Vdoc - Pub Power System Analysis A Dynamic Perspective
Vdoc - Pub Power System Analysis A Dynamic Perspective
K. N. Shubhanga
Department of Electrical Engineering
National Institute of Technology Karnataka
Karnataka, India
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]
Foreword xvii
Preface xix
About the Author xxiii
Index 779
Foreword xvii
Preface xix
About the Author xxiii
Index 779
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
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.
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.
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
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.
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.
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. 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
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.
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.
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.
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.
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.
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)
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 )
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 )
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 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
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.
x F
i
− v +
dy
v = Ri + (1.18)
dt
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
dx 1 2 dL
F = i
dt 2 dt
1 2 dL dx
= i . (1.23)
2 dx dt
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)
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
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)
di 2 dL(q )
vi = Ri 2 + L(q )i +i (1.27)
dt dt
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,
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
or
v = Ri + e(1.35)
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
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
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,
1.4
L11(θ) = Ls − Lm cos (2θ)
1.2
0.8
(pu) 0.6
0.4
0
L12(θ) = L21(θ) = −M sin(θ)
−0.2
−0.4
−0.6
0 1 2 3 4 5 6 2π 7
(rad)
1 T
Wmg = i [ L (q )] i (1.44)
2
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
dWmg d i 1 T d [ L (q )]
= i T [ L (q )] + i i (1.47)
dt dt 2 dt
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
and
d i d [ L(q )]
v = −[ R] i − [ L(q )] − i (1.55)
dt dt
di d[ L(q )]
p = i T v = − i T [ R] i − i T [ L(q )] − iT i (1.56)
dt dt
d i dWmg 1 T d [ L (q )]
i T [ L (q )] = − i i (1.57)
dt dt 2 dt
dWmg 1 T d [ L(q )]
p = i T v = − i T [ R] i − − i i (1.58)
dt 2 dt
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
However, under sinusoidal steady-state, for a three-phase generator, expression (1.59) simplifies to
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
uF iF = iF2RF
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.
[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.
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. 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.
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.
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.
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
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.
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
V = Ve j 0 = V ∠0°
p
−j
I = Ie 4 = I ∠ − 45° (2.6)
Vm Im
where, V = and I = .
2 2
v1 (t ) = Vm cos(w t )
v2 (t ) = Vm sin(w t )
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.
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.
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.
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
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).
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.
i(t)
+
+ vR
R
−
v(t)
+
_ L vL
−
R = 1.4142 Ω and L = 4.5016 mH
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
−
v = vR + v L (2.9)
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)
R R
− t Vm − t
itr (t ) = i (0)e L −
|Z|
[ sin(a − f ) ] e L
Vm
iss (t ) = sin(w t + a − f )
|Z|
Vm d V
v (t ) = R sin(w t + a − f ) + L m sin(w t + a − f ) (2.12)
|Z| dt | Z |
or
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)
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)
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 |
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
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.
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)
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)
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.
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 )
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
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
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 )
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
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 )
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.
−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
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)
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 )
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
0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Time (s )
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
V
Since we know that I = , we get,
|Z|
VR = IR and VL = IX L
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).
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
i(t)
+
+ vR
R
−
v(t)
+
_ C vC
−
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 )
1 V
I = V−
R (1 + jw RC )
jw CV
= (2.39)
(1 + jw CR)
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
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
VR
VC
f
o
V=V 0 (90 − f )
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 )
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 )
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.
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
−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
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 )
+ L − V
+ +
f
R Vq
−
Is
Vp
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.
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.
v(t) v L (t)
v1(t) i s (t)
+ + L − + R L2 +
vR(t) v2(t)
I
− II −
i 2(t)
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
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)
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 )
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
VL2
P= = 0.3299 W
R
The reactive power associated with the inductors L and L2 (with load convention) is obtained as
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.
v(t) vL (t)
v1 (t) iS (t)
+ L −
+ +
+
R vC(t)
− C
−
i (t) i (t)
R
C
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
is (t ) = iR (t ) + iC (t ) (2.61)
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 )
− 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
IC = = 0.3922∠56.31° A
− jX C
Similarly,
V1 = I s ( jwL) = 0.4385∠82.88° V
90 90
1 1
120 60 120 60
V1 Ic
Vq
V V
180 −Is 0 180 0
Is Is
VL
210 330 210 330
VL2
P= = 0.6154 W
R
5. The reactive power associated with the inductor, L (with load convention), is obtained as
and the reactive power associated with the capacitor C, (with load convention), is given by
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
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.
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
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.
V Is VL
o
+ − V=V 0 (90 −f )
+ L +
V1 R Vq
−
Is
Vp
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 .
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
I
+
Vp
+ VR
R
−
I
V
+ Vq
_ C VC
o
− V=V 0 (90− f )
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)
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)
Table 2.1 Generator Convention for Source-element in RL-series and RC-series Circuits.
Table 2.2 Load Convention for Load-element in RL-series and RC-series Circuits.
=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
=0 =−0.1752 =0 =0.1752
=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.
+ +
Element Element
Ve Ve
− −
=0 = −0.3846 =0 = 0.3846
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 )
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)
V2
= e ∠f (2.74)
| Z |
2
1 I
+ a xL = 0.5 a +
1 2
V1 V2
− −
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
= 0.6465∠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.
R 0.0418 0
I 2R
xL I 2 ( jx L ) 0 0.2090
Total ∑ SL = 0 ∑ PL = 0 ∑ QL = 0
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.
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.
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
Table 2.7 Load Convention-based Power Calculation for Two Source Interconnection-bus Voltage
Magnitude Modified.
R 0.0418 0
I 2R
xL I 2 ( jx L ) 0 0.2090
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
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
V12 VV
P1 + jQ1 = j − j 1 2 (cosd + j sin d ) (2.79)
XL XL
V V
P1 = 1 2 sind (2.80)
XL
and
(V12 − V1V2 cos d )
Q1 = (2.81)
XL
V V 0.70712
P1 = 1 2 sin d = sin( −25o ) = −0.4226 W
L
X 0 . 5
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:
2p 2p
vbn = Vm1sin w t − + Vm3sin 3 w t −
3 3
Now obtain the line-to-line voltage, vab as
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.
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
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.
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 N = I a + I b + I c (2.86)
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
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
I N = Ia + Ib + Ic
(
= 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
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.
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.
v1a (t ) = Vm sin(w t + fv )
ia (t ) = I m sin(w t + fi )
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
and
Vm I m
pb (t ) = [cos(f v − fi ) − cos(2(w t − 120°) + fv + fi )] (2.90)
2
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.
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
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 )
R 3 I a2 R 0.1254 0
xL 3 I a2 ( jx L ) 0 0.6270
V V
P3f = 3 1 2 sind (2.95)
XL
and
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
*
S3f = 3V ph I ph
= 3 3V ph ∠fv ( I ph ∠fi )*
*
S3f = 3V ph I ph
= 3 3V ph ∠f v ( I ph ∠fi )*
= 3V ph ∠f v ( 3 I ph ∠fi )*
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
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].
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
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)
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
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
or
V1 puV2 pu
Ppu = sind (2.105)
X Lpu
and
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.
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
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:
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
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
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.
Section−1 Section−3
Section−2
3 4 Load
1 2
G T T SL
1 2
Y Y Y
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.
Table 2.11 Per Unit Impedances on SBN = 100 MVA and VB1 = 22 kV in Section-1.
On the chosen power base, the per-unit value of the complex load is given by
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,
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.
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
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
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
(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
and
3.64 × I B3 3.64 × 3.0303
= = 6368.34 A
3 3
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.
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
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
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
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.
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
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].
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
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).
VRp
V 22
Bp
22
a= = = 1.1111
VRs
N 360
V 400
Bs
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,
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.
Armature b’ Rotor
coils c
N N
+
+
+
a a’ +
+
S S
c’ Field
Stator b coils
Stator
Salient pole construction Non−salient pole construction
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].
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
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.
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
Cooling water
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.
Head
Water
Reservoir
Generator
Penstock Valve
control Generator hall
Pelton turbine
with buckets
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.
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.
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.
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.
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].
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.
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.
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
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.
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.
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.
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.
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.
∆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.
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.
∆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
(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.
[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.
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?
+ ia1 ia2 +
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
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.
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
Magnetic axis of aˇ -a
(Ref-axis)
θ q-axis
d-axis c
bˇ
iF
a
ω
aˇ
cˇ b
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.
)
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,
= fa [cos q1 + jsin q1] + fb [cos (120 + q1) + j sin (120 + q1)]
+ fc [cos (120 − q1) - j sin (120 − q1)]
2p 2p
f d = k d f a cos q + f b cos q − + f c cos q +
3 3
f0 fa
f d = [ P ] f b (3.5)
fq f c
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
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
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.
α
Magnetic axis a−a' coil
D ω ot (Stationary frame)
q
d δ
θ
Q
D−Q: Synchronous-frame
d−q : Rotor-frame
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
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
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,
p= ∑ vi = ∑ vi (3.12)
abc 0 dq
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
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
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)
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
( )
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.
q
d
δ Q
θ
ω ot
β
D−Q: Synchronous-frame
d−q : Rotor-frame
q
d
γ δ Q
θ
ωot
β
D−Q: Synchronous-frame
d−q : Rotor-frame
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
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.
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
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.
{[ 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 )
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
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.
PLL controller
Transformation Module
i xcw
La
PI A
iLb αβ DQ Oscillator
abc αβ iD − Controller
iLc iD error +
+ + x
sw
0 ω
The different sections in synchronous reference fame PLL (SRF-PLL) are briefly explained 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
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:
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.
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
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
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 )
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
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)
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
dia
va = Ria + L + vCa + ea (3.35)
dt
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
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,
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.
vCabc = [ P ] vC0 dq
T
eabc = [P]Te0dq(3.40)
[ P ]T v0 dq = [ R][ P ]T i0 dq + [ L]
d
dt
( )
[ P ]T i0 dq + [ P ]T vC0 dq + [ P ]T e0 dq
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
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)
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
Combining the first two terms on the RHS, we get the desired equation in the 0dq-frame as
d 1
vC = iabc
dt abc C
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
0 0 0
d
( )
[ P ] [ P ] = 0 0
dt
T
w
0 −w 0
0 0 0
d 1
(vC0 dq ) = − 0 0 w vC0 dq + i0 dq (3.49)
dt C
0 −w 0
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
dvCq 1
= iq + w vCd (3.51)
dt C
(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)
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
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:
(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
f 0 dq = [ PD ] f 0 DQ
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
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
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.
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.
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
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
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.
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.
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.
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.
(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
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)
y cc ′
Lcc = = Ls + Lm cos 2(q + 2p /3)
ic
−p
0+
3 p
q = =−
2 6
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,
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.
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.
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,
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.
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.
ω
F
a' vF
b'
ib
'
vbb
b
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)),
q q
d δ d
θ δ
Q Q
θ ωo t
since it is customary to take q (or Q) axis as the reference for angle measurement. Further, note that
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)
• 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.
[ P ] [ 0 ] [ L11 (q )] [ L12 (q )] [ P ]T [ 0 ]
= T [ L (q )] [ L (q )]
[0] [ I ] 21 22 [0]T [ I ]
or
[ P ][ L11 (q )][ P ]T [ P ][ L12 (q )]
[ LB ]( 7× 7) =
[ L21 (q )][ P ]T [ L22 (q )]
Notes:
• [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
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
y d Ld M dF M dH id
y F = M dF LF LFH iF (4.14)
y H M dH LFH LH iH
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.
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
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
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
0
y
q
−y d
dy B
v B = −[ R] i B − 0 w − (4.23)
dt
0
0
0
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.
vq
d
v
K q-axis
d-axis
G
H
iF
F
ω
vF
i = [ B]T i B
(4.32)
and from the expression for [ LB ] = [ B][ L(q )][ B]T , we get [ L(q )] as
Teg = −
P
2
1
2
i TB [ B]
d
dq
(
[ B]T [ LB ][ B] ) [ B]T i B
(4.34)
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
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)
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)
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
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
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)
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
Time (s)
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
0 = − Raid − w 0y q (4.60)
0 = − Raiq + w 0y d (4.61)
v F = RF iF (4.62)
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
2
ia (t ) = id cosq (4.72)
3
2 M dF iF0
ia (t ) = − cosq (4.73)
3 Ld
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.
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
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
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
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
x (t ) = e At x (0)
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
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
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
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
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.
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
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
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
−500
id (A )
−1000
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).
(iaP + − iaP − )
iacP = (4.100)
2
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 )
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
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).
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
1000
i (t)
a
500
ia and iacP+−
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
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
1 Vm 1 150.8279
Ld = = = 4.966 mH (4.103)
w 0 I sm w 0 80.559
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
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)
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
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
M dF iF0 1 1 1 Td′
id ( s) = − − M dF iF0 ′ −
Ld s Ld Ld (1 + sTd′ )
M dF iF0 1 1 1 sTd′
=− + ′ − (4.110)
s Ld Ld Ld (1 + sTd′ )
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
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.
(
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
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.
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)
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
1 1 1−z 2
= T
s 2p z
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%.
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
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 )
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.
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.
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
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
Consider
−49 A21 B21
G21 = = +
( p − p1 )( p − p2 ) ( p − p1 ) ( p − p2 )
where,
Consider
p +1 A22 B22
G22 = = +
( p − p1 )( p − p2 ) ( p − p1 ) ( p − p2 )
where,
Thus, we have,
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
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.
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
x = Ax
[V ] y = A[V ] y
y = [V ]−1 A[V ] y
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
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.
• 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
−49
x1 = − x1 + 2 x
99 1
x1 = −1.9898 x1
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.
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:
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.
[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
Rotor ra
Stator
sA
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
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.
i1
Vs = 1.0 V + vc
L2 = 1 H C = 100 µF
i2
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
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.
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
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
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
dy F v Fo
v F = iF RF + with iFo =
dt RF
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
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.
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
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
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
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
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
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
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
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.
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
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
M2
∆y d = Ld − dF ∆id = LÄd ∆id (5.43)
LF
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.
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
dy d
vd = −id Ra − wy q −
dt
dy q
vq = −iq Ra + wy d −
dt
vd i R wy q 1 dy d
=− d a − −
VB I B Z B w By B y Bw B dt
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
Ä
xÄ d = L d
xq = Lq
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
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
y d ( s) G ( s)
id ( s) = − E fd ( s) (5.48)
X d ( s) X d ( s)
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
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:
E fd ( s)
x x fl
d
=−
x fl (1 + sTdÄ )
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.
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.
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).
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′
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
id = id1 + id 2 (5.62)
where
id1 = id |y ′ =0 and id 2 = id |y d = 0
F
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 (5.63)
id1 =
xÄd
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
y q Lq M qG iq
y = M LG i
G qG G
Also,
dy G
0 = − RG iG − (5.67)
dt
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.
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
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
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
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
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,
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.
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.
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.
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)
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.
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:
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 )
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
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:
y d ( s) G ( s)
id ( s) = − E fd ( s) (5.104)
X d ( s) X d ( s)
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.
N ( s)
= (5.108)
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 )
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).
• 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.
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
TdÄ
Cf = (5.118)
x fl
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
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)
y q ( s) = X q ( s)iq ( s) (5.129)
where
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)
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).
iq(s) i′G(s)
xkl xgl
i′K(s)
yq(s) xq
1
1
sCg
sCk y ′K(s) y ′G(s)
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
iq = F2 (y q ,y ÄG ,y ÄK ) (5.143)
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
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
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:
( 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)
Notes:
• It is also noted that in general the following relationship is true among the reactances.
• 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)
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
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
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
w = w r (5.170)
P
2
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)
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)
We know that
Teg
Teg =
SB (5.185)
w rB
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)
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
In per-unit,
Pm T w w
= Pm = m r = Tm
SB TrB w rB wB
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.
dd w w
Setting = 0, we have = o =1
dt wB wB
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)
y q0 = xq iq0 (5.206)
Substituting eqs (5.205) and (5.206) in eqs (5.198) and (5.199), 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 )
I 0 = I aq + I ad (5.213)
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 + jxs I ad + I aq (5.215)
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.
Equation (5.209) is re-written by subtracting and adding the term xq id0 to the RHS,
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
1. Obtain Eq0 as
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 = E∠0
Vt∠q
jxL
Generator Peg
vd 0 = − xq iq0 (5.224)
vq0 = xd id 0 + E fd 0 (5.225)
− 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
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
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
fave = Bavet p L ,
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
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.
d−axis
lf
E
q−axis
Fag1
Tia
2
−p p
0 2 2
0 p 2p
−Tia b
2
Rotor surface
Stator surface
If phase-a winding is spatially spread over P poles, the MMF expression becomes
4Tia
Fag1 = cos( b ) (5.236)
pP
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.)
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
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
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
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.
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
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
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
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)
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
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
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
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.
2.1
xq″ = xq′ and Tqo
′′ ≠ 0 xq′ = xd″
1.0
xd″ = xd′ and Tdo
″
≠0 xq = xd′
′ ≠0
Tqo
′′ ≠ 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′ .
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.
vq = −iq Ra +y d (5.257)
Using eqs (5.157) and (5.158) in eqs (5.256) and (5.257), we have
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
Ra jX″
I
+
(E″q + j E″d)∠d Vt∠q
Figure 5.24 Dynamic Equivalent Circuit of the Machine Neglecting Dynamic Saliency.
( )
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,
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.
The expression for torque developed by the machine can be written from eq. (5.196) as
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)
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.
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
d 0 = ∠Eq0 (5.274)
2. Compute
iq0 + jid 0 = I 0 e − jd 0
= I 0 ∠(f 0 − d 0 )
vq0 + jvd 0 = Vt 0 e − jd 0
= Vt0 ∠(q 0 − d 0 )
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)
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
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
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)
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.
′ = 4.3 s
Tdo ″
Tdo = 0.032s
′ = 0.85 s
Tqo ′′ = 0.05 s
Tqo
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
pfg 0.8346 (lag) 0.9285 (lag) 1.0000 0.9748 (lead) 0.7507 (lead)
(Continued)
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
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′
RL 0 0
vabc = 0 RL 0 iabc
0 0 RL
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.
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)
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
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)
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″
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)
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
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).
−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)
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)
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)
T
where y dq = y dy q
T
and y r = [y F y H y G y K ]
∆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
where
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.
Ra = 0.01 Ra = 0
-38.5059 -38.5459
-21.5886 -21.5948
-9.2898 -9.3160
-2.5011 -2.4999
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)
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.
−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).
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.
−2
−4
−6
id
−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).
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′ )
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.
0.3
with stator transients
0.25
0.2
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)
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)
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)
Generator Pg
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
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
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.
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.
VS
Figure 5.44 A Single-time Constant Static Exciter Used with SMIB System
for a Transient Stability Run.
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.
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
1.05
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.
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.
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
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
Figure 5.50 Plot of ia(t), va(t), and Vt with AVR for Clearing Time Equal to 0.252 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.
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.
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
(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:
Generator Sb = Pb + jQb
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.
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.
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)
Note: Only the LTI system can be represented by a set of equations in the matrix-form as
x = Ax + Bu
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.
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!
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.
+ 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.
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
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.
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
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:
In the following section, some examples of numerical integration of ODEs are discussed.
x = F ( x , t )
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.
iL
R
v L
The governing equation is obtained by applying KVL to the loop and is given by
diL
L + R iL = v
dt
diL R v
= − iL +
dt L L
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
n t=nh in K
0 0 0 10
1 0.01 0.1 9
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.
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.
2.5
1.5
0.5
0
1 1.5 2 2.5 3
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.
R(Ω) L(H)
iL C(F)
v vc
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.
Table 6.4 System Eigenvalues for Different h- Forward Euler Method Series RLC Circuit.
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.
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.
In the following section, the application of the method is illustrated for a series RL and RLC
circuits.
R 1
in+1 = in + − in+1 + vn+1 h (6.18)
L L
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).
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)
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
i i
( I − [ A]h) Ln+1 = Ln + h[ B ]vn+1
vC n+1 vC n
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.
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.
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
and
x2 = −( x12 − 1) x2 − x1
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.
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.
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)
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
h h
K 3 = f xn + K 2 , t n +
2 2
K 4 = f ( x n + hK3 , tn + h)
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
y(0.5) = 3.21875
By actual integration, we get
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.
R 1 h
− − iLn + K11 1
K12 L L 2
K = + L vn
22 1 h
0 vC n + K 21 0
C 2
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
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
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.
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
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.
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.
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.
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
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.
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
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.
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
jxi2 = 1.8 pu
i 2
j0.5 1 j0.3
j0.15 j0.15
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
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
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.
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
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 )
w (k ) − w 0 ( )
(
F2 x ( k ) = w ) (6.50)
0
(
2 H 0.8 − 0.65 sin d ( k ) )
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,
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.
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.
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.
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.
−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
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.
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.
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.
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.
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.
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.
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.
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.
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
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
Now, substituting the values and solving for the critical clearing-angle, we get,
d cr = 1.7248 rad
Note that the above value is close to that indicated in Table 6.8 for Tclear = 0.411 s
dw m
M = Pm − Pmax sin d = f (d )
dt
2H
where M = and wm is the relative speed rad/s.
w0
dd dw m
dt = =M
wm f (d )
f (d )dd = M w m dw m
d wm
∫ f (d )dd = M ∫ w m dw m
ds 0
1 2
Pm (d − d s ) + Pmax (cos d − cos d s ) = Mw m
2
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)
P
d uep = (p − d s ) = p − sin −1 m
Pmax 3
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
1 2H 2 (6.65)
WKE = w
2 w 0
m
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.
Delta (rad) 2
1 B D
0
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.
dd
= w − w 0 (6.67)
dt
2 H dw
= Pm − Pmax 2 sin d (6.68)
w 0 dt
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.
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)
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.
[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
v(t) C
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.
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.
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.
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
SL (Complex power)
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 )
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°
Notes:
1. The following criterion is used to stop the iteration process:
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.
SL (Complex power)
Xc = 1.2 pu
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 )
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%.
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.
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
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.
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.
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.
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:
f1 ( x ) x1
x
f 2 ( x )
F (x) = x=
2
with
f n ( x ) xn
( ) ( )
−1
Dx( k ) = − J x( k ) F x( k ) (7.12)
x( k +1) = x( k ) + Dx( k )
(
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)
f1 ( x, y ) : x 2 + xy − 10 = 0
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 )
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] .
PL + jQL
*
=I =
(V1 − V2 ) (7.15)
V2
ZL
where Z L = ( RL + jX L ) =| Z L | ∠q L .
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)
(
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)
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
(
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 )
( )
max DP ≤ e
DQ
With Z L = 0.2062∠75.96° and V2( 0) = 1.0 ∠0°, for k = 0, we can write eq. (7.23) as
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.
(1 − V2 ∠ − d ) V22
PL + jQL = V2 ∠d + j (7.29)
ZL ∠ −q L Xc
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
(
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)
DV2( 0) 0.025
Dd =
( 0) −0.218750
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.
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
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
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.
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.
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:
PL3
T1 Y Y
1 2 3 4
G4
G1
Y Y T2
PL5
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.
j0.1 j0.135
j0.135
+ 5 +
E1 =1.0 0° E4 =1.0 0°
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.
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.
j0.1 j0.135
j0.135
IfG1
+ 5 IfG4 +
E1 =1.0 0° E4 =1.0 0°
If5
0.871258
= −j
2.724998
I f 5 = I fG1 + I fG 4 = − j 3.596256 pu
ZThf bus f
VPRf
I
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 j0.135
j0.135
ZTh2
5
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)
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 j0.135
j0.135
ZTh5 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)
1.0 ∠0o
If5 = = − j 3.596255 pu
j 0.278067
E1 E4
I GS1 = and I GS 4 = (8.5)
jxd′′1 jxd′′4
where I GS1 and I GS 4 represent current injections into the buses as shown in Figure 8.8.
+ +
j0.1
IGS1 j0.135
IGS 4
5 j0.135
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
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
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
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,
[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.,
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)
− 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)
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)
+ +
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.
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
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)
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
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,
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
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)
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.
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)
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.
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,
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.
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
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
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,
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
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.
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.
VPF2 0 0 0
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.
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,
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
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
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
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.
VPF 3 − VPF 2
I 32 = = ( −0.7408 − j1.1445) pu
jx L
*
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
Area−2
Area−1
3
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
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
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.
jx bus k
kk
+
o
1.0 0
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.
+ +
o
1.0 0
o 1.0 0
The above example clearly shows that with interconnection of sources, the fault level at a bus increases.
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
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.
a = e j120 (8.35)
o o o
1 + a + a 2 = 0 (8.37)
and
o o
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
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
= 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
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
1 1 1
1
[Ts ] = 3 1 a a 2 (8.43)
2
1 a a
−1
Vabc = [Ts ] Va( 012) (8.44)
1 1 1
−1
[Ts ] = 1 a 2 a (8.45)
1 a a 2
Thus, from eq. (8.44), we have,
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)
Similar to eqs (8.42) and (8.44), when sequence component transformation is applied to phase
currents, we get,
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).
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
= I a Z + ( I a + I b + I c ) Zn
= I a ( Z + Zn ) + I b Zn + I c Zn
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 )
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
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:
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:
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
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).
a Z a Ia (0) = 0 Z
n n
Ia (0)
Va (0) Va (0)
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
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)
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)
The above result implies that the sum of line-to-line voltages is always zero.
xd′′ + xq′′
Z2 g = (8.55)
2
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
A generator can carry zero-sequence current depending on the terminal conditions if Zng is finite.
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.
Zs
Ia
Zm Zs
Ib
Ic Zm Zs Zm
V1a V2a
V1b V2b
V1c V2c
Using KVL, we can write in matrix form the voltage drop across the line as
Using eqs (8.44) and (8.47) in the above expression and rearranging, we get,
where
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.
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
4. Obtain [PI] matrix such that the voltage drop across a line is expressed in terms of the vector
of bus voltages, [V] as
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)
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)
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)
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,
0 1 0 −1
[ PI ] = 0
1 −1 0
5. Obtain [Ym] = [Zm]−1 [PI] as
− 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
(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.
Y Y ∆ ∆
p s p s
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)
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
∆ Y ∆
Y
p s p s
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)
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
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.
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.
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)
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
P 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
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.
Zns Z0
Z0 = Z0L + 3Zns
Zero−sequence current paths
P S
P S
Znp
Z0 = Z0L + 3Znp
P S
P S
P S
Z0P Z0S
xd′′ + xq′′
Z2 g = j = Z1g ( = jxd′′ )
2
for the generators. This also implies that [YBUS(2)] is identical to [YBUS(1)].
j0.1 j0.135
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.
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)
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].
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.
I fb = I fc = 0 (8.66)
Writing the voltage equations for the Thevenin equivalent circuit of the sequence networks with
respect to the fault bus, we have,
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
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.
Z_i2=Ybus\e2
where e2 = [0,1,0,0,0]T.
j0.1 j0.135
j0.135
o + 5 + o
E1 =1.0 0 E4 =1.0 0
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
Reference
(a) Reference
(b)
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)
j 0.085 × j1.585
ZTh2( 0) =
j (0.085 + 1.585)
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
∆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)
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
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
= 0.90926∠ − 107.74o
= 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
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
= 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 .
a
I fa
.f I fa (1)
f f
I fa (2)
+ +
b .
f
+
Z Thf (1) Zf Z Thf (2)
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)
1
I fa( 2) = ( a 2 − a ) I fb
3
This implies that,
Ifa(0) = 0
Ifa(1) = – Ifa(2) (8.81)
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,
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).
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
Zf
Reference
= −4.8635 pu
= 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
4. The sequence components of the generator-1 currents neglecting the phase-shifts offered by the
∆-Y transformer, T1, are
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.
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)
Using eq. (8.90) in the above expression and noting that a + a 2 = −1 , we get,
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,
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,
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
( ) ( )
( 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.
j0.135 j0.1
j0.1 j0.5 j0.1 4
1 2 3 5 j0.135
Reference
j1.5
1 2 3 4
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
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
4. The sequence components of the generator-1 currents neglecting the phase-shifts offered by
the ∆ - Y transformer, T1, are
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).
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.
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.
dSmi
dt
=
1
2 Hi
( )
Tm0i − Tegi (8.103)
Tegi = Real Ei I Gi
*
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)
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
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
(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.
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
0.2
Rotor angle wrt COI (rad)
0.1
G−1
0
G−4
−0.1
−0.2
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)
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.
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)
−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.
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.
2. Given that T
2p 2p
vabc = 10 sin(w 0 t ) 9 sin w 0 t − 10 sin w 0 t +
3 3
1 2
3 4
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
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.
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.
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.
Turb Gen
Pg RL xFC Eb = Eb∠0°
xL
Vg = Vg ∠d g
Generator Infinite bus
Tmt −Teg
Ktg
Turbine Generator
(Ht) (Hg)
Dtg
Dt Dg
Tmt
Mechanical system dg
Teg
The equations of the two rotor-mass system are summarised as follows [4]:
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
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
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.
1
2p f er = (9.6)
xL 1
w 0 x FC w 0
or
x FC
f er = f 0 (9.7)
xL
Pg
Generator Infinite bus
Since normally xFC < xL, this implies that fer < f0, signifying that the resonance phenomenon occurs at
a frequency below the nominal frequency.
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
2 2
with Vm = Vg and Em = Eb .
3 3
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)
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)
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.
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)
Tmt
dg Teg
Mechanical system Electrical system
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)
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)
Tmt
dg Teg
Mechanical system Electrical system
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.
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).
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
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)
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)
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,
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
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
( )
∆x t = [ At ]( 4 × 4) + [ Bt ]( 4 ×1) [Sr 2 ](1× 4) ∆x t + [ Bt ]( 4 ×1) [Sr1 ](1× 4)) ∆x N (9.17)
∆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
∆x N = [ AN ]( 4 × 4) ∆x N + [ BN ]( 4 × 4) ∆x t (9.19)
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.
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.
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.
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
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
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.
[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
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.
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.
HP IP LPA LPB G E
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)
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.
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:
Eb = 1.0∠0°
xT = 0.14 Eb
Vg ∆
865.0 MVA
All data are in pu on generator MVA base
Following the modal transformation procedure indicated in [1], the modal frequencies are obtained
as shown in Table 10.2.
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.
∆TmH
[ ∆Tm′ ] (2,1) = 2H H
∆TmI
[ ∆Tm′ ] (4,1) = 2H I
∆TmLA
[ ∆Tm′ ] (6,1) = 2 H LA
∆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.
where
1
[ DTe ](12×1) (10,1) =
2 HG
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,
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
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 LALB DLALB
Asm (6, 7) = ; Asm (6, 8) =
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 LBG DLBG
Asm (8, 9) = ; Asm (8,10) =
2 H LB 2 H LB
9. 9th row:
Asm (9,10) = w B
KGE DGE
Asm (10,11) = ; Asm (10,12) =
2 HG 2 HG
KGE D + DE
Asm (12,11) = − ; Asm (12,12) = − GE
2H E 2H E
Table 10.3 Generator Standard Parameters (on Machine Base 865 MVA, 22 kV), IEEE FBM.
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′
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′
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)
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′
∆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′′
or
∆Te(1×1) = [Ce ](1×6) ∆xe (6×1)(10.32)
where
[Ce3 ](1× 2) = −y q0 y d 0
∆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)
[ R9 ](1×12) = 0 0 0 0 0 0 0 0 1 0 0 0
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
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
−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)
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)
Vg
Efdmax
−
+ KA
Vref Σ Efdst
1 + sTA
+ Efdmin
VS
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)
(
∆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
v vQg 0
[Vtr ](1×2) = VDg 0
Vg 0
g0
∆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)
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’
and ‘b ’ networks are identical, it is to be noted that the currents and voltages are different in both
the networks.
R L C
+
ia vCa
vga eba
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
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
d vCD iD
[TK ] ( 2× 2) = [C1 ]( 2× 2) [TK ] ( 2× 2)
T T
v (10.55)
dt CQ ( 2×1) iQ ( 2×1)
Pre-multiplying both sides of the above equation by [TK](2 × 2) and making use of the identities,
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.
∆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
and
where
R 0 L 0
[ R](2×2) = 0 R
[ L ](2×2) = 0L
and
xE
L= ; x E = xT + x L + xSYS
wB
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)
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,
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)
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.
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)
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)
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)
∆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
[ 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)
In order to remove the redundant states associated with the line inductor, a procedure described below
is employed:
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).
∆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
and
d ∆iD
= C DQi ∆ x np( 21×1) (10.91)
dt ∆iQ ( 2×1) ( 2 × 21)
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.
∆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)
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,
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.
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
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
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
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)
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′′
[Ms3](2 × 6) is given by
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
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 − 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 )
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 )
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)
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
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
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
E fd 0
Vref = + Vg 0 (10.136)
KA
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.
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.
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.
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
The eqs. (10.139) to (10.144) are combined and written in matrix form as,
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:
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.
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.
−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.
Table 10.6 Eigenvalues for the IEEE FBM System with Pg0 = 0.5 pu and RL = 0.
(Continued)
Table 10.6 (Continued)
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.
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.
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.
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.
+
Vto ∠0°
Efdo ∠d
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.
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)
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].
Amplifier
Slip ring
Field Field Armature PT
Armature
Regulator
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
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.
The time constant of the bus voltage measuring transducer can be taken as 0.02 s.
Two typical types of AC excitation systems are discussed in the following sections.
N
Armature Armature Armature
S
PT
Field of
AC exciter
Three phase AC
Regulator
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
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)
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.
The time constant of the bus voltage measuring transducer can be taken as 0.02 s.
Main generator
AC Exciter Stationary Field
Controlled rectifier
Field Slip ring
Armature Armature
PT
Exc.
reg.
Regulator
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
Typical values of the parameters of IEEE-type AC4A excitation system are tabulated in Table 11.3.
The time constant of the bus voltage measuring transducer can be taken as 0.02 s.
Armature
Three
phase AC*
PT
* Alternatively, from
auxiliary bus
Regulator
The IEEE-type ST1A exciter model (see Figure 11.10) represents a potential-source controlled recti-
fier systems.
+ 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
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.
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.
VC
Efdmax
−
+ KA
VREF Σ Efd
1 + sTA
+
Efdmin
VS
The time delay associated with the bus voltage measuring transducer is neglected.
Main generator
Saturable
current
transformer Armature
CT
Current PT
source Field Power
transformer
Source
regulator
Slip ring
Voltage source
Regulator
The IEEE-type ST2A exciter model (see Figure 11.13) represents a compound-source rectifier
excitation system.
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
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.
The time delay associated with the bus voltage measuring transducer is neglected.
Y ( s) sK F
= (11.7)
U ( s) 1 + sTF
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
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
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
sU ( s) (11.21)
Y ( s) =
1 + sT
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.
E fd ( s) KA
= (11.25)
[VREF − Vc ]( s) 1 + sTA
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.
VC
−
+ + 1 z 1 Efd
Σ KA Σ TA s
VREF0
−
VS
KC2
+
+ H
VC V1 KA EFD
− Σ
Σ 1 + sTA
+ +
VREF
KC1
1 + sTB
V2
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
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)
where Vc0 is the terminal voltage of the machine under steady state.
Speed deviation
PM = Mechanical power
LFC = Load frequency control
Speed Speed deviation
governor
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.
w
Tm10
− +
+ w
Turbine
Σ Kp1 Σ
w Ref1 generator−1
+ ∆Tm1
w Tm20
− +
+ Turbine
Kp2
w Ref2 Σ Σ
+ ∆Tm2 generator−2
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
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
(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.
PGV (1 − sTW) PM
(1 + 0.5 sTW)
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
2
B = Y (s)
+ s =
(1 − sTw )
2 Tw
Tw s =−
s
Tw 2
2 s =−
Tw
1+ 2
=
Tw 2
−
2 Tw
= −3
1 −3
Y ( s) = +
s 2
T + s
w
1 3
= −
s 2
T + s
w
−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.
Reheater Crossover
+ +
Σ Σ
PM
+ +
PGV
1 1 1
1 + sTCH 1 + sTRH 1 + sTCO
Figure 11.21 Tandem Compounded, Single Reheat Type Steam Turbine Model.
Table 11.6 Typical Values of Tandem Compounded, Single Reheat Type Steam Turbine Parameters.
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
Po
Pmax
+
K (1 + sT2)
Σ
p.u. slip (1 + sT1)(1 + sT3) ∆Pe − PGV
Pmin
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.
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.
PO Pmax
+
K (1 + sT2) − 1 1 PGV
Σ
p.u slip 1 + sT1 T3 s
−
Pmin
Typical values of parameters for speed-governor of steam turbines are tabulated in Table 11.8.
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
4
3V
2
Signals (V )
−2
−4
−6
−8
0 1 2 3 4 5 6
Time (s )
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.
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 )
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.
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 )
+ 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.
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
+
VC − V1 1 + sTC KA EFD
Σ 1 + sTA
1 + sTB
+
VRMIN
VREF V1MIN
Figure 11.32 Figure for Analysing the Small-signal Stability of the Operating Point
without Speed Governor.
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.
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.
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
Vt∠q
E = E∠0
jXL
Generator Peg
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
Now, linearising the swing equation around the operating points, we have,
d ∆d
= ∆Smw B (12.5)
dt
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)
d ∆Sm
2H = − D ∆Sm + ∆Tm − ∆Teg (12.6)
dt
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
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.
x
K
F M
Friction is zero
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
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
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.
This shows that the speed deviation phasor leads the angle deviation phasor by 90°. These phasors
are shown in Figure 12.6.
∆Sm( 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
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
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
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.
∆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
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.
PT
Measurement
block
Main th
Vg Field i machine
−
VRef Power
Σ AVR Exciter system
+
+/−
VS
PSS
PSS input signal(s)
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.,
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
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 )
−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
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
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.
18
16
14
12
Phase−deg
10
0
0 0.5 1 1.5 2 2.5 3 3.5 4
Frequency in Hz
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]:
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
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.
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)
∆Sm( jΩ)
∆Teg3( jΩ)
∆TegD( jΩ)
FL
∆Teg1( jΩ)
∆d ( jΩ)
∆TegE( jΩ)
Vt = 1.0∠q Eb = 1.0∠0 pu
jXL = 0.4 pu
Generator Pg
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 Eb
where Teg = sin d and Tm is assumed to be constant at Tm0 = Pg0, neglecting losses.
( x′d + xL )
∆d 0 wB
∆d
= Ks (12.33)
∆Sm − 0 ∆Sm
2H
E q′ 0 Eb
where K s = cos d 0 .
( xd′ + x L )
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.
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.
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.
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 )
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.
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
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′ )
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,
∆id = B1∆x
∆iq = B3 ∆x (12.43)
∆y d = B4 ∆x
∆y q = xq B3 ∆x
(12.44)
with B4 = ( xd′ B1 + B2 )
= 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
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
∆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.
0 376.9911 0
A(3×3) = −0.1106 0 −0.1063 (12.50)
−0.3543 0 −0.4630
1.34
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.
V = 1.0∠q E = 1.0∠0pu
jxL = 0.7 pu
Generator Pg
Table 12.3 Standard Parameters for a 865 MVA, 22 kV Generator in pu on the Machine Base.
∆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.
∆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
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,
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)
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)
where
− E cos d 0 0 0 xL
[ Avs ](2×2) = [ Ave ](2×6) = [Ce1 ]
− E sin d 0 0 − x L 0
∆xe (6×1) = [ AG ](6×6) ∆xe (6×1) + [CWm ](6× 2) [ Avs ]( 2× 2) ∆xs ( 2×1)
Now, from eqs (12.61) and (12.55), we can combine the state equations as
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
T
where y dq = y d y q and xrs = [y F y H yG y K d w y ]T
∆y dq = − [ AES11 ]−1
( 2 × 2) [ ES12 ]( 2 ×6) [ rs ]( 6 ×1) (12.64)
A ∆x
Using eq. (12.64) in eq. (12.65), we have the reduced state-model given by
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
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.
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).
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.
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
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
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.
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.
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.
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.
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).
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.
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
where
1
Atf
(1×1) = − T
A
KA
Btf
(1×1) = − T
A
KA
Etf
(1×1) = T
A
V = vd2 + vq2
or
or
∆vd
∆V = [Vv ] (12.72)
∆vq
where
v vq 0
[Vv ] = Vd 0 V0
0
where
∆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 .
∆y dq = − [ AFD11 ]−1
( 2 × 2) [ FD12 ]( 2 × 7) rsf ( 7 ×1) (12.77)
A ∆x
Using eq. (12.77) in eq. (12.78), we have the reduced state-model given by
where
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.
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.
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:
1.5
Delta (rad)
0.5
0
0 2 4 6 8 10 12 14 16 18 20
Time (s)
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.
where
[Cet ] = [[Ce ] 0 0 0 ]
∆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
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).
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,
∆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)
−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 )
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 )
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.
Yw ( s) sTw
= + 1 − 1 (12.88)
U ( s) (1 + sTw )
1
= 1 − (12.89)
(1 + sTw )
1 1
xw = − xw + u (12.90)
Tw Tw
and
yw = u − xw (12.91)
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
K c1 1 K
xc = − xw − xc + c1 u (12.99)
T2 T2 T2
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
(
rsf ( 7×1) = [ AFDR ]( 7× 7) ∆xrsf ( 7×1) + [ EC1 ]( 7×1) ∆Vref + ∆Vs
∆ x )(1×1(12.103)
)
∆u = [ I 6 ]∆xrsf (12.104)
∆ 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
[ 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.
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.
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)
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.
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.
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
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
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
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.
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
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.
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.
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.
x
2
F1
x
3
k k
M 12 23
1
M M F3
2 3
Damping B1
F1 F3
M M M
1 2 3
B1
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
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
M1 = 10 kg; M 2 = 1 kg; M 3 = 1 kg
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
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.
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.
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
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
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
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
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.
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 )
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.
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 )
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.
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
l5 = 0 l6 = −0.4279
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,
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.
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 )
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.
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 )
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 )
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.
M1 x1 + M 2 x2 + M 3 x3
xCOI = (13.15)
M1 + M 2 + M3
M1v1 + M 2 v2 + M 3v3
vCOI = (13.16)
M1 + M 2 + M 3
and
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.
2. Now, unlike in Figure 13.3, xiCo and viCo contain only the low frequency component.
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 )
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 )
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.
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.
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.
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 )
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.
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.
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 )
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
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.
( )
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
( )
= 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,
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
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 )
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
−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:
ith generator
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:
Further, we have,
T T
∆Vg = P g ∆VQD , and ∆VL = P l ∆VQD
T T
∆I g = C g ∆xg + Dg P g ∆VQD , and ∆I L = [YL ] P l ∆VQD (13.27)
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
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
dy G 1
= −y G + y q (13.34)
dt Tq′
dy K 1
= −y K + y q (13.35)
dt Tq′′
where
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)
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)
( 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′′
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′′
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′′
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′′
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)
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
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
Ag (1, 2) = w B
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
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′
1
Ag (5, 5) = −
Tq′
Vg 0 cos(d 0 − q g 0 )
Ag (6,1) =
Tq′′
1
Ag (6, 6) = −
Tq′′
C3y K 0 sin(d 0 − q g 0 )
+ + C5Vg 0 cos(2(d 0 − q g 0 ))
xq′′
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′
cos(d 0 − q g 0 )
Bgp (6,1) = −
Tq′′
sin(d 0 − q g 0 )
Bgp (6, 2) =
Tq′′
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′′
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′′
VS
dE fd
dt
=
1
TA
( )
− E fd + K A Vref + Vs − Vc (13.74)
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
xE = [ vR x B x F ](3×1)
T
+ +
PM
+ +
F F F
HP IP LP
P
GV 1 1 1
1 + sT 1 + sTRH 1 + sT
CH CO
P P
O max
+
K(1+sT2) − 1 1 PGV
Pmin
The above model is modified as shown in Figure 13.17 for the purpose of linearisation.
KT2
T1
T y1 +
pu slip K 1− 2
T1
+
Q
1+ sT1
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
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)
(
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
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
)
1
Vg 0 ∆q g = −VDg 0 ∆VQg + VQg 0 ∆VDg (13.83)
Vg 0
We know that
Vg2 = VQg
2 2
+ VDg
Now consider,
VQg 0 VDg 0
∆Vg = ∆VQg + ∆VDg (13.84)
Vg 0 Vg 0
Denoting
Vg 0 ∆q g ∆VQg
∆Vgp = and ∆Vgr =
∆Vg ∆VDg
Thus,
∆Vgp = [ P ] ∆Vgr
where
1 −VDg 0 VQg 0
[P] = V VDg 0
Vg 0 Qg 0
[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
nb
∑ (G1 jVDj + B1 jVQj ) = I D1 (13.86)
and j =1
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
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)
∆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.
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
V = VQ2 + VD2
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
Q
∆QL = nk L 0 ∆V (13.101)
V0
where
nk = n p r1 + ni r2 + nz r3
VQ 0 VD 0
∆V = ∆VQ + ∆VD (13.102)
V0 V0
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
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
1 0
PG (i , j ) =
0 1
0 0
PG (i , j ) =
0 0
The generator current vector, ∆IG is a collection of the quantities ∆Ig1, ∆Ig2, ∆Ig3, …, ∆Igng and using
(13.73), ∆Ig can be expressed as
where
T
∆X G = ∆xTg1 ∆xTg 2 ∆xTgng
∆VGT = ∆VgT1r T r
∆VgT2r ∆Vgn
g
T
∆I G = ∆I Tg1 ∆I Tg 2 ∆I Tgng
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
′
∆X G = [ AG ] ∆X G + [ BG ] ∆VG + [ EG ] ∆U c (13.118)
where
[ BG ] = BlkDiag Bgr1 r
Bgr 2 Bgng
∆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.
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
∂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.
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.
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°.
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.
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
Table 13.1 Eigenvalue Listing for Unreduced State Matrix, Two-machine System.
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
(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
( ∆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:
Note: The time–domain simulations are generally carried out using the original unreduced variables
which preserves the zero-mode effects.
Table 13.2 Eigenvalue Listing for Reduced State Matrix, Two-machine System.
5 −41.0237 1 0 DampH-3
6 −38.8606 1 0 DampH-1
9 −3.2860 1 0 DampG-3
10 −2.6531 1 0 DampG-1
(Continued)
Table 13.2 (Continued)
11 −20.4799 1 0 DampK-3
12 −20.2055 1 0 DampK-1
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
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).
−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.
Table 13.3 Eigenvalue Listing for Reduced State Matrix, Two-machine System with Turbine on M1.
1 −2.5000 1 0 Turbx3-1
2 −0.1000 1 0 Turbx2-1
3 −99.9827 1 0 PG-1
6 −50.0682 1 0 y1gv-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.
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).
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.
−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
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.
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 )
Table 13.7 Oscillatory Modes-Modified 4 Machine System-without AVR in the Base Case.
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).
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).
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.
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
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
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
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].
∆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
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
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.
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 e2 = [ 0 1 0 0 0 ](1×16n ) , with 1 corresponding to slip state variable of third machine.
T
g
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.
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
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
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.
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
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.
−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.
[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. MartÃn, J. M. RodrÃguez, 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.
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.
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
Construct the bus admittance matrix accounting x′d of the machines. Show that the matrix can be
written including the internal nodes, as
[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)
1 xt 2 3 4
L1
xsys
L2
Vg = 1.05 Eb = 1.0∠0°
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).
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.
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.
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
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]:
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
where
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 ) = − (14.18)
( Ra + jxd″ ) ( Ra + jxd″ )
(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.
Add and subtract xd″ iq from the LHS of the above equation, we get
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
″
0 ≈ − Edummy − ( xq″ − xd″ )iq
(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″
(iQ + j iD )
Ra
Vg = (vQ + j vD )
IDQS = (iQS + j iDS)
x´´
d
Augmented network
Eq″
idS = −
xd″
( x′ q − x ″ q ) xq − x ′ q x ″ q
E″d = − yK + yG
x′ q xq x ′ q
( x′ − x ″ d ) x − x′ d x ″ d
E″q = d yH + d yF
x′ d xd x ′ d
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
where,
y d = x ″ d id + E ″ q and y q = x ″ q iq − E ″ d
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″
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:
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.
y y
2 2
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.
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
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.
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
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.
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.
dH1d 1 dH 2d 2
= ( H1w1 − H1w 0 ) and = ( H 2w 2 − H 2w 0 ) (14.40)
dt dt
dd COI
HT = H1w1 + H 2w 2 − ( H1 + H 2 )w 0 (14.41)
dt
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
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.
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
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
ddi dd i dd COI
= − = ( Smi − SmCOI )w B (14.54)
dt dt dt
or
ddi
= Smiw B (14.55)
dt
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
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
Subtransmission
EHV bus level Large customers
their wiring
Distribution
Feeders
Substation level
PL , Q L
(System Load) HV customers
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.
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.
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
Thus,
*
S
I Lo = Lo (14.62)
Vo
PLo − jQLo
= (14.63)
Vo*
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
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.
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.
Augmented Network
YBUS
ILn YL
(Negative
Current injection)
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.
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
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
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).
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
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
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:
− 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
− 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
0.5
Slip wrt COI
−0.5
−1
−1.5
0 1 2 3 4 5 6 7
TIME (s)
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.
0.5
Slip wrt COI (pu)
−0.5
−1
−1.5
0 1 2 3 4 5 6 7
TIME (s)
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.
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.
∆f
PG1 + PG 2 = PLo 1 + k pf
f o (14.76)
−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)
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.
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.
SmCOI =
( Sm1 + Sm1 )
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)
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.
2
Slip wrt COI
−2
−4
−6
0 1 2 3 4 5 6 7 8 9 10
Time (s)
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)
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.
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)
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)
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.,
−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.
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)
(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.
−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.
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.
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)
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
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.
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)
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 )
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
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 )
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
0.5
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.
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.
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.
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
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.
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:
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.
10
−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.
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.
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).
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.
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.
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).
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).
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.
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
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.
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.
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.
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.
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)
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.
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.
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.
0.25
0.2
Load Bus freq. deviations (pu)
0.15
0.1
0.05
−0.05
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.
[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.
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
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.
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)?
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.
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
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).
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)
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
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)
SB
Base torque ( rotor ),TB =
w rmB
2
where, wrmB (in mech. rad/s) is the base speed for the rotor = wB .
P
Rs
2. p.u stator resistance/phase Rs =
ZB
Lls
3. p.u stator leakage inductance/phase Lls =
LB
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)
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
1 2
J w rmB
9. Inertia constant, H = 2 in MJ/MVA, where J is moment of inertia in kg-m2.
SB
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
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
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
By setting the derivative equal to zero in (15.9), the steady-state value of the load torque is
obtained as TL0 = Te0.
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)
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
2.4 × 103
SB = = 2962.96 VA
0.9 × 0.9
VB = 460 V
w B = 376.9911 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
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,
isD 0 −0.574448
i 0.829639
sQ 0 =
irD 0 0.092932
irQ 0 −0.85366
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
−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:
= 0.829639 pu
2 2
In the above calculations, (isD 0 + isQ 0 ) Rs represents the three-phase copper losses for the stator
windings.
I r20 Rr 0.013836
Air − gap power = = = 0.80442 = Te 0 pu
sr 0.0172
= Te 0 − I r20 Rr
(1 − sr )
= I r20 Rr
sr
and
= Te 0 + I s20 Rs
= 0.82964 pu
= 0.574448 pu
= 0.57446 pu
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
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
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)
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
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 )
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.
TL = TL 0 (w r )m
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.
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).
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
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
y rD − Lm isD
irD = (15.24)
Lrr
Lm L 2
y sD = Lss isD + y rD − m isD
Lrr Lrr
or
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)
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´
w −wr 1 dy rD
0 = Rr irD + B y rQ + (15.28)
wB w B dt
Lrr
y rQ = ′ (15.29)
vD
Lm
Let us define
1 Lrr
T0′ =
w B Rr
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
vQ′ Lm
irD = − − isD (15.31)
Lm Lrr
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
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′
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
or
Rr Lrr ′
1 Lrr dvD
0= irQ Lm + (w B − w r ) vQ′ + (15.36)
Lm w B Rr w B Rr dt
y rQ Lm
= − isQ
Lrr Lrr
′
vD L
irQ = − m isQ (15.37)
Lm Lrr
′
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′
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
Te = y rD irQ −y rQ irD
Lrr L
=− ′ rr irD
vQ′ irQ − vD
Lm Lm
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)
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.
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′
dw r
2H = Te − TL (15.45)
dt
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.
or
PS = vsQisQ and QS = −vsQisD
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).
2. The real power delivered to the mains is, Ps = −0.7802 pu. The reactive power absorbed from
mains is Qs = 0.6061 pu.
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:
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.
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.
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.
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:
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)
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.
c
closed at t = 0
Efd
Table 15.1 Standard Parameters for a 230 V, 1000 VA Generator in pu on the Machine Base.
T ′d = 18 ms T ″d = 11 ms
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
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
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
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
dd w a
= w0 − w0
dt w rB
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.
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
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
0.679 0.97
× × (3 × S B ) = 987.95 W
2 2
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)
5. Further, it can be verified that the total real power consumed is,
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
−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
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
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
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
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
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)
2
0.5367 1
(3 × S B ) = 43.2 W
2 Ropen
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)
ir = ig − iL(15.61) ir Rc
Since vg Ropen
+ vc
vg = Ropenir = Ropen (ig − iL)(15.62) C
(
vOpen = Rc + Ropen C) dvC
dt
+ vC (15.65)
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
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
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
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.
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
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
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
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)]
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
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)
0.8654 0.05365
P2 = × 3 × S B = 69.64 W
2 2
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.
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.
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).
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.
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)
−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.
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
3 4 5 6 7 8 9 10
t(s)
Delta of G1 wrt G2
0.2
Delta (rad)
3 4 5 6 7 8 9 10
t(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
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.
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.
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
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.
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.
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.
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.
0.7854 0.5504
P1 = × 3 × S B = 648.42 W
2 2
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°.
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,
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.
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).
100
Vdc1 = 205 V
50 94.83 VAR
Qout (VAR)
−50
R = 0.7 pu
−100
−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
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
and
( )
V = I g1 + I g 2 0.7 = 0.96683∠ − 10.615o pu
( )* × S B = 906.836 − j94.836 VA
V I g1
( )* × 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
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.
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.
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).
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
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)
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
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)
and
( )
V = I g1 + I g 2 10 = 1.0636∠0.25o pu
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
( )* × 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.
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
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.
1. From the RL-circuit shown in Figure 15.18, for a three-phase balanced condition, we
can write
iaL iaL va
d R 1
ibL = − l ibL + L vb
dt Ll l
icL icL vc
vabc = RSiabcr
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.
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)
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.
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.
2. The speed of the set reduces to 1.0491 pu. At this speed, the load impedance is computed as
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
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.
2
0.7241 1
PRs = × 3 × S B = 78.64 W
2 10
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.
250
Qout (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.
Son
i 0dq1 i abcg1 jx l
Rl
+
SC RO i abcL
vdq1 G1 [P (θ1)] −1
Mains
RS
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.
−1
iabcg1 = [ P (q1 ) ] i0 dq1 (15.76)
3. The angle q1 = wa1t and wa1 (in electrical rad/s) is obtained from (15.49).
4. The network equations are not enabled.
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.
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:
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)
and
( )
V = I g1 − I L 10 = 0.99695∠1.26o pu
( )* × S B = 208.682 − j19.469 VA
V I g1
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)
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)
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
2
ig1 peak = × 0.21023 = 0.1716 pu
3
and
2
iLpeak = × 0.11135 = 0.090919 pu
3
Losses in Rl:
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.
500
Pg1/Qout (W/VAR)
−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.
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.
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.
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
−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
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.
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)
−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.
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.
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