = YLV

(14.74)
Now, from (14.70) and (14.72), the net load current components are obtained as
diQLd 1
=  −iQLd + (iQL − iQYL ) 
dt TL  

diDLd 1
dt
=
TL
[ −iDLd + (iDL − iDYL )]
I Ln = (iQLd + jiDLd ) (14.75)

The initial value on these states is set to zero. This I Ln is vectorised along with I DQS in (14.57).

14.6  A SUMMARY OF THE IMPLEMENTATION PROCEDURE

The steps involved in the time–domain simulation of the power system stability analysis are enumerated
below:
1. Perform the load flow analysis of the power system for a given load-generation profile and
network connectivity.
2. Using the converged load flow results, obtain the initial value of the state variables as shown earlier.
Pre F
3. Construct the augmented bus admittance matrix for the pre-fault, YBUS , faulted, YBUS , and post-
Pos
fault system, YBUS .
4. While assembling the blocks, the signal flow between various blocks is as shown in Figure 14.11.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 654 9/12/2017 4:58:28 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  655 

5. In Figure 14.11, the augmented YBUS is switched so as to realise different network conditions.
Pre
Before, the application of the network disturbance (at t = tf ), the YBUS is set to YBUS . If a dis-
turbance is in the form of a three-phase symmetrical fault, then at t = tf , the YBUS is switched to
F
YBUS and is continued until the fault is cleared at t = tcl. To simulate the fault clearing, the bus
Pos Pos
admittance matrix is changed to YBUS . Here, if fault clearing involves line tripping, then YBUS
Pre Pos Pre
is different from the YBUS , otherwise, YBUS is identical to YBUS .
6. If the disturbance is in the form of a line trip without a fault, then at t = tf, the YBUS is simply switched
Pre Pos F
from YBUS to YBUS without using YBUS . To achieve this in the programming environment, a zero
duration three-phase fault is chosen at one of the bus so that tcl coincides with tf.
7. In the time–domain simulation, disturbances can be chosen other than the network disturbances.
For example, if a small perturbation of the reference voltage of the exciter is considered, then the
Pre
YBUS is left with YBUS throughout the simulation.
8. If a power system stability analysis is to be carried out without an excitation controller on a gen-
erator, then its Efd is set to Efd0. This effectively disables the controller. A similar disabling setting
is made even for the prime mover controller where, Tm is set to Tm0.
In this programming, partitioned explicit solution approach has been employed. For numerical
integration of differential equations, a variable step method, ODE-45, has been used by choosing
the relative error as 0.001 and absolute error as 0.01. The maximum step-size is selected as 0.01 s.

iq
Load current
Edummy injection

−I Ln
i qS & idS I QDS
−1
calculations [T] i i
Eq Ed QLd & DLd
v + jv calculations
iQ & i D Qg Dg
Augmented
[T ] calculations
YBUS
iq
v + jvDL
Rotor circuit id V QL
Efd g
Differential eqns.

Excitation
Eq E
d systems
V Ref
Torque Vs
calculation Power system
T eg stabilizer (PSS)

di Swing Turbine
Equation Tm systems

Smi PGV
p.u. slip Speed Po
governor
i th generator

Figure 14.11  Time–domain Implementation of Power System Stability Analysis Programme.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 655 9/12/2017 4:58:40 PM

 656  |  Power System Analysis

While programming in SIMULINK, vectorisation and sparsity solution techniques inherent to

MATLAB [15] have been utilised. The solution for bus voltages is obtained by declaring YBUS as
sparse and employing backslash command: V = YBUS\I. This procedure reduces the number of float-
ing point operations considerably.

14.7 DEMONSTRATION OF SYSTEM FREQUENCY

WITH TWO-MACHINE POWER SYSTEM
For the two-machine power system used earlier (see Figure 14.12), consider the following cases to
demonstrate the system frequency after the occurrence of the disturbances.

Pg0 = 1.0 pu
V1 =1.0 0 V3 =1.0

M/c−2

j0.35 j 0.35
PL3 = 1.7 pu
M/c−1 1 2 3
j 0.01

4 PL4 = 0.1 pu

Figure 14.12  Two-machine Power System.

1. A short duration (0.01 s) 3-phase fault at bus 4 without any line tripping.
2. A line between buses 3 and 4 is tripped to isolate the load at bus 4 at 0.5 s.
Note: In the above two cases, speed governors are not enabled on any machine. This implies that
the individual machines’ Tm is set to their respective Tm0.
3. A line between buses 3 and 4 is tripped to isolate the load at bus 4 at 0.5 s. In this case a speed-­
governor on machine 1, is enabled with Tm2 = Tm20.
4. A line between buses 3 and 4 is tripped to isolate the load at bus 4 at 0.5 s. In this case speed-­
governors on machine 1 and machine 2 are enabled.
The above cases are repeated with and without considering frequency-dependency for loads. The IEEE
FBM machine parameters are chosen for the generators. Inertia constant, H = 2.8941 MJ/MVA and
damping is set to zero. The nominal frequency is 60 Hz and all parameters are chosen on the machine
power base. Exciter data: KA = 200 TA1/2 = 0.025/0.02 s, and Efd max/min = ± 6 pu. Turbine data:
Reheat-type turbine with the associated speed-governor system. The block parameters are given below:

T1 = 0.02 s T2 = 0 s T3 = 0.01 s s  = 0.05 Pgmax = 1.1

Pgmin = 0.1 TCH = 2.0 s TRH = 10 s TCO = 0.4 s
FHP = 1.0 FIP = 0 FLP = 0
Load model: Constant impedance-type load model with kpf = 1.5 and kqf = 2. The base case power
flow results are tabulated in Table 14.1.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 656 9/12/2017 4:58:41 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  657 

Table 14.1  Base Case Power Flow Results-two-machine Power System.

Bus No. V p (deg) Pg0 Qg0 PL0 QL0
1 1.0000 0.0 0.8 0.2446 – –
2 0.9563 −17.023 – – – –
3 1.0000 −34.046 1.0 0.2447 1.7 0
4 0.9999 −34.104 – – 0.1 0

In the following lines, the YBUS structure is shown at various stages:

1. Accounting only transmissions lines:
 − j 2.8571 j 2.8571 0 0 
 j 2.8571 − j 5.7143 j 2.8571 0 
YBUS1 =  
0 j 2.8571 − j102.8571 j100 
 
0 0 j100 − j100 
2. Inclusion of load admittance:
 − j 2.8571 j 2.8571 0 0 
 j 2.8571 − j 5.7143 j 2.8571 0 
YBUS 2 = 
0 j 2.8571 1.7 − j102.8571 j100 
 
0 0 j100 0.1 − j100 

3. Inclusion of generator d-axis sub-transient admittance ( = 1/ j 0.135 = − j 7.4074) pu:

 − j10.265 j 2.8571 0 0 
 j 2.8571 − j 5.7143 j 2.8571 0 
Pre
YBUS 3 = 
0 j 2.8571 1.7 − j110.265 j100 
 
0 0 j100 0.1 − j100 

4. The bus admittance matrix during a three-phase fault at bus-4:

 − j10.265 j 2.8571 0 0 
 j 2.8571 − j 5.7143 j 2.8571 0 
F
YBUS 4 = 
0 j 2.8571 1.7 − j110.265 j100 
 
 0 0 j100 100000. 1 − j100 
A three-phase bolted short-circuit fault is simulated by setting a large conductance at the fault bus.
5. The bus admittance matrix to depict removal of a line between buses 3 and 4 in the post-fault system:
 − j10.265 j 2.8571 0 0 
 j 2.8571 − j 5.7143 j 2.8571 0 
Pos
YBUS 5 = 
0 j 2.8571 1.7 − j10.265 0 
 
0 0 0 0.1

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 657 9/12/2017 4:58:46 PM

 658  |  Power System Analysis

14.7.1  Three-phase Fault without Frequency-dependent Loads

A short duration self-clearing fault is considered at bus 4 with the post-fault system same as the pre-
F
fault system. To obtain this, at t = 0.5 s, the YBUS is switched to YBUS 4 and at t = 0.51 s the YBUS is
Pre
switched back to YBUS 3 . Further, the frequency sensitivity coefficients are set to zero. Figure 14.13
shows the slip-COI, i.e., (Smi − SmCOI), plots for the machines. Following the disturbance, the two
machines swings relative to each other and the relative oscillations are stable, indicating that the oper-
ating point is small-signal stable.
Before the fault, since simulation is started at an operating point, the electrical outputs of the
machines will be exactly equal to their respective mechanical inputs. However, during the fault, for a
short time, the electrical outputs of the machines deviate from their initial value. Hence, an additional
energy is imparted due to this imbalance and remains trapped in the rotor-mass system. This energy
increases the system frequency to a new constant value. This is evident in Figure 14.14 where the
COI-frequency deviation, i.e., SmCOI (or COI-speed deviation in pu) of the system is plotted. A small
reduction in the system frequency is seen in the latter part of the simulation due to a minor numerical
error in the computations. This case is similar to the multi-mass spring system where, an external
force F3 is applied for a short duration, without considering any damping.

−3 Fault at bus−4 without Freq. dependent loads

x 10
1.5
M/c−1
M/c−2

0.5
Slip wrt COI

−0.5

−1

−1.5
0 1 2 3 4 5 6 7
TIME (s)

Figure 14.13  Slip-COI of Machines-Fault without Frequency-dependent Loads.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 658 9/12/2017 4:58:48 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  659 

x 10
−3 Fault at bus 4 without Freq. dependent loads
3

2.5

1.5
COI−speed

0.5

−0. 5
0 2 4 6 8 10 12 14 16 18 20
Time (s)

Figure 14.14  System Frequency Deviation (in pu)- Fault without Frequency-dependent Loads.

14.7.2  Three-phase Fault with Frequency-dependent Loads

Here also, a short duration self-clearing fault is considered at bus 4 (with the post-fault system same
as the pre-fault system). The frequency sensitivity coefficients are set to typical values as indicated
earlier. Figure 14.15 shows the slip-COI plots for the machines. As in the previous case the relative
oscillations are stable indicating that the operating point is small-signal stable.
In this case also, an additional energy is imparted to the rotor-mass system during the fault. This
energy increases the system frequency to a new value. However, unlike the previous case, here, the
COI-frequency (or COI-speed deviation) of the system decays as time progresses due to frequen-
cy-dependency nature of real power loads (see Figure 14.16). Eventually, the system frequency settles
back to the nominal value. This also implies that the electrical outputs of the machines return to their
initial values. This case is similar to the multi-mass spring system where, an external force, F3, is
applied for a short duration, considering damping.

14.7.3  Line Trip without Frequency-dependent Loads

A line between buses 3 and 4 is tripped to isolate the load at bus 4 at 0.5 s. This, disconnects a real
power load of 0.1 pu from the system. To obtain this, at t = 0.5 s, a zero duration three-phase fault is
Pos
applied at bus-4 followed by line tripping. This switches the bus admittance matrix to YBUS 5 from
Pre F
YBUS 3 without using YBUS 4 . This triggers a relative oscillation among two machines which damps
out as time progresses. This denotes that the swing-mode is stable (see Figure 14.17).

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 659 9/12/2017 4:58:50 PM

 660  |  Power System Analysis

−3 Fault at bus−4 with Freq. dependent loads

x 10
1.5
M/c−1
M/c−2

0.5
Slip wrt COI (pu)

−0.5

−1

−1.5
0 1 2 3 4 5 6 7
TIME (s)

Figure 14.15  Slip-COI of Machines-Fault with Frequency-dependent Loads.

For a small reduction in real power load, both the machines respond as they are synchro-
nously connected to each other. This causes their electrical power out to reduce to a new value as
shown in Figure 14.18. From Figure 14.18, it is clear that the sum of Tegi is almost equal to PL3.
Since no speed governors are considered, their mechanical inputs are constant at Tm1= 0.8 pu and
Tm2= 1.0 pu.
From Figure 14.19, it can be seen that the difference between the rotor-angles (measured with
respect to COI, i.e., d i − d COI ), reduces due to a reduction in the real power flow in the line.
Since the existing real power load is frequency-independent, its value remains almost con-
stant. Further, it should be noted that the effect of speed deviation on the voltage is neglected
in the stator voltage equation as per the assumption. Hence, a change in load is also negli-
gible even though they are modelled as constant impedance-type. Thus, due to a continued
difference between the electrical outputs and mechanical inputs, the system frequency, as depicted
by COI-speed deviation (pu), continuously increases without bound -see Figure 14.20. This indi-
cates that the system is exhibiting frequency instability. This observation is identical to that in the
multi-mass spring system where, the force, F3, is chosen as a step-input without any damping. As a
result, the COI-displacement ( d COI ) depicts a t2-response and the COI-velocity ( SmCOI ) displays a
ramp-response.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 660 9/12/2017 4:58:56 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  661 

x 10
−3 Fault at bus−4 with Freq. dependent loads
3

2.5

1.5
COI−speed

0.5

−0. 5
0 2 4 6 8 10 12 14 16 18 20
Time (s)

Figure 14.16  System Frequency Deviation (in pu) - Fault with Frequency-dependent Loads.

14.7.4  Line Trip with Frequency-dependent Loads

This is similar to the previous case, except that here, frequency-dependent loads are considered. A
line trip triggers an inter-machine oscillations which are stable as shown in Figure 14.21. The corre-
sponding rotor-angle plots are depicted in Figure 14.22 where they almost settle to their initial value
under steady-state.
From the system frequency plot (see Figure 14.23), it is clear that due to the frequency-dependent
loads, the system frequency does not increase indefinitely; rather, it settles to a new higher value. This
increase in frequency in turn increases the real power load on the system.
The steady-state system frequency can be approximately calculated as follows: From (14.69) and
assuming PL to remain close to PLo= 1.7, we have,

 ∆f 
PG1 + PG 2 = PLo 1 + k pf 
 f o  (14.76)

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 661 9/12/2017 4:58:57 PM

 662  |  Power System Analysis

−4
x 10 Line trip, No turbine, without Freq. dependent loads

M/c−1
5 M/c−2

2
Slip wrt COI

−1

−2

−3

−4

−5

0 1 2 3 4 5 6 7
Time (s)

Figure 14.17  Slip-COI of Machines-Line Trip without Frequency-dependent Loads.

Note: Since the speed deviation is neglected in the stator voltage equations (in addition to neglecting
the stator transients), the electromagnetic torque (i.e., air-gap torque), Teg, developed by the machine,
now corresponds to the speed w = wB. Further, the mechanical input to the machine is assumed to be
of torque input from the turbine. Hence, the powers associated with a machine in per-unit are numer-
ically equal to the respective torque values in per-unit.
Since no turbine controllers are enabled, PG1 + PG2 = 1.8. For kpf = 1.5, the per-unit frequency
∆f
deviation can be computed as = 0.039 pu. The corresponding bus frequency plot is shown
fo
in Figure 14.24. This frequency is slightly different from the value indicated in Figure 14.23
because the SmCOI is computed as an accumulated frequency error following a differential torque,
whereas in the above equation, the load-bus frequency is calculated using the bus voltage-angle
deviation rates.
The variation of the electrical power output of the generators is shown in Figure 14.25. It can be
noted that with frequency-dependent real power loads, there exists a load-generation balance without
causing frequency instability. However, the system frequency reaches a value above the nominal.
This observation is identical to that in the multi-mass spring system where the force, F3, is chosen as
a step input considering damping.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 662 9/12/2017 4:58:59 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  663 

1.05
M/c−1
M/c−2

0.95

0.9
Teg

0.85

0.8

0.75

0.7
0 2 4 6 8 10 12 14 16 18 20
Time (s )

Figure 14.18  Electrical Power Output of Machines-Line Trip without Frequency-dependent Loads.

14.7.5 Line Trip, Speed-governor on M-1 Enabled

without Frequency-dependent Loads
In this case, the speed governor on machine 1 alone is enabled. A line trip leads to system oscilla-
tions as shown in Figure 14.26. Figure 14.26 (a) shows that the disturbance triggers relatively small
inter-machine oscillations and dominant common-mode oscillations. When the common mode SmCOI
is removed from the slip-variables, it leaves behind the relative oscillations as shown in Figure 14.26(b).
The corresponding rotor-angle plots are depicted in Figure 14.27. A reduction in the angle difference
shows that the real power flow through the line decreases to balance the new load on the system.
Since a speed governor is enabled on machine 1, the system frequency does not increase in an
unbound fashion, even without considering the frequency-dependent loads. The system frequency,
SmCOI, settles to a new higher value as shown in Figure 14.28. Note that, here, the SmCOI is calculated
as follows (since the machines are with equal H):

SmCOI =
( Sm1 + Sm1 )
2

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 663 9/12/2017 4:59:03 PM

 664  |  Power System Analysis

Line trip, no turbine, without Freq. dependent loads

0.25
M/c−1
M/c−2
0.2

0.15

0.1

0.05
deltacoi (rad.)

−0.05

−0.1

−0.15

−0.2

−0.25
0 2 4 6 8 10 12 14 16 18 20
TIME (s)

Figure 14.19  Rotor-angles-Line Trip without Frequency-dependent Loads.

14.7.6 Line Trip, Speed-governor on M-1 Enabled

with Frequency-dependent Loads
This case is similar to the previous case except that here, the frequency-dependent loads are­
considered to show their influence explicitly. As regards the system frequency, the oscillation ampli-
tudes are less and the steady-state frequency error is smaller than the previous case (see Figures 14.28
and 14.29).

14.7.6.1  Performance of Speed-governor Controller on Machine 1

In order to demonstrate the control action of the speed governor on machine 1, the torques developed
by the generators are plotted in Figure 14.30. It is clear from Figure 14.30 that the torque output of
machine 2 settles to 1.0 pu as there is no controller on machine 2, whereas machine 1 regulates its
torque output so that the total generation is equal to the new load. The new load on the system is
slightly higher than 1.7 pu due to its dependency on bus frequency.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 664 9/12/2017 4:59:05 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  665 

Line trip, no turbine, without Freq. dependent loads

0.18

0.16

0.14

0.12

0.1
COI−speed

0.08

0.06

0.04

0.02

−0.02
0 2 4 6 8 10 12 14 16 18 20
Time (s)

Figure 14.20  System Frequency Deviation (in pu)-Line Trip without Frequency-dependent Loads.

x 10−4 Line trip, No turbine, with Freq. dependent loads

6
M/c−1
M/c−2

2
Slip wrt COI

−2

−4

−6
0 1 2 3 4 5 6 7 8 9 10
Time (s)

Figure 14.21  Slip-COI of Machines-Line Trip with Frequency-dependent Loads.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 665 9/12/2017 4:59:07 PM

 666  |  Power System Analysis

Line trip, no turbine with Freq. dependent loads

0.25
M/c−1
M/c−2
0.2

0.15

0.1

0.05
deltacoi (rad)

−0.05

−0.1

−0.15

−0.2

−0.25
0 2 4 6 8 10 12 14 16 18 20
Time (s)

Figure 14.22  Rotor-angles-Line Trip with Frequency-dependent Loads.

14.7.7 Line Trip, Speed-governors on M-1 and M-2 Enabled

with Frequency-dependent Loads
In this case, the speed governor is enabled on both the machines. This implies that the torque input to
each of the machines is regulated by tracking the speed of the respective machines. In addition, the
frequency-dependent loads are considered. The slip-COI and rotor-angle plots are shown in Figures
14.31 and 14.32, respectively. It is clear from Figures 14.31 and 14.32 that rotor swings are relatively
smaller because of the control action of speed governors.
As regards the system frequency, the oscillation amplitudes reduce further and the steady-state
frequency error is much smaller than the previous case (see Figures 14.29 and 14.33).

14.7.7.1 Performance of Speed-governor Controllers

on Machines 1 and 2
In Figure 14.34, the torque outputs of machines 1 and 2 are shown, where both the speed governors
are enabled. Here, it can be seen that the torque output of both the machines is controlled so that the
net generation is again equal to the new load at a new system frequency.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 666 9/12/2017 4:59:12 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  667 

Line trip, no turbine, with Freq. dependent loads

0.045

0.04

0.035

0.03

0.025
COI−speed

0.02

0.015

0.01

0.005

−0.005
0 2 4 6 8 10 12 14 16 18 20
Time (s)

Figure 14.23  System Frequency Deviation (in pu)-Line Trip with Frequency-dependent Loads.

Line trip, No turbine, with Freq. dependent loads

0.04

0.035

0.03
Bus−3 frequency deviation (pu)

0.025

0.02

0.015

0.01

0.005

−0.005
0 2 4 6 8 10 12 14 16 18 20
Time (s)

Figure 14.24  Bus-3 Frequency Deviation-line Trip with Frequency-dependent Loads.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 667 9/12/2017 4:59:14 PM

 668  |  Power System Analysis

1.05
M/c−1
M/c−2
1

0.95

0.9
Teg (pu)

0.85

0.8

0.75

0.7
0 2 4 6 8 10 12 14 16 18 20
Time (s)

Figure 14.25  Electrical Power Output of the Generators-line Trip with

Frequency-dependent Loads.

For a power system to reach an equilibrium point, it is required that both the electrical system and
the mechanical system attain their equilibrium. Such a condition is established when a load generation
balance is achieved with regard to the following quantities on power system:
1. Reactive power balance: This implies that the reactive power generated must be equal to the
reactive power loads and reactive power losses, i.e.,

∑QG = ∑QL + Qlosses (14.77)

Since reactive power is related to the energy associated with electrical elements, it strongly
decides the system voltage.
2. Real power balance: This implies that the real power generated must be equal to the real power
loads and real power losses, i.e.,

∑PG = ∑PL + Plosses (14.78)

Since the real power delivering ability of a system is decided by the mechanical power input to
the system, it decides the generator speed and hence the system frequency. However, the reactive
power handled by a generator does not involve any electro-mechanical energy conversion, except
for I2Ra losses.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 668 9/12/2017 4:59:16 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  669 

−3
x 10
12
M/c−1
10 M/c−2

6
Slip (pu)

−2
0 1 2 3 4 5 6 7 8 9 10
(a)
−4
x 10

M/c−1
4 M/c−2

2
Slip wrt COI

−2

−4

0 1 2 3 4 5 6 7 8 9 10
Time (s)
(b)

Figure 14.26  Slip-variable and Slip wrt COI of Machines-line Trip, Speed-governor on M-1,
without Frequency-dependent Loads.

With regard to system variables, such as voltage and frequency, the following observations
are made:
(a) Response time of the electrical systems is much smaller compared to that of the mechanical
system. Therefore, the bus voltage settles much faster than the system frequency.
(b) Since reactive power controlling devices can be placed at any desired location, voltage con-
trol can be locally achieved. Such a flexibility does not exist with respect to frequency con-
trol. This is because as per the aggregated speed-COI swing equation, the system frequency
is decided by the accumulated load-generation real power mismatches. Therefore, the real
power load control at any single location cannot affect desired changes in the system fre-
quency. Thus, to regulate the system frequency, a coordinated global-level control effect
should be exercised.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 669 9/12/2017 4:59:17 PM

 670  |  Power System Analysis

Line trip, Turbine on M−1, without Freq. depen. loads

0.25
M/c−1
M/c−2
0.2

0.15

0.1

0.05
deltacoi

−0.05

−0.1

−0.15

−0.2

−0.25
0 2 4 6 8 10 12 14 16 18 20
Time (s)

Figure 14.27  Rotor-angles-line Trip, Speed-governor on M-1, without Frequency-dependent

Loads.

(c) It is noted that an excursion in system frequency is much more detrimental to system oper-
ation as it causes widespread effects, for example, turbine tripping, transformer overfluxing,
and decrease in steam-station outputs [2]. In a generation-limited power system, it is not
generally feasible to enable the speed-governors. Further, the frequency-dependent loads in
a power system cannot be expected to regulate the frequency adequately. Hence, in order
to curtail wide frequency deviations, it is generally desired to have large interconnection of
control areas so that the system possesses large inertia, which, in turn, decreases the rate of
change of frequency.

14.8  LARGE DISTURBANCE PERFORMANCE

In Section 14.7, frequency stability analysis is carried out ensuring that the post-disturbance
operating point is small-signal stable with regard to relative oscillations. In the current section,
large disturbances are considered so that the relative rotor-angle excursions are large.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 670 9/12/2017 4:59:17 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  671 

−3
x 10 Line trip, Turbine on M−1, without Freq. depen, loads
12

10

8
COI−speed

−2
0 2 4 6 8 10 12 14 16 18 20
Time (s)

Figure 14.28  System Frequency Deviation (in pu)-Line Trip, Speed Governor on M-1, without
Frequency-dependent Loads.

x 10
−3 Line trip, Turbine on M−1, with Freq. depen. loads
10

6
COI−speed

−2
0 2 4 6 8 10 12 14 16 18 20
Time (s)

Figure 14.29  System Frequency Deviation (in pu)-line Trip, Speed Governor on M-1, with
Frequency-dependent Loads.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 671 9/12/2017 4:59:19 PM

 672  |  Power System Analysis

Line trip, Turbine on M−1, with Freq. depen. loads

1.05
M/c−1
M/c−2
1

0.95

0.9
Teg (pu)

0.85

0.8

0.75

0.7

0.65
0 2 4 6 8 10 12 14 16 18 20
Time (s)

Figure 14.30  Torque Output of Machines-line Trip, Speed Governor on M-1, with
Frequency-dependent Loads

x 10−4 Line trip, Turbine on M−1 and 2, with Freq. depen. loads
6
M/c−1
M/c−2

2
Slip wrt COI

−2

−4

−6
0 1 2 3 4 5 6 7 8 9 10
Time (s)

Figure 14.31  Slip-COI of Machines-line Trip, Speed Governor on M-1 and M-2, with
Frequency-dependent Loads.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 672 9/12/2017 4:59:30 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  673 

Line trip, turbine on M-1 and 2, with Freq. depen. loads

0.25
M/c−1
M/c−2
0.2

0.15

0.1

0.05

0
deltacoi

−0.05

−0.1

−0.15

−0.2

−0.25
0 2 4 6 8 10 12 14 16 18 20
Time (s)

Figure 14.32  Rotor-angles-line Trip, Speed Governor on M-1 and M-2,

with Frequency-dependent Loads.

14.8.1  Two-machine Power System

To simulate a large disturbance, a long duration, three-phase fault with line trip is assumed in a case
where speed governors on M-1 and M-2 are enabled with frequency-dependent loads. A three-phase
fault is considered on line between buses 3 and 4 very close to bus 4. A fault-clearing strategy is
considered in which the fault is cleared by tripping the line. This scenario is generally stated as an
application of a three-phase fault at bus 4, followed by line tripping to clear the fault. Thus, in the
post-fault system, the said line will not be present. To obtain this, at t = 0.5 s (tf), the YBUS is switched
F Pre Pos
to YBUS 4 from YBUS 3 and at t = tcl, the YBUS is switched to YBUS 5 . In the following section, various
case studies are presented to demonstrate the rotor-angle swings due to such a large disturbance.

14.8.1.1  Critically Stable System

Here, the fault is applied at t = 0.5 s. The fault duration, Tclear, is progressively increased in suita-
ble steps. Thus, tcl = tf + Tclear. For each value of the clearing time, the time–domain simulation is

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 673 9/12/2017 4:59:32 PM

 674  |  Power System Analysis

x 10
−3 Line trip, Turbine on M−1 and 2, with Freq. depen. loads
7

4
COI−speed

−1
0 2 4 6 8 10 12 14 16 18 20
Time (s)

Figure 14.33  System Frequency Deviation (in pu)-line Trip, Speed Governor on M-1 and M-2,
With Frequency-dependent Loads.

repeated and the relative rotor-angle plot is observed. It is noted that at a certain clearing time the
rotor-angles exhibit the highest relative angular deviation and then it decreases. Since the post-fault
operating point is small-signal stable, these oscillations will damp out and eventually reach a steady
state. For a clearing time slightly greater than this value, the rotor angle of M-1 just separates from
the other machine. This denotes that the system has lost synchronism. In practice, such conditions
call for emergency control actions such as initiation of controlled system-separation action to prevent
complete blackouts.
For a fault duration equal to 0.55 s, Figure 14.35 shows the rotor-angle oscillations which are
just stable. This clearing time, Tclear, is referred to as the critical clearing time. Since the relative
rotor-angle deviation is large in the first cycle of the swings and can threaten the synchronous oper-
ation of the system, this large disturbance performance of the system is generally considered as the
first-swing stability of the system. In Figure 14.36, the slip of machines is plotted. The figure shows
that the oscillations damp out in the later part of the cycles indicating that the post-fault system is
stable.
For this critically stable case, real power flow in line 1 (between buses 1 and 2) and angle across
the line are also shown in Figures 14.37 and 14.38, respectively. The pre-fault value of these variables
can be verified from Table 14.1. The bus voltage magnitude at all buses is plotted in Figure 14.39.
It is clear from Figure 14.39 that during the fault and after the fault, the bus-4 voltage remains

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 674 9/12/2017 4:59:33 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  675 

Line trip, Turbine on M−1 and 2, with Freq. depen. loads

1.05
M/c−1
M/c−2

0.95

0.9
Teg (pu)

0.85

0.8

0.75

0.7
0 2 4 6 8 10 12 14 16 18 20
TIME (s )

Figure 14.34  Torque Output of Machines-line Trip, Speed Governor on M-1 and M-2,
with Frequency-dependent Loads.

zero, and other bus voltages are low during fault; once the fault is cleared, they recover close to the
pre-fault value.

14.8.1.2  Critically Unstable System

Here, the fault is applied at t = 0.5 s and is cleared at t = 1.051 s so that the system is just unstable. This
corresponds to a fault duration of 0.551 s. In Figure 14.40, the rotor angles are plotted to demonstrate
that the machines just separate from one another leading to loss-of-synchronism. This is also evident
in Figure 14.41 where the slip of machines is plotted.
For this unstable case, real power flow in line 1 and angle across the line are also shown in Figures
14.42 and 14.43, respectively. These figures clearly show that power fluctuations are quite large and
the normal power flow is no longer feasible when the system loses synchronism. This is also seen in
bus-1 voltage plot (see Figure 14.44). The fluctuation in these quantities is mainly due to sweeping of

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 675 9/12/2017 4:59:34 PM

 676  |  Power System Analysis

1.5
M/c−1
M/c−2

1
Rotor angle wrt COI (rad)

0.5

−0.5

−1

−1.5
0 5 10 15
Time (s)

Figure 14.35  Rotor-angle of Machines-two-machine System, Fault Duration = 0.55 s

0.025
M/c−1
0.02 M/c−2

0.015

0.01

0.005
COI

0
Slip

−0.005

−0.01

−0.015

−0.02

−0.025
0 2 4 6 8 10
Time (s )

Figure 14.36  Slip-COI of Machines-two-machine System, Fault Duration = 0.55 s

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 676 9/12/2017 4:59:38 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  677 

Real power vs time

1.5
Line1

1
Real power (pu)

0.5

−0.5

0 2 4 6 8 10 12 14 16 18 20
TIME (s )

Figure 14.37  Real Power Flow Through Line-1, Two-machine System, Fault Duration = 0.55 s

Angle vs time
60
Line1

50

40

30
Angle (in deg)

20

10

−10

−20
0 2 4 6 8 10 12 14 16 18 20
Time (s)

Figure 14.38  Bus Angle Across Line-1, Two-machine System, Fault Duration = 0.55 s

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 677 9/12/2017 4:59:40 PM

 678  |  Power System Analysis

0.8
(pu)

0.6
Bus
V

0.4
Bus−1
Bus−2
0.2
Bus−3
Bus−4

0
0 2 4 6 8 10
Time (s )

Figure 14.39  Bus Voltage Magnitudes, Two-machine System, Fault Duration = 0.55 s

5
M/c−1
4 M/c−2

2
Rotor angle (rad)

−1

−2

−3

−4

−5
0 0.5 1 1.5 2 2.5 3
Time (s )

Figure 14.40  Rotor-angle of Machines-two-machine System, Fault Duration = 0.551 s

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 678 9/12/2017 4:59:42 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  679 

M/c−1
0.25 M/c−2
0.2

0.15

0.1

0.05
SlipCOI

−0.05

−0.1

−0.15

−0.2

−0.25

0 0.5 1 1.5 2 2.5 3 3.5 4

Time (s )

Figure 14.41  Slip-COI of Machines-two-machine System, Fault Duration = 0.551 s.

Real power vs time

1.5
Line1

0.5
Real power (pu)

−0.5

−1

−1.5
0 0.5 1 1.5 2 2.5 3
Time (s)

Figure 14.42  Real Power Flow Through Line-1, Two-machine System, Fault Duration = 0.551 s.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 679 9/12/2017 4:59:44 PM

 680  |  Power System Analysis

Angle vs time
80
Line1

60

40

20
Angle (in deg)

−20

−40

−60

−80
0 0.5 1 1.5 2 2.5 3
Time (s)

Figure 14.43  Bus Angle Across Line-1, Two-machine System, Fault Duration = 0.551 s.

1.2

1.15

1.1

1.05
Vbus−1 (pu)

0.95

0.9

0.85

0.8

0.75
0 0.5 1 1.5 2 2.5 3
Time (s )

Figure 14.44  Bus-1 Voltage Magnitude, Two-machine System, Fault Duration = 0.551 s.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 680 9/12/2017 4:59:47 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  681 

rotor poles, known as ‘slipping of poles’, across the synchronously revolving stator field at a certain
rate, depending on the energy imparted to the rotor during the fault.

14.8.2  Four-machine, 10-bus Power Systems

The single line diagram of a 50 Hz, four-machine power system is shown in Figure 14.45. The system
details are adopted from [1].

1 5 6 9 10 8 7 3
8 4 1 6 10
2
1 3
3
9 5 7 11

Load A Load B
2 4

2 4

Figure 14.45  Four-machine Power Systems.

The generator data is chosen to be pertaining to 2.2 model as shown in Table 14.2 and power base of
100 MVA is chosen.

Table 14.2  Four-machine Power System Generator Standard Parameters.

Gen. No. xd x′d x″d T′do T″do xq x′q x″q T′qo T″qo H D

1 0.2 0.033 0.0264 8.0 0.05 0.190 0.061 0.03 0.4 0.04 54 0
2 0.2 0.033 0.0264 8.0 0.05 0.190 0.061 0.03 0.4 0.04 54 0
3 0.2 0.033 0.0264 8.0 0.05 0.190 0.061 0.03 0.4 0.04 63 0
4 0.2 0.033 0.0264 8.0 0.05 0.190 0.061 0.03 0.4 0.04 63 0

The base case power flow results are tabulated in Table 14.3. From bus-9 to bus-10, there is a net real
power flow of 0.6654 × 3 = 1.9962 pu.
The following controllers and load models are employed:
1. Excitation controllers:
(a) Generators 1 and 2 with static ST-1A -type exciter.
(b) Generators 3 and 4 with static single-time constant -type exciter.
2. Power system stabiliser: A slip-signal -type PSS is enabled on generator 2.
3. Prime mover controllers:

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 681 9/12/2017 4:59:50 PM

 682  |  Power System Analysis

Table 14.3  Base Case Power Flow Results-four-machine Power System.

Bus No. V p (deg) Pg0 Qg0 PL0 QL0

1 1.0300 8.216 7.0 1.3385 – –

2 1.0100 −1.504 7.0 1.5918 – –
3 1.0300 0 7.2172 1.4464 – –
4 1.0100 −10.205 7.0 1.8078 – –
5 1.0108 3.662 – – – –
6 0.9875 −6.243 – – – –
7 1.0095 −4.698 – – – –
8 0.9849 −14.944 – – – –
9 0.9761 −14.419 – – 11.59 2.12
10 0.9717 −23.292 – – 15.75 2.88

(a) Generators 1 and 2 with hydro -type turbines.

(b) Generator 3 with steam turbine of reheat-type.
(c) No turbine is enabled on generator 4.
4. Load models:
(a) Real power component is modelled as 30% constant power, 30% constant current, and 40%
constant impedance type.
(b) Reactive power component is modelled as constant impedance type.
In addition, the frequency dependency of loads are accounted by specifying the following variables
kpf = 1.5 and kqf = 2.
In the following section, case studies have been presented for a three-phase fault at bus 9, followed
by tripping of line 1 (between buses 9 and 10) to clear the fault.

14.8.2.1  Critically Stable Case

For a fault duration equal to 0.29 s, the rotor-angle plots are shown in Figure 14.46. It is clear
from Figure 14.46 that after exhibiting a large angular deviation, the rotor angles decrease and
oscillations damp out in later cycles, indicating that the system is large-signal stable for the chosen
clearing time.
It is to be noted that generator 2 is provided with a potential-source controlled rectifier- based
exciter system (ST1A) where the ceiling voltage of the exciter is a function of the generator terminal
voltage, whereas for generator 4, a simplified static exciter is used where the ceiling voltage is fixed at
± 6 pu. This is evident in Figure 14.47.
When field voltage is being ‘forced’, the corresponding field current variations are shown in Figure
14.48. It can be noted that even when the field voltages are ‘forced’ in the negative direction, the cur-
rent does not become negative. In Figures 14.47 and 14.48, oscillations eventually damp out since the
post-fault operating point is small-signal stable.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 682 9/12/2017 4:59:51 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  683 

Rotor angle for fault at bus−9, line tripped −1 (9−10)

2

G1
1.5

1 G2
Rotor angle (rad) wrt COI

0.5

−0.5

G3
−1

G4

−1.5
0 2 4 6 8 10 12 14 16 18 20
Time (s )

Figure 14.46  Rotor Angle for Tclear = 0.29 s, Four-machine Power Systems.

For the critically stable system, the bus frequency deviation at load buses is depicted in Figure
14.49. From Figure 14.49, the following observations are made:
1. For the post-fault system, the bus frequency deviations damp out and finally reach a non-zero
steady-state. This is clearly due to the presence of prime mover controllers and frequency-
dependent load.
2. During the oscillations, the bus-9 frequency swings almost out-of-phase with respect to bus-10
frequency. This denotes that these buses belong to different coherent areas.

14.8.2.2  Critically Unstable Case

Here, the fault-clearing time is increased so that Tclear= 0.291 s. This makes the system unstable as
shown in Figure 14.50. Here, generators 1 and 2 together separate from the generator group 3 and
4. This grouping is referred to as mode-of-instability (MOI). The clearing time, Tclear, pertaining
to the first-swing stability of the system, for which the system is just stable, is attributed as the
critical clearing time. Thus, the critical clearing time for this case is 0.29 s. For Tclear= 0.291 s, the

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 683 9/12/2017 4:59:53 PM

 684  |  Power System Analysis

10

Efd Gen 2 (pu) 5

−5

−10
0 2 4 6 8 10 12 14 16 18 20

10

5
Efd Gen 4 (pu)

−5

−10
0 2 4 6 8 10 12 14 16 18 20
Time (s)

Figure 14.47  Field Voltage Variation for Tclear = 0.29 s, Four-machine Power Systems.

Field currents for G2 and G4

22

20
G2
18

16
If (pu)

14
G4

12

10

6
0 2 4 6 8 10 12 14 16 18 20
Time (s)

Figure 14.48  Field Current Variation for Tclear = 0.29 s, Four-machine Power Systems.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 684 9/12/2017 4:59:54 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  685 

Load bus frequency deviation

0.05

0.04
Bus−9

0.03
∆ frequency (pu)

0.02

0.01
Bus−10

−0.01

−0.02
2 4 6 8 10 12 14 16 18 20
Time (s)

Figure 14.49  Load-bus Frequency Deviations for Tclear = 0.29 s, Four-machine Power Systems.

corresponding field voltage and field current fluctuations are denoted in Figures 14.51 and 14.52,
respectively. It is noted that these oscillations are quite large and do not damp out as time progresses.
A similar observation is made with respect to the load bus frequencies for this unstable case (see
Figure 14.53).

14.8.2.3  Performance of COI-variables with Different Disturbances

In order to understand the behaviour of system frequency under different disturbance conditions, the
following case studies are carried out:
1. Case-1: A three-phase fault at bus-9 with tripping of line-1 so that the post-fault system is differ-
ent from the pre-fault system (as considered in the previous sections with composite loads).
2. Case-2: A three-phase fault at bus-9 without tripping of any line so that the post-fault system is
identical to the pre-fault system.
3. Case-3: This case is similar to Case-1, except that instead of the composite load models for the
real power loads, here, constant impedance load model is employed for the real power.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 685 9/12/2017 4:59:55 PM

 686  |  Power System Analysis

Rotor angle for unstable case

20

15

10
G1

G2
Rotor angle (rad) wrt COI

G3
−5
G4

−10

−15

−20
0 1 2 3 4 5 6
Time (s )

Figure 14.50  Rotor-angle Deviation for Tclear = 0.291 s, Four-machine Power Systems.

In all the above cases, the COI-variables are observed for a given fault duration of 0.29 s.

Case 1: Composite loads with line clearing

In Figures 14.54 and 14.55, the rotor-angle and slip variables are plotted, respectively.
From the plots, the following observations are made:
1. This case is similar to multi-mass spring system where the force F3 is in the form of a step input
with damping.
2. The COI-delta shows a slowly increasing ramp-response, whereas COI-speed reaches a steady-
state with a constant DC offset. This clearly indicates that the frequency of post-fault system is
slightly above the nominal value.

Case 2: Composite loads without line clearing:

In Figures 14.56 and 14.57, the rotor angle and slip variables are plotted, respectively.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 686 9/12/2017 4:59:56 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  687 

Field voltage for unstable case

8
G2

4 G4

2
E (pu)

0
fd

−2

−4

−6

−8
0 1 2 3 4 5 6
Time (s )

Figure 14.51  Field Voltage Variation for Tclear = 0.291 s, Four-machine Power Systems.

From the plots, the following observations are made:

1. This case is similar to multi-mass spring system where the force F3 is in the form of a short dura-
tion pulse with damping.
2. The COI-delta reaches a steady-state with a constant DC offset, whereas COI-speed decays to
zero, indicating that the frequency of post-fault system eventually attains the nominal value.

Case 3: Constant impedance load model with line clearing:

In Figure 14.58, slip variables are plotted for the same fault duration of 0.29 s. The following infer-
ences are in comparison to Case-1:
1. Here, the maximum deviation in the system frequency is just above 0.015 pu, whereas in Case-1,
the deviation is well above 0.025 pu.
2. It is noted that the critical clearing time with constant impedance loads is 0.36 s, whereas, for
Case-1, it is 0.29 s.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 687 9/12/2017 5:00:00 PM

 688  |  Power System Analysis

Field currents for unstable case

22

20

G4

18

G2

16
If (pu)

14

12

10

8
0 1 2 3 4 5 6
Time (s )

Figure 14.52  Field Current Variation for Tclear = 0.291 s, Four-machine Power Systems.

The above points show that a fault with composite type of real power load models is more severe
than that with constant impedance-type load models.
3. The steady-state frequency error with constant impedance type loads is around 0.0025 pu and in
Case-1, this is relatively small (0.001 pu).

14.8.2.4  Performance of Real Power Load Models

To understand the influence of composite-type loads and constant impedance-type loads on the system
performance, the voltage magnitudes at the load buses are plotted for Case-1 and Case-3 as shown in
Figure 14.59.
From Figure 14.59, it is clear that with composite-type loads, the bus voltage reduction is much
larger than that with constant impedance-type loads. This is because a composite-type load tries to
maintain the power by drawing more current, in turn, leading to more voltage drop at majority of
the buses. The effect of this is to cause a larger dip in the electrical power output of the genera-
tors. However, with constant impedance-type loads, a reduction in voltage magnitude results in a

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 688 9/12/2017 5:00:01 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  689 

Load bus frequency for unstable case

0.07

0.06

0.05

0.04

Bus−9
∆ Frequency (pu)

0.03

0.02

0.01

Bus−10
0

−0.01

−0.02
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (s )

Figure 14.53  Load-bus Frequency Deviations for Tclear = 0.291 s, Four-machine Power Systems.

corresponding decrease in the real power demand. The leads to a load-release effect, thereby, causing
a lower voltage dip.
To appreciate the net loading behaviour of the generators and their influence on the system fre-
quency, a centre-of-inertia torque, TCOI, is computed as follows using (14.44):

4 4
TCOI = ∑ Tmi − ∑Tegi (14.79)
i =1 i =1

The plots of TCOI are shown in Figure 14.60 for these two load models. It shows that except for a
short period during the fault, the TCOI for constant impedance-type loads possesses smaller amplitudes
for most of the duration than those with composite-type loads. Thus, a higher amplitude TCOI (with
composite-type loads) produces a larger deviation in the system frequency than in Case-3. However,
its decay rate being large, the system frequency deviation decreases faster reaching a lower steady-
state frequency error (compare Figures 14.55 and 14.58).

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 689 9/12/2017 5:00:02 PM

 690  |  Power System Analysis

100

80
δ (rad)

60

40

20

0
0 5 10 15 20 25 30 35 40

100

80
COI−delta

60

40

20

0
0 5 10 15 20 25 30 35 40

1
δ wrt COI

−1

0 5 10 15 20 25 30 35 40
Time (s )

Figure 14.54  Plots of Rotor-angle Variables for Composite Loads with Line Clearing.

14.8.3  50-machine, 145-bus the IEEE Power Systems

The data for a 50-machine IEEE system has been obtained following the web link [17] on 100 MVA
base. This system has 145 buses, 401 lines, 52 transformers, 64 loads, and 97 shunts. Out of 50 gen-
erators, 44 generators follow the classical model and six generators come with a 1.1 model. These
six generators are provided with IEEE-type AC4A exciters. To demonstrate large disturbance perfor-
mance of the system, a three phase fault is applied at time t = 0.5 s at bus 7 and is cleared by tripping
line 8 (between buses 6–7). This line number is decided by its row number in file nt.dat. The loads
are modelled as constant impedance type with frequency-dependency. For a fault-clearing time of

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 690 9/12/2017 5:00:03 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  691 

0.04

0.02
Slip (pu)

−0.02
0 5 10 15 20 25 30 35 40

0.03
COI-speed

0.02

0.01

0
0 5 10 15 20 25 30 35 40

0.02

0.01
Slip wrt COI

−0.01

−0.02
0 5 10 15 20 25 30 35 40
Time (s )

Figure 14.55  Plots of Slip Variables for Composite Loads with Line Clearing.

0.143 s (Tclear), the rotor-angle plots are shown in Figure 14.61. The corresponding load bus frequency
deviations are plotted in Figure 14.62.
When the fault-clearing time is increased to 0.144 s, machines at buses 104 and 111 just separate
together from rest of the machines (see Figure 14.63). The corresponding load bus frequency devia-
tions are plotted in Figure 14.64.
It should be noted that once a generator or a group of generators lose synchronism with respect
to the rest of the system, the only control action that can be initiated is to separate the generator or
the group from the system. This is done to prevent cascaded tripping of transmission lines due to
violation of safe loading limits, thereby causing system-wide instability. While separating the unit(s),
care is taken to see that the system is split in a controlled fashion so that in each island there exists a

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 691 9/12/2017 5:00:04 PM

 692  |  Power System Analysis

40

δ (rad) 30

20

10

0
0 5 10 15 20 25 30 35 40

40

30
COI−delta

20

10

0
0 5 10 15 20 25 30 35 40

0.5
δ wrt COI

−0.5

−1
0 5 10 15 20 25 30 35 40
Time (s )

Figure 14.56  Plots of Rotor-angle Variables for Composite Loads without Line Clearing.

load-generation balance without leading to frequency instability. In order to carry out these control
actions in a coordinated manner in case of loss-of-synchronism, a system protection scheme (SPS)
is designed to detect abnormal system conditions and take pre-planned, corrective actions (other
than the isolation of faulted elements) to provide acceptable system performance [18, 19]. This is
also known as remedial action scheme (RAS) and typical schemes are generator tripping, load rejec-
tion, under-frequency load shedding, system separation, out-of-step relaying, FACTS-based control
action, etc.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 692 9/12/2017 5:00:05 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  693 

0.04

0.02
Slip (pu)

−0.02
0 5 10 15 20 25 30 35 40

0.02
COI-speed

0.01

−0.01
0 5 10 15 20 25 30 35 40

0.02

0.01
Slip wrt COI

−0.01

−0.02
0 5 10 15 20 25 30 35 40
Time (s )

Figure 14.57  Plots of Slip Variables for Composite Loads without Line Clearing.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 693 9/12/2017 5:00:12 PM

 694  |  Power System Analysis

0.03

0.02
Slip (pu)

0.01

0
0 5 10 15 20 25 30 35 40

0.02

0.015
COI-speed

0.01

0.005

0
0 5 10 15 20 25 30 35 40

0.01

0.005
Slip wrt COI

−0.005

−0.01
0 5 10 15 20 25 30 35 40
Time (s )

Figure 14.58  Plots of Slip Variables for Constant Impedance Load Model with Line Clearing.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 694 9/12/2017 5:00:13 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  695 

1.5

1
Vbus−9 (pu)

0.5

0
0 2 4 6 8 10 12 14 16 18 20
(a)

0.9
Composite
Vbus−10 (pu)

0.8 Const−Z

0.7

0.6

0.5
0 2 4 6 8 10 12 14 16 18 20
Time (s)
(b)

Figure 14.59  Plots of Voltage Magnitudes at the Load Buses for Two Types of Load Models.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 695 9/12/2017 5:00:14 PM

 696  |  Power System Analysis

20
Composite
Const−Z

15
Tcoi=Sum(Tm)−Sum(Teg) (pu)

10

−5
0 2 4 6 8 10 12 14 16 18 20
Time (s)

Figure 14.60  Plots of TCOI for Two Types of Load Models.

Fault bus−7, Line cleared−8

4
104

3
111
Rotor angle wrt COI (rad)

−1

−2
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (s)

Figure 14.61  Rotor-angle Deviation for Clearing Time = 0.143 s, 50-machine Power Systems.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 696 9/12/2017 5:00:20 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  697 

0.08

0.07

0.06
Load Bus Freq. deviations (pu)

0.05

0.04

0.03

0.02

0.01

−0.01

−0.02
0 1 2 3 4 5 6
Time (s)

Figure 14.62  Load-bus Frequency Deviations for Clearing Time = 0.143 s, 50-machine
Power Systems.

Fault at bus 7 and Line trip−8

70

60
M/cs 111 and 104
50
Rotor angle wrt COI (rad)

40

30

20

10

Rest of m/cs.
0

−10
0 0.5 1 1.5 2 2.5 3 3.5
Time (s)

Figure 14.63  Rotor-angle Deviation for Clearing Time = 0.144 s, 50-machine Power Systems.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 697 9/12/2017 5:00:28 PM

 698  |  Power System Analysis

0.25

0.2
Load Bus freq. deviations (pu)

0.15

0.1

0.05

−0.05

0 0.5 1 1.5 2 2.5 3 3.5

Time (s)

Figure 14.64  Load-bus Frequency Deviations for Clearing Time = 0.144 s, 50-machine Power
Systems.

References
  [1] K. R. Padiyar, Power System Dynamics - Stability and Control, BS Publications, Hyderabad,
India, 2002.
  [2]  P. Kundur, Power System Stability and Control, McGraw-Hill Inc., NY, 1994.
  [3]  A. R. Bergen and V. Vittal, Power System Analysis, Pearson Education Asia, India, 2001.
  [4] P. M. Anderson and A. A. Fouad, Power System Control and Stability, Iowa State University
Press, Ames, Iowa, 1977.
  [5] K. R. Padiyar, Analysis of Subsynchronous Resonance in Power Systems, Kluwer, Academic
Publishers, Boston, 1999.
  [6]  IEEE Std 115-2009, IEEE Guide for Test Procedures for Synchronous Machines.
  [7] IEEE Std 1110-2002, IEEE Guide for Synchronous Generator Modeling Practices and
Applications in Power system Stability Analysis.
  [8]  E. W. Kimbark, Power System Stability- Vol-III: Synchronous Machines, Jhon Wiley, NY, 1956.
  [9] IEEE Committee Report, ‘Load representation for dynamic performance analysis’, IEEE
Transactions on Power Systems., vol. 8(2), pp. 472–482, May 1993.
[10] IEEE Committee Report,’Standard load models for power flow and dynamic performance simu-
lations’, IEEE Transactions on Power Systems, vol. 10(3), pp. 1302–1312, August 1995.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 698 9/12/2017 5:00:30 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators  |  699 

[11] IEEE Task Force on Load representation for dynamic Performance, ‘Bibliography on load mod-
els for power flow and dynamic performance simulation’, IEEE Trans on Power Systems., vol.
10(1), pp. 523–538, February 1995.
[12] O. I. Elgerd, An Introduction Electric Energy systems Theory, Tata Mc Graw-Hill Publishing
Company Limited, Mumbai, 1983.
[13] K. N. Shubhanga, ‘Transient Stability-Constrained Generation Rescheduling and Compensation
Placement Using Energy Margin and Trajectory Sensitivities’, Ph.D. Thesis submitted to IIT
Bombay, 2003.
[14] P. W. Sauer and M. A. Pai, Power System Dynamics and Stability, Prentice Hall, Upper Saddle
River, New Jersey, 1998.
[15]  Using MATLAB, Version 5.3, Release 11, The Math Works Inc.
[16]  Using Simulink, Version 3, Release 11, The Math Works Inc.
[17]  Power System Test Cases Archive, UWEE, Available at: www.ee.washington.edu/research/pstca.
[18] Arun G. Phadke and James S. Thorp, Computer Relaying for Power System, Wiley India Pvt.
Ltd. Edition-2, New Delhi, 2012.
[19] M. Prithwish, P. Seshadri, et al., ‘WAMS and Synchrophasor Experience During Synchronization
of Large Grids in India’, The IEEE International Conference on Signal Processing, Informatics,
Communication and Energy Systems, (IEEE SPICES 2015), India, February 2015.

Review Questions
1. What is the effect of accounting dynamic saliency of a generator while interfacing it to the
network? How is it handled in transient stability studies?
2. What is the purpose of using a transformation matrix while modelling the generator as a source
of current injection into the network?
3. What is the significance of COI-frame-of-reference? How is it calculated using the machine rotor
variables?
4. Give reasons for calculating the machine swing plots with respect to COI-reference.
5. Explain different schemes used to solve differential-algebraic-equations.
6. It noted that for transient stability solution, variable-step integration method of explicit kind is
generally used, whereas, for a power electronics system simulation, variable-step integration
method of implicit kind is employed. Give reasons.
7. What are system loads? How are they modelled for power system stability analysis?
8. How are network disturbances simulated in transient stability analysis?
9. The single-line diagram of a two-machine, 6-bus 50-Hz power system is shown in Figure 14.65.
The system details are adopted from reference J.J. Grainger., et al. Choose 100 MVA as the
system base.

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 699 9/12/2017 5:00:30 PM

 700 | Power System Analysis

1 2

3 4

Figure 14.65 Two-Machine 6-bus Power System for Transient Stability Analysis.
The line data and transformer data are given in Tables 14.4 and 14.5, respectively, on 100 MVA
base.
Table 14.4 Line Data for Two-machine 6-bus Power System.

From To r x B(total)
(pu) (pu) (pu)
1 2 0.01008 0.05040 0.1025
1 3 0.00744 0.03720 0.0775
2 4 0.00744 0.03720 0.0775
3 4 0.01272 0.06360 0.1275
1 4 0.01 0.05 0.06

Table 14.5 Transformer Data for Two-machine 6-bus Power System.

From To r x Tap Ratio
(pu) (pu) (pu)
1 5 0.0 0.01 1
4 6 0.0 0.01 1

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 700 9/12/2017 5:00:37 PM

 Transient Stability Analysis of Power Systems with Interconnected Generators | 701

The bus data is given in Table 14.6.

Table 14.6 Bus Data for 2-machine 6-bus Power System.

Generation Load
Bus No. p PGp (MW) QGp (MVAR) PLp (MW) QLp (MVAR) | V p | (pu) Type
1 0 0 50 30.99 – PQ-bus
2 0 0 170 105.35 – PQ-bus
3 0 0 200 123.94 – PQ-bus
4 0 0 80 49.58 – PQ-bus
5 – – – – 1.0 Slack-bus
6 318 – – – 1.02 PV-bus

The per-unit values of the standard parameters for the generators are given on the respective
machine’s MVA ratings:
(a) Generator-6: 400 MVA, 22 kV, 50 Hz:
′ ″
xd = 2, x ′ d = 0.32, x ″ d = 0.2, xq = 1.9, x ′ q = 0.75, x ″ q = 0.2, Tdo = 6 s, Tdo = 0.05 s,
′ ″
Tqo = 1 s, Tqo = 0.05 s, H = 2.5 s.

(b) Generator-5: 250 MVA, 16 kV, 50 Hz:

′ ″
xd = 1.8, x ′ d = 0.3, x ″ d = 0.23, xq = 1.7, x ′ q = 0.65, x ″ q = 0.25, Tdo = 5 s, Tdo = 0.03 s,
′ ″
Tqo = 1 s, Tqo = 0.07 s, H = 3 s.

Exciter on G-5 and G-6: Single-time constant static exciter, K A = 200, TA = 0.02 s, Efdlimits = ± 6 pu.
Loads are modelled as constant impedance type with frequency-dependent characteristics. Using
the transient stability package, perform the following case studies:
(a) For a three-phase fault at bus 1, a line between buses 1 and 4 is tripped to clear the fault. This
line is designated as line 5 as it occupies fifth row in file nt.dat.
(b) For a three-phase fault at bus 2, a line between buses 2 and 4 is tripped to clear the fault. This
line is designated as line 3.
For the above mentioned cases, set the fault duration as 0.2 s and obtain the following plots:
- Machine speeds and rotor angles with respective to COI.
- System frequency denoted by COI speed.
- Magnitude of the bus voltages.
- Real power flow in all lines except the line which is tripped to clear the fault.
10. Derive an expression for TCOI .
11. From Figure 14.20, approximately determine the rate of rise of COI-speed deviation. Verify this
value analytically using TCOI .
12. For two-machine power systems, considering frequency independent loads in the base case, with-
out governor controllers, if a real power load of 0.1 pu is suddenly connected at bus-4 in addition
to the existing load, estimate the trend of the system frequency and its rate of deviation.
13. Comment on the importance of system frequency following a disturbance.
14. What are the different components of a typical wide-area measurement systems (WAMS)?

M14_POWER-SYSTEM-AN00_SE_XXXX_CH14.indd 701 9/12/2017 5:01:14 PM

 Dynamic Modelling of Some
Electrical Machines and their
Power-flow Analysis 15
CHAPTER OUTLINE

15.1 Induction Motor Model 15.2 Modelling of DC Motor-driven

Synchronous Generator

In this chapter, dynamic models of some rotating machines such as induction motor, DC motor, and
DC motor-driven synchronous generators are described.
The induction machine is discussed without a detailed derivation of the mathematical model. The
main objective is to understand the nature of these models since nearly 60%–70% of power systems
loads are of induction motor type. In most power system stability studies, the dynamic models of
induction machines are approximated by voltage- and frequency-dependent static loads. However,
a realistic representation is necessary as it contributes to system inertia, system faults, and system
damping [1]. In this chapter, a reduced order model is also introduced which is generally used in most
system studies. Such a model description also helps in realising induction generator operation with the
wind turbine-based energy conversion systems [2].
In addition, the modelling details of a DC motor-driven synchronous generator are presented to
understand the working of a hardware set-up which exists in most laboratories. Here, a separately
excited DC shunt motor is modelled with the lab estimated parameters. Even for the synchronous
generator, the parameters obtained by conducting tests in the lab are employed [3, 4]. Several case
studies are carried out using this model to demonstrate the effect of real power load on the generator,
synchronisation of one motor-generator (MG) set with another to share a common load, synchronisa-
tion of an MG set to main supply, etc.

15.1 INDUCTION MOTOR MODEL

In this section, the dynamic model of a single-cage/phase-wound induction motor is discussed. Here,
all quantities including the rotor parameters are referred to the stator. Hence, this model is also referred
to as the fundamental frequency dynamic model.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 702 9/12/2017 8:52:53 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  703 

15.1.1 Fifth Order Induction Motor Model

A fifth order model of the induction motor is listed below. It should be noted that all quantities are at
actuals pertaining to a starr-connected system.
1. The stator flux-linkage differential equations are

dy sD
= vsD − Rs isD − wy sQ
dt
dy sQ
= vsQ − Rs isQ + wy sD (15.1)
dt

By setting w appropriately, an induction motor model in different reference frames such as sta-
tionary frame, rotor frame, or synchronous frame can be obtained [5]. For power system applica-
tions, synchronous frame-of-reference, is desired. By setting w = wB (in electrical rad/s), we get
an induction motor model in the synchronous reference-frame (SRF). In this frame, vsQ and vsD
denote the real and imaginary components of the stator line-to-line voltage. The zero-sequence
equation is not considered in the above model assuming a balanced system.
2. The rotor flux-linkage differential equations are
dy rD
= vrD − Rr irD − (w − w r )y rQ
dt

dy rQ
= vrQ − Rr irQ + (w − w r )y rD (15.2)
dt

In the above equation, vrD and vrQ are set equal to zero to simulate a cage rotor. For a doubly-fed
induction machine (DFIM) operation, these voltage components are derived from a converter sys-
tem [6]. Further, a setting w = wB electrical rad/s, provides the SRF-based simulation. In the above
expressions, wr represents the rotor speed in electrical rad/s  = w rm  and wrm is obtained using
P
(15.5) below.  2 
3. Flux linkage-current relationship is given by

y sD   Lss 0 Lm 0  isD 
y  0 Lss 0 Lm  isQ 
 sQ  =   (15.3)
y rD   Lm 0 Lrr 0  irD 
    
y rQ  0 Lm 0 Lrr  irQ 

Notes:
1. All inductances are at actuals and are referred to the stator side. They are obtained using open-­
circuit and blocked-rotor tests.
2. Inductances Lss and Lrr per phase values and are given by

Lss = Lls + Lm

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 703 9/12/2017 8:52:56 AM

 704  |  Power System Analysis

Lrr = Llr + Lm

where Lls is the leakage inductance per phase of the stator and Llr is the leakage inductance per
phase of the rotor circuit referred to the stator circuit. Lm is the mutual inductance per phase
measured with respect to stator. Rs represents the stator winding resistance per phase and Rr
denotes the rotor resistances per phase referred to stator circuit. The iron, friction, and windage
losses, obtained from the open-circuit test, are usually neglected.
3. The expression for the developed torque, Te (in N-m), is given by

(
Te = y rD irQ −y rQ irD ) P2 (15.4)
Note that Te drives the rotor against the load torque, and P represents the number of poles.
4. The swing equation is given by
dw rm
J = Te − TL (15.5)
dt
where wrm is the rotor speed in mechanical rad/s.
J is the moment of inertia of the motor and load in kg-m2.
TL is the load torque in N-m.
Te is the air-gap torque acting on the rotor and is obtained from (15.4).

15.1.1.1 Example: Free-run of the Induction Motor Using

Quantities at Actuals
A three-phase 2.4 kW, 460 V, 60 Hz, 1750 rpm, four-pole, Δ-connected induction motor has the
following Y-equivalent per-phase parameters: Rs = 1.77 W, Rr = 1.34 W, xls = 5.25 W, xlr = 4.57 W,
xm = 139 W, J = 0.025 kg-m2, and rated slip = 1.72%. Assume full-load efficiency of 90% and full-load
power factor = 0.9 lagging. It is required to obtain free acceleration characteristics for the induction
machine. For this, TL is set to zero and the rated voltage is applied to the motor, assuming direct-on-
line (DOL) starting. The electromagnetic torque developed, rotor speed in electrical rad/s, and torque
versus speed characteristics are drawn in Figures 15.1, 15.2, and 15.3, respectively.
It is noted that even when an induction motor is simulated in different reference frames, the air-gap
torque, speed of the machine, and the stator currents in the abc-frame remain identical. For a free-run
demonstration of the induction motor, since TL is set equal to zero, it only forces the developed air-gap
torque to be very small, as the rotor speed can never attain the synchronous value.

15.1.2  Per-unit Representation of the Induction Motor Model

Here, the machine equations are given in the synchronous-frame-of-reference [7]. The base quantities
are selected for the stator circuit, as all rotor quantities are referred to the stator circuit. Choosing
these base quantities as
• Base power SB = Three-phase apparent power rating of the machine.
• Base voltage VB = VLL (Rated line-line voltage of the machine)
• Base frequency (in elec. rad/s) wB = 2p fB, where  fB = 50 (or) 60 Hz.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 704 9/12/2017 8:52:56 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  705 

Te vs time
60

50

40

30
−−−>Te (Nm)

20

10

−10

−20

−30
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−−−>Time (s)

Figure 15.1  Torque Developed by the IM on Free-run.

The other quantities are computed, as shown below, similar to a synchronous generator:

SB
Base current, I B =
VB

VB
Base impedance, Z B =
IB

VB
Base flux − linkages,y B =
wB

yB
Base inductance, LB =
IB

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 705 9/12/2017 8:52:58 AM

 706  |  Power System Analysis

Wr vs time
400

350

300

250
−−−>Wr (rad/s)

200

150

100

50

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−−−> Time (s)

Figure 15.2  Rotor Speed of the IM on Free-run.

SB
Base torque ( rotor ),TB =
w rmB

2
where, wrmB (in mech. rad/s) is the base speed for the rotor = wB .
P

Per-unit values of parameters are obtained as follows:

w0
1. p.u nominal frequency w 0 = , (w0 = 2pf0), f0 is the nominal frequency in Hz.
wB

Rs
2. p.u stator resistance/phase Rs =
ZB
Lls
3. p.u stator leakage inductance/phase Lls =  
LB

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 706 9/12/2017 8:53:03 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  707 

Torque vs speed
60

50

40

30
−−−>Te (Nm)

20

10

−10

−20

−30
0 50 100 150 200 250 300 350 400
−−−>Wr (rad/s)

Figure 15.3  Torque-speed Characteristic of the IM on Free-run.

Llr
4. p.u rotor leakage inductance/phase Llr =
LB
Lm
5. p.u mutual inductance/phase Lm =
LB
Rr
6. p.u rotor resistance/phase Rr =
ZB

7. p.u stator inductance Lss = Lm + Lls

8. p.u rotor inductance Lrr = Lm + Llr

1 2
J w rmB
9. Inertia constant, H = 2 in MJ/MVA, where J is moment of inertia in kg-m2.
SB

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 707 9/12/2017 8:53:05 AM

 708  |  Power System Analysis

The machine equations in per-unit are listed below. Note that all variables are also in per-unit.
1. Stator flux-linkage differential equations are

1 dy sD
= vsD − Rs isD − w 0y sQ
w B dt

1 dy sQ
= vsQ − Rs isQ + w 0y sD (15.6)
w B dt

2. The rotor flux-linkage differential equations are

1 dy rD
= − Rr irD − (w 0 − w r )y rQ
w B dt

1 dy rQ
= − Rr irQ + (w 0 − w r )y rD (15.7)
w B dt

The above expressions are obtained by dividing (15.1) and (15.2) by VB.

wr w
where, w r = = rm represents the per-unit rotor speed and is obtained using (15.9) below.
w B w rmB

3. Flux linkage-current relationship is given by

isD  y sD 
i  y 
 sQ  = [ K ]−1  sQ  (15.8)
irD  y rD 
   
irQ  y rQ 

where

 Lss 0 Lm 0 
 
0 Lss 0 Lm 
[ K ] = 
Lm 0 Lrr 0 
 
0 Lm 0 Lrr 

4. The swing equation in per-unit is given by

dw r
2H = Te − TL (15.9)
dt

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 708 9/12/2017 8:53:08 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  709 

where the developed electromagnetic torque in per-unit is given by

Te = y rD irQ −y rQ irD (15.10)

By setting the derivative equal to zero in (15.9), the steady-state value of the load torque is
obtained as TL0 = Te0.

15.1.2.1 An Equivalent Circuit for Induction Motor Under Steady-state

For a given slip, sr = (w 0 − w r 0 ) , the machine currents under steady-state can be computed as
follows:

vsD   Rs w 0 Lss 0 w 0 Lm  isD 

v   −w L Rs −w 0 Lm
 
0  isQ 
 sQ  =  0 ss (15.11)
0   0 sr w 0 Lm Rr sr w 0 Lrr  irD 
    
0   − sr w 0 Lm 0 − sr w 0 Lrr Rr  irQ 

The above equation is obtained by substituting the derivatives equal to zero in (15.6) and (15.7)
and, in turn, using the current-flux-linkage relationship. Under a balanced sinusoidal steady-state
condition, vsD = 0; and vsQ = 1 (in per-unit). This procedure helps us to obtain the initial values of
variables.
To obtain an equivalent circuit (in per-unit), from (15.11), we can write the quantities in phasor
form as,

vsQ + jvsD = Rs (isQ + jisD ) + jw 0 Lss (isQ + jisD ) + jw 0 Lm (irQ + jirD ) (15.12)

0 = Rr (irQ + jirD ) + jsr w 0 Lm (isQ + jisD ) + jsr w 0 Lrr (irQ + jirD ) (15.13)

Using, Lss = Lls + Lm and Lrr = Llr + Lm in (15.12) and (15.13), respectively, we have,

Vs = Rs I s + jw 0 Lls I s + jw 0 Lm ( I s + I r ) (15.14)

0 = Rr I r + jsr w 0 Lm ( I s + I r ) + jsr w 0 Llr I r (15.15)

where, Vs = (vsQ + jvsD ) , I s = (isQ + jisD ) and I r = (irQ + jirD ) .

To arrive at an equivalent circuit, (15.15) is further modified as

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 709 9/12/2017 8:53:12 AM

 710  |  Power System Analysis

Rr
I r + jw 0 Lm ( I s + I r ) + jw 0 Llr I r = 0 (15.16)
sr

From (15.14) and (15.16), an equivalent circuit on per-phase basis, in per-unit, can be drawn as shown
in Figure 15.4. In the figure, over-bars have been neglected and I m = ( I s + I r ) .

Rs Is xls Rr Ir xlr

Im
Rr (1− sr)
Vs xm
sr

Figure 15.4  An Equivalent Circuit for IM on Steady-state at Fundamental Frequency.

15.1.2.2 Example: Steady-state Condition Calculations in Per-unit

for a Loaded Case
For the previous machine, the following base values are chosen:

2.4 × 103
SB = = 2962.96 VA
0.9 × 0.9

VB = 460 V

w B = 376.9911 rad / s

w rmB = 188.495 rad / s

I B = 6.4412 A

Z B = 71.4151 Ω

y B = 1.2202 Wb

LB = 0.1894 H

TB = 15.719 Nm

The per-unit values of the machine parameters are

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 710 9/12/2017 8:53:18 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  711 

Rs
Rs = = 0.024785
ZB

Rr
Rr = = 0.018764
ZB

Lls
Lls = = 0.073514
LB

Llr
Llr = = 0.063992
LB

Lm
Lm = = 1.9464
LB

w 0 = 1.0

1 2
J w rmB
H= 2 = 0.14989 s
SB

Assuming full-load on the machine, we set sr = 0.0172 pu. Using (15.11), we have,

 0   0.02478 2.01988 0 1.94637  isD 0 

1   −2.01988 0.02478 −1.94637 0  i 
 =   sQ 0 
0 0 0.03348 0.01876 0.03458 irD 0 
    
 0   −0.03348 0 −0.03458 0.01876 irQ 0 

and solving for currents we get,

isD 0   −0.574448
i  0.829639 
 sQ 0  =  
irD 0  0.092932 
   
irQ 0   −0.85366 

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 711 9/12/2017 8:53:21 AM

 712  |  Power System Analysis

The stator current, I s0 = (isQ0 + jisD0) = 1.0091 ∠ – 34.67° pu and rotor current, I r0 = (irQ0 + jirD0) =
0.8587 ∠ 173.79°. Therefore, the magnetising current, I m0 = I s0 + I r0 = 0.48211 ∠ – 92.86° pu. The
–
supply voltage, Vs = 1.0 ∠ 0° pu.
The initial values of the flux-linkages can be obtained from (15.8) as

y sD 0   2.01988 0 1.94637 0   −0.574448

y   0 2.01988 0 1.994637   0.829639
 sQ 0  =   
y rD 0  1.94637 0 2.01036 0   0.092932
    
y rQ 0   0 1.94637 0 2.01036  −0.85366 

 −0.979438
0.014238 
= 
 −0.931261
 
 −0.10138 

Now, from (15.10), the developed electromagnetic air-gap torque in per-unit is,

(
Te 0 = y rD 0 irQ 0 −y rQ 0 irD 0 )
= [( −0.9313 × −0.8537) − ( −0.1014 × 0.0929) ]

= 0.80440 pu

This air-gap torque developed in per-unit drives the rotor at speed w r0 = (1 – sr) = (1 - 0.0172) =
0.98280 pu against the load torque TL0 in per-unit. This Te0 is numerically equal to the air-gap power
developed in per-unit at synchronous speed, w 0 = 1.0 pu.
The complex input power is given by S s = Ps + jQs = Vs I s* . The real and reactive power balance
can be verified as follows:

Real Power Balance:

Input real power = vsD0isD0 + vsQ0isQ0

= 0.829639 pu

Stator copper losses, I s20 Rs = (isD

2 2
0 + isQ 0 ) Rs = 0.025238

2 2
In the above calculations, (isD 0 + isQ 0 ) Rs represents the three-phase copper losses for the stator
windings.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 712 9/12/2017 8:53:24 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  713 

Air − gap power = Input real power − Stator copper losses

= 0.829639 − 0.025238
= 0.80440 = Te 0 pu

Rotor copper losses, I r20 Rr = irD

2
( 2
)
0 + irQ 0 Rr = 0.013836 pu

Also, note that

I r20 Rr 0.013836
Air − gap power = = = 0.80442 = Te 0 pu
sr 0.0172

Mechanical power output = Te 0w r 0 = 0.80440 × 0.98280

= 0.79057 pu

= Te 0 − I r20 Rr
(1 − sr )
= I r20 Rr
sr
and

Input real power = Mechanical power output + ( I r20 Rr + I s20 Rs )

= 0.79057 + (0.013836 + 0.025238)
= 0.80440 + 0.025238

= Te 0 + I s20 Rs
= 0.82964 pu

Reactive Power Balance:

The reactive power input can be obtained as

Input reactive power, Qs = vsD 0 isQ 0 − vsQ 0 isD 0

= 0.574448 pu

Now, the net reactive power losses are given by

Qloss = I s20 Lls + I m

2 2
0 Lm + I r 0 Llr

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 713 9/12/2017 8:53:29 AM

 714  |  Power System Analysis

= 0.57446 pu

The power factor of operation is cos(34.67°) = 0.822 lagging.

15.1.2.3 Example: Free-run and Constant Load Torque

Characteristics
Here, to begin with, the induction motor is started on no-load by setting TL = 0. Once the motor picks
up speed on DOL starting, a constant load torque of TL0 = 0.8044 pu is applied at t = 0.4 s. The per-
unit values of the electromagnetic torque developed, rotor speed, and stator phase currents are shown
in Figures 15.5, 15.6, and 15.7, respectively.

Te vs time in per unit values

4

2
−−−>Te (pu)

−1

−2
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−−−>Time (s)

Figure 15.5  Per Unit Torque Developed by the IM on Free-run and on Loading.

From Figure 15.5, we can see that the maximum value of the torque in Nm is given by 3.3173 times
the base torque, TB. This is equal to 52.145 Nm, which can be verified from Figure 15.1. On loading

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 714 9/12/2017 8:53:30 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  715 

Wr vs time in per unit values

1.4

1.2

1
−−−> Wr (pu)

0.8

0.6

0.4

0.2

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−−−>Time (s)

Figure 15.6  Rotor Speed of the IM in Per Unit on Free-run and on Loading.

the machine with a constant torque load, the steady-state speed of the motor is 0.9828 per-unit which
leads to a slip of 0.0172 pu, as stated. From Figure 15.7, it can be seen that the stator currents during
starting is relatively large (nearly five to six times the rated current) and as the motor picks up speed
this current reduces.

15.1.2.4  P
 ower Balance: Some Observations Under
Dynamic Conditions
A power balance analysis is carried out during the dynamic conditions as well, in terms of compo-
nents, in SRF as shown in Figure 15.8. The input power represents the instantaneous power given by

T
p(t ) = iabcs vabcs = vsD isD + vsQ isQ

The sum of mechanical output power and copper losses is computed as

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 715 9/12/2017 8:53:31 AM

 716  |  Power System Analysis

Stator currents vs time in per unit values

10

5
−−−>isa (pu)

−5

−10
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−−−>Time (s)
10

5
−−−>isb (pu)

−5

−10
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−−−>Time (s)
10

5
−−−>isc (pu)

−5

−10
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−−−>Time (s)

Figure 15.7  Stator Phase Currents of the IM on Free-run and On Loading.

2 2 2 2
pmR (t ) = Te × w r + (isD + isQ ) Rs + (irD + irQ ) Rr (15.17)

From Figure 15.8, it is clear that upto t = 0.2 s, the waveform p(t) differs from pmR(t) plot. Once high
frequency (electrical circuit) transients decay appreciably, the difference between them becomes rela-
tively small, and in the later part, where low-frequency (mechanical system) transients are dominant,
this difference is negligible. Therefor, the difference between p(t) and pmR(t) during the initial tran-
sient period can be attributed to the rate of change of magnetic energy as per the generalised theory
of rotating machines, which is given by

dWmg
p(t ) = i 2 R losses + + (Te × w r )
dt

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 716 9/12/2017 8:53:33 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  717 

7
Mechanical output power+copper losses
Input inst. power
6

4
power (pu)

−1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Time (s )

Figure 15.8  Power Balance Under Dynamic Conditions.

dWmg
During the electrical circuit transient period, the component is not zero. However, once a
dt
three-phase balanced sinusoidal steady-state is reached, this term becomes zero and p(t) = pmR(t).
This simply denotes a real power balance in the machine. This condition also implies a state where a
steady-state is reached with the magnetic system under sinor excitation. This, in turn, shows the reac-
tive power balance. Figure 15.8 also shows that electrical circuits reach their steady-state much faster
than mechanical systems. This means that for every changing state of the mechanical variables, we can
assume that electrical systems are already at their sinusoidal steady-state.

15.1.2.5 Example: Influence of Speed-dependent Load

Torque Characteristics
In this case, the load torque applied at time t = 0.4 s is assumed to vary as a function of speed, instead
of maintaining it at a constant value. Thus, TL in per unit is expressed as a function of the per-unit
rotor speed as

TL = TL 0 (w r )m

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 717 9/12/2017 8:53:34 AM

 718  |  Power System Analysis

where m is decided by the load characteristics to model fan or blower type of loads. In this simula-
tion analysis, m is set to 0, 1, and 2, in each run. The developed torque Te in each case is plotted in
Figure 15.9.

Te vs time in per unit values

1.2
m=0
m=1
m=2

1.1

1
−−−>Te (pu)

0.9

0.8

0.7

0.6
0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75
−−−>Time (s)

Figure 15.9  The Plot of Electromagnetic Torque for Different Speed Functions.

From Figure 15.9, it is clear that for m = 2, the amplitude of torque oscillations is relatively small as
against that with m = 1 or m = 0. Further, it is noted that under steady state, the slip values are sr =
0.0172 (m = 0), sr = 0.0169 (m = 1), and sr = 0.0166 (m = 2).

15.1.3 Reduced Order Model for the Induction Machine

In this section, a third order model for the induction motor is derived using the above fifth order model.
dy sD dy sQ
In this model, stator transients are neglected. This done by setting = 0 and = 0 . Further,
dt dt

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 718 9/12/2017 8:53:35 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  719 

these equations are derived in the synchronous reference-frame with w = wB. These changes make
the stator voltage equation purely algebraic and thus enable easy interfacing of the induction machine
model to the network algebraic equations in a power-swing type of analysis. In the following lines, the
process of reducing the order of the model is described.
From (15.6), the stator voltage equation becomes

vsD = Rs isD +y sQ

vsQ = Rs isQ −y sD (15.18)

From (15.7), the rotor voltage differential equations become

 w −wr  1 dy rD
vrD = 0 = Rr irD +  B  y rQ +
 wB  w B dt

 w −wr  1 dy rQ
vrQ = 0 = Rr irQ −  B  y rD + (15.19)
 wB  w B dt

From (15.3), the flux linkage-current relations are written as follows:

For the stator,

y sD = Lss isD + Lm irD (15.20)

y sQ = Lss isQ + Lm irQ (15.21)

For the rotor,

y rD = Lm isD + Lrr irD (15.22)

y rQ = Lm isQ + Lrr irQ
(15.23)

15.1.3.1 Modification of the Stator Equations

From (15.22), we can write,

y rD − Lm isD
irD = (15.24)
Lrr

Using the above expression in (15.20), we get,

Lm L 2
y sD = Lss isD + y rD − m isD
Lrr Lrr

or

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 719 9/12/2017 8:53:38 AM

 720  |  Power System Analysis

Lm  L 2
y sD = y rD +  Lss − m  isD
Lrr  Lrr 

Similarly, writing an expression for irQ from (15.23) and substituting it in (15.21), we get

Lm  L 2
y sQ = y rQ +  Lss − m  isQ
Lrr  Lrr 

Let us define

Lm
vQ′ = − y rD
Lrr
Lm
′ =
vD y rQ
Lrr
 L 2
and X s′ =  Lss − m  (15.25)
 Lrr 
Notes:
1. X′s (in p.u) is defined as the per unit transient reactance of the induction motor which is also equal
to L′s in per unit.
2. v′Q is defined as the transient Q-axis internal voltage and v′D is defined as the transient D-axis
internal voltage.

Now, with the above defined transient voltages and the reactance, the ysD and ysQ expressions become

y sD = X s′isD − vQ′
y sQ = X s′isQ + vD
′ (15.26)

Using (15.26) in (15.18), we get,

vsD = Rs isD + X s′isQ + vD

′

vsQ = Rs isQ − X s′isD + vQ′

Now, using the phasor notations to represent the quantities, we get,

vsQ + jvsD = Rs isQ − X s′ isD + vQ ′ + j ( Rs isD + X s′ isQ + vD ′ )

= Rs (isQ + jisD ) + jX s′ (isQ + jisD ) + (vQ

′ ′
+ jvD )

Defining Vs = vsQ + jvsD ; I s = isQ + jisD and V ′ = vQ′ + jvD

′ , we have,

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 720 9/12/2017 8:53:42 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  721 

Vs = ( Rs + jX s′ ) I s + V ′ (15.27)

From the above equation, it is evident that an induction machine can be represented by a simple tran-
sient equivalent circuit as shown in Figure 15.10.

Rs X´s

Is

Vs V´

Figure 15.10  Induction Machine Transient-equivalent Circuit.

15.1.3.2 Modification of the Rotor Equations

D-axis Equations:
From (15.19), we have,

 w −wr  1 dy rD
0 = Rr irD +  B  y rQ + (15.28)
 wB  w B dt

From (15.25), we have,

Lrr
y rD = − vQ′
Lm

Lrr
y rQ = ′ (15.29)
vD
Lm

Using the above expressions in (15.28), we get,

 w − w r   Lrr  1  Lrr  dvQ′

0 = Rr irD +  B    ′ +
vD  − L  dt
 w B   Lm  wB m

Rr   Lrr  1  Lrr  dvQ′ 

0= irD Lm + (w B − w r )   ′ +
vD  − R  dt  (15.30)
Lm   w B Rr  wB r 

Let us define

1  Lrr 
T0′ =
w B  Rr 

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 721 9/12/2017 8:53:49 AM

 722  |  Power System Analysis

which is referred to as transient open-circuit time-constant in seconds. Note that Lrr and Rr are in
per unit.
Rewriting (15.24), we have

y rD − Lm isD
irD =
Lrr

y rD Lm
= − isD
Lrr Lrr

Using (15.29) in the above equation, we get,

vQ′ Lm
irD = − − isD (15.31)
Lm Lrr

Using the above expression for irD in (15.30), we get,

L2m dvQ′
0 = −vQ′ − isD + (w B − w r ) T0′vD
′ − T0′ (15.32)
Lrr dt
From (15.25), we have,

L2m
X s′ = Lss −
Lrr

or

L2m
= ( X ss − X s′ ) (15.33)
Lrr

where Lss in per unit is denoted by Xss in per unit.

Now, (15.32) becomes,

dvQ′
0 = −vQ′ − ( X ss − X s′ )isD + (w B − w r ) T0′vD
′ − T0′
dt

or

dvQ′ 1
=  −vQ′ − ( X ss − X s′ )isD  + (w B − w r ) vD
′ (15.34)
dt T0′  

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 722 9/12/2017 8:53:52 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  723 

Q-axis Equations:
From (15.19), we have,

 w −wr  1 dy rQ
0 = Rr irQ −  B  y rD + (15.35)
 wB  w B dt

Using (15.29) in the above equation, we get,

 w − w r   Lrr  1  Lrr  dvD

′
0 = Rr irQ −  B   −  vQ′ +  
 w B   Lm  w B  Lm  dt

or

Rr   Lrr  ′ 
1  Lrr  dvD
0= irQ Lm + (w B − w r )   vQ′ +    (15.36)
Lm   w B Rr  w B  Rr  dt 

From (15.23), we get,

y rQ − Lm isQ
irQ =
Lrr

y rQ Lm
= − isQ
Lrr Lrr

Using (15.29) in the above equation, we get,

′
vD L
irQ = − m isQ (15.37)
Lm Lrr

Employing (15.33) and (15.37) in (15.36), we get,

′
dvD
′ − ( X ss − X s′ )isQ + (w B − w r ) T0′vQ′ + T0′
0 = vD
dt

or
′
dvD 1
= ′ + ( X ss − X s′ )isQ  − (w B − w r ) vQ′ (15.38)
 −vD
dt T0′ 

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 723 9/12/2017 8:53:54 AM

 724  |  Power System Analysis

Note:
If wound rotor is employed, then as in the case of a doubly-fed induction generator (DFIG), voltages
may be injected to the rotor circuit. In such cases, the following changes are made in the fundamental
frequency model of the induction machine:

dvQ′ 1  Lm 
=  −vQ′ − ( X ss − X s′ )isD − vrD  + (w B − w r ) vD
′ (15.39)
dt T0′  Rr 

′
dvD 1  L 
= ′ + ( X ss − X s′ )isQ + m vrQ  − (w B − w r ) vQ′ (15.40)
 −vD
dt T0′  Rr 

(w B − w r ) = sr w B = slip speed in electrical rad/s.

15.1.3.3 Modification of the Torque Expression

Using (15.29) in torque equation (15.10) we have,

Te = y rD irQ −y rQ irD

Lrr L
=− ′ rr irD
vQ′ irQ − vD
Lm Lm

Further, using (15.31) and (15.37) in the above expression, we get,

Lrr  v′ L  L  vQ′ L 
Te = − vQ′  D − m isQ  − vD
′ rr  − − m isD 
Lm L
 m Lrr  Lm  Lm Lrr 

or

Te = vQ′ isQ + vD
′ isD (15.41)

15.1.3.4 A Summary of the Reduced Order Model

The induction motor model generally employed for power system stability studies is listed below:
1. Stator voltage equation:

Vs = ( Rs + jX s′ ) I s + V ′ (15.42)

The above equation denotes the quasi-steady-state condition of the stator, where transients are
neglected.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 724 9/12/2017 8:53:58 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  725 

2. Rotor electrical equations:

dvQ′ 1
=  −vQ′ − ( X ss − X s′ )isD  + (w B − w r ) vD
′ (15.43)
dt T0′  

′
dvD 1
= ′ + ( X ss − X s′ )isQ  − (w B − w r ) vQ′ (15.44)
 −vD
dt T0′ 

Note that wr is the rotor speed in electrical rad/s.

3. Rotor mechanical equation:

dw r
2H = Te − TL (15.45)
dt

where the developed electromagnetic torque in per-unit is given by

Te = vQ′ isQ + vD
′ isD (15.46)

In order to compare the performance of the third order model with the fifth order model, the free-run
followed by constant load torque application-test is conducted with the third order model. The result-
ing plots for electromagnetic torque and per unit speed are shown in Figures 15.11 and 15.12, respec-
tively. From the figures, it is clear that the third order model closely follows the fifth order model,
except for the initial transients mainly due to the stator circuit.

15.1.3.5  Induction Generator Operation of Induction Machine

(Reduced Order Model)
In this section, induction generator operation of the induction machine is demonstrated using third-or-
der model for the induction machine. As per the motor convention, if the variable TL used in (15.45)
is positive, it denotes a torque delivered to the load. Thus, to get generator operation, TL should be
reversed in sign to represent a prime mover which drives the machine. To show this, the following
steps are employed (see Figure 15.13):
1. At t = 0, a supply voltage of 1 pu is applied with TL set to zero, until t = 0.8 s. This implies that
until this time, the induction machine behaves like a motor with vsQ + jvsD = 1.0 ∠ 0°. During this
period, the motor draws a negligible isQ, denoting a small amount of real power and an appreci-
able iSD which represents the reactive power consumed by the motor. This is because, as vsD = 0,
we have,

PS + jQS = (vsQ) × (isQ + jisD)*

or

PS = vsQisQ  and  QS = −vsQisD

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 725 9/12/2017 8:54:00 AM

 726  |  Power System Analysis

Te vs time in per unit values

4
Detailed model
Reduced model

2
−−−>Te (pu)

−1

−2
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−−−>Time (s)

Figure 15.11  The Plot of Electromagnetic Torque for Reduced Order IM Model.

2. At t = 0.8 s, the torque TL is switched to −0.8044 pu. This is equivalent to driving the rotor of the
machine by an external prime mover. This causes the machine to run at a speed higher than 1 pu
(i.e. at a supersynchronous speed) and makes the machine to run with a negative slip, providing
a generator operation, and delivering real power to the mains. This is denoted by a negative isQ0,
i.e., isQ0 = −0.7802 pu. However, the machine continues to absorb reactive power from the mains
denoted by a negative isD (which continues to be negative).

The power flow under new steady-state condition is verified as follows:

1. The components of the stator current drawn are,

isQ0 + jisD0 = - 0.7802 - j0.6061 = 0.98797 ∠ - 142.16° pu.

2. The real power delivered to the mains is, Ps = −0.7802 pu. The reactive power absorbed from
mains is Qs = 0.6061 pu.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 726 9/12/2017 8:54:01 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  727 

Wr vs time in per unit values

1.4
Detailed model
Reduced model

1.2

0.8
−−−>Wr (pu)

0.6

0.4

0.2

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−−−>Time (s)

Figure 15.12  The Plot of Per-unit Rotor Speed for Reduced Order IM Model.

3. The rotor power input = TL × w r = - 0.8044 × 1.0158 = - 0.81715 pu. As per (15.45), under steady-
state, Te = TL. Hence, the power at the air gap is given by Te × w0  = - 0.8044 × 1.0 = - 0.8044 pu.
4. The stator copper losses = I2s0 × Rs = 0.987972 × 0.024785 = 0.024192 pu.
5. The real power available at the mains = - 0.8044 + 0.024192 = - 0.78021 pu.

15.2 MODELLING OF DC MOTOR-DRIVEN
SYNCHRONOUS GENERATOR
In order to demonstrate the interfacing of a DC motor model to a synchronous generator model, DC
shunt motor modelling equations are listed below:

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 727 9/12/2017 8:54:01 AM

 728  |  Power System Analysis

IG operation of IM
3
Te (pu) 2
1 −0.8044 pu
0
−1
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (s)

1
Speed (pu)

0.9 1.0158 pu

0.8
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (s)
0
isD (pu)

−1
−0.6061 pu

−2
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (s)
2

1
isQ (pu)

−0.7802 pu
0

−1
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (s)

Figure 15.13  The Plot for Induction Generator Operation of the Induction Machine.

15.2.1 DC Motor Modelling

A separately excited DC shunt motor is employed for driving the synchronous generator. The state
modelling equations are highlighted below without derivation.
1. Armature circuit equation:

dI a R k V
= − a I a − bt w a + a (15.47)
dt La La La
where
La = Inductance of the armature, Ra = Armature resistance.
Va = Voltage applied to the armature in V, Ia = Armature current in A.
kbt = Machine coefficient for a given constant excitation.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 728 9/12/2017 8:54:02 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  729 

2. Mechanical equation:

dw a B k T
=− w a + bt I a − L (15.48)
dt Jm Jm Jm
where
   Jm = Moment of inertia of the motor.
      B = Damping coefficient of the motor-generator (MG) setup.
   TL = Load torque in N-m.
     wa = Mechanical speed of the motor in rad/s.

For a given shunt field current, we know that the back-emf, Eb = kbtwa, and under steady-state we can
compute it as Eb = Va − IaRa. Hence, for a set of applied voltage, Va, the resulting Ia and wa are noted
under no-load condition of the machine. From these observations, a series of values for kbt is obtained.
Assuming no saturation, an average of these values is chosen as the final kbt. Similarly, an estimate
for B is determined by calculating Tm = kbtIa and using the relationship, Tm = Bwa under no-load. A
typical value is selected for La.

15.2.1.1 Example: No-load Run of a DC Motor

The typical parameters for a 200 V, 3 kW, four-pole, 1500 rpm, shunt wound DC motor located in the
lab is given below:

La = 0.0063 H, Ra = 1.8W, kbt = 1.1607 V-s.

Jm = 0.08735 kg-m2, B = 0.023 Nm-s.

In order to understand the motor performance, a no-load test (where TL = 0) is conducted on the
motor by suddenly applying a DC voltage of 200 V without using any starting resistance. For this
condition, the armature current, speed of the motor and the torque developed by the machine are
plotted as shown in Figure 15.14. The figure shows that since no starting resistance is used in the sim-
ulation, the armature current transiently reaches a large value equal to 102 A which is approximately
200
given by = 111.1 A.
1.8
The steady-state values are verified as follows:

Power developed = Va × I a0 − I a20 × Ra

= 200 × 3.313 − 3.3132 × 1.8

= 642.84 W

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 729 9/12/2017 8:54:03 AM

 730  |  Power System Analysis

DC Armature Current (Ia)

150

100
Ia(A )

50 3.313 A

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
t (s)
DC Motor Speed (N)
2000

1500
N (rpm)

1000
1596.44 rpm

500

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
t (s)
Torque Developed (Tm)
150

100
Tm (N−m)

50 3.8452 Nm

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
t (s)

Figure 15.14  The Plots Pertaining to No-load Test on a DC Motor.

Torque developed, Tm0 = kbtIa0= 1.1607 × 3.313 = 3.8454 N-m. From Figure 15.14, we also have

2p N rpm0
Power developed = Tm0 ×
60
2p 1596.44
= 3.8452 ×
60
= 642.84 W

Further, note that under steady-state, Tm0 = Bwa0= (0.023 × 167.18) N-m.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 730 9/12/2017 8:54:04 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  731 

15.2.2 DC Motor Driving a Synchronous Generator

Here, the DC motor described in the previous section is mechanically coupled to a three-phase
­synchronous generator (see Figure 15.15). The generator equations are identical to that considered
for the short-circuit test earlier, and a large terminal resistance Ropen is used here to simulate the
open-circuit condition. The machine parameters considered for the simulation are given in Table 15.1.
The lab machine is rated for 230 V line-to-line, four-pole, 1000 VA, 50 Hz, and is a starr-connected
synchronous generator.

R is switched from Ropen to RL at t = 4 s

DC motor armature winding

DC motor field winding + Ia Starr

connected
A a
resistors
+ Va +
− − N
M
R
b

c
closed at t = 0

Field of the generator

closed at t = 3 s
+
−

Efd

Figure 15.15  DC Motor Coupled to a Synchronous Generator with Resistive Load.

Table 15.1  Standard Parameters for a 230 V, 1000 VA Generator in pu on the Machine Base.

xd = 0.423 x′d = 0.219 x″d = 0.128

xq = xd x′q = xd x″q = xd

T ′d = 18 ms T ″d = 11 ms

T ′q = 0.85 s T ″q = 0.05 s

Ra = 0.035 Jg = 0.08735 kg-m2 50 Hz

In order to ‘couple’ the generator to the DC motor, the mechanical equation of the DC motor indicated
above is modified as

dw a
J m′ = − Bw a + Tm − Teg (15.49)
dt

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 731 9/12/2017 8:54:05 AM

 732  |  Power System Analysis

where
J ′m = (Jm + Jg) = the combined moment of inertia of the MG setup.
B = Damping coefficient of the MG setup.
wa = Mechanical speed of the MG setup in rad/s.
Tm = kbtIa = Torque developed by the motor in N-m.
Teg = Electromagnetic torque developed by the generator in N-m. The per-unit value of the gener-
ator torque is converted into actuals (N-m) by multiplying it by the base torque. The base values are
given by
Voltage base, VB = 230 V

Power base, S B = 1000 VA

Speed base, w B = 314.16 rad / s

Current base, I B = 4.3478 A

Impedance base, Z B = 52.9 Ω

Rotor speed base, w rB = 157.08 rad / s

Rotor Torque base, TB = 6.3662 Nm

15.2.2.1 Synchronous Generator Supplying a Pure Resistive Load

A pure resistive load is interfaced to the generator as follows:
1. The terminal voltage in the machine-frame is given by

vgd = igd × R and vgq = igq × R(15.50)

where the machine currents, igd and igq are obtained as functions of machine state-variables and
i0 is set to zero.
2. Upto t = 4 s, the value of R is set to Ropen and beyond 4 s, it is set to RL = 0.7 pu.
3. The machine terminal voltages and load currents are obtained by transforming the machine-
frame quantities to abc-frame by employing Park-inverse transformation matrix as

vabc = [ P ]−1 v0 dqg

(15.51)

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 732 9/12/2017 8:54:08 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  733 

where

 1 
 cosq sinq 
 2 
2 1  2p   2p  
[ P ]−1 = cos q −  sin q − 
3 2  3   3 
 
 1  2p   2p  
 cos q +  sin q + 
 2  3   3  

with q = w t = (w0t + d  ) and d is obtained by solving

dd  w a 
= w0 − w0
dt  w rB 

where w0 is set to wB.

4. The the line-to-line voltage vLL(t) is computed as ( va (t ) − vb (t ) ) .

In this case, to begin with, the DC motor is started by applying 200 V as in the previous case. The
generator terminal resistance Ropen is set to 10 pu to approximate an open-circuit condition. The MG
set speed attains 1.0642 pu. At t = 3 s, a field voltage of Efd0= 1 pu is applied to the generator. The

line-to-line voltage (RMS) builds up to a value equal to 342.8 = 242.4 V. The following observations
are made (see Figure 15.16): 2

1. Once the field voltage is applied, the speed of the MG set drops to 1.0588 pu from 1.0642 pu as
the generator loads the DC motor due to a finite Ropen.
2. Since Efd0= 1 pu, the open-circuit line-to-line voltage (RMS) should be VB. Since the speed of the
generator is 1.0588 pu, the expected terminal voltage is 230 × 1.0588 = 243.52 V.

Now, at t = 4 s, the Ropen is switched to RL whose value is set to 0.7 pu, keeping the Efd0 at 1 pu.
270.54
This causes the terminal voltage to drop to a value equal to = 191.3 V. It should also be
2
noted that while the generator is made to deliver real power, the speed of the MG set decreases to
1.0103 pu. This implies that the frequency of the generator drops from the earlier 50 × 1.0588 = 52.94
Hz to 50 × 1.0103 = 50.51 Hz. Figure 15.17 shows the plots of per-phase voltage and current in
per-unit.
The following observations are made:
0.679
1. From Figure 15.17, we can see that equals the load resistance of 0.7 pu and the ia (t )
0.97
waveform is in phase with the voltage-wave. Also, note that the period of the wave is 0.0198 s and
the frequency is given by 50.505 Hz.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 733 9/12/2017 8:54:10 AM

 734  |  Power System Analysis

DC Armature Current (Ia)

8
3.313 A
Ia (A ) 6 8.779 A
3.863 A

4
2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5
t (s)
DC Motor Speed

1.06
w (pu)

1.04 1.0643 1.0103

1.0588
1.02

2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5

t (s)
Torque Developed (Tg)

1
Teg (pu)

0.1053
0.5 1.0273

0
2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5
t(s)
Generator terminal voltage (line−to−line) (V )

200
VLL (V )

0
−200 342.8

2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5

t (s) 270.54 V

Figure 15.16  The Plots Pertaining to R-load on the Generator.

2. Three-phase load power can be computed as

0.679 0.97
× × (3 × S B ) = 987.95 W
2 2

3. The total generator armature copper loss is given by

2
 0.97 
  × 0.035 × (3 × S B ) = 49.397 W
2

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 734 9/12/2017 8:54:11 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  735 

Phase−a voltage and currents

1
0.97 pu va(t)
0.8 ia(t)

0.6

0.4

0.679
Per unit values

0.2

−0.2 0.0198 s

−0.4

−0.6

−0.8

−1
5.005 5.01 5.015 5.02 5.025 5.03 5.035 5.04 5.045 5.05
t (s)

Figure 15.17  The Instantaneous Phase-voltage and Phase-current Plots Pertaining

to R-load on the Generator.

4. The total real power developed by the generator is

Teg × 1.0103 = 1.0273 × 1.0103 = 1.0379 pu × S B = 1037.9 W

5. Further, it can be verified that the total real power consumed is,

= Load real power + armature copper loss

= 987.95 + 49.397 = 1037.3W

15.2.2.2 Synchronous Generator Supplying a Series RL-Load

In this case, after building the voltage on open circuit as in the previous case, at t = 4 s, a series RL
circuit is connected across the terminals. The circuit connection is as shown in Figure 15.18.
In order to interface the RL-series circuit, the circuit equation in the abc-frame, is transformed into
the machine-frame or 0 dq-frame as follows:
The circuit equation in the abc-frame is given by

v ga  ia  ia 
    d  
v gb  = Rl ib  + L dt ib  (15.52)
v gc  ic  ic 
 
xL
where Rl and L = are scalar values.
w0

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 735 9/12/2017 8:54:14 AM

 736  |  Power System Analysis

−1
Now, using the relationships v gabc = [ P ] v0 dq and iLabc = [ P ] iL 0 dq and following the pro-
−1

cedure indicated in the earlier chapters, the above equation can be transformed into the 0 dq-frame as

v g 0  i L 0  i L 0  i L 0 
    d    
v gd  = Rl iLd  + L dt iLd  + Lw 0w pu iLq  (15.53)
v gq  iLq  iLq   −iLd 
     

dq w
In the above derivation, is set to w 0w pu where w pu = a .
dt w rB
Since the zero component does not exist for a balanced condition, the above equations are
rewritten as

diLd R v gd
= − l iLd + − w 0w pu iLq (15.54)
dt L L

diLq Rl v gq
=− iLq + + w 0w pu iLd (15.55)
dt L L

In (15.50), it is noted that the machine currents, idg and iqg, are functions of the machine state variables
and even in the above expressions, the load currents are con-
sidered as state-variables. Therefore, to retain the identity of ig iL
the machine terminal voltage, vgd and vgq, the following steps
ir
are employed as per Figure 15.18. Rl
vg
ig = ir + iL(15.56) Ropen

Re-writing the above equation for d- and q-components, L

we have

ird = igd − iLd(15.57) Figure 15.18  Connection of

RL-series Circuit Across
irq = igq − iLq(15.58)
the Generator Terminals.
Further, we can see that the machine terminal voltages are
given by

vgd = Ropen ird  and  vgq = Ropen irq(15.59)

For carrying out the load test, Ropen is set to 10 pu. An RL load with Rl = 0.05 pu and xL = 0.698 pu
is applied at t = 4 s, keeping the Efd0 at 1 pu. This causes the terminal voltage to drop to a value equal

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 736 9/12/2017 8:54:16 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  737 

213.85
to = 151.21 V. From Figure 15.16, it can be seen that when the generator is made to deliver
2
R = 0.7 pu, the speed of the MG set decreased to 1.0103 pu. However, in the current case, where the
real power load corresponds only to Rl = 0.05 pu, the speed drops to 1.0588 pu (see Figure 15.19).
This clearly denotes that under steady state, the real power load on the generator decides the amount
electromagnetic torque developed by the machine and hence causes a speed drop (see mechanical
equation (15.49)), in addition to a terminal voltage decrease due to armature reaction effects. Further,
until the prime mover responds to this speed change, the real power load is met by the rate of release of
stored kinetic energy of the rotor mass. Thus, a primemover controller which changes the mechanical
input to the generator as per the speed change is a requirement for all conventional type generators
such as thermal, hydro systems, etc. However, in the current case, the DC motor acts as a prime mover
which does this speed regulation inherently. Suppose an alternator is required to deliver only reactive
power, then the speed drop of an MG set is negligible except corresponding to a small amount of
machine losses.
Figure 15.20 shows the plots of per-phase voltage and current in per-unit with RL load. The fol-
lowing observations are made:
1. The frequency of the generator is fm = 50 × 1.0588 = 52.94 Hz. The period measured from the
figure is 0.0189 s and the frequency is given by 52.91 Hz.
2. The phase-angle between the voltage and current is given by 0.0045 × fm × 360° = 85.76°
lagging.
3. The impedance of the load is obtained as

0.5367
ZL = ∠85.76° = 0.74058∠85.76° = 0.0547 + j 0.7385 pu
0.7247

4. Further, note that the new xL = 0.698 × 1.0588 = 0.7390 pu. Thus, the actual ZL = 0.05 + j 0.7390
= 0.7407 ∠ 86.13° pu.
The following observations are made with regard to power balance:
1. Three-phase real power load can be computed as

 0.5367   0.7247  o
   cos(85.76 ) × (3 × S B ) = 43.1039 W
2  2 

2. Three-phase reactive power load can be computed as

 0.5367   0.7247 

2
 
2
( o
)
 sin 85.76 × (3 × S B ) = 581.773 VAR

3. The real and reactive power supplied by the generator to the load is given by

PGL + jQGL = (vgq + jvgd) (iLq + jiLd)*

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 737 9/12/2017 8:54:18 AM

 738  |  Power System Analysis

DC Armature Current (Ia)

4
3.8
Ia (A )
3.6
3.4 3.8685 A
3.2
2.5 3 3.5 4 4.5 5 5.5 6
t (s)
DC Motor Speed

1.064
1.062
w (pu)

1.0588 pu
1.06
1.058
2.5 3 3.5 4 4.5 5 5.5 6
t (s)
Torque Developed (Tg)

1
0.1045 pu
Teg (pu)

0.5
0
−0.5
2.5 3 3.5 4 4.5 5 5.5 6
t (s)
Generator terminal voltage (line−to−line) (V )

200
VLL (V)

0
−200

2.5 3 3.5 4 4.5 5 5.5 6

t (s) 213.85 V
342.83 V

Figure 15.19  The Plots Pertaining to RL-load on the Generator.

Comparing the real parts on both sides we get,

PGL = (vgqiLq + vgdiLd) × SB = 39.39W

Similarly, by comparing the imaginary parts, we get,

QGL = (vgdiLq − vgqiLd) × SB = 582.2VAR

4. The total real power developed by the generator is given by

Teg × 1.0588 = 0.1045 × 1.0588 = 0.11064 pu × SB = 110.64 W

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 738 9/12/2017 8:54:19 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  739 

RL load on the generator

0.8
va(t)
0.5367 pu iaL(t)
0.6
0.7247 pu
0.4

0.2
Per−phase

−0.2 0.0045 s
0.0189 s

−0.4

−0.6

−0.8
5.005 5.01 5.015 5.02 5.025 5.03 5.035 5.04 5.045 5.05
t (s)

Figure 15.20  Instantaneous Phase-voltage and Phase-current Plots Pertaining

to RL-load on the Generator.

5. Losses in the Ropen is given by

2
 0.5367  1
  (3 × S B ) = 43.2 W
2 Ropen

6. The total generator armature copper loss is given by

0.51632 × 0.035 × (3 × SB) = 27.99W

where 0.5163 pu represents the generator per-phase RMS current.

7. Further, it can be verified that the total real power consumed is,
= Load real power + loss in Ropen + armature copper loss

= 39.39 + 43.2 + 27.99 = 110.58 W

15.2.2.3 Synchronous Generator Supplying a Series RC-Load

In this case, after building the voltage on open circuit as in the previous case, at t = 4 s, a series RC
circuit is connected across the terminals instead of an RL-series circuit. The circuit connection is
shown in Figure 15.21.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 739 9/12/2017 8:54:20 AM

 740  |  Power System Analysis

Since the generator current and the capacitor voltage are considered as state variable, in order to
interface the RC-series circuit across Ropen, the following procedure is employed:

v g = Rc iL + vC (15.60)

From Figure 15.21, we can write that ig iL

ir = ig − iL(15.61) ir Rc

Since vg Ropen
+ vc
vg = Ropenir = Ropen (ig − iL)(15.62) C

from (15.60), we have,

Figure 15.21  Connection of
Ropen (ig − iL) = RciL + vC(15.63) RC-series Circuit Across the
Generator Terminals.
Thus, the new relationship is given by

Ropen ig = (Rc + Ropen) iL + vC(15.64)

dvC
Since we know that iL = C , the final differential equation to be handled is given by
dt

(
vOpen = Rc + Ropen C) dvC
dt
+ vC (15.65)

where vopen = Ropenig.

Writing the above equation in the matrix form for abc-quantities, we get,

vOpen,a  vCa  vCa 

  


( )d  
vOpen,b  = vCb  + Rc + Ropen C dt vCb  (15.66)
vOpen,c  vCc  vCc 
   
1
Where Rc, Ropen, and C = are scalar values.
xC w 0
−1 −1
Now, using the relationships vOpen,abc = [ P ] vOpen,0 dq and vCabc = [ P ] vC 0 dq and follow-
ing the procedure indicated in the earlier chapters, the above equation can be transformed into the
0dq-frame as

vOpen,0  vC 0  vC 0  vC 0 

    d  
v = v
 Open,d   Cd  + Rc (
+ Ropen C )
dt
v
 Cd  + R(
c + Ropen Cw) w
 
0 pu  vCq  (15.67)
vOpen,q  vCq  vCq   −vCd 
     

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 740 9/12/2017 8:54:27 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  741 

Since the zero component does not exist for a balanced condition, the above equations are rewritten as

dvCd 1 vOpen,d
=− vCd + − w 0w pu vCq (15.68)
dt (
Rc + Ropen C ) Rc + Ropen C( )
dvCq 1 vOpen,q
=− vCq + + w 0w pu vCd (15.69)
dt ( Rc + Ropen ) C ( Rc + Ropen ) C
After solving the above equations, the load current is calculated from (15.64) as

Ropenigd − vCd
iLd =
( Rc + Ropen )
Ropenigq − vCq
iLq = (15.70)
( Rc + Ropen )
Once the components of load currents are obtained the terminal voltage components, vgq and vgd, of
the machine can determined using (15.62).
Here, Ropen is set to 10 pu and a RC load with Rc = 0.7 pu and xC = 0.8 pu is applied at t = 4 s, keep-
ing the Efd0 at 1 pu. This causes the terminal voltage to increase due to the magnetising effect of the
408.8
load current to a value equal to = 289.07 V. Since the generator supplies a considerable amount
2
of real power the speed drops to 1.0015 pu from 1.0642 pu (see Figure 15.22).
Figure 15.23 shows the plots of per-phase voltage and current in per-unit with RC load. The fol-
lowing observations are made:
1. The frequency of the generator is fm = 50 × 1.0015 = 50.07 Hz. The period measured from the
figure is 0.0199 s and the measured frequency is given by 50.25 Hz.
2. The phase-angle between the voltage and current is given by 0.0027 × fm × 360° = 48.67° leading.
3. The impedance of the load is obtained as

1.0265
ZL = ∠ − 48.67° = 1.0618∠ − 48.67° = 0.70117 − j 0.79729 pu
0.9668
0.8
4. Further, note that the new xC = = 0.7988 pu. Thus, the actual ZL = 0.7 - j 0.7988 =
1.0015
1.0621 ∠ - 48.74° pu.
The following observations are made with regard to power balance:
1. Three-phase real power load can be computed as
 1.0265   0.9668 
   cos( 48.67°) × (3 × S B ) = 981.1 W
2  2 

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 741 9/12/2017 8:54:29 AM

 742  |  Power System Analysis

DC Armature Current (Ia)

10
8

Ia (A ) 6 9.6615 A
4
2.5 3 3.5 4 4.5 5 5.5 6 6.5 7
t (s)
DC Motor Speed
1.06
1.04
w (pu)

1.02 1.0015 pu
1
0.98
2.5 3 3.5 4 4.5 5 5.5 6 6.5 7
t (s)
Torque Developed (Tg)

1
Teg (pu)

0.5 1.1931 pu

0
2.5 3 3.5 4 4.5 5 5.5 6 6.5 7
t(s)
Generator terminal voltage (line−to−line) (V )
400
200
VLL (V )

0
−200
−400
2.5 3 3.5 4 4.5 5 5.5 6 6.5 7
342.8 V t (s) 408.8 V

Figure 15.22  The Plots Pertaining to RC-load on the Generator.

2. Three-phase reactive power load can be computed as

 1.0265   0.9668  o
   sin( −48.67 ) × (3 × S B ) = −1119.6 VAR
2  2 
3. The real power supplied by the generator to the load is given by
PGL = (vgqiLq + vgdiLd) × SB = 980.59 W

The reactive power supplied by the generator to the load is given by

QGL = (vgdiLq − vgqiLd) × SB = −1119 VAR

4. The total real power developed by the generator is given by

Teg × 1.0015 = 1.1931 × 1.0015 = 1.1949pu × SB = 1194.9 W

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 742 9/12/2017 8:54:31 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  743 

1.5
1.0265 pu va(t)
ia(t)
0.9668 pu
1

0.5

−0.5 0.0027 s
0.0199 s

−1

−1.5
5.005 5.01 5.015 5.02 5.025 5.03 5.035 5.04 5.045 5.05

Figure 15.23  The Instantaneous Phase-voltage and Phase-current Plots Pertaining

to RC-load on the Generator.

5. Losses in the Ropen is given by

2
 1.0265  1
  (3 × S B ) = 158.06 W
2 Ropen
6. The total generator armature copper loss is given by
0.733622 × 0.035 × (3 × SB) = 56.511 W

7. Further, it can be verified that the total real power consumed is

= Load real power + loss in Ropen + armature copper loss
= 981.1 + 158.06 + 56.511 = 1195.7 W

15.2.3  S
 ynchronisation of Two DC Motor-driven
Synchronous Generators
Here, two DC motor-driven synchronous generators, MG-1 and MG-2, are considered for the analysis.
For MG-1, the DC motor supply voltage is kept at 200 V and for the MG-2, the applied DC voltage is
set at 200.25 V. With this, the two MG sets run at slightly different speeds before synchronising them.
For example, before the application of the field voltage, the speed of the machines is1.0643 pu and
1.0656 pu, respectively. Once a field voltage of 1 pu is applied to the generators, at t = 3 s, the MG-1
runs at a speed of 1.0588 pu, whereas MG-2 runs at 1.0601 pu.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 743 9/12/2017 8:54:32 AM

 744  |  Power System Analysis

In order to carry out the synchronisation, the interfacing equations are written in the abc-frame
(instead of the SRF) as this frame is common to both the MG sets. The model equations are as follows:

Before synchronisation:

1. For generator-1:
The terminal voltage of the machine in the abc-frame is obtained as
vabc1 = R1 iabcg1(15.71)
where
−1
iabcg1 = [ P (q1 ) ] i0 dq1

Note that in the above equation, q1 = wa1t and wa1 (in electrical rad/s) is obtained from (15.49) corre-
P
sponding to MG set 1. From (15.49) we get wa1 in mechanical rad/s, hence, multiply it by before
2
using it in q1. Further, idq1 is obtained as a function of the machine’s state variables, and i0 is set to zero.
Having obtained vabc1, the terminal voltage of the machine in the 0dq-frame of G-1, is obtained as

v0 dq1 = [ P (q1 ) ] vabc1

2. For generator-2:
The terminal voltage of the machine in the abc-frame, is determined as
vabc2 = R2 iabcg2(15.72)
where
−1
iabcg 2 = [ P (q 2 ) ] i0 dq2

Note that in the above equation q2 = wa2t and wa2 (in electrical rad/s) is obtained from (15.49)
corresponding to MG set 2. As in the previous case, the terminal voltage of the machine in the
0dq-frame of G-2, is obtained as
v0 dq2 = [ P (q 2 ) ] vabc 2

After synchronisation:
The two generators ‘see’ a common terminal voltage in the abc-frame and is given by
vabc = RS (iabcg1 + iabcg2)(15.73)
where
−1 −1
iabcg1 = [ P (q1 ) ] i0 dq1; and iabcg 2 = [ P (q 2 ) ] i0 dq2

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 744 9/12/2017 8:54:34 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  745 

In the above equation, q2 = wa2t as before, however, q1 is now determined as

q1 = w a2t + d 1
with
dd 1
= (w a1 − w a2 )
dt
Further, the machine terminal voltages in their respective 0 dq-frame are obtained as

v0 dq1 = [ P (q1 ) ] vabc ; and v0 dq2 = [ P (q 2 ) ] vabc

The complete process has been diagrammatically represented in Figure 15.24.

i 0dq1 i abcg1 i abcg2

2 2 i 0dq2

1 1 vdq2
vdq1 G1 [P (θ 1)] − 1 R1 RS R2 [P (θ 2)] − 1 G2
1' 1’

2' 2'
vabc1 = R 1i abcg1 vabc2= R 2i abcg2

[P(θ 1 ) ] [P (θ 2)]

After the switch is moved to 2-2'

vabc1 = vabc2 = vabc = R s (i abcg1+ i abcg2)

Figure 15.24  Synchronisation of Two MG Sets.

At t = 4 s, the two generators are synchronised by moving the terminal switches from position 1-1′
to 2-2′ simultaneously, for both the machines. The corresponding plots are shown in Figure 15.25.
The following observations are made:
1. After reaching a steady state, both the MG sets run at an identical speed of 1.0621 pu. Before reach-
1
ing this steady-state, they oscillate relative to one another with a frequency of = 2.065 Hz.
0.4842
2. Before synchronisation, each generator is made to deliver a real power of 111.1 W pertaining to
its own R1 = R2 = Ropen = 10 pu. However, after synchronisation, both generators together supply
a common terminal resistance of RS = 10 pu. Now, G-1 ends up supplying 42.78 W, whereas G-2
supplies 69.59 W. This is because the no-load speed of G-1 is, wa1= 1.0588 pu, and that of G-2 is,
wa2= 1.0601 pu. This is verified from Figures 15.26 and 15.27.
For G-1, we have

 0.8654   0.03292 
P1 =    × 3 × S B = 42.73 W
 2  2 

For G-2, we have

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 745 9/12/2017 8:54:37 AM

 746  |  Power System Analysis

DC Armature Current (Ia)

6 MG−1
3.6661 A
Idc (A) MG−2
4
3.5272 A
2
3 4 5 6 7 8 9
t (s)
DC Motor Speed
1.0656 pu
1.08 1.0621 pu
w (pu)

1.06

1.04 1.0643 pu

3 4 5 6 7 8 9
t(s)
Real power output
400
69.59 W
200
Pout (W)

0
−200 42.78 W
−400
3 4 5 6 7 8 9
t(s)
Delta of G1 wrt G2
1
Delta (rad)

0 −1.5755 rad

−1
−0.0108 rad = −0.618 deg.
0.4842 s
−2
3 4 5 6 7 8 9
t(s)

Figure 15.25  Generator Variables After Synchronisation of Two MG Sets.

 0.8654   0.05365 
P2 =    × 3 × S B = 69.64 W
 2  2 

The total load power is given by

2
 0.8654  1
PT =   × 3 × S B = 112.41 W
 2  10

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 746 9/12/2017 8:54:39 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  747 

1
0.8654 pu vag1(t)
0.8 iag1(t)

0.6

0.4 0.3292/10 pu

0.2
Phase voltage

−0.2

−0.4

−0.6

−0.8

−1
8 8.005 8.01 8.015 8.02 8.025 8.03 8.035 8.04 8.045 8.05
t(s)

Figure 15.26  Generator Terminal Voltage and Current for G-1 After Synchronisation.

1
0.8654 pu vag2(t)
0.8 iag2(t)

0.6 0.5365/10 pu

0.4

0.2
Phase voltage

−0.2
0.0188 s
−0.4

−0.6

−0.8

−1
8 8.005 8.01 8.015 8.02 8.025 8.03 8.035 8.04 8.045 8.05
t(s)

Figure 15.27  Generator Terminal Voltage and Current for G-2 After Synchronisation.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 747 9/12/2017 8:54:41 AM

 748  |  Power System Analysis

1
3. From Figure 15.27, the time period can be noted as 0.0188 s. Hence the frequency = =
0.0188
53.191 Hz. From the speed of the setup, we get fm = 1.0621 × 50 = 53.105 Hz.

4. The DC motor-1 armature current is 3.5272 A, whereas the DC motor-2 armature current is
3.6661 A. This clearly denotes that motor-2 draws more power than that of motor-1 even when
they are running at the same speed.

15.2.3.1 Demonstration of Influence of Phase-angle

on Synchronisation Transients
In the previous results, the synchronisation is carried out at t = 4 s. From Figure 15.25, it can
be  observed that this chosen time corresponds to a phase angle of −1.5755 rad = −90.27°. This
phase-­angle difference is clearly seen between the generator terminal voltages, va1(t) and va2(t),
just prior to the synchronisation (see Figure 15.28), where it is denoted by a time delay of 0.0049
s. However, once the synchronisation is successful, there is no phase-angle difference between the
generator terminal voltages showing synchronous operation of the machines.
In order to understand the influence of the phase-angle difference that exists at the time of synchro-
nisation on the nature of transients, another time instant is chosen corresponding to t = 15.45 s instead
of t = 4 s. This offers a phase-angle difference of −6.345 rad = −363.53° = −3.53°. This is shown in
Figure 15.29.

1.5
0.0049 s va1(t)
va2(t)

0.5
Phase voltage

−0.5

−1

Synchronized
−1.5
3.95 4 4.05 4.1 4.15
t(s)

Figure 15.28  Generator Terminal Voltages Before and After Synchronisation (t = 4 s).

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 748 9/12/2017 8:54:42 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  749 

1.5
va1(t)
va2(t)

0.5
Phase voltage

−0.5

−1

Tsyn=15.45 s
−1.5
15.4 15.42 15.44 15.46 15.48 15.5 15.52 15.54 15.56 15.58 15.6
t(s)

Figure 15.29  Generator Terminal Voltages Before and After Synchronisation (t = 15.45 s).

The transients in the rotor angle and real power delivered by G-1 are plotted in Figures 15.30 and
15.31, respectively. From Figure 15.30, it can be seen that for t = 4 s, the deviation in the angle is
quite large and is equal to 0.787 − (−0.0108) = 0.7978 rad, whereas for t = 15.45 s, the angle devi-
ation is very small, less than 0.05 rad. This influence is dominantly observed in real power output
of generators. For example, in Figure 15.31, the power excursions are quite large in the range of
thousands for t =4 s, whereas with t = 15.45 s, the power swings are well within 100 W. It should be
noted that in either of the cases, the steady-state powers and angle (−6.294 + 2p = −0.01081 rad)
remain identical.

15.2.3.2 Demonstration of Influence of Field Voltage Variation

In this case, after synchronising the generators at t = 4 s, at time t = 7 s, the field voltage of G-1 is
changed to 1.1 pu and at the same time, the field voltage of G-2 is decreased to 0.9 pu (from their
earlier 1 pu values) so that the terminal voltage does not change significantly. The following transients
are plotted in Figure 15.32.
From Figure 15.32, the following observations are made:
1. The generator speed does not change appreciably. The speed changes to 1.062 pu from the earlier
value of 1.0621 pu.
2. There is a negligible change in the real power output of the generators.
3. The rotor angle slightly increases to −0.0312 rad from the earlier −0.0108 rad.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 749 9/12/2017 8:54:42 AM

 750  |  Power System Analysis

Delta of G1 wrt G2 (Tsyn=4 s)

1

0.787 rad
0
Delta (rad)

−1
−0.0108 rad

−2
2 3 4 5 6 7 8 9 10
t(s)
Tsyn=15.45 s
−6.25
Delta (rad)

−6.3

−6.294 rad
−6.35
−6.345 rad

14 15 16 17 18 19 20
t(s)

Figure 15.30  Transients in Rotor Angle for Synchronisation Time t = 4 s

and t = 15.45 s.
Tsyn = 4 s
1000

−1000
P1 (W )

42.78 W
−2000

−3000

−4000
0 1 2 3 4 5 6 7 8 9 10
t (s)
Tsyn=15.45 s
200

100
P1 (W )

111.1 W
0
42.78 W

−100
0 2 4 6 8 10 12 14 16 18 20
t (s)

Figure 15.31  Transients in Real Power Output of G-1 for Synchronisation Time t = 4 s
and t = 15.45 s.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 750 9/12/2017 8:54:43 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  751 

1.0601 pu DC Motor Speed

1.065
w (pu)
1.06
1.055 1.0621 pu
1.0588 pu 1.062 pu
1.05
3 4 5 6 7 8 9 10
t(s)
Real Power output
200
69.55 W
Pout (W)

0
42.79 W
−200
3 4 5 6 7 8 9 10
t(s)
Reactive Power output
Qout (VAR)

200
252.4 VAR
0

−200 −252.4 VAR

3 4 5 6 7 8 9 10
t(s)
Delta of G1 wrt G2

0.2
Delta (rad)

−0.2 −0.0312 rad

3 4 5 6 7 8 9 10
t(s)

Figure 15.32  Generator Variables After Varying their Field Voltages at t = 7 s.

4. Since G-1 is over-excited, G-2 is under-excited, and the load is purely resistive, Rs, the G-1 gen-
erates (inductive) reactive power of 252.4 VAR and G-2 absorbs this reactive power (which is a
negative inductive reactive power generation).
The above observations are further verified from the generator terminal variables. Figures 15.33 and
15.34 show the generator terminal voltage and supply current for G-1 and G-2, respectively.
1. For generator-1: From Figure 15.33, we can see that the phase-angle between the voltage va1(t)
(=va2(t) ) and current iag1(t) is lagging and is equal to

f 1 = 0.0042 × 1.062 × 50 × 360° = 80.23°

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 751 9/12/2017 8:54:45 AM

 752  |  Power System Analysis

For G1, R=10 pu, Ffd1=1.1 pu and Efd2=0.9 pu

1
0.8653 pu vag1(t)
0.8 iag1(t)

0.6

0.4 0.1972 pu

0.2
Per phase

−0.2
0.0042 s
−0.4

−0.6

−0.8

−1
9 9.01 9.02 9.03 9.04 9.05
t (s)

Figure 15.33  Generator Terminal Voltage and Current for G-1 After Field Voltage Change.

For G2, R=10 pu, Efd=1.1 pu and Efd2=0.9 pu

1
0.8653 pu
vag2(t)
0.8 iag2(t)

0.6

0.4
0.2016 pu
0.2
Per phase

−0.2
0.0039 s
−0.4

−0.6

−0.8

−1
9 9.01 9.02 9.03 9.04 9.05
t (s)

Figure 15.34  Generator Terminal Voltage and Current for G-2 After Field Voltage Change.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 752 9/12/2017 8:54:45 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  753 

and hence the reactive power output is given by

0.8653 × 0.1972
Qout1 = sin80.23o × 3 × S B = 252.28 VAR
2

2. For generator-2: From Figure 15.34, we can see that the phase-angle between the voltage va2(t)
and current iag2(t) is leading and is equal to

f 2 = −0.0039 × 1.062 × 50 × 360° = −74.55°

and hence the reactive power output is given by

Qout 2 =
0.8653 × 0.2016
2
( )
sin −74.55o × 3 × S B = −252.21 VAR

For this condition, the corresponding field currents are shown in Figure 15.35.

Field current of G1
3
2.6004 pu
If1 (pu)

2.5

2
0 2 4 6 8 10
t (s)
Field current of G2
3
If2 (pu)

2.5 2.1277 pu

2
0 2 4 6 8 10
t (s)

Figure 15.35  Generator Field Currents for G-1 and G-2 When Field Voltages are Changed.

15.2.3.3  Real Power Share Between Generators

In this case, after synchronising the generators, with no field voltage change, at t = 8 s, the resistance
RS is changed to 0.7 pu. Having reached a steady state with this load, G-1 supplies a real power of

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 753 9/12/2017 8:54:47 AM

 754  |  Power System Analysis

648  W and G-2 supplies 673 W. It is to be noted that G-2 supplies slightly more power than G-1
as G-2’s no-load speed is more than (wa2 = 1.0601 pu) that of G-1’s speed (wa1= 1.0588 pu) (see
Figure 15.32) (between 3 and 4 s). The related plots are shown in Figure 15.36 upto 10 s.

DC Armature Current (Ia)

10
6.88 A (M−2) 8.18 A
8
Idc (A)

5.54 A
6

4
6.74 A
7.5 8 8.5 9 9.5 10 10.5 11 11.5 12 12.5 13
t (s)
DC Motor Speed
1.07
1.06 1.0435 pu MG1
MG2
w (pu)

1.05 1.0304 pu
1.04
1.03 1.0621 pu
7.5 8 8.5 9 9.5 10 10.5 11 11.5 12 12.5 13
t (s)
Real power output
1000
673 W
Pout (W)

906 W
500
648 W
0 428.7 W
7.5 8 8.5 9 9.5 10 10.5 11 11.5 12 12.5 13
t (s)
Delta of G1 wrt G2
0.4
Delta (rad)

0.2
0.2263 rad

0
7.5 8 8.5 9 9.5 10 10.5 11 11.5 12 12.5 13
t (s)

Figure 15.36  Generator Plots for a Change of R-load and VDC1 Change.

The following observations are made:

1. Once the load is changed, the speed of the set drops from 1.0621 pu to 1.0304 pu.
2. To meet the new load demand the armature current of DC machines increases to 6.88 A
(DC  motor-2) and 6.74 A (DC motor-1). The corresponding generator currents are shown in
Figures 15.37 and 15.38.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 754 9/12/2017 8:54:48 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  755 

For G1 RL=0.7 pu
1
0.7854 pu vag1(t)
0.8 iag1(t)
0.5504 pu
0.6

0.4

0.2
Per phase

−0.2

−0.4

−0.6

−0.8

−1
9.2 9.21 9.22 9.23 9.24 9.25
t (s)

Figure 15.37  Terminal Voltage and Current for G-1 When RS = 0.7 pu.

For G-1, we have

 0.7854   0.5504 
P1 =    × 3 × S B = 648.42 W
 2  2 

For G-2, we have

 0.7854   0.5715 
P2 =    × 3 × S B = 673.28 W
 2  2 

At t =10 s, the DC voltage applied to motor-1 is changed to 205 V. For the remaining duration,
the generator plots are shown in Figure 15.36. From the figure, the following performance is
noted:

1. Until 10 s, the rotor angle d of G-1 (taking G-2 as reference) is not appreciable. However, after
the VDC1 change and on reaching the new steady-state, the angle is given by d = 0.2263 rad
or 12.97°.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 755 9/12/2017 8:54:49 AM

 756  |  Power System Analysis

For G2, R =0.7 pu

1
0.7854 pu vag2(t)
0.8 iag2(t)
0.5715 pu
0.6

0.4

0.2
Per phase

−0.2

−0.4

−0.6

−0.8

−1
9.2 9.21 9.22 9.23 9.24 9.25
t (s)

Figure 15.38  Terminal Voltage and Current for G-2 When RS = 0.7 pu.

2. Since G-1 is leading G-2, it shares more real power than that of G-2. Thus, Pout1 = 906 W and
Pout2 = 428.7 W. The total real power absorbed by the load can be computed as
2
 0.7895  1
PT =   × 3 × S B = 1335.7 W
 2  0.7

whereas,

PT = Pout1 + Pout 2 = 906 + 428.7 = 1334.7 W

3. The set now runs at a new speed of 1.0435 pu which is higher than the previous speed of
1.0304 pu.
4. The armature current of motor-1 increases to 8.18 A and that of motor-2 decreases to 5.54 A sig-
nifying increase of load on the DC motor-1. The generator terminal voltage and current delivered
are shown in Figures 15.39 and 15.40.
5. Following this VDC1 change, the reactive powers associated with the generators change as shown
in Figure 15.41.
The above observations are verified using an equivalent circuit shown in Figure 15.42. Note that
this equivalent circuit is drawn in the synchronous frame where the speed of the machine is w0 ×
1.0435 rad/s.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 756 9/12/2017 8:54:50 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  757 

For G1
1
0.7895 pu vag1(t)
0.769 pu (Ia)
0.8 iag1(t)

0.6

0.4

0.2
Per Phase

−0.2

−0.4

−0.6

−0.8

−1
12 12.005 12.01 12.015 12.02 12.025 12.03 12.035 12.04 12.045 12.05
t (s)

Figure 15.39  Terminal Voltage and Current for G-1 After VDC1 is Changed (RS = 0.7 pu).

For G2
1
0.7895 pu vag2(t)
0.8 iag2(t)

0.6
0.3718 pu
0.4

0.2
Per phase

−0.2

−0.4

−0.6

−0.8

−1
12 12.005 12.01 12.015 12.02 12.025 12.03 12.035 12.04 12.045 12.05
t (s)

Figure 15.40  Terminal Voltage and Current for G-2 After VDC1 is Changed (RS = 0.7 pu).

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 757 9/12/2017 8:54:52 AM

 758  |  Power System Analysis

Reactive power output

200
QG1
QG2
150

100
Vdc1 = 205 V
50 94.83 VAR
Qout (VAR)

−50
R = 0.7 pu
−100

−150 −94.83 VAR

−200
7.5 8 8.5 9 9.5 10 10.5 11 11.5 12 12.5 13
t (s)

Figure 15.41  The Reactive Powers of the Generators After VDC1 is Changed.

Ra jxd V̄ Ra jxd

+ +
Ē f d 1 R S = 0.7 pu
I¯g 2 Ē f d 2
I¯ g1

Ē f d 1 = 1.0435 12..97o and Ē f d 2 = 1.0435 0 o

R a=0.035pu jxd = j0.423 × 1.0435 = j 0.4414 pu

Figure 15.42  The Equivalent Circuit of the Generators After VDC1 is Changed.

Now, writing KVL to each of the loops and arranging them in matrix form we get,

  I g1  1.0435∠12.97  (15.74)
o
0.735 + j 0.4414 0.7
0.7   =  
 0.735 + j 0.4414   I g 2  1.0435∠0o 


M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 758 9/12/2017 8:54:53 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  759 

Solving for the currents, we get,

I g1 = 0.94306∠ − 4.644o and I g 2 = 0.45397∠ − 23.093o

and

( )
V = I g1 + I g 2 0.7 = 0.96683∠ − 10.615o pu

Now, we can compute the following:

1. The real and reactive power delivered by G-1 is given by

( )* × S B = 906.836 − j94.836 VA
V I g1

2. The real and reactive power delivered by G-2 is given by

( )* × S B = 428.545 + j94.836 VA
V I g2

3. The phase-angle between V and Ig1 is 5.97° leading. The phase-angle between V and Ig2 is 12.47°
lagging. These can be verified from Figures 15.39 and 15.40. This shows that G-1 absorbs reactive
power, whereas G-2 generates reactive power.

4. From the line-to-line voltage (in per-unit), the phase-voltage in per-unit is obtained as

2
v peak = × 0.96683 = 0.78942 pu
3

5. Similarly, the peak value of per-phase current is determined as

2
ig1 peak = × 0.94306 = 0.77 pu
3

and
2
ig 2 peak = × 0.45397 = 0.37066 pu
3
Further, the current in amperes can obtained by simply multiplying these quantities by the base
current, IB.

15.2.3.4 Demonstration of Synchronous Motor Operation of G-2

In this case, synchronous motor operation of G-2 is demonstrated. As in the previous case, two gener-
ators are synchronised at t = 4 s. Once a steady state is reached after the synchronisation (but before

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 759 9/12/2017 8:54:55 AM

 760  |  Power System Analysis

t = 8 s), G-1 supplies 42.78 W and G-2 supplies 69.59 W to the terminal resistor, RS = 10 pu. At t = 8 s,
the DC voltage applied to DC motor-1 is increased to 201.5 V from 200 V. Following this, the variation
of quantities are depicted in Figure 15.43.

DC Armature Current (Ia)

5
MG−1
3.9579 A
Idc (A)

MG−2
4
3.2634 A
3
4 5 6 7 8 9 10 11
t (s)
DC Motor Speed
1.07

1.065 1.0621 pu
w (pu)

1.0661 pu
1.06

1.055
4 5 6 7 8 9 10 11
t (s)
Real power output
200
123.8 W
Pout (W)

100
−10.7 W
0

4 5 6 7 8 9 10 11
t (s)
Delta of G1 wrt G2
0.1
Delta (rad)

0.05

0 0.0539 rad

4 5 6 7 8 9 10 11
t (s)

Figure 15.43  Synchronous Motor Operation of G-2 After VDC1 is Changed (RS = 10 pu).

From Figure 15.43, the following observations are made:

1. The G-1 continues to work as a generator now generating a real power of 123.8 W. Of this,
113.1 W is consumed by RS. It is calculated as
2
 0.8684  1
PRs =   × 3 × S B = 113.1 W
 2  10

Remaining 10.7 W is consumed by G-2, which now works as a synchronous motor.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 760 9/12/2017 8:54:56 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  761 

2. This releases the load on the motor-2 and the speed of the set rises to 1.0661 pu (from the ear-
lier 1.0621 pu). This is denoted by a decrease in its armature current. The real power input to
G-2 is now utilised to overcome the shaft-losses. In other words, the shaft losses are now met by
both DC motor-2 and the reversed G-2 power of 10.7 W, except a minor armature copper losses
of G-2.
3. The rotor angle of G-1 increases to 0.0539 rad = 3.088°.
4. The generator G-1 and synchronous motor terminal voltages and currents are shown in
Figures 15.44 and 15.45, respectively.

M2 as Synchronous motor
1
0.8684 pu vag1(t)
0.8 iag1(t)
0.9517/10 pu
0.6

0.4

0.2
Per phase

−0.2

−0.4

−0.6

−0.8

−1
11 11.01 11.02 11.03 11.04 11.05
t (s)

Figure 15.44  Terminal Voltage and Current of G-1: Synchronous Motor

Operation of G-2.

The machine currents and power factor of operation can be verified by performing a circuit analysis as
shown in the previous case. Changes are made related to the equivalent circuit shown in Figure 15.42
so that the circuit equation appears as follows:

  I g1  1.0661∠3.088 
o
10.035 + j 0.423 × 1.0661 10
10  = 
 10.035 + j 0.423 × 1.0661  I g 2  1.0661∠0o 


M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 761 9/12/2017 8:54:57 AM

 762  |  Power System Analysis

M2 as Synchronous motor
1
0.8684 pu vag2(t)
0.8 iag2(t)

0.6
0.0969/10 pu
0.4

0.2
Per-phase

−0.2

−0.4

−0.6

−0.8

−1
11 11.01 11.02 11.03 11.04 11.05
t (s)

Figure 15.45  Terminal Voltage and Current of the Synchronous Motor.

Solving for the currents we get,

I g1 = 0.116547∠3.37o and I g 2 = 0.011854 ∠ − 147.42o

and

( )
V = I g1 + I g 2 10 = 1.0636∠0.25o pu

Now, we can calculate the following:

1. The real and reactive power (or complex power) delivered by G-1 is given by

( )* × S B = 123.7742 − j6.7413 VA
V I g1

2. The real and reactive power delivered by G-2 is given by

( )* × S B = −10.6536 + j6.7413 VA
V I g2

3. The phase-angle between V and Ig1 is 3.12° leading. The phase-angle between V and Ig2 is 147.67°
lagging. These can be verified from Figures 15.44 and 15.45.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 762 9/12/2017 8:54:59 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  763 

4. From the line-to-line voltage (in per-unit) the phase-voltage in per-unit is obtained as

2
v peak = × 1.0636 = 0.8684 pu
3

5. Similarly, the peak value of per-phase current is determined as

2
ig1 peak = × 0.116547 = 0.09516 pu
3
and

2
ig 2 peak = × 0.011854 = 0.0096785 pu
3

It is to be noted that such an operation of a generator is not permitted as the prime mover is not
designed for a reverse torque. Therefore, in such cases the generator is tripped by employing a
reverse-power relay in practice. However, a DC motor-driven generator does not impose such
constraints.

15.2.3.5 Interfacing of an RL-series Load to the Two Generators

in Parallel
In this case, the two generators are synchronised as explained earlier at t = 4 s. At t = 8 s, a series
connected RL load is connected across the terminals. Here, the interfacing is carried out in the abc-
frame as follows:

1. From the RL-circuit shown in Figure 15.18, for a three-phase balanced condition, we
can write

iabcr = iabcT − iabcL(15.75)

2. In the above expression, iabcT = iabcg1 + iabcg2.

3. The load current iabcL is obtained by solving the load-differential equation,

iaL  iaL   va 
d   R   1  
 ibL  = − l ibL  + L vb 
dt Ll l  
icL  icL  vc 

4. The common terminal voltage is given by

vabc = RSiabcr

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 763 9/12/2017 8:55:04 AM

 764  |  Power System Analysis

In the above expressions, the generator currents and load currents are treated as state variables. For the
x
load, Rl is set to 0.7 pu and xl = 0.698 pu and Ll = l .
wB
At t = 11 s, the field voltage of G-1 is increased to 1.05 pu from the earlier 1.0 pu, keeping the field
excitation of G-2 unaltered. The plots are noted as in Figure 15.46.

DC Armature Current (Ia)

5
Idc (A)

5.065 A (M − 2)
4 4.9252 A

3
8 9 10 11 12 13
t (s)
DC Motor Speed
1.065
MG−1
1.06
w (pu)

1.0491 pu MG−2 1.0483 pu

1.055
1.05
1.045
8 9 10 11 12 13
t (s)
Real power output
800
320.4 W 337.3 W
600
Pout (W)

400
200
294.2 W 308.2 W
0
8 9 10 11 12 13
t (s)
Delta of G1 wrt G2
0 Efd−1 change
Delta (rad)

−0.01
−0.024 rad
−0.02
−0.0127 rad
−0.03
8 9 10 11 12 13
t (s)

Figure 15.46  Plots of Variables for RL-series Load on the Paralleled Generators.

From Figure 15.46, the following observations are made:

1. After synchronisation, under steady state (before 11 s), G-1 supplies P1 + jQ1 = (294.2 + j283.4)
VA and G-2 supplies P2 + jQ2 = (320.4 + j277.7) VA. The total real power supplied = 614.6 W.
The total reactive power supplied = 561.1 VAR.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 764 9/12/2017 8:55:05 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  765 

2. The speed of the set reduces to 1.0491 pu. At this speed, the load impedance is computed as

Zl = 0.7 + j 0.698 × 1.0491 = 0.7 + j 0.73227 pu

Hence, the power factor angle can be obtained as fl = 46.291°.

3. From Figure 15.47, it can be noted that

0.7241 × 0.7146
PL = cos( 46.291o ) × 3 × S B = 536.33 W
2

0.7241 × 0.7146
QL = sin( 46.291o ) × 3 × S B = 561.05 VAR
2

RL load with G−1 and G−2

0.8
va(t)
0.7241 pu
iaL(t)
0.6

0.7146 pu
0.4

0.2
Per-phase

−0.2 0.00245 s
0.01906 s

−0.4

−0.6

−0.8
10.01 10.015 10.02 10.025 10.03 10.035 10.04 10.045 10.05
t (s)

Figure 15.47  Plots of Terminal Voltage and Load Current with RL Load
On the Set.

4. The real power consumed by RS = 10 pu is given by

2
 0.7241 1
PRs =   × 3 × S B = 78.64 W
 2  10

Hence, the total real power is 536.33 + 78.64 = 614.97 W.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 765 9/12/2017 8:55:07 AM

 766  |  Power System Analysis

5. After t =11 s, having reached a steady-state, from Figure 15.48, we can see that the reactive power
supplied by G-1 increases to 352.3 VAR and that supplied by G-2 decreases to 236.4 VAR. This
field voltage change also causes a minor change in speed and real power shared by the generators.

Reactive power output

350

283.4 VAR 352.3 VAR

300

250
Qout (VAR)

277.7 VAR Efd−1 change

200 236.4 VAR

150

100
8 9 10 11 12 13
t (s)

Figure 15.48  Plots of the Reactive Power Shared After Changing the Field
Voltage of G-1.

15.2.4 Synchronisation of a DC Motor-driven Generator

to Mains Supply
In this case, a DC motor-driven generator is synchronised to the mains supply. This operation is quite
different from the parallel operation of two generators, in that the mains is assumed to be of infinite
capacity where the voltage and frequency deviations are negligible. In order to demonstrate this oper-
ation, the interfacing of the finite capacity generator to the mains is done in the abc-frame. The circuit
employed is shown in Figure 15.49.
1. The DC motor is made to run on a 190 V DC supply. At t = 3 s, the generator field is energised
with 1 pu voltage when the generator is on effective open-circuit represented by RS = 10 pu.
2. With the switch SC kept open, the frequency of the generated voltage is 1.011 × 50 = 50.55 Hz
which is close to the mains nominal frequency of 50 Hz, but slightly on the higher side so that
upon synchronisation, the machine starts delivering real power to the mains.
3. For Efd0 = 1 pu, the open-circuit line-to-line voltage is 1.011 × Efd0 which is close to the mains
2
voltage which is also set to 1 pu (line-to-line) or per-phase peak is × 1 pu.
3

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 766 9/12/2017 8:55:08 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  767 

Son
i 0dq1 i abcg1 jx l
Rl
+
SC RO i abcL
vdq1 G1 [P (θ1)] −1
Mains
RS

vabc1 = R S i abcR Before SC is closed i abcR = i abcg1

After SC is closed i abcR = i abcg1− i abcL

[P (θ1)]

Figure 15.49  Circuit Diagram for Synchronising the Generator to the Mains.

4. The phase sequence of the generated voltage is made identical to the mains phase-sequence,
a − b − c.
5. Keeping the switch Son open, at t.= 4 s, switch SC is closed in order to connect the mains network
to the generator terminal through a large series resistance of RO. such that RO + Rl = 10 pu. This
series resistance represents the resistance of the lamps normally used in three dark-lamp method
of synchronisation.
6. At a desired time instant, t = Tsyn, depending on the phase-angle denoted by d, the switch Son is
closed to synchronise the generator to the mains.

Before closing the switch SC :

1. The generator terminal voltage in the abc-frame is given by
vabc1 = RS iabcR

where iabcR = iabcg1,

  and iabcg1 is obtained as

−1
iabcg1 = [ P (q1 ) ] i0 dq1 (15.76)

2. The generator terminal voltage in the 0 dq-frame is given by

v0 dq1 = [ P (q1 ) ] vabc1 (15.77)

3. The angle q1 = wa1t and wa1 (in electrical rad/s) is obtained from (15.49).
4. The network equations are not enabled.

After closure of the switch SC :

1. From Figure 15.49, for a three-phase symmetrical line, we can write

iaL  iaL  va1  vaM 

d   R + RO   1   1  
ibL = − l + −
 bL  L  b1  L vbM  (15.78)
i v
dt   Ll l  l 
icL  icL  vc1  vcM 

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 767 9/12/2017 8:55:09 AM

 768  |  Power System Analysis

where vabcM denotes the mains voltage at frequency w B .

2p

2. In the above expression, the generator terminal voltage is computed as

vabc1 = RS iabcR

with iabcR = (iabcg1 − iabcL), where iabcg1 and iabcL are treated as state variables.

After Synchronisation:

1. At t = Tsyn, the switch Son is closed to short the RO resistor, signifying synchronisation of the
generator to the mains. From this instant onwards, in the above expression, i.e., in (15.78), RO is
x
set to zero. For the line, Rl is set to 0.01 pu, xl = 0.2 pu, and Ll = l .
wB
2. The angle q1 in (15.76) and (15.77) is changed to

q1 = w B t + d 1

with
dd 1
= (w a1 − w B )
dt

3. The current through RS, i.e., iabcR continues to be equal to iabcg1 − iabcL.
Setting Tsyn = 5 s, different plots are obtained as in Figure 15.50.

From Figure 15.50, the following observations are made:

1. Before closure of the switch Son, the speed of the DC motor is 1.0045 pu. However, once the
generator is synchronised and on reaching a steady state, the generator speed is driven to
1.0 pu.
2. Since the no-load speed of the generator is kept slightly higher than 1.0 pu, on synchronisation to
the mains it supplies a real power of 109.11 W at the mains bus.
3. The armature current of the DC-motor increases due to a reduction in its back-emf, thus balanc-
ing the new power demand on the generator.
Under this condition, the generator’s and mains’ terminal quantities are plotted as in Figures 15.51 and
15.52, respectively.
The following changes are made related to the equivalent circuit shown in Figure 15.42 with the
speed of the machine set to 1.0 pu:

  I g1  1.0 ∠6.387  (15.79)

o
10.035 + j 0.423 10
=  
10
 10.01 + j 0.2  − I L  1.0 ∠0o 


M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 768 9/12/2017 8:55:11 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  769 

DC Armature Current (Ia)

6

Idc (A)
4
4.2654 A
2

2 4 6 8 10
t (s)
DC Motor Speed
1.0045 pu
1.02 1.0 pu
w (pu)

1
1.011 pu
0.98
2 4 6 8 10
t (s)
Real power received at Mains

0
Pout (W)

−2000 109.11 W
Tsyn=5 s
−4000
2 4 6 8 10
t (s)
Delta of G1 wrt Mains
−56
−58
Delta (rad)

−60
−62
−56.4372 rad =6.387 deg
−64
−66
−68
2 4 6 8 10
t (s)

Figure 15.50  Synchronisation of DC Motor-driven Generator to Mains

with Tsyn = 5 s.
Solving for the currents, we get,

I g1 = 0.21023∠6.59o and ( − I L ) = 0.11135∠ − 168.63o

and

( )
V = I g1 − I L 10 = 0.99695∠1.26o pu

Further, we can calculate the following:

1. The real and reactive power (or complex power) delivered by G-1 is given by

( )* × S B = 208.682 − j19.469 VA
V I g1

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 769 9/12/2017 8:55:13 AM

 770  |  Power System Analysis

G−1 terminal voltage and current

1
0.858/5=0.1716 pu
0.8

0.6

0.4
0.814 pu
0.2
Per-phase

−0.2

−0.4

−0.6
vg(t)
−0.8 iag(t)

−1
10 10.01 10.02 10.03 10.04 10.05
t (s)

Figure 15.51  G-1 Terminal Quantities for Synchronisation of DC Motor-driven Generator to

Mains with Pmains = 109.11 W.

Mains voltage and current entering

1
0.9088/10 pu
0.8165 pu
0.8

0.6

0.4

0.2
Per-phase

−0.2

−0.4

−0.6
vM(t)
−0.8 iaL(t)

−1
10 10.01 10.02 10.03 10.04 10.05
t (s)

Figure 15.52  Mains Terminal Quantities for Synchronisation of DC Motor-­driven Generator

to Mains with Pmains = 109.11 W.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 770 9/12/2017 8:55:14 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  771 

2. The real and reactive power supplied to the mains is given by

VM ( I L ) × S B = 109.168 − j 21.949 VA
*

3. From the line-to-line voltage (in per-unit), the phase-voltage in per-unit is obtained as

2
v peak = × 0.99695 = 0.814 pu
3

4. Similarly, the peak value of per-phase current is determined as

2
ig1 peak = × 0.21023 = 0.1716 pu
3

and
2
iLpeak = × 0.11135 = 0.090919 pu
3

5. The real power delivered by G-1 is verified as

Losses in RS :
2
 0.814  1
PRs =  × 3 × S B = 99.389 W
 2  10

Losses in Rl:

PRl = 0.111352 × 0.01 × S B = 0.12399 W

Hence, the total real power is 109.168 + 99.389 + 0.12399= 208.68 W.

6. The mains absorbs a negative reactive power. This implies that it delivers reactive power to the
network, of which 0.111352 × 0.2 × SB = 2.4798 VAR is consumed by the line reactance and the
rest is absorbed by G-1.

15.2.4.1 Phase-angle Influence-Synchronising a DC Motor-driven

Generator to Mains Supply
Having satisfied three conditions for synchronising a DC motor-driven generator to mains supply,
the time instant at which the switch Son is closed decides the phase angle between the voltages V
and VM. When Tsyn is set to 5 s, the phase angle denoted by d1 is equal to −67.10°. This is obtained
 180 
by converting −57.72 rad into degrees. Thus, we can see that  −57.72 ×  / 360° gives −9.1864
 p 
revolutions. This, in turn, can be written as −0.1864 × 360°. For this angle, the real power output of

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 771 9/12/2017 8:55:16 AM

 772  |  Power System Analysis

G-1 Pg1, and reactive power received at the mains bus are plotted in Figure 15.53. From the figure, it
can be seen that the transient amplitudes are quite large.

Pg1 and Q entering the mains for Tsyn = 5 s

1500
P
Q
1000
208.62 W

500
Pg1/Qout (W/VAR)

−500 −21.94 VAR

Tsyn = 5 s

−1000

−1500

−2000
3 4 5 6 7 8 9 10 11
t (s)

Figure 15.53  The Pg1 and Qout-Received at the Mains Bus for Tsyn = 5 s.

However, when G-1 is synchronised by setting Tsyn = 5.91 s, the phase-angle turns out to be only 0.61°.
This corresponds to an angle of −56.538 rad. For this phase angle, the powers are noted as shown in
Figure 15.54. From the figure, it is clear that for the same steady-state power levels, the transient peaks
are quite small with this phase-angle.

15.2.4.2 DC Motor Supply Voltage Change - Operation of a DC

Motor-driven Generator on Mains Supply
In this case, after synchronising the generator to the mains, the DC voltage applied to the motor
is increased to 195 V from 190 V at t = 9 s. This tends to increase the speed of the setup above
1.0 pu. However, the speed of the generator is held constant at 1.0 pu corresponding to the mains
supply frequency unlike in the two-generator case discussed earlier. This initiates the following
events:

1. The constancy of speed causes the motor to draw more armature current, and hence the torque
developed by the motor increases to 8.1745 Nm from 4.95 Nm as shown in Figure 15.55.
2. The real power delivered by the generator increases thus raising the real power received at the
mains to 599.2 W.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 772 9/12/2017 8:55:16 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  773 

Pg1 and Q Entering the Mains for tsyn = 5.91 s

350

300
208.62 W

250
Tsyn = 5.91 s

200
Pg1/Qout (W/VAR)

150
P
Q
100

50
−21.94 VAR

−50

−100
3 4 5 6 7 8 9
t (s)

Figure 15.54  The Pg1 and Qout-Received at the Mains Bus for Tsyn = 5.91 s.

3. Since the generator is provided with a constant field voltage, it demands more reactive power
from the mains. Therefore, the reactive power supplied by the mains increases to 191.3 VAR from
the earlier 21.94 VAR. In Figure 15.55, Qout denotes the power absorbed at the mains.
4. Following a change in VDC, the rotor speed oscillates at a frequency of 1.63 Hz. This is generally
termed as swing-mode of the system.

15.2.4.3 Generator Field Voltage Change-Operation of a DC Motor-

driven Generator on Mains Supply
Here, in the previous scenario, after increasing the VDC, the field voltage applied to the generator is
increased from 1.0 pu to 1.1 pu at t = 12 s. Following this change, the response of the set-up is high-
lighted below (see Figures 15.56 and 15.57).
1. G-1 no longer absorbs reactive power. It changes from −112.3 VAR (shows absorption) to
61.2 VAR, denoting reactive power supply. This releases the reactive power demand on the mains.
2. Following a VDC change (in the previous case), the rotor angle of the generator has increased to
25°. Now, after increasing the field voltage the rotor angle decreases to 22.07°.
3. The real power received at the mains slightly decreases from 599.2 W to 594 W.
4. The torque developed by the motor does not change appreciably. This clearly demonstrates that
a change in field voltage does not influence the real power output of the generator significantly;
however, it changes the reactive power shared by the generator dominantly.
5. The speed of the set-up is held fixed due to the constancy of the supply frequency.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 773 9/12/2017 8:55:17 AM

 774  |  Power System Analysis

DC Motor

8
4.95 Nm

Tm (Nm)
6
4 8.1745 Nm
2
2 4 6 8 10 12
t (s)
DC Motor Speed

1.02 1.0 pu
w (pu)

0.98
2 4 6 8 10 12
t (s)
Real power received
1000
109.11 W
Pout (W)

0
599.2 W

−1000
2 4 6 8 10 12
t (s)
Reactive power received
500
−21.94 VAR −191.3 VAR
Qout (VAR)

−500
2 4 6 8 10 12
t (s)

Figure 15.55  The Plots of Motor Torque, Speed, Real and Reactive Power Received at the Mains
Bus for VDC Change.
A similar equivalent, circuit-based analysis is carried out using Figure 15.42 with the speed of the
machine set to 1.0 pu, and the matrix equation is written as

  I g1  1.1∠22.07  (15.80)
o
10.035 + j 0.423 10
=  
10
 10.01 + j 0.2  − I L  1.0 ∠0o 


Solving for the currents, we get,

I g1 = 0.6946∠1.75° and ( − I L ) = 0.5939∠ − 179.10o

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 774 9/12/2017 8:55:20 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  775 

Reactive power received at mains

500
−9.16 VAR
Qout (VAR) −191.3 VAR
0

−500
4 6 8 10 12 14
t (s)
Reactive power output of G−1
200
−112.3 VAR
Qout (VAR)

61.2 VAR
0

−200
4 6 8 10 12 14
t (s)

Delta of G1 wrt G2
−55.5
−56.1635 rad (22.07 deg.)
Delta (rad)

−56

−56.5
−56.112 rad. (25 deg.)
−57
4 6 8 10 12 14
t (s)

Figure 15.56  The Plots of Generator Reactive Power and Rotor Angle and the Reactive Power
Received at the Mains Bus for Efd Change.

and

( )
V = I g1 − I L 10 = 1.0111∠6.75° pu

The real and reactive power supplied to the mains is given by

VM ( I L ) × S B = 593.87 − j 9.31 VA
*

A negative sign associated with the reactive power supplied to the mains denotes that the mains is
absorbing a capacitive power, implying a generation of reactive power.
The per-phase terminal voltages at the generator bus and the mains bus are shown in Figure 15.58.
From the line-to-line voltage (in per-unit), the phase-voltage in per-unit is obtained as

2
v peak = × 1.0111 = 0.82556 pu
3

Also, note the phase angle lead of the generator voltage with respect to the mains voltage in Figure 15.58.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 775 9/12/2017 8:55:22 AM

 776  |  Power System Analysis

DC Motor
8

Tm (Nm)
6
4 8.17 Nm 8.17 Nm
2
4 6 8 10 12 14
t (s)

DC Motor Speed
Efd is changed to 1.1 pu
1.02 1.0 pu
w (pu)

0.98
4 6 8 10 12 14
t (s)
Real power received
1000
Pout (W)

0
599.2 W
594 W
−1000
4 6 8 10 12 14
t (s)

Reactive power received

500
−191.3 VAR
Qout (VAR)

−9.16 VAR
−500
4 6 8 10 12 14
t (s)

Figure 15.57  The Plots of Motor Torque, Speed, Real and Reactive Power Received at the Mains
Bus for Efd Change.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 776 9/12/2017 8:55:23 AM

 Dynamic Modelling of Some Electrical Machines and their Power-flow Analysis  |  777 

Vg and VM per−phase voltage for VDC=195 V and Efd=1.1

1
0.8255 pu (Vg) 0.8165 pu
0.8

0.6

0.4

0.2
Per-phase

−0.2

−0.4

−0.6
Vg
−0.8
Vg leads Vm by 6.7 deg. VM
−1
14 14.01 14.02 14.03 14.04 14.05
t (s)

Figure 15.58  The Per-phase Terminal Voltages at the Generator Bus and the Mains Bus
for Efd Change.

References
[1]  P. Kundur, Power System Stability and Control, McGraw-Hill Inc., NY, 1994.
[2]  T. Ackermann, Wind power in Power Systems, John Wiley and Sons, England, 2005.
[3]  E. W. Kimbark, Power System Stability-Vol-III: Synchronous Machines, John Wiley and Sons,
NY, 1956.
[4]  D. Surendra and K. N. Shubhanga, ‘Development of a power system laboratory supported by real-
time systems’, Int. Conf. on Power and Energy Systems (ICPS), Chennai, December 2011, pp. 1–6.
[5]  R. Krishnan, Electric Motor Drives Modelling, Analysis and Control, Prentice-Hall, NY, 2008.
[6]  Vijay Vittal and Raja Ayyanar, Grid Integration and Dynamic Imapcts of Wind Energy, Sringer,
NY, 2013.
[7]  P. C. Krause, O. Wasynczuk, and S. D. Sudhoff, Analysis of Electrical machinery and Drives
Systems, The IEEE Press, Wiley Interscience, 2002.

Review Questions
1. A 3-phase, 2 MW, 690 V, 50 Hz, 4-pole, ∆-connected induction motor, has the following Y-equivalent
per-phase parameters: Rs = 0.00488 pu, Rr = 0.00549 pu, xls = 0.09241 pu, xlr= 0.09955 pu,
Xm = 3.95279 pu, H = 0.5 s. Assume full-load efficiency of 90% and full-load power factor
= 0.9 lagging. Implement a time-domain simulation model to determine the following:
(a) The maximum value of the inrush current drawn by the machine if the machine is on no-load
with DOL starting.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 777 9/12/2017 8:55:23 AM

 778  |  Power System Analysis

(b)  The no-load current and power factor under steady-state.

(c) The value of constant load torque which makes the machine to draw full-load current under
steady-state.
(d)  The full-load slip and full-load power factor.
2. Develop the fundamental frequency steady-state model for the above induction machine and
perform the power flow analysis at full-load.
3. For a DC motor data given in section 15.2.1.1, develop a steady-state model using the differential
equations and determine the no-load speed and armature current.
4. Consider a DC motor-driven synchronous generator, which is delivering a star-connected resistive
load with RL = 0.7 pu. In the time-domain simulation, it is required to reach the final steady state
with the load, without sequentially switching different sources. To achieve this, determine the
initial value of different state variables for VDC = 200 V and Efd = 1.0 pu so that these values can
be entered in the respective integrator blocks before running the simulation.
5. In a case study presented in section 15.2.2.1, where a DC motor-driven synchronous generator
delivers a pure resistive load, perform a time-domain simulation employing a single time constant
excitation controller to regulate the terminal voltage at rated value. Use an appropriate gain for the
controller.
6. In section 15.2.4, a DC motor-driven alternator is connected to main supply. The steady-state
speed of the setup is 1.0 pu. Perform the following tasks:
(a) Carry out the power flow analysis of the DC motor and calculate the mechanical power input
to the generator in watts.
(b)  Using the above power value, determine the rotor-angle of the alternator.
Note that VDC = 190 V and Efd0= 1.0 pu.
7. In the above example (given in section 15.2.4), a DC motor-driven alternator is connected to main
supply at 1.0 pu terminal voltage. Repeat this time-domain simulation where the alternator is
synchronized to the mains at a reduced terminal voltage say, 0.75 pu. Determine the steady-state
power angle of the machine for VDC = 190 V. Perform the power flow analysis.

M15_POWER-SYSTEM-AN00_SE_XXXX_CH15.indd 778 9/12/2017 8:55:23 AM

 Index

∆ connection 75 basic parameters 226

0.0 Model 521 brushless excitation system 481
0dq-frame 134 bus frequency deviation 683
1.0 Model 524
1.1 Model 209, 218, 533 C
2.2 Model 221, 230, 251, 528 calculation of bus voltages 361
50-Machine 690 capability curves 95
ceiling voltage 478
A centre of inertia 642
characteristics of loads 103
Abc-frame 134, 239 choice of frequency 23
ac1a Excitation system 482 circuit-1 46, 57
ac4a Excitation system 483 circuit-2 47, 57
aggregate models 79 circuit-3 50, 57
agnetic axis 149 clarke’s reference 119
air-gap torque 712 classical model 254, 507
algebraic loop 639 classification of power system stability 2
angle across line 677 cOI-reference-frame 572
angle reference 123 common-mode 566
angular frequency 24 common-mode frequency 643
approximate differentiator 490 complex power 54, 55, 74
armature current 729 composite loads 685
asynchronous link 101 computational issues 4
augmented YBUS 655 constant power-type load 652
controllers 476
B convergence property 341
cosine-based reference 26
back-emf 729 critical clearing-angle 324
back-slash function 359 critical energy 327, 330
backward euler method 299 critically stable system 673
balanced systems 74 critically unstable 281
ball in a valley example 507 critically unstable system 675
base values 76, 208 current-injection approach 580

Z02_POWER-SYSTEM-AN00_SE_XXXX_INDEX.indd 779 9/16/2017 10:40:40 AM

 780 | Index

D F
Damping ratio 187 Fast-acting and high-gain static exciter 546
Damping torque 513 Fast-acting differential equations 653
Dc1a excitation system 480 Fast mode 194
Dc-motor 164, 183, 283 Fault analysis 350
Dc shunt motor 728 Fault current 367
Deactivated source 30 Fault-on trajectory 331
Decrement factor 173, 626 Field forcing 478
Differential-algebraic equations 5 Field voltage 164, 684, 749
Differential algebraic equations (DAEs) 578 First-swing stability 252, 478
Differential equations 288 First-swing transient stability 277
Direct-on-line (DOL) 704 Fixed-point iteration method 335
Distribution systems 101 Flat-line 98
Dormand-prince method 308 Flexible ac transmission systems 100
Double line-to-ground fault 406 Flux-linkage 209, 215
Doubly-fed induction generator (DFIG) 724 Flux linkage-current 153
Dq-frame 138 Forward euler method 293
Dummy-coil voltage 636 Four-machine 681
Dynamic equivalent circuits 253 Four-machine modified power system 612
Dynamic saliency 252, 633 Four-wire and three-phase 67
Free-run of the induction motor 704
E Frequency-dependent 106
Frequency-dependent load 651
Eigenvalue 192 Frequency instability 660
Eigenvalue analysis 185, 259, 430 Frequency regulation 53
Eigenvalue analysis with pss 548 Frequency sensitivity coefficient 106
Eigenvalues 607 Frequency stability 2
Eigenvalues and eigenvectors 5 Frequency stability analysis 642
Eigenvector 192 Fundamental frequency component 175,
Electromagnetic torque 19 185, 202
Electromagnetic transient 6 Fundamental frequency dynamic model 702
Electromagnetic transient analysis 76, 288
Electromechanical energy conversions 9
G
Electromechanical oscillations 3, 273, 557
Electromechanical transient analysis 289 Gauss seidel iteration method 341
Energy-function 288 Generation convention 34
Energy function method 326 Generator 23, 92
Equal-area criteria 324 Generator convention 15, 56, 58, 152
Equivalent circuit 85, 132, 214, 217, 241, Generator equations 582
709, 758 Generator source current 637
Equivalent field current 272 Generator winding current 637
Equivalent impedance 32, 41, 48, 50 Gep 516
Exciter-modes 608 Grid frequency 105
Explicit methods 291 Ground current 379

Z02_POWER-SYSTEM-AN00_SE_XXXX_INDEX.indd 780 9/16/2017 10:40:40 AM

 Index  |  781 

H Linear discrete-time system 293

Linearisation 445, 578, 595
Harmonic oscillator 130 Linear systems 2, 185
Hvdc link 100 Linear time-invariant (LTI) 7, 25, 289
Hydraulic turbines 495 Line-to-line fault 403
Hydro-turbine 95 Line tripping 656
Loadability of lines 98
I Load convention 34, 56, 58
Load currents 596
Ieee first benchmark system 439 Load equivalent circuit 652
Impedance base 77 Load-flow problem 348
Impedance diagram 352 Load frequency control (LFC) 493
Implicit methods 292 Load-generation balance 104
Inductance matrix 145 Load modelling 648
Induction generator effects 472 Load share 107
Induction generator operation 725 Local mode 559
Induction motor 702 Local swing-mode 620
Inertia constant 233 Loss-less 164
Initial condition calculation 255, 314, 464, 490 Lossless line 98
Initial condition perturbation 523 Lossless system 63
Instantaneous power 13, 33, 36 Loss-of-synchronism 283, 510, 558, 692
Inter-area mode 560
Interconnected generators 632
Interconnected operation 102 M
Interconnected sources 60, 62 Magnetic axis 145
Interconnection of two sources 65 Magnetic energy 716
Magnetic field energy 10
J Mains bus 775
Mains supply 766
Jacobian 343 Mechanical equation 184
Mechanical input perturbation 523
K Mechanical power 18
Mechanical torque 18
Kinetic energy 18, 233, 327 Modal analysis 187
Kron’s reference-frame 119 Modal frequencies 439
Modal-speed calculations 467
Mode identification 432
L Modelling issues 5
Large disturbance 670 Modelling of dynamical systems 6
Large power systems 1 Mode-of-instability (MOI) 683
Lead compensator 516 Mode shape 189
Left-eigenvector 186 Motor convention 14, 146
Linear combination 30 Mutual-inductance 145, 148

Z02_POWER-SYSTEM-AN00_SE_XXXX_INDEX.indd 781 9/16/2017 10:40:40 AM

 782 | Index

N Phasor analysis 24, 31, 41

Phasor representation 26, 33
Natural frequency 422 Positive peak envelope 179
Natural response 167, 190, 205 Positive-sequence 85
Negative-sequence network 374 Positive-sequence network 374
Neper frequency 30, 185 Potential energy 327
Network devices 99 Power angle 73
Neutral wire 69 Power-angle curve 324
Newton-raphson iteration method 341 Power balance 57, 60, 668, 715
Nominal frequency 104 Power-flow equations 348
Nominal tap-ratio 83 Power-invariant 120, 153
Non-linear algebraic equations 334 Power-ramping exercises 535
Non-linear systems 2 Power system oscillations 558
Non-minimum phase system 497 Power system stabiliser 621
Non-salient pole 93, 144 Power-variant 126
Numerical integration algorithms 290 Preventive control 619
Numerical iterative techniques 334 Primemover 92, 95
Numerical methods 289 Prime mover controllers 492
Primemover system 273
O Primitive machine 14
Primitive parameters 163
Off-line analysis 289 PSS block 519
Off-nominal voltage ratings 89 PSS location 515
Off-nominal tap-ratio 91 Pure resistive load 732
Online simulations 289 Pure resistor circuit 33
Open-circuit time constant 163, 224, 228
Operational admittance 183
Operational inductance 202 R
Ordinary differential equations (ODE) 288 Radial lines 102
Radian frequency 30, 187
RC-series circuit 40, 55, 740
P
Reactive power 9, 37, 46, 477
Park-inverse transformation 732 Reactive power loss 53
Park transformation 115, 125, 153, 374 Real and reactive powers 62, 70, 104
Participation factors 433, 532 Real or active power 34
Participation matrix 602 Real power 37, 46, 477
Partitioned solution 647 Real power share 753
Passivation 355 Real-time simulations 289
Performance of real power load 689 Reduced order mode 252, 724
Per-unit 209 Reduced system matrix 606
Per-unit representation 76, 704 Redundancy of a state 564
Phase-angle difference 24, 748 Relative-angular stability 506
Phased-locked loop 129 Remedial action scheme (RAS) 692
Phase-plane trajectories 318 Resistance matrix 153
Phase-sequence 66 Reverse-power relay 763
Phasor 122, 219 Revolving field 66, 155, 171, 245

Z02_POWER-SYSTEM-AN00_SE_XXXX_INDEX.indd 782 9/16/2017 10:40:40 AM

 Index  |  783 

Right-eigenvector 186 Stability analysis 2, 126

RL series circuit 34, 54, 177, 735 Stability of numerical integration methods 293
Rotor circuits 210, 216, 226, 229 Stable torsional mode 428
Rotor coils 150 Standard parameters 163, 178, 444
Rotor-frame-of-reference 239 Standstill frequency response (SSFR) tests 219,
Rotor position 145 255
Rotor reference-frame 118 State-space form 289
Rotor-reference frame 153 State-space model 7, 210, 216
Round rotor 240, 243, 248 State variable 7, 30
Runge-kutta (RK) fourth order method 304 State-vector 30
Static exciter 514
Stationary reference-frame 119
S Stator coils 147
Salient pole 93, 146, 240, 242, 250 Stator transient 194, 200, 260, 267, 531
Self-inductance 145, 147 Steady-state operation 237
Sequence impedances 379 Steady-state short-circuit 164
Sequence networks 394 Steam-turbine 95
Series RL circuit 29 Stiff system 193
Short-circuit 167, 183, 200, 259, 267 Sub-synchronous resonance (SSR) 97,
Short-circuit analysis 351 419, 437
Short circuit capacity 371 Subtransient 224, 228
Short-circuit time constant 173, 224, 228 Subtransient internal voltage 231
Shunt-fault currents 409 Surge impedance loading 98
Simultaneous implicit solution 647 Swing-equation 273
Simultaneous non-linear equation 340 Swing equation 232, 236, 645, 584
Sine-based reference 26 Swing mode 435, 511
Single-line diagrams 75 Swing-mode oscillations 559
Single line-to-ground fault 397 Symbolic math 156
Single-phase circuits 33, 57 Symmetrical component 374, 379
Single-time constant static exciter 449 Symmetrical fault calculations 368
Sinor steady-state 32 Symmetrical three-phase 353
Sinor waveform 22 Synchronisation 272, 743
Sinusoidal steady-state 20 Synchronisation transients 748
Slip-signal pss design 622 Synchronising torque 513
Slow mode 194 Synchronising torque coefficient 510
Small-signal stability 3, 106, 506, 524 Synchronous generators 68, 76, 93, 164
Smib system 253, 275, 326, 507 Synchronous grid 103
Solution of swing equation 311 Synchronous inductances 156
Source voltage 84, 88 Synchronous machines 6, 143, 202, 384
Space-phasor 122, 150, 151, 244 Synchronous motor operation 759
Speed-governing systems 498 Synchronous operation 61, 73
Speed governor 104, 663 Synchronous reactance 478
Speed instability 3 Synchronous reference-frame (SRF) 119, 703
Spring-mass system 510, 561 System frequency 616, 656
St1a excitation system 485 System function 6
St2a excitation system 487 System protection scheme (SPS) 692

Z02_POWER-SYSTEM-AN00_SE_XXXX_INDEX.indd 783 9/16/2017 10:40:40 AM

 784 | Index

T Two generators in parallel 763

Two-machine power system 605, 673
Taylor series 290 Two-mass mechanical system 419
The IEEE 14-bus power systems 99 Types of excitation systems 477
Thevenin equivalents 375
Thevenin impedance 356
Thevenin’s theorem 355 U
Three dark-lamp method 767
Unbalanced system 374
Three-phase systems 64, 70
Unloaded generator 171, 178, 224
Three-wire power systems 67
Unreduced system matrix 606
Thyristor-controlled series capacitor 100, 474
Unsymmetrical faults 373
Tie-lines 99
Time-constant 30, 216, 223, 224
Time-domain analysis 32 V
Time-domain simulation 255, 265, 277, 461
Time-invariant systems 7 Variable-step methods 308
Time-varying algebraic equation 635 Variation of inductances 16
Time-varying parameters 1 Voltage build-up 162, 164, 220
Time-varying systems 7 Voltage dependency 106
Time-varying transformation 114 Voltage equation 152, 158, 209
Torque 161, 235, 242, 512 Voltage instability 3
Torques deviation phasors 514 Voltage matrices 593
Torsional frequency 422
Torsional mode 434
Transformers 80, 89, 390, 641 W
Transient 216 Wavelength 98
Transient inductance 176 Windup and non-windup type limiters 500
Transient internal voltage 219
Transient power-angle 255
Transient power angle curve 509 Y
Transient reactance 720
Transient stability 275, 507 Y connection 74
Transient stability analysis 3, 410, 632
Transmission line 132, 385, 640
Transmission systems 97
Z
Trapezoidal method 303 zero-eigenvalue 607
Triplen currents 379 Zero-sequence impedance 386
Triplen harmonics 65 Zero-sequence network 375, 392

Z02_POWER-SYSTEM-AN00_SE_XXXX_INDEX.indd 784 9/16/2017 10:40:40 AM