Dynamic Security of Interconnected Electric Power Systems - Volume 2: Dynamics and Stability of Conventional and Renewable Energy Systems

Power System Dynamic Security - Volume 2
This book is written for engineers involved in the operation, and control of
electric power systems. The book also provides information and tools for
researchers working in the fields of power system security, stability,
operation and control. It consists of two volumes. In the first volume, the
traditional techniques for stability and dynamic equivalence are presented.
In addition, an overview of the main drivers and requirements for
modernization of the traditional methods for online applications is
discussed. The second volume includes advances in the security, stability,
control, and stabilization of electrical power systems. The given material is
written in graded complexity for facilitating their inclusion in
undergraduate, postgraduate, and technical training courses. For online
dynamic security systems, a major part of this volume includes the
derivation, analysis, and stabilization of the SMIB equivalence of power
systems. This volume also includes the operation and dynamical
characteristics of variable renewable power generation (wind and solar-PV)
as well as their dynamical interactions with power systems and the
interconnection requirements are presented in details.
Mohamed EL-Shimy
Dr. M. EL-Shimy is currently an Assoc. Prof. in the

Dynamic Security of
department of Electrical Power and Machines – Ain
Shams University. He is also an electromechanical Interconnected Electric
specialist, and a freelance trainer. He is a technical
reviewer for some major journals and conferences.
Please visit http://shimymb.tripod.com and
Power Systems – Volume 2
http://goo.gl/tCNBLU for more details Dynamics and stability of conventional and
renewable energy systems
EL-Shimy
978-3-659-80714-5
DOI: 10.13140/RG.2.2.36832.07683
DYNAMIC SECURITY OF INTERCONNECTED
ELECTRIC POWER SYSTEMS – VOLUME 2
Dynamics and stability of conventional and renewable energy systems
Mohamed EL-Shimy
Ain Shams University – Cairo, Egypt
Contact info:
mohamed_bekhet@eng.asu.edu.eg
shimymb@yahoo.com
002 01005639589
http://shimymb.tripod.com
M. EL-Shimy. Dynamic Security of Interconnected Electric Power Systems – Volume 2. LAP, 2015.
To my wife Hala, and my children (Sara, Malak, Omar, and Aly)
2
PREFACE
This book is written for engineers involved in the operation, control, and
planning of electric power systems. In addition, the book provides
information and tools for researchers working in the fields of power system
security and stability. The book consists of two volumes. The first volume
provides traditional techniques for the stability analysis of large scale power
systems. In addition, an overview of the main drivers and requirements for
modernization of the traditional methods for online applications is discussed.
The second volume (i.e. thus volume) provides techniques for offline and
online stability and security studies. In addition, the impact of variable
generation on the stability, and security of power systems is considered in the
second volume.
The book covers some essential aspects related to the modeling,
simulation, and analysis of power systems. Its contents are useful for
educational and research objectives as well as training of engineers and high
level technicians. Fast and online assessment of the dynamic security and
stability of power systems are given significant attention in this volume. The
stability of grid-connected variable renewable energy sources is presented
considering wind and solar-PV technologies. In addition, the impact of these
renewable sources on the transient stability of power systems is presented
with sufficient details for providing an understanding of various dynamic
phenomena. Several numerical examples and elaborated case studies are
given to enhancing the capability of the engineers in the modeling,
simulation, analysis, and control of power systems.
Corrective actions for transient stability preservation and restoration are
also presented with a focus on the load shedding for restoring and
enforcement of power system stability. For these targets, the minimization of
load shedding is considered as a techno-economical solution for solving
stability problems associated with a sudden drop in the power generation.
These sudden drops may be caused by several reasons such as forced outage
of generating units, or the intense reductions in the renewable resources. The
load shedding minimization for dynamic security preservation is also
considered in this book. Reasonable approaches for online security and
corrective actions studies are presented. These approaches are based on the
electromechanical equivalence of power systems that is based on online
3
measurements. Single-machine infinite-bus (SMIB) equivalences of power

systems are presented. These equivalences can be estimated and studied at
high speeds that are reasonable for online decision making and
implementation of controlled changes or corrective actions. Consequently,
the modeling, the simulation, the control of the SMIB are given in details in
this volume. These issues are presented considering the steady and various
dynamic states as well as various levels of the modeling details. Many
toolboxes of the MATLAB and the power system analysis toolbox (PSAT)
are described and used in for implementing various models and simulations.
The first volume 1 consists of three chapters and starts with a detailed
overview of the operational requirement of recent and future power systems
considering the integration of variable generation resources into the
electricity grid (chapter 1). The fundamentals and advances of power system
security requirements are also presented in chapter 1. An overview of the
electromechanical equivalence techniques is presented in chapter 2. An
improved coherency-based equivalence technique is presented in this volume
(chapter 3). The presented technique uses the traditional data sets for the
construction of the equivalence. In addition, the concept of remote areas is
introduced for the maximization of the dynamic model reduction of very
large-scale systems. Several case studies are presented for the evaluation,
validation, and analysis of the presented theories and models.
This volume consists of 13 chapters and 9 appendices. The numbering of
chapters in this volume starts from 4 as the first three chapters form the first
volume. The titles of various chapters and appendices in this volume are:
- Chapter 4: dynamic modeling, simulation, control, and analysis of
simplified power systems
- Chapter 5: measurement-based SMIB equivalence of multi-machine
systems and transient stability
- Chapter 6: load shedding and online applications
- Chapter 7: damping improvement and stabilization through facts-pod
controllers
- Chapter 8: overview, modeling, and performance analysis of grid-
connected wind energy sources
1
M. EL-Shimy. Dynamic Security of Interconnected Electric Power Systems – Volume 1.
ISBN 978-3-659-71372-9, LAP LAMBER Academic Publisher, Germany, 2015.
4
- Chapter 9: dynamical impacts of wind power and damping

enhancement
- Chapter 10: overview, characterization, and modeling of solar-pv
systems
- Chapter 11: modeling, dynamic analysis, and control of grid-connected
solar-PV generators
- Chapter 12: consistence of wind power technologies with the fault
ride-through capability requirements
- Chapter 13: overview of the operational requirements of grid
connected solar-PV systems
- Appendix 1: steady state modeling of the SMIB considering the effect
of the excitation control
- Appendix 2: linearized transfer function of a simulink model, and root
locus based controller design
- Appendix 3: modeling of individual and aggregate loads in power
systems
- Appendix 4: fundamentals of speed governors
- Appendix 5: transient droop compensators for speed-droop governors
of hydroelectric units
- Appendix 6: interrelation between the dynamic and the static security
- Appendix 7: numerical integration methods
- Appendix 8: data of the detailed two-area system
- Appendix 9: nomenclature of the symbols used in chapter 10 and 11
The following table summarizes the main subject(s) associated with

each chapter and appendix. This table is also valuable in organizing the
reading and the use of the manuscript.
I would like to express my gratitude for some of my students that
contribute in executing some of the numerical examples and case studies
presented in this volume of the textbook. Significant parts of the numerical
examples of the analysis of wind energy conversion systems are partially
performed by Eng. Nisma Ghaly and Eng. Omnia Rassem. The simulation
results FACTS-POD design and analysis are partially performed by Eng.
Mohamed Mandour. Some of the results of the modeling and analysis of the
solar-PV generators are performed by Eng. Hossam Khairy. Finally, some of
5
the literature review of solar-PV technologies is performed by Eng. Taha

Abdo.
Stability
Equival-
analysis & Renewable energy
ence
control
Stability control
Nomenclature
Fundamentals
Modeling and
Conventional
Introductory
Impacts and
Renewable
Grid codes
simulation
and data
sources
sources
Offline
Online
Ch. 1
Volume 1
Ch. 2
Ch. 3
App.
Ch. 4
Ch. 5
Ch. 6
Ch. 7
Ch. 8
Ch. 9
Ch. 10
Ch. 11
Volume 2
Ch. 12
Ch. 13
App. 1
App. 2
App. 3
App. 4
App. 5
App. 6
App. 7
App. 8
App. 9
Generally, nothing is either absolutely perfect or final. Therefore, please do

not hesitate to send me your feedback, and suggestions.
M. EL-Shimy, Nov. 2015
6
TABLE OF CONTENTS (VOLUME 2)

Preface 3
Table of contents 7
Chapter 4
13 -
Dynamic Modeling, Simulation, Control, and Analysis of
115
Simplified Power Systems
4.1 Introduction 13
4.2 Identification and characteristics of steady and dynamic states
of power systems
4.3 Frequency dynamics of power systems 17
4.4 Linearized classical modeling and simulation of the traditional
23
SMIB system
4.4.1 The swing equation 24
4.4.2 The transient power-angle characteristics 31
4.4.3 Linearization and linearized analysis 34
4.4.4 Time-domain solutions of the linearized model 51
4.4.5 Small-signal stability and eigenvalues of the state matrix 64
4.5 Inclusion of field circuit dynamics in the linearized model of
65
the SMIB system
4.6 Including AER and PSS in the linearized third-order model of
80
the SMIB system
4.7 Power System Stabilizers (PSS) 86
4.8 Frequency response and Speed Governors 90
4.8.1 Frequency response of a governor-less generator 93
4.8.2 Types of speed governors 97
4.8.3 Supplementary control action for unit output power
108
control
Chapter 5
117 -
Measurement-based SMIB Equivalence of Multi-Machine
197
Systems AND Transient Stability
5.1 Introduction 117
5.2 Measurement-based electromechanical equivalence 122
5.2.1 Stage I: Inertia estimation at the interface buses 124
5.2.2 Stage II: Electrical parameters of the equivalent generator 125
5.3 SMIB equivalence of multi-machine systems 127
7
5.4 Transient stability and the Equal Area Criterion (EAC) 131
5.4.1 Classification of power system stability 133
5.4.2 The equal area criterion (EAC) and advancements 135
5.4.3 Transient Stability Enhancement 170
5.5 Measurement-based SMIB Equivalence of Multi-machine
193
power systems – A case study
Chapter 6 199 -
Load Shedding and Online Applications 246
6.2 Voltage Stability and UVLS – an overview 202
6.2.1 Basic Definitions 202
6.2.2 Classifications of voltage stability studies 205
6.2.3 Connection between Voltage Stability and Rotor Angle
207
Stability
6.2.4 Fundamentals of Voltage stability analysis 207
6.2.5 UVLS schemes 230
6.3 UFLS – AN OVERVIEW 234
6.4 Dynamic load shedding based on online measurements 237
Chapter 7
247 -
Damping Improvement and Stabilization through FACTS-POD
306
controllers
7.2 FACTS devices – Definitions 249
7.3 Modeling and Modal Analysis 254
7.3.1 Linearized Modeling for POD design 254
7.3.2 TCSC Dynamic Modeling and Control 256
7.4 POD Design and Tuning 258
7.4.1 Frequency Response Method 259
7.4.2 Residue Method 262
7.5 Study System 1 – SMIB system 265
7.6.1 Impact of the TCSC on the small- signal stability 267
7.6.2 Observability and controllability of various input signals 267
7.6.3 POD designs 268
7.7 Case Study 2 – Weakly interconnected multi-machine system 274
7.8 Placement of FACTS Devices 286
7.8.1 Preliminary mathematical modeling 287
8
7.8.2. General form of static-sensitivity analysis 291

7.8.3 TCSC Placement Problem 293
7.8.4 Study System 298
7.8.5 Implementation and Results 299
Chapter 8
307 -
Overview, Modeling, and Performance Analysis of Grid-
382
connected Wind energy Sources
8.2 Wind Energy Conversion Technologies (WECTs) 310
8.3 Modeling of Wind Turbine Generators (WTGs) 315
8.3.1 Wind turbine model and its controls 317
8.3.2 Shaft model 319
8.3.3 Modeling of Generators, and their controls 321
8.4 Case study 1 - Steady-state characteristics and steady-state
332
stability of DFIG
8.5 Case study 2 - Transient performance of FSWT based on SCIG
341
considering the single-mass model (rigid rotor)
8.6 Case study 3 – Design of Pitch Angle Controllers 353
8.7 Case study 4 – Modal Analysis of FS-IGIB System 366
Chapter 9 383 -
Dynamical Impacts of Wind Power and Damping Enhancement 418
9.2 Impact of wind power on the inertia of power systems and the
383
equivalence SMIB
9.3 Impact of wind power on the stability of power systems 387
9.4 Response of WTGs in comparison with Conventional
399
Synchronous Generators
9.5 Design of PODs in the presence of wind power generation 411
Chapter 10 419 -
Overview, Characterization, and Modeling of Solar-PV Systems 477
10.2 Qualitative Overview of solar-PV technologies 422
10.3 Quantitative Characterization of solar-PV technologies 425
10.4 Fundamentals and Applications of solar-PV systems 442
10.4.1 Solar-PV arrays 442
10.4.2 PCU and MPPT 449
9
10.4.3 Earth-Sun Geometry and Observer Angles 457

10.4.4 Sun tracking mechanisms 463
10.4.5 Modes of operation of solar-PV generators 466
10.5 Modeling of grid-connected solar-PV generators 469
10.5.1 Resource Assessment and Energy Production (RA&EP)
470
Models of solar-PV modules
10.5.2 Equivalent Circuit based Models (ECBMs) of solar-PV
473
generators
10.6 Case Studies 477
Chapter 11
479 -
Modeling, Dynamic Analysis, and Control of Grid-Connected
527
Solar-PV generators
11.2 Modeling of solar-PV generators for dynamic analysis 481
11.2.1 Solar-PV array modeling 482
11.2.2 DC-DC Converters and MPPT control 493
11.2.3 DC-AC Converters (Inverters) and Power control 508
11.3 Matlab-Based Modeling and Simulation of the grid-connected
514
Solar-PV generator
11.3.1. Modified MPPT techniques 516
Chapter 12
529 -
Consistence of Wind Power Technologies with the Fault Ride-
540
through Capability Requirements
529
12.1 Introduction
12.2 FRT Capability Requirements 530

12.3 Numerical Examples 534
Chapter 13
541 -
Overview of the Operational Requirements of Grid Connected
556
Solar-PV Systems
13.2 Grid code and standard requirements 543
13.2.1. Normal operation requirements 543
13.2.2. Under grid disturbance requirements 552
13.2.3 SCADA Integration Requirements 555
Appendix 1: Steady state modeling of the SMIB considering the effect 557
10
of the excitation control

Appendix 2: Linearized transfer function of a simulink model, and
591
root locus based controller design
Appendix 3: Modeling of individual and aggregate loads in power
595
systems
Appendix 4: Fundamentals of speed governor 617
Appendix 5: Transient droop compensators for speed-droop governors
621
of hydroelectric units
Appendix 6: Interrelation between the dynamic and the static security 625
Appendix 7: Numerical integration methods 629
Appendix 8: Data of the detailed two-area system 633
Appendix 9: Nomenclature of the symbols used in Chapter 10 and 11 637
11
12
Chapter 4
Dynamic Modeling, Simulation, Control, and Analysis of
Simplified Power Systems
4.1 Introduction
In the first volume of this book, the construction and evaluation of the
coherency-based electromechanical equivalence is presented. The presented
method uses the traditional approach which relies on the availability of the
required data sets of system parameters as well as its variables. The method
presumes that the required data sets are complete and accurate; however, not
only the operating conditions of power systems vary with time, but also the
parameters of various components; for example, aging and operating stresses
affect the values of the parameters of various components and may cause
significant change in their recorded historical values. In addition, the
information provided by the data sets is highly uncertain and may suffer
incompetence or at least they may not be in agreement with the actual system
data. The modeling and analysis presented in the first volume consider only
conventional power systems. The integration of recent generation
technologies causes changes in the dynamic behavior and interaction between
system components. Therefore, the impact of these technologies as well as
advanced modeling approaches and methods for online stability and security
studies are presented in this volume.
The problems associated with the availability and accuracy of the data sets
of power systems as well as the estimation of the operating conditions are
highly reduced by the use of synchronized online monitoring of power
systems. Therefore, electromechanical equivalency techniques that solely use
online measurements instead of the traditional data sets are presented in this
volume. The method can be effectively executed online and this becomes
possible due to the availability of WAMS and PMUs. The WAMS and PMUs
are recently introduced in power systems for time-synchronized
measurements, telemetry, and recording of critical quantities (such as state
variables, and status of components). Consequently, accurate online analysis
and management of large-scale power systems becomes possible. An
overview of the functionality and characteristics of these systems is
summarized in the first volume of the book (chapter 1 and 2).
13
The objectives of this chapter also include the analysis of some critical
issues associated with the modeling, stability, security, operation, and control
of power systems.
4.2 Identification and characteristics of steady and dynamic

states of power systems
A great variety of different dynamics occurs in power systems. The
physical origin and the time scale of these dynamics are very different. If the
time derivative of any system’s quantity is non zero, then the system is
conceptually is in a dynamic state. In that sense, a power system will never
reach a steady state condition because there are normal continuous changes in
the system loads. This is too strict from engineering, mathematical, or even
physical points of view. A power system may be subjected to many changes
that may not be considered as disturbances. For example, slow and small
changes in the system loading is a situation where a power system is
considered in a steady state rather than a dynamic state; however, sudden and
large changes in the system loading activate the system’s dynamic state.
Generally, sudden changes are disturbances that cause fast changes in the
system operating conditions. From mathematical point of view, the
magnitude of a disturbance defines the appropriate mathematical modeling
approach of a power system. If the magnitude of a disturbance is large, then
modeling a power system using differential equations is common for
handling the large nonlinearities in the performance of power systems. On the
other hand, linearized models are generally acceptable for modeling power
systems subjected to small disturbances. In the two situations (i.e. large and
small disturbances), the desired accuracy level, the simulation time, and the
available computer memory defines the model details. More modeling details
are always associated with higher accuracy; however, these higher accuracy
levels are at the expenses of larger simulation time and memory requirements
as well as mathematical complexity. In addition, highly detailed models may
result in too large number of time dependent quantities. Consequently the
results may be difficult to be understood and clear conclusions may be also
difficult to be extracted. Therefore, reduced order models with reasonable
simplifications may be appropriate for fast assessment of the stability of
power systems. This is of special importance when the high accuracy is not a
14
main target and reasonable accuracy as well as high simulation speed is

demanded (e.g. online applications).
According to this discussion, mathematically and physically, a power may
be in one of three states; steady state, dynamics associated with small
disturbances, and dynamics associated with large disturbances. The big
questions are then: what is the threshold magnitude of a disturbance that
makes it large disturbance? What is the maximum magnitude of a slow
change at which the system is considered in a steady state? Actually, there
are no universal values of these values as they are highly dependent on the
parameters, topology, operating conditions, and the scale of the power
system. The reaction (or the response) of a power system to a specific change
defines the type of the change and the corresponding state of the system;
however, the power balance mismatch associated with a given change is
suggested in this book as an indicator to the type a change and the
corresponding state of a power system. Generally, the steady state usually
fulfills the secure operation requirements i.e. the normal state requirements
(stated in volume 1/chapter 1). Switching actions such as distribution
capacitor bank or small load switching may cause limited local disturbances
that may not propagate through wide parts of a power grid and vanish fast.
Therefore, their impact on the power system is insignificant. Changes in a
power system may be classified into three categories,
I. Slow changes with small magnitudes: the reaction of a power system to

such changes is insignificant and the system may be considered in a
steady state. In this case, time-independent mathematical models are
sufficient for representing power system. These types of models are
called ‘load flow models’. The system is then considered in a steady
state if it could continually keep the power balance or in other words
the active and reactive power mismatch (P, and Q) is practically
zero. The power mismatches are defined by,
15
II. Fast (or sudden) changes with small magnitudes: In this case, the ratio
of the active and reactive power mismatches to the respective total
generation ( are small. In this case linearized
modeling of the system nonlinearities is usually acceptable. For the
linearized modeling to be acceptable, the steady-state deviation of the
system variables should be small (e.g.  10% of the initial steady state
values).
III. Fast (or sudden) changes with large magnitudes: In this case, the ratio
of the active and reactive power mismatches to the respective total
generation ( are large. System nonlinearities
should be carefully modeled for such kind of disturbances. Large
disturbances usually cause violations in the normal state requirements
and the system in this case is said to be in an emergency state or
extreme state as defined in chapter 1 of the first volume of this
textbook.
It should be noted that the impact of active power mismatches is

significantly different in comparison with reactive power mismatches 2. Active
power mismatches significantly affect the frequency of the system while their
impact on the voltage is minor. On the other hand, reactive power
mismatches have a significant impact on the voltage while their impact on the
system frequency is minor. From disturbance spread point of view, active
power mismatches cause widespread impact on the system performance
while reactive power mismatches have a local impact. For example, an active
power mismatch causes frequency upset all over the power grid while a
2
Further readings:
 Kundur, P. (1994). Power system stability and control (Vol. 7). N. J. Balu, & M.
G. Lauby (Eds.). New York: McGraw-hill.
 Venikov, V. A. (1977). Transient processes in electrical power systems. Mir
Publishers.
 Machowski, J., Bialek, J., & Bumby, J. R. (1997). Power system dynamics and
stability. John Wiley & Sons.
 Savulescu, S. C. (Ed.). (2014). Real-time stability in power systems: techniques for
early detection of the risk of blackout. Springer.
 Kundur, P., Paserba, J., Ajjarapu, V., Andersson, G., Bose, A., Canizares, C., &
Vittal, V. (2004). Definition and classification of power system stability
IEEE/CIGRE joint task force on stability terms and definitions. Power Systems,
IEEE Transactions on, 19(3), 1387-1401.
16
reactive power mismatch causes local voltage problems at the mismatch bus
and its vicinity.
4.3 Frequency dynamics of power systems

Understanding of the frequency dynamics of power systems 3,4,5,6 is
essential for several issues including frequency control design and estimation
of the system inertia through measurements. Consider a power system with
several conventional synchronous generators. If this power system is
subjected to a disturbance such as a sudden outage of a generator, then
dynamical changes in the system start instantaneously. These dynamics are
mainly caused by the instantaneous power imbalance between the
instantaneous generation and consumption of electric power. Consequently,
the remaining synchronous generators are subjected to acceleration and
deceleration effects. Since a generator’s turbine governor cannot increase the
input mechanical power instantaneously, hence the additional power has to
be extracted from the kinetic energy stored in the rotor. Since each
generator has an inertia, then its power angle (or the angle of its induced emf;
) cannot change as rapidly as needed for exactly restoring the power
balance. Consequently, a disturbed generator will be subjected to the power
and speed swings (or oscillations) which are of decaying amplitudes in stable
dynamics. Due to the strong connection between the mechanical and
electrical frequency, the changes in the rotor speed results in changes in the
electrical frequency.
As an example, the frequency behavior of a generic power system with
frequency controls following a loss of generation is shown in Fig. 4.1.
Initially, the system electrical loading is higher than the mechanical input
power inputs. Therefore, the speed of generators and grid frequency decrease.
Consequently, the remaining generators increase their output power by
3
Ravalli, P. (1986). Frequency Dynamics and Frequency Control of a Power System,
Particularly Under Emergency Conditions (Doctoral dissertation, University of
Melbourne).
4
Andersson, G. (2012). Dynamics and control of electric power systems. Lecture notes
227-0528-00, ITET ETH, EEH - Power Systems Laboratory, ETH Z¨urich
5
Eremia, M., & Shahidehpour, M. (Eds.). (2013). Handbook of electrical power system
dynamics: modeling, stability, and control (Vol. 92). John Wiley & Sons.
6
Wall, P., Gonzalez-Longatt, F., & Terzija, V. (2012, July). Estimation of generator
inertia available during a disturbance. In Power and Energy Society General Meeting,
2012 IEEE (pp. 1-8). IEEE.
17
releasing kinetic energy stored in their rotors (Fig. 4.1: from t = 0 sec to 10
sec). As a result, the power imbalance is reduced while the turbine governors’
increase the input mechanical power to the generators. The lowest frequency
(or the frequency nadir) is reached when the mechanical power equals to the
electrical power). Since at the frequency nadir the frequency is still below the
normal value, then the governors increase the mechanical power input to the
generators. Therefore, the initially lost kinetic energy values from the rotors
are now starting to be restored and the system frequency increases.
Eventually, the power balance and the frequency are restored in the system
has sufficient capacity to compensate the lost generation. This control action
is called primary Automatic Load Frequency Control (ALFC)7. This
frequency control loop causes a stationary frequency error determined by the
droop characteristics (or function) of the turbine governor. Fine tuning of the
system frequency is achieved by another control loop called the secondary
ALFC. These control loops are shown in Fig. 4.2. The overall objective of
the ALFC is to maintain an active power balance (or load tracking) at the
generator buses by maintaining a constant frequency. Consequently, the
ALFC systems collectively maintain active power balance and frequency on
the system wide basis. The response of the primary ALFC loop is faster in
comparison with the secondary ALFC loop. The primary ALFC loop
responds to frequency (or speed) error signal, which is an indirect measure of
the active power imbalance. Via a speed governor and control valves, the
steam (or hydro) flow is regulated with the intent of achieving an active
power balance to relatively fast load fluctuations. On the other hand, the
secondary ALFC loop maintains fine adjustment of frequency (i.e. it
eliminates the small frequency error which still remaining after the action of
the primary ALFC loop); however, the secondary ALFC is insensitive to
rapid load and frequency variations, but it responds to drift-like (or slow)
changes that take place over a period of minutes.
The mechanism by which a generator supplies power to the grid is
summarized in the following points. Consider a generator operating in its
normal state, delivering a steady state active power Pgo MW to the network.
In this case,
7
Elgard, O. I. (1982). Electric energy systems theory. New YorkMc Graw-Hill.
18
 The generator currents and rotor magnetic field create a constant

electro-mechanical decelerating torque Tgo = Pgo/mo, where mo is the
rational speed of the turbine-generator in rad/sec. The generator torque
Tgo is also referred to the counteracting electromagnetic torque (Teo)
 The turbine delivers a constant accelerating torque TTo which, if
expressed in turbine power, amount to Pto = mo Tto. The turbine
torque is also referred to the mechanical input torque (Tm).
 Under the stated conditions, both torques (and powers) are in complete
balance and the frequency fo is maintained constant i.e. Tt = Te.
Assume now this equilibrium is suddenly upset by an electrical load

change, Pg. This a typical situation when either the local load is changed by
an active power PD or the transmission power flow is changed by Pline
(which may be caused by faults) or both.
 As a result the generator power changes instantaneously with the
amount of PG electrical power balance require that Pg = PD +
Pline
 In the moment following this electrical change no changes take place
in the turbine torque due to the time delay associated with the response
of the prime mover. Consequently, the turbine-generator experience a
slight torque or power imbalance.
 If the generator’s load change is positive (load increase), the turbine-
generator will decelerate, and if Pg is negative, the turbine-generator
will accelerate. In either case the generator frequency will undergo a
change f, which thus becomes the indicator of the existing power
imbalance.
 The ALFC loops are designed to maintain the power balance by an
appropriate adjustment of turbine torque.
The frequency shown in Fig. 4.1(a) is the system frequency which is the
average frequency of the system. As shown in Fig. 4.3, the frequencies of the
individual generators can be regarded as comparatively small variations over
the system frequency. The system frequency is also called the frequency of
the Center of Inertia (COI).
19
(a)
(b)
Fig. 4.1: Frequency and power dynamics of an interconnected power system
as subjected to a generator outage. (a) Frequency dynamics. (b) Active power
and mechanical power dynamics
Fig. 4.2: Model for the ALFC loops 8
8
Elgerd, O. (1981). Control of electric power systems. IEEE Control Systems Magazine,
2(1), 4-16.
20
Fig. 4.3: Average frequency of the system and frequencies of individual

generators
The frequency dynamics can be classified according to the strength of

the disturbances in two categories. The first category includes dynamics
associated with small disturbances (such as normal small load variations)
while the second category concerns with large disturbances (such as faults).
Despite the strength of the disturbance, the frequency deviations are
associated with the power imbalance in the system. Practical examples 9 of
each category are shown in Fig. 4.4 and 4.5. Most utility systems set a
maximum threshold of ±1% frequency variations for continuously supplying
system loads. Therefore, frequency variations of amplitudes less than ±1%
will not be considered as a problem and no control is required for such cases;
however, larger variations need to be controlled for restoring the system
frequency to the normal bandwidth.
9
Eremia, M., & Shahidehpour, M. (Eds.). (2013). Handbook of electrical power system
dynamics: modeling, stability, and control (Vol. 92). John Wiley & Sons.
21
Fig. 4.4: Typical small frequency variations 7
Fig. 4.5: Large frequency variations resulting from a generation trip of

1250 MW in the WECC on May 18, 20017
As previously mentioned, the inertia of the generators, and the ALFC

control loops as well as other controls such as the Automatic Generation
Control (AGC) define the frequency control actions as illustrated in Fig. 4.6.
In the following sections, the modeling, analysis, and computer simulation
of the dynamics of power systems will be presented considering simplified
models and simple systems. This chapter considers the SMIB while multi-
machine power systems will be considered in the next chapter. The main
objective of the chapter is providing the fundamental understanding of the
dynamics of power systems, control design, system simulation, and analysis.
22
(a)
(b)
Fig. 4.6: Frequency variations; (a) Variations under small and large
disturbances; (b) various ranges of frequency deviations and associated
control actions
4.4 Linearized classical modeling and simulation of the traditional SMIB

system
The Single Machine Infinite Bus (SMIB) system is popularly used in the
simplified analysis of the performance of a generator connected to an infinite
bus (a bus with constant voltage magnitude and frequency; represents a large
power grid connected to the generator); see Fig. 4.7.
23
Fig. 4.7: The traditional SMIB system
In the following it is assumed that,

1. The normal-state automatic controls (shown in Fig. 4.8), such as
Automatic Excitation Regulators (AER) and governors, are not
active. This assumption is valid for the first instants of a
transient process due to the inherent time delay associated with
controlled quantities and their control systems.
2. The network active power losses are neglected. This assumption
is commonly acceptable as the X/R ratio of power transmission
systems is high values (typical value ranges from 3 to 10 for
HVAC cables and from 5 to 10 or more for HVAC overhead
line). Therefore, the transmission network, which is commonly
constructed as an overhead system, is usually considered as a
pure reactive network.
3. The synchronous generator is represented by classical model
(shown in Fig. 4.9). With this model, the transient emf of the
generator is assumed to be constant and the field circuit
dynamics are neglected during the transient process. An
assumption which is valid for the initial stage of the transient
process; however, in reality the transient emf and the machine
equivalent reactance changes with the time during the transient
process. The assumption is shown to provide high simplicity
with reasonable accuracy.
4.4.1 The swing equation

For stability analysis, the shaft connecting the rotating elements in a
turbine-generator is usually assumed rigid and the free-body rotation is
assumed. Therefore, the total inertia of the rotating elements is simply the
sum of the individual inertia of various connected rotating elements such as
the turbine, the rotor, and the exciter. The equation of motion of such rotating
24
system obeys the Newton’s second law of rotational motion which takes the
form,
where J is the total moment of inertia in kg.m2, m is the rotor shaft

velocity in rad/s, Tt is the turbine or input torque in N.m, Te is the
electromagnetic or the counteracting torque in N.m, and TD is the damping
torque in N.m.
(a)
(b)
Fig. 4.8: Normal-state controls for conventional synchronous generators; (a)
block diagram; (b) connection layout
25
Fig. 4.9: Classical dynamic model of conventional synchronous generators
This equation is called the equation of motion or the swing equation. The
term represents the rotor acceleration. According to the torque
balance presented on the RHS of eq. (4.3), the rotor will accelerate if the
RHS term of eq. (4.3) is positive while rotor deceleration is associated with a
negative value of that term. In case of perfect torque balance (i.e. the RHS of
eq. (4.3) = 0), then the rotor will rotate at a fixed speed which is the
synchronous speed. It is clear from eq. (4.3) that the impact of torque
imbalance on the rate of change of the rotor speed is inversely proportional to
the monument of inertial. Therefore, for the same torque imbalance subjected
to two identical generators except their moment of inertia, the generator
having a higher moment of inertia is less sensitive to torque imbalance in
comparison with the generator with lower inertia. It is also important to note
that the changes in the turbine torque are significantly slow in comparison
with the electromagnetic torque. This is attributed to the long time constants
associated with the prime mover system while the electromagnetic changes
are almost instantaneous.
Since the power angle ((shown in Fig. 4.9) represents the rotor position
with respect to a synchronously rotating reference frame axis (see Fig. 4.10),
then the rotor speed as a function of the rate of change of (takes the form,
26
Fig. 4.10: the power angle 
In eq. (4.4), m presents the mechanical angle of the rotor in mech. rad.,
m presents the rotor speed in mech. rad/s. Based on (4.4) can be written as
. It is clear that the deviation in the rotor speed can be
represented by . The damping torque in (4.3) can be
presented by,
where Dm is the damping torque coefficient in N.m.sec.
Substituting (4.4) in (4.5) and the resulting equation as well as eq. (4.4) in
(4.3), then
Since, the synchronous speed is a constant value, then
The term presents the net mechanical shaft torque (Tt) and
the term presents the rotor speed deviation , then (4.7)
becomes,
27
Since the power = torque x angular speed, then .

If small disturbances are considered, then the speed deviations may small
enough for neglecting its impact on the power-torque interrelation. Therefore,
and eq. (4.8) becomes the swing equation,
where is the damping torque coefficient and Mm = Jsm

which is the angular momentum of the rotating elements at the synchronous
speed. Pm is the shaft net power input to the generator, and Pe is the
electromagnetic power which is equal to the air gap power.
It is worthy to be noted that the angular momentum Mm is related to
kinetic energy (K.E) stored in the rotating elements in the synchronous speed.
This K.E is usually called the inertia constant H in MJ. The inertia constant H
‘quantifies the K.E stored in the rotating elements at the synchronous speed
( ) in terms of the seconds it would be taken by the generator to provide
an equivalent amount of energy when operating at rated power’. With Sn
defined as the rated MVA of the generator, the inertia constant H and the
momentum Mm take the form,
Therefore, the momentum equals to double the value of the K.E of the
rotating elements. The momentum can also be presented in terms of the
mechanical time constant . The mechanical time constant can be explained
by considering a generator at rest and suddenly a mechanical torque equals to
is applied to the shaft. In this case, the rotor speed will be linearly
increased and the rotor will take a time to reach the synchronous speed.
28
The previous modeling of the swing includes mechanical quantities while

it is preferable to express the model using electrical quantities. To do so, the
following relations are to be used, the mechanical angle = number of
pole pairs (p) x electrical angle , and mechanical angular velocity ( ) =
number of pole pairs x electrical angular velocity ( ) i.e.
Since all the upcoming equations will include electrical quantities, the
suffix e will be removed. Equation (4.9) can be transformed to include the
electrical quantities as follows,
or
where
The swing equation can be represented in p.u by dividing (4.16) by Sn,
In this representation, the inertia constant in seconds as well

as the inertia constant H. The synchronous speed s = 2fo in rad/s and fo is
the normal state frequency in Hz. Therefore, or M = 2 H with
the synchronous speed is taken as 1.0 p.u. The power values are in p.u while
the angles are in rad. Therefore, (4.17) can be written as,
29
H d 2
 Pm  Pe  PD (4.18)
 f o dt 2
Although we assumed that the damping coefficient is a constant value

such that the damping power takes the form , the damping
coefficient is a nonlinear function of the power angle highly dependent on its
value. For small deviations in the rotor speed, it can be shown that 10,
 
PD  Dd sin 2   Dq cos2    D ( ) (4.19)
The function D() takes the form shown in Fig. 4.11. It is depicted from
the figure that the damping increases with the increase of the power angle till
a power angle of 90o. Afterward, the damping declines. It is known that the
maximum allowable steady state power angle of an uncontrolled cylindrical
pole machine is 90o; higher angle values cause instability of the machine. In
most types of power systems studies, the average value Dav of the damping
coefficient is used instead of representing the actual variations of the
damping coefficient.
Fig. 4.11: Variations of the damping coefficient with the power angle
The value of the damping coefficient D depends on many factors such as:
10
Machowski, J., Bialek, J., & Bumby, J. R. (1997). Power system dynamics and stability.
John Wiley & Sons.
30
1. The induction motor action (i.e. the electromagnetic damping) due to

the damper winding and rotor body eddy currents.
2. Speed – torque characteristics of the prime mover.
3. Parameters of the excitation control system.
4. The load characteristics.
5. The loading conditions.
The instabilities that may be seen with small disturbances are always
attributed to low damping of oscillations.
4.4.2 The transient power-angle characteristics

Referring to eq. (4.18), the electrical power Pe, can be represented by
the transient power-angle characteristics of the system. Considering
cylindrical and salient pole generators, the transient power angle
characteristics will be derived in the following part. Given that all resistances
neglected, the vector diagram of Fig. 4.12 represented the SMIB (Fig. 4.7)
with cylindrical (non-salient) pole machine while Fig. 4.13 shows the vector
diagram of the system considering a salient pole generator. It will be used for
obtaining the transient power-angle characteristics. The details of the
derivations are presented in Appendix 1. In addition, the steady state power-
angle characteristics of various types of conventional synchronous generators
can be found in Appendix 1. The equations are derived based on the vector
diagrams of voltage and currents.
As shown in Appendix 1, the steady state and transient power-angle
characteristics of various machine types and their corresponding plots are
shown in Table 4.1.
The power-angle characteristics representing the SMIB system is described
by equation (4.21) for non salient pole generators and by (4.23) for salient
pole generators. In comparison with the steady state characteristics of the
non-salient pole generator, it is seen from Fig. 4.15 and 4.17 that the
maximum angle increased above 90 o. This is attributed to the impact of
saliency during the transient state even with non-salient pole machines.
Detailed explanations can be found in reference 11. In addition, the saliency
11
Krause, P. C., Wasynczuk, O., Sudhoff, S. D., & Pekarek, S. (2013). Analysis of electric
machinery and drive systems (Vol. 75). John Wiley & Sons.
31
causes an increase of the maximum power. For simplification, it is assumed

in the following modeling and analysis that,
Fig. 4.12: vector diagram of the SMIB system with cylindrical pole generator
Fig. 4.13: vector diagram of the SMIB system with salient pole generator
32
Table 4.1: Power-angle characteristics of the SMIB system considering

non-salient and salient pole generators
Steady state characteristics Transient characteristics
V 2 1 1 
'
E qV EV
P sin  (20) P
q
sin    '   sin 2 (21)
X d X d'  2  X d X d 
Non-salient pole generator
Fig. 14 Fig. 15
E qV V 2 1 1  E q'V V 2 1 1 
P sin      sin 2 (22) P sin    '   sin 2 (23)
X d 2  X q  X d   X d'  2  X d  X q  
Salient pole generator
Fig. 16 Fig. 17
 The difference (’) shown in Fig. 4.12 and 4.13 is small enough to
consider that,
E q'  E ' cos(   ' )  E ' (4.24)
33
 It is also assumed that the saliency effects on the characteristics are

negligible. Therefore, pure sinusoidal characteristics are obtained.
Therefore, the transient power-angle characteristics of salient and non-
salient pole machines can be represented approximately by,
E 'V
Pe  ' sin   Pmax sin  (4.25)
xd
Fig. 4.18 illustrates a comparison between exact and approximate power –

angle characteristics. It can be seen that the approximations result in an
increase in the maximum power and reduction of the maximum angle to 90 o.
As a result, the stability of the system is overestimated; however, the
mathematical treatment will be simplified. For performance demonstration
purpose, the approximations are acceptable; however, practical applications
should consider the impact of these approximations on the results.
Fig. 4.18: Exact and approximate power-angle characteristics
4.4.3 Linearization and linearized analysis

The swing equation (18) can be now written as,
H d 2 d
 P  P sin   D (4.26)
 f o dt 2 m max
dt
34
The swing equation is then a nonlinear function of the power angle ;

however, for small disturbances, the swing equation may be linearized with
little loss of accuracy. The swing equation can also be written in the form of
two first-order differential equations as,
f o
  Pm  Pe  D 
H (4.27)
  
where  is a differential operator i.e. d/dt operator.
Consider a small deviation oin power angle from the initial operating point
o, i.e.
 (4.28)
Linearizing (4.27) about an initial operating condition represented by o as

follows:
f o
  Pm  Pe  D 
H (4.29)
  
where
Pe
Pe    Pmax cos  o (  ) (4.30)
 o
The term Pmax cos o is known as the synchronizing power coefficient. The
higher the synchronizing power coefficient, the more stable the system. The
Ps is given by:
Ps  Pmax cos  o (4.31)
Therefore, (4.29) becomes,

35
f o
   Pm  Ps   D   (4.32)
H
  
Equation (4.32) can be written matrix form:
 f o f o 
 
  f o 
D Ps 
d    H H
      H  Pm (4.33)
dt          0 
 1 0 
Which takes the standard state-space form form:

 x  A x  B u (4.34)
It is clear that the state matrix A is dependent on the system parameters D,

H, X’d, and the initial operating condition represented by the value of E’ and
o. The block diagram representation of the linearized swing equation is
shown in Fig. 4.19. This diagram can be implemented, for example, on the
Simulink for the simulation of the small-signal performance of the SMIB.
Fig. 4.19: Blok-diagram of the SMIB system with classical model

linearization
36
Example
Simulation and analysis of the linearized classical model of the SMIB
system
As an example, the SMIB system shown in Fig. 4.7 consists of a 50-Hz

equivalent generator having inertia constant H = 10 sec and a transient
reactance of 0.3 p.u is connected to an infinite bus through a purely reactive
circuit with Xe = 0.4 p.u. The generator is delivering power of 0.6 p.u at 0.8
lagging power factor to the infinite bus at 1.0 p.u voltage. The damping
power coefficient is assumed to be D = 0.1 p.u. The linearized model of the
system is implemented on the simulink as shown in Fig. 4.20. Based on the
models presented in Appendix 1, various values of the variables are
calculated. It is found that o = 18.5940o, E’ = 1.3172 p.u, and Ps = 1.7835
p.u. The following simulations are performed to show the response of the
system and for the analysis of the system’s frequency response. In addition,
the impact of various quantities on the system performance is demonstrated.
The system linearized model is implemented on the simulink as shown in
Fig. 4.20. The following cases are considered.
Fig. 4.20: The simulink model of the linearized dynamic model of the
SMIB system of the numerical example – the machine is represented by the
classical model
37
 Case #1: The response of the system as subjected to a sudden

decrease in the mechanical power input of 10%. The disturbance is started
at t = 0.5 sec from the simulation time. This is to show the initial conditions
in the pre-disturbed conditions. The results are shown in Fig. 4.21.
 Due to the disturbance, an oscillatory stable transient process occurs
and the system settles to a new steady state point after about 5 seconds from
the instant of the disturbance.
 Since, the disturbance is a sudden reduction in the mechanical power
input, it seems that the situation is impossible in reality due to the inherent
relatively long time constant of the prime mover subsystem; however,
reduction in the mechanical power input as a disturbance is used here for only
showing the transient response of the SMIB system.
 The disturbance causes a negative mismatch in the power balance
(defined by the RHS of the swing equation; eq. (4.17)). Therefore, the
generator initially decelerates and its speed as well as its frequency is
reduced. It is clear from the transient power-angle characteristics (eq. (4.25))
that the electrical active power is directly proportional to the sine of the
power angle. Therefore, as expected, the disturbance initially causes
reduction in the power angle due to the input power reduction. The electrical
power is also reduced. The damping power - which is mathematically a
scaling of the speed deviation by the damping coefficient (see eq. (4.19)) - is
shown, as expected, to be in-phase with the speed deviation. In addition, the
electrical active power and the power angle are in-phase.
 Case #2: in this example, the impact of the damping on the system
response is evaluated. The original value of the damping coefficient of the
system is assumed to be 0.1; new values are given to the damping coefficient
for the same disturbance of case #1 and comparative analysis is performed.
The new values are 0.2, 0.0, and -0.1 and the results are shown in Fig. 4.22
and Fig. 4.23. The value of the damping coefficient associated with each
curve is shown in the figures.
38
Time (s)
Fig. 4.21: Response to a step change of -10% in the mechanical power
input
Time (s)
Fig. 4.22: Impact of the positive values of the damping coefficient on the
linearized transient performance of the SMIB system – Classical model (D
are the shown numbers)
39
Fig. 4.23: Impact of the negative value of the damping coefficient on the
linearized transient performance of the SMIB system – Classical model
 It is clear from Fig. 4.22 that the amplitude and the settling time of the
transient oscillations are highly dependent on the damping coefficient value.
The higher the damping coefficient, the better the response (lower amplitudes
and settling time). It is also shown that the frequency of the oscillations is
insensitive to the damping coefficient.
 Based on Fig. 4.22, and 4.23, it can be easily depicted that the positive
damping results in damped oscillatory response (a stable situation) while the
negative damping is associated with oscillatory response with increasing
amplitudes (instability situation). The case of zero damping is associated with
sustained oscillations of constant amplitude (critical or marginal stability).
 Negative damping can be realized in reality in the situations where the
parameters of the generator controllers are incorrectly tuned. In addition,
positive damping is associated with some types of loads.
 Zero damping is a sort of a theoretical assumption or a situation where
negative damping sources compensate the positive damping of the system.
Generators have the inherent damping capability due to their electric power
40
losses, and eddy current flow in the magnetic core of the machine as well as
other energy losses such as windage, and friction.
 Case #3: in this study the impact of the generator’s inertia on the
transient response is considered. In comparison with the data of case #1, in
this case only the inertia constant of the generator is changed. The original
value of the inertia constant H = 10 sec. In this case, two additional values
are considered which are 5 sec, 20 sec, and 100 sec. The simulation results
are shown in Fig. 4.24.
 It is clear from the results that increasing the inertia reduces the
magnitudes of the speed oscillations; however, the frequency and the settling
time of the speed oscillations increase. This is attributed to the increase in the
K.E stored in the rotating system with the increase of the inertia constant.
This means that the energy needed to alter the speed of the generator.
Consequently, the amplitude decreased with the increase in the inertia while
restoration of the normal K.E takes more time, which results in increasing the
settling time and reducing the frequency of the speed oscillations.
 From the power angle deviations point of view and considering the
increase in the rotor K.E energy with increase in the inertia, it is clear that the
increase of the inertia increases the amplitudes of the angle oscillations. In
addition, the settling time is increased and the frequency of the angle
oscillations is reduced. Same comments can be given to the electrical power
due to its tight relation to the power angle.
 Although the increase of the inertia constant causes some drawbacks
such as a reduction in the frequency of oscillations and increase of the
settling time, the speed (or the frequency) deviations are reduced. This is of
great importance in keeping the system synchronized. Therefore, with higher
inertia, a disturbance is unlikely to move the system to the extremes state in
comparison with a system with lower inertia.
 From the damping power point of view and considering the impact of
increase the inertia on the speed deviations, higher initial amplitudes of the
damping power are available with lower inertia values. This is because the
damping power is directly proportional to the speed deviations. Therefore,
increasing the damping ratio of low inertia generators can be a good measure
for enhancing its stability. This is can be achieved via Power System
Stabilizers (PSS) for the conventional synchronous machines or generally
41
with Power Oscillation Damping (POD) controllers. PODs can be designed

to improve the system damping and can be placed at any location on a power
grid and they are usually integrated with FACTS devices such as SVCs or
TCSCs12,13. The design of PSSs and PODs will be considered in the
upcoming chapters of this textbook.
Fig. 4.24: Impact of the inertia constant on the linearized transient

performance of the SMIB system – Classical model (H are the shown
numbers)
12
M. Mandour, M. EL-Shimy, F. Bendary, and W. Ibrahim, “Damping of Power Systems
Oscillations using FACTS Power Oscillation Damper – Design and Performance
Analysis”, 16th International Middle East Power Systems Conference (MEPCON’14),
Dec. 23-25, 2014, Cairo, Egypt, pp. 1- 8.
13
M. Mandour, M. EL-Shimy, F. Bendary, W.M. Mansour. Impact of Wind Power on
Power System Stability and Oscillation Damping Controller Design. Industry Academia
Collaboration (IAC) Conference, 2015, Energy and sustainable development Track, Apr. 6
– 8, 2015, Cairo, Egypt
42
 Case #4: In this example the impact of the grid coupling strength on
the performance of the SMIB system is considered. The grid coupling
strength is inversely proportional to the value of the network connecting the
generator to the infinite bus. In the considered simplified model, the
equivalent reactance network is the external reactance (xe). If xe is small, then
strong coupling is achieved while high value of xe is referred to weak
coupling. With the SMIB system data as given in case #1, only the external
reactance is given some new values. These considered values are 0.1 p.u, 0.4
p.u (the original value), 0.7 p.u, and 1.0 p.u. Consequently, the values of the
initial power angle, the transient emf, and the synchronizing power are
changed. Table 4.2 lists these values for various values of the external
reactance. These values are shown in Fig. 4.25 and the simulation results are
shown in Fig. 4.26.
Table 4.2: Impact of variations of xe on E’, o,and Ps

xe (p.u) E’(p.u) o (deg) Ps(p.u)
0.1 1.1916 11.6190 2.9180
0.4 1.3172 18.5940 1.7835
0.7 1.4318 24.7751 1.3000
1.0 1.5379 30.4775 1.0195
 The changes shown in Table 4.2 and Fig. 4.25 reveal some important
stability and operational issues. Fig. 4.25(b) shows a general shape of the
synchronizing power and transient power functions. This figure shows that
increasing the electrical power is associated with an increase in the power
angle and a decrease in the synchronizing power. In addition, Table 4.2
shows that increasing the external reactance results in a decrease in the
synchronizing power. Therefore, the synchronizing power decrease with
either an increase in the generator loading or decrease in the network
coupling strength or both of them. Actually, Fig. 4.25(b) shows that the
increase in the external reactance simultaneously increases the power angle
and decreases the synchronizing power. As previously stated and also shown
in Fig. 4.26, better transients are associated with higher synchronizing power.
The phrase better refers to lower the amplitude of oscillations, and settling
time. For example, in Fig. 4.26, the increase in the external reactance which
43
(as shown in Table 2) decreases the synchronizing power, results in an

increase in the amplitude of the speed oscillations and reduction in their
frequency of oscillations.
(a) Values of main variables (b) Impact on small signal stability
Fig. 4.25: Impact of the external reactance on the variables and stability
SMIB system
 Fig. 4.25(b) also shows that increasing the external reactance requires a
significant increase in the transient emf (the steady state emf also should be
increased as shown in Table 4.1). This required increase in the emfs
mandates an increase in the design rating of the generator excitation system.
In addition, for a given design, increasing the emf reduces the active power
production capability. This is because the active and reactive power of a
generator are dependent quantities and related to the generator MVA rating
by . From a practical point of view, the reactive power needed
for compensating long lines is not only provided by the centralized
generators. Instead, distributed reactive power compensators are installed
along the long lines for reactive power support, voltage control, and voltage
stability enhancement.
44
Fig. 4.26: Impact of the grid-coupling strength on the linearized transient

performance of the SMIB system – Classical model (xe are the shown
numbers)
 Case #5: The given linearized model can be used for representing
some other disturbances such as a sudden permanent disconnection of a
line. Consider the SMIB system shown in Fig. 4.27 in which two lines
connect the generator to the infinite bus. The generator normal frequency is
50-Hz generator and having an inertia constant M = 2H = 20 sec. The
transient reactance equals to 0.3 p.u. The generator is connected to the
infinite bus through a purely reactive circuit with XT = 0.2 p.u., and XL = 0.4
p.u. The generator is delivering power of 0.6 p.u at 0.8 lagging power factor
to the infinite bus at 1.0 p.u voltage. The p.u damping coefficient is assumed
to be 0.10. In this example, it is assumed that the line outage disturbance is
small enough for the linearized model to be valid.
45
Fig. 4.27: The SMIB system of case #5
 Recalling that with the classical model, a constant transient emf is

assumed during the transient process. Therefore, before and after the outage
of the line, the transient emf will not be changed; however, the transient
power-angle characteristics will be changed as shown in Fig. 4.28. In that
figure, the curve number I refers to the pre-outage characteristics while the
curve number II refer to the post-outage characteristics. Since, the outage of
the line results in an increase in the transfer reactance (xe) between the
generator and the infinite bus, the maximum power of curve II is low in
comparison with curve I. This shows that the line outage can be simulated by
an equivalent sudden change in the mechanical power. Numerical values and
calculation methodology are shown below.
Fig. 4.28: Pre-outage and post-outage transient power-angle

characteristics
 Based on the models of Appendix 1,

46
E 'V
P  I sin  Pmax
I I
sin
X 12
The pre-outage (I) system reactance = X d  X 12
'I I
= 0.3 + 0.4/2 + 0.2 = 0.7 p.u.
The transient emf E’ can be obtained using (34), it is calculated

at prefault conditions.
2 2
P I   Q I 
E   o X 12
'
  V  o X 12 
V   V 
Hence, E’ = 1.380 p.u. = constant. Therefore,
PI = 1.97 sin 
The initial operating angle =  oI  sin1 Po Pmax

I

= 17.732  o
=
0.309485 rad.
II
The post-outage reactance = X 12 = 0.9 p.u.
The post-outage approximated transient power angle
characteristics is then:
E 'V
P  II sin  Pmax
II II
sin = 1.53 sin 
X 12
Based on Fig. 4.28, it can be concluded that at the instant of line

tripping, a sudden change in the system topology occurs
resulting in a sudden change in power-angle characteristics of
the system. The operating point moves suddenly from point (a)
to point (b) causing a sudden power imbalance between the
mechanical input and electrical output of the generator; Pmo.
The operating point corresponding to point (b) cannot be stable
due to the mismatch between mechanical and electrical power.
As the difference is positive the rotor accelerate following the
post-outage characteristics (II) and then after some power
47
oscillations the system reaches a new steady state operating

point corresponding to point (c) in Fig. 4.28. Where

 oII  sin1 Po Pmax
II

is the final steady state angle (after power-
oscillations settles) = 23.089 o = 0.402975 rad. The initial change
in generator input power

Pmo  P I  P II    oI
 Pmax
I
sin oI  Pmax
II
sin oI
Pmo = 1.97 sin( 17.732) – 1.53 x sin(17.732) = 0.1340 p.u
Fig. 29: Simulation of a line outage by an equivalent power imbalance
Therefore, the considered disturbance can be simulated in the same way as

the previous cases. In this case, Ps  Pmax
II
cos  oI = 1.457 p.u. The results are
shown in Fig. 4.29.
48
 From a physical point of view, the line outage causes an increase in xe.
Therefore, the final steady state angle is expected to be high in comparison
with the initial steady state angle. The response verifies this issue. In
addition, the sudden increase in xe causes sudden decrease in the electrical
loading on the generator. Consequently, it is expected that the frequency will
increase just after the line outage.
 Case #6: simulation of a temporary outage of a line. This situation

may be faced due to, for example,
temporary fault or operation of Automatic
Circuit Reclosers (ACR). In this case, the
mechanical power input remains
unchanged during the transient process.
The disturbance in this case can be
simulated by an equivalent sudden
change in the power angle. As shown in
Fig. 4.30, the initial operating angle of
the system is ; corresponding to point Fig. 4.30: temporary line
(a) on curve I. When the line is outage
disconnected, the transient power curve changed to curve II and the operating
point moves to point (b) on curve II. It is assumed that the line will return to
service after a very short duration. In the figure, it is shown that the line
reconnected at an angle see which is corresponding to point (c) on curve II.
Now, the system characteristic return to curve I and at the instant of line
reconnection, the operating point moves to point (d) on the curve I. Due to
the power mismatch, a transient process acts and afterward the system
operating point settles at the initial operating point . The overall changes in
the system due to the disturbance are null; however, the difference (
acts as a disturbance which is equivalent to the momentary outage
of the line. In this case, the data of the system of case #5 is considered such
that one of the lines is temporary disconnected and the equivalent sudden
angle change equals to 10 degrees (0.17453 rad). The original simulink
model is modified as shown in Fig. 4.31 by setting an initial non-zero angle
which equals to o. Since the transient process will mainly act on the curve
49
I. Hence, = 1.8764 p.u. The results are shown in Fig. 4.32

which are in confirmation with the stated analysis of this case (Fig. 4.30).
The response of the system for momentary disturbances is called zero-input
response while the response for sustained disturbances is called forced
response.
Fig. 4.31: The Simulink model for simulation of a momentary outage of a

line in the SMIB system – classical model.
Fig. 32: Simulation of a momentary outage of a line in the SMIB system

(zero-input response) – classical model.
50
By the end of this numerical example, it is worthy to be noted that the

conclusions derived from the analysis of the presented simple system and
simplified models are still valid for larger system and more complex models.
This is will be demonstrated later in this textbook.
4.4.4 Time-domain solutions of the linearized model

Fortunately, it is possible to derive the analytical time domain expressions
that describe the solution of the linearized swing equation (4.26). The
equation presents a linearized standard second order differential equation and
its solutions will be presented in the following part. The solutions
corresponding to various zero-input and forced responses will be considered.
From the block diagram of Fig. 91, we have in the s-plane
1  f 
   o  Pm  Ps   D   (4.35)
s  Hs 
Eliminating  from (4.35) using s results in:
1 f 
   o Pm  Ps   Ds  (4.36)
s  Hs 
Rearranging (4.36) in the form of a quadratic equation results in
f o D f o Ps f o
s 2 (  )  s(  )  (  )  Pm (4.37)
H H H
Therefore, the characteristic equation is given by:
f o D f o Ps
s2  s  0 (4.38)
H H
Equation (4.38) is similar to the standard 2nd order differential equation of the
general form:
51
s 2  2 n s   n2  0 (4.39)
Therefore, the undamped natural frequency n in rad/sec is
f o
n  Ps (4.40)
H
and the damping ratio  is given by:
D f o
 (4.41)
2 HPs
The equations describing the free-motion zero-input response (u = 0) of the

system can then be derived as follows. Referring to the state equation (4.34),
we see that the free motion (with zero input) is given by:

 x  Ax (4.42)
Considering the SMIB system shown in Fig. 4.33, suppose that the breakers
of one of the transmission circuits open and close quickly. As previously
described, the disturbance is equivalent to a small deviation  in the power
angle from the initial operating point o i.e.
 = o +  (4.43)
Fig. 4.33: SMIB system for the analysis of the zero-input response
In this case u = Pm = 0, and the differential equation of motion becomes:
d 2  d
2
 2 n   n2   0 (4.44)
dt dt
52
with the characteristics equation:
s 2  2 n s   n2  0 (4.45)
The roots of the characteristic equation 1,2 (which are the eigenvalues of the
state matrix A) are given by:

1, 2    n   n  2  1  (4.46)
Generally, eigenvalues can be represented as
 =   j  (4.47)
where frequency of oscillation is f = /(2) Hz.
It is known that the response depends on the roots of the characteristic

equation. Three different cases can be realized as shown in Fig. 4.34.
Fig. 4.34: Possible conditions of the eigenvalues
 Case 1: 1  2 and both are real. This is case occurs when  2 > 1. The
general solution in this case is
 (t )  A1e 1t  A2 e 2t (4.48)
The constants A1 and A2 can be found from the initial conditions:
53
d (t )
 (0)  0 and  (0)  0 (4.49)
dt t 0
Therefore, (4.47) becomes:
 (t ) 
 o
2  1

2 e 1t  1e 2t  (4.50)
which is an over-damped response. This case describes a situation

where 1 = 1 and 2 = 2. In addition, 1 = 2 = 0. It is clear that three
situations can be realized according to the values of the 1 and 2.
(i) 1 > 0 and 2 > 0. In this case, as shown in eq. (4.50), the
time domain (TD) response is a rising exponent with time
constant  = 1/ as shown in Fig. 4.35. This situation, of
course, is corresponding to an unstable response.
Fig. 4.35: Response with 1 > 0, 2 > 0 and  = 0
(ii) 1 < 0 and 2 < 0. In this case, as shown in eq. (50), the time
domain (TD) response is a falling exponent with time
constant  = -1/ as shown in Fig. 4.36. This situation, of
course, is corresponding to a stable response.
54
Fig. 4.36: Response with 1 < 0, 2 < 0 and  = 0
(iii) Either 1 > 0 or 2 > 0 or both of them > 0. In this case,

as shown in eq. (50), the time domain (TD) response is a
rising exponent similar to that shown in Fig. 4.35. This
situation is corresponding to an unstable response.
 Case 2: 1 = 2 = and both are real. This is case occurs when  2 = 1.

The general solution in this case is
 (t )  e t  A1  A2t  (4.51)
Using (4.51) to find the constants A1 and A2. Then the solution
becomes:
 (t )  e t  o 1  t  (4.52)
which is a critically damped response. The situations described in case

1 are also valid in this case.
 Case 3: 1 and 2 are complex conjugate. This is case occurs when  2

< 1 resulting in an oscillatory TD solution. The roots of the
characteristic equation take the form:
1, 2    n  j d  (4.53)
55
where d is the damped frequency of oscillations,
d  n 1   2 (4.54)
and the solution takes the form:
 o
 (t )  e  nt sin( d t   ) (4.55)
1 2
where cos-1 . In this case, the TD response is oscillatory. And the

response time constant is given by:
1
 (4.56)
 n
If the response is stable then the response settles in approximately 4

times  Therefore, the settling time is s  4 . The frequency of the
oscillations takes the form.
 n  o
 (t )   e  nt sin( d t ) (4.57)
1 2
The stability of the system in this case is governed by the signs of the
real parts of the eigenvalues according to the following situations.
(i) i > 0 and   0. The time domain response of this situation
is harmonic oscillations with exponentially rising amplitude
(i.e. unstable response) as shown in Fig. 4.37.

56
Fig. 4.37: Response with i > 0 and   0
(ii) i < 0 and   0. The TD response of this situation is stable

damped harmonic oscillations as shown in Fig. 4.38.
Fig. 4.38: Response with i < 0 and   0
(iii) i = 0 and   0. The TD response of this situation is

harmonic oscillations with constant amplitude of o and
frequency  as shown in Fig. 4.39. The system in this case is
called critically stable or in bifurcation.
Fig. 4.39: Response with i = 0 and   0

57
Demonstrating problem
The reader may use the following data for applying the time domain
solutions.
a) A 50-Hz generator having inertia constant M = 2H = 20 sec and a

transient reactance of 0.3 p.u is connected to an infinite bus through a purely
reactive circuit with equivalent reactance of 0.4 p.u. The generator is
delivering power of 0.6 p.u at 0.8 lagging power factor to the infinite bus at
1.0 p.u voltage. Assume the per unit damping power coefficient is 0.14.
Consider a small disturbance  = 10o resulting of breakers open and quick
close of one of the two lines. It is required to obtain and plot the functions
describing the motion of the rotor i.e. (t) and f(t).
Hint: in this case  = 0.2025 < 1. Hence, oscillatory solution governed by
(4.55) and (4.57). The plots of the solutions take the form shown in Fig. 4.40
and 4.41. This solution is in confirmation with the TD solution obtained with
the Simulink.
Fig. 4.40: Angle response of problem 1
58
Fig. 4.41: Frequency response of problem 1
b) The data for the system of Fig. 33 are: fo = 50 Hz. H = 9.95 sec. x’d =
0.30 p.u. xe = 0.40 p.u. V = 1.0 p.u. D = 0.15. Consider a small disturbance
o = 10o. It is required to plot the response (t) and (t) for various
values of the active power delivered to the infinite bus form, say zero to 0.95
Pmax. Also, plot the functions P(), Ps(), , n(). Assume that E’ = 1.4
p.u. is kept constant for all load values. Analyze the results.
Expressions describing the TD response of the SMIB as subjected to

sudden changes in the mechanical power inputs can also be derived. The
system response in this case is a forced response in which the disturbance is
sustained. Assume that the input power is increased by a small amount Pm.
In this case Pm  0 i.e. u  0. The linearized equation of motion of the
system can be rewritten as:
d 2 d  f o
 2n  n
2
  Pm  u (4.58)
dt 2 dt H
59
The characteristic equation of the system is given by (4.39). Equation

(4.58) can be written in the state space form as,
The state variables are defined as x1 =  and x2 = . This equation can be

written as and the output takes the form
. The s-space of this model can be obtained using the Laplace
transform such that . This is shown in eq. (4.60) while eq. (4.61)
shows the solution of the state variables. The Laplace transform for a sudden
(or step) change in the input u is equals to u/s which is called U(s) in eq.
(4.60).
Based on (4.61), the state variables take the form,
The TD solutions of the s-space solutions (i.e. eq. (62) and (63)) take the
forms,
 f o Pm  1 
 (t )   1  e n t
sin  d
 t    (4.64)
H n2  1  2 
and
60
f o Pm
 (t )  e  nt sin d t  (4.65)
H n 1   2
Demonstrating problem
The reader may use the following data for applying the time domain
solutions. A 50-Hz generator having inertia constant M = 20 sec and a
transient reactance of 0.3 p.u is connected to an infinite bus through a purely
reactive circuit shown in Fig. 4.42. With XT = 0.2 p.u., and XL = 0.4 p.u. The
generator is delivering power of 0.6 p.u at 0.8 lagging power factor to the
infinite bus at 1.0 p.u voltage. Assume a per unit damping power coefficient
is 0.14. While the system is operating under steady state condition, the
breakers of one of the lines are suddenly open and remain open. Assuming
the initial power imbalance caused by the disturbance is small enough to
utilize linearized system equations, and assume that the transient emf E’
remains constant at prefault value. Obtain and plot the functions describing
the motion of the rotor i.e .(t) and f(t).
Fig. 4.42: SMIB for demonstration of the forced response
Solution hints. The disturbance can be simulated by as a step change in the

mechanical power input. The pre-disturbance transient power-angle
characteristics take the form
E 'V
P  I sin  Pmax
I I
sin = 1.97 sin 
X 12

The initial operating angle =  oI  sin1 Po Pmax
I

= 17.732 o = 0.309485 rad.
61
The post-disturbance approximated transient power-angle characteristics take the
form,
E 'V
P  II sin  Pmax
II II
sin = 1.53 sin 
X 12

The pre- and post-disturbance characteristics are shown in Fig. 4.43.
Fig. 4.43: Pre- and post-disturbance characteristics and equivalent

disturbance
The equivalent disturbance is the initial change in generator input power

Pmo  P I  P II    oI
 Pmax
I
sin oI  Pmax
II
sin oI 0.1340 p.u
Since the system dynamics occurs along the post-disturbance power angle
characteristics. Then, the synchronizing power is calculated as:
Ps  Pmax
II
cos  oI = 1.457 p.u.
The responses are then shown in Fig. 4.44 and 4.45.

62
Fig. 4.44: Angle response
Fig. 4.44: Frequency response
63
4.4.5 Small-signal stability and eigenvalues of the state matrix

The previous analysis shows the interrelation between the system small-
signal stability and the eigenvalues of the system. In addition, it is shown that
the shape of the response can be easily predicted from the eigenvalues of the
system. Although the considered system is just SMIB and the adopted model
is very simple, the derived conclusions are general regardless of the system
size and the model complexity.
Generally, the roots of the characteristic equation of a power system are
the eigenvalues of the state matrix A. The time response of a dynamic system
is dependent on the values of the eigenvalues. The eigenvalues of a state
matrix A can also be found by solving:
det(A - I) = 0 (4.66)
In fact, expansion of the determinant of (4.66) gives the characteristic

equation. The n solutions of  are the eigenvalues of A. Each eigenvalue is
corresponding to a dynamic mode and the time dependent characteristic of a
mode corresponding to an eigenvalue i is given by exp(i t). The stability of
the system is determined by the eigenvalues as follows:
1. A real eigenvalue corresponds to a non-oscillatory mode, such that:

a. A negative real eigenvalue represents a decaying mode.
The larger its magnitude ,the faster the decay.
b. A positive eigenvalue represents aperiodic instability.
2. Complex eigenvalues occur in conjugate pairs, and each pair
corresponds to an oscillatory mode. The real component of the
eigenvalues gives the damping, and the imaginary component gives
the frequency of oscillation. Such that:
a. A negative real part represents a damped oscillation.
b. A positive real part represents oscillation of increased
amplitude.
Thus, the equilibrium points may be classified as follows:
64
1. Stable if the real part of all eigenvalues are negative i.e. all eigenvalues
are on the left-half complex plane  , j.
2. Unstable if the real part of at least one of the eigenvalues is positive i.e.
at least one of the eigenvalues is on the right-half complex plane  , j.
3. Bifurcation o if the real part of at least one of the eigenvalues is zero
i.e. at least one of the eigenvalues lies on the imaginary axis complex plane 
, j.
4.5 Inclusion of field circuit dynamics in the linearized model of

the SMIB system
The classical model of the synchronous machines assumes that the
transient emf is kept constant (E’ = 0) during the transient processes.
Therefore, the d-axis flux density is assumed to be constant during transient.
Note that the transient emf is proportional to the direct axis flux linkages.
Consequently, the field circuit electromagnetic dynamics are neglected. The
linearized classical model can be considered as an elementary model for the
analysis of the damping and synchronizing power of synchronous generators.
Higher order models of the synchronous machines can simulate the field
circuit dynamics14,15,16,17. The following simplifying assumptions are
considered in the following model,
1. The effects of the damper windings (amortisseur effects) are neglected.
These effects include additional damping; however, selecting an
appropriate damping coefficient is a possible way of considering the
amortisseur effects in a simple way. This assumption is quite
acceptable when the rotor speed deviations are small.
2. The electromagnetic transient of the stator windings are neglected.
Therefore, the time-dependent changes in the stator flux linkages (p)
14
Demello, F. P., & Concordia, C. (1969). Concepts of synchronous machine stability as
affected by excitation control. IEEE Transactions on power apparatus and systems, 88(4),
316-329.
15
Kundur, P. (1994). Power system stability and control (Vol. 7). N. J. Balu, & M. G.
Lauby (Eds.). New York: McGraw-hill.
16
Rogers, G. (2012). Power system oscillations. Springer Science & Business Media.
17
Padiyar KR. Power System Dynamics: Stability and Control: Anshan; 2004.
65
are neglected18. It can be shown that neglecting these transients is

compensated by neglecting the electromagnetic transients associated
with the power network. Therefore, modeling the power network
components by time-independent parameters is also acceptable. In
addition, the phasor diagram can be used to define the stator equations
(Fig. 4.13).
3. The machine saturation is neglected.
4. The resistances associated with the machine’s armature and the
network are neglected.
5. For generality, a salient-pole machine model is considered.
Given the stated assumptions, the machine model becomes a third order
model instead of a second order model. The additional differential equation
describes the field circuit dynamics. This model is as follows.
or
or
18
Kundur, P. ‘Chapter 5: Simplifications essential for large-scale studies – example 5.1’
in: ). Power system stability and control (Vol. 7). N. J. Balu, & M. G. Lauby (Eds.). New
York: McGraw-hill.
66
Note that the generator field voltage (Efd) equals to 1.0 p.u for 1.0 p.u
terminal voltage at open circuit conditions (i.e. on the air gap line). The
linearized form of this model is defined by the Heffron-Phillips Constants as
follows,
The block diagram representing the linearized model of the SMIB system
with field circuit dynamics included is shown in Fig. 4.45. As shown in the
figure, additional control systems can be included. These control systems
include the prime mover and the excitation control systems as well as
additional stabilization controls. In addition, the model can now simulate the
machine terminal voltage.
The equations representing these constant can be found through the
linearization of the machine model represented by equations (4.67) to (4.75).
In order to do so, linearization of equations (4.67), (4.68-a), (4.69-a), (4.71)
to (4.75) are prepared as they will be needed during the derivation of the K-
constants.
67
Fig. 4.45: Block diagram representing the linearized model of the SMIB
system with field circuit dynamics included with approximate modeling of
the machine damping (i.e. damper windings, eddy currents, electrical losses,
friction, and windage)
Equation (4.67):
Equation (4.68-a):
Equation (4.69-a):
Equation (4.71):
68
Equation (4.72):
Equation (4.73):
Equation (4.74):
Equation (4.75):
The constant K1 represents the change in electrical torque for a change in

rotor angle at constant flux linkage in d-axis i.e. at Eq’ = constant.
Based on (4.82),
Based on (4.83) and (4.84),
69
Substituting (4.89) and (4.90) in (4.88) and considering (4.87), results in:
Equations (4.70) shows that the term equals to .

Therefore, equation (4.91) becomes,
The constant K2 represents the change in electrical torque for a change in

d-axis flux linkage at constant rotor angle i.e.
Based on (4.82):
Using equations (4.83), (4.84), and (4.73) we get,
Considering the s-space representation of eq. (4.85)
From which the values of K3 and K4 can be determined after eliminating

id using (4.83) and arranging the equation in the form of (4.77).
70
It is clear that K3 represents the impedance factor that correlates the steady
state and transient reactances while K4 represents the demagnetizing effect of
change in rotor angle i.e. armature reaction as it is negative emf that oppose
the transient emf that corresponding to Efd.
Based on (4.78), the constants K5 and K6 are defined as follows.
The constant K5 Represents the change in terminal voltage with change in

rotor angle at constant while the constant K6 represents the change in
terminal voltage with change in at constant . Using (4.83) and (4.84) for
eliminating and from (4.80) and (4.81) then substitute the resulting
equations in (4.79) for eliminating and . The definitions (4.98) and
(4.99) are then used to determine the constants. The results are:
The values of K1, K2, K3, K4, and K6 are usually positive while the
value of K5 is usually negative. In fact, the increase in the machine active
power loading (in other words, the increase in the machine power angle,
71
reduces the terminal voltage) while the increase in the transient emf increases
the terminal voltage. Given that . Therefore, the
negative value of K5 and the positive value of K6 are expectable.
The required initial conditions for estimating the K-constants are: EQo, Vo,
o, Vtdo, Vtqo, and Vto. These variables are determined based on the
mathematical models presented in Appendix 1. The value of Vo is supposed
to be a known quantity as it is the infinite bus voltage magnitude while the
other variable can be estimated as follows.
2 2
P   Q 
EQo   o X q    V o  o X q   (4.102)
V o   Vo 
(Po /V o )X q 
tan   (4.103)
V o  (Qo /V o )X q 
2 2
P   Q 
V to   o X e   V o  o X e  (4.105)
V o   Vo 
V tqo V to cos(o  eo ) (4.106)
V tdo V to sin(o  eo ) (4.107)
(Po /V o )X e
tan eo  (4.108)
V o  (Qo /V o )X e
Referring to Fig. 4.25, the synchronizing power coefficient (Ps) is

significantly affected by the inclusion of the field circuit dynamics.
Generally, the synchronizing power coefficient is defined by:
With the field circuit dynamics neglected (i.e. the classical model), Ps and
the natural frequency of oscillations are given by,
72
With the field circuit dynamics considered, by replacing in (4.70) by

(4.77), the results is,
or
Therefore,
Equation (4.113) shows that at steady state (i.e. s  0), the synchronizing
power coefficient is reduced by ( ) with the field circuit dynamics
included in comparison with the classical model. This reduction is due to the
demagnetizing effect of armature reaction. The reduction in Ps is expected to
have a negative impact on the settling time and the frequency of oscillations.
Since the values of K1, K2, and K3 are usually positive, then it is possible to
say that the due to the demagnetizing effect of armature reaction adds a
positive damping (DAR). This can be easily seen by replacing s by j in
(4.112) and decomposing the resulting equation into real and imaginary parts.
73
Since then or , then (4.115)

becomes,
Equation (4.115) can be then written as
where Kd represent the armature reaction p.u damping coefficient while Ks

represents the synchronizing power coefficient. Both coefficients are rotor-
speed dependent and they are quadrature to each other as shown in Fig. 4.46;
where
Fig. 4.46: Synchronizing and damping components of the electrical power
74
The equivalent damping constant Deq can be obtained from,
Note: that if o is represented in p.u (i.e. equals to 1.0 p.u), then D = Kd.
The natural frequency of oscillations in the absence of voltage regulator

(Efd = 0) is then,
Note: that, if which is popular situation, then
Example
Simulation and analysis of the linearized model of the SMIB system
considering field circuit dynamics
The presented linearized third-order model is implemented on the

Simulink as shown in Fig. 4.47. The parameters of the considered system and
main variables of the considered system are given as follows such that the
reactances, voltages, and power values are represented in p.u while the time
75
and inertia constants are represented in seconds. The system data are typical
values of generators in nuclear power plants19.
The steady state initial conditions and the K-constants as well as some
other parameters are consequently determined using the presented model.
Their values of the K-constants are,
Fig. 4.47: simulink model of the linearized model SMIB system

The main objective of this example is to show the effect armature reaction
damping. Since D is set to zero, the performance of the classical model was –
19
http://mkpalconsulting.com/
76
as shown in the previous simulations - sustained harmonic oscillations;

however, with field circuit dynamics considered, the performance is
oscillatory with decayed amplitudes. This is demonstrated by each of the
following disturbances:
1. A step decrease in the mechanical power of 10% (Fig. 4.48 and 4.49)
as an example of the forced response of the system due to sudden change in
the load.
2. A step decrease in the field voltage of 10% (Fig. 4.50 and 4.51) as an
example of the forced response of the system due to sudden change in the
field voltage caused by a very fast acting excitation regulator, and
3. A sudden change in the power angle by 10 o (Fig. 4.52 and 4.53) as an
example of the zero-input response of the system due to sudden change in the
network topology.
All simulations show clearly the damping effect of the field circuit
dynamics. The model can provide many other simulations and analysis that
may be implemented by the interested readers.
Fig. 4.48: Linearized response of the SMIB system for Pm = -10%
77
Fig. 4.49: Linearized response of the SMIB system for Pm = -10%
considering field circuit dynamics – the electrical power components
Fig. 4.50: Linearized response of the SMIB system for Efd = -10%
78
Fig. 4.51: Linearized response of the SMIB system for Efd = -10%
Fig. 4.52: Linearized response of the SMIB system for o = 10o
79
Fig. 4.53: Linearized response of the SMIB system for o = 10o
4.6 Including AER and PSS in the linearized third-order model of the
SMIB system
As stated in Appendix 1, an excitation system and an excitation control
system are a combination of devices designed to generate a field current and
control it by means of Automatic Excitation Regulators (AER) which is also
called Automatic Voltage Regulator (AVR). An excitation control system is
composed of three main components. These components are the field
winding, exciter (i.e. a DC power source), and a regulator (AER or AVR).
The AER controls the value of Efd as shown in Fig. 4.54. The regulator is
nonlinear due to the presence of the Efd limiter.
80
Fig. 4.54: Generic schematic of excitation control systems
The excitation control system model can be integrated with the 3 rd order
linearized model of the machine as shown in Fig. 4.55.
Fig. 4.55: SMIB 3rd order linearized machine model with excitation
control
The main objective of this model is to investigate the impact of the

excitation control system on the damping of the system. There are many other
applications in which the model of Fig. 4.55 can be used. Measurement-based
estimation of the system inertia is an example that will be considered later in
this chapter. Since measurement-based equivalence is a target of this book,
no more detailed mathematical modeling will be given to the system of Fig.
81
4.55; however, the interested reader may refer, for example, to Kundur’s
textbook20.
Example
Simulation and analysis of the linearized model of the SMIB system with
excitation control system
The system of Fig. 4.47 is considered in this example; however, an

excitation control system is included in the system model as shown in Fig.
4.55. The combined transfer function of the exciter and the regulator is given
the simple form,
With the gain of G(s) is set to zero, the effect of the excitation control
system is deactivated. In this case, the effect of the effect of the field circuit
dynamics is considered as shown in Fig. 4.47. In this example, it is assumed
that the terminal voltage regulator is fast enough for neglecting the time delay
effect of its time constant (TR). In addition, it is assumed that the capacity of
the AER is large enough for neglecting the AER output limiter; consequently,
the AER becomes a linear network. A fast acting excitation control system
with a time constant TA = 0.1 sec and gain KA = 50 (i.e. high gain) is
considered. Proper selection of the gain KA is usually based on a stability
criterion for the AVR loop21. A sudden decrease of 10% in the mechanical
power input is considered as a disturbance. The effect of the excitation
control system on the system damping in comparison with the unregulated
system is investigated considering the mentioned gain (KA = 50) and a gain
KA of a zero value. The simulations of the system with each of the mentioned
gains are performed. The Simulink model is shown in Fig. 4.56 while the
results are shown in Fig. 4.57. The yellow lines are corresponding to the KA =
50 while the other lines are corresponding to the KA = 0. It is clear from the
20
21
Elgard, O. I. (1982). Electric energy systems theory. New YorkMc Graw-Hill.
82
results that the considered AVR significantly reduces the system damping
and causes instability due to negatively damped oscillations.
Fig. 4.56: Lumped simulink model of the system
Time (s)
Fig. 4.57: Effect of the excitation system control in comparison with the
unregulated system
83
Reducing the AER gain improves the stability. This is shown in Fig. 58
with KA = 30 (yellow lines) and KA = 10. In both situations, the system is
stable; however, the stability suffers from two issues: (1) The steady state
error in the terminal voltage is high and its value is increased with the
increase in the gain KA.; (2) The settling time of the oscillations is also high.
Therefore, the stabilization gain (or damping enhancement) caused by the
reduction of the AER gain is counteracted by the high steady state errors and
the long settling time.
Fig. 4.58: Stabilization of the system by AER gain reduction
84
The impact of the gain KA can be demonstrated and explained using the
root locus of the system which is shown in Fig. 4.59. Fig. 4.59(a) shows the
full root locus of the system while Fig. 4.59(b) shows the root locus focused
near the imaginary axis for gain selection purposes.
(a)
(b)
Fig. 4.59: Root locus of the regulated SMIB with AER – the effect of
AER gain on stability and performance; (a) Full view; (b) Zoomed view
85
It is clear from the figure that the critical value of the gain is 42.9 at which
the system is critically stable. Higher values of the gain result in system
instability while with lower values the system is stable but its damping is
poor. Appendix 2 provides an overview of the extraction of linear transfer
from a Simulink model and the linearized analysis of transfer function as well
as gain selection based on the root locus analysis.
4.7 Power System Stabilizers (PSS)

Power system stabilizers are supplementary stabilizing controllers of
the AER. As shown in Fig. 4.57, AER can provide negative damping to
power systems; a situation which leads to system instability. This AER
caused instabilities depend on the system parameter, the operating conditions,
and the AER structure as well as its parameters. For example, the responses
of Fig. 58 are stable; however, the characteristics of the oscillations (settling
time, overshoot, rise time… etc.) are poor. PSSs can also provide a tool for
enhancing the system performance. Tuning of PSSs and the control design
issues are out of the scope of this chapter; however, stabilization of the
system of Fig. 4.56 will be considered.
Generally, the main objective of PSSs is to artificially produce a
damping torque, which in-phase with the rotor speed. This torque is called a
supplementary stabilizing signal and the controllers used to generate these
signals are known as power system stabilizers. Fig. 4.60 shows PSS
integrated with the AER such that the speed deviation is the PSS input. The
frequency or acerbating power may be also used as input signals to the PSS.
As shown in Fig. 4.60(b), the speed deviation is used as an input to the PSS.
The transducer measures and converts the input signal to voltage. The high
frequency filter is designed to attenuate the generator’s torsional
frequencies as well as high frequency measurement noise. The lead-lag
stages are usually composed of identical a number of phase compensators.
The function of the lead-lag stages is to provide an overall phase lead over
the frequency range of interest to compensate for the lag produced in the
generator-excitation system. The minimum number of phase compensators is
one while the maximum known number is three while the most popular
designs contain two compensators. The output signal of the lead-lag stages is
then amplified by an amount Ks and sent through a washout stage. The
washout prevents voltage offsets during steady-state or prolonged speed or
86
frequency change. The output limiter prevents the stabilizer output signal
from interfering with the action of the voltage regulator during severe system
disturbances.
(a)
(b)
Fig. 4.60: AER with PSS supplied from the speed deviation; (a) general
block diagram; (b) the PSS details are shown
87
Example
Simulation and analysis of the linearized model of the SMIB system with
excitation control system and simplified PSS
With an AER gain of 50, the system of Fig. 4.56 shows instabilities due to
the negative damping cause by the high-gain AER. In this example, PSS is
used for system stabilization. The parameters of the PSS in this example are
selected based on the experience; however, the parameters should be
optimized for the best additional damping, performance enhancement, and
stability. A simplified PSS is considered where the transducers and high pass
filters are assumed to be fast enough for their transfer functions to be
neglected. In addition, the effect of the PSS output limiter is neglected. One
stage lead-lag compensator is considered. The parameters of the PSS are
shown in Fig. 4.61 and the system response in comparison with the system
performance without PSS (yellow lines) is shown in Fig. 4.62(a) and the PSS
output is shown in Fig. 4.62(b).
Fig. 4.61: The linearized third-order model of the SMIB system with AER
and PSS
88
Although the PSS parameters in this example are roughly estimated (i.e.
they are not well tuned) and the PSS model is simplified, the results show
that the PSS stabilized the system even if the AER gain is too high (caused
instability without the PSS). Better performance can be obtained by PSS
optimized tuning; however, this issue is out of the scope of this chapter.
Time(s)
(a)
Time(s)
(b)
Fig. 4.62: Performance of the linearized third-order model of the SMIB
system with AER and PSS for a sudden drop in the mechanical power by
10% (a) and PSS output (b)
89
4.8 Frequency response and Speed Governors

The generator response to load changes is simply described by the
equation of motion which is called the swing equation as described by
equations (4.17) or (4.18). With the speed expressed in p.u, the linearized
form of the swing equation can be written as,
or
where M = 2H, is the damping power caused by the damping

associated with the machine’s electrical and mechanical losses as well as the
control system and operating conditions. The damping power coefficient (D)
relates this power to the p.u change in the rotor speed ( ) as shown in eq.
(4.126) or (4.127). The transfer function of the linearized swing equation can
be represented by the block diagram of Fig. 4.63. The damping power term is
shown by dotted lines as it can be omitted from the block diagram if the
machine electromagnetic transients (for example, the dynamics of the field
circuit and damper windings as well as other dynamics associated with the
controls).
Fig. 4.63: Transfer function of the linearized swing equation
The electrical changes in the generator load are instantaneously reflected

as changes in the electrical power. Consequently, the machine speed (or
90
frequency) deviates due to the mismatch between the electrical and

mechanical power.
The load also responds to the changes in the frequency and voltage
changes. These changes depend on the load mix supplied by the generator
(see chapter 7 in Kundur’s textbook 22 and Appendix 3). Two popular static
load models are usually used for load response modeling; the exponential
load model and the ZIP load model. The exponential load model can be
represented by (4.128) while ZIP takes the form of (4.129).
128)
where Po and Qo are the load active and reactive power at the nominal
voltage Vo (usually 1.0 p.u). Kpf and Kqf represents respectively the active and
reactive load sensitivity to frequency variations. The exponents  and 
represent respectively the load active and reactive power sensitivity to
voltage variations. The coefficients p, p and p represents respectively the
p.u share of constant power, constant current, and constant impedance in the
active power load while the coefficients q, q and q represents the same
quantities for the reactive power load. The relation between these coefficients
is,
From frequency sensitivity point of view, equations (4.128) and (4.129)

show that the load can be represented by two components; insensitive
component and sensitive component. Given that in p.u, f =  and
22
91
considering the load sensitivity to frequency variations, the changes in the

electrical power can be represented by
where is the frequency independent changes, ( ) is the

frequency dependent load changes, represents the load sensitivity to
frequency variations and also called the load damping constant. The load
damping constant is in the range of 1% to 2%; and its value represents the
percentage change in the load for 1% change in the frequency. For focusing
the analysis on the load frequency response, the damping constant D will be
neglected in the following analysis. Based on the presented load reaction to
frequency variations, the swing equation can then take the form of equation
(132) and can be represented by Fig. 4.64.
Based on block reduction theory, the block diagram of Fig. 4.64 can be
reduced to Fig. 4.65 which can be represented as shown in Fig. 4.66.
where
Therefore, at steady state (i.e. t   or s  0),
92
Fig. 4.64: The linearized swing equation considering load components
Fig. 4.65: Simplified block diagram of the linearized swing equation

considering load components
Fig. 4.66: Equivalent block diagram of the linearized swing equation

considering load components
4.8.1 Frequency response of a governor-less generator

With (i.e. no speed governor; see Fig. 4.67), eq. (4.134) shows
that . Therefore, the steady state speed is the speed at which
the load change ( ) is exactly compensated by the frequency-dependent
load change ( ). In addition, Fig. 4.66 reduced to Fig. 4.68.
93
Fig. 4.67: SMIB with the speed governor deactivated
Fig. 4.68: Equivalent block diagram of the linearized swing equation

considering load components and without a speed governor
The time domain (TD) solution of the system is can be then obtained as
follows,
If = constant, then and (4.135) becomes,
The inverse Laplace can be easily determined by partial fraction and the
use of the Laplace transform table23,24 i.e.
23
http://tutorial.math.lamar.edu/Classes/DE/InverseTransforms.aspx
24
http://tutorial.math.lamar.edu/Classes/DE/Laplace_Table.aspx
94
This function is shown in Fig. 4.69 and the steady state speed deviation is
then,
Fig. 4.69: Speed deviation for a sudden change in the load in governor-
less system
As shown, a sudden increase in the load causes steady state reduction in

the speed while a sudden reduction in the load causes a steady state increase
in the speed or frequency).
The torque-speed characteristics for a governor-less generator can be
determined based on the torque-power relation,
The linearized form of (139) takes the form,
95
For ,
or in p.u
Therefore,
The integration constant can be obtained from the steady state initial
conditions i.e. . Hence, in p.u,
This relation shows that the torque-speed relation in a governor-less

system is linear. In addition, this relation is correct over a limited at rated
speed (see Fig. 4.70). As shown, an increase in the frequency results in a
reduction in the mechanical torque as the input power is constant.
Fig. 4.70: Torque-speed relation of a governor-less generator near rated

speed
96
4.8.2 Type s of speed governors

From the operational point of view, a power system must be capable of
exactly supplying the load demand. Since power systems do not contain bulk
power storage and the load also is a stochastic variable, then the total power
generated should be sufficient to provide a power balance at every instant.
The dilemma is that the time constants associated with the mechanical power
changes are significantly larger in comparison with most load variations;
however, this problem is solved by the proper operation of dispatchable
generators as explained in the following.
Consider a system with n-dispatchable generators. (n-1) generators are set
to operate in the PV-mode while only one generator is set to operate in the
slack mode. The daily load forecast is used as a main input data to the system
operators. The continuous load forecast curve is transformed into a discrete
function with a sampling time in the order of 30 min or 60 min. Then the load
level at each time step is used to allocate the power over the generators
operating in the PV-mode. Usually security-based economic dispatch is used
to determine the appropriate load sharing between the PV-generator. Up to
that point, the power balance requirement cannot be achieved due to the
mismatch between the forecasted load curve and the generation as well as
other mismatch sources such as forecast errors, power losses, and random
events (e.g. forced outage of generators or sudden change in the load).
Therefore, the remaining generator is set to operate as a slack generator. The
slack generator is not actually has a specific load sharing demand. Instead,
this generator operates as a power balancer for securing the power balance
requirement and consequently keeping the system operating at an acceptable
frequency.
The described operation of the system can only be possible with the use of
speed governors. Two main types of speed governors are available according
to the desired operational function of a generator. PV-generators are required
to share the load. Therefore, they are equipped with speed governors with
droop characteristics. A slack generator is required to balance the system.
Therefore, it is equipped with an isochronous governor. The ideal steady state
speed-power characteristics of these types of governors are shown in Fig.
4.71. It can be seen from the figure that a droop governor reacts to changes in
the frequency by opposite changes in the mechanical power. These two types
97
of governors will be considered in the following. In addition, Appendix 4

provides an overview of the fundamentals of speed governors.
Fig. 4.71: Steady state characteristics of isochronous and droop governors
A. Isochronous (constant speed) governors

The isochronous governor adjusts the valve/gate position for adjusting
the generator speed (or frequency) to the desired value (usually the nominal
frequency). This is shown in Fig. 4.72 where the turbine shaft speed is
assumed to be identical with the rotor generator’s speed i.e. a single mass
model is adopted.
Fig. 4.72: Block diagram of a generator with an isochronous governor
In the figure, Y represents the valve/gate position. In load frequency

response analysis, the generator model is represented by Fig. 4.65 or 4.66.
Therefore, Fig. 4.72 becomes,
98
Fig. 4.73: Generator with isochronous governor model for load-frequency

response
There many designs of steam and hydro turbines. Three popular designs
are considered in the following analysis. These designs are non-reheat steam
turbines, reheat steam turbines, and hydro-turbines. Details about these
turbines can be found in Kundur’s textbook. Table 4.3 summarizes the
transfer functions of these turbines as well as the typical values of their
parameters.
Table 4.3: Transfer functions of turbines

Turbine Transfer function Typical parameters
Non-reheat TCH = 0.3 sec
steam
Reheat steam25 FHP = 0.3

TCH = 0.3 sec
TRH = 7.0 sec
Hydro* TW = 1.0 sec
* Important notes about hydroelectric turbines are presented in Appendix 5.
25
The model of the reheat steam turbine is applicable to non-reheat steam turbines but the
value of TRH should be set to zero.
99
EXAMPLE
LOAD-FREQUENCY RESPONSE OF GENERATOR WITH AN ISOCHRONOUS
GOVERNOR
In this example, the load-frequency response of a generating unit equipped
with an isochronous governor is considered. The non-reheat steam turbine
model shown in Table 4.3 will be simulated. The model shown in Fig. 4.73 is
implemented on the Simulink (Fig. 4.74) and typical parameters are used. For
all models a disturbance of 10% step increase in the load is applied. The
typical parameters of the turbine transfer functions are shown in Table 4.3
while the rest of parameters are: H = 5 sec, Dpf = 1.0 while the impact of the
gain KG is tested by setting its value to 0.035, 0.1, 3.0, and 5.0; the results are
shown respectively in Fig. 4.75 to 4.78.
Fig. 4.74: A generator with isochronous governor and non-reheat steam

turbine
The results show that with a gain of 0.035, a critically damped response is
achieved. Higher values of the gain result in oscillatory performance while at
a gain of 5.0 the oscillations become unstable. Therefore, proper design of the
gain for a desired load-frequency performance is essential for stable and
satisfactory dynamic performance of the governor. The most important
characteristics of the isochronous governors are that the steady state errors in
the speed (or frequency) are zero. This is due to the reset action of the
integrator, which results in continuous control actions of the governor till the
zero error in the speed is achieved.
100
Time (sec)
Fig. 4.75: Performance of a generator with isochronous governor and non-
reheat turbine; KG = 0.035
Time (sec)
101
Time (sec)
Time (sec)
102
Time (sec)
B. Speed droop governors

The function and performance of an isochronous governor are
satisfactory when the generator is operating off the grid or for grid balancing
function; however, in a grid-connected operation, more than one isochronous
governor generator results in problems in the load sharing between the
generators and the overall stability of the system.
Unless they have exactly the same settings, if two generators having
isochronous governors operate in parallel they will fight each other to set the
system to the individual settings of each of them. Since the optimal economic
secure operation of generators requires specific optimal settings to be
allocated to each generating unit. Therefore, using more than one isochronous
generator is practically impossible. Stable load sharing between parallel units
requires that a change in a generator frequency is to be compensated by its
power output. In other words, an increase in the frequency should be
compensated by a reduction in the power output and vice versa. This action
can be achieved by the use of governors having speed-droop or speed-
103
regulation characteristics26 (Fig. 4.80). The ideal steady state characteristics

of a speed-droop governor are shown in Fig. 4.81. As shown in Fig. 4.80, the
isochronous governor is transformed into a droop governor by adding a
feedback loop with a constant gain R around the terminals of the isochronous
governor. The value of R defines the droop of the governor as shown in Fig.
4.81. The droop defines the steady state change in the frequency due to a
steady state change in the output power. For example, a droop of 2% means
that 2% deviation in the frequency causes 100% change in the output power
or the valve/gate position. As shown in Fig. 4.80(c), the governor equivalent
model is a proportional controller with a gain of a value (1/R).
Example
Load-frequency response of generator with a speed-droop governor
In this example, the considered disturbance and the parameters of system

of Fig. 4.74 are used for simulating a generator with a droop governor. The
block diagram of Fig. 4.80(b) is implemented on the Simulink as shown in
Fig. 4.82. The droop of the generator is set to 10% (0.1 p.u). The chosen
value of the droop is high for clearly showing the steady state errors. Two
values for the governor gain are selected to show the critical damped and
oscillatory damped responses. These values are 0.035 and 1.0; the results are
shown in Fig. 4.83 and 4.84. The results show that with either value of the
governor gain, there is a steady state error in the frequency, output power,
and valve position. In addition, this steady state error equals to the droop. It
can also be easily seen that the settling time and overshoots associated with
the generator equipped with a droop governor are much lower in comparison
with the generator equipped with an isochronous governor. The addition of
the droop constant to the governor control loop reduces the effect of the
integrator and improves the dynamic performance; however, the steady state
performance is degraded due to the steady state error.
26 There are some distinguishing characteristics of hydro-turbines that mandate an

additional block called ‘transient droop compensation’ to be added to the governor.
Appendix 3 describes these special characteristics of hydro-turbines and the transient
droop compensator.
104
(a)
(b)
(c)
Fig. 4.80: Droop Governors; (a) Block diagram of a generator with a
droop governor; (b) Generator with a droop governor model for load-
frequency response; (c) Reduced block diagram model of the governor
( )
105
Fig. 4.81: Steady state characteristics of a speed-droop governor
Fig. 4.82: Simulink model of a generator with a droop governor and non-
reheat steam turbine
106
Time (sec)
Fig. 4.83: Performance of a generator with a droop governor and non-
reheat steam turbine; KG = 0.035
Time (sec)
Fig. 4.84: Performance of a generator with a droop governor and non-
reheat steam turbine; KG = 1.0
107
4.8.3 Supplementary control action for unit output power control

For generator output control (or prime mover control), droop governors
require a sustained change in system frequency27. Therefore, speed-droop
governors alone cannot restore the power system frequency to the pre-
disturbance level. This is can be explained based on Appendix 4 and Fig.
4.85. It is clear from Fig. 4.85 that an increase in the load, the frequency (or
speed) drops and consequently the governor increase the steam (or hydro)
flow (see Fig. A3.1) for compensating the speed drop. Due to the droop
characteristics of the governor, the frequency will not be totally restored;
however, a remaining error f will sustain. The governor will not restore the
frequency to the original value; however, this is can be achieved by
increasing the reference power (Pref) settings of the governor by the use of the
speed-changer (shown in Fig. A3.1). Consequently, the droop characteristics
move upward as shown in Fig. 4.85. Proper increase in the reference power
(Pref) will result in restoring the frequency to the original value while the
generator is supplying the higher load. Therefore, the speed-changer provides
a supplementary control to the governor for adjusting its the settings.
Fig. 4.85: Operation of a droop-governor and the supplementary control

action
27
WECC Control Work Group. WECC Tutorial on Speed Governors. Une 20002. WECC.
108
The load sharing between generators can also be provided by the droop
governors and the speed-changers. If two generating units operating in
parallel for supplying a load and both of them equipped with isochronous
governors with a very slight difference in their speed settings (design
parameter), then one of the units will try to carry the entire load and the other
will shed all of its output. Therefore, neither load sharing nor control of
sharing is possible when units operating in parallel have isochronous
governors. This is can also be explained by considering two isochronous
units coupled together on the same load and the speed settings are not the
same.. Since there cannot be two different speeds (or frequencies) on one
system, one unit will have to decrease its actual speed and the other unit will
have to increase its actual speed to an average speed between the two units.
The governor on the unit that decreased speed will move to increase steam to
try to correct for the decrease in speed, and the governor on the other unit that
increased speed will move to decrease the steam to try to correct for the
increase in speed. The result will be that the unit with the higher speed setting
will continue to take all of the load until it reaches its power limit, and the
other unit will shed all of its power and become motored (driven by the other
unit). Therefore, the system will become unbalanced when isochronous units
coupled together. Consequently, in the power generation system, only one
generator can be equipped with an isochronous governor while the rest of
generators are to be equipped with droop governors.
Fig. 4.86: Load sharing between units with droop governors
109
With droop governors, the load sharing is possible as shown in Fig. 4.86.
As shown, the governors will share the load increase (P) at a lower common
frequency value. This new frequency is defined by
while P= P1+P2. The system frequency can be restored by adjusting the
speed changers (or the reference power settings). Consequently, the speed-
power relation is changed such that the frequency is restored.
(a)
(b)
(c)
Fig. 4.87: Droop Governors; (a) Block diagram of a generator with a
droop governor and load reference control; (b) Generator with a droop
governor model and load reference control for load-frequency response; (c)
Reduced block diagram model of the governor with the load reference control
( )
110
When a generator is paralleled with a large utility grid, it is important

to consider that:
1. The utility will act as an isochronous generator. Therefore, the utility
will set the speed/frequency for any units being paralleled to it.
Consequently, a simple isochronous unit cannot be paralleled to the
utility.
2. When an isochronous generator is connected to a utility bus, the utility
will determine the speed/frequency of the generator. If the governor
speed reference is less than the utility frequency, the utility power will
flow to the generator and motor the unit. On the other hand, if the
governor speed is even fractionally higher than the frequency of the
utility, the governor will go to full load in an attempt to increase the
interface bus frequency. Since the definition of a utility is a frequency
which is too strong to influence, the generator will remain at full load.
3. Droop governors provide the solution to this problem. The droop
causes the governor speed reference to decrease as load increases. This
allows the governor to vary the load with the speed setting since the
speed cannot change.
With the speed-changer considered, the block diagram of Fig. 4.80 is

modified to that shown in Fig. 4.87.
Example
Load-frequency response of generator with a speed-droop governor and
load reference control
In this example, the example of Fig. 4.82 is repeated considering the

impact of step changes in the load reference control. The disturbance in this
case is 10% increase in the reference power setting. The Simulink model of
the system is shown in Fig. 88 while the results are shown in Fig. 4.89 and
4.90 for KG of 0.035 and 0.1 respectively. Based on the previous discussion,
it is expected that the droop governor is not capable of restoring the system
frequency; as shown, there are steady state frequency errors associated with
the change in the power input to the generator. Since the input power
111
increased by 10%, the positive steady state frequency error is appeared with
both values of KG. The solution to this issue is demonstrated in the next
example.
Fig. 4.88: Simulink model of a generator with a droop governor and load
reference control with non-reheat steam turbine
Fig. 4.89: Performance of a generator with a droop governor and load

reference control with non-reheat steam turbine; KG = 0.035
112
Fig. 4.90: Performance of a generator with a droop governor and load

reference control with non-reheat steam turbine; KG = 1.0
Example
Load-frequency response of generator with a speed-droop governor and
load reference control – AGC for restoring system frequency
In this example, the example of Fig. 4.82 is repeated considering the

impact of the load reference control in restoring the system frequency by
adjusting the frequency-load relation. For adjusting the reference power (Fig.
4.87) to restore the system frequency, an integrator and a stabilization gain
are connected between the speed deviation and the reference power.
Therefore, Pref = -Kperf/s i.e. a feedback PI loop is formed. The integrator
is used for keeping Pref changed till the frequency error  becomes zero.
The value of Kperf is set to 0.035 for control loop stabilization (see Appendix
2). The added reset or integral control of the reference power makes the
113
system operates in an AGC mode for restoring the system frequency. The
simulink model of the system is shown in Fig. 4.91 while the results are
shown in Fig. 4.92 and 4.93.
Fig. 4.91: Simulink model of a generator with a droop governor and load
reference control with non-reheat steam turbine – AGC mode
Time (sec)
Fig. 92: Performance of a generator with a droop governor and load
reference control with non-reheat steam turbine – AGC mode; KG = 0.035
114
Fig. 93: Performance of a generator with a droop governor and load

reference control with non-reheat steam turbine – AGC mode; KG = 1.0
It is clear from the results that the proper load reference control reduces
the speed error signals to zero and also enhances the overall dynamic
performance of the system.
This chapter presents the fundamentals of power system dynamics.

Simplified models and system topology (i.e. the SMIB) are considered for
their easy handling and simulation; however, the presented conclusions are
quite applicable to larger systems from scale point of view and also to more
complex models. The following chapters include further analysis of larger
power systems. In addition, the dynamics of recent energy production
technologies will also be presented.
115
116
Chapter 5
Measurement-based SMIB Equivalence of Multi-
Machine Systems AND Transient Stability
5.1 INTRODUCTION
It can be seen from Fig. 4.1 that a system subjected to a disturbance reacts
to that disturbance by frequency dynamics. The system frequency is then the
average of the frequency of individual generator in the system. Just after a
disturbance, the system frequency dynamics (Fig. 4.1) can be characterized
by two quantities; the frequency gradient ((df/dt), and the frequency
deviation.
In the previous chapter, the linearized modeling and simulation of
various components and controllers in the SMIB system (Fig. 4.7 or 4.27) is
presented. The model of Fig. 4.55 represents a synchronous machine and an
AER. In that model, the synchronous machine is represented by the third-
order linearized model i.e. the field circuit electromagnetic dynamics are
considered. The PSS model is represented by Fig. 4.60 while the droop
governor model with load reference control is represented by Fig. 4.87. The
AGC supplementary control for restoring the system frequency is represented
in Fig. 4.91. The stated models represented the overall structure of the SMIB
system with machine controls included. This overall model is shown in Fig.
5.1.
In this section, this model is simulated considering various levels of
modeling details. The main objective is to investigate the impact of the
dynamic response of various components in the SMIB system on the
frequency dynamics of the system. The changes in the frequency gradient and
frequency deviations are the main focus. This is because the values of these
quantities will be shown to have a major impact on the measurement-based
equivalence that will be presented in this chapter. A secondary objective of
the considered simulations includes a more clear understanding of power
system dynamics and their governing issues.
The parameters of the considered system and main variables of the
considered system are given as follows such that the reactances, voltages, and
power values are represented in p.u while the time and inertia constants are
represented in seconds.
117
Fig. 5.1: The linearized 3rd order model of the SMIB system with various
controllers included.
Generator data:
The steady state initial conditions and the K-constants as well as some
other parameters are consequently determined using the presented model.
Their values of the K-constants are,
AER data: The transfer function of the AER is represented by equation

(4.125) and its parameters are: TSR = 0, TA = 0.1 sec, KA = 30
PSS data: The general block diagram of the PSS is shown in Fig. 4.60. In
this example, one lead-lag compensator is considered while the transfer
functions of the frequency transducer and the high frequency filter are
neglected. Therefore, the transfer function of the PSS is shown in Fig. 5.2
while its parameters are,
118
Fig. 5.2: Simplified transfer function of the PSS
T1 = 0.16, T2 = 0.03, TW = 1.0, KS = 10

Turbine data: Non-reheat steam turbine is considered. The transfer
function of the turbine is stated in Table 4.3 (TCH = 0.3 sec)
Governor data: KG = 0.035, R = 10%
AGC (for restoring system frequency) loop data: KAGC = 0.035
The considered case studies are summarized in Table 5.1. The
transmission network connecting the generator terminals to the infinite bus is
constructed of a step-up transformer and two parallel lines. The considered
disturbance is a momentary outage of one of the lines; a situation that can be
theoretically represented – as explained in the previous chapter - by a sudden
change in the power angle. This change in the angle is assumed to be 5
degrees. The system model is simulated considering the cases listed in Table
5.1 for the stated disturbance. A single disturbance is used for comparative
analysis purposes. The results are shown in Fig. 5.3. The period of the TD
simulations are set to 5.0 sec for illustrating some important characteristics of
the transient processes.
Table 5.1: Case studies for the system of Fig. 5.1

Case Conditions
Case 1: Classical model K2 to K6 = 0.0, field dynamics neglected, no
governor, no AER, and no PSS. D = 0.0
Case 2: Field dynamics No governor, no AER, and no PSS i.e. unregulated
only machine
Case 3: case 2 + AER No governor, and no PSS
Case 4: case 3 + PSS No governor
Case 5: case 4 + No supplementary control
governor
Case 6: the overall All controllers
structure
119
The most important conclusion that can be easily depicted from the results
is that the characteristics of the initial frequency dynamics are the same
regardless of the modeling details as well as the available controllers. In all
cases, the initial frequency gradient, deviation, and nadir are the same. This
can be easily explained based on the fundamentals given in the previous
chapter. At the beginning of the transient process, the system dynamics are
dependent mainly on the system inertia. This is because the changes in the
electrical power response are instantaneous (Fig. 5.3(b)); the electromagnetic
transient causes by, for example, the machine’s field circuit dynamics and the
other dynamics caused by various controllers are inherently suffering from
time delays.
(a)
(b)
Fig. 5.3: Frequency and electrical power transients for various cases
shown in Table 5.1
120
These time delays depend on the parameters of the machine and

thecontrollers as well as the initial conditions, and the severity of the
disturbance. The mechanical power transients associated with the prime
mover system and their controllers are shown in the previous chapter to be
characterized by very large response time in comparison with the
electromagnetic transient. Therefore, their effects will not act on the
considered TD duration.
The basic characteristics of the SMIB system and impact of various
controllers are demonstrated in the previous chapter and the cases shown in
Fig. 5.3. Although the derived conclusions were based on a simple system
and linearized models, they are also applicable to larger systems with more
complex details as will be shown later. In this section, an efficient simple
method for estimating the equivalent system inertia constant at a specific is
presented, performed, and evaluated. The method is solely dependent on
synchronized bus frequency and power measurements under disturbed
conditions 28,29,30,31,32 (refer to chapter 1/volume 1 of this textbook for more
details about synchrophasor measurements). Based on the swing equation
(4.17) or (4.18), the inertia constant M can be determined as,
PMUs can measure the AC power frequency (i.e. 50 Hz or 60 Hz) voltage

and current waveforms at typical rate of 48 samples per cycle i.e. 2400
28
El-Shimy, M. (2015). Stability-based minimization of load shedding in weakly
interconnected systems for real-time applications. International Journal of Electrical
Power & Energy Systems, 70, 99-107.
29
Tsai, S. J., Zhang, L., Phadke, A. G., Liu, Y., Ingram, M. R., Bell, S. C., ... & Tang, L.
(2007). Frequency sensitivity and electromechanical propagation simulation study in large
power systems. Circuits and Systems I: Regular Papers, IEEE Transactions on, 54(8),
1819-1828.
30
Wall, P., González-Longatt, F., & Terzija, V. (2010, August). Demonstration of an
inertia constant estimation method through simulation. In Universities Power Engineering
Conference (UPEC), 2010 45th International (pp. 1-6). IEEE.
31
Chassin, D. P., Huang, Z., Donnelly, M. K., Hassler, C., Ramirez, E., & Ray, C. (2005).
Estimation of WECC system inertia using observed frequency transients. Power Systems,
IEEE Transactions on, 20(2), 1190-1192.
32
Zhang, Y., Bank, J., Wan, Y. H., Muljadi, E., & Corbus, D. (2013). Synchrophasor
Measurement-Based Wind Plant Inertia Estimation: Preprint (No. NREL/CP-5500-
57471). National Renewable Energy Laboratory (NREL), Golden, CO..
121
samples per second for 50 Hz systems and 2880 samples per second for 60Hz
systems. Therefore, the changes in the mechanical power in a sampling
duration t can be neglected. This is due to the relatively slow mechanical
system dynamics in comparison with electromechanical and electromagnetic
dynamics. In addition, the changes in the rotor speed during are small enough
for the PD = D to be neglected. Just after a disturbance, typical inertial and
frequency responses of power systems (Fig. 5.3) are initially governed by the
system inertia as demonstrated in the previous examples. Therefore, the
measurements used for estimating the inertia of the system should be just
after a disturbance (i.e. at to+ where to is the disturbance instant).
Consequently, for a small sampling time of measurements ( t), the inertia
constant M can be estimated at the location of measurements using,
In (5.2), the power and frequency are in p.u while the inertia constant and
the time are in seconds.
5.2 Measurement-based electromechanical equivalence

An improved approach for fast and flexible electromechanical
equivalency for online applications will be presented in this chapter. The
approach is based on the availability of WAMS and PMU (see volume
1/chapter 1) at limited selected locations in the power system. Coherency
grouping of generators is also required for the construction of the
equivalence; however, the grouping is not necessary to be performed online.
As demonstrated in the previous volume (chapters 2 and 3), a coherent
group of generators remains coherent regardless of the disturbance type,
location, severity; however, large changes in the system topology may cause
upset of the coherency grouping. This issue can be solved by identification of
the coherency grouping from the performance of the generators under various
disturbed conditions raised from natural changes in the system. With this
coherency identification approach, the coherency grouping remains up-to-
date without the need of performing a mathematical simulation for that
purpose. Consequently, the accuracy and speed of coherency grouping will
122
be significantly improved. Again, the identification step is only needed in

situations of large topological changes in the power system.
For simplifying the presentation of the approach, the system shown in Fig.
5.4 is considered. The considered interconnected power system is assumed to
be composed of two radially connected areas. This system is said to be
weakly interconnected if the transmission system interconnecting the two-
areas are of a capacity is less than the capacity of the smaller area by more
than 15% or 20%. The configuration of Fig. 5.4 is popular in interconnected
systems and systems integrated with large amounts of renewable sources
where the location of these sources are remotely placed in comparison with
the grid access.
Fig. 5.4: Radial connection of power areas or systems
The buses at which the tie-link is connected are called the interface buses.
These buses are required to be equipped with WAM devices for online
measurement of the voltage phasors at the interface buses via PMUs,
frequency, and tie-line active and reactive power values. Either area 1 or area
2 may be considered as the study area (SA) while the other one is considered
as the external area (EA). In the following modeling and analysis, the
equivalency of both areas will be considered. The objective is the
minimization of the dynamic order of the overall interconnected systems for
dynamic security studies that will be considered in later in this book. It is
assumed that the generators of each area are coherent. Therefore, their
electromechanical aggregation is feasible. In this chapter, only conventional
sources are considered while various renewable resources will be considered
later in this book.
The equivalency approach is based on two stages both of them is based on
online synchronized measurements of the frequency, voltage phasors and tie-
line power flow. The first stage as shown in Fig. 5.5, involves the estimation
123
of the equivalent inertia at the interface buses while the second stage (Fig.
5.6) involves the estimation of the electrical parameters of the equivalent
generator of each area.
Fig. 5.5: Inertia estimation (stage I)
Fig 5.6: Electrical parameters of the equivalent generator (stage II)
Classical generator models are considered for the equivalent generator.

This simplification can be justified based on the application of the equivalent
simplified models. These models are to be used for fast prediction of the
transient stability of power systems, for example, using the energy functions
or the equal area criterion (EAC). Therefore, the 0+ transient performances of
the generators are of the major importance for the identification of the
stability conditions. As shown in the previous chapter as well as this chapter,
during the initial stages of a transient process (Fig. 5.3), the impact of
generator controls may be neglected and the classical model may be
considered as an adequate model for that purpose.
5.2.1 Stage I: Inertia estimation at the interface buses

The equivalent inertia of the coherent group of generators comprising an
area can be estimated at the area interface bus. This is based on the use of the
measurement of active power and frequency measurements on that bus.
These measurements are performed under disturbed conditions. The
measurements in a very short window after a disturbance are used for
124
estimating the equivalent inertia. Since the inertia estimation is based solely
on measurements, the internal structure of the considered area is not required
and the area can be considered as a black-box. Therefore, the method is
general and can be applied to any mix of generating technologies as well as
large areas. This kind of flexibility cannot be provided with traditional
methods of equivalency which are mainly based on system data sets. For the
purpose of the estimation of the inertia at a specific interface bus, equation
(5.2) is to be used.
5.2.2 Stage II: Electrical parameters of the equivalent generator

The equivalent generator is assumed to be represented by the classical
model (i.e. fixed transient emf ( ) behind a transient reactance ( ) as
shown in Fig. 4.9. Based on the standard model of the steady-state and
approximate transient characteristics of synchronous generators (chapter 4
and Appendix 1), the following equations can be easily proved for the
equivalent generator shown in Fig. 5.7 with ,
Fig. 5.7: An equivalent generator representing an area connected to the

rest of the system through an interface bus.
125
The interface quantities are assumed to be known from measurements

through WAMs. These quantities are the interface bus voltage phasor (Vt and
t) and power flow (Pt and Qt) from the considered area to the rest of the
system. The unknown quantities associated with the equivalent generator are
its emf (Eg), the phase angle of the emf (g), and its reactance (Xg). The nature
quantities are based on the measurements’ conditions i.e. steady state or
transient.
Equations 5.3 to 5.5 are to be solved simultaneously for estimating the
unknown quantities of the equivalent generator. Since these equations are
nonlinear transcendental equations, then the iterative approach is a good
choice for their simultaneous solution. For this purpose, with the index (i)
represents an iteration number, the following equations are derived,
126
With is selected, equations (5.6) to (5.8) respectively can be used to

perform cyclic iterations. The initial value of the equivalent generator emf
can be set to reasonable values such as 1.0 p.u or Vt. the termination criterion
of the iterations process is where is each of the estimated
quantities and is the desired accuracy tolerance (e.g. 10-5). This method is
simple and straight forward in comparison to previous method such as that
presented by Chow et.al 33,34.
With the presented method in this section, the weakly interconnected

power system in Fig. 5.4 can be reduced to the two-machine form presented
in Fig. 5.5 or 5.6.
5.3 SMIB equivalence of multi-machine systems

The online dynamic security assessment and control is required for
enabling power systems to withstand unexpected contingencies in a secure
way i.e. without causing system instabilities. Due to the complexity and
inherent nonlinearity of power systems, the online dynamic security
assessment and control is considered one of the major challenges in recent
power systems. Chapter 1 and 2 (in Volume 1) discuss in details the speed
and accuracy requirements for handling the stability or the dynamic security
of power systems online.
One of the most efficient and fast enough approaches for online
security studies is the online based equivalence such as that presented in the
previous section. With the possibility of reducing a multi-machine system to
a single machine equivalence (SME), not only the speed of dynamic security
assessment is significantly enhanced, but the fast stability analysis methods -
such as the Equal Area Criterion (EAC) - can be implemented on the SME or
33
Chow, J. H. (2013). Power system coherency and model reduction. London: Springer.
34
Chow, J. H., Chakrabortty, A., Vanfretti, L., & Arcak, M. (2008). Estimation of radial
power system transfer path dynamic parameters using synchronized phasor data. Power
Systems, IEEE Transactions on, 23(2), 564-571.
127
the equivalent SMIB system. The SME approach is a hybrid direct-temporal

method for fast handling of the dynamic security or transient stability. The
transient stability of power systems includes two main aspects; analysis and
control. The speed of solution, its accuracy, and the nature as well as the
strength of the disturbances define the appropriate modeling and solution
techniques. For example, the EEA provides a very fast way for assessing the
transient stability of the power system; however, in its original form, the size
of the system is limited to one or two machines. Methods such as the
Extended Equal Area Criterion (EEAC) can be used for extending the
application limits of the tradition EEA. This section provides an improved
EEAC by presenting the online-based SME. Therefore, an interconnected
system such as that shown in Fig. 5.4 can be dynamically transformed to an
equivalent SMIB (see Fig. 5.8) through the presented online equivalency
approach.
Fig. 5.8: SMIB equivalence
High-order nonlinear modeling of individual components forming a power

system results in highly accurate results. The solution, in this case, is in the
time domain; however, neither the execution time nor the volume of the
results to be analyzed is acceptable especially for online applications. Other
methods such as linearized analysis methods are significantly faster in
comparison with the solution of nonlinear models; however, linearized
models are only applicable for simulating small disturbances that result in
small deviations in the operating conditions.
Due to its high economical pressure and constraints, the deregulation of
power systems causes near security limit operation of power systems. In
addition, the deregulation leads to larger interconnections for energy security,
and large power transactions for economical profit maximization. Therefore,
the time needed for system monitoring and security analysis becomes too
small for the current computational technologies and method to handle power
128
systems in real-time. The SMIB equivalence of power system is shown to be

a direct approach to enhancing the real-time handling of the stability and
security of power systems 35.
The derivation of the SMIB equivalence of a power system will be
presented and evaluated in the following. The SMIB equivalence such that
shown in Fig. 5.8 is to be determined based on the two-machine equivalence
(Fig. 5.5 or 5.6) which is estimated as an equivalence to the multi-machine
radially connected areas shown in Fig. 5.4. For simplicity, all resistances are
neglected in this analysis too. This is assumption is also usual in the EAC
based analysis. Based on the equivalent circuit shown in Fig. 5.6, the power
mismatches at the interface buses can be represented by
where Pmax is the maximum power across the interconnection; Pg1 and Pg2
are respectively the power production of the equivalent generators of area 1
and area 2 respectively; Pd1 and Pd2 are the equivalent active power demands
on area 1 and area 2 respectively; 12 is the phase angle difference across the
interconnection which is equal to 1 - 2. Therefore, the swing equation at
each interface bus and the equation of the relative motion can be represented
by,
Therefore,
35
Power & Energy Systems, 70, 99-107
129
In addition, the equation of motion of the SMIB equivalence shown in

Fig. 5.8 can be represented by
where
The equivalent transient reactance of the SME is simply the equivalent

reactance of the two machines. Therefore, the SMIB equivalence of the
system is presented by eqs. (5.14) to (5.16). Although, this model abstracts
the original system (Fig. 5.4) in the very simple form shown in Fig. 5.8, the
SMIB equivalence preserves some important characteristics in the two-
machine equivalence as well as the original system. For example, eq. (5.16)
can be rewritten in the following perturbation form,
It is clear from eq. (5.17) that a drop in Pg2 increases the equivalent power
(Peq) by M1/(M1+M2) while a reduction in Pd2 reduces Peq by M1/(M1+M2).
Therefore, changes of Pg2 and Pd2 have opposite impact on the value of Peq
but its sensitivity to both of them is the same. This is also applicable to the
impact of the changes in Pg1 and Pd1 on Peq; however, a drop in Pg1 reduces
Peq and a drop in Pd1 increases Peq. In this case the sensitivity factor is
M2/(M1+M2). As a direct application of the SMIB equivalence in security
analysis is the estimation of the maximum sudden drop in the power
generation Pg2. In addition, the presented model can be used to determine
corrective actions such as the minimum amount of load shedding for ensuring
the stability of the system if the drop in the generation is higher than the
maximum limit. The validity of this analysis as well as the entire equivalency
approach and the efficiency of the EAC in assessing the stability of the
130
system will be demonstrated later in this book considering the nonlinear

models of power systems for stability analysis.
5.4 Transient stability and the Equal Area Criterion (EAC)

A system is in a dynamic state if the time derivative of any system
quantity (y) is non zero. A dynamic system can be described mathematically
differential or difference equations. According to the IEEE/CIGRE joint task
force on stability terms and definitions 36, power system stability can be
defined as, “The ability of an electric power system, for a given initial
operating condition, to regain a state of operating equilibrium after being
subjected to a physical disturbance, with most system variables bounded so
that practically the entire system remains intact”. This definition indicates
that there are three conditions of system stability; stable, unstable, and
critically stable. Based on section 4.4.4, stable and unstable conditions may
show either non-oscillatory or oscillatory responses. The shape of the
responses depends mainly on the system damping. Stable conditions as
associated with positive damping while unstable conditions are associated
with negative damping. Critical stability conditions are a special situation
where the damping is absent. Fig. 5.9 illustrates various stability conditions.
The IEEE/CIGRE definition of stability is accompanied with two
important comments that are also illustrated in chapter 4. These comments
are as follows:
 Comment 1: “It is not necessary that the system regain the same steady
state operating equilibrium as prior to the disturbance. This would be the
case when e.g. the disturbance has caused any power system component
(line, generator, etc.) to trip. Voltages and power flows will not be the same
after the disturbance in such a case. Most disturbances that are considered in
stability analyses incur a change in system topology or structure”. See Fig.
5.10.
36
Kundur, P., Paserba, J., Ajjarapu, V., Andersson, G., Bose, A., Canizares, C., ... &
Vittal, V. (2004). Definition and classification of power system stability IEEE/CIGRE
joint task force on stability terms and definitions. Power Systems, IEEE Transactions on,
19(3), 1387-1401.
131
(a) (b)
(c)
Fig. 5.9: Stability conditions; (a) Stable transients; (b) Unstable transients;
(c) Critically stable transients (sustained transients)
Fig. 5.10: Illustration of comment 1
 Comment 2: “It is important that the final steady state operating

equilibrium after the fault is steady state acceptable. Otherwise protections
or control actions could introduce new disturbances that might influence the
stability of the system. Acceptable operating conditions must be clearly
132
defined for the power system under study”. For example, the steady-state
acceptable operating points of an unregulated non-salient pole synchronous
generator connected to an infinite bus (see Appendix 1) are shown in Fig.
5.11; as illustrated, an increase in the mechanical power must be associated
with an increase of the power angle for stable operation. Therefore, the left-
hand side of the power-angle curve with respect to the 90o power angle is
stable while the other side is unstable.
Fig. 5.11: Illustration of acceptable steady-state operating points
5.4.1 Classification of power system stability

Power system stability can be classified according to many criteria such
as: type of disturbances, and the physical instability mechanism and time.
According to the type the disturbances, the power system stability can be
classified as:
 Local stability that defines the behavior of the system around the
particular equilibrium point being analyzed. This type of stability is
associated with small disturbances. In this case, the nonlinear models of
power systems which are usually in the form of Algebraic Differential
Equations (ADEs) are linearized around a specific operating point.
Linearized system analysis and model analysis can then be performed for this
type of disturbances.
 Finite stability, which defines the behavior of the system in a finite
region of space. If the region of interest is extended to all possible state
space, it is called global stability for the system. Finite and global stabilities
are usually associated with large disturbances. Due to the nonlinear
133
behaviors of power systems to large disturbances, stability assessment is

required in the simulation of the dynamic evolution of the system or
energy based methods such as the Equal Area Criterion (EAC).
According the physical instability mechanism and time, the stability
problem can be classified according to Fig. 5.12.
Fig. 5.12: Classification of the stability problem according to the physical

instability mechanism and time
Rotor angle stability refers to “the ability of synchronous machines of a

power system to remain in synchronism after disturbance”. The ideal system
for the study of the rotor angle stability is the SMIB system. This type of
stability studies can be classified according to the strength of the disturbance
to either large disturbance stability (i.e. transient stability) or small-signal
stability. The main difference between the two studies is the way power
systems modeled. Small-signal stability also refers to local stability while
transient stability is referred to either finite or global stability.
Frequency stability refers to the ability of a power system to maintain
steady frequency following a severe system disturbance resulting in a
significant imbalance between generation and load. Simply, this type of
stability studies concerns the ability of the system to sustain power balance.
134
Voltage stability refers to the ability of a power system to maintain steady

voltages at all buses in the system after being subjected to a disturbance from
a given initial operating condition. It concerns the ability to maintain/restore
equilibrium between load demand and load supply from the power system.
Instability that may result occurs in the form of a progressive fall or rise of
voltages of some buses. Small disturbance or static voltage stability analysis
usually represents power systems of algebraic equations while large
disturbance voltage stability (also called transient voltage stability) models
similar to those used in the transient stability analysis.
This classification shows specific studies of major interests, according to
the system engineers, while the system dynamics due to one disturbance may
show all kinds of classified studies.
5.4.2 The Equal Area Criterion (EAC) and advancements

The transient stability of power system can be generally studied by
solving the ADEs representing power systems by numerical methods.
Consequently, the time domain (TD) response of the system can be
determined. According to the TD results, the system stability can be assessed.
Fig. 5.9 illustrates some TD responses for various stability conditions.
Despite the certainty of the results of this approach is high, it suffers from
many drawbacks such as,
 The quality of the input data to the models are usually suffering from
uncertainties.
 The pre-disturbance operating conditions and exact topology of the
system are also suffering from uncertainties.
 For a given disturbance in a typical power system, the simulation
results are usually huge to be easily analyzed.
 Real-time dynamic security requires too fast decisions and corrective
actions to be made. A situation that makes the conventional TD approach
impotent to fulfill these time limits.
The EAC37 ,38 is a graphical method for fast assessment of the transient
stability and possibly fast analysis of the feasibility of some corrective
37
Kimbark, E. W. (1995). Power system stability (Vol. 1). John Wiley & Sons.
38
Venikov, V. A. (1977). Transient processes in electrical power systems. Mir Publishers.
135
actions. The method is traditionally used in the analysis of the SMIB system
or two generators. In addition, the method suffers from oversimplifications
such as classical modeling of synchronous generators with damping
neglected; however, this issue is not critical if the method is used for
assessing the first stage of the transient process (also called first swing
stability). In addition, methods for enhancing the EAC modeling are
constructed as will be shown in this section. This scope of stability analysis is
shown to be indicative about the transient stability, but not sufficient under
some conditions. The possibility to produce high quality equivalent models
of power systems and the possibility, for a given large disturbance, to model
a large scale power system by only two machines, extend the applications of
the EAC to cover real power system sizes. For example, if a group of
considered generators connected to a large system through weak link(s) (i.e.
long transmission line(s)), then the study system can be transformed to SMIB
if the group of generators shows sufficient coherency level between them. In
another situation, if two groups of generators tied through weak link(s), the
overall system can be represented by two generators given that each group
show sufficient coherency between its generators. These situations will be
clearly presented in this book. It is worthy to me mentioned that the
equivalences presented in this book are independent on a specific
disturbance.
The Extended Equal Area Criterion (EEAC) is another well-known
method for the Single Machine Equivalence (SME) of power systems;
however, the produced equivalence is dependent on a specific disturbance or
a set of disturbances. The MSE using the EEAC method can be determined
based on classical dataset approaches or using online measurements. The core
idea of the EEAC is based on the fact that regardless of the system size, the
mechanism of loss of synchronism originates from the irrevocable separation
of the system machines into two groups: critical machines, which comprise
the units responsible for the loss of synchronism; and noncritical generators,
that is, the remaining machines. Accordingly, when analyzing an unstable
situation, the SME procedure starts with the identification of the critical
group of machines as soon as the system enters the postfault phase.
Therefore, the EEAC equivalence approach gathers temporal information
about the system dynamics. An equivalent two machine system is then
constructed such each group of machines are s reduced to one machine. Then
136
the two machine equivalent system is reduced to the SMIB system by the
Kimbark’s method. Originally the EEAC is based on datasets and
mathematical modeling of various components in the detailed system while
its recent applications are based on on-line measurements.
A. Fundamental EAC concepts

Consider the SMIB equivalence shown in Fig. 5.8. The following
assumptions are used in the traditional EAC transient stability method,
1. The synchronous machine is represented by the classical model.

2. The approximate transient power-angle characteristics (chapter 4) are
used to represent the system.
3. The input torque to the synchronous machine is constant.
4. The angular momentum of the synchronous machine is constant.
Therefore, the inertia constant is assumed to be constant.
5. The resistances of lines and machines are neglected.
6. All damping torques are neglected.
Suppose the system of Fig. 5.8 is operating in the steady state delivering a
power of Po at an angle of 0. This is operating point a in Fig. 5.13. The
initial power Po equals to the mechanical power input Pm. Due to malfunction
of the line, its circuit breakers open reducing the real power transferred to
zero. This is also equivalent to a temporary 3-phase to ground short-circuit at
the generator terminals. Consequently, the generator power-angle relation
becomes zero instead of the initial relation Pe = Pmax sin . This is illustrated
in Fig. 5.13. Due to its inertia, the speed of the rotor cannot be changed
suddenly i.e.  = d/dt = 0 at the instant of the fault. Therefore, the
transition from the initial power-angle curve to the new power-angle relation
is done with constant power angle i.e. vertical transition a  b. At point b,
due to the power imbalance between the electrical and mechanical power, the
speed of the generator starts to change. Since, the mechanical power is higher
than the electrical power, the generator accelerates i.e. gaining kinetic energy
(KE); the accelerating power Pa becomes equal to Pm; which gives rise to the
rate of change of stored kinetic energy in the rotor masses. Therefore, the
rotor will accelerate under the constant influence of non-zero accelerating
137
power (equals to Pm) and hence the load angle will increase in a continuous
manner. Consequently, the operating point moves from b toward c.
Fig. 5.13: EAC basic concepts
Now suppose the circuit breaker re-closes at an angle cl (point c). The
power will then revert back to the normal operating curve. Therefore, the
operating point moves suddenly from c to e. At point e, the electrical power
will be more than the mechanical power and the accelerating power will be
negative i.e. it becomes decelerating power. This will cause the machine to
decelerate; however, due to the inertia of the rotor masses, the load angle will
still keep on increasing along the path e  f. The increase in this angle may
eventually stop if the rotor losses the excess KE during the acceleration stage.
This is shown at point f where the rotor is at higher KE with respect to
normal KE at the synchronous speed but at that point the KE gained due to
the inertia of the generator is lost and the rotor starts to decelerate to move
back along the f  e  d  a path. When the rotor reaches point a, its speed
equals to the synchronous speed but also due to its inertia, the rotor continue
to decelerate and moves along the path a  0 while it is subjected to
accelerating power as Pm > Pe. This action continue till the rotor starts to
accelerate again. The rotor in this case oscillate around point a (which is the
post-disturbance steady state operating point) and settles at that point again
after a period of oscillations (Fig. 5.14).
138
Fig. 5.14: Stable oscillations
It can be easily depicted from Fig. 5.13 that the stable situation is limited
by a rotor angle cr after which the subsequent operating points are associated
with acceleration (Pm > Pe) rather than deceleration. The possible
deceleration region (dehd) defines the maximum acceleration energy gain
which can be absorbed by the system without causing loss of synchronism
i.e. without causing instability. Otherwise, the system will be unstable. The
KE gain due to the considered disturbance will be shown to be proportional
to the area abcda. Note that,
d d   d   d  
2 2
   2   2  (5.18)
dt  dt   dt   dt 
Hence, multiplying both sides of the swing equation by d dt and

rearranging we get
H d d  d
2
    Pmo  Pe  (5.19)
 f o dt  dt  dt
Multiplying both sides of (5.19) by dt and then integrating between two

arbitrary angles 0 and  we get
139
2 
H d 
 f o  dt 
   Pmo  Pe d  (5.20)
0 0
The left-hand side (LHS) of (5.20) is proportional to rotor KE energy

while its right-hand side (RHS) is the area under the curve given the range of
the angle. Therefore, (5.20) represents that rotor KE. Now suppose the
generator is at rest at 0. Therefore, at  = o, d/dt = 0. Once a fault occurs,
the machine starts accelerating. Once the fault is cleared at  = cl, the
machine starts to decelerate. Therefore, the acceleration area is given by,
cl
A1  Aacc    Pmo  0d   0 (5.21)
0
In a similar way, we can define the area of possible deceleration.
cr
A 2  A possdec    Pe  Pmo d   0 (5.22)
cl
Let Adec (or A3) be a deceleration area that counteracts the acceleration
area Aacc i.e. Adec = Aacc. If the fault is cleared such that Adec  Apossdec, then
the system is stable and the maximum angle of swing is m. In this case, the
difference A = Apossdec – Aacc is considered as the transient stability margin
(Kt); as described by (5.23), it can be used as an indicator of the stability. In
addition, as A increases, the stability is stronger. The value of A can be
used also to study the effect of various factors on the transient stability.
As shown in Fig. 5.14, the system is unstable if Adec  Apossdec. If Adec =

Apossdec, then the system is critically stable and the fault clearing angle is the
critical clearing angle limit (CCL) which is given the symbol ccl. In
140
addition, the maximum angle of swing in this case equals to cr. This is
represented by (5.24) while the general stability criterion is stated by (5.25).
ccl cr
  Pmo  0d     Pe  Pmo d  (5.24)

0 ccl
where
For the considered case shown in Fig. 5.13, the critical clearing angle
(CCA) ccl can be determined using (5.24) and the result is,
For the considered pure sinusoidal power-angle relation, .

Therefore, and then (5.27) can be written as,
Since , then equation (5.28) can be reduced to,
Based on (5.29), the relation between critical clearing duration ( )

and the generator loading can be clearly viewed. This relation is illustrated in
Fig. 5.15. It is clear that the increase in the generator loading results in a
significant decrease in the available clearing duration in radians. The usual
units of time is seconds. Therefore, for the considered disturbance, the critical
141
clearing time can be determined analytically by solving the swing equation

during the fault period. During that period, Pe = 0. Therefore,
d 2  f o
 Pmo (5.30)
dt 2 H
d t f o f
 Pmo dt  o Pmot (5.31)
dt 0 H H
t
f o s
  Pmot dt  Pmot 2   0 (5.32)
0
H 4H
with s  2 f o , the time corresponding to an operating angle  such that o 

  ccl takes the form
4H
t     0  (5.33)
s Pmo
and, the critical clearing time (CCT) is then,
4H
t ccl     0  (5.34)
s Pmo ccl
Fig. 5.15: Impact of initial loading on the available critical clearing

142
Example 1
Critical clearing and transient simulation of a temporary three phase
fault at the generator terminals
In this example, a nuclear power station connected to an infinite bus

through pure reactive transmission link is considered. The rating of the
station is 2220 MVA and the link voltage is 400 kV; these values are the
basis of the per unit quantities. The generator is with non-salient pole and its
steady state reactance equals to 1.82 p.u while its transient reactance equals
to 0.2 p.u. The transmission link reactance is 0.3 p.u. The generator inertia
constant M is 8.0 seconds. The generator delivers 0.9 p.u active power to the
system. At this operating point, the measured terminal voltage magnitude
equals to 1.0432 and its phase angle is 0.261905 rad (15 o) while the infinite
bus voltage magnitude is 1.0 p.u and its phase angle is taken as the reference
angle.
It is required to determine the critical clearing time of a temporary three-
phase at the terminals of the generator. In addition, it is required to evaluate
the accuracy the result through TD simulation of the non-linear swing
equation. The generator damping as well as all the losses is neglected. The
critical clearing angle can be determined using (5.28) i.e.
Therefore, the initial transient power angle of the generator is to be

determined. Based on chapter 4 and appendix 1,
The value of Po is given as 0.9 p.u while the value of V is given as 1.0
p.u. The transient emf is unknown but can be determined using,
143
All values in this equation are known except the value of the reactive
power Qo; however, this value can be determined using the measured
terminal voltage phasor. Considering that,
The value of the reactive power injected to the infinite bus is 0.024097
p.u. Consequently, the value of E’ is 1.107584 p.u and its phase angle can be
obtained from,
and it is found to be 0.41839 rad (23.96233o) as explained in Appendix 1,

this value is approximately the power angle of the generator. Now the CCA
can be calculated; its value is 1.547603 rad (88.6344 o). The corresponding
CCT can be found using (5.33) and its value is then 0.252 seconds.
For evaluating the results, the system is simulated using the Power
System Analysis Toolbox39 (PSAT) as shown in Fig. 5.15. The three-phase
fault is applied at the generator terminals at t = 0.5 second and cleared after a
specific clearing time. The clearing time is set to three different values in
seconds; Case 1: 0.2 < CCT, Case 2: 0.252 = CCT, and Case 3: 0.253 > CCT.
This is to illustrate the validity of the results and also to study the
performance of the system as well as to illustrate the causes of stability and
instability from the TD prospective. The TD simulation results are shown in
Fig. 5.16 to 5.20.
39
Milano, F., Vanfretti, L., & Morataya, J. C. (2008). An open source power system
virtual laboratory: The PSAT case and experience. Education, IEEE Transactions on,
51(1), 17-23.
144
Fig. 5.15: PSAT model of the study system
The ID of each shown quantity refers to the clearing time case. It is clear
from the results that the estimated TTC is accurate. The results also reveal
some dynamic characteristics of power systems without considering various
damping providers and impact of controllers. Approaching the CCT results in
an increase in the amplitude of the angle oscillations and reduces their
frequency. This is also shown in the voltage, speed, and active power
responses. Clearing the fault at a time larger than the CCT results in an
unlimited increase in the power angle. The transient voltage performance
shows a voltage collapse when the clearing time is larger than the CCT.
Therefore, the TVS as well as the angle are lost. The active power transients
show a tight relation with the EAC. This is illustrated by the acceleration and
deceleration areas from the power-time prospective. It is clear that the
instability (case 3) is caused by the increase of the acceleration areas in
comparison with the deceleration areas. In addition, in this case, the generator
could not decelerate after the second swing and it continuously accelerates. In
the other cases (i.e. 1 and 2), the acceleration areas are less than the
deceleration areas.
It is also interesting to see the (-) and (Pe-) trajectories for the
considered cases. The results are shown in Fig. 5.21 and 5.22. As shown in
these figures, that the stable cases (i.e. cases 1 and 2) are characterized by
sustained oscillations around the initial steady state operating point. As
shown in Fig. 5.21 and 5.22, the instability is shown as an unlimited increase
in the speed and the outage of the generator power.
145
Fig. 5.16: Power angles for cases 1 and 2
Fig. 5.17: Power angle for case 3
Fig. 5.18: Rotor speeds for cases 1 and 2

146
Fig. 5.19: Terminal voltage
Fig. 5.20: Active power
147
(a) (b)
(c)
Fig. 5.21: The - trajectories for various cases
It is worthy to be mentioned here that in real systems, there are many

sources of damping of oscillations (see chapter 4). Therefore, the amplitudes
of the oscillations are reduced with time till the final steady state operating
point is reached. For the considered zero-input disturbance, the final and
initial operating points are the same. This is illustrated in Fig. 5.23
148
(a) (b)
(c)
Fig. 5.22: The Pe- trajectories for various cases
149
(a) -t (b) -t
(c) Pe-t
(d) - trajectories (e) Pe- trajectories

Fig. 5.23: Stable transient response with damping considered
150
B. EAC-based analysis of 3-phase faults away from the generator

terminals
This section concerns with the EAC-based stability analysis of arbitrary
balanced 3-phase faults in the transmission link connecting the generator to
the infinite bus. As shown in Fig. 5.24, a balanced 3-phase fault occurs at the
one of the transmission lines connecting the generator to the infinite bus. The
fault is assumed to be occurred at a specific distance from the generator
terminals over one of the lines. The fault is then provides a reactance aXl
between its location and the terminal bus; where a is a factor proportional to
the length of the faulty section w.r.t the terminal bus. The reactance of the
remaining section of the faulty line is bXl where a + b = 1.0.
Fig. 5.24: The SMIB system with 3-phase fault on a transmission line
As shown in Fig. 5.25, three stages characterize the fault process each of
which is characterized by a power-angle curve. These curves are:
 The pre-fault curve (I); describes the initial power-angle curve of the
system. The operating angle in this case is .
 The during-fault curve (II); describes the power-angle characteristics
during the fault i.e. before clearing the faulty line.
 The post-fault curve (III); describes the power-angle characteristics
after clearing the fault. If the system is stable, then the final steady state
operating angle is . Otherwise, the system will
not reach a post-fault steady-state equilibrium point due to its instability.
The figures also show the stable and unstable conditions from the P- and
P-t prospective.
151
(a) Stable conditions; P- domain (b) Stable conditions; P-t domain
(c) Unstable conditions; P- domain

(f) Unstable conditions; P-t domain
Fig. 5.25: EEA-based analysis of general balanced three-phase faults
The three curves are modeled by the same equation which is the
approximate transient power-angle characteristics; however, in each of the
three conditions the transfer reactance is different. This is illustrated in Table
5.2.
152
Table 5.2: Various power-angle curves

Curve (i) System topology
i = I, II, and III

(I) Pre-fault
(II) During-fault
(III) Post-fault
The values of X12 are determined using standard circuit reduction methods
(i.e. series, parallel, -Y, and Y- transformations) as will be illustrated in
the following example. It can be easily shown that the CCA (i.e. the angle at
which Aacc = Apossdec) takes the following form. Due to the nonlinearities of
the system with the considered fault, the CT and CCT cannot be expressed by
analytical expressions; however, numerical methods such as the trapezoidal
rule or the Euler method may be utilized for obtaining TD solutions.
It is important to note that for the same CCA, the CCT may be changed
from a system to another system. This is because the CCA is based on the
P.E/K.E balance while the CCT is a function of the system inertia. The
increase in the system inertia leads to increase in CCT while the CCA
remains the same for the same electrical quantities, topology, and
disturbance. It is of critical importance to understand that the dynamic
security is mainly guaranteed by the transient stability; however, the transient
153
stability is not a guarantee of the static security (or adequacy); see Appendix
6 for more details.
Example 2
Critical clearing and transient simulation of a three phase fault in the
transmission system
Consider a simple system in which a generator is connected to an infinite

bus through a double circuit transmission line as shown in Fig. 5.26. The per
unit system reactances that are converted into a common base, are also shown
in this figure. Let us assume that the infinite bus voltage is 10. The
generator is delivering 1.0 per unit real power at a lagging power factor of
0.9839 to the infinite bus. While the generator is operating in steady state, a
three-phase bolted short circuit occurs in the transmission line connecting
buses 2 and 4  very near to bus 4. The fault is cleared by opening the circuit
breakers at the two ends of this line. The objectives of this example are to
illustrate the analysis of the considered faults via the EAC and to determine
the impact of the generator inertia on the permissible (i.e. critical) clearing
time via TD analysis.
Fig. 5.26 SMIB with fault in the transmission system
Let the current flowing from the generator to the infinite bus be denoted
by I. Then the power delivered to the infinite bus is
From the above equation,

154
I  1.0164  cos 1  0.9839 

p.u
 1.0164  10.3
The total impedance during the time when both the lines are operational
(Fig. 5.27 (a)), the impedance between the generator and the infinite bus is
j(0.3+0.1+0.1) = j0.5 p.. Then the generator internal emf is
E   1.0  j 0.5 1.0164  10.3  1.224.625
Therefore the machine internal voltage is E = 1.2 per unit its angle is
24.625 or 0.4298 rad. The equivalent circuit of the system for various stages
of the disturbance (i.e. pre-fault, during-fault, and post-fault) are shown in
Fig. 5.27. In the during-fault phase the equivalent reactance X12 is determined
by the two method described in section A1.4 of Appendix 1.
Fig. 5.27: Various transfer reactances; (a) Pre-fault; (b) During-fault; (c)
Post-fault
Based on Fig. 5.27,
155
1.2
Pre  fault: PeI  sin   2.4sin 
0.5
During  fault: PeII  0.857sin 
1.2
Post  fault: sin   2.0sin 
PeIII 
0.6
The three power-angle curves are shown in Fig. 5.28. From this figure we
find that
1
cr    sin 1    2.618 rad
 2
Fig. 5.28: Power-angles curves for the three phases of operation of the
system
The CCA can be directly calculated using (5.36) or by the calculations of

the areas as follows. The accelerating area is given by,
ccl
Aacc   1  0.857sin  d 
0.4298
 ccl  0.4298  0.857cos ccl  0.857cos  0.4298

 ccl  0.857cos ccl  1.2089
and the decelerating area is
ccl
Aacc   1  0.857sin  d 
0.4298
 ccl  0.4298  0.857cos ccl  0.857cos  0.4298

 ccl  0.857cos ccl  1.2089
156
and the decelerating area is

2.618
A possdec    2sin   1d 
ccl
 2cos  2.618   2cos ccl  2.618  ccl

 ccl  2cos ccl  0.8859
Equating the two areas we get
 0.323 
ccl  cos 1    1.8573 rad  106.41
 1.1429 
As mentioned earlier, the critical clearing angle depends on the system
configuration. The critical clearing time, however, is dependent on the value
of the inertia constant H (or M = 2H). Usually numerical methods are
employed in finding out the clearing time and for solving the swing equation
in the TD. Appendix 7 presents some numerical integration methods for this
purpose.
In this example, the CCT is estimated using the MALAB ODE solver and
the results are shown in Fig. 5.29. It is clear from the results that the CCT is
very sensitive to the inertia constant. It can be seen that as the value of H
increases, the CCT is also increasing; however, the critical clearing angle
remains the same. This is obvious; as H increases, the response of the system
becomes more sluggish due to larger inertia (see chapter 4). Therefore, the
rotor takes more time to accelerate. The results best fit with the following
function,
In this example, k = 0.1968.
Fig. 5.29: Impact of the inertia constant on the CCT
157
C. EAC-based analysis of unbalanced faults

With unbalanced faults, the stability analysis is to be performed in the
same way described in the previous sections; however, the impact of the
unbalanced fault on the during-fault transfer reactance must be
considered. The determination of the during-fault transfer is based on the
standard fault analysis methods40,41. This requires the construction of the
positive, negative, and zero sequence networks of the system, then a
connection between these networks according to the considered fault. Table
5.3 summarizes the connection between various sequence networks according
to various unbalanced short-circuit faults. Table 5.4 summarizes the zero-
sequence equivalent circuits of major system components.
Table 5.3: Sequence networks for unbalanced S.C faults

Fault
SLG LLG LL
Representation
Sequence network connection
40
Grainger, J. J., & Stevenson, W. D. (1994). Power system analysis (Vol. 31). New York:
McGraw-Hill.
41
Nagrath, I. J. (2003). Modern power system analysis. Tata McGraw-Hill.
158
Table 5.4: Sequence impedances of system components

Representation Impedances Z012
Balanced loads
Lines
Generators
Three single-phase units and five-legged

core three-phase units
; Zo 
Transformers*
Three-legged core three-phase units
* The magnetization current and core losses represented by the shunt branch are
neglected (they represent only 1% of the full load current). The transformer is modeled
with the equivalent series leakage impedance Zl.
159
Example 3
EAC for unbalanced faults
Consider a simple system with two identical transmission lines, as shown
in Fig. 5.30. A single-line-to-ground (SLG) fault is assumed in this example
and the fault is cleared by disconnected the three phases of the faulted line.
Fig. 5.30: The SMIB with an LG fault
Fig. 5.31: During-fault transfer reactance for an SLG fault
The pre-fault and post-fault transfer reactances can be found as

demonstrated in the previous example. During the fault, the connection
between the positive, negative and zero sequence networks for the SLG fault
is shown in Table 5.3 and the sequence impedances of various components
are shown in Table 5.4. For the considered system, the during-fault transfer
160
reactance can be obtained as shown in Fig. 5.31. The rest of the EAC-based
analysis or TD simulations can be performed using the resulting power-angle
diagrams. It can be easily depicted that the during fault transfer reactance
with the SLG fault will be high in comparison with a three-phase at the same
location. Consequently, the peak of the during-fault will be higher and the
acceleration area will be smaller. As a result, the impact of the SLG fault on
the transient stability is lower than the impact of three-phase fault.
D. EAC-based treatment of the damping

In the fundamental modeling for the EAC, it is shown than the machines
are represented only by the swing equation with the damping neglected as
well as all losses. Therefore, the original EAC only considers the
electromechanical transients while the electromagnetic transients as well as
the damping power are neglected. It is shown in chapter 4 and in this chapter
that the damping has a key role the stabilization of power systems. The
damping is provided by many causes including the machine design, field-
circuit dynamics, tuning controllers, and various losses. In chapter 4, a simple
representation of the damping power is given as where  = -
o. During rotor acceleration, the speed deviation  > 0. Consequently, the
damping power is positive. During the deceleration an opposite situation is
faced i.e.  > 0 and PD < 0. It is shown in Fig. 4.11, that the damping
coefficient is a function of the power angle; however, the average value of
the damping coefficient is usually used in the simplified stability and EAC
analysis. Therefore, the damping power is linearly related to the speed
deviation. Consequently, in the P- domain, it is shown by Willems42 that the
effect of damping can be represented with straight inclined lines with a
constant slop,
This is also valid in multi-machine power systems 43. In all cases, the
generators are assumed to initially running at the synchronous speed. The
42
Willems, J. (1969). Generalisation of the equal-area criterion for synchronous machines.
Electrical Engineers, Proceedings of the Institution of, 116(8), 1431-1432.
43
Willems, J. L. (1971). Direct method for transient stability studies in power system
analysis. Automatic Control, IEEE Transactions on, 16(4), 332-341.
161
modification in the EAC for inclusion of the damping effect is shown in Fig.
5.32 considering a sudden increase in the mechanical power input.
Fig. 5.32: Modified EAC for inclusion of the effect of the damping –
sudden increase in the mechanical power
In the original EAC, the stability condition is that A1  A2 while with the
damping power considered the acceleration and deceleration areas are
modified as shown in the figure and the stability condition becomes:
where
In this case,
(5.41)
It can be easily depicted from Fig. 5.32 and eq. (5.40) and (5.41) that A3 <
A4. Therefore, the resultant increase in the acceleration area is less than the
resultant increase in the possible deceleration area. Consequently, as
expected, the system is more stable with the damping considered. In other
162
words, neglecting the damping underestimates the system stability; a

situation which does not cause harmful decisions and also enhances the
avoidance of the system uncertainties. As shown in Fig. 5.33, the same
approach can be used for either balanced or unbalanced faults.
Fig. 5.33: Modified EAC for inclusion of the effect of the damping –
short-circuit faults
In this case,
(5.41)
It is important to note that the single-machine equivalence presented in

section 5.3, can be analyzed as presented in this section; however, the
equivalence with damping considered will be valid if .
Otherwise, it is better to neglect the effect of the damping.
G. Demonstration of the impact of AER and PSS on the EAC stability

In the previous chapter, it is shown that the increase in the gain of
AERs can result in system instability. This is due to the added negative
damping to the system; however, with low AER gain the stability may be
preserved, but the steady-state voltage error may be high. It is also shown that
PSSs can be designed and tuned for improving the system stability. PSSs can
163
be used for restoring the system stability when the instability is caused by
negative damping caused, for example, by high-speed high-gain AERs.
In this section, analysis of the impact of various controllers on the stability
and the areas of the EAC will be presented considering stable and unstable
conditions. The system of Example 1 is considered, but the following
modeling and operational conditions are implemented:
 The generator is represented by the third-order dynamic (non-linear)
model; however, the same initial power produced.
 The transmission system is constructed of two-parallel lines as shown
in Fig. 5.34.
Fig. 5.34: SMIB system for demonstration of the impact of controllers
From control points of view, the following cases are considered,

 Case 1: The generator is unregulated.
 Case 2: The generator is equipped with high-speed low-gain AER (KA
= 20).
 Case 3: The generator is equipped with high-speed low-gain AER (KA
= 20) and tuned PSS.
 Case 4: The generator is equipped with high-speed high-gain AER (KA
= 200) that is selected to cause instability due to its added negative
damping.
The considered disturbance is a rapid switching on one of the transmission
lines due to a temporary fault on the line. The rapid switching is provided by
Automatic Circuit Reclosers (ACR). The response of the system for each of
the stated control conditions is simulated using the PSAT and the results are
compared. A small simulation time is considered to show some major
differences in the responses for various cases. The results are shown as
follows.
Fig. 5.35 shows that cases 1, 2, and 3 are stable while case 4 is unstable. It
can also be shown that the low gain AER reduces the damping of the system
164
in comparison to the unregulated case; however, the PSS is capable of

increasing the damping to a level higher than that of the unregulated system.
The presence of high gain AER shows sever reduction in the system damping
and the net damping became negative causing the system instability.
Fig. 5.36 shows that the high gain AER causes out of step of the generator
while the other generators of the cases remain synchronized. It is also clear
that, as previously explained, the initial gradient of the speed (or frequency)
is not affected by the presence of the controller even in the unstable case.
This is because the initial frequency gradient is mainly dependent on the
generator inertia.
Fig. 5.37 confirms the previous results and also shows that the speed and
frequency convergence to – at the end of the transient process – to the initial
point if the system is stable while divergence is detected when the system is
unstable.
Fig. 5.35: (t) relations
Fig. 5.36: (t) relations

165
(a) Case 1
(b) Case 2
(c) Case 3 (d) Case 4

Fig. 5.37: - trajectories
Fig. 5.38: Field input voltage
166
Fig. 5.39: The transient emf (i.e. the internal voltage)
Fig. 5.38 shows that case 1 is characterized by constant field voltage as

the machine is unregulated. The field voltage in the other cases is time-
dependent due to the presence of the AER. In these cases, oscillatory
transient variations of the field voltage are detected. These variations are with
decaying amplitudes in stable cases, while raising amplitudes are detected in
the unstable case.
Fig. 5.39 shows the variations of the transient emf. In the unregulated
machine case, it is clear that the variations of the transient emf are small in
comparison with the regulated machine. Therefore, the constant transient emf
assumption is valid in the unregulated systems. With the presence of the
AER, the stable and unstable oscillatory responses are clearly shown.
Fig. 5.40: Terminals voltages of the generators
Fig. 5.40 shows the variations of the terminal voltages. It is shown that the
high gain AER (case 4) could not surpass the other cases, although it
167
provided the highest field forcing at the beginning of the transient process;
however, due to its inherent negative damping, the magnitudes of the voltage
oscillations could not be bounded.
Fig. 5.41: Electrical active power
Fig. 5.41 shows the electrical power as a function of time. These functions
can be used to study the EAC in the time-domain as illustrated in Fig. 5.25. It
is shown that the high-gain AER causes small reduction in the acceleration
area and a relatively large reduction in the deceleration area. This is due to
the negative damping provided by this AER. This phenomena can also be
explained through the P- curves shown in Fig. 5.41. These curves are
obtained from the TD simulations using appropriate time and other settings.
The impact of positive and negative damping on the transient stability can
be explained using the Willems’s method presented in the previous section.
This is shown in Fig. 5.42. In comparison with a system with damping
neglected, the positive damping increases both the acceleration and possible
deceleration areas; however, the gain in the possible deceleration area is
much higher. With negative damping, both areas are reduced, but the
reduction in the possible deceleration area is very high. Therefore, the
instability in this situation is very likely.
168
Case 1 Case 2
Case 3 Case 4
Fig. 5.41: P- relations
169
Unregulated system
System with positive damping System with negative damping

Fig. 5.42: Impact of the sign of the damping coefficient on the stability
5.4.3 Transient Stability Enhancement

The main approaches for enhancing the stability of power systems
include, but not limited to:
1. Reduction of the disturbing influence. For example, use of high-speed
fault-clearing and rapid switching may be used for for minimizing the fault
severity and duration. Disconnection of faulted phases only is also a way for
reducing the impact of faults. This can be realized by the use of single-phase
breakers.
2. Increasing the restoring synchronizing forces. This approach includes
the proper design of additional controllers. These controllers may be designed
for controlling the dynamic performance of individual generator or for a
170
group of generator or on system-wide scope. The main objective of generator

control for stability enhancement should reduce the accelerating area or
increase the possible decelerating area or both of them. Generator controllers
include PSS, fast-valving, and braking resistance. Network controllers
include FACTS devices, and POD.
3. Enhancing the system configuration and operating conditions for
stability enhancement. For example, reducing the power flow over long weak
links and reducing the dependency between power areas are possible ways
for enhancing the stability. Increasing the transmission as well as the
generation reserve can serve as a strong defense to instabilities.
4. Proper choice of intrinsic limits and setting of operating limits for
secure operation of system components. Protective devices and systems play
a major role in providing that security. In addition, proper testing, modeling,
simulation, and analysis of system components and system configurations are
the main way of determining the proper settings.
5. Optimally set corrective actions for various contingencies for
minimizing the impact of disturbances on the system intactness considering
the techno-economic feasibility of the actions. Load shedding is an example
of corrective actions that provide a tool for keeping system stability when
large amounts of power production drops; however, as will be shown later,
the load shedding should be optimized. The optimization includes the time of
implementation, and the amount of load curtailment.
In this section, some methods for the transient stability enhancement will
be summarized; however, some other methods such as AER and PSS are
already presented, while some other techniques such as FACTS-POD will be
presented in the next chapters. Consider the SMIB system shown in Fig. 5.43.
Fig. 5.43: SMIB system with a fault at one of the transmission lines
171
A. Fast Valving. The impact of fast changes in the mechanical power for
stability enhancement on the EAC-based stability is demonstrated in Fig.
5.44(a). As shown, these changes cause significant reduction in the
acceleration area and simultaneously significant increase in the possible
deceleration area. Consequently, the system stability is highly enhanced. The
control action required within the turbine to produce such rapid changes in
the mechanical power is called fast valving44 which is also called early valve
actuation. The sequence of operations encountered and characterizing
durations of a fast valving process are shown in Fig. 5.44(b).
(a)
(b)
Fig. 5.44: Turbine fast valving; (a) EAC analysis; (b) Sequence of valve
operations
44
Edwards, L., Gregory, J. D., Osborn, D. L., Doudna, J. H., Pasternack, B. M., &
Thompson, W. G. (1986). Turbine fast valving to aid system stability: benefits and other
considerations. Power Systems, IEEE Transactions on, 1(1), 143-153.
172
Fast valving requires a very fast response from the turbine. To be useful,
the reduction in the mechanical power should be as soon as the fault
occurred. For example, after 30% of the first swing period, which is in the
order of a few tenths of a second following a disturbance. The mechanical
power restoration to the initial pre-fault value should also be fast provided
that the reduction causes system stability. For example, the mechanical power
should be restored in less than a second.
Due to their slow capability of dynamic changes in the mechanical power
(see Appendix 5), hydro-turbines cannot usually provide fast valving. Steam
turbines, on the other side, have faster response in comparison with hydro-
turbines and fast valving is possible. As shown in Fig. 5.44, the change in the
mechanical power is a function of the rotor speed; however, additional
supplementary control action should be added to the standard closed-loop
STG governors. These controllers are needed for compensating the slow
changes in the speed at the beginning of a transient process due to the STG
inertia. Therefore, this supplementary control should be capable of
identifying fault conditions and perform fast-valving within a reasonable
time.
B. Generator tripping 45 can be considered as an alternative to the fast

valving. Selective tripping of generators can be used in improving the
stability of power systems during severe disturbances. By doing so, the
power transfer to a fault is prevented through rapid trip of selected
generators. Unless special facilities are provided, the generating-unit follows
the normal sequences of shutdown and start-up. The look-up table method
presents a fast way to deciding the generators to be tripped for dynamic
security of power systems. In addition, the look-up table only needs to be
updated in the cases of significant topological changes in the system.
C. Rapid switching and high-speed fault clearing. Since the kinetic

energy gained by the generators during a fault is directly proportional to the
duration of the fault, then reducing the fault clearing time is a way for
45
Karady, G. G. (2002). Improving transient stability using generator tripping based on
tracking rotor-angle. In Power Engineering Society Winter Meeting, 2002. IEEE (Vol. 2,
pp. 1113-1118). IEEE.
173
enhancing the stability. The positive impact of rapid switching on the system
stability is demonstrated in the previous section. Fig. 5.45 shows the impact
of reducing the fault clearing time on EAC areas. As shown, the reduction
the clearing angle results in a reduction in the acceleration area and an
increase in the possible deceleration area. Consequently, the stability is
enhanced.
Fig. 5.45: Impact of fault clearing reduction
D. Reclosers. The use of reclosers also secures the system stability for the
momentary faults conditions. Two-cycle circuit-breakers, together with high
speed relays and communication, are now widely used in locations where
rapid fault clearing is important. Recently, a one-cycle circuit breaker by the
Bonneville Power Administration (BPA) combined with a rapid response
over-current type sensor is constructed for special applications that need very
fast fault clearing. Recently, ultra-high-speed relaying systems for EHV lines
are available. The operation of these relays is based on traveling wave
detection.
E. Dynamic braking by braking resistors is also shown to be an effective

method for stability enhancement using variable structure control 46. In this
46
Wang, Y., Mohler, R. R., Spee, R., & Mittelstadt, W. (1992). Variable-structure FACTS
controllers for power system transient stability. Power Systems, IEEE Transactions on,
7(1), 307-313.
174
method, an artificial electrical unity power-factor load is added at the

generator terminals during a transient disturbance. The main objective of the
dynamic braking is to increase the generator’s electrical power output.
Consequently, the rotor acceleration is reduced. One form of dynamic
braking involves temporary switching in shunt resistors for about 0.5 seconds
following a fault for enhancing the stability nearby generator(s).
Consider the SMIB system of Fig. 5.46. Without the braking resistor,
the power-angle characteristics of the system take the form,
with the resistance included, the power-angle characteristics can be

determined using the YBUS-based approach presented in volume 1/chapter 3 or
the network reduction methods presented in Appendix 1; however, as the
system size is small, the power-angle relations will be derived based on the
basic circuit analysis theories. The generator current phasor takes the form,
The terminal voltage as a function of the infinite bus voltage takes the
form,
The line current as a function of the generator and resistor currents takes
the form,
175
(a)
(b)
Fig. 5.46: Braking resistors for stability enhancement; (a) System
topology; (b) Equivalent circuit
Eliminating Il from (5.44) using (5.45) and separating Vt from the

resulting equation, the result is,
Eliminating Vt from (5.43) using (5.46) and separating Ig, the result is,
The apparent power can be then determined as,
Therefore, the active power production with the presence of the braking
resistor takes the form,
176
It can be easily shown that (5.49) is reduced to (5.42) when Rbr =  i.e.
disconnected.
As an example, the parameters and operating conditions of the system of
Fig. 5.46 are Po = 0.6 p.u, Qo = 0.45 p.u, E’ = 1.3172 p.u, V = 1.0 p.u, x’d =
0.3 p.u, xe = 0.4 p.u, and M = 20 sec. Consequently, Vt = 1.3804 p.u, and Pmax
= 1.8817 p.u. Using (5.49), the changes in the power-angle relation as
affected by Rbr are shown in Fig 5.47. It is clear from the figure that reducing
the braking resistance shifts-up the P- relation. The braking power
consumption for a given resistance can be determined at the initial voltage
applied to the resistance i.e.
The initial braking power is reduced with the increase in the barking
resistor; see Fig. 5.47. As the braking resistors increased, the braking power
decreased and consequently the impact of the barking resistor on the P-
relation is reduced.
Equation (5.49) can be simplified by considering that
. The resulting equation is then,
where is the maximum power before inserting the

braking resistor; and
177
(a)
(b)
(c)
Fig. 5.47: Braking resistance; (a) Effect on the P- relations; (b) Braking power; (c)
Increase in the maximum power with increase in the braking power
Considering the trigonometric relation,
with and ; if the angle is within the first

quadrant than . Otherwise, the angle is
determined according to Fig 5.48.
178
Fig. 5.48: Determination of the phase-shift in eq. (5.53)
Based on (5.53), (5.51) can be written as,
where
.In comparison with the detailed power-angle model (5.49), the

approximate model (5.51) and its alternative form (5.54) nearly give the same
power-angle characteristics. This is demonstrated using the sample system
data of this section with Rbr = 2.0 p.u as shown in Fig. 5.49.
Fig. 5.49: P- characteristics with the presence of a braking resistor – various
models and approximations
179
The value of c3 is very close to unity as .Furthermore, if

the phase-shift (5.55) is also neglected, the resulting equation takes the form,
Therefore, the P- relation with the presence of the braking resistor is
reduced to the P- relation without the braking resistor added but shifted up
by the factor KP. The stated assumptions show acceptable accuracy as shown
in Fig. 5.49.
It important to note that the braking power is of a constant value;
however, it is not constant as the resistor is connected to a variable bus
voltage. With the classical model considered (i.e. E’ assumed to be constant),
the variations of the braking power with the power angle can be found using
where
The relation Pbr- is shown in Fig. 5.49 which reveals that the braking
power is decreased with the increase of the power angle. The maximum value
of the braking power occurs at  = 0. This characteristics is beneficial in the
transient stability enhancement as the acceleration area (which is usually at
low power angles < 90 o) will be highly reduced due to the braking power.
The impact of the braking power on the possible deceleration area is positive;
however, its impact is minor as the values of the braking power at high
angles are low.
The impact of sudden addition of a braking load on the power-angle
characteristics as well as the associated dynamics is illustrated in Fig. 5.50.
The figures show the insertion of a braking resistor at  = br. The model of
(5.57) is used for this demonstration.
180
Fig. 5.50: Dynamics of the insertion of a braking resistor
At the insertion of the braking resistor (point a), the generator will
decelerate and goes through oscillatory response till the operating point
settles at point (c). The angle at point c is less than the initial angle due to the
braking power. Braking resistors are connected to the system just after a fault
that causes system acceleration. Usually, braking resistors are disconnected
from the system after a short duration of a transient process.
As an illustrative example, consider the SMIB system of Fig, 5.46. The
system is subjected to a forced outage of one of the lines due a fault. The
resistor in this case operated only during the acceleration and deceleration
periods. As shown in Fig. 5.51(a), the braking resistor causes a decrease in
the acceleration area and increase in the possible deceleration area.
Therefore, it can enhance the system stability or even stabilize the system.
During the second and possibly during higher order of swings, the braking
resistor may also be connected to securing the transient stability.
From the techno-economic feasibility point of view, braking resistors are
only inserted during the acceleration periods i.e. during the increase of the
system frequency (as shown in Fig. 5.51(b)). The braking control in this case,
inserts the resistor when the frequency of the system increases above the
normal. Therefore, the resistors are provided by bang-bang control (i.e. A
control mode in which an element, such as a braking resistor is switched on
and off) at appropriate moments. By doing so, the cost of the resistor and its
heat dissipation capability are reduced. In addition, as shown in Fig. 5.49, it
will be well utilized as its impact during the deceleration periods if
effectively minor.
181
The resistors can be switched by the use of either mechanical circuit

breaker or electronic switches (Fig. 5.52). An example, of an electronic
switch is the back-to-back thyristors configuration where the thyristors in this
case are controlled by the integral-cycle control (ICC) algorithm. Unlike the
phase angle control (PAC), the ICC allows continuous full connection and
disconnection of the resistor. The PAC is suitable for devices such as
thyristor controlled reactors (TCR) where partial amounts of the controlled
inductance are desired, as a typical example, for voltage control function.
(a)
(b)
Fig. 5.51: Impact of braking resistors on the transient stability; (a) The
braking resistor operates during the acceleration and deceleration; (b) The
braking resistor only operates during the acceleration.
182
Fig. 5.52: Thyristor-switched braking resistor (TSBR)
A special situation where the braking resistor becomes ineffective in

enhancing the stability is when the fault occurs at the point of connection of
the resistor with the system. In this case, switching on the resistor will not
cause an increase in the electrical power during the fault.
F. Fault Current Limiters (FCLs) are devices that are capable of limiting
fault currents47,48,49. This is achieved by providing impedance against the
excessive increase of abnormal currents; for example, during faults. Resistive
FCL can be considered as a series switched barking resistor. Therefore,
resistive FCLs help to reduce the impact of faults on the transient stability of
generators by increasing the electrical power consumption during fault
conditions. Therefore, reduce the acceleration energy and enhance the
transient stability. For this purpose, the FCL must be resistive or containing
large resistance for consuming the active power during the during fault period
which causes the stabilization effect.
FCLs are also sometimes called Fault Current Controllers (FCC);
however, they not a real controller that acts all the time. FCLs only acts when
a fault occurs i.e. in the ‘during-fault’ period. In comparison with series
current limiting devices such as reactors or high-impedance transformers,
FCLs do not add significant impedance to the system during normal
47
Japan, I. E. E. (1999). State-of-the art and trends in fault current limiting technologies.
IEEJ Report, (709).
48
Youjie, Z. X. X. L. M., & Xiaoning, X. (2004). The Development and Application of
Fault Current Limiting Device [J]. Transactions of China Electrotechnical Society, 11,
000.
49
Noe, M., & Steurer, M. (2007). High-temperature superconductor fault current limiters:
concepts, applications, and development status. Superconductor Science and Technology,
20(3), R15.
183
conditions; the functional concept of FCLs is very similar to the surge

arresters (SA). Both of them engage only during abnormal conditions while
during normal operation, they have not a noticeable impact on the system
performance. Fast valving and braking resistors are also of this nature. In
comparison to FCLs, the reactors as fault limiting components cannot
provide direct stability enhancing by reducing the acceleration area as they
are not active power consumers. In addition, series reactors may lead to
voltage stability problems especially during transients.
From economical point of view, the cost of FCLs is highly compensated
by the use of protective devices with lower ratings in comparison with the
absence of FCLs. In addition, FCLs enhances the system stability and
reliability. They also can be placed at any desired location in a power
network at various voltage levels and for various system components. The
main functional requirements of FCLs include:
 Fully automated function.

 Invisible during the normal operation of power systems,
 Insert a large controllable and manageable impedance when a fault
occurs,
 Operate as soon as a fault occurs or at least within the first cycle of the
fault current,
 The recovery time should be as short as possible i.e. the FCL should
return to its normal operating conditions within a short interval in the
post-fault time,
 FCLs should be capable to re-engage in a short-duration for protecting
the system from successive faults, and
 The coordination of the system protection should be shifted by the
presence of FCLs.
From the technological point of view, there are two main categories of
FCLs; superconducting and solid state FLCs. Technological details of FCLs
are out of the scope of this book; however, the Yadav’s paper 50 entitled
‘Review on Fault Current Limiters’ is a good source about that issue. The
50
S. Yadav, G. K. Choudhary, and R. K. Mandal (2014).Review on Fault Current
Limiters.IJERT 3(4), 1595 - 1603.
184
common components of FCLs are (1) Fault detector; (2) FLC controller; (3)
Limiting impedance. A typical characteristic of FCLs is shown Fig. 5.53.
Typical values for the detection and starting times are 2 ms and 1 ms
respectively. As shown in the figure, the impact of the fault detection and
starting durations is a partial stress on the system by the fault current;
however, the magnitude and duration of this stress are small.
(a)
(b)
(c)
Fig. 5.53: Typical characteristics of FCLs; (a) Connection; (b)
Characteristics; (c) Impact on fault currents
The analysis of the inclusion of the FCLs and their impact on the
transient stability can be performed in the same manner as the braking
resistance; however, in this case the resistance is added in series with the
faulty line during the during-fault period. Consider the system of Fig. 5.54
185
which will be used to demonstrate the derivation of the during-fault

characteristics. The system is subjected to a fault at a random point within
one of the lines. The fault impedance is ZF i.e. a general unbalanced fault; the
value of ZF is zero for balanced three-phase faults. Both lines are equipped
with FCLs; however, the one on the faulty line is only considered in the
analysis as the other FCL is not subjected to fault currents and consequently
its impedance is nearly zero as shown in Fig. 5.53. The equivalent circuit of
the system can be reduced to a simple form by successive Y-
transformations are shown in Fig. 5.54. The resulting shunt impedances at the
infinite bus are omitted as they have not an effect on the generator’s power
flow.
(a) (b)
(c)
(d)
Fig. 5.54: SMIB with fault current limiter and its equivalent circuit
reductions
The generator current and the during-fault power can be then represented
by,
186
With the absence of the FCL and for a three-phase fault (i.e. ZF = 0), the
during-fault power is then . Therefore, the difference between
the two characteristics is the power consumed by the FCL. The main impact
of the FCLs on the stability is a reduction in the acceleration area. Therefore,
the stability is enhanced by the presence of FCLs.
As an example, the system of Fig. 5.55 is subjected to a momentary three-
phase fault at the generator’s terminals. In this special case, ZF = 0, Za = ,
and Zb = ZFCL + jX’d. Therefore, equation (5.62) is reduced to,
(a)
(b)
Fig. 5.55: Impact of FCL on the transient stability
Since in this case the during-fault power in a system without an FCL is
zero, then (5.63) represents the FCL power consumption. This power
187
consumption is also a constant in this special case; however, it is generally a

function of the power angle.
Due to the detection and starting periods, the action of the FCL is delayed
by a small amount of time as shown Fig. 5.53. Therefore, initially, the
during-fault power is zero; however, after this period, the during-fault power
is moved upward by the amount as a result of the FCL. It
should be noted that the equivalent power angle of the detection and starting
times (FCL) is dependent on the system inertia. Therefore, their effect on the
power-angle curves can be neglected if the system inertia is high. As shown
in Fig. 5.55, the acceleration area is significantly reduced. At the fault
clearing instant, the current goes back to low values. Therefore, the FCL
impedance is reduced to its normal near-zero value while the system
characteristic returns to the initial one. Consequently, the possible
deceleration area is not affected by the presence of the FCL while the overall
stability is enhanced.
Unlike the braking resistor, and fast valving, the FCL will not interact
with the system speed variations. This is because FCLs respond to a current
increase rather than frequency changes. Braking resistors and fast viewing,
on the other hand, operates in a bang-bang mode i.e. they will interact during
the entire transient process. Consequently, FCLs enhances only the first
swing stability while braking resistors and fast valving can possibly enhance
all the swings. The distinguishing feature of FCLs is that they can be placed
at any location in a power network for fault current limiting. Therefore, FCLs
may be found in the high-voltage bulk power grid as well as the low-voltage
light-current distribution systems.
In the next chapters, detailed analysis of the impact of FACTS devices on
the stability of power systems will be presented. In addition, stabilization of
power system through oscillation damping controllers integrated with
FACTS devices will be designed. In this section, highlights about the impact
of series and shunt regulated compensators on the system stability will be
presented.
G. Reduction of the transmission system reactance. The transmission

reactance has a significant impact on the maximum of a power curve.
Therefore, reduction of the transmission reactance can be used for increasing
188
the possible deceleration area and consequently enhancing the transient

stability. The reduction can be achieved by either improving the design of
transmission components or adding compensating components. Reducing
leakage reactances of transformers and reducing the reactances of
transmission lines are among the component design-based enhancement
methods. Installing series capacitor compensators can be an effective way for
the reduction of the transmission reactances. Generally, the main drawback of
reducing the transmission reactances is the resulting increase of the fault level
of the system. In addition, the costs associated for over-sizing power
transformers for reducing their leakage reactances is not usually economical
while the series capacitive compensations have a better relative economic
feasibility. The main critical issues associated with series capacitive
compensations are the possibility of resonance and sub-synchronous
resonance51 (SSR). These phenomena results in limiting the amount of a
series compensation that can be added to a transmission system. The
switching of compensating devices should be carefully adjusted or optimized
for enhancing the system stability52.
H. Regulated shunt compensators. Regulated shunt compensators (such

as SVCs and STATCOMs) are devices that are mainly designed for
controlling bus voltage magnitude; however, they are also results in an
increase in the maximum transmission power transfer. Therefore, they
conceptually have a positive impact on the transient stability; however, they
may reduce the stability of a system if their controllers are not well tuned.
I. Single-pole switching or independent-pole braking operations.

Independent-pole operation refers to the use of separate operating
51
Kumar, R., Harada, A., Merkle, M., & Miri, A. M. (2003, June). Investigation of the
influence of series compensation in AC transmission systems on bus connected parallel
generating units with respect to subsynchronous resonance (SSR). In Power Tech
Conference Proceedings, 2003 IEEE Bologna (Vol. 3, pp. 6-pp). IEEE.
52
Rahim, A. H. M. A., & Al-Sammak, A. I. J. (1991, January). Optimal switching of
dynamic braking resistor, reactor or capacitor for transient stability of power systems. In
Generation, Transmission and Distribution, IEE Proceedings C (Vol. 138, No. 1, pp. 89-
93). IET.
189
mechanisms for each phase of the circuit breaker 53,54. In such breakers, the
three poles are mechanically separate and have separate tripping mechanisms.
The outcomes of such mechanisms are the isolation of only faulty lines as
well as reduction of the impact of failures in three-phase circuit breakers.
Knowing that the majority of three-phase faults started as single line to
ground faults, the value of singe-pole switching in reducing, from stability
and economically, the impact of fault is high.
J. Load shedding. Shedding of pre-selected loads is a possible way for

balancing the load and generation. Although, this action has a large
economical impact due to the intentional disconnection of customers,
sometimes it is the possible feasible way for keeping the system stability and
synchronization55. Load shedding can also be used for balancing load and
generation is a forced isolated area which may be developed during the
extremes state of a power system (see chapter 1). There are three main
categories of load shedding schemes 56.
1. Reactive load shedding is a fully automated scheme that depends on
the value of the absolute frequency and look-up tables by which the amount
of load to be shed are determined.
2. Proactive load shedding schemes are also fully automated shedding
systems, but more sophisticated in comparison with the reactive schemes. In
this scheme, not only the absolute frequency is used as an indicator for load
shedding requirements, but also the rate of change of frequency or even the
acceleration of frequency. These inputs are used by these schemes for
initiating the load shedding process while the amount of load to be shed are
determined via look-up tables.
53
Balu, N. (1980). Fast turbine valving and independent pole tripping breaker applications
for plant stability. IEEE Transactions on Power Apparatus and Systems, 4(PAS-99),
1330-1342.
54
Taylor, C. W., Mittelstadt, W. A., Lee, T. N., Hardy, J. E., Glavitsch, H., Stranne, G., &
Hurley, J. D. (1986). Single-pole switching for stability and reliability. IEEE transactions
on power systems, 2(1), 25-36.
55
56
Grewal, G. S., Konowalec, J. W., & Hakim, M. (1998). Optimization of a load shedding
scheme. Industry Applications Magazine, IEEE, 4(4), 25-30.
190
3. Manual load shedding is the slowest scheme in terms of reaction time

in comparison with other schemes. The reaction time in this case is mainly
dependent on the operator response time. This scheme has application where
the operator has been informed of an impending or the likely loss of
generation. Consequently, the operators can counteract the potential
imbalance by shedding load.
K. Islanding. Forced islanding is typically follows an unsuccessful

attempt to restore system frequency by shedding loads. Selective islanding
can be performed manually for special severe situations (see chapter 1).
Intentional Islanding57, or controlled system separation, may be used to
prevent a severe disturbance in one part of an interconnected system from
propagating and causing a widespread system cascaded outages and breakup.
Once an impending instability condition is detected system separation is
initiated in a controlled manner by opening specific tie-lines. Considerations
when assessing the possibility of implementing an islanding scheme include
planning studies, stability studies, and fault-level studies. Planning studies
ensure that there is no power imbalance within the proposed island and
that steady-state stability is possible. Stability studies ensure that the
island will maintain transient and voltage stability. Fault-level studies ensure
that the new fault-levels in the islanded system are within acceptable limits.
L. Parallel HVDC and HVAC transmission. Unlike, the standard HVAC

transmission systems, HVDC systems provide control of the power flow over
the HVDC lines through their power electronic converters. In addition,
operating HVDC line in parallel with HVDC line can be used for direct
control of the power flow over the HVAC line 58. Therefore, they can be used
for enhancing the system stability. Since the traditional control of HVDC
lines does not provide either damping effect or synchronizing, supplementary
57
Pilo, F., Celli, G., & Mocci, S. (2004, April). Improvement of reliability in active
networks with intentional islanding. In Electric Utility Deregulation, Restructuring and
Power Technologies, 2004.(DRPT 2004). Proceedings of the 2004 IEEE International
Conference on (Vol. 2, pp. 474-479). IEEE.
58
Rahman, H., & Khan, B. H. (2008). Stability improvement of power system by
simultaneous AC–DC power transmission. Electric Power Systems Research, 78(4), 756-
764.
191
control is needed for controlling the converters in order to improve the

transient stability. For example, the power flow of the HVDC line can be
controlled according to the frequency deviation.
(a)
(b)
(c)
Fig. 5.56: Operation of parallel DC-AC transmission for transient stability
enhancement
As an illustrative example of the use of HVDC for enhancing the transient

stability, consider the system of Fig. 5.56 in which HVDC line is operating in
parallel redundant with HVAC line i.e. each of them can separately transfer
the pre-fault power (Po). Suppose that a three-phase fault occurs on the
HVAC line very close to the generating station bus. Therefore, the during-
fault power transfer is nearly zero. By its controls, the HVDC line increases
its power flow such that the total initial power transfer is restored in the post-
fault conditions (see Fig. 5.56). If the DC line is replaced by an AC line, then
this is action may be impossible due to the post-fault power transfer limit
192
which is limited by the reactance of the available line. Therefore, as shown in

Fig. 5.57, in comparison with an alternative HVAC line, the HVDC – with
the proper control - results in a significant increase in the possible
deceleration area while the original pre-fault operating point is restored.
Fig. 5.57: Impact of regulated HVDC line for transient stability control on
the transient stability in comparison with uncontrolled HVAC line
5.5 Measurement-based SMIB Equivalence of Multi-machine

power systems – A case study
The presented equivalency approach is applied to the IEEE two-area
system. As shown in Fig. 5.58, the system consists of two weakly
interconnected areas. All generators are identical. In addition, all
transformers are identical. The machine data are shown in Tables 5.5 while
the rest of the system data and operating conditions are available at my
paper59 entitled ‘Stability-based minimization of load shedding in weakly
interconnected systems for real-time applications’. Under the given loading
conditions, area 1 is exporting 400 MVA to area 2. The initial power flow
results are shown in Tables 5.6 while the line flow results are shown in Table
5.7.
59
193
Table 5.5: Machine data of the two-area system on 900 MVA base
Ra Xd Xq Xl Xd’ Xq’ Xd’’ Xq’’ Tdo’ Tqo’ Tdo’’ Tqo’’ H D
0.0025 1.8 1.7 0.2 0.3 0.55 0.25 0.25 8 0.4 0.03 0.05 6.5 0
Table 5.6: Initial power flow results

P
Bus V phase P gen Q gen load Q load
[p.u.] [rad] [p.u.] [p.u.] [p.u.] [p.u.]
Bus 01 1.03 0.575014 6.9 2.52699 0 0
Bus 02 1.01 0.403633 7.1 2.92232 0 0
Bus 03 1.03 0 6.34 1.92492 0 0
Bus 04 1.01 -0.14569 7 1.85999 0 0
Bus 05 0.995251 0.46203 -8.5E-14 -3E-15 0 0
Bus 06 0.968819 0.28216 -6.8E-13 8.86E-14 0 0
Bus 07 0.955458 0.130335 -1.6E-12 4.45E-13 0 0
Bus 08 0.935301 -0.13942 2.64E-13 9.96E-13 0 0
Bus 09 0.983239 -0.40137 4.4E-13 1.78E-13 0 0
Bus 10 0.986062 -0.26334 -4.1E-13 1.53E-14 0 0
Bus 11 1.004066 -0.10256 -4.8E-14 4.93E-15 0 0
Bus 12 0.956271 0.119751 1.03E-12 1.44E-14 9.67 -0.82891
Bus 13 0.98551 -0.4196 1.18E-12 1.11E-13 17.67 -2.3993
Table 5.7: Initial line flow results

From Bus To Bus Line P Flow Q Flow P Loss Q Loss
[p.u.] [p.u.] [p.u.] [p.u.]
Bus 05 Bus 06 1 6.9 1.674485 -8.9E-16 1.27240
Bus 08 Bus 09 2 2.165 -0.12241 4.44E-16 0.59127
Bus 06 Bus 07 3 14 2.359337 0 2.14749
Bus 07 Bus 12 4 9.67 -0.7259 -1.8E-15 0.10300
Bus 07 Bus 08 5 2.165 0.46887 0 0.59127
Bus 13 Bus 09 6 -17.67 2.399305 0 0.32740
Bus 11 Bus 10 7 6.34 1.233858 -8.9E-16 1.0345
Bus 10 Bus 09 8 13.34 1.200513 0 1.84504
Bus 08 Bus 09 9 2.165 -0.12241 4.44E-16 0.59127
Bus 07 Bus 08 10 2.165 0.46887 0 0.59127
Bus 01 Bus 05 11 6.9 2.526995 0 0.85251
Bus 02 Bus 06 12 7.1 2.922322 -2.7E-15 0.96506
Bus 04 Bus 10 13 7 1.859988 -8.9E-16 0.85881
Bus 03 Bus 11 14 6.34 1.924917 0 0.69106
194
The objectives of this case study are to estimate the electromechanical

single-machine equivalence of each area and to determine the equivalent
SMIB of the system.
(a)
(b)
Fig. 5.57: The IEEE two-area weakly interconnected system; (a) Single-
line diagram (SLG); (b) PSAT model
195
The equivalent inertia and transient reactances of the study system are
determined using the presented method. The disturbance is a temporary 3-
cycle three-phase fault started at t = 1 sec. The fault is applied on bus 8. The
PSAT is used as a simulation environment and the PSAT model is shown in
Fig. 5.57(b). The absolute angle (in radians) responses of the system is shown
in Fig. 5.58 while the speed of the generators are shown in Fig. 5.59. These
results reveal that the system is stable.
Fig. 5.58: The absolute angle responses (in radians) of the detailed system
Fig. 5.59: Rotor speed responses (in p.u) of the detailed system
The estimations of the equivalent machine of each area at the associated

interface bus are determined based on section 5.2. The outputs of the PMUs,
bus frequency, and area active power measurements are shown in Fig. 5.60.
On 100 MVA base, Table 5.8 shows the determined inertia values.
196
(a) (b)
(c) (d)
Fig. 5.60: Measurements at the interface buses
Table 5.8: Equivalent inertia (M) in sec, p.u transient reactances, and p.u
transient emf (E’)
Area 1 equivalence Area 2 equivalence SMIB equivalence
Inertia Reactance E’ Inertia Reactance E’ Inertia Reactance
29.11 0.0236 1.047 53.15 0.0253 1.062 18.81 0.0122
In the next chapter, the determined equivalents will be evaluated using the
EAC and TD. In addition, these equivalents will be used for demonstrating an
EAC-based fast method for load shedding minimization.
197
198
Chapter 6
Load Shedding and Online Applications
6.1 Introduction
Load shedding is a controllable reduction of a pre-determined amount
of the load power consumption according to specific shedding criteria. The
pre-determined amount of load to be shed is traditionally determined
according to an analysis of the dynamic security for a set of contingencies. In
this case, look up tables are prepared and the implementation of the load
shedding is performed according to them. A shedding look-up table is
dependent on the operating conditions of the systems, its available topology,
the available reserve, and the contingency. In addition, a look-up table must
be updated for changes in the system. A look-up table determined for a given
operating conditions and a specific set of single contingencies might be not
valid in the situation of multiple contingencies such as cascaded outage of
lines.60. The problems associated with look-up tables will be eliminated in
this chapter through the online dynamic security analysis and load shedding
estimation61.
The main objective of the load shedding is correcting an abnormal
system state (see chapter 1) to either the normal state or the alert state. For
example, the balance of the generation and the load during abnormal
operating conditions is a way for keeping the system stability. As explained
in chapter 5, the load shedding can be applied manually or automatically. In
addition, the automatic load shedding has many philosophies. For example,
the automatic reactive load shedding depends on the value of the absolute
frequency as a shedding criterion while with the proactive load shedding not
only the absolute frequency is used as a shedding criterion but also the rate of
change of frequency or even the acceleration of frequency.
60
Hafiz, H. M., & Wong, W. K. (2004, November). Static and dynamic under-frequency
load shedding: a comparison. In Power System Technology, 2004. PowerCon 2004. 2004
International Conference on (Vol. 1, pp. 941-945). IEEE.
61
199
Based on its definition, the load shedding is a corrective action by

which system overloads can be relieved. The system overloads can be
classified according to its root causes as,
1. Active power shortage: in this case the available power production

sources or generators are insufficient to meet the system demand. In
this case, the main symptom of the system overload is a drop in the
average system frequency. The shedding criterion in this case is the
value of the system frequency. In the frequency drops below a pre-
determined value (see chapter 4), the load shedding is activated. This is
called Under Frequency Load Shedding (UFLS).
2. Reactive power shortage: in this case the available capacity is capable
to meet the load active power demand while the voltage at specific
locations in the network is too low for appropriate power flow and load
requirements. The shedding criterion in this case is the violation of the
bus voltage magnitudes. For this purpose, it is called Under Voltage
Load Shedding (UVLS).
The UFLS and UVLS schemes can be considered as a special protection

or wide area protection systems that attempt to minimize the impact of
disturbances and prevent either brownouts or blackouts in power systems.
Both types of load shedding should have shedding algorithms. A load
shedding algorithm defines the method(s) by which several loads will be
switched-off (and switched-on again) automatically to keep the power
consumption below a defined secure level.
Based on their time-frames and magnitudes of the associated changes
in the operating conditions, system overloads may be also classified as,
1. Sudden and large changes: in this case a disturbance causes to the

system to rapidly move from the initial stable operation conditions and
the time needed to take appropriate corrective actions before a possible
system collapse is very small. In this case, fully automated, fast, and
appropriate corrective actions must be taken.
2. Slow and small changes: in this case a disturbance causes slow
changes with small amplitudes. Consequently, manual corrective
actions are possible.
200
Based on this classification, the appropriate load shedding scheme can be

determined.
The UVLS schemes are integrated into utility electrical systems to operate
as a last resort for the controllable shedding of specific amounts of loads at
specific locations in the grid. This action can prevent the loss of a large
amount of the load or the entire load due to uncontrolled cascading events. In
contrast, the UFLS is designed for use in either the emergency or extreme
states (see chapter 1). The main objective is to stabilize the balance between
the available generation and load before or after an electrical island has been
formed i.e. the UFLS drops enough load to allow the frequency to stabilize.
Therefore, the UFLS helps to prevent the complete blackout, and allows
faster system restoration in case of islanding. Typically, an UVLS responds
directly to voltage conditions in a local area. The goal of a UVLS scheme is
to shed load to restore reactive power relative to demand, to prevent voltage
collapse and to contain a voltage problem within a local area rather than
allowing it to spread in geography and magnitude. Load shedding is generally
applied in steps. In each step, a specific amount of the load is dropped. If the
first load-shedding step does not allow the system to rebalance, and voltage
continues to deteriorate, then the next block of the load is dropped. The
UFLS operates in the same manner.
The online determination of the load to be shed seems impossible if the
estimation of the shedding is based on the TD simulation of the full system
with detailed models. This is due to the available time limits to do the
dynamic security analysis, set decisions, take actions, and monitor the system
response to the taken actions. In this chapter an extremely fast method will be
presented for estimation of the minimum amount of load shedding for
securing the transient stability of power systems. Consequently, the method is
applicable for active power shortage problems. The method is based on the
on-line based electromechanical equivalency and the single machine
equivalence of power systems which are explained in chapter 5. The
equivalent SIMB is then analyzed and the dynamic security analysis is
performed using the standard EAC. The minimum amount of load shedding
as a corrective action is also determined based on the EAC.
201
6.2 Voltage Stability and UVLS – an overview

In this section the fundamentals of voltage stability and methods for
stability assessment will be presented. In addition, the UVLS schemes will be
explained.
6.2.1 Basic Definitions

Historically, the analysis of power system stability pertained to the
synchronous machine dynamics. On the other hand, in the analysis of
voltage stability, the major concern is the voltage collapse, although the
generators may remain in synchronism. As defined in chapter 4, Voltage
stability refers to the ability of a power system to maintain steady acceptable
voltages at all buses in the system after being subjected to a disturbance from
a given initial operating condition. In other words, the power system is
voltage stable if voltages after disturbances are close to voltages at normal
operating conditions. A power system becomes unstable when voltages
uncontrollably decrease due to outage of equipment, increment in load,
decrement in production or in voltage control. Voltage instability has been
given much attention by power system researchers and planners in recent
years, and is being regarded as one of the major sources of power
system insecurity. Based on the CIGRE definition, the voltage instability is
the absence of voltage stability, and results in progressive voltage collapse
(or increase). The IEEE defines the voltage collapse as the process by which
the sequence of events accompanying voltage instability leads to a
blackout or abnormal low voltages in a significant part of the power system.
Traditionally, voltage stability problems are associated with weak systems
and long lines; however, voltage problems are now also a source of concern
in highly developed networks as a result of heavier loading. Defining the
power transfer corridor (PTC) as a transmission system separating the two
portions of an interconnected power system (this type is called close PTC
interface), or a subset of transmission circuits exposed to a substantial
portion of the power exchange between two parts of an interconnected
system (this type is called open PTC interface). The circuits connecting buses
7 and 9 in the two-area system of Fig. 5.57 is an example of the closed PTC
interface. The power transfer limits of a PTC are limited by many constraints,
including the system rotor-angle stability, voltage stability, and thermal (i.e.
202
the N-1 static security transfer limit) limits. The impact of various
constraining limits on the transfer limits is highly dependent on the operating
conditions of the system where the system loading is the major governing
factor62. This is illustrated in Fig. 6.1. It is clear that at light and medium
loading conditions, the transient stability presents the main constraint. On the
other side, at high loading conditions, the voltage stability becomes the
dominant constraint. The secure operating limits are determined by the
minimum value of various constraints.
Fig. 6.1: Secure transfer limits and various constraints
Unlike the frequency stability, the voltage stability is generally a local

problem; however, the consequences of voltage instability may have a
widespread impact. The result of this impact is the voltage collapse, which
usually results from a sequence of contingencies rather than from one
particular disturbance. It leads to very low profiles of voltage in a major part
of a power system.
The voltage instability generally results in monotonically (or
aperiodically) decreasing voltages. Sometimes the voltage instability may
manifest as undamped (or negatively damped) voltage oscillations prior to
voltage collapse (see Appendix 3). A possible outcome of voltage instability
is loss of load in an area, or tripping of transmission lines and other elements
by their protective systems leading to cascading outages. Loss of
62
Uhlen, K., Pálsson, M., Time, T. R., Kirkeluten, Ø., & Gjerde, J. O. (2002, June).
Raising stability limits in the Nordic power transmission system. In 14th Power Systems
Computation Conference.
203
synchronism of some generators may result from these outages, or from

operating conditions that violate field current limit.
Several factors contribute to the voltage collapse such as increased
loading on transmission lines, reactive power constraints, on-load tap
changer (OLTC) dynamics and load characteristics. The problem of low
voltages in steady state conditions should not be confused with voltage
instability. It is possible that the voltage collapse may be precipitated even if
the initial operating voltages may be at acceptable levels. Voltage collapse
may be fast (due to induction motor loads or HVDC converter station) or
slow (due to on-load tap changers and generator excitation limiters).
The main reason for voltage instability is that the reactive power cannot
be transmitted over long distances and has to be delivered directly to the
point, which needs reactive-power support. The reasons for reducing reactive
power transfer include, but not limited to,
 The reactive power cannot be transmitted across large power angles,

even with substantial voltage magnitudes. High angles are due to long lines
and high power transfers. This is a physical constraint and can be simply
explained by considering the 2-bus system shown in Fig. 6.2. With the line
resistance neglected, the active and the reactive power transfer take the
forms,
Fig. 6.2: An infinite bus supplying a load center via a transmission line
204
where and is the maximum power

transfer between node 1 and node 2.
Fig. 6.3: Active and reactive power transfer characteristics
It can be seen from Fig. 6.2, that either increasing the active power
transfer or the angle separation between the buses, the reactive power transfer
is reduced.
 Active and reactive power losses constraints. Real losses should be

minimized for economic reasons; reactive losses should be minimized to
reduce investments in reactive power devices. Both active and reactive losses
depend on reactive power transfer. Therefore, minimizing the reactive power
transfer and keeping the voltage high is essential for loss minimization. This
is can be explained by the following equations that can be derived from the
system of Fig. 6.2.
205
 Transfer of large amounts of reactive power will result in over-sizing

the ampacity of the network components.
6.2.2 Classifications of voltage stability studies

The voltage stability is divided into short-term and long-term voltage
stability and both types are load-driven as shown in Table 6.1. The distinction
between long and short-term voltage stability is according to the time scale of
load component dynamics. Short term voltage stability is characterized by
components like induction motors, excitation of synchronous generators and
devices like high voltage direct current (HVDC) or static VAR compensators.
The time scale of short-term voltage stability is the same as rotor-angle
stability. The distinction between these two phenomena is sometimes
difficult, because voltage stability does not always occur in its pure form and
it goes hand-to-hand with rotor-angle stability; however, the distinction
between these two stabilities is necessary for understanding of the underlying
causes of the problem in order to develop appropriate designs and operating
procedures. The duration of long-term dynamics is up to several minutes.
Table 6.1: Classification of power system stability according to the main

driving forces, disturbance strength, and the time scale
Time scale Generator-driven Load-driven
Pure rotor angle stability problem Pure voltage stability problem

Rotor-angle stability Voltage stability
Short
Small-signal Large-signal Small-signal Large-signal
Voltage stability
Long Frequency Stability
Small-signal Large-signal
The voltage stability can be also classified according to the strength of the
disturbances (see Table 6.1). Consequently, two main classifications are
considered; small-disturbances and large-disturbances. Large-disturbance
voltage stability examines the response of power system to a large
disturbance, like faults, loss of load, or loss of generation. The ability to
control voltages following large disturbance is determined by the system load
characteristic and the interactions of both continuous and discrete controls
and protections. The small-disturbance voltage stability considers the power
206
system’s ability to control voltages after small disturbances like for instance
small changes in load. It is determined by load characteristics, and controls at
a given instant of time. The analysis of small-disturbance voltage stability
can be done in steady state conditions by static methods like for examples
load-flow programs; however, the voltage stability is a single problem on
which a combination of both linear and non-linear tools can be used.
6.2.3 Connection between Voltage Stability and Rotor Angle Stability

Voltage stability and rotor angle stability are interlinked. Transient
voltage stability is often interlinked with transient rotor-angle stability, and
slower forms of voltage stability are interlinked with small disturbance rotor-
angle stability. There are examples of pure voltage stability; for instance a
synchronous generator or large system connected by transmission line to
asynchronous load. Pure angle stability, for example, identified in case of a
remote synchronous generator connected by transmission lines to a large
system. However, rotor-angle stability, as well as voltage stability, is affected
by reactive power control. In particular, small disturbance instability
involving aperiodical increasing angles was a major problem before
continuously acting generator automatic voltage control regulators become
available. We now can see the connection between small-disturbance angle
stability and long-term voltage stability: generator current limiting prevents
normal automatic voltage regulation.
Voltage stability is concerned with load area and load characteristics.
Rotor angle stability is concerned with integrating remote power plants to a
large power system over a long transmission line. Voltage stability is
basically load stability, and rotor angle stability is basically generator
stability. For instance, if the voltage collapses at a point in a transmission
system remote from loads, it is an angle stability problem. If the voltage
collapses in a load area, it is probably mainly a voltage instability problem.
6.2.4 Fundamentals of Voltage stability analysis

Dynamic or transient voltage stability (TVS) includes non-linear modeling
of power system components while static voltage stability involves steady-
state representation of the system components. The load models and the TVS
analysis are illustrated in the previous chapters and Appendix 3. Here in the
static voltage stability modeling and analysis are explained.
207
For the simple 2-bus system shown in Fig. 6.2, the system’s voltage and
active power relation (V2-P2 characteristic) at the load bus can be determined
by writing equations (6.1) and (6.2) in a closed form by eliminating 12 and
using, , and (2 is the power factor
angle at bus 2) as well as considering that and .
Without considering the thermal limit and other physical limits, it can be
seen from (6.5) that the receiving end capability is a circle centered at (0, -B)
and with radius A. This is shown in Fig. 6.4. It can be seen that the maximum
range of possible reactive power flow occurs at zero active power flow. The
reactive power range is reduced with the increase in the active power flow.
The reactive power at the maximum active power flow is negative of a value
–B.
Fig. 6.4: Ideal capability limits at the load bus
The P2-V2 relation can be found by solving the quadratic equation (6.5);
208
It can be seen from (6.6) that a real solution exists if

. In addition, for a single value of the load bus voltage, two
power solutions are available. The maximum loadability of the system is
determined by the maximization of (6.6). The maximization conditions are
and . The corresponding voltage and power are called the
critical voltage (Vcr) and the power transfer limit ( ) from voltage
stability point of view respectively. These values are,
The function (6.6) is usually written for describing the relation between
V2 and P2. Therefore, its preferable alternative form is,
This equation shows that positive real solutions are only possible if
. The functions represented by (6.8) and
(6.8) represent the equation of a parabola and their general form is shown in
Fig. 6.5. Due to its shape, curve (b) is commonly called the nose curves.
209
(a) (b)
Fig. 6.5: Static voltage stability and practical limits; (a) P-V curve; (b) V-P
curve.
As shown in Fig. 6.5(b), for a single of the power consumption (Po), there
are two values of the bus voltage; Va and Vb. This situation is physically
impossible and one solution is stable (Vb) while the other is unstable (Va).
The stability is explained using Fig. 6.6 where the active power Po is
increased by P. With an increase in the active power, the voltage drop over
the transmission line (V = I212Z12) increases. Therefore, the new operating
point should be characterized by a lower voltage magnitude. As shown in the
figure, the increase in the active power, results in a decrease in the voltage
magnitude from b to b’ over the upper part of the V-P characteristics while
the lower part shows an increase in the bus voltage magnitude from a to a’. In
other words, the higher values of the voltage magnitude solutions are the
stable conditions. Consequently, the positive sign of the inner root of (6.9)
represents the stable solution. Therefore, the upper part is stable. The stability
criteria can be then represented by,
210
Fig. 6.6: Operating point static voltage stability
From voltage stability point of view, the critical voltage and the maximum
loadability limit are represented by (6.7) and (6.8); however, from load
requirements point of view, the load bus voltage should be kept within an
acceptable limits or bandwidth (see Fig. 6.5). Consequently, if bus 2 is
supplying a load, then the load voltage requirement should be respected. This
is required for proper operation of the load equipment and for preventing
their malfunction. Typically, depending on the load-voltage sensitivity (see
Appendix 3) the allowable voltage bandwidth is 3%, or 5% or 10%.
Generally, the normal state demanded voltage (VD) limits are represented by,
Usually, the critical voltage is much less than the allowable minimum load
voltage; however, in the cases where the critical voltage is less than the
minimum load voltage, the system operator restricts the bus voltage to the
critical voltage rather than the load minimum voltage. The main operational
objective is to simultaneously securing the voltage stability and the load
voltage requirements. Due to these limits, not only the allowable bus voltage
is modified but also the maximum loadability limit. For example, in Fig. 6.5
a significant reduction of the voltage stability based maximum loadability
limit due to the load voltage limits. The maximum loadability limit is reduced
from to .
211
Equation (6.9) can be arranged to show a very important feature of the

system. Considering that the short-circuit power at the grid access point (bus
2; also called point of common coupling (PCC)) is given by,
Consequently, (6.9) can be written as,
This relation shows that the V-P curve is universal and only dependent on
the short-circuit capacity of the system and the load point power factor angle.
The relation describes what is called the normalized V-P curve in which the
load voltage is represented in p.u of the grid voltage while the load active
power is represented in p.u of the grid short-circuit capacity. The V-P relation
is illustrated in Fig. 6.7 considering the effect of the power factor of the
receiving end.
Fig. 6.7: The normalized V-P relation as affected by the power factor at
the receiving end
212
The figure shows that the critical voltage locus is highly dependent on the
power factor; from lagging to leading power factors, the critical voltage is
increased. As previously explained, the operation of the system should secure
satisfactory voltage level at the load bus while keeping the system stable.
Therefore, the minimum of the critical voltage and the load minimum voltage
limit is selected as the operation voltage i.e.
As an example, at the operating point h, the critical voltage is higher that

the load minimum allowable voltage. Therefore,
.
Doing so ensures both the satisfaction of the load and the voltage stability.
At point d, . In the situation where the critical
voltage is very close or within the normal operation voltage, the monitoring
of power system security becomes more complicated (see chapter 1). These
situations can be realized when the system is over compensated by capacitive
compensators. With high capacitive compensations (leading power factor), it
is shown that the transfer limit is significantly increased but the system
becomes more vulnerable to voltage collapse as the critical voltage also
increases. An ideal situation is when the transfer limit is as high as possible
while the critical voltage is as low as possible. This point will be considered
later while discussing the voltage stability enactment and control options.
So far, the V-P relation model is in agreement of many physical facts;
however, the model as plotted in Fig. 6.7, did not show the effect of the
power factor on the voltage at zero active power. An open-circuited
transmission lines show receiving voltage value that is high in comparison
with the sending end voltage. This is called the Ferranti effect and it is
attributed to the effect of the line capacitance. Recalling, that the considered
line model is simply a series impedance with shunt elements neglected, the
resulting model will neither the Ferranti effect nor the accurate impact of the
load power factor on the V-P relation. It is worthy to be mentioned here that
the impact of the line resistance can be neglected in HV and EHV lines. This
213
is because the X/R ratios at these high voltage levels are high. On the other
hand, the X/R ratios in medium and low voltage circuits are not usually high
enough for the resistance to be neglected. More sophisticated models are
available for modeling these issues considering the distributed parameter
models of transmission lines 63. For this purpose the line of Fig. 6.2 is
represented by its long-line model parameters i.e. the characteristic (or surge
impedance (Zo), the electrical line length (), and the surge impedance
loading (PSIL) as shown in Fig. 6.8.
Fig. 6.8: Long line representation
The standard performance equations for a long AC transmission line is

written as,
The receiving end power (P2 + j Q2) is related to the receiving end current
(I2) by:
Eliminating I2 from (6.13) using (6.14),
Through mathematical treatment of equation (6.15) and considering the

phase angle of the voltage of bus 2 as a reference, the voltage solutions for
equation (6.15) can be derived to take the form,
63
El‐Shimy, M. (2014). Modeling and analysis of reactive power in grid‐connected
onshore and offshore DFIG‐based wind farms. Wind Energy, 17(2), 279-295.
214
where
Equation (6.16) gives the actual V-P characteristics of the system. It is

plotted for an OH-HVAC and UG-HVAC (or submarine cable) lines with
average parameters and the results are shown in Fig. 6.9.
(a)
(b)
Fig. 6.9: Exact V-P relations; (a) OH-HVAC; (b) UG-HVAC cable
It can be easily depicted from the exact relations that the impact of the
Ferranti effect on OH-HVAC lines is minor and can be neglected in a
simplified analysis. On the other hand UG-HVAC cables are characterized by
215
a strong Ferranti effect. This is can be explained using the -model shown in
Fig. 6.10.
Fig. 6.10: The -model for OH and UG transmission lines
This model is valid for both overhead (OH) and underground (UG)
transmission line; however; the major difference between the two
transmission types is the value of the shunt admittance or capacitance. In OH
lines, the shunt capacitance is very small in comparison with UG lines. This
is mainly due to the insulation type which is just air in OH installations while
it is a man-made insulating material in UG installations. The ratio of the UG
capacitance to the OH capacitance is nearly equal to the relative permittivity
of the cable’s insulating material. Based on the -model, it is clear that
the steady state normal reactive power production (i.e. the capacitive) from a
line is nearly constant as the bus voltages are nearly fixed close to 1.0 p.u. On
the other hand, the inductive power absorption is dependent on the line
loading. At no load conditions, the series impedance current is corresponding
to the current flow with the internal capacitances of the line and its value is
very small. With the increase of the load, the inductive reactive power
consumption increases. This is illustrated in Fig. 6.11 considering two
identical lines; one of them is OH while the other is UG cable. Both lines are
represented by the -model such that their terminal voltages are kept
constant. It is shown that the UG line is reactive power producers for the
entire load range. The OH line produces capacitive reactive power at light
loading conditions while they consumes reactive power during the high
loading conditions. The main cause of this effect is the shunt capacitances.
216
Fig. 6.11: Reactive power vs. loading level; (a) OH lines; (b) UG cables
The critical voltage shown in the previous analysis is corresponding to the

system characteristics as the system is supplying constant power loads. As
shown in Appendix 3, the loads have a verity of P-V and Q-V characteristics.
The previous analysis only considers the system characteristics while the
load-voltage sensitivity is not considered. The static representation of various
load models is presented in Appendix 3. The load-voltage sensitivity has a
significant impact on the operating point as well as the critical conditions.
This is can be illustrated using the system of Fig. 6.2 such that V1 = 1.0 p.u
and X12 = 0.06 p.u. The exponential load model of equation (4.128) is
considered. The initial voltage Vo is set to 1.0 p.u while the initial power Po is
set to have multiple values with a constant lagging power factor of 0.9. The
exponent  is set to be either 0.0 or 0.6 (note:  = 0.6 represents a composite
load model; see Appendix 3). The model of equation (6.12) is used to
simulate the system performance. For the considered load, the acceptable
voltage bandwidth is 10% around the unity voltage magnitude. The results
217
Fig. 6.12: Impact of load models on the voltage stability
It can be seen from the figure that with  = 0.6, the increase in the load
power (Po) causes a reduction in the operating voltage. Up to a load power of
2.7, the load bus voltage magnitude do not violate the acceptable limits while
higher load consumptions causes a violation of the lower limit of the load bus
voltage magnitude. It can be also depicted from the figure that with  = 0, the
maximum power transfer and the critical voltage are found to be 5.223 p.u
and 0.59 p.u respectively. Recalling that the critical voltage is defined as the
point at which the system and load characteristics are tangent to each other.
Therefore, it is shown that with  = 0.6, the critical voltage is reduced to 0.41
p.u and the maximum power transfer is also reduced to 4.7 p.u. Therefore, the
nose point of the V-P curve is the critical voltage stability point only when a
constant power load is considered while other load models will result in
different critical conditions. With  = 0.6, the reduction of the critical voltage
is beneficial as the system becomes more stable at low voltage values in
comparison with the constant power case; however, this benefit is
counteracted by the reduction in the maximum power transfer. Considering
the nose point as a critical operating point underestimates the system voltage
stability; however, there will be a significant stability margin from the critical
voltage point of view. On the other hand, the nose point overestimates the
218
maximum power transfer. Therefore, it is recommended to improve the

loading modeling as well as improve the voltage stability assessment for a
secure operation64.
The impact of contingencies or the N-1 security requirements on the
static voltage stability will be illustrated considering the system of Fig. 6.2
with the reactance X12 represented by its details; a transformer and two
parallel lines. This is shown in Fig. 6.13. As a numerical demonstration, the
transformer reactance is taken as 0.01 p.u while the reactance of each line is
0.1 p.u. The load at unity voltage is taken as 1.0 p.u. In addition, the load
power factor is fixed at 0.9 lagging. The load is modeled as a constant
impedance ( = 2).
(a)
(b)
Fig. 6.13: A system for demonstration of a line outage contingency; (a)
The initial topology (I); (b) the post-contingency topology (II)
The static voltage stability is dealing with slow changes of small

magnitudes. For this type of analysis, the transient associated with a
contingency are neglected. For example, in the considered system, the
64
Rassem OM, EL-Shimy M, Badr MAL. Assessment of Static Voltage Stability Limits
as Affected by Composite Load Models. ASJEE. 2008;2:201 - 9
219
electromechanical and electromagnetic transients associated with a line

outage contingency are neglected; the focus in the static analysis is on the
initial and final static voltage stability. Therefore, the during-fault conditions
will not be considered; however, these conditions are of major importance in
dynamic analysis studies such as the transient stability and the transient
voltage stability. In this illustration, a static outage of one of the lines is
considered. In comparison with the pre-fault reactance ( ), the post-fault
reactance ( ) is increased. As a result, the pre-fault V-P curve of the system
is changed to the post-fault V-P curve as shown in Fig. 6.14.
(a)
(b)
Fig. 6.14: Static voltage stability – contingency analysis using V-P curves
220
It is clear from Fig. 6.14 that the contingency causes a reduction of the
nose point from c to c’. From the critical voltage point of view, the nose point
can be considered – as explained earlier – as a conservative indicator of the
stability limit. As shown in the figure, the considered disturbance did not
alter the nose point voltage; however, the nose point power is significantly
reduced as a result of the increase of the transfer reactance X12. The
contingency also moves the operating point from a to b while if a constant
power load is considered, the load point will move from a to b’. It is clear
that the voltage associated with point b is high in comparison with point b’.
Therefore, the operating point is more secure with the constant impedance
model in comparison with the constant power model. As explained earlier,
the constant power model is a highly theoretical load while the constant
impedance load and the exponential load model prove their capability to
model actual load performance.
In this demonstration, if the load power is increased to 1.8 p.u and bus
voltage deviations should be less than or equal to 10%, then the simulation
of the system with a constant power load model result in a violation of the
minimum voltage limit and the system is considered insecure. On the other
hand, the constant impedance load model shows acceptable operation, but
very close to the lower limit (i.e. an alert condition). If the load is increased
to 8.0 p.u, the stability will be highly degraded as the operating point will be
at the post-fault nose point. Higher power consumptions will cause static
voltage instability.
The popular method for the analysis and the assessment of the static
voltage stability is usually through the V-P curves; however, the Q-V curves
can also be used for this purpose. In addition, they also can provide some
additional insight of the static voltage stability. A Q-V curve expresses the
relationship between the reactive support Qc at a given bus and the voltage
on that bus. As shown in Fig. 6.15, the Q-V curves can be determined by
connecting a fictitious generator with zero active power at the load bus, then
the reactive power of the compensator is recorded as the terminal voltage is
being varied. Since it does not produce the active power, the fictitious
generator can be treated as a synchronous condenser or a static VAR system
(SVS). SVSs include SVCs, and STATCOMs.
221
Fig. 6.15: Use of a fictitious generator to produce Q-V curves
As previously explained, the V-P curves define the system characteristics

while the load characteristics are merged with the system characteristics for
finding the operating point. Unlike the V-P curves, the Q-V curves show the
static characteristics of the combined system and load.
The Q-V relation can be determined in a similar way as that presented in

section 6.2.D but equations (6.1) and (6.2) should be written as,
Therefore,
Which can be written as,
222
Equations (6.20) and (6.21) describe the Q-V curve considering a constant
power load model. Real solution is only possible if from

which the maximum demand can be found.
The load can be represented, for example, by the exponential or the ZIP
models as descried (4.128) and (4.129). As a numerical demonstration, the
parameters of the system of Fig. 6.13(a) are considered. The Q-V curve for a
load active power of 3.0 p.u while the load is operating at a constant lagging
power factor of 0.9 is shown in Fig. 6.16.
Fig. 6.16: The Q-V curve of the demonstration system for 3.0 p.u active
load @ 0.9 lagging power factor
The figure shows the plots of the positive and negative roots of equation
(6.20). The values of the positive roots do not provide meaningful operating
conditions. The negative root allocates operating equilibrium for a single
value of the injected reactive power from the hypothetical compensator. Of
course, this situation is impossible and one operating point is stable while the
other one is unstable. For a given value of Qc, the possible operating
equilibrium is a and b. If the compensator capacitive power increased these
operating points move such that a  a’ and b  b’. From the physical point
of view, the increase in the capacitive power injection at the load bus should
increase the voltage magnitude. As seen, the voltage at a’ is less than the
223
voltage at a while the voltage at b’ is higher than the voltage at b. Therefore,

the right-hand side of the negative root is stable while the other side is
unstable. The operating point d presents the critical point. The stability
considering the negative root of (6.20) is then defined as,
The value dQc/dV2 is also called the voltage sensitivity factor (VSF).
The VSF measures the change in voltage magnitude at a given node as a
consequence of a reactive power injection in that node. Since we are dealing
with a fictitious reactive power compensator, then the operating equilibrium
of the actual system is either at point c or c’. Based on (6.10), the stable
equilibrium point is c. It is then clear that the distance from point d (i.e. the
critical point) and the horizontal axis presents the margin to the loss of a
stable operating point. Therefore, this distance is called the reactive power
margin (KQ). With the load active power changed, a family of Q-V curves is
obtained as shown in Fig. 6.17.
Fig. 6.17: Impact of increase in the load consumption on the Q-V curve
and the static voltage stability; the compensator sizing is also shown
224
It is clear from the figure that the increase of the load results in a reduction
in the load bus voltage magnitude (the operating point moves from a to b to c
as the load increase from 1.0 to 3.0 to 5.3 p.u). It is also shown that at a
power demand of 5.3 p.u, one operating point is available which is point c.
This point is called the critical point at which the reactive power margin is
zero. Increasing the consumption above the critical value, no possible
operating equilibrium can be achieved with zero compensation. This is
shown in the fourth loading level (8.0 p.u). In this case, a capacitive
compensation is essential for system operation. From voltage stability point
of view, the minimum reactive power distance that can bring the system at
least to a critical condition; see Fig. 6.17; however, this minimum value
result in an insecure system operation and a higher value should be selected
for achieving a reasonable reactive power margin and acceptable operating
voltage magnitude. From the load satisfaction point of view, points b and c
violate the lower voltage limit. Therefore, a capacitive compensator may be
used to increase the bus voltage magnitude to an acceptable value.
Considering the N-1 criteria or a contingency of a static outage of one
of the lines in the system of Fig. 6.13, the Q-V curves for active power
consumptions of 1.0 p.u and 3.0 p.u are illustrated in Fig. 6.18.
Fig. 6.18: Static voltage stability – contingency analysis using Q-V curves
225
The results show that the lightly loaded operating conditions results in a
secure system operation in the pre-fault and post-fault conditions. On the
other hand, the heavily loaded system could not find an operating point in the
post fault conditions. In addition, the initial operating conditions lightly
violates the lower limit of the bus voltage magnitude.
Practically, the impact of various contingencies on the voltage stability
or voltage security can be performed considering the long-term (static) or
short-term (transient) voltage stability. The main purpose of these studies is
the determination of the voltage security level of power systems. The
standard contingency analysis methods (chapter 1) are used for this purpose.
In addition, the data required for the simulation of various contingencies can
be based on offline datasets or online measurements. The online
measurements are currently possible due to the availability of WAMs and
PMUs (chapter 1). Fig. 6.19 shows a schematic diagram of voltage security
assessment algorithms.
Fig. 6.19: Voltage security assement algorithms
If the system could not provide an acceptable operation (see Fig. 6.17
and 6.18), then the voltage control is required for that purpose. Based on Fig.
226
6.20, the relation between the voltages V1 and V2 as a function of the voltage
drop and voltage deviation takes the form,
(a)
(b)
Fig. 6.20: Basics of voltage control; (a) Phasor diagram; (b) Control
methods
where
227
If the HV or the EHV transmission is considered, then the X/R ratio is

high enough for the resistance to be neglected. Therefore, the voltage drop is
mainly dependent on the reactive power flow over the line while the voltage
deviation is mainly dependent on the active power flow.
It can be seen from (6.24) that five variable affects the voltage drop;
however, the impact of these variables on the voltage controls is not equal.
For example, the control of the active power flow has a minor impact on the
voltage drop as it is weighted by the line resistance which is typically low. In
addition, improvement in the resistance is commonly attributed to its
reduction. Therefore, the development in the wired transmission technologies
is expected to provide a further reduction of the impact of the active power
control on the voltage drop. The active power supplied from the system (P2)
to the load can be altered, for example, by injecting active power directly to
the load via a conventional diesel generator or a solar-PV generator. In this
case, where PDG is the active power produced by a distributed
generator connected at the load bus.
On the other hand, controlling the reactive power has a significant impact
on the voltage drop. This is because the high weight of the reactive power
caused by the line reactance. The reactive power supplied from the system to
the load (Q2) can be controlled via direct injection or absorption of reactive
power at the load bus. This is can be provided by passive or active
compensators (Fig. 6.21). Passive compensators includes fixed capacitors
(FC), and fixed reactors (FR) while active compensators include synchronous
condensers, thyristor controlled reactors (TCR), thyristor switched capacitors
(TSC), static var compensators (SVC), and static synchronous compensators
(STATCOM). In this case, where Qc is the reactive power of a
shunt connected compensator at the load bus.
Fig. 6.21: Shunt compensation
228
The reactance itself can also be controlled by the use of either passive or
active series compensators (Fig. 6.22). Fixed capacitors and thyrsitor
controlled series compensators (TCSC) are commonly used for that purpose.
In this case, where Xc is the reactance of the series
compensator. Series compensators are usually capacitive for reducing the
voltage drop and increasing the power transfer limit. Therefore, Xc is usually
negative. Consequently, the use of series compensators is accompanied with
dangers from resonance and sun-synchronous resonance (SSR). A situation
that can be avoided by proper selection of the possible range of capacitive
series compensations.
Fig. 6.22: Series compensation
The use of under-load tap-changing (ULTC) transformers also provide a

tool for a direct control of the load bus voltage magnitude. ULTC
transformers can be controlled manually or automatically. Based on the
expected load curve and load flow, timetables are usually constructed for the
manual control of ULTC transformers. On the other hand, automatic ULTC
transformers provide a semi-immediate correction of the bus voltage
magnitude; however, their effectiveness is limited by the small range of tap
changers. In addition, automatic ULTC transformers are usually subjected to
frequent failures due to excessive operation of the tap-changers. These
excessive operations may also be unnecessary in the situation of momentary
voltage variations (such as voltage dip or swell) where the voltage returns to
the acceptable range in a short-duration which is small in comparison with
the time constant of the tap-changing mechanism. In addition, devices such as
uninterruptable power supplies (UPS) and dynamic voltage restorers (DVR)
provide a more effective way for securing sensitive loads from these voltage
disturbances. Practically, as shown in Fig. 6.23, the control systems of ULTC
transformers include a deadband and a time delay blocks. The deadband
defines the acceptable limits of the voltage magnitude of the controlled bus.
229
These limits are defined according to the load sensitivity. The time delay
block serves mainly as a protection device for preventing the tap-changer
from responding to short-duration voltage variations.
Fig. 6.23: Automatic control of ULTC transformers
The Q-V and V-P curves can serve as a vital tool in the analysis of the
impact of various voltage control alternatives as well as sizing them 65.
6.2.5 UVLS schemes

As a result of a disturbance, the voltage magnitude may drop to a pre-
selected level for a pre-determined time. In such cases, the UVLS sheds a
selected amount of loads at selected locations in the system. The objective is
then to prevent the widespread of voltage problems or voltage collapse.
System planning engineers perform numerous studies (see Fig. 6.19) using
the V-P curves as well as other analytical methods for the proper
determination of the minimum amount of load that must be shed for securing
the system to retain voltage stability under credible contingencies.
Various voltage control equipments provide a defense against voltage
collapse; however, in the situations the system is subjected to severe
disturbances or cascaded outages, various voltage control and voltage
restoration equipment may fail to restore the normal state voltage range. In
these situations, the load shedding provides an effective corrective action for
65
Van Cutsem, T., & Vournas, C. (1998). Voltage stability of electric power systems (Vol.
441). Springer Science & Business Media.
230
preventing either the voltage collapse or the islanding of the system. This is
illustrated in Fig. 6.24. The shown voltage problems are attributed to the
insufficiency of the reactive power sources needed for restoring an acceptable
voltage level. It should be noted that system operators usually shed load as
the last resort. Under Voltage Load Shedding (UVLS) schemes, drops a load
when the voltage gets too low. The dropping of load will alleviate the system
by eliminating the current flowing to the dropped load. UVLS usually
triggers distribution feeders to open when voltage of the bulk electric system
is around 90%. Definite time relays usually act when all three phases show
low voltage for around 10 seconds at 90% voltage magnitude, this would be
after some of the ULTC transformers have been acted. Certain critical
customers cannot be dropped from load despite the help it may present to the
system. Critical customers include hospitals or customers that would loose
lots of revenue from being dropped.
Fig. 6.24: Long-term (slow dynamic) voltage security and corrective

actions
The order of various corrective events following a contingency is shown

in the figure. The events started as a response to a contingency such as a
forced outage of a generator or a line. Erroneous operation or vandalism may
also cause disturbances of the similar impact.
231
There are two main UVLS schemes; the decentralized (also called
distributed UVLS) and the centralized schemes. With a decentralized scheme,
protection relays are installed at the loads that are candidate to be shed upon
severe voltage problems. As voltage conditions at these locations begin to
collapse, load assigned to that relay is automatically shed. On the other hand,
a centralized scheme has undervoltage relays installed at key system buses
within system areas. The trip information is transmitted to shed loads at
various locations. Both schemes require high-speed and reliable
communication for proper operation. Fig. 6.25 illustrates a typical
distribution substation with integrated UVLS and UFLS special protections.
Fig. 6.25 A typical distribution substation with integrated UVLS and

UFLS special protections
In this system, the UV relay is installed on the HV side of the transformer

for correct detection of the grid voltage. This is because the voltage on the
secondary side is not a real indicative of the grid voltage due to the actions of
the ULTC transformer (or any other load side voltage controllers). The UF
relay is installed on the LV side of the transformer because the transformer
does not affect the frequency while the LVPT is more economical in
comparison with the HVPT. Fig. 6.26 shows examples of the setting of the
UVLS considering a contingency. The results of Fig. 6.14 and 6.18 are used
for this purpose. The operating margin is left for allowing the natural
232
reduction of the load due to the load-voltage sensitivity. The main

complication associated with the setting UVLS is the inaccuracies associated
with the PT and the UV relay. Therefore, proper and secure setting should be
chosen for considering the probable inaccuracies.
(a)
(b)
Fig. 6.26 Contingency-based setting of the UVLS protection
233
6.3 UFLS – AN OVERVIEW

The frequency response and stability are briefly considered in the previous
chapters. In this section an overview of the UFLS schemes and technologies
will be presented. Generally, the system frequency is a good indicator of the
power balance and overload conditions in a power system. The UFLS is the
last resort for the treatment of serious frequency declines in power systems
when subjected to large disturbances. Under the emergency state or the
extreme state, the ability to maintain the power balance and stabilize the
frequency is directly related to the effectiveness of the employed UFLS
Strategy. An effective UFLS strategy should be capable of:
(1) Restrain the frequency decline,

(2) Restore the normal frequency,
(3) Minimization of the load shedding,
(4) Minimization of the frequency recovery time,
(5) Minimization of the frequency fluctuations, and
(6) Provide the desired protection functions as economical as possible.
Typically, an UFLS scheme sheds the loads in several stages. In each

stage a pre-defined amount of the load is disconnected and the shedding of
the load is continued till the normal frequency (i.e. the power balance) is
restored. This is illustrated in Fig. 6.27. In 50 Hz systems, the common
practices by most utilities use 49.3 Hz (i.e. 1.4% drop in the frequency) as the
first frequency step, and between 48.5 and 48.9 Hz for the last step 66. In
proper dynamic control of power systems, a sufficient time delay must be left
between the shedding of each load block. This is of major importance for
monitoring the correct impact of disconnecting a load block on the system
frequency and for avoiding excessive, unnecessary load shedding as well as
for avoiding subjecting the system to over-frequency conditions due to the
over - shedding of loads. The proper timing of shedding each load block
should not only depend on the frequency but also the rate of change of the
frequency. This results in an adequate time separation between the shedding
66
Hafiz, H. M., & Wong, W. K. (2004, November). Static and dynamic under-frequency
load shedding: a comparison. In Power System Technology, 2004. PowerCon 2004. 2004
International Conference on (Vol. 1, pp. 941-945). IEEE.
234
of blocks. Using small load shedding blocks in conjunction with shedding

timing based on the rate of change of frequency can be an effective way for
prevention of over-shedding.
The blocks of load shedding shown in Fig. 6.27 can be selected based on
two criteria; the static criterion and the dynamic criterion. Fig. 6.28 shows
flowcharts describing the logic of each criterion. In the static criterion, fixed
load blocks are disconnected in each load shedding stage. This criterion may
reduce the impact and the effectiveness of the load shedding, especially in
large disturbance conditions that are associated with a steep decline in the
frequency. The dynamic load shedding is constructed for solving this
problem. In the dynamic load shedding the amount of load to be disconnected
at each shedding stage is dynamically selected based on the system
frequency, the rate of change in the frequency, the voltage, and the severity
of the disturbance(s). In other words, the amount of a load shedding block is
a function of the magnitude of the power imbalance.
Fig. 6.27 UFLS conceptual operation – blocks of load shedding in k-stages

scheme
235
There are three main methods for the implementation of UFLS strategies.
These methods are
(a) (b)
Fig. 6.28: UFLS criteria; (a) Static (semi-adaptive) UFLS; (b) Dynamic
(adaptive) UFLS
1. The traditional method; when the frequency is lower than the first
setting value, the first level of load shedding will be implemented. If
the frequency continues to decline, it is clear that the first load shed
amount is insufficient. When the frequency is lower than the second
setting value, the second stage of load shedding is then
implemented. If the frequency continues to decline, the further load
shed stages are activated until the normal frequency value is restored.
The traditional method follows the static shedding criteria and the
amount of load shedding per each shedding stage is determined based
on the analysis of the worst possible expected events. Therefore, for
236
less severe events, the first stage of shedding may result in an over-
shedding and may also cause over-frequency problems
2. The semi-adaptive method; to some extent, this method is similar to
the traditional method; however, the specific amount of load to be shed
is determined in terms of the measuring value of the rate of change of
frequency.
3. The self-adaptive method follows the dynamic shedding criterion for
more accurate estimation of the proper amounts of the load to be shed
in each stage and the timing of each stage.
6.4 Dynamic load shedding based on online measurements

As shown in Fig. 6.28, the major challenge in the practical
implementation of the dynamic load shedding method is the proper
estimation of the load to be shed in each stage. The load shedding timing can
be easily determined based on the measured df/dt while the amount of the
load to be shed requires deep analysis of electromechanical transient
associated with the disconnection of a specific amount of load. This is mainly
for avoiding over-shedding, which may lead to over-frequency problems. In
addition, from economy, reliability, and power quality points of view, the
load to be shed should be minimized. The minimum value of the load to be
shed can be defined as the minimum amount of disconnected loads for
restoring the system transient stability (or dynamic security). On the other
hand, the estimation of the minimum load shedding using detailed
electromechanical modeling of power systems requires huge time in
comparison with the under frequency phenomena. In addition, the amounts of
simulation results to be analyzed are very huge for reaching appropriate
conclusions and decisions in a reasonable time.
In this section, the value of the presented online-based SMIB equivalency
(chapter 5) in the analysis and decision making considering the under-
frequency fast phenomena will be demonstrated and evaluated. The two-area
system of Fig. 5.57 will be used to demonstrate the use of the SMIB
equivalence (Table 5.6) in the estimation of the online load shedding
minimization67. The system consists of two weakly interconnected areas or
67
237
systems. It is assumed that the generators in area 2 are highly variable while
area 1 comprises less variable power sources. In both areas sudden drop in
the power generation may be attributed to forced outage of generators, or
faults, or unavailability of the primary energy resources. In area 2, the
variability of the power sources causes sudden changes in the output power
of the generators. These sudden changes may cause emergency or extreme
security and transient stability problems if the drop in the power sources is
intense and rapid. In such situations, load shedding is implemented to ensure
system stability by curtailing sufficient system loads for matching the
available generation with the remaining loads and keeping the system
stability.
Fig. 6.29: Equivalent representation of a sudden change in Pg2 for the

system of Fig. 5.57 using its SMIB equivalence shown in Fig. 5.7
As early state, the load shedding is implemented as a last resort to

protect the system against the dangerous decline of either the frequency or
the voltage or both of them. In this analysis, the load shedding is
implemented to protect the system interconnection against frequency declines
that cause system transient instability. Due to the delay in the propagation of
frequency changes in weakly interconnected systems, there is a tendency to
localize the power adjustments following large contingencies. In addition, the
localized power adjustments have a significant improvement on the system
stability. Therefore, drop of generation in a specific area will be primarily
238
compensated by load shedding within that area unless more necessary actions
should be taken.
With the SMIB equivalence of the considered interconnected system (Fig.
5.8 and Table 5.6), the maximum sudden drop in the power generation Pg2
can be determined using the EAC. Consider that the system shown in Fig.
5.57 operates at initial conditions Pg1o, Pg2o, Pd1o, and Pd2o. As shown in Fig.
6.29, this is equivalent to operation at Peqo and o; refer to section 5.3 for
associated the mathematical modeling. The value of Peqo equals to 4.330 p.u.
This value is not dependent on the nature of the power sources. The value of
1 is 64.5o.
In the following the symbol 12 will be written as . Due to the variability
of generation in area 2, it is assumed that a sudden drop in the generation
(Pg2 < 0) occurs. Consequently, Peq is suddenly increased by –M1Pg2/(M1 +
M2). The corresponding acceleration area (Aacc) is the area (abca), the
deceleration area (Adec) is the area (cdec), and possible deceleration area
(Apossdec) is the area (cdfec) are shown in Fig. 6.29. On Fig. 6.29, o, 1, max,
cr indicate the initial steady state operating angle, the final steady state
operating angle, the maximum angle of oscillations, and the critical transient
angle respectively. It is clear from the figure that cr and 1 are dependent
variables such that cr =  - 1. The stability of the system is governed by the
ratio of the possible deceleration area and the acceleration area (refer to
section 5.4). The system is stable if this ratio is higher than one while it is
unstable if the ratio is less than one. A unity ratio indicates critical stability
conditions. The maximum or the permissible sudden drop in Pg2 can be then
determined by determining the value of Peq1 at which the Aacc equals to
Apossdec. In this case max = cr. Therefore,
The integration in eq. (6.26) results in the following equation which is to

be solved for 1.
239
In eq. (6.27), cr =  - 1, o = sin-1(Peqo/Pmax). The maximum permissible

increase in Peq and the maximum generation drop in Pg2 can be then found by
equals to 6.32 p.u. Consequently, the is found to be

approximately -5.6 p.u. It is clear from (6.28) and (6.29) that the permissible
changes are highly dependent on the initial operating conditions of the
system. As shown in Fig. 6.30, a generation drops higher than results
in system instability. This is can be mitigated by fast load shedding in the
form of immediate reduction in Pd2.
Fig. 6.30: Transient stability regain by load shedding following an

emergency sudden drop in Pg2.
Recalling that the impact of sudden reduction in Pd2 is a reduction in Peq
by M1Pd2/(M1 + M2). Therefore, if the generation drop ,
then the minimum load shedding as shown in Fig. 6.30 is defined by
240
Equation (6.30) for the considered parameters and variables is then,
It is known that the decrease in the mechanical driving power has the
same impact on the rotor angle swing as that of increasing the electrical
output power. In this analysis, eq. (6.30) depicts that the change electric
power demand shows that the same impact as the changes in the mechanical
power. The verification of the estimated maximum drop in power generation
and minimum load shedding requirements is performed through time domain
simulation of the two-machine equivalent system (shown in Fig. 5.6). The
perturbation cases shown in Table 6.2 are considered.
Table 6.2: The considered disturbances

Case Perturbation
1 5.0 p.u drop in the generation of area 2
4 6.0 p.u drop in the generation of area 2 &
0.4 p.u simultaneous load shedding
Case 1 represents a drop on the generation below the critical value (5.6
p.u). Case 2 represents the critical drop in the generation in area 2 while case
3 present situation where the drop of generation is higher than the critical
values. Case 4 presents mitigation by the minimum amount of load shedding.
The dynamic performances of the system are shown Fig. 6.31. These figures
show the rotor angle swing of area 2 relative to area 1. The time-domain
simulations are performed using the PSAT. The results ensure the accuracy
of the presented method in predicting the system transient response and
minimization of the load shedding requirements.
241
(a)
(b)
Fig. 6.31: Dynamic performance associated with sudden changes in the
generation and load; (a) Cases 1, 2, and 4; (b) Case 3.
From a practical point of view, the simultaneous drop in the power

generation and the required load shedding is technically impossible. In
addition, introducing a time delay in the load shedding process is useful for
avoiding over-shedding of loads (Fig. 6.28). Therefore, the impact of delayed
load shedding on the system stability and the validity of the presented results
is investigated through the EAC and time domain dynamic simulation.
In comparison with Fig. 6.30, the simultaneous load shedding with the
values presented by eqs. (6.30 and 6.31) will not cause changes in the
acceleration area (area abca) shown in Fig. 6.32(a). Therefore, the system
will be stable after the perturbation in the power generation and load. On the
contrary, the delayed load shedding, as shown in Fig. 6.32(b), causes an
increase in the acceleration area by the amount bfghb. Therefore, the
minimum amounts of load shedding determined by eq. (6.30) will not
effectively restore the system stability. Therefore, the (eq. (6.28) and
242
Fig. 6.32) must be reduced to (shown in Fig. 6.32(b)); is

reduced by an amount that increase the possible deceleration area by an
amount jcekj such that the increase in the acceleration area is precisely
compensated by the increase in the possible deceleration area i.e. area jcekj =
area bfghb.
(a) (b)
(c) (d)
Fig. 6.32: Impact of load shedding delay; (a) Simultaneous load shedding
(shed = o); (b) Delayed load shedding (shed > o): energy balance; (c)
Delayed load shedding: shed < h; (d) Delayed load shedding: shed  h
In Fig. 6.32, shed denotes the angle at which the load shedding takes place
while h is the angle defined by where Peq1 is the
equivalent power due to the reduction in Pg2. The load shedding delay time is
nonlinearly related to the difference between the load shedding angle (shed)
and the initial angle (o). Fig. 6.32(c) and 6.32(d) show two situations defined
243
according to the relation between shed and h. These situations are shed < h
(Fig. 6.32(c)) and shed  h (Fig. 6.32(d)). The acceleration and possible
deceleration areas are clearly illustrated in the figures. It can be easily shown
that the angle 1 can be determined for both situations by solving eq. (6.32)
for 1. Equation (6.32) is determined by equalizing the acceleration and
deceleration areas in each situation.
It is clear that eq. (6.32) will be reduced to eq. (6.27) if simultaneous load
shedding (i.e. shed = o) is applied. Based on the value of 1, (shown in
Fig. 6.32) can be determined using . Consequently the
minimum reduction in peak (denoted by ) is equal to .
As a result, the minimum load shedding ( ) can be determined using
. It is depicted from eq. (6.32) that the value of 1 is
dependent on the initial operating conditions of the system represented by o
and the shedding delay which is represented by (shed - o).
Fig. 6.33: Impact of load shedding delay on the minimum required load
shedding – EAC approach
Considering Pg2 of -6.0 p.u, Fig. 6.33 illustrates the impact of delayed
load shedding on the minimum load shedding requirements. The results show
that with the increase in the shedding delay, the minimum load shedding
requirements are increased.
The relation between the minimum load shedding and the shedding time
delay (rather than (shed - o)) can be determined by solving the nonlinear
244
mathematical model representing the system or performing numerical

integration considering the EAC-based classical model. This is can be easily
adopted in contingency analysis/corrective action schemes in the system
security center. The relation between the minimum load shedding and the
shedding time delay is determined as shown in Fig. 6.34.
Fig. 6.34: Impact of shedding delay time on the minimum amount of load
shedding – TD analysis
The results shown in this figure are determined through TD dynamic

simulation of the two-machine equivalence of the system considering a drop
of 6.0 p.u in the power generation of area 2. Recalling that the maximum
drop of generation in this area is found to be 5.6 p.u. Therefore, as
determined, at least 0.4 p.u of area 2 load must be immediately disconnected.
With the delayed load shedding, the results shown in Fig. 6.34 show that the
minimum value of the load shedding must be increased for restoring the
system stability. This is in agreement with the EAC based analysis shown in
Fig. 6.32 and 6.33; however, with the time domain analysis, the shedding
delay is represented in time rather than angles. In addition, the application
and the accuracy of the EAC is limited to the initial stages of the dynamic
processes; however, the EAC shows significant potential and accuracy in the
fast estimation of the transient stability of power systems.
The time domain responses of the two-machine equivalence considering
zero cycles (case 5) and eight cycles (case 6) load shedding delay times are
245
shown in Fig. 6.35. With the delayed shedding, the minimum load shedding
is increased from 0.4 p.u (no delay) to 0.48 p.u. The transient associated with
the load shedding action is clearly shown on the figure.
Fig. 6.35: Dynamic response of properly reduced loads with and without
time delay.
246
Chapter 7
Damping Improvement and Stabilization through
FACTS-POD controllers
7.1 Introduction
The power system stability has been recognized as an important problem
for secure system operation since the early power systems. The importance of
this phenomenon has emerged due to the fact that many major blackouts in
recent years caused by power system instability. Therefore, there is a great
need to improve electric power utilization while still maintaining reliability
and security. As power systems have evolved through continuing growth in
the interconnections and the increased operation in highly stressed
conditions, different forms of power system instability have emerged (see
chapters 4 and 5).
Historically, the control of power systems is based on the generators’
controllers such as AVRs, PSSs, and ALFCs. Most power systems contain
elements that help regulate the network’s power flow, such as phase
changers, series compensation, and shunt compensation. Historically, these
devices were mechanically switched and may not be capable of reacting fast
enough to prevent cascading failures. With the demand of improving the
power quality, the reliability, and the security of modern power systems,
these controllers cannot fulfill the recent and future modernization
requirements. Consequently, the need for better high-speed control of power
flow led to an initiative by the Electric Power Research Institute (EPRI) to
develop power-electronic based devices 68, employing high speed, high power
semi-conductor technology, to help better regulate power flow. All these
devices are collectively known as Flexible AC Transmission Systems
(FACTS) devices. The family of FACTS devices includes high speed
versions of traditional devices like phase changers, and series and shunt
compensators, as well as devices other electronically controlled controllers.
According to the Edris 69, the FACTS devices are defined as “power
electronic based controllers and other static equipment which can regulate
68
www.epri.com
69
Edris, A. A. (1997). Proposed terms and definitions for flexible AC transmission system
(FACTS). IEEE Transactions on Power Delivery, 12(4).
247
the power flow and transmission voltage through rapid control action”.
Currently, the FACTS devices are considered a powerful tool for enhancing
both steady-state and dynamic performance of power systems 70. The advent
of the FACTS devices has required additional efforts in modeling and
analysis, requiring engineers to have a wider background for a deeper
understanding of power system's dynamic behavior.
From an economical point of view, the FACTS provides an excellent
alternative for increasing the transmission capacity by improving the efficient
use of the existing power grid. This can be accomplished through increased
control.
One of the most promising recent decentralized network controllers is the
family of (FACTS) devices. FACTS devices have been shown to be effective
in controlling power flow and damping power system oscillations. By
controlling power flow of an individual line, power can be redirected to/from
various parts of the power grid. Redirecting power flow allows for utilization
of power lines that physics of power flow alone would not allow. The
variable series compensation is highly effective in both controlling power
flow in the line and in improving stability. With series compensation, the
overall effective series transmission impedance from the sending end to the
receiving end can be arbitrarily decreased thereby influencing the power
flow. This capability to control power flow can effectively be used to
increase the transient stability limit and to provide power oscillation
damping. Therefore, this chapter will present the design of a supplementary
controller called Power Oscillation Dampers (PODs) for FACTS-based series
compensation with a main objective of enhancing the damping of power
systems71. A variety of design methods can be used for tuning the parameters
of various power system controllers. The most common conventional
techniques are based on frequency response 72 [5], pole placement73 [6],
eigenvalues sensitivity74 [6, 7] and residue method 75 [8].
70
EL-Shimy M. Multi-objective Placement of TCSC for Enhancement of Steady-State
Performance of Power System. Scientific Bulletin - Faculty of Engineering - Ain Shams
Uni. 2007;42(3):935 - 50.
71
M. Mandour, M. EL-Shimy, F. Bendary, and W. Ibrahim, “Damping of Power Systems
Oscillations using FACTS Power Oscillation Damper – Design and Performance
Analysis”, MEPCON 2014.
72
N. Martins and L. Lima, "Eigenvalue and Frequency Domain Analysis of Small-Signal
Electromechanical Stability Problems," IEEE Symposium on Application of Eigenanalysis
and Frequency Domain Method for System Dynamic Performance, 1989.
248
Due to their popularity, POD designs are presented in this chapter

using the frequency domain and residue methods for control design. The
Thyristor Controlled Series Compensator (TCSC) that belongs to the family
of FACTS devices is considered for its significant impact on both power flow
control (see Fig 7.1) and damping of oscillations. The TCSC are mainly used
for power flow control and as active series compensators for AC power
transmission lines; however, with the POD as a supplementary controller to
the TCSC, the capability of TCSCs in damping power oscillations will be
presented.
The small signal stability of power systems as affected by TCSC
devices and PODs are evaluated and compared with the base power system
where no FACTS devices are included. Both modal analysis and time domain
simulation (TDS) are presented to show the impact of the designed PODs on
damping the electromechanical oscillations in power systems. Several
examples are given to show the impact of POD input signals on the design
and system response.
7.2 FACTS devices – Definitions

FACTS devices are being carefully studied and installed for their fast
and accurate control of the transmission system voltages, currents, impedance
and power flow. The fundamental objective of FACTS devices is to improve
power system performance without the need for generator rescheduling or
topology changes. Therefore, they found numerous applications in
deregulated as well as regulated power systems. The fast development of
reliable power electronic devices facilitates the construction of the FACTS
devices. The main objectives of FACTS devices include, but not limited to:
 Increase the power transfer capability of a transmission network in a
power system,
73
B. C. Pal, "Robust pole placement versus root-locus approach in the context of damping
interarea oscillations in power systems," IEE Proceedings on Generation, Transmission
and Distribution, vol. 149, Nov 2002.
74
R. Rouco and F. L. Pagola, "An eigenvalue sensitivity approach to location and
controller design of controllable series capacitors for damping power system oscillations,"
IEEE Transactions on Power Systems, vol. 12, Nov 1997.
75
R. Sadikovic, et al., "Application of FACTS devices for damping of power system
oscillations," presented at the IEEE Power Tech, St. Petersburg, Russia, June 2005.
249
 Provide the direct control of power flow over designated transmission

routes,
 Provide secure loading of transmission lines near their thermal limits,
and
 Improve the damping of oscillations as this can threaten the security or
limit usage of the line capacity.
FACTS devices include multiple technologies that can be applied

individually or in coordination with other devices to control one or more
interrelated power system parameters such as series impedance, shunt
impedance, current, voltage and damping of oscillations. These controllers
were designed based on the concept of FACTS technology known as FACTS
Controllers.
According to the IEEE76, the term FACTS is defined as “The Flexible AC
Transmission System(FACTS) is a new technology based on power electronic
devices which offers an opportunity to enhance controllability, stability and
power transfer capability of AC Transmission Systems”. For power system
security, various control strategies can be implemented to FACTS devices to
guarantee the avoidance and survival of emergency conditions and to operate
the system at lowest cost.
FACTS controllers are historically introduced for solving specific power
system problems; however, with research efforts, their applications extend
their original objectives. From a chronological point of view, FACTS devices
can be classified into three generations as shown in Fig. 7.1. The figure
shows some common examples of the FACTS devices in each generation.
The classification also shows the method of connecting various devices to the
power system. The hybrid-connected FACTS are having series and shunt
connected components. The abbreviations are defined in Table 7.1.
As shown in Fig. 7.1, there are a large variety of FACTS devices currently
available. These devices comprise shunt devices, series devices, and hybrid
or (combined) devices. Each of these devices can be characterized by
different characteristics during the steady state and transient operation of
each of them. In addition, each device has a number of operational modes.
76
NGHingorani, L. G. (2000). Understanding facts concepts & tech nology of flexible AC
transmission system. NewYork, IEEEPowerEngineering Society.
250
For example, the TCSC can be controlled to provide either a constant power
control or a constant admittance control (see section 7.3.2).
(a)
(b)
(c)
Fig. 7.1: FACTS devices; (a) Classification; (b) Concepts of power flow
control using FACTS devices (active and also reactive); (c) Generalized
model of FACTS devices
251
Table 7.1: Definition of the abbreviations in Fig. 7.1 as well some other
devices
SVC Static VAR compensator
TCR Thyristor controlled reactor
TSC Thyristor switched capacitor
TSR Thyristor switched reactor
1st
Thyristor controlled series capacitor or
Generation TCSC
compensation
TCSR Thyristor controlled series reactors
TCPST Thyristor controlled phase shifting transformer
TCPAR Thyristor controlled phase angle regulator
STATC Static synchronous compensator
OM
2nd SSSC Static synchronous series compensator
Generation BESS Battery Energy Storage System
UPFC Unified power flow controller
IPFC Interline power flow controller
3rd GUPFC Generalized unified power flow controller
Generation HPFC Hybrid power flow controller
Fig. 7.1(c) shows a generalized model for FACTS devices. According to

its design and operational characteristics, some of the FACTS devices
contain all the shown components while some other FACTS devices have
some of them. The definitions of various variables and parameters shown in
Fig. 7.1 are as follows,
N ID of the FACTS device.

i Sending end bus number.
Terminal end (or sending end) bus number (0 for pure shunt connected
j
devices such as SVCs and STATCOMs).
Mode Control mode.
Pref Reference or desired active power flow arriving at the terminal end bus.
Qref Reference or desired reactive power flow arriving at the terminal end bus.
Vref Voltage set point at the sending end bus.
Maximum shunt element current at sending end bus at unity voltage.
Ishmax
Some devices such as TCSC do not contain a shunt element.
Bshmax Maximum value of the shunt element susceptance. This parameter is not
252
applicable to some devices such as STATCOMs while it is a main
parameter in SVCs. This value is also not available in pure series devices
such as TCSCs.
Minimum value of the shunt element susceptance. This parameter is not
applicable to some devices such as STATCOMs while it is a main
Bshmin
parameter in SVCs. This value is also not available in pure series devices
such as TCSCs.
Maximum bridge active power transfer between the series element and
Ptrmax the shunt element. This power transfer is possible in hybrid devices while
it is not available in either pure series or pure shunt devices.
VSEmax Maximum acceptable voltage magnitude at the terminal end bus.
VSEmin Minimum acceptable voltage magnitude at the terminal end bus.
Vsemax Maximum series voltage.
Isemax Maximum series current in MVA at unity voltage (default 0.0)
Reactance of the series element. This reactance is a dummy reactance in
Xse
some devices and used in certain solution states
Xsemax Maximum value of the reactance of the series element.
Xsemax Minimum value of the reactance of the series element.
The proposed generalized FACTS model has a series element that is

connected between the two shown buses and a shunt element that is
connected between the sending end bus and ground. The shunt element at the
sending end bus is used to hold the sending end bus voltage magnitude to Vref
subject to the sending end shunt current limit Ishmax or other limits such as the
shunt susceptance operational limits. In the steady state analysis, the shunt is
handled in power flow solutions in a manner similar to that of locally
controlling synchronous condensers and continuous switched shunts for the
operation within the linear control regions. The shunt element is considered
as a fixed susceptance when the shunt element hits its operational or control
limits. The series element is mainly controlling the power flow on a line
through appropriate changes. These changes include, for example, change in
the line reactance by TCSC or phase separation in TCPAR.
One or both of the shown elements (i.e. the shunt and the series) may be
used depending upon the type of device. An adaptation of the generalized
model for a specific device is a simple matter. UPFCs have both the series
and shunt elements active, and allows for the exchange of active power
between the two elements. SSSCs are modeled by deactivating the shunt
element i.e. by setting both the maximum shunt current limit and the
253
maximum bridge active power transfer limit to zero. STATCONs and SVCs
are modeled by setting the terminal end bus to zero (i.e. the series element is
deactivated). IPFC are modeled by using two consecutively numbered series
FACTS devices. By setting the control mode, one device will be assigned, as
the IPFC master device while the other becomes the slave device. Both
devices have a series element without a shunt element. The conditions of the
master device define the active power exchange between the devices.
The mathematical modeling of some of these devices will be presented in
the next sections and chapters. According to the nature of the analysis and the
simplifying assumptions, some of the models will be applicable for dynamic
studies while the others will be applicable for static studies.
7.3 Modeling and Modal Analysis

7.3.1 Linearized Modeling for POD design
The power systems are dynamic systems that can be represented by
differential algebraic equations in combination with non-linear algebraic
equations. Hence, a power system can be dynamically described by a set of n
first order nonlinear ordinary differential equations. These equations are to be
solved simultaneously. In vector-matrix notation, these equations are
expressed as,
where: , , ,
, , is the order of the system, is the
number of inputs, and is the number of outputs.
The column vector x is called the state vector and its entries are the state
variables. The vector u is the vector of inputs to the system, which are
external signals that have an impact on the performance of the system. The
output variables are those that can be observed in the system. The column
vector y is the vector of system output variables, referred as output vector and
is the vector of nonlinear functions defining the output variables in terms of
state and input variables.
254
The design of POD controllers is based on linear system techniques. After

solving the power flow problem, a modal analysis is carried out by
computing the eigenvalues and the participation factors of the state matrix of
the system. The dynamic system is put into state space form as a combination
of coupled first order, linearized differential equations that take the form,
where represents a small deviation, is the state matrix of size ,B

is the control matrix of size , is the output matrix of size , and
is the feed forward matrix of size The values of the matrix D define
the proportion of input which appears directly in the output.
The eigenvalues of the state matrix can be determined by
solving Let be the ith eigenvalue of the
state matrix A; the real part gives the damping, and the imaginary part gives
the frequency of oscillation. The relative damping ratio is then given by:
If the state space matrix A has n distinct eigenvalues, then the diagonal
matrix of the eigenvalues (Λ), the right eigenvectors ( ), and the left
eigenvectors (Ψ) are related by the following equations.
In order to modify a mode of oscillation by a feedback controller, the

chosen input must excite the mode and it must also be visible in the chosen
output [8]. The measures of those two properties are the controllability and
255
observability, respectively. The modal controllability and modal

observability ( ) matrices are respectively defined by,
The mode is uncontrollable if the corresponding row of the matrix is

zero. The mode is unobservable if the corresponding column of the matrix
is zero. If a mode is neither controllable nor observable, the feedback
between the output and the input will have no effect on the mode.
7.3.2 TCSC Dynamic Modeling and Control

The TCSC as shown in Fig. 7.2(a) can be defined as capacitive
reactance compensator which consists of a series fixed capacitor (FC) bank
shunted by a thyristor-controlled reactor (TCR) in order to provide a
smoothly variable series capacitive reactance. When placed in series with a
transmission line as shown in Fig. 7.2(b), the TCSC can change the power
flow on the line as a result of the changes made by the TCSC on the line
reactance; the following algebraic equations approximately govern the power
flow on a line connecting buses k and m when the line resistance is neglected.
The TCSC can be controlled to provide either a constant power control or

a constant admittance control. The constant power control scheme is shown
in Fig. 7.3(a). In this case, the state variables of the TCSC are and
. Therefore, the state space model of the constant power
regulator takes the form,
256
where: , ,
, and
(a)
(b)
Fig. 7.2 TCSC structure and control modes: (a) Basic structure, (b) A line
with TCSC
The constant admittance regulator for TCSC takes the form shown in Fig.
7.3(b). In this case, one state variable ( ) represents the TCSC and the
state space model takes the form,
The constant admittance operation of the TCSC will be considered in the

following analysis for compensating differences between the reactances of
two parallel transmission lines.
257
(a)
(b)
Fig. 7.3 Control modes of TCSC: (a) Constant power regulator; (b)
Constant admittance regulator
7.4 POD Design and Tuning

The POD controller is designed using two methods. These are the
frequency response method and the residue method. The main design
objective is to achieve a predefined damping level of the electromechanical
oscillations. The general control diagram of the power system controlled by
the POD is shown in Fig. 7.4. The structure of the POD controller (Fig. 7.5)
is similar to the classical power system stabilizer (PSS) described in chapter
4. The controller consists of a stabilizer gain, a washout filter, and phase
compensator blocks. The washout signal ensures that the POD output is zero
in steady-state. The output signal vPOD is subjected to an anti-windup limiter
and its dynamics are dependent on a small time constant Tr (in this analysis Tr
= 0.01 s). The gain Kw determines the amount of damping introduced by the
POD and the phase compensator blocks provide the appropriate phase lead-
lag compensation of the input signal.
258
Fig. 7.4 General feedback control system
Fig.7.5: Scheme of the POD controller
7.4.1 Frequency Response Method

The POD controller is designed using the frequency response method
through Nyquist plots of a given Open Loop Transfer Function (OLTF).The
Nyquist criterion allows to assess the closed-loop stability of a feedback
system by checking the OLTF poles and plotting its frequency response.
Closed-loop stability of the open-loop unstable system is obtained by
ensuring an anti-clockwise encirclement of the (-1) point of the complex
plane in the Nyquist plot of the OLTF after applying feedback compensation.
The main steps of the procedure for POD design using the frequency
response method can be described by a flowchart as shown in Fig. 7.6. The
main design steps in the POD design using the frequency response method
can be summarized as follows 77,
1. Eigenvalue analysis. In this design, the critical modes of the

uncompensated system (i.e. without the POD) are identified based on
eigenvalues and the participation factors of the state matrix. The
participation factors ( ) of the state variables to each eigenvalue are
computed by using right and left eigenvectors. If and represent
77
H. M. Ayres, I. Kopcak, M. S. Castro, F. Milano and V. F. d. Costa, "A didactic
procedure for designing power oscillation damper of facts devices", Simulation Modelling
Practice and Theory, vol.18, no.6 June 2010
259
respectively the right and the left eigenvector matrices (Eqs. (7.6) to
(7.8)), then the participation factor of the state variable to the
eigenvalue can be defined as,
2. State-space form. In this step, all output and input matrices ( , , ,

and ) are determined. The observability and controllability as defined
by Eqs. (7.9) and (7.10) can be determined based on these matrices.
3. Nyquist analysis. In this step, the value washout filter time constant is
randomly selected between 1 and 20 Sec then the Nuquist plot of the
uncompensated loop including the washout filter is constructed. The
required phase compensation  is then determined from the constructed
Nyquist plot. The objective is to obtain a good phase margin based on
the critical frequency .
4. Compensator blocks tuning. Based on the value of that is determined

in the previous step, the parameters of the phase compensator blocks
are determined in this step using,
where is the number of the lead-lag blocks and is the frequency

of the critical mode to be damped. The value of is usually one or
two; Fig. 7.7 shows a POD with two lead-lag blocks (i.e. = 2)
which is considered in this paper. In this layout, T3 and T4 are equal to
T1 and T2.
5. Damping ratio adjustment. In this step, the root locus plot of the
compensated system is used to determine the value of Kw that provide
260
an acceptable damping ratio (i.e.  ≥ 10%). The POD design is

completed by completing this step; however, further adjustment of the
design can achieved by fine tunning the POD parameters as described
in the next step.
\
Fig. 7.6: Flowchart describing the frequency response method
6. Fine tuning of the POD design. The POD parameters have to be

specified and chosen to fulfill specific performance parameters. The
261
damping is one of the most important performance parameters;

however, the performance is also governed by many parameters such
as the maximum rise time ( ), the maximum overshoot ( ), the
desired damping ratio ( ), and the settling time ( ). The fulfillment of
these performance parameters can be achieved by fine tuning of the
POD parameters keeping in mind that the damping ratio is the main
specification in power system control design and, for large power
systems, 10% of damping is considered sufficient for POD controllers.
7.4.2 Residue Method

The residue method for POD design will be described based on the
general feedback control system shown in Fig. 7.7. The transfer function of
the system is G(s) and the feedback control is H(s). The open loop transfer
function of a SISO system78 is
can be expanded in partial fractions of the Laplace transform in

terms of the matrix , the matrix, the right eigenvectors, and the left
eigenvectors as:
Each term in the nominator of the summation is a scalar called residue.

The residue for a particular mode gives the sensitivity of the eigenvalue of
that mode to the feedback between the output and the input of the SISO
system. The residue is the product of the mode’s observability and
controllability.
When applying the feedback control, eigenvalues of the initial system
are changed. It can be proved that when the feedback control is applied,
movement of an eigenvalue is calculated by,
78
R. Sadikovic, et al., "Application of FACTS devices for damping of power system
oscillations," presented at the IEEE Power Tech, St. Petersburg, Russia, June 2005.
262
Fig. 7.7 Closed-loop system with POD control.
It can be observed from (7.23) that the shift of the eigenvalue caused by a
feedback controller is proportional to the magnitude of the residue. For
improving the damping of the system, the change of eigenvalue must be
directed towards the left half side of the complex plane. This is can be
achieved by the use of the FACTS-POD controller. The compensation phase
angle required to move an eigenvalue to the left in parallel to the real
axis is illustrated in Fig. 7.8. This phase shift can be implemented using the
lead-lag function of the POD represented by Fig. 7.5 and equation (7.25). The
parameters of the lead-lag compensator are determined using,
Fig. 7.8 Shift of eigenvalues with the POD controller.
263
Fig. 7.9 Flowchart describing the residue method
264
where: is the phase angle of the residue , is the frequency of

the mode of oscillation in rad/sec, is the number if compensation stages
(in this paper, ).
The controller gain is computed as a function of the desired eigenvalue

location according to Eq. 7.29.
The flowchart summarizing the previous design procedures is shown in

Fig.7.9.
7.5 Study System 1 – SMIB system

The study system is shown in Fig.7.10. This system will be studied and
analyzed with the aid of the Power System Analysis Toolbox (PSAT), the
Simulink and the control system toolbox of Matlab 2012a. The study system
consists of four 555 MVA, 24 kV, 60 HZ units supplying power to an infinite
bus through two transmission circuits. The four generators are represented by
one equivalent generator that is represented by the second order dynamic
model. On 2220 MVA and 24 kV base, the transient reactance of the
equivalent generator ( ) is 0.3 p.u, its inertia constant (H) is 3.5 sec, and its
damping coefficient (D) is 10 in p.u torque/p.u speed. The initial conditions
of the system in p.u on the 2220 MVA, 24 KV base are ,
, P = 0.9, and Q = 0.3 (overexcited).
265
Fig. 7.10 The study system with the p.u network reactances are shown on
2220 MVA base
The TCSC will be placed on line 2 as shown in Fig.7.11 and will be used
to compensate the difference between the reactances of line 1 and line 2 of
the study system. In this case, series compensation ratio is 0.462. The input
variables to the PSAT block for modeling the TCSC are: sec,
p.u, p.u.
Fig. 7.11 The study system after connecting the TCSC
7.6 CASE STUDY 1 – THE SMIB SYSTEM

The results will be presented through studying the system described in
Fig. 7.10 in three scenarios as shown in Fig. 7.12. In the Time Domain
Analysis (TDS), the considered small-signal disturbance is a +10 % step
increase in the mechanical power input (Pm) to the equivalent generator of the
study system. The changes in the mechanical power will be started at t = 2
sec.
Fig. 7.12 Study Scenarios

266
7.6.1 Impact of the TCSC on the small- signal stability

Tables 7.2 and 7.3 show the system dominant eigenvalues and their
participation factors of scenario 1 and scenario 2 respectively.
Table 7.2: Scenario 1 - dominant eigenvalues and participation factors

f  Participation factors
Eigenvalues Most Associated States
(Hz)
(%) δ1 ω1
-5.49711  j7.6085 9.19.1 9.34 5.0 5.0 δ1, ω1
Table 7.3: Scenario 2 - dominant eigenvalues and participation factors

f  Participation Factors
Eigenvalues Most associated states
(Hz) (%) δ ω X1_TCSC
-5.49711j8.0854 9.1191 8.79 0.5 0.5 0 δ ,ω
-100 0 100% 0 0 1 X1_TCSC
It is clear from these tables that both scenarios are stable; however, the
eigenvalues of the system are changed as an effect of adding the TCSC to the
system. The TCSC adds a non-oscillatory eigenvalue. The frequencies of the
oscillatory modes of the system with TCSC are increased by 6.216% in
comparison with the system without the TCSC while their damping ratios are
reduced by 5.888%; (the percentage changes are calculated according to: %
change = 100*(new value – old value)/old value). Therefore, the inclusion of
the TCSC degrades the system stability. The damping ratio is less than 10%.
Therefore, inclusion of POD is recommended to elevate the damping ratio to
a value higher than or equal to 10%. POD designs according to the frequency
response and residue methods are presented in the next sections; the objective
is to increase the damping ratio to an acceptable value i.e.  ≥ 10%. Various
input signals to the POD will be considered. In addition, the observability and
controllability of them will be determined using equations (7.9) and (7.10).
7.6.2 Observability and controllability of various input signals

The observability and controllability of candidate feedback signals to
the POD will be determined. Based on Fig. 7.10, the varies feedback signals
are the current across the transformer, the sending end active power, and the
sending end reactive power. The modal controllability and modal
267
observability ( ) matrices associated with the considered feedback signals

are shown in Table 7.4.
Table 7.4: Modal observability and controllability of various feedback signals

Modal
Modal observability C’
Feedback signal controllability B’
matrix
matrix
The current across the [ 1.3707 1.3707 -0.2843]  0.08  j 0.93 
transformer  
 0.08  j 0.93
The sending end active power [1.223 1.223 -0.2785  
The sending end reactive [0.1438 0.1438 0.0209  100 
 
power
Considering the critical electromechanical modes shown in Table 7.3

(highlighted by gray shading), it is depicted from Table 7.4 that all the
considered signals are observable and controllable. Highest observability is
associated with the current across the transformer feedback signal followed
by the sending end active power then the sending end reactive power. Due to
space limits, POD designs will be presented considering only the current
across the transformer as a feedback signal; however, the presented design
algorithms are general and can be applied to design PODs considering any
acceptable feedback signal.
7.6.3 POD designs

Based on the flowcharts presented in Fig. 7.6 and 7.9, POD designs
using the frequency response and residue methods are respectively presented
in this section. Designs with each of the considered feedback signals will be
determined. The initial stages of the design i.e. building the input and output
matrices, analysis of the eigenvalues, modal controllability, and modal
observability are already performed in the previous subsections. The washout
filter time constant (Tw) is chosen to be 7. This value is arbitrary selected
between 1 and 20.
A) The Frequency Response Method

With the transformer current as an input signal to the POD, the
Nyquist plot (for positive frequencies) of the uncompensated OLTF (pre-
design) and the compensated OLTF (post-design) is shown in Fig. 7.13.
268
Fig. 7.13 Nyquist plots of SMIB system with and without POD
It is depicted from Fig. 7.13 and Table 7.3 that the OLTF for the
system is stable, but presents poorly damped poles. For a good POD design,
the resulting polar plot should be approximately symmetric with respect to
the real axis of the complex plane. Based on the Nyquist plots, the value of
the angle  required to relocate the critical frequency is 100.23o. Therefore,
using equations (7.19) and (7.20), the parameters of the lead-lag
compensators are T1 = 0.3408 sec. and T2 = 0.0449 sec. The gain Kw is
determined based on the root locus of the system including the POD. The
Matlab control system toolbox is used to construct the root locus as shown in
Fig. 7.14. The gain Kw is determined by dragging the critical mode to an
acceptable damping ratio which is chosen to be higher than 10%. As shown
in Fig. 7.14, the value of the damping of the critical mode in the compensated
system is set to 15.63% and the corresponding gain is 0.0641. The transfer
function of the POD is then takes the form,
With the POD connected to the system shown in Fig. 7.11 as shown in
Fig. 7.15, the design will be evaluated by both the eigenvalue analysis and
the TDS of the compensated system. The results of the eigenvalue analysis of
the compensated system is shown in Table 7.5 which indicates that the
minimum damping of the system is improved to 15.63% as set by the POD
design. Tables 7.2 and 7.3 show respectively that the damping of the system
269
without TCSC is 9.34% and 8.79% in the uncompensated system with TCSC.
This ensures the success of the POD design for improving the damping of the
system.
Fig. 7.13 Root locus of the compensated system and selection of the gain
Kw
Fig. 7.15 Modelling of the SMIB in the 3rd scenario
Table 7.5: Frequency domain method based eigenvalue analysis of the

compensated system
f 
Eigenvalues Most associated states
(Hz) (%)
-91.1111 j61.9306 95.50 19.53% X1_TCSC, V3_POD
-13.0749+j0 0 100% V2_POD
-9.1117j7.8266 9.1.91 15.63% δ ,ω
-0.14262+j0 0 100% V1_POD
The TDS is performed considering a 10% step increase in the

mechanical power input to the equivalent synchronous generator. This
270
disturbance started at t = 2 sec. The simulation is performed using the Matlab

control system toolbox. The responses of the systems of the three scenarios
shown in Fig. 7.12 are compared as shown in Fig. 7.16.
(a)
(b)
Fig. 7.16 TDS for 10% increase in the mechanical power: (a) Rotor
angles; (b) Rotor angular speeds.
It is depicted from Fig.7.16 that the POD improves the dynamic

performance of the system through increasing the system damping,
decreasing the overshoots, and decreasing the settling time.
B) The Residue Method

The design is based on the flowchart of Fig. 7.9. With the transformer
current as an input signal to the POD, the residues for all eigenvalues of the
system without POD should be obtained to determine the residue of the most
critical mode. This is shown in Table 7.6. Afterward, the POD parameters
can be determined as previously described. The transfer function of the POD
is then takes the form:
271
Table 7.6 Residues of the eigenvalues

Eigenvalues Residues
-0.1043-
-5.49711+j8.0854
j1.281
-
-5.49711-j8.0854
0.1043+j1.281
-100 -28.4305
Table 7.7 shows the eigenvalue analysis of the system after connecting
the POD to the system which indicates the improvement in the system
damping in comparison to the systems of scenarios 1 and 2. The TD
responses as various scenarios subjected to the considered disturbance are
shown in Fig. 7.16. The results validate the POD design using the residue
method which results in approximately the same TD response of the system.
Table 7.7 Residue method: Eigenvalue analysis of the compensated system

Most
f 
Eigenvalues associated
(Hz) (%)
states
X1_TCSC,
1.1.97 58.57%
-15.441.j42.5808 V3_POD
-9.1114j7.9995 9.1117 15.33% δ ,ω
-12.4271+j0 0 100% V3_POD
-0.14255+j0 0 100% V1_POD
C) Further Analysis
In this section a summary of some other related results will be
presented to show the effect of some critical issues in damping of oscillations
in power systems. These issues are the impact of POD input signal and the
value of the time constant of the washout filter (Tw) on damping of power
system oscillations. The impact of POD input signal is shown in Fig. 7.17
which indicates that better dynamic performance can be achieved with the
transformer reactive power as a feedback signal while the other feedback
signals (i.e. the transformer current and the transformer active power ) have
the same impact on the dynamic performance of the system. Therefore,
272
careful choice of the input signal is important for damping maximization

through POD design. High damping can be achieved with the transformer
reactive power as an input signal because of the less control loop stability
restrictions on the POD parameters in comparison with other signals.
(a)
(b)
Table 7.17 TDS for 10% increase in the mechanical power with various
feedback signals: (a) Rotor angles; (b) Rotor angular speeds.
The results shown in Table 7.7 are obtained with Tw = 7. Although the
literature recommended to select a random value for Tw between 1 to 20 sec,
detailed analysis shows that the acceptable range of Tw is dependent on the
system parameters and operating conditions. This is demonstrated in Table
7.8 for the same design conditions shown in Table 7.7. Three values of Tw are
shown. These values are 1 sec, 7 sec, and 14 sec.
273
Table 7.8: Impact of Tw on the dynamic performance

Tw = 1 Tw = 7 Tw = 14
 f  f  f
(%) (Hz) (%) (Hz) (%) (Hz)
27.54% 1.09.4 19.53% 95.50 9.15% 16422.0
100% 0 100% 0 100% 0
16% 9.1011 15.63% 9.1.91 16.21% 9.10..
100% 0 100% 0 100% 0
Table 7.8 shows that for all the considered values of Tw, the damping ratio
of the critical electromechanical modes can be successfully increased to
values higher than 15% which is practically acceptable damping level. The
interesting part here is that, as shown in Table 7.8, that increasing Tw results
in decreasing the damping ratios and increasing the frequencies of some of
the electromechanical modes that was originally not critical (i.e. their
damping ratio was higher than 10%). It is also shown that high value of T w
such as 14 could result in creating new critical modes in the compensated
system. Therefore, careful selection of Tw should be considered in the initial
stages of the design. It is also important to know that a suitable value of Tw
for a specific system may be not suitable for another system. In addition, the
impact of Tw on the dynamic performance is also sensitive to the operating
conditions of a power system.
7.7 Case Study 2 – Weakly interconnected multi-machine

system
The two-area system that is considered in the previous chapter is
considered here to but with some structural and data changes for the inclusion
of FACTS devices and consideration of higher-order models of the
generators.
Fig. 7.18: The two area study system
274
The new study system is shown in Fig. 7.18. The detailed system data are
listed Appendix 8.
In the original system, each area consists of two synchronous generator
units. The rating of each synchronous generator is 900 MVA and 20 kV.
Each of the units is connected through transformers to the 230 kV
transmission line. There is a power transfer of 400 MW from Area 1 to Area
2. The PSAT model of the two area system is shown in Fig. 7.19.
Fig. 7.19 The PSAT model of the study system
This system will be studied considering three scenarios. In the first two
scenarios, the synchronous generators will be controlled by AVRs only (i.e.
without PSS) and the PODs in these scenarios will be designed for the
purpose of improving the small signal stability by increasing the critical and
unacceptable damping ratios to acceptable levels (>=10%). In the first
scenario, one PODfor one SVC will be designed and the SVC will be
installed in the central bus between the two areas (bus 8). In the second
scenario, two PODs for two SVCs will be simultaneously designed. In this
case, the old capacitors located at bus 7 & bus 9 will be replaced by SVCs. In
the third scenario, the PSS will be included in the system and the POD will
be designed to improve the system stability in the presence of the PSS. The
description of study scenarios is shown in Fig. 7.20.
The basic structure of SVC is shown in Fig. 7.20(a). The SVC is
connected to a coupling transformer that is connected directly to the ac bus
whose voltage is to be regulated. The SVC is composed of a controllable
shunt reactor and a fixed shunt capacitor(s). When placed in shunt with the ac
bus as shown in Fig. 7.20, total susceptance of SVC can be controlled by
275
controlling the firing angle of thyristors; however, the SVC acts like fixed
capacitor or fixed inductor at the maximum and minimum limits.
Two Area Study

System
Scenario 1 Scenario 2 Scenario 3

With AVRs and one with AVRs and two with AVRs, PSS
POD PODs and POD
Fig. 7.19 Study scenarios of the two area system
The SVC regulators can be controlled to provide either two models. The
first one assumes a time constant regulator, as depicted in Fig. 7.21(a). In this
model, a the dynamics of the SVC takes the form,
(a)
(b)
Fig. 7.20 SVC structure and control modes: (a) Basic structure, (b) A bus
with SVC
276
The model is completed by the algebraic equation expressing the

reactive power injected at the SVC node,
As shown, the regulator has an anti-windup limiter. Therefore. the

reactance bSVC is locked if one of its limits is reached and the first derivative
is set to zero.
The second model, as depicted in Fig. 7.21(b) takes into account the firing
angle α, assuming a balanced, fundamental frequency operation. Thus, the
model can be developed with respect to a sinusoidal voltage. The differential
and algebraic equations are then,
The state variable α also undergoes an anti-windup limiter.
A. Modal analysis of the original system (no SVCs)

Before starting the study scenarios, the system eigenvalues without SVCs
will be obtained to show the impact of the added SVCs on the system
stability. The system eigenvalues that have the lowest damping ratios will be
included in Table 7.9.
Table 7.9: Dominant eigenvalues and participation factors of the test system
without FACTs
Most
f  Status of
(Hz) (%) Eigenvalues
states
-0.54657j6.5963 9.5017 8.25% δ2 ,ω2 Unacceptable
-0.06528+j3.3288 5.0111 1.96% δ3 ,ω3 Critical
277
(a)
(b)
Fig. 7.21 Control modes of SVC: (a) SVC Type 1 Regulator; (b) SVC
Type 2 Regulator
It is clear that the system without FACTs is stable but presents three
poorly damped eigenvalues with damping ratios 8.25%, 8.25% and 1.96%.
Hence, the system has very poor damping of oscillations.
B. Scenario 1
In this scenario, an SVC will be added to the system at bus 8; Fig. 7.22.
The synchronous generators will be controlled by AVRs but without PSSs.
Fig. 7.22: Modelling of first scenario
278
The eigenvalues of the system with SVC which have the lowest damping
ratios will be shown in Table 7.10. In comparison with Table 7.9, it is clear
that the SVC degrades the system damping.
Table 7.10: Dominant eigenvalues and participation factors of the test system
in Scenario 1 with SVC
Most
f  Status of
states
-0.53283j6.6197 9.504 8% δ2 ,ω2 Unacceptable
-0.05755+j3.5689 5.0.151 1.6% δ3 ,ω3 Critical
Table 7.10 Root locus of the compensated system and selection of the gain
Kw
With the SVC, the critical eigenvalue has a damping ratio 1.6% while the
two unacceptable eigenvalues have damping ratios 7.77% and 8%. Since the
damping ratio is less than 10%. Therefore, inclusion of POD is required to
elevate the damping ratio to a value higher than or equal to 10%. The
frequency response method will be used for that purpose. The POD washout
filter time constant will be chosen as Tw=1 sec. By testing all possible POD
stabilizing signals according to its own root-locus leads to that there is no any
signal that can achieve the design objective. As demonstration, the POD is
designed with the sending current between Bus 6 and Bus 7 are used as
279
stabilizing signal to the POD. Based on the Nyquist plot, the value of the
angle  required to relocate the critical frequency is found to be be 103.71 o.
Therefore, using equations (7.20) and (7.21), the parameters of the lead-lag
compensators are T1 = 0.0969 sec. and T2 = 0.8104 sec. The POD gain (Kw) is
selected based on the root-locus of the system as shown in Fig. 7.23. It is
shown that with a gain of 1.21 the 1.6% damping ratio becomes 11% while
the 7.77% and 8% damping ratios will not be increased to the acceptable
levels. The transfer function of the POD then takes the form,
With the POD connected as shown in Fig. 7.11, the eigenvalues are shown
in Table 7.11. The results indicates that it is not possible to increase all the
damping ratios to accepted level by designing one POD in the two area
system which does not contain PSS.
Fig. 7.11Modelling of the two area system in the 1st scenario with POD
Table 7.11 Dominant eigenvalues and participation factors of the test system
in Scenario 1 with SVC & POD
Most
f  Status of
states
-0.53477j6.6161 9.50.7 8% δ2 ,ω2 Unacceptable
-0.39502j3.4775 5.00459 99% δ1 ,ω1 Accepted
280
C. Scenario 2
In this scenario, the two old capacitors installed at Bus 7 and Bus 9 are
replaced by two SVCs for the purpose of designing two PODs for the system.
The two area system with the installed SVCs is shown in Fig. 7.12. The
synchronous generators will also be controlled by only AVRs as the previous
scenario.
Fig. 7.12 Modelling of the two area system in the 2nd scenario
The eigenvalues of the system with SVCs which have the lowest damping
ratios will be shown in Table 7.12.
in Scenario 2 with SVCs
Most
f  Status of
states
-0.52663j6.6178 9.50.. 8% δ2 ,ω2 Unacceptable
-0. 0.03599+j3.7071 5.01551 0.45% δ1 ,ω1 Critical
According to Table 7.12, the presence of the SVCs degrads the system in
comparsion with the original topology and scenario 1. There is a critical
eigenvalue with damping ratio 0.45% and two unacceptable eigenvalues with
damping ratios 7.92% and 8%.
The design of the PODs will be achieved in two stages. In the first stage,
one POD is designed and the system minimum damping ratio is determined.
If the minimum damping is acceptable, then there is no need for the inclusion
of the second POD. Otherwise, the second POD is designed for achieving
acceptable damping ratios.
281
For the first POD, the sending current between Bus 10 and Bus 9 is
chosen as a stabilizing signal to the POD for the SVC located at bus 9. The
POD transfer function is then found to be,
With this POD connected to the system, the system eigenvalues are shown
in Table 7.13.
in Scenario 2 with SVCs & POD1
Most
f  Status of
states
-0.52629j6.6183 9.50.4 8% δ2 ,ω2 Unacceptable
-0.38416j3.7492 5.01111 10.1% δ3 ,ω3 Acceptable
It is depicted from Table 7.13 that the first POD has a positive impact on
the critical eigenvalues since it has the ability to increase the damping of the
critical eigenvalues to 10.2% and it has no significant impact on the other
eigenvalues. The design objective of the second POD is to increase the
unacceptable damping ratios to accepted level. The root-locus of all possible
stabilizing signals to the second POD are determined to select the best signal
which can improve the damping ratios of the rest of eigenvalues to accepted
levels. Unfortunately, through testing all possible POD stabilizing signals, no
signal can be found suitable for achieving the design target; however, the best
found POD has the following transfer function and the system than has the
eigenvalues shown in Table 7.14 which are unacceptable but slightly
improved.
282
in Scenario 2 with SVCs & PODs
Most
f  Status of
states
- 0.31606 j 2.7477 5.7751 99.4% δ3 ,ω3 Acceptable
D. Scenario 3:
In this scenario, another control will be added on the synchronous
generators of the two area system shown in the previous scenario. PSS will
be added to generator 4 as shown in Fig. 7.13. The PSS is added to generator
4 for the purpose of improving one of the unacceptable damping ratios and
the others will be increased by the POD.
Fig. 7.13 modelling of the two area system in the 3rd scenario
The eigenvalues of the system with SVCs which have the lowest damping
ratios are shown in Table 7.15.
in Scenario 3 with SVCs
Most
f  Eigenvalue
(Hz) (%) Status
states
-1.6434j7.5784 9.1171 21.17% δ4 ,ω4 acceptable
-0.52845j6.6191 9.50.1 7.95% δ2 ,ω2 Unacceptable
- 0.13075j 3.8027 5..5004 3.4% δ3 ,ω3 Critical
According to Table 7.15, there is a critical eigenvalue with damping ratio

3.4% and unacceptable eigenvalues with damping ratios 7.95%. It is clear
283
that the PSS has a positive impact on the damping; however, the damping
ratios are still unacceptable. Therefore, inclusion of POD is performed to
elevate the damping ratio to a value higher than or equal to 10%. It is found
that the sending current between Bus 5 and Bus 6 as a stabilizing signal to the
POD can be used for achieving this target. Based on the Nyquist plot, the
value of the angle  required to relocate the critical frequency will be -21.77o.
The parameters of the lead-lag compensators are T1 = 0.3184 sec. and T2 =
0.2172 sec. The POD gain (Kw) is selected based on the root-locus of the
system. With a gain of 0.0767 the 3.4% damping ratio becomes 18.1% while
the 7.95% damping ratio becomes 10%; see Fig. 7.14.
Fig. 7.14 Root locus of the compensated system and selection of the gain
Kw
The transfer function of the second POD then takes the form,
This chapter presented successful designs of PODs for enhancing the

stability and stabilization of power systems. The design approach was based on
conventional methods in which the design is carried out at a specific operating
point. The controllers are then performs well at that specific operating point as
284
well as operating points in the close proximity of the design point; however,
power systems are subjected to large normal changes of the operating
conditions during its normal operating conditions. Therefore, the capability of
the designed controllers to serve their objectives and functions are doubtful
when there are large deviations of the operating conditions in comparison with
the design-based operating conditions. Therefore, the use of intelligent and
adaptive systems that combine knowledge, techniques and methodologies from
various sources are essential for the real-time control of power systems79.
These methods are out of the scope of the current volume of the book;
however, a brief survey is given here for interested readers and researchers.
In practice different conventional control strategies are being used for
POD design as described in this chapter; however, handling of system
nonlinearities is the main advantage of the AI design techniques over the
conventional design techniques. Artificial Intelligence techniques such as
Fuzzy Logic, Artificial Neural networks and Genetic Algorithm can be
applied for POD design, which can overcome the limitations of
conventional controls. The POD based FACTs devices such as UPFC and
STATCOM has been designed by neural network80,81 in a multi-machine
power system to suppress the power system oscillations and to enhance the
transient stability. The POD has also been designed by fuzzy logic 82,83 to
improve the damping the power system oscillations. The genetic algorithm 84,85
79
Patel, R. N., S. K. Sinha, and R. Prasad. "Design of a Robust Controller for AGC with
Combined Intelligence Techniques." Proceedings of World Academy of Science:
Engineering & Technology, Vol. 47, No.1, pp. 95-101 , 2008.
80
D. Ravi Kumar, N. Ravi Kiran " Hybrid Controller Based UPFC for Damping of
Oscillations in Multi Machine Power Systems," International Journal of Engineering
Research & Technology (IJERT) , ISSN: 2278-0181, Vol. 2, Issue 10, October - 2013.
81
Mozhgan Balavar,” Using Neural Network to Control STATCOM for Improving
Transient Stability” Journal of Artificial Intelligence in Electrical Engineering, Islamic Azad
University, Ahar, Iran, Vol. 1, No. 1, June 2012.
82
Dakka Obulesu, S. F. Kodad, B V. Sankar Ram, " Damping of oscillations in multi-
machine integrated power systems using hybrid fuzzy strategies," International Journal of
Research and Reviews in Applied Sciences, Vol. 1, Issue 2, November 2009.
83
Prasanna Kumar Inumpudi and Shiva Mallikarjuna Rao N,” Development of a Fuzzy
Control Scheme with UPFC’s For Damping of oscillations in multi machine integrated
power systems” International Journal of Engineering Research and Applications (IJERA),
Vol. 1, Issue 2, pp.230-234.
84
Aghazade, A.; Kazemi, A.”Simultaneous coordination of power system stabilizers and
STATCOM in a multi-machine power system for enhancing dynamic performance”4th
285
method can be also used to design the POD and the POD gain was obtained
through the minimization of an objective function based on the damping ratio.
Moreover, the GA-based methodology can be used to solve power system
problems86 such as the inter-area congestions and oscillations issues in a HV
Transmission system.
Recently, many researches focus on the robust control design of the POD to
enhance the system dynamic performance. H∞ technique has been applied87 to
design the POD in large wind power plants (WPPs) to improve small signal
stability by increasing the damping of electromechanical modes of oscillation.
While in , the optimization of POD parameters is formulated based on a
mixed H2 H∞ control88, and carried out under all system outage events such as
line tripping, and load/generation shedding.
7.8 Placement of FACTS Devices

The placement and the selection of FACTS devices in power networks are
mainly determined by the problem to be solved. Usually, the fundamental
function(s) of a device is used for the allocation of a device or a set of devices
for solving problems. For example, solution of bus voltage magnitude
problems are usually allocated to devices such as SVCs, STATCOMs, and
ULTC transformers. On the other hand, treatment of problems associated with
power flow control and line flow overloads or congestions are usually
allocated to, for example, TCSCs. The primary functions of various FACTS
devices for implementing the proper corrective actions to common problems
are illustrated in Table 7.16. The use of FACTS devices in their traditional
International Power Engineering and Optimization Conference (PEOCO), IEEE, pp: 13 –

18, Shah Alam, 23-24 June 2010
85
M. Mandour, M. EL-Shimy, F. Bendary, and W.M. Mansour. Design of power
oscillation damping (POD) controllers in weakly interconnected power systems including
wind power technologies. Accepted for publications in the JEE, Oct. 2015.
86
A. Berizzi, C. Bovo, V. Ilea,” Optimal placement of FACTS to mitigate congestions and
inter-area oscillations” PowerTech, 2011 IEEE Trondheim, pp: 1 – 8, Trondheim, 19-23
June 2011
87
J. Mehmedalic, T. Knuppel and J. Ostergaard,” Using H∞ to design robust POD
controllers for wind power plants” Universities Power Engineering Conference (UPEC),
2012 47th International, IEEE, pp.1-6, London, 4-7 Sept. 2012
88
Issarachai Ngamroo and Tossaporn Surinkaew,” Optimization of Robust Power
Oscillation Dampers for DFIG Wind Turbines Considering N-1 Outage Contingencies”
Innovative Smart Grid Technologies Conference Europe (ISGT-Europe), 2014 IEEE PES,
Istanbul, October 12-15
286
control form in the enhancement of power system damping, as shown in this

chapter, requires additional supplementary control actions (i.e. the PODs).
Supplementary control for FACTS devices (not considered in Table 7.16) can
be provided for making them effective in enhancing the voltage stability,
transient stability, dynamic security, and reliability. Some devices have
inherent control and capabilities to contribute in enhancing the system
dynamics.
In this section, the multi-objective placement of TCSC in power networks
will be considered89. The basic objectives of this section is to quantify the
effect of controllable series reactances on power systems and to find a simple,
but accurate method for multi-objective placement of TCSC based on
sensitivity analysis for enhancing static performance of power systems. This
section presents a new multi-objective placement technique of TCSC based on
sensitivity analysis for:
1) Maximum relief of network loading (or congestion),
2) Power flow control,
3) Maximum reduction in active and reactive losses in a particular line, and
4) Maximum reduction in active and reactive power loss in the entire
network.
The OPF with FACTS constraints is used to show the validity of the
placement technique to relief network congestions. The preliminary
mathematical modeling of a system with TCSC, and a general form of static-
sensitivity analysis are first considered then the placement criterion will be
presented. A case study is then presented.
7.8.1 Preliminary mathematical modeling
The power system AC power flow model can be found in any power
system analysis text book for example. The fundamental model is based on
the system section shown in Fig. 7.15. the symbols in that figure are defined
as,
89
EL-Shimy M. Multi-objective Placement of TCSC for Enhancement of Steady-State
Performance of Power System. Scientific Bulletin - Faculty of Engineering - Ain Shams
Uni. 2007;42(3):935 - 50
287
Table 7.16: Primary applications of some major FACTS devices
FACTS devices
Main Subjects
Problems
Corrective
STATCOM
TCPAR
action(s)
TCSC
UPFC
SSSC
TCR
SVC
TSC
Supply Q
at heavy
loading
voltage
Low
Reduce line
Xl
voltage at
loading
High
Absorb Q
low
Voltage limits
contingenc
following
voltage
High
Absorb Q
a
y
contingenc
following
voltage
Low
Supply Q
a
y
overloadin
Transmiss
ion circuit
Increase the
Thermal
limits
transfer
g
limit
distribution over
Adjust
parallel line
reactances
Power
Load flow
Adjust
phase angles
Load flow
reversal
Adjust
phase angles
transmissi
on power
Limited
Stability
Reduce line
Xl
Limit the
High short circuit
short circuit
Short circuit
currents
current
power
AND
increase line
Xl
288
Fig. 7.15: Power flow model; (a) Single-line diagram; (b) Detailed
representation in terms of network elements.
bsi: Susceptance of all shunt devices connected to bus i (e.g. capacitor or

inductor banks)
yij = 1/zij = yji = gij + j bij: series admittance of line m connecting bus i
and bus j.
ysij = ysji = gsij + j bsij: shunt admittance of line m.
Vi: bus voltage magnitude of bus i.
i: bus voltage phase angle of bus i.
m: line index.
SGi: apparent power generated at bus i = PGi + jQGi.
SDi: apparent power demanded from bus i = PDi + jQDi.
The active and reactive power injected to bus i are given by,
Pi  PGi  PDi V i 2 g ii V i  V j  g ij cos  ij  bij sin ij  (7.43)

j K (i )
Qi  QGi  Q Di  V i 2bii V i  V j  g ij sin  ij  bij cos  ij  (7.44)

j K (i )
289
where K(i) is the set of all buses connected to bus i; ij = i - j;
g ii  g
jK ( i )
sij  g ij ; bii  ( b
jK ( i )
sij  bij )  bsi

The active and reactive power flow on line m from bus i to bus j are given by,
Plm  Pij V i 2 g ij  g ijV iV j cos ij  bijV iV j sin ij (7.45)
Qlm  Qij  V i 2 (bsij  bij )  bijV iV j cos ij  g ijV iV j sin ij (7.46)
The active and reactive power losses on line m are given by,
PLm  Pij  Pji  g ij (V i 2 V j2 )  2g ijV iV j cos ij (7.47)
Q Lm  Qij  Q ji  V i 2 (bsij  bij ) V j2 (bsij  bij )  2bijV iV j cos ij (7.48)
The network active and reactive power losses are given by:
nl nl
S L  PL  jQ L   PLm  j  Q Lm (7.49)
m 1 m 1
where nl is the number of network lines.
Now, consider a line with a TCSC as shown in Fig. 7.16. The TCSC can be
defined as capacitive reactance compensator which consists of a series fixed
capacitor (FC) bank shunted by a thyristor-controlled reactor (TCR). This
structure is required for providing a smoothly variable series capacitive
reactance. See Fig. 7.17. The line power flow, line losses are obtained by
modifying equations (7.45) to (7.48) by replacing bij by (bij+bclm) where bclm
is the susceptance of the TCSC on line m, and equals to the inverse of the
TCSC reactance xclm. This is the static-reactance model of a line with TCSC.
290
Fig. 7.16: A line with TCSC
Fig. 7.17: Thyristor-Controlled Series compensation or capacitor (TCSC)
7.8.2. General form of static-sensitivity analysis

The static power flow equations (7.43) and (7.44) can be written in a
generalized vector form as,
F (X ,U , D )  0 (7.50)
where F is a vector function; X is the dependent or the state variable vector;

U is the independent, input, or control variables; D is the demand variables.
If the system is consisted of N buses such that buses 2 to m are PV-
buses, buses m+1 to N are PQ-buses, and bus 1 is the slack bus then,
X T   2  N V m 1 V N  (7.51)
U T  PG 2 PGm V 2 Vm x cl 1 x c lnl  (7.52)
D T   PD 1 PDN QD 1 Q DN  (7.53)
It should be noted that the control vector U may contain other variables
such as switch status of switched- capacitor or inductor banks, phase-angle of
ULTC or TCPAR. The solution of (7.50) is the state vector X from which the
output variables can be calculated. These output variables include active and
reactive line power flow, active and reactive line power loss, system active
and reactive power loss, and output power of slack bus generator. All the
291
desired output variables form the output vector W. For a given operating
point (Xo, Uo, Do), the output vector Wo can be determined as a function of
that operating point. This function takes the general form,
W o
 (X o ,U o , D o ) (7.54)
Applying the first-order Taylor expansion to (7.54) for a change U in the

control vector and linearizing (i.e. 1st order sensitivity). This results in the
following equation,
W  X X  U U  D D (7.55)
Assuming that D = 0 and multiplying the 1st term by U/U, equation

(7.55) becomes,
X
W   X U  U U
U
 X 
  X  U  U
 U 
   X K XU  U  U
 K W U U (7.56)

where KXU is state-to-input sensitivity matrix, and KWU is the output-to-input
sensitivity matrix.
The state-to-input sensitivity matrix KXU is obtained from the 1st order
sensitivity of equation (7.50) with D = 0. This results in the following
equations
K XU  FX1FU (7.57)
K UW  X FX1FU  U (7.58)
Equation (7.56) represents a general form of output-to-input static-sensitivity

analysis. This general formulation will be used as the core for the multi-
292
objective placement of TCSC in power system for steady-state performance

enhancement.
7.8.3 TCSC Placement Problem

The selection of input and output variable depends on the study being
carried. Consider an N-bus system with win nl-lines such that a TCSCs are
placed on all lines and let the vector Xcl be the vector of TCSC reactances i.e.
X clT  U T  x cl 1 x clm x c lnl  (7.59)
The output vector W takes the form,
W T  | Pl 1 | ... | Plnl | | Ql 1 | ... | Qlnl | PLl 1 ... PL lnl QLl 1 ... Q L lnl PL Q L  (7.60)
The modulus of line active and reactive power flow is taken for
avoiding the error and the misunderstanding of the sensitivity results. For
example, if the modulus is not taken and the sensitivity of power flow on line
m’ due to placement of a TCSC on line m is negative, this will give one of the
following meanings:
 The effects of the TCSC is to reduce the flow on line m’, or

 The flow on line m’ is increased in opposite direction i.e. from bus j to
bus i if the sensitivity is calculated for flow on line m’ from bus i to
bus j.
As the modulus of line active and reactive power flow is taken, a negative
sensitivity will only mean that the flow on line m’ is reduced due to
placement of TCSC on line m, and this flow is increased with positive
sensitivity.
The output-to-input static-sensitivity is obtained using the general form
(7.56) where the matrices required in the calculation of the output-to-input
static-sensitivity matrix KUW are explained in the following. The Jacobian
matrix FX is the Newton-Raphson power flow standard Jacobian matrix and
takes the form,
293
 P    
 Q    F 
X   (7.61)
   V 
The Jacobian matrix FU takes the form:
 P 
 Q    FU  X l 
 
 P X cl 

    X cl  (7.62)
 Q X cl 
A general element in the P/Xcl matrix (and similarly Q/Xcl) takes

the form Pi/xclm. The derivatives Pi/xclm and Qi/xclm are obtained by
differentiating equations (7.43) and (7.44) after modifying them due to the
existence of the TCSC as previously described. These derivatives takes the
form:
Pi
 
 lm V i 2 V iV j cos  ij  lmV iV j sin  ij

 for i  slack, and m  K (i ) (7.63)
X clm 
 0 Otherwise
Q i
 
 lm V i 2 V iV j cos  ij   lmV iV j sin  ij

 for i  K (PQ ), and m  K (i ) (7.64)
X clm 
 0 Otherwise
Where K(PQ) is the set of all PQ-buses.
g lm 2r (x  x clm )
 lm   2 lm lm (7.65)
x clm (rlm  (x lm  x clm )2 ) 2
blm rlm2 -(x lm  x clm )2

lm   (7.66)
x clm (rlm2  (x lm  x clm )2 )2
z lm  rlm  jx lm (7.67)
294
The vectors X and U takes the form:  X  W X and U  W U

where the input vector U is defined in equation (7.59), the output vector W is
defined in equation (7.60), and the state vector X is defined in equation
(7.51). Theses derivatives can be easily found by direct differentiation similar
to that performed in equations (7.63), and (7.64).
With series compensation the overall effective series transmission
impedance from the sending end to the receiving end can be arbitrarily
decreased thereby influencing the power flow on the entire lines causing
redistribution of the power flow all over the network. A basic constraint on
placement of TCSC is that the placement will not cause transmission
congestions (line overload); a situation that can initiate a cascaded failure
event. Therefore, this reduces system security and reliability. Congestions
also results in reduction in the competition opportunity in deregulated
electricity markets, load shedding, and finally system collapse. One way to
avoid such situation is to use the sensitivity of line real power flow
performance index (PPI) to line power flow as restraining factor on
placement of TCSC on electrical networks90. The PPI itself can be used to
describe the severity of system loading during normal and contingency
conditions91. This PPI is given by,
2n
   Plmo 
nl
PPI    lm   max  (7.68)
m 1  2n   Plm 
where Plmmax is the rated continuous capacity of line m; n is an exponent; lm

is a real non-negative weighting factor which may be used to reflect
importance of line m; Plmo is the base-case power flow on line m.
The value of PPI is small when all network lines normally loaded (i.e.
significantly below their thermal limits), and its value increases as the
network lines loading increases. The PPI becomes high when there is a single
major congestion or a number of small line loading violations. The inability
to discriminate between a single major congestion and a number of small line
90
S.N. Singh, A.K. David, “Optimal Location of FACTS Devices for Congestion
Management”, Electric Power System Research, Vol. 58, 2001, pp. 71-79.
91
G.C. Ejebe, and B.F. Wollenberg, “Automatic Contingency selection”, IEEE, PAS, Vol.
98, No. 1, 1979, pp. 92-104.
295
loading violations is called masking effect. Because a single major congestion

may have severe effect on system security as compared to a large number of
small violations, the masking effect should be omitted. This is can be
achieved through proper selection of the exponent n. A value of n > 1 may be
a solution leading to reduction of the probability of occurrence of the
masking effect. Herein, n is taken equals to 2, and lm is taken equals to unity
for all lines (equal importance).
The sensitivity of line real power flow performance index PPI to line
power flow as affected by placement of TCSC on line m’ is represented by
the index  Plcm’ which is defined by,
4
PPI nl
 1  Plm
 lcm ' 
P
 3
| Plmo |  max  (7.69) 
x cl ' x cl ' 0 m 1  Plm  x clm '
The value of  P is anticipated as follows

 A positive value of  P indicates that the network loading increases with
the considered location of the TCSC,
 A negative value of  P indicates that the network loading decreases
with the considered location of the TCSC.
Hence, for maximization of network loading relief (or maximization of

network congestion relief) TCSC should be placed on the location
corresponding to the most negative value of  P. Although calculation of  P is
initiated to represent a basic constraint on the placement of TCSC on
electrical networks, the most negative value of  P itself satisfy the objective
of maximization of network congestion relief. This can be proven using a
network with severe congestions such that the traditional optimal power flow
OPF algorithms cannot relief this congestion without load curtailment. Then,
the optimal placement of TCSC is implemented on the system based on the
most negative  P-criteria and a modified OPF algorithm incorporating FACS
devices with power flow control constraints92 is utilized to show the effect of
the TCSC on the relief of severe network congestions.
92
G Shaoyun, and T S Chung, “Optimal Power Flow Incorporating FACTS Devices with
Power Flow Control Constraints”, Electrical Power & Energy Systems, Vol. 20, No. 5,
1998, pp. 321-326
296
By inspecting equation (7.68) that was extensively used in reliability,

security, and stability studies, it is clear that the line loading is represented by
its maximum (rated) active power flow which is far from the reality. This is
because the actual limits of a line is given as either rated current flow (Ilmmax),
or rated apparent power flow (Slmmax). Hence, for sack of accuracy an
apparent power performance index (SPI) is constructed instead of the PPI.
The SPI takes the form,
2n
 lm   Plmo  Q lmo 
2n
   S 
nl nl 2 2
SPI    lm   lmo max 
    (7.70)
m 1  2n   S lm  m 1  2n   
max
S lm
 
The sensitivity of line apparent power flow performance index SPI to

line power flow as affected by placement of TCSC on line m’ is represented
by the index  Slcm’ which is given by (for n = 2, and lm = 1):
4
SPI nl
 1   Plm Q lm 
 lcm ' 
S
 2
S lmo  max   | Plmo |  | Q lmo |  (7.71) 
x cl ' x cl ' 0 m 1  S lm   x clm ' x clm ' 

The value of  is treated in the same manner of treating the value of
S
 P. In this analysis both  S and  P are used and compared. In doing so, in the
calculation  P the line active power limit is taken equals to its apparent power
limits as widely used based on the assumption that the line power flow is
mostly active power. But this justification is not general.The objectives of
the TCSC placement in this analysis are, 

 Maximum relief of network loading (or congestion),
 Power flow control,
 Maximum reduction in active and reactive losses in a particular line,
and
 Maximum reduction in active and reactive power loss of a power
network.
The proposed procedure to solve the stated multi-objective placement

problem is to construct a set of placement solutions that satisfy each of the
297
objectives and then an overlapping approach is used for identifying the

solution(s) that satisfy all the constraints. This procedure will be illustrated in
the results section.
7.8.4 STUDY SYSTEM

The multi-objective placement and sensitivity analysis are performed
on the six-bus system shown in Fig. 7.18. The system consists of 6-buses, 4-
generators, and 7-lines. Line indexes (m) are shown within a circle. The
system bus-data and base-case AC power flow, line-data, base-case line
power flow and loses, and generator-cost functions and generator limits are
shown in Tables 7.17, 7.18, 7.19, and 7.20 respectively, all in p.u on 100-
MVA, 138-kV base. A TCSC is assumed to be placed on all lines such that -
0.7xlm  xclm  0 to avoid overcompensation.
Fig. 7.18: Six-bus Study System
Table 7.17: Bus Data and Base-Case AC Power Flow

Bus No. Type V  (deg.) PG QG PD QD
1 Slack 1.020 0 0.22 0.14 0.70 0.1
2 PV 1.040 2.734 0.99 0.21 0.45 0.1
3 PV 1.010 -0.364 0.86 0.25 0.50 0.1
4 PV 1.030 2.695 1.30 0.09 0.30 0.1
5 PQ 1.009 -0.313 0 0 0.35 0.1
6 PQ 0.953 -2.919 0 0 1.00 0.3
298
Table 7.18: Line Data

Line index, Line
rlm xlm blm Slmmax
m From bus To bus
1 1 2 0.04 0.08 0.02 1.0
2 1 5 0.08 0.16 0.04 1.0
3 2 4 0.04 0.08 0.02 1.0
4 3 5 0.08 0.16 0.04 1.0
5 3 6 0.08 0.16 0.04 1.0
6 4 5 0.08 0.18 0.08 0.5
7 4 6 0.08 0.16 0.04 1.3
Table 7.19: Base-Case Line Power Flow and Loses

Line index, From To
Pij Qij Pji Qji PLm QLm
m bus i Bus j
1 1 2 -0.537 0.014 0.548 -0.013 0.01110 0.02220
2 1 5 0.057 0.023 -0.057 -0.064 0.00040 0.00080
3 2 4 -0.008 0.123 0.009 -0.143 0.00067 0.00130
4 3 5 -0.001 -0.011 0.001 -0.030 0.00001 0.00001
5 3 6 0.361 0.165 -0.348 -0.178 0.01293 0.00259
6 4 5 0.301 -0.03 -0.294 -0.007 0.00684 0.01540
7 4 6 0.690 0.159 -0.652 -0.122 0.03839 0.07680
Losses 0.07034 0.14240
Table 7.20: Generator-Cost Functions and Generator Limits*

Gen. on    PGmin PGmax
2
bus $/h $/MWh $/MW h MW MW
1 373.5 7.62 0.0020 10 100
2 388.9 7.57 0.0013 20 280
3 194.3 7.77 0.0019 20 200
4 253.2 7.84 0.0013 20 300
*Generator cost functions takes the form C( PG )    PG  PG2 $/h
7.8.5 Implementation and Results

It should be noted that some of the lines will be rejected from
placement list based on the values of  in equation (7.69) or (7.71). The
calculation of  P requires the calculation of line active power flow sensitivity
to TCSC location and the calculation of  S requires the calculation of both
line active power flow and line reactive power flow sensitivities to TCSC
location.
The line active power flow sensitivity matrix is shown in Table 7.21.
The values can be used as a guide for TCSC placement for line active power
flow control as shown in Table 7.22.
299
Table 7.21: Line active power flow sensitivity matrix (Plm/xclm’)

Loc. 1 Loc. 2 Loc. 3 Loc. 4 Loc. 5 Loc. 6 Loc. 7
Line 1 0.9375 0.2875 -0.125 0 0.45 -0.8875 -0.75
Line 2 0.46875 0.15 -0.05625 0 0.225 -0.44375 -0.375
Line 3 0.975 0.3 -0.1375 0 0.475 -0.925 -0.775
Line 4 0.11875 0.0375 -0.01875 -0.00625 0.61875 0.28125 0.91875
Line 5 0.11875 0.0375 -0.01875 -0.00625 0.61875 0.28125 -1.05625
Line 6 -0.32222 -0.1 0.038889 -0.005556 0.377778 0.677778 -0.63333
Line 7 -0.13125 -0.04375 0.0125 0 -0.66875 -0.3 1.1
Table 7.22: Priority sets of placemat of TCSC for line active power flow
control based on line active power flow sensitivity matrix of Table 7.21
Location of TCSC Placement for
Line, m
Increase active power flow Decrease active power flow
1 1, 5, 2 6, 7, 3
2 1, 5, 2 6, 7, 3
3 1, 5, 2 6, 7, 3
4 7, 5, 6, 1, 2 3, 4
5 5, 6, 1, 2 3, 4
6 6, 5, 3 7, 1, 4
7 7, 3 5, 6, 1, 2
The priority locations in Table 7.22 for placement of TCSC are

arranged in priority sets for TCSC placement i.e. from the most effective
locations to the least effective locations for placement of TCSC for line
active power flow. The diagonal elements of the sensitivity matrix of Table
7.21 represents the self-sensitivity to active power flow i.e. sensitivity of line
active power flow as a result of placement of a TCSC on that line, while the
off-diagonal values represent the mutual-sensitivity i.e. i.e. sensitivity of line
active power flow as a result of placement of a TCSC on another line.
It is clear from Table 7.21 that the active power flow on a line can be
increased, or decreased, or not affected significantly by placement of TCSC
on another specified line. For example, the most effective TCSC location for
reduction of active power flow on line 1 is placement of TCSC on line 6 as
the value of sensitivity is the most negative index while placement of TCSC
on line 1 results on the maximum increase of active power flow on line 1 as
affected by the TCSC as the value of sensitivity is the most positive index. It
is also shown that placement of TCSC on line 4 has no significant effect on
the active power flow on line 1. An interesting situation appear with
placement of TCSC, for example, on line 3, the active power flow sensitivity
300
on line 3 as affected by TCSC on that line results in active power flow

reduction (this verified through full AC power flow). Hence, it is not a
general rule that placement of a series capacitor on a line within a network
increases the active power flow on that line, actually the full AC power flow
shows that the reactive power flow on line 3 increases by placing a TCSC on
that line, this is clarify the importance of using the most negative  S-criteria
instead of most negative  P-criteria as a constraint explained earlier for
enhancing the system static security. In addition, the full AC power flow
shows that the placement of TCSC has insignificant effect on bus voltage
magnitudes.
The value of  P as affected by TCSC location is shown in Table 7.23
which shows that placement of TCSC either on line 5 or line 6 increases
network loading ( P > 0) with maximum increase associated with placement
of TCSC on line 5. Therefore, both lines are rejected from the TCSC
placement set.
Table 7.23: The value of  P as affected by TCSC location

Location, m 1 2 3 4 5 6 7
-0.00249
-0.00179
-0.00272
-0.31602
0.18693
0.13766
-0.0053
P
Placement priority 2 4 5 3 Rejected Rejected 1 (Opt.)
The optimal placement of TCSC for network loading relief based on the
most-negative P-criteria is a TCSC placed on line 7. The effectiveness of
placement of TCSC on other lines for network loading relief is arranged as a
priority list as shown in the table. Hence, one of these location (i.e. placement
on lines 1, 2, 3, 4, 7) will be the selected location if it satisfy all the desired
objectives. The correctness of the rejections will be verified by calculation
the performance index PPI for separate placement of TCSC on line 7 and line
5 and simulating the changes in PPI through multiple loadflow as the TCSC
reactance changes all over its range. Fig. 7.19 shows these changes in the
PPI, it is clear from that figure that placement of a TCSC on line 7 decreases
the PPI all over the range of variation of the TCSC and the opposite occurs
with placement of TCSC on line 5.
301
Fig. 7.19: Changes in the PI through Multiple Loadflow
The line reactive power flow sensitivity matrix is shown in Table 7.23
and can be used as a guide for TCSC placement for line reactive power flow
control this shown in Table 9 arranged in priority sets for TCSC placement
i.e. from the most effective locations to the least effective locations for
placement of TCSC for line active power flow.
Table 7.23: Line reactive power flow sensitivity matrix (Qlm/xclm’)

Line Loc. 1 Loc. 2 Loc. 3 Loc. 4 Loc. 5 Loc. 6 Loc. 7
1 0.6125 0.15 -0.075 -0.0125 0.25 -0.4875 -0.4125
2 -0.23125 0.1125 0.025 -0.01875 -0.10625 0.24375 0.18125
3 0.5 0.1625 1.8 0 0.25 -0.4625 -0.3875
4 -0.05625 0.08125 0.0125 -0.0375 -0.3125 -0.1625 0.53125
5 -0.05625 -0.01875 0.00625 0 0.13125 -0.1375 -0.08125
6 -0.12222 0.0388889 0.011111 0.016667 0.144444 0.383333 -0.24444
7 0.04375 0.0125 -0.00625 0 -0.3875 0.10625 -0.33125
Table 7.24: Priority Sets of Placemat of TCSC for line reactive power flow
control
Based on Line reactive Power Flow Sensitivity Matrix of Table 7.23
Line, m
Increase reactive power flow Decrease reactive power flow
1 1, 5, 2 6, 7, 3, 4
2 6, 7, 2, 3 1, 5, 4
3 3, 1, 5, 2 6, 7
4 7, 2, 3 5, 6, 1, 4
5 5, 3 6, 7, 1, 2
6 6, 5, 2, 4, 3 7, 1
7 6, 1, 2 5, 7, 3
302
Table 7.25: The value of  S as affected by TCSC location

Location, m 1 2 3 4 5 6 7
-0.00825
-0.00024
-0.00213
-0.34721
0.00180
0.19112
0.15468
S
Placement
2 4 Rejected 3 Rejected Rejected 1 (Opt.)
priority
Based on the values of the line active- and line reactive- power flow
sensitivity matrices, the sensitivity of line apparent power flow performance
index SPI to line power flow as affected by placement of TCSC on line m’ is
calculated and these values and TCSC placement priority are shown in Table
7.24 and 7.25. Based on the most-negative  S criteria an additional line is
rejected from TCSC placement list which is line 3. It is logical that the results
obtained from  S criteria are more accurate than those obtained from  P
criteria. Therefore, the final TCSC placement list consists of 4 locations on
lines {7, 1, 4, 2} arranged according to their effectiveness in relieving
network overloads (congestions). Based on that, the rejected locations are
identified in Table 7.22 and Table 7.24 by non-bold numbers. In addition,
from this point the rejected locations will not be mentioned in the following
placement lists.
The line- active power loss and reactive power loss sensitivity matrices
are shown in Table 7.26 and Table 7.27 respectively. Those matrices can be
used as a guide for TCSC placement for line active and reactive power loss
reduction as shown in Table 7.28. Hence, the overlapping placement
decisions of active and reactive power loss reduction represent the candidate
locations of TCSC for line apparent power reduction.

Table 7.26: Line active Power loss Sensitivity Matrix (PLlm/xclm’)
Line Loc. 1 Loc. 2 Loc. 3 Loc. 4 Loc. 5 Loc. 6 Loc. 7
1 0.04 0.0125 -0.005 0 0.02 -0.0375 -0.03125
2 0.003125 0.001875 0 0 0.00125 -0.00188 -0.00188
3 0.005 0.00125 0.01875 0 0.0025 -0.005 -0.005
4 0 0 0 0 0.00125 0 0.000625
5 0.005 0.001875 -0.01937 0 0.04 0.0125 -0.06063
6 -0.01444 -0.004444 0.001667 0 0.017778 0.031667 -0.02833
7 -0.0125 -0.004375 0.00125 -0.28125 -0.07937 -0.02875 0.106875
303
Table 7.27: Line reactive Power loss Sensitivity Matrix (QLlm/xclm’)

Line 1 0.6125 0.15 -0.075 -0.0125 0.25 -0.4875 -0.4125
Line 2 -0.23125 0.1125 0.025 -0.01875 -0.10625 0.24375 0.18125
Line 3 0.5 0.1625 1.8 0 0.25 -0.4625 -0.3875
Line 4 -0.05625 0.08125 0.0125 -0.0375 -0.3125 -0.1625 0.53125
Line 5 -0.05625 -0.01875 0.00625 0 0.13125 -0.1375 -0.08125
Line 6 -0.12222 0.0388889 0.011111 0.016667 0.144444 0.383333 -0.24444
Line 7 0.04375 0.0125 -0.00625 0 -0.3875 0.10625 -0.33125
Table 7.28: Priority Sets of Placemat of TCSC for line active, reactive, and
apparent power loss reduction
Line, m Decrease line Decrease line Decrease line
active power loss reactive power loss apparent power loss
1 7 7, 4 7
2 7 1, 4 NO
3 7 7 7
4 NA* 1, 4 NO
5 7 7, 1, 2 7
6 7, 1, 2 7, 1 7, 1
7 4, 1, 2, 7 NO
* NA: Not Applicable
** NO: No Overlapping (in this case one line power loss can be reduced
via a sigle TCSC)
Table 7.29: Network power loss sensitivity matrices and placement priority
sets
PL
0.0005 0.00125 0.0175 -0.00063 -0.00625 -0.00278 -0.00437
Sensitivity
PL
Priority Rejec Rejec Rejec 2 Rejec Rejec 1 (Opt.)
sets
QL
-0.2875 -0.00625 0.0125 0 -0.175 -0.08889 -0.51875
Sensitivity
QL Rejec Rejec Rejec Rejec
2 3 1 (Opt.)
Priority sets
SL Rejec Rejec Rejec Rejec
NO NO 1 (Opt.)
Priority sets
The network active, and reactive power loss sensitivity matrices are
shown in Table 7.29. Those matrices can be used as a guide for TCSC
placement for network power loss reduction. Hence, the overlapping
placement decisions of active and reactive power loss reduction represent the
candidate locations of TCSC for line apparent power reduction.
304
Based on the presented results, the optimal placement of TCSC

network loading relief maximization and network apparent loss minimization
is a TCSC is placed on line 7. Tables 7.22 and 7.24 gives a priority list for
TCSC placement for line active and reactive power flow control which is
very valuable, especially, in deregulated electricity markets where
controllable power flow paths are an important requirements for
maximization of electricity markets deregulation. Although, placement of
TCSC on line 7 minimize the overall network power loss, Table 7.28 lists a
priority sets for placement of TCSC for power loss reduction on a particular
line.
To evaluate the proposed technique in the congestion relief reduction,
the load parameters of Table 7.17 are doubled and a traditional OPF
algorithm (with no load curtailment option) is applied to solve the system
with no TCSC installed. The results in this case is that the OPF could not get
an acceptable operating point for the system because line 5 is congested and
the system operating cost is 6371.1 $/hr. To solve this problem a TCSC is
placed on two locations separately, the first location on line 7 (the optimal
location) and on line 3 (the rejected location with lowest positive  S). The
FACTS-OPF algorithm of is applied and the results show that with a TCSC
placed on line 7, an acceptable operating point is obtained with no
congestions and the system operating cost is 6368.58 $/hr. With a TCSC
placed on line 3, the congested line remains congested. By forcing the
reactance of the TCSC on line 3 to a value of -0.1xl3, additional congestion is
occurred on line 6 and the system violations increases. These results prove
that the optimal placement of TCSC is on line 7 and also prove the accuracy
of the most-negative  S-criteria over the most-negative  P-criteria.
305
306
Chapter 8
Overview, Modeling, and Performance Analysis of Grid-
connected Wind energy Sources
8.1 Introduction
Generally, the electric power is produced through an energy conversion
process in which a primary energy source is converted into electric power. As
illustrated in Fig. 8.1, the primary energy resources can be classified
according their capability of replenishment (or renewal) into two broad
categories; non-renewable and renewable energy resources93. The
conventional or non-renewable energy resources are available on the earth in
limited quantity and will be vanished in the future. On the opposite side,
renewable resources are natural sources of energy that are continually
renewed, or replenished by nature, and hence will never run out.
Energy plays an important role in the national security of any given
country as a fuel to power the economic engine 94. Therefore, the energy
security or in other words the access to cheap energy and availability of
energy resources for the future is one of the major challenges in the energy
sector. Non-renewable energy resources are distributed in an uneven way
throughout the world while many types of renewable energy resources such
as wind and solar energies are available for all locations on the earth.
Therefore, renewable energy resources can contribute in enhancing the
worldwide energy security, reducing the energy threat, and crisis. In addition,
renewable energy sources are sustainable due to their low pollution levels in
comparison with conventional sources of energy as well as their natural
availability. It should be noted that the phrase renewable and the phrase
sustainable are different. “Renewable energy includes all those natural
sources which can be replaced by natural ecological cycles. On the other
hand sustainable energies are those energies that help this world in meeting
its present needs without depleting ability of serving next generation’s needs.
93
Maczulak, A. (2010). Renewable Energy: Sources and Methods. New York: Infobase
Publishing, 2010
94
Sovacool, B. K.; Brown, M. A. (2010). Competing dimensions of energy security: An
international perspective. Annual Review of Environment and Resources, 35, 77-108.
307
Both kinds of energy leaves approximately zero effect on environment and

remain in same form throughout existence on this universe” 95.
Fig. 8.1: Conventional and renewable energy resources
As an electrical power source, renewable energy sources suffer from some

significant problems in comparison with conventional energy sources 96 (see
chapter 1). These problems set a major challenge on the reliable use of
renewable energy resources as electrical power sources especially when they
significantly contribute by a large amount in the energy mix of power
systems. A major change in the energy mix (see chapter 1) of power systems
caused by the integration of large amounts of variable renewable energy
sources raises a number of challenges regarding grid stability, reliability,
security, power quality and behavior during fault conditions. Consequently,
elaboration of specific technical requirements or grid codes for the
connection of large amounts of variable renewable energy has been
constructed. The grid codes stipulate that these energy sources should
95
Difference between renewable and sustainable energy,
http://www.solarpowernotes.com/difference-between-renewable-and-sustainable-
energy.html#.UnAsrlPnURQ
96
Hand, M.M.; Baldwin, S.; DeMeo, E.; Reilly, J.M.; Mai, T.; Arent, D.; Porro, G.;
Meshek, M.; Sandor, D. (Eds. 4 Vols). (2012). Renewable Electricity Futures Study
NREL/TP-6A20-52409. Golden, CO: National Renewable Energy Laboratory
308
contribute to power system operation and control in a similar way the

conventional generating systems do 97,98.
Given that it is not techno-economically feasible or may be impossible
to store massive amounts of electric power, the operational problems are
mostly related to the characteristics of the renewable energy resource. Some
major renewable energy resources such as wind and solar (shown in Fig. 8.1)
are inherently variable and intermittent. Therefore, despite of the capability
of predicting them, such grid-connected variable and intermittent energy
sources cannot provide the main operational requirements of power systems
such as dispatch, unit commitment, security, and reliability. Consequently,
significant R&D is needed for solving these operational problems. It is
worthy to be mentioned that some renewable resources such as biomass,
biofuel, and hydropower with reservoir are characterized by low variability
and high dispatchability. Therefore, such energy sources have better
operational characteristics in comparison with other sources.
Essential grid code requirements are related to frequency, voltage and
behavior in case of grid faults. The most common requirements include 99
active power control, frequency control, frequency and voltage acceptable
ranges, voltage control, voltage quality, fault ride-through (FRT) capability,
power plant modeling, and communication and external control. The main
aim of the requirements is to ensure that renewable power plant do not
adversely affect the power system operation and control with respect to
security of supply, reliability, and power quality. The fulfillment of the grid
code requirements is mostly related to the technological advances of the
renewable power generators; however, the fulfillment of the operational
requirements need more sophisticated efforts because these requirements are
mostly related to the characteristics of the primary renewable resources.
97
Tsili, M.; Papathanassiou, S. (2009). A review of grid code technical requirements for
wind farms. IET Renewable Power Generation, 3(3), 308-332.
98
onshore and offshore DFIG‐based wind farms. Wind Energy.
http://dx.doi.org/10.1002/we.1575, Dec. 6, 2012
99
Holttinen, H.; Bettina Lemström, V. T. T.; Meibom, F. P.; Bindner, H. (2007). Design
and operation of power systems with large amounts of wind power. State-of-the-art report.
VTT (Espoo))(= VTT Working Papers, 82). URL: http://www. ieawind.
org/AnnexXXV/Publications W, 82
309
Probably the hydrogen storage and hydrogen technologies will play a key
role in the future of variable renewable energy sources100.
In chapter 1, an overview of various energy sources and their operational
characteristic is given. Both renewable and non-renewable energy sources are
considered. In this chapter, detailed overview of the technological aspects of
wind and solar-PV energy production technologies will be presented. In
addition, the grid-interconnection requirements (i.e. the grid codes) for
connecting these energy sources to the grid will be explained considering
various international codes.
8.2 Wind Energy Conversion Technologies (WECTs)

For thousands of years, man has utilized wind energy to sail ships,
grind grain and pump water. The kinetic energy in the wind can be converted
to a mechanical power using windmills. Historically, windmills are created
from thousands of years for direct use of mechanical power in applications
such as grinding grains, and pumping water. Windmills were in use in ancient
Egypt fifty centuries ago for grinding flour and water pumping windmills
have been recorded in Kautalya’s Arthashastra, indicating their existence in
India from 400 BC. Recently as the late 20 th century, the wind power is used
in Europe for electric power production through double energy conversion
systems called wind turbines.
Generally, as shown in Fig.
8.2, two energy conversion
processes comprise a wind
turbine. The first process is the
conversion the wind kinetic
energy to a mechanical energy
Fig. 8.2: Wind energy conversion processes
which is converted to electrical
energy in the second energy
conversion process through an electricity generator.
As shown in Fig. 8.3 to 8.10, wind turbines can be classified according
to many indicators 101,102 such as the axis of rotation (Fig. 8.3), airflow path
100
Parfomak, P. W. (2012). Energy storage for power grids and electric transportation: A
technology assessment. Congressional Research Service, Tech. Rep. R, 42455
101
EL-Shimy, M. (2013). Probable power production in optimally matched wind turbine
generators. Sustainable Energy Technologies and Assessments, 2, 55-66
310
relative to the turbine rotor (Fig. 8.4), turbine capacity (Fig. 8.5), drivetrain
condition (Fig. 8.6), power supply mode (Fig. 8.7), location of the wind
turbine (Fig. 8.8), type of generator (Fig. 8.9), and operating concept (Fig.
8.9).
Fig. 8.3 shows the classification of wind turbines based on the direction of
the axis of rotation i.e. the horizontal axis wind turbine (HAWT) and the
vertical axis wind turbine (VAWT). Due to the low aerodynamic efficiency
of the VAWT, the HAWT dominate the wind turbine industry and the
market.
Fig. 8.3: Classification of wind turbines according to axis of rotation
Fig. 8.4: Classification of HAWTs according to airflow path relative to the

turbine rotor
102
Spera, D. A. (2009). Wind Turbine Technology: Fundamental Concepts of Wind
Turbine Engineering. American Society of Mechanical Engineers
311
According to the airflow path relative to the turbine rotor, the HAWT can
be classified into upwind and downwind wind turbines as shown in Fig. 8.4;
the majority of HAWT is upwind.
The classification according to the turbine capacity is shown in Fig. 5.
Turbine capacities are designed according to the intended applications;
however, most recent designs are multi-
megawatt wind turbines for large-scale
wind farm applications.
Table 8.1 provides a brief comparison
between various WECTS. Although Fig.
8.9 and Table 8.1 shows a large variety of
generator/concept mix, a recent
technological status 103 shows the DFIG-
based VSWT present about 78% of the
wind power installed capacity, 12% for
Fig. 8.5: Classification of wind
full-scale converter based VSWT, 5% for turbines according to turbine
the limited variable speed concept, and 5% capacity
for the SCIG-based FSWT.
Fig. 8.6: Classification of wind

turbines according to drivetrain Fig. 8.7: Classification of wind turbines
conditions according to power supply mode
103
312
Fig. 8.8: Classification of wind turbines according to location of the wind

turbine
Fig. 8.9104: Classification of wind turbines according to generator types

and operating concepts
104
313
Table 8.1: WTG configurations and their comparison

Configuration Gen. Conv. Advantages Disadvantages
type size
Fixed speed operating concept (geared)
* No slip-rings/brushes * Gearbox related problems
* Simple and robust design * Low aerodynamic efficiency
* Low cost * High mechanical loading
NA
SCIG * High reliability and * High electric power
durability pulsations
* No reactive power capability
Semi-variable speed operating concept (geared)
* Improved speed range * Slip-rings/brushes related
* Reduced mechanical problems
stresses * Limited speed (above
Partial- * Reduced electric power the synchronous speed)
WRIG
scale pulsations * Higher power losses in the
* Improved aerodynamic rotor
efficiency circuit
* Gearbox related problems
Variable speed geared concept
* Control of active and * Limited speed range
reactive power depending
DFIG
* Improved aerodynamic on the converter size
Partial-scale
and * Slip-rings/brushes related

overall efficiency problems in partial-scale
BDFIG converter configurations
* Gearbox related problems
BDFRIG
* Control of active and * Gearbox related problems

Full-scale
PMSG reactive power

* Improved aerodynamic
and
overall efficiency
SCIG * Full speed range
* No brushes
Variable speed gearless conecpt
* Control of active and * Relatively complex control
reactive power
Full-scale
EESG * Improved aerodynamic

efficiency
* Full speed range
PMSG
FSWT Fixed Speed Wind Turbine. WRIG Wound-Rotor Induction Generator.

Limited Variable Speed Wind
LVSWT Turbine. DC Direct Current Generator.
VSWT Variable Speed Wind Turbine. BDFIG Brushless DFIG.
Brushless Double-Fed Reluctance Induction
PSC Partial-Scale Converter. BDFRIG Generator.
FSC Full-Scale Converter. PMSG Permanent Magnet Synchronous Generator.
Squirrel-Cage Induction
SCIG Generator. EESG Externally Excited Synchronous Generator.
DFIG Doubly-Fed Induction Generator.
314
Fig. 8.10 illustrates a typical configuration and main components of

geared HAWT105,106.
Fig. 8.10: A typical configuration and main components of geared HAWT
8.3 Modeling of Wind Turbine Generators (WTGs)

The list of symbols used in the modeling of the considered WTGs is
shown in Table 8.2. The models of various system components are
summarized here and they are based on107,108,109,110,111,112,113.
105
EL-Shimy, M. (2010). Alternative configurations for induction-generator based geared
wind turbine systems for reliability and availability improvement. Proc. of IEEE 14th
international middle-east power conference MEPCON2010. Cairo, Egypt.(pp. 617-623).
IEEE
106
Ciang, C. C.; Lee, J. R.; Bang, H. J. (2008). Structural health monitoring for a wind
turbine system: a review of damage detection methods. Measurement Science and
Technology, 19(12), 122001
107
Milano, F. (2010). Power system modelling and scripting. Springer.
108
Milano, F. (2008). Documentation for PSAT version 2.0.0, 2008.
109
EL-Shimy, M. (2010) Steady State Modeling and Analysis of DFIG for Variable-Speed
Variable- Pitch Wind Turbines. Ain Shams Journal of Electrical Engineering, 1, 179 - 89
110
Slootweg, J. G.; Kling, W. L. (2002). Modelling and analysing impacts of wind power
on transient stability of power systems. Wind Engineering, 26(1), 3-20
315
Table 8.2: List of Symbols

Pitch control gain Stator resistance (p.u)
Number of blades Rotor resistance (p.u)
Number of poles Rotor leakage reactance (p.u)
Power rating (MVA) Magnetizing reactance (p.u)
Pitch control time constant (s) Stator leakage reactance (p.u)
Number of machines that compose
Stator reactance =
the wind park
ηGB Gear box ratio Rotor reactance =
Air density (kg/m3) p Blade pitch angle (deg)
Turbine performance coefficient
Rated frequency (rad/s) cp
or power coefficient
Wind turbine angular speed (p.u)  Tip speed ratio
Generator rotor speed (p.u) Ar Area swept by the blades (m2)
Relative angle displacement of the
vbt Blade tip speed (m/s)
two shaft (rad)
Wind upstream the rotor i.e. The
Electrical torque (p.u) v
wind speed (m/s)
The mechanical power extracted
Mechanical torque (p.u) P
from the wind
Machine rotor inertia constant
Rated output mechanical power
(MWs/MVA)
Wind turbine inertia constant
uc Cut-in wind speed (m/s)
(MWs/MVA)
Shaft stiffness (p.u) ur Rated wind speed (m/s)
Furling or cut-out wind speed
Shadow effect factor uf
(m/s)
Stator flux bc Shunt capacitor conductance
Converter current Converter voltage
Generator active power Generator reactive power
Rotor active power on the rotor Rotor reactive power on the rotor
side side
Convertor active power on the grid Convertor reactive power on the
side grid side
Stator ( ) and Stator ( ) and
rotor ( ) voltages rotor ( ) currents
Voltage behind transient reactance
Voltage at generator terminals
d_q components
111
El-Shimy, M.; Badr, M. A. L.; Rassem, O. M. (2008, March). Impact of large scale
wind power on power system stability. In 12th International Middle-East Power System
Conference MEPCON 2008. Aswan, Egypt. (pp. 630-636). IEEE
112
113
Ghaly, N.; EL-Shimy, M.; Abdelhamed, M. Parametric study for stability analysis of
grid-connected wind energy conversion technologies. In 15th International Middle-East
Power System Conference MEPCON 2012. Alexandria, Egypt. (pp. 1-7). IEEE
316
8.3.1 Wind turbine model and its controls

The modeling of the pure mechanical items of the wind turbine is
independent of the generator configuration. Fig. 8.11 illustrates a general
schematic diagram of a geared grid-connected WECS. The output mechanical
of the WTG can be mainly controlled by either stall-control or pitch-control
of the turbine blades.
Fig. 8.11: Generic Grid-connected geared WECS
In the stall controlled turbines, the turbine blades are firmly bolted to the
turbine hub and the rotor speed is adjusted using a gearbox attached between
the turbine and generator shaft while the pitch angle is fixed at a value that
results in limiting the power output at high wind speed by blades stalling.
In pitch control, the stall angle of the blades is controlled via pitch angle
variations to limit the power output to a set value. Stall controlled turbines
are mainly implemented in the FSWT concept; however, the current
technological standard utilizes pitch-control for both FSWT and VSWT
concepts. The stall controlled turbines are not currently common due its
problems such as gearbox wear and blade design complications.
The considered model here is for pitch-controlled wind turbines which are
currently dominating the wind turbine industry. This is because this model is
adequate for the speed control of recent wind turbines. In this model, turbine
blades can rotate in order to reduce the rotor speed and output power in case
of high wind speeds. The angular position of the blades is called pitch
angle. The mechanical power extracted from the wind is a function of the
317
wind speed , the rotor speed and the pitch angle . The mechanical
power model can be approximated by
The tip speed ratio is the ratio between the blade tip speed and the
wind upstream the rotor . The tip speed ratio is defined by
(a)
(b)
Fig. 8.12: Wind turbine; (a) Power curve and operational regions of a
pitch controlled wind turbine; (b) Pitch angle control scheme
The curve is popularly approximated as follows
where
318
The power curve and the operational regions for a typical pitch-controlled
wind turbine are shown in Fig. 8.12(a). As shown, there are four operational
regions each of which has its own characteristics and limitations. In region 1,
the wind speeds are insufficient to run the turbine. Therefore, the output
power is zero. Region 2 starts when the wind speed exceeds the cut-in value
and extends up to the rated wind speed at which the rated output power is
produced. In this region, maximization of the power extraction is the main
concern. Therefore, the pitch angle is kept zero. This is achieved by an anti-
windup limiter in the pitch angle controller. In the third region, the rotor
speed and consequently the power extraction is limited to the rated power to
avoid overloading the turbine. This is achieved on the turbine level by using
the pitch angle control which adjusts the pitch angle to an appropriate
controlled value. In region 4 where the wind speeds reach dangerous limits,
the turbine is taken out of service for protection against mechanical damage.
A schematic diagram of a pitch-angle controller is shown in Fig. 8.12(b) and
it can be described mathematically by
where  is a function which allows varying the pitch angle set point only
when the difference exceeds a predefined value
depending on the operating concept of the WTG.
In the Fixed-Speed Wind Turbine (FSWT) concept the turbine drive a
Squirrel-Cage Induction Generator (SCIG) that is directly connected to the
grid expect during start-up where a soft-starter is used to minimize the startup
stresses on the system and the grid. In this operating concept,  should not
exceed 1% or 2% above the synchronous speed and the speed is kept constant
within this range. In the Variable-Speed Wind-Turbine (VSWT) concept, the
allowable speed range is high. For example it is 30 around the synchronous
speed in DFIG-based systems.
8.3.2 Shaft model

The consideration mechanical dynamics of the shaft depends on the
operational concept of the WTG. In FSWT concept, the shaft dynamics
should be considered due to its significant impact on the WTG performance
319
including power and voltage fluctuations. Therefore, popularly a two mass

model is adopted to represent the shaft in the FSWT concept. In the VSWT
concept, the available controls minimize or prevent the impact of the shaft
dynamics. Therefore, usually the shaft dynamics are not considered in the
modeling of the WTG and one mass model is adopted. In the two-mass model
(Fig. 8.13), the shaft dynamics are represented by,
where is the displacement of the two shafts, is the electrical torque,

and is the mechanical torque which can be represented by
Fig. 8.13: Two-mass representation of WTG shaft
A periodic torque pulsation can be added to Tt to simulate the tower

shadow effect. The shadow-effect frequency depends on the rotor speed t,
the gear box ratio ηGB, and the number of blades nblade. The torque pulsation
model is
320
In the VSWT concept, the Voltage Source Converter (VSC) controls can
effectively damp shadow effect modes and shaft oscillations. If the control is
efficient enough, the shaft can be considered rigid, i.e., = and can be
modeled using,
8.3.3 Modeling of Generators, and their controls

A. SCIG model and control
The equivalent circuit of a SCIG (with a single-cage) is shown in Fig.
8.14. In comparison with a conventional induction motor, the induction
generator currents are positive if injected into the network. The equations are
formulated in terms of the real (d) and imaginary (q) axes, with respect to the
network reference angle. Considering the third-order model of the machine,
the equations of the machine are as follows,
(a)
(b)
Fig. 8.14: SCIG; (a) Equivalent circuit of a SCIG; (b) dq-coordinates
321
The link between the voltages, currents, and state variables is modeled as
follows,
where is the stator reactance, ,

and . Note that is the rotor reactance
In the synchronously rotating reference frame, the link between the

network and the stator voltages of the machine is as follows,
The active and reactive power flow can be determined by
where is the fixed capacitor conductance which is connected to the

generator terminals. The value of this capacitor can be determined at the
initialization step to impose the desired initial bus voltage level.
In the FSWT concept, the only available control is the pitch angle
control and no additional control is provided to the system. The impact of the
pitch-controller SCIG-based FSWT is illustrated in Fig. 8.15. With zero
pitch-angle, the turbines produce its maximum power at a wind speed of 10
m/s while the output power is reduced to 0.725 (i.e. 27.5% reduction) when
the pitch-angle becomes five degrees. The characteristics shown in the figure
322
are obtained by simulating equation (1) using MATLAB SimPowerSystems

Toolbox.
(a) (b)
Fig. 8.15: Impact of pitch-angle control on the output power of a FSWT: (a)
Power-speed characteristics at zero pitch-angle; (b) Power-angle
characteristics at a pitch-angle of five degrees
B. DFIG model and control

Fig. 8.16 shows the basic building blocks of a DFIG-based VSWT. The
system consists of a wound-rotor induction machine controlled by two back-
to-back connected PWM converters with a controllable DC intermediate link
scheme. The first converter is an AC-DC PWM-converter called the rotor-
side-converter (RSC) connected between the rotor AC-windings (via slip
rings and brushes) and the DC-link. The second converter is a DC-AC PWM-
converter called the grid-side-converter (GSC) connected between the AC-
grid (i.e the DFIG stator windings) and the DC-link. A wind turbine (WT),
with energy-control via the pitch-angle controller, is coupled to the generator
shaft via a gearbox. The RSC controller provides control of both the WT
output power and the DFIG reactive power output or terminal voltage or
power factor. The main function of the GSC control is to regulate the DC-
link voltage as well as possible other control functions such as the generation
or absorption of reactive power. In power control, the turbine output is
controlled in order to follow a pre-defined power-speed characteristic
corresponding to the maximum wind-energy-capture tracking called the
maximum power tracking characteristics (MPTC). This achieved through
integration between the RSC and the pitch-angle controllers.
323
Fig. 8.16: Basic building blocks of a DFIG-based VSWT
Fig. 8.17: Speed-power characteristic of VSWT wind turbine at zero

pitch-angle
Fig. 8.18 Equivalent circuit of a DFIG
324
The speed control is aimed to maximize the power production of the wind
turbine. Fig. 8.17 shows the dependence of the mechanical power
produced by the wind turbine on the wind speed and the turbine rotor
speed . In addition, the maximum mechanical power locus for each wind
and rotor speeds is illustrated on the figure. This curve is used for defining,
for each value of the rotor speed, the optimal mechanical power that the
turbine has to produce. For super-synchronous speeds, the reference power is
fixed to 1.0 p.u to avoid overloading the generator. For < 0.5 pu, the
reference mechanical power is set to zero. For , the detailed (
, ) characteristic is,
This is can be simplified to

Due to the fast dynamics of the stator flux comparison with the grid
dynamics, the electromagnetic equations of the DFIG are usually represented
by a steady-state model. The equivalent circuit of a DFIG is shown in Fig.
8.18. The three – phase stator and rotor windings of an induction machine can
be represented by two sets of orthogonal fictitious coils. The DFIG is
controlled in a rotating d-q reference frame, with the d-axis aligned along the
stator-flux vector. The machine stator and rotor voltages are functions of the
stator and rotor currents as well as the rotor speed . The d-q representation
of the machine is as follows,
The links between stator fluxes and generator currents are modeled using
325
The network interface is modeled as follows,
The generator active and reactive power productions depend on the

stator currents ( ), the converter currents , the stator
voltages ( ), and the converter voltages as follows,
The above expressions can be rewritten as a function of stator and rotor

currents and stator and rotor voltages. In fact, the converter powers on the
grid side are represented by (8.31) and (8.32) while the converter powers on
the rotor side are represented by (8.33) and (8.34).
Assuming a loss-less converter model, the active power of the

converter coincides with the rotor active power, thus = . The reactive
power injected into the grid can be approximated by neglecting stator
326
resistance and assuming that the d-axis coincides with the maximum of the
stator flux. Therefore, the powers injected in the grid can be simply
represented by,
C. Steady-state performance
It is clear from Fig. 8.14 and 8.14, that the model of the DFIG includes
the model of the SCIG. If the rotor circuit of a WRIG is shorted, the machine
becomes conceptually similar to a SCIG. Therefore, the following analysis is
given for the DFIG as it is the general case. The 3rd order fundamental
frequency model for the DFIG with flux oriented control scheme is
considered here for extraction of the steady state performance
characteristics 114.
Fig. 8.19 dq-representation of DFIG
The 3rd-order model is based on the following assumptions. (1) Neglecting

magnetic saturation; (2) Single-mass representation of all rotating masses; (3)
Stator electromagnetic transients as well as stator resistance are neglected; (4)
114
EL-Shimy M. Steady State Modeling and Analysis of DFIG for Variable-Speed
Variable- Pitch Wind Turbines. ASJEE. 2010;1:179 - 89
327
Generator current and power convention is used; (5) Synchronous reference

frame. In addition, for stator flux-oriented control scheme, the stator flux s
is aligned to the q-axis (i.e. = 0) as shown in Fig. 8.19. This is providing
decoupling between active and reactive power output control. Based on the
above assumptions and model requirements, the DFIG 3 rd-order model is as
follows.
where is the equivalent transient reactance of the machine

and s is the slip of the rotor.
Based on equations (8.41) and (8.42), it is noticed that stator active and
reactive power equations are decoupled due to the stator flux oriented-control
scheme.
It is assumed that the GSC operates at unity power factor for all
operating conditions i.e. Qr = 0. In addition, it is assumed that both
328
mechanical losses and converter losses are neglected. The rotor active power
is also neglected under steady-state conditions. Therefore,
The stator active and reactive power represented by equations (8.41) and
(8.42) can be obtained in terms of the rotor voltage d- and q-components by
solving, under steady-state conditions, equations (8.39) and (8.40) for the
rotor current d- and q-components. These are results in the following
equations,
Therefore, by substituting (8.47) in (8.41) and (8.48) in (8.42), the stator

active and reactive power characteristics are obtained and take the form,
For assessing the steady-state stability, based on the DFIG Ps-s

characteristics, for a given operating point on the Ps-s characteristics of an
329
induction machine the dPs /ds criteria can be used. Since generator power
convention is used in the machine modeling, a steady-state stable operating
point requires that dPs /ds < 0. Based on (8.49), dPs /ds take the form,
where
Since we are dealing with DFIG characteristics control via rotor injection,
it is valuable to determine the values of the slip at zero output power (sR) as
well as the generator-mode critical slip scr_gen. The former is obtained by
setting Ps = 0 in (8.49), and the later is obtained by setting dPs /ds = 0 in
(8.51). Two solutions for scr are obtained, where the value of scr
corresponding to positive Ps is the generating-mode critical slip scr_gen and the
other (corresponding to negative stator power) is the motoring-mode critical
slip scr_mot. The value of sR and scr_gen can be computed using,
where , ,
The control strategy of DFIG system includes controlling the active

and reactive power injections from the generator. For active power control,
optimum operating point tracking of the VSWT is considered to provide
maximum energy capture from wind. The objective of tracking control is to
keep the turbine on this optimum tracking curve as the wind speed varies.
330
A general optimal tracking strategy for DFIG system is shown in Fig.

8.20 and can be represented by the following equation,
for
for
for
for
for
for
Based on Fig. 8.20, the function of the pitch-angle control is to limit the
turbine output energy at higher than rated wind speeds through variable-pitch
operation. For lower than rated wind speeds the wind energy capture is
maximized through fixed-pitch operation with pitch-angle set to zero degrees.
Fig. 8.20 Optimal tracking strategy for DFIG system
Two methods are used, for DFIG converters, to control output real-
power production of the DFIG according to the optimal tracking
characteristics 115. These are (i) current-mode control, and (ii) speed-mode
control. In current-mode control, that can be considered as a standard
tracking mode, the electrical real-power output from the DFIG is controlled
115
R. Pena, J.C.Clare, and G. M. As her, Doubly fed induction generator using back-to-
back PWM converters and its application to variable-speed wind-energy generation, IEE
Proc.-Electr. Power Appl, Vo1 143, No. 3, May 1996, pp. 231-24
331
according to the generator rotor speed. In speed-mode control, the generator

rotor speed is controlled according to a desired electrical real-power output
allowing extra merits over current-mode control in allowing flexible
dispatchable-like operation of the DFIG system; however, the available wind
resource defines the maximum available generation.
In the current-control mode as illustrated in Fig. 8.16, the RSC is used
to set the appropriate electrical active power output reference of the DFIG to
follow the optimal tracking characteristics shown in Fig. 8.20 according to
equation (8.41). In addition, the RSC can also be used to regulate the reactive
power output from the DFIG according to equation (8.42).
The steady-state rotor circuit d-q current components required for
active- and reactive- power control are obtained directly from equations
(8.49) and (8.50). However, the steady-state d-q voltage components
impressed on the DFIG rotor circuits for active- and reactive- power control
can be obtained for a given rotor speed and demanded reactive power output
through simultaneous solution of equations (8.49) and (8.50) where the DFIG
active power output is calculated from the tracking characteristics of Fig.
8.20. These d-q voltage components take the form,
where
8.4 Case study 1 - Steady-state characteristics and steady-state

stability of DFIG
Consider DFIG of 2MW, 690V, 50Hz. The DFIG p.u parameters are listed
in Table 8.3, which also include definitions of DFIG parameters and symbols.
332
Table 8.3 DFIG parameters

Parameter Symbol Value, p.u
Stator resistance Rs 0.00488
Stator leakage reactance Xs 0.09241
Rotor resistance Rr 0.00549
Rotor leakage reactance Xr 0.09955
Magnetizing reactance Xm 3.95279
Inertia constant (lumped) H 3.5, sec.
A 2MW variable-pitch variable-speed wind turbine with characteristics

shown in Fig. 8.21 (obtained using the VSWT aerodynamic model available
within SimPowerSystems Toolbox). The turbine tracking characteristics data
are given in Table 8.4.
Fig. 8.21 VSWT power characteristics at zero-pitch angle and optimal

tracking characteristics
Fig. 8.21 Tracking characteristics data

Parameter Symbol Value, p.u
Rotor speed at point A rA 0.7
Rotor speed at point B rB 0.71
Rotor speed at point C rC 1.2
Rotor speed at point D rD 1.21
Output Power at point C PC 0.73
Output power at point D PD 1
333
In the following steady-state simulation, the DFIG terminal voltage is

assumed to be kept constant at 1 p.u which can be realized practically when
the DFIG is connected to an infinite-bus through a network with infinite
strength.
A. Operation over full tracking characteristics

With generator-rotor speed varies from 0.6 p.u to 1.3 p.u, the demanded
active power output from the DFIG is obtained according to the optimal
tracking characteristics shown in Fig. 8.21 and the DFIG reactive power
output is set to zero. The resulting rotor voltage and current components are
shown in Fig. 8.22 while Fig. 8.23focuses on the rotor voltage components
only.
Fig. 8.22 Operation over full tracking characteristics with Qs = 0; rotor

current and voltage components
As depicted from equations (8.41) and (8.42) and Fig. 8.22 and Fig.
8.23(b), the resulting rotor current components that are decoupled from each
other are linearly related with the demanded active and reactive power for all
rotor speeds. Based on Fig. 8.23(a), for rotor speed below 0.7 p.u where both
active and reactive power demands are zero, the Vqr is of a fixed negative
value independent on the rotor speed while Vdr is linearly related and
decreasing with the rotor speed. The described changes of Vdr are also
applicable for the rotor speed range with positive values of slip at sub
334
synchronous speeds and negative value at supersynchronous speeds. The sign

of Vqr is negative for speeds up to 1.13 p.u and positive for supersynchronous
speeds above 1.13 p.u while its value is zero at the synchronous speed.
Although Vqr is of a fixed value independent on the rotor speed for
demanded zero output active power, its value is increasing with the rotor
speed above 1.2 p.u with fixed demanded output active power of 1.0 p.u. This
is shown as steady-state instability as depicted from Fig. 8.24 that shows
dPs/ds values for the entire tracking characteristics under different demanded
reactive power output settings. Therefore, it can be concluded that positive
sign of Vqr indicates steady-state instability; however, these results are highly
dependent on the considered parameters and may significantly vary for
different parameters. Since dPs/ds is positive for rotor speeds above 1.13 p.u
(for Qs = 0), the DFIG is unstable in this region. This is occurring with
different values of reactive power output demanded at rotor speeds higher
than 1.13 p.u. The causes of the steady-state instability for rotor speeds
greater than 1.13 p.u can be explained by calculating the DFIG maximum
steady-state power, critical generator slip, and the slip at zero power output as
shown in Fig. 8.25 with the instability zone focused in Fig. 8.26. Based on
Fig. 8.24, Fig. 8.25 and Fig. 8.26 for stable operating points where dPs/ds <
0, the demanded rotor speeds and active power demands are less than the
critical generator speed and the maximum generator power respectively. In
addition, for rotor speeds between 1.13 p.u and 1.2 p.u, the demanded active
power output is higher than the maximum generator output. In addition, for
all the stable operating points, the demanded rotor speeds are higher than the
rotor speed at zero active power output. The steady-state unstable operating
points are realized by positive dPs/ds and positive Vqr. These points are
characterized by violating the critical generator speed or the maximum
generator power or both as depicted from Fig. 8.23 and Fig. 8.26. The
considered reactive power output demands do not significantly affect the
steady-state stability of the DFIG as shown in Fig. 8.24. Therefore, the
considered DFIG should not run at supersynchronous speeds ( rmax shown in
Fig. 8.20) higher than 1.13 p.u in order to achieve stable steady-state
operation. In addition, positive sign of rotor voltage d-component can be
considered as an indicator of DFIG steady-state instability and injection of
positive Vqr should be prohibited to avoid hindering steady-state stability.
335
(a) (b)
Fig. 8.23 Operation over full tracking characteristics with Qs = 0; (a) Rotor
voltage components; (b) Rotor currents and powers
Fig. 8.24 Operation over full tracking characteristics with different Qs;
Steady-state stability
Fig. 8.25 Operation over full tracking characteristics with Qs = 0 - Causes of

steady-state instability
336
Fig. 8.26 Operation over full tracking characteristics with Qs = 0 - Focus on

the steady-state instability region
B. Solo effects of rotor voltage components on DFIM performance

The solo effects of each of the d- and q- components of the rotor
voltage on the steady-state performance and steady-state stability of the
doubly-fed induction machine (DFIM) are analyzed based on the assumption
of unity terminal voltage. The analysis considers the motor as well as
generator operation modes. The resulting steady-state responses are plotted
for a limited range of rotor speeds (slips) for better visualization. However,
due to the careful selection of the scale range, this will not affect the derived
conclusions.
With Vqr = 0, the effects of variations of Vdr on the DFIM Ps-s
characteristics, steady-state stability, and critical limits are shown in Fig.
8.27-a, Fig. 8.27-b, and Fig. 8.27-c respectively. The following are depicted
from Fig. 8.27. With positive Vdr, the DFIM Ps-s characteristics shifted to
right (Fig. 8.27-a) and a significant extension of the steady-state stability
region, with respect to zero rotor voltage case, occurs in subsynchronous
speeds as positive Vdr increased. In addition, a significant reduction of the
stability region in supersynchronous speeds occurs as positive Vdr increased.
In addition, a significant linear increase of the maximum power in the
generating mode as positive Vdr increased; this is in contrary to the motoring
mode maximum power, which is very close to zero with all positive values of
Vdr. Therefore, with positive Vdr, wide range of subsynchronous speeds are
available for steady-state stable operation of the DFIM. However, due to the
significant increase in generating-mode maximum power and the significant
337
reduction in motoring-mode maximum power, positive Vdr improves the

generating-mode of the DFIM for subsynchronous speeds. The opposite is
obtained for negative Vdr. Therefore, with negative Vdr, wide range of
supersynchronous speeds is available for steady-state stable operation of the
DFIM. However, due to the significant increase in motoring-mode maximum
power and the significant reduction in generating-mode maximum power,
negative Vdr improves the motoring-mode of the DFIM for supersynchronous
speeds. Generally, linear increase of the slip at zero power output is obtained
with increasing Vdr.
Fig. 8.28 shows the solo effect of Vqr with Vdr = 0. Based on the figure,
neither positive nor negative Vqr affect either the critical slips of the DFIM or
the slip at zero power output. With positive Vqr, the steady-state stability
region of rotor slips is normally within the critical slip values corresponding
to generating and motoring modes of operation; however, with negative Vqr
the steady-state stability region of rotor slips is inverted.
With positive Vqr, the steady-state stability of both generating and
motoring modes of operation are enhanced via increase in the maximum
power; however, this is valid for subsynchronous rotor speeds in the
generating mode and supersynchronous speeds in the motoring mode.
With negative Vqr, enhancement in the steady-state stability through
increase in the output power occurs for supersynchronous speeds higher than
1.03 p.u in the generating mode and for subsynchronous speeds less than 0.97
p.u in the motoring mode. Within the mentioned speed range the machine is
unstable in either modes of operation. It is clear from Fig. 8.28-c that an
approximately linear transitions between the stable and unstable regions of
rotor speeds occur as Vqr is reduced from 0.05 p.u to -0.05 p.u.
338
(a)
(b)
(c)
Fig. 8.27 Solo effects of Vdr on DFIM performance and stability; Vqr = 0
339
(a)
(b)
(c)
Fig. 8.28 Solo effects of Vqr on DFIM performance and stability; with Vdr
=0
340
The above analysis of the solo effects of rotor voltage components on

the steady-state performance and steady-state stability of DFIM show that
infinite number of machine characteristics can be obtained through mix
selection of Vqr and Vdr even with limited amplitude of rotor voltage Vr. Each
of these characteristics has its own merits and steady-state stability limits.
8.5 Case study 2 - Transient performance of FSWT based on

SCIG considering the single-mass model (rigid rotor)
This case study presents a detailed analysis of the transient response
considering two control situations of a pitch controlled wind turbine (see Fig.
8.30). In the first situation, the pitch angle control loop is working properly
while in the second situation the pitch controller is stalled due to an internal
failure. The first situation will be called Active Stall Control (ASC) while the
second situation will be called Stalled (or disconnected) Controller (SC). The
failure may be a disconnection in the closed loop pitch-control system or
problems in the hydraulic pitch changer.
The objective is to demonstrate the dangers associated with losing the
only available controller which is responsible of limiting the output power of
the turbine during abnormal conditions. These abnormal conditions include
high wind speed in comparison with the rated wind speed or a network fault.
The single-mass model will be utilized in this example for simplification;
however, this is only applicable to rigid shafts while the usual shafts of
FSWTs are soft and the multi-mass model is recommended for accurate
analysis. A detailed comparison between the stated situations is conducted.
Two types of disturbances are considered. The first is an intense change in
the wind speed and the second is a three-phase fault at the induction
generator terminals.
A. Study system and modeling requirements

The study system of Fig. 8.29 is implemented on SimPowerSystems based
Matlab Toolbox. The low-speed shaft dynamics and the pitch-angle rate
limiter are neglected. This is to allow investigating the FSWT system
dynamics under design improvement in both the low-speed shaft material
stiffness and the rate of change of pitch angle capability. High rates of pitch-
341
angle change may be obtained via fast acting hydraulic/mechanical actuator

designs that capable of handling high speed change in pitch angle variations
either using PI controllers or neural network based controllers 116. From
technical and operational points of view, appropriate delay-logic should be
included in the pitch-angle control loop. This delay may be optimized based
on the existing magnitude and rate of change of the wind speed considering
also a statistical survey analysis of wind speed patterns. This is to avoid
excessive unnecessary operation, and wear out of the pitch-angle control
system. The basic building blocks needed to simulate the study system of Fig.
8.29 are shown in Fig. 8.30.
Fig. 8.29: Study System
Fig. 8.30: Basic building blocks of study system model
116
M. EL-Shimy, “ Modeling and Control of Wind Turbines Including Aerodynamics”,
Scientific Bulletin - Faculty of Engineering - Ain Shams Uni. Vol. 41, No. 2, June 30,
2006
342
The equations describing each of the shown components are already stated
in the previous sections. For stalled controller, the pitch angle  is fixed to
zero regardless of the available wind speed. This concept is used for
implementing the model in the SimPowerSystems toolbox. In the
SimPowerSystems library, the induction generator is represented by 5 th order
dynamic model. Therefore, stator dynamics as well as network dynamics are
included in the results. In addition, a pitch angle controller of a PI
configuration is also included in this library. The system model initialization
is conducted by simulating the undisturbed system and then appropriate
initial conditions are recorded in the initial conditions slot of each block. This
method shows better model initialization to steady-state conditions in
comparison with the model initialization method described in the toolbox
manual.
The data of the wind turbine and the pitch-angle controller are shown in
Table 8.22. The turbine power characteristics at zero and at five degrees
pitch-angles are plotted and they are shown in Fig. 8.31 and Fig. 8.32
respectively.
Table 8.22: Wind turbine data

For both stall and active stall controlled turbine
Nominal Mechanical Power (MW) 1.5
Base wind speed* (m/s) 10
Maximum power at base wind speed
1
(pu of nominal mechanical power)
Base rotational speed (pu of base generator speed) 1
For Active stall controlled turbine
Pitch angle controller gain, kp 5
Pitch angle controller gain, ki 25
Maximum pitch angle (deg) 40
*The base wind speed is the mean value of the expected wind speed.
343
Fig. 8.31: Turbine power characteristics at zero pitch angle
Fig. 8.32: Turbine power characteristics at five degrees pitch angle
The controlled stalling effect using pitch-angle control is depicted from

Fig. 8.31 and Fig. 8.32. Based on Fig. 8.31, with zero pitch-angle the nominal
mechanical of the turbine occurs at the base wind speed, which obeys the
turbine data. However, based on Fig. 8.32, with five degrees pitch angle the
output power is limited due to blades stalling results in 0.725 pu mechanical
power output at the base wind speed. Induction generator data are shown in
Table2 with pu values are based on generator ratings.
The p.u data of the system based on the turbine rating are shown in Table
8.23 and Table 8.24.
B. Simulation results
For showing the capability of the considered FSWT systems under large
disturbances originates from the nature or from the system faults, two types
of disturbances are considered. The first is an intense change in the wind
344
speed and the second is a three-phase fault at the induction generator

terminals.
Table 8.23: Induction generator data
Nominal Power (MVA) 1.6667
Nominal L-L rms voltage (V) 575
Nominal frequency (Hz) 60
Stator resistance (pu) 0.01
Rotor resistance (pu) 0.014
Stator leakage reactance (pu) 0.1
Rotor leakage reactance (pu) 0.098
Magnetizing reactance (pu) 3.5
Inertia constant including the wind turbine (pu) 3
Number of pole 6
Table 8.24: Network data

Transformer 1 2
Nominal power (MVA) 2 47
Nominal frequency (Hz) 60 60
Turns ratio 25e3/575 120/25
Resistance (pu) 0.0017 0.0053
Reactance (pu) 0.05 0.160
Transmission Line
Resistance () 0.0825
Reactance () 0.1037
Grid
Nominal voltage (kV) 120
nominal frequency (Hz) 60
S.C level (MVA) 2500
S.C ratio 10
Compensating capacitor
Nominal voltage (V) 575
Rated capacity (kVAr) 700
B.1. Wind Speed Disturbance

Starting at the base wind speed (i.e. the mean value of the expected
wind speed = 1 pu), 60% step increase in the wind speed occurs at t = 1
second and sustained at this level for two seconds till this change in wind
speed is vanished in step manner at t = 3 second. The selected base wind
speed is 10 m/s, which is also corresponding the maximum mechanical power
345
output. The considered wind speed disturbance is shown in Fig. 8.33. This
disturbance is applied to the considered situations i.e. the stalled controller
(SC) and the active stall control (ASC). The corresponding response of the
mechanical torque (Tm) and the electrical torque (Te) for both situations are
shown in Fig. 8.34.
Fig. 8.33: Wind speed disturbance
Fig. 8.34: Torque response for the wind speed disturbance
Based on Fig. 8.34, the mechanical torque and electrical torque of the SC
turbine nearly follow the wind speed pattern. Therefore, upon
turbine/generator protection failure, this may be representing a hazard to the
system security during high wind speeds. However, with ASC, a significant
and continuous drop in the torque is obtained resulting in lower system stress
during disturbance. This results from the pitch angle control that limits the
346
generator output power to its nominal value for high wind speeds. The pitch-
angle variation is shown in Fig. 8.35. The maximum rate of the required
change in the pitch angle is about 9.0 deg/s.
Fig. 8.35: Pitch-angle variation for the wind speed disturbance
The generator active/reactive power response and the generator terminal

voltage response for the wind speed disturbance of Fig. 8.33 are shown in
Fig. 8.36 and Fig. 8.37 respectively.
Fig. 8.36: Generator active/reactive power response for the wind speed
disturbance
Based on Fig. 8.36, a better response of active power generation is

obtained as a result of the pitch-angle control. With the controller stalled, the
violation of the rated power of the turbine is sustained during the considered
wind speed disturbance while the ASC results in smaller violations that are
decreasing with time. In addition, with pitch-angle control, the generator bus
347
reactive power demand is greatly reduced resulting in lower generator

terminal voltage fluctuations (Fig. 8.37) during high wind speed disturbance.
Therefore, better power quality is obtained and less effect on local sensitive
load equipment is resulted from the ASC. If the acceptable steady state
terminal voltage magnitude is 5%, then with the SC, the generator is
subjected to an unacceptable drop in its terminal voltage while the situation is
absent with the ASC.
Fig. 8.37: Generator terminal voltage response for the wind speed
disturbance
Fig. 8.38: Generator slip response for the wind speed disturbance
The generator slip during wind speed disturbance is shown in Fig. 8.38. It
is shown that in either wind turbine control conditions, the generator is stable
and did not temporarily operate as motor during the considered wind speed
348
disturbance; however, a better response is obtained with the ASC. Better

generator stability is obtained with ASC as depicted from the torque-slip
trajectory of Fig. 8.39.
Fig. 8.39: Torque-slip trajectory for wind speed disturbance
B.2. Three-Phase Fault at Generator Terminals

The analysis the considered FSWT systems during severe fault
disturbances is considered in this section. A 5-cycle, 3-phase (LLLG) fault
started at t = 1.0 sec is applied at the generator terminals and the critical-
clearing time (CCT) is determined. The results show that CCT is 6-cycle (0.1
second) for both SC and ASC conditions. Unlike the considered wind speed
disturbance, the fault will not cause an increase in the generated power during
the fault conditions. Therefore, the impact of the pitch angle control did not
reflect on the CCT.
The generator terminal voltage response for both stall and active stall
controlled turbine systems is nearly identical and is shown in Fig. 8.40. The
mechanical torque response comparison and pitch-angle response of the ASC
are shown in Fig. 8.41 and Fig. 8.42 respectively.
349
Fig. 8.40: Generator terminal voltage response for the 3-phase fault
Fig. 8.41: Mechanical torque response for the 3-phase fault
Fig. 8.42: Pitch-angle response for the 3-phase fault
Based on Fig. 8.41, a significant reduction in the turbine mechanical

torque output is resulted with the ASC due to pitch-angle control as depicted
350
from Fig. 8.4; however, this significant reduction in the mechanical torque
due to the fault will not be reflected significantly on other system variables
especially the active/reactive power generated as will be illustrated below.
Therefore, it is clear that the generator and the network mainly govern the
electromechanical transient process and, to high extent, independent on the
turbine control. This is depicted from the generator torque-slip trajectory
shown in Fig. 8.43; however, the benefit of the input mechanical torque
reduction due to pitch-angle control is a consequent significant reduction of
the torsional stress applied to the shaft during the transient process.
Fig. 8.43: Torque-slip trajectory for the 3-phase fault
Fig. 8.44: Electrical torque response for the 3-phase fault
351
The responses of the electrical torque, active power generated, and

reactive power generated, and generator slip are shown in Fig. 8.44 to Fig.
8.47 respectively.
Fig. 8.45: Active power response for the 3-phase fault
Based on Fig. 8.43 to Fig. 8.47, insignificant gain in the overall system
response is obtained with the ASC for the considered 3-phase fault at the
generator terminals; however, both types of control results in stable FSWT
system without the complications associated with RSC and the voltage
recovery following a close fault associated with DFIG based systems 117.
Fig. 8.46: Reactive power response for the 3-phase fault
117
Mustafa Kayikçi and J. V. Milanovic´, “Assessing Transient Response of DFIG-Based
Wind Plants—The Influence of Model Simplifications and Parameters”, IEEE
Transactions On Power Systems, Vol. 23, No. 2, May 2008, pp. 545-554
352
Fig. 8.47: Generator’s slip response for the 3-phase fault
8.6 Case study 3 – Design of Pitch Angle Controllers118

This section presents a model for a variable-pitch, constant-speed
horizontal-axis wind turbines including aerodynamics and mechanical parts.
In addition, the pitch angle control design is considered. Two techniques of
control are applied; the PID control and ANN-NARMA-L2-based control.
The main objective of both techniques is to keep the speed of the turbine
constant at a desired level.
A. Turbine Description and Preliminary Modeling

The wind turbine rotor is connected to a generator. The generator output
can be controlled to follow the commanded voltage. The wind turbine has
pitchable blades to control the aerodynamic power extracted from the wind.
In addition, there is a mechanical speed-changer (i.e. gearbox) between the
low-speed rotor shaft and the high-speed generator shaft. The low-speed shaft
is driven by the turbine blades, which generates aerodynamic power. The
high-speed shaft is loaded by the electric generator in the form of an
electrical load. As the wind speed fluctuates, the wind turbine is controlled by
changing the pitch angle to fix the rotor speed following the variation of the
wind speed. Therefore, the wind-turbine-generator (WTG) system converts
rotational energy to electrical energy, which is may be directly supplied to the
118
EL-Shimy M. Modeling and Control of Wind Turbines Including Aerodynamics.
Scientific Bulletin - Faculty of Engineering - Ain Shams Uni. 2006;41(2):527 - 42.
353
utility grid at the distribution level or to the HV grid. The choice depends on
the sacle of the wind farm and the intended application.
A.1. Available Wind Power Modeling

The kinetic energy, U of a packet of wind of mass m flowing at
upstream speed u in the axial direction of the wind turbine is given by,
1 1
U  mu 2  ( Ax)u 2 (8.57)
2 2
where A is the cross-sectional (swept) area of the wind turbine, is the air
density and x is the thickness of the wind packet. The wind power, Pw in the
wind, which represents the total power available for extraction, is given by,
dU
Pw  (8.58)
dt
Therefore,
1 dx 1
Pw  Au 2  Au 3 (8.59)
2 dt 2
The mechanical power, Pm extracted from the available power in the wind Pw
is expressed by the turbine power coefficient of performance CP which is a
nonlinear function of tip speed ratio  and pitch angle . Therefore,
Pm  C p  ,  Pw (8.60)
In ideal conditions, the turbine cannot extract more than 59% of the
total power of undisturbed tube of air with cross sectional area equals to wind
turbine swept area. This called Betz limit. The tip speed ratio  is a variable
that combines effect of rotational speed of the turbine and wind speed. It is
defined as the ratio between the rectilinear speed of the turbine tip (R) and
the wind speed (u).
354
R
 (8.61)
u
where R is the maximum radius of the wind turbine swept area. The
following equation can be used to approximate the CP curve,
C p ( ,  )  0.22116K  0.4  5e 12.5 K (8.62)

where
1 0.035
K  3 (8.63)
  0.08   1
Variation of CP with tip speed ratio at constant pitch angle and variation of
CP with pitch angle at constant tip speed ratio are shown in Fig. 8.48 and Fig.
8.49 respectively. As shown in Fig. 8.48, the maximum value of Cp reduced
with the increase in the pitch angle. In addition, the tip-speed ratio at
maximum Cp increases with the decrease in the pitch angle. This clarifies the
impact of the variations of the pitch angle on the power coefficient which
consequently affects the power production.
The wind-turbine rotor performance can also be evaluated as function
of the coefficient of torque Cq. As the wind power Pw equals to the product of
the aerodynamic torque TA and the rotor rotational speed , the torque
coefficient can be related to the power coefficient by,
C p ( ,  )  Cq ( ,  ) (8.64)
Based on (8.59), (8.60), (8.61), and (8.64), the aerodynamic torque that turns
the rotor shaft takes the form,
1 
TA   AR Cq ( ,  )u 2 (8.65)
2 
355
Fig. 8.48: Variations of the power coefficient with the tip-speed ratio at
constant pitch-angle
Fig. 8.49: Variations of the power coefficient with the pitch angle at
constant tip-speed ratio
A.2. Wind Turbine Linearized Modeing for Controller Design

The most special feature about wind turbines is the fact that, unlike other
generation systems, the power inflow rate is not controllable. In most
conventional power generation systems, the fuel flow rate, or the amount of
energy, applied to the generator controls the output voltage and frequency;
however, the wind speed varies with time and so does the power demand.
Therefore, conventional generation systems can be referred to as controlled
energy sources, while the wind is an uncontrolled energy resource. The
power demand is an uncontrolled energy sink.
Sometimes the wind speed can be very high resulting in power generation
that exceeds either the rating of the turbine or the demanded power. This
356
might lead to the turbine exceeding its rotational speed limit and
subsequently the turbine may be damaged. On the other hand, the wind speed
can be too low for any power production and therefore alternative energy
sources should be used. This forms a hybrid-electric power source.
The wind-turbine-generator (WTG) linearized model will now be derived
for the use in the pitch-angle control design. The model is divided into two
main parts. The first part is the wind turbine, which included a turbine rotor
on a low-speed shaft, a gearbox and high-speed shaft. The inputs for this part
of the plant are the wind speed and the blade pitch angle while the outputs are
the high-speed shaft angular rotation and the mechanical power, Pm. The
second part is the electric generator whose input is constant angular rotation
from the turbine plant and whose output is electrical power. Fig. 8.50 shows a
block diagram of the wind turbine system.
Fig. 8.50: Grid-connected WTG
The equation of motion of wind turbine system is given by,
d
JT  T A T L (8.66)
dt
where JT is the equivalent combined moment of inertia of the rotor,

gearbox and both the low-speed and high-speed shafts; TL is the wind turbine
load torque representing the input torque to the electrical generator and
opposed by its electrical torque.
For the purpose of dynamic analysis and for designing a linear controller,
such as PID controller, equation (8.66) is linearized around an initial
operating point (uo, o, o). Substituting for TA in (8.66) using (8.65), then
the linearized form of (8.66) takes the form,
357
d T
 u      L (8.67)
dt JT
The parameters , and are calculated at the initial operating conditions
(uo, o) and they are given by:
1  dC 
 ARuo  2Cqo  o q  (8.68)
2JT  d 
 o
1 dC
 AR 2uo q (8.69)
2JT d o
1 dC
 ARuo2 q (8.70)
2JT d o
The magnitude of  and  respectively show the relative weight of the effect
of wind speed and pitch angle on the wind turbine dynamics. In s-space,
(8.67) takes the form:
1  T 
 ( s )  u( s )   ( s )  L  (8.71)
s   JT 
Equation (8.71) represents the linearized form of the wind turbine transfer
function. The turbine power output is given by,
Pm  TA (8.72)
The linearized form of output power equation (8.72) takes the form:
Pm   o TA  TAo (8.73)
358
Based on (8.67) and (8.74), the block diagram of wind turbine plant is
given in Fig. 8.51.
Fig. 8.51: Linearized model of the WT
A.3. Wind Turbine Control

A general block diagram for wind turbine pitch-control control system is
shown in Fig. 8.52.
Fig. 8.52: General block diagram of the pitch-control system for WTS
The transfer function of a hydraulic actuator that changes the blade pitch
angle can be represented by first-order transfer function,
 ( s ) kA
GA (s)   (8.74)
 c ( s ) s  k A
359
In this case study two controllers are considered; a PID and a neural
network NARMA-L2 controllers (that are available in the MATLAB). PID
controllers regulate the error, or difference between the measured input and
the desired input (). This error value along with its derivative and integral
with respect to time provides a signal to the actuator(s), which affects the
controlled plant. The PID controller is a linear, single-input single- output
controller limited to three gains. The transfer function of the PID controller is
given by,
 c ( s ) k
GC ( s )   k p  I  k D s (8.75)
 ( s ) s
The central idea of the NARMA-L2 neurocontroller is to transform

nonlinear system dynamics into linear dynamics by canceling the
nonlinearities. There are typically two stages involved when using neural
networks for control. The first stage is system identification in which a neural
network model of the plant to be controlled is developed by training a neural
network to represent the forward dynamics of the system. The second stage is
the control design in which the neural network plant model is used to design
(or train) the controller.
B. ANALYSIS AND CONTROL DESIGN

The mathematical model in previous sections is applied to develop the
response of controlled and uncontrolled wind turbine plant. The wind turbine
parameters are given in Table 8.24.
Table 8.24: Wind turbine parameters

Rated Power (kW) 20
Radius (m) 5
2
Drive train inertia (kg.m ) 1270
Gear ratio 11.43
Operation angular speed (rpm) 105
Rated wind speed (m/s) 11.7
Cut-in speed (m/s) 6.5
Furling speed (m/s) 23
360
B.1. Uncontrolled Response of Wind Turbine Plant

With initial operating point is (uo = 7 m/s, o = 10.5 rad/s, o = 9 deg.) the
parameters , and  calculated; their values are 0.2071, -0.0668, and –
0.0298 respectively. With uncontrolled wind-turbine plant, the response to a
unit step in wind speed, a unit step in pitch angle, and a 10% step increase in
load torque are determined and shown in Fig. 8.53, Fig. 8.54, and Fig. 8.55
respectively. These responses are mainly dependent on the rotational inertia
of the wind turbine plant, the scaling factors  and and the parameter . The
parameter  represents the wind turbine aerodynamic characteristics and it
does not affect the wind turbine plant inputs.
(a) (b)
Fig. 8.53: Uncontrolled response to a unit step in the wind speed
(a) (b)
Fig. 8.54: Uncontrolled response to a unit step in the pitch angle
361
(a) (b)
Fig. 8.55: Uncontrolled response to a unit step in the load troque
The shown simulation results verify the conceptual correctiness of the

model. For example, an increase in the wind speed results in an increase in
the mechanical power and the speed while an increase in the pitch angle
causes reduction in both the speed and the power. The increase on the load
torque has a similar impact to the increase in the pitch angle but with a
different time constant.
B.2. Controlled Response of Wind Turbine Plant

PID and neural network NARMA-L2 controllers are used to compensate
the wind turbine speed deviations by changing the pitch angle . Based on
Routh-analysis of the wind turbine transfer function and trial and error
approach, the gains of PID controller are selected to be (k P = 60, kI = 50, and
kD = 20). The plant response to a unit step in wind speed, and a 10% step
increase in load torque are shown in Fig. 8.56, and Fig. 8.57 respectively.
It is shown that PID controller succeeded in keeping wind turbine speed
and output power. In order to show the validity of the PID controller, a
variable wind speed is assumed as shown in Fig. 8.58, the plant response to
this variable wind speed is shown in Fig. 8.59. It is clear that plant response
suffers from a small amount of control errors due to the fast changes in the
wind speed.
362
(a) (b)
(c)
Fig. 8.56: PID controlled response to a unit step in the wind speed
(a) (b)
(c)
Fig. 8.57: PID controlled response to a unit step in the load troque
363
Fig. 8.58: Variable wind speed
(a) (b)
(c)
Fig. 8.59: PID controlled response to variations in the wind speed
The neural network NARMA-L2 controller with on-line training is used

instead of PID controller. The plant response to a unit step in wind speed, a
10% step increase in load torque, and variable wind speed are shown in Fig.
8.60, Fig. 8.61, and Fig. 8.62 respectively.
364
(a) (b)
(c)
Fig. 8.60: ANN controlled response to unit step in the wind speed
(a) (b)
(c)
Fig. 8.61: ANN controlled response to 10% increase in the load torque
365
(a) (b)
(c)
Fig. 8.62: ANN controlled response to variable wind speed
Although a small amount of control errors are obtained when using PID
controllers, high accuracy of plant response to follow the objective of a
nearly zero speed deviation is obtained with neural network NARMA-L2
controller. The main critical issue associated with the presented use of the
NARMA-L2 controller is the required high rate of the pitch angle changes.
These high rates require fast acting actuators and mechanical withstand
capability of the affected components.
8.7 Case study 4 – Modal Analysis of FS-IGIB System119

This section presents a linearized model for a simple system consisting of
an induction generator (IG) connected to an infinite bus through a linear
passive transmission network. The linearized model is obtained by linearizing
the third-order dynamic model presented in section 8.3.3.B. This model will
be used for studying the small signal stability of the induction generator and
modal analysis of the system as affected by the network parameters and
119
EL-Shimy M, Badr MA. Modal Based Analysis of Induction Generator Infinite-Bus
System. ASJEE. 2008; 2:249 - 61.
366
initial operating conditions. The considered network parameters are the series
impedance of the interconnecting network and the shunt capacitive
susceptance at the generator terminals.
A. System Modeling
The system under study consists of an induction generator connected to an
infinite bus (IGIB) through a linear passive transmission network as shown in
Fig. 8.63. This system is practically acceptable for modeling the integration
of wind-energy conversion systems utilizing induction generators with large-
scale power systems.
Fig. 8.63: Induction generator infinite-bus (IGIB) system
Fig. 8.64: 3rd order model for induction machine
Fig. 8.65: Equivalent circuit of induction machines
B. Nonlinear model for the IGIB system

The third-order model of the induction generator presented in section
8.3.3.B will be presented here using compact forms using the phasor or
matrix representations. The machine equations are based on the generator
367
current direction convention. The stator electromagnetic transients are

neglected using the 3rd order induction machine model in the synchronously
rotating reference frame axis, Fig. 8.64. Therefore, the stator windings
equations are represented by algebraic equations. The 3 rd order machine
model considers two rotor windings on the d- and q- axis. Hence, two state
variables define the electromagnetic transients of the rotor. The steady state
equivalent circuit of induction machines is shown in Fig. 8.65. Based on Fig.
8.64, and Fig. 8.65, the stator algebraic equations take the form (8.12) and
(8.13). These equations can also be represented in the phasor form (8.76) or
the matrix form (8.77). Equations (8.14) and (8.15) can be either combined in
the phasor form of (8.78) or the matrix form (8.79). The rotor
electromechanical transients can be represented by (8.80).
E' Vt  I s ( rs  jx' ) (8.76)
 E d'   rs x '   I d  V d 
 '  '      (8.77)
 E q   x rs   I q  V q 
 pE d'  1/ To' s o   E d'   0   I d 

 '   '     
 0   I q 
(8.79)
 q  
pE s  o 1/ T o   q  
'
E
(x ss  x ' )
where   , and p is a deferential operator (i.e. d/dt).
T o'
2Hps  Te  Tm (8.80)
Where
Te  Ed' I d  E q' I q (8.81)
A general linear passive network connecting the induction generator to an

infinite bus (IB) with voltage magnitude Eb and with zero phase-angle can be
368
represented as hybrid two-port network120. If, as shown in Fig. 8.66, the

induction generator is connected to bus 1 and the IB bus is bus 2, the hybrid
two-port network model of the network takes the form,
Vt  h11 I a  h12 Eb (8.82)
Fig. 8.66: General linear passive network connecting IG to IB
The parameters h11 and h12 are related to the reduced admittance matrix
of the network by the following equations; keeping in mind that induction
generator and IB buses are only the retained buses in that matrix:
1  Y21
h11  and h12  (8.83)
Y11 Y11
Generally, h11 and h12 can be expressed as:
h11  z R  jz I 
 (8.84)
h12  h1  jh2 
For induction machines, Kron’s transformation is not needed and the

machine can be directly analyzed in the common DQ-frame121. Therefore, the
following expression is generally valid,
Vd   z R  z I   I d   h1 
V      Eb  (8.85)
 q   zI z R   I q  h2 
The output active and reactive power supplied to the grid are presented by
120
K R Padyiar. Power System Dynamics: Stability and Control. Interline Publishing Pvt
Ltd; 1996
121
B. M. Nomikos, C. D. Vournas. Evaluation of Motor Effects on the Electromechanical
Oscillations of Multimachine Systems. IEEE Bologna PowerTech Conference, Bologna,
Italy 2003; (June 23-26)
369
Ptogrid  Vd I d  Vq I q (8.86)
Qtogrid  Vq I d  Vd I q (8.87)
Linearizing equations (8.77), (8.79), (8.80), and (8.85) and eliminating

the non-state variable Vd, Vq, Id, and Iq, the following state input equation is
obtained as,
 p E d'   c 5 c6 o E qo'   E d'   0 

      
 p E q    c 7
'
c8 o E do   E q    0   T m  (8.88)
' '
 p s   k 1    
   2H
k2 0   s   1 
2H   2H 
This takes the general state-space input equation px  Ax  Bu , where,
1
c5   c3 c 6  soo  c 4
T o'
1
c 7  soo  c1 c8   c 2
T o'
k1  I do  Edo
'
c1  E qo
'
c3 k 2  I qo  Edo
'
c2  E qo
'
c4
1
c1 c 2   rs  z R x '  z R 
c c    ' 
 3 4   x  z I rs  z r 
The initial conditions are calculated from the steady-state system

model with the initial slip is calculated separately using,
so  max ( so1 , so2 ) (8.89)
Vto2 rr rr Vto4
so1,2   1 (8.90)
 2Tmo xt2 xt 2 2
4Tmo xt
370
An accurate value of so can be obtained through an iterative procedure

by setting the initial guess of Vto to 1.0 p.u and updating it after solving the
system equations under steady-state conditions.
The linearized system equations are represented by the block diagram
of Fig. 8.67 which can be implemented on the SIMULINK for the linearized
time domain (TD) simulation of the system.
Fig. 8.67: Linearized IGIB system block diagram for linearized TD

simulation
Equation (8.88) describes the connection between of the rotor mechanical

equations and stator as well as the rotor electrical equations including the
network block. The terminal voltage and stator current block is obtained
eliminating Vd and Vq from (8.77) using (8.85) and linearizing the resultant
equation to get Id and Iq then Vd and Vq are obtained from the
linearization of either (8.85) or (8.77). The output power block is obtained
directly by linearizing (8.86) and (8.87). The terminal voltage and stator
current vectors are obtained from the relations,
Vt  Vd2  Vd2 (8.91)
371
 V 
 t   Vt  tan 1  q  (8.92)
 V q
I a  I d2  I q2 (8.93)
a   I a  tan 1 I q I d   (8.94)
C. Problem Formulation and Solution Algorithm

For ensuring dynamic stability as well as acceptable operating
conditions, the shunt-capacitor and the equivalent external reactance should
be carefully selected. This is can be stated as,
For a specified initial conditions

Search for bc and k e
s.t. (8.95)
Vtmin  Vt  Vtmax
real(eig(A ))  0
The value of the external line impedance (re+jxe) is defined as a percentage of

the induction generator transient impedance (rs+jx’s) by the variable ke
defined as:
re  jx e 
ke 
r
s  jx s'  (8.96)
The problem stated by (8.95) can be used in either the design stage or
for the analysis of an existing system. A general solution algorithm for (8.95)
is illustrated in the flowchart of Fig. 8.68, which is implemented on
MATLAB, with the terminal voltage limits of the induction generator are
assumed to be Vtmin = 0.9, Vtmax = 1.1
372
Fig. 8.68: IGIB SSSA algorithm Flowchart
D. System Parameters
The induction generator parameters data on 6.45 MVA, 25 kV, 50 Hz base
are listed in Table 8.25.
373
Table 8.25: Induction generator data

Parameter Value Unit
Stator resistance, rs 0.03 p.u
Stator leakage reactance, xs 0.27 p.u
Rotor resistance, rr 0.03 p.u
Rotor leakage reactance, xr 0.22 p.u
Magnetizing reactance, xm 8.06 p.u
Total inertia Constant*, H 3.6 sec
* Includes the wind turbine inertia constant
E. Results and Discussion

In this part, the results for the following cases are illustrated and
discussed:
 With infinite network strength (ke = 0, bc = 0) the effects of variations

of Tmo (i.e. initial conditions) on system stability are investigated.
 With finite network strength (ke > 0) the effects of bc on system
stability are investigated for various values of ke.
E.1. Infinite network Strength

The eigenvalues (’s) of the system as affected by the input mechanical
torque with Eb = 1 p.u are listed in Table 8.26 and their real-part values are
illustrated in Fig. 8.69. The maximum of the real-part of the eigenvalues and
the torque-slip relation are shown in Fig. 8.70 and Fig. 8.71 respectively.
Based on Table 8.26, for low input mechanical torque values up to 0.3 p.u,
only real eigenvalues exist; however,with further increase in the input
mechanical torque, two of the three eigenvalues become complex numbers.
The critical stability is obtained at 1.0 p.u input mechanical torque as also
depicted from Fig. 8.69, and Fig. 8.70. These results are in agreement with
the Te/s criterion which is based on the torque-slip characteristics of the
induction generator for a generator model that is based on the generator
current direction convention (shown in Fig. 9.71). The case where Tmo = 0.9
p.u is selected as a base-case for subsequent analysis.
374
Table 8.26: Eigenvalues (’s) of the system as affected by the input

mechanical torque with Eb = 1, ke = 0
Tmo (p.u) 1 2 3
0.001 -20.8466 +j0 -16.242 +j0 -5.3816 +j0
0.1 -21.246 +j0 -5.4474 +j0 -15.7768 +j0
0.3 -19.1165 +j0 -18.264 +j0 -5.0897 +j0
0.5 -19.1336 +j4.8918 -19.1336 -j4.8918 -4.203 +j0
0.7 -19.7408 +j8.4033 -19.7408 -j8.4033 -2.9885 +j0
0.9 -20.5146 +j13.1638 -20.5146 -j13.1638 -1.4409 +j0
1 -21.2351 +j20.6006 -21.2351 -j20.6006 0 +j0
Fig. 8.69: Real-part of the eigenvalues for infinite network strength
Fig. 8.70: Maximum of the real-part of the eigenvalues for infinite network
strength
Fig. 8.71: Torque-slip relation for infinite network strength

375
E.2. Finite network Strength

The effects of external network strength on the system stability and
generator terminal-voltage magnitude are investigated in two phases. First,
without shunt capacitor installed at generator terminals. Second, with shunt
capacitor installed at generator terminals.
With varying ke for zero shunt capacitive susceptance at generator
terminal, the eigenvalues of the system and the generator terminal voltage are
shown in Table 8.27. Fig. 8.72 shows the variations of both the maximum of
the real-part of the eigenvalues and generator terminal voltage for different
values of ke at bc = 0.
Fig. 8.72: Effect of ke on maximum of the real-part of the eigenvalues and

the generator terminal voltage for Eb = 1, bc = 0, and Tmo = 0.9
Table 8.27: Eigenvalues (’s) of the system as affected by ke with

Eb = 1, bc = 0, and Tmo = 0.9
ke re xe 1 2 3 Vto
0 0 0 -20.5146 +j13.1638 -20.5146 -j13.1638 -1.4409 +j0 1
0.1 0.003 0.048 -19.5752 +j12.9693 -19.5752 -j12.9693 -1.3506 +j0 0.9694
0.2 0.006 0.096 -18.7859 +j12.7755 -18.7859 -j12.7755 -1.2666 +j0 0.9389
0.3 0.009 0.144 -18.1128 +j12.5913 -18.1128 -j12.5913 -1.1863 +j0 0.9087
0.4 0.012 0.192 -17.5315 +j12.4205 -17.5315 -j12.4205 -1.1087 +j0 0.8792
0.5 0.015 0.24 -17.0236 +j12.2638 -17.0236 -j12.2638 -1.0332 +j0 0.8504
4 0.12 1.92 -10.9804 +j10.7593 -10.9804 -j10.7593 0.0105 +j0 0.3449
Based on Table 8.27 and Fig. 8.72, it is clear that the reduction in the
network strength (i.e. increase of ke) provokes both system stability and
generator terminal voltage in an approximate linear manner. With ke greater
than 0.3 p.u the generator terminal voltage drops below the minimum
376
allowable value which is 0.9 p.u. In addition, very high values of ke (i.e. very
weak external network) results on system instability.
Now, the combined effects of bc and ke on the system stability and the
generator terminal voltage are investigated for ke = 0.4 p.u (corresponding to
voltage level violation problem) and 4.0 p.u (corresponding to both voltage
level violation and stability problems). By varying bc with ke = 0.4 p.u, the
eigenvalues of the system and the generator terminal voltage are shown in
Table 8.28. Fig. 8.73 shows the variations of both the maximum of the real-
part of the eigenvalues and generator terminal voltage for different values of
bc at ke = 0.4 p.u.
Table 8.28: Eigenvalues (’s) of the system as affected by bc with

Eb = 1, ke = 0.4, and Tmo = 0.9
bc 1 2 3 Vto
0 -17.5315 +j12.4205 -17.5315 -j12.4205 -1.1087 +j0 0.8792
0.1 -17.4622 +j12.442 -17.4622 -j12.442 -1.1451 +j0 0.8942
0.2 -17.3903 +j12.4647 -17.3903 -j12.4647 -1.1832 +j0 0.9097
0.4 -17.238 +j12.5145 -17.238 -j12.5145 -1.2649 +j0 0.9424
0.8 -16.8946 +j12.6347 -16.8946 -j12.6347 -1.4534 +j0 1.0154
1 -16.5966 +j12.7489 -16.5966 -j12.7489 -1.6204 +j0 1.0779
1.2 -16.4883 +j12.7928 -16.4883 -j12.7928 -1.6816 +j0 1.1005
Fig. 8.73: Effect of bc on maximum of the real-part of the eigenvalues and the
generator terminal voltage for Eb = 1, ke = 0.4, and Tmo = 0.9
Based on Table 8.28 and Fig. 8.73 it is clear that the increase of bc
enhances both the generator terminal voltage level and the system stability.
Moreover, the maximum limit of bc is determined by the maximum allowable
voltage level at generator terminals. Therefore, it is clear that the role of
377
shunt capacitor bank at the induction generator terminals is not only enhances
the generator terminal voltage but also improves system dynamic stability;
however, this role is hindered by lowering external network strength as
shown in the following.
By varying bc at ke = 4.0 p.u, the eigenvalues of the system and the
generator terminal voltage are shown in Table 8.29. Fig. 8.74 shows the
variations of both the maximum of the real-part of the eigenvalues and
generator terminal voltage for different values of bc at ke = 4.0 p.u.
Table 8.29: Eigenvalues of the system as affected by bc with

Eb = 1, ke = 4.0, and Tmo = 0.9
bc 1 2 3 Vto
0 -10.9804 +j10.7593 -10.9804 -j10.7593 0.0105 +j0 0.3449
0.1 -9.8377 +j10.9833 -9.8377 -j10.9833 0.0812 +j0 0.3641
0.3 -6.0232 +j11.8514 -6.0232 -j11.8514 0.3761 +j0 0.4068
0.5 -1.7876 +j12.8235 -1.7876 -j12.8235 0.6748 +j0 0.4548
1 -7.2939 +j11.5301 -7.2939 -j11.5301 2.5833 +j0 0.5676
1.2 -4.6959 +j6.3858 -4.6959 -j6.3858 -0.8555 +j0 0.5846
1.3 -3.9976 +j1.9998 -3.9976 -j1.9998 -0.5162 +j0 0.5825
1.4 -9.2755 +j0 3.7197 -j0 -0.6827 +j0 0.5734
1.5 -11.3805 +j0 8.486 -j0 -0.4805 +j0 0.5581
Fig. 8.74: Effect of bc on maximum of the real-part of the eigenvalues and the
generator terminal voltage for Eb = 1, ke = 4.0, and Tmo = 0.9
Based on Table 8.29 and Fig. 8.74, it is clear that the increase of bc to 1.2
p.u increases the generator terminal voltage and brings the system to stable
operation. However, this stable operating point is not valid due severe
violation in the generator terminal voltage. Further increase of bc brings the
system back to unstable operating points with reduction the generator
378
terminal voltage which results from the voltage instability. Therefore, based
on Fig. 8.73 and Fig. 8.74, the role of shunt capacitor bank at the induction
generator terminals in enhancing both generator terminal voltage and system
stability is hindered by lowering the external network strength. As the
external network strength increases, properly selected shunt capacitor bank at
the induction generator terminals can provide good voltage regulation and
better stabilization. However, in case of very weak external network strength,
neither adequate voltage regulation nor stabilization effects of the capacitor
bank can be provided. In this case, acceptable operation of the system can be
achieved by distributed capacitive compensation over the transmission line.
For reasonable external network strength, it is recommended to use at the
terminals of the generator thyristor switched capacitor banks (TSC) with
optimized control algorithm instead of fixed shunt capacitor banks (FC). This
is can provide optimal stabilization in addition to providing adequate voltage
control for different operating points of the system.
E.3 Time Domain Simulation

The block diagram of Fig. 5 is implemented on SIMULINK for providing
time domain simulations of the system. The considered disturbance is a 10%
step increase in the mechanical power input at t = 1.0 sec. In order to
compare the system response to the prescribed disturbance, the following
cases are considered:
 Base-case: Eb = 1, Tmo = 0.9, bc = 0, ke = 0.

 Case#1: Eb = 1, Tmo = 0.9, bc = 1.2, ke = 0.4
 Case#2: Eb = 1, Tmo = 0.9, bc = 1.0, ke = 4.0
The variations of Te and s are shown in Fig. 8.75, the Te-s
trajectories are shown in Fig. 8.76, the variations of Ptogrid are shown in Fig.
8.77, and the variations in Qtogrid are shown in Fig. 8.78.
379
Fig. 8.75: Variations of Te and s
Fig. 8.76: Te-s trajectories
Fig. 8.77: Variations of Ptogrid

380
Fig. 8.78: Variations of Qtogrid
The time domain simulation as shown in Fig. 8.75 to Fig. 8.78 validate the
modal analysis results as the base-case and case#1 are stable and case#2 is
unstable. The following point presents a summary of the results and
conclusions,
1. The reduction in the network strength (i.e. increase of ke) provokes

both system stability and generator terminal voltage in approximately
linear manner.
2. The role of shunt capacitor bank at the induction generator terminals is
not only enhances the generator terminal voltage but also improves
system stability. However, this role is hindered by lowering external
network strength.
3. As the external network strength increases, properly selected shunt
capacitor bank at the induction generator terminals can provide good
voltage regulation and better stabilization. However, neither adequate
voltage regulation nor stabilization effects of the capacitor bank can be
provided in case of very weak external network. For very weak links,
distributed compensators instead of terminal compensators are
required.
4. The TSC at the generator terminals instead of the usual fixed
capacitors is recommended for optimized injection of the reactive
power at the generator terminals. As shown, the proper amount of the
capacitive compensation at the terminals is highly dependent on the
system parameters and the operating point.
381
The next chapter will present more analysis of the impact of wind power
on power systems. In addition, the combined impact of FACTS devices and
wind power on the dynamic stability of power systems will also be
demonstrated. The design of POD for enhancing the stability of power
systems that contain wind power sources will also be given.
382
Chapter 9
Dynamical Impacts of Wind Power and Damping
Enhancement
9.1 Introduction
In this chapter a series of case studies will be presented for
demonstrating some critical aspects about the dynamical characteristics of
power systems as affected by wind power. In the previous chapter, it is
shown that the FS-SCIG WT needs a controllable capacitive compensation at
its terminals for voltage control and stability enhancements. In addition,
FACTS devices may be used in power systems for many additional reasons
such as power flow control. Therefore, the impact of FACTS devices such as
SVCs and TCSCs on the dynamics of power systems in the presence of wind
power will also be presented. The chapter also includes the stabilization and
stability enhancement in power systems that contain wind power generators.
9.2 Impact of wind power on the inertia of power systems and

the equivalence SMIB
In chapter 5, the two-machine and single-machine equivalences of the
weakly interconnected areas of Fig. 5.57 is presented in section 5.5. The
power generation technologies in that system are totally conventional power
generators.
In this section, the same system is considered; however, wind power
generation is integrated with the system122 as shown in Fig. 9.1; 150 MW of
the conventional power generated in area 2 is replaced by wind power.
Therefore, the 700 MW power generations from G4 in the first scenario
becomes 550 MW in the second scenario. The wind farm is formed of 100
identical fixed-speed squirrel-cage induction-generator (FS-SCIG) based
122
383
wind turbine generators (WTGs) each of 2 MVA rating. The parameters of

the WTGs123,124,125 are listed in Table 9.1.
Table 9.1: Wind turbine and induction machine parameters

Parameter Value
SCIG DFIG
Number of blades 3
Rated wind speed, u [m/s] 16
Blade length [m] 75
Inertia constant of turbine Hwr [kWs/kVA] 2.5
Inertia constant of machine Hm [kWs/kVA] 0.5
Shaft stiffness for FSWT126 [p.u.] 0.3 ------
Pitch control gain [p.u] 10
Pitch control time constant [s] 2
Voltage control gain Kv [p.u] ------ 10
Power control time constant Te [s] ------ 0.01
Gear box ratio[int -] [41/89]
Number of poles 4
Rated voltage [kV] 0.69
Frequency [Hz] 60
Resistance of the stator, RS [p.u.] 0.01
Resistance of the rotor, Rr [p.u.] 0.01
Leakage inductance of the stator, xs [p.u.] 0.1
Leakage inductance of the rotor, xr [p.u.] 0.08
Mutual inductance, xm [p.u.] 3
123
El-Shimy M, Badr M, Rassem O. Impact of large scale wind power on power system stability.
In: Power system conference, 12th international middle-east power conference (MEPCON).
Aswan, Egypt: IEEE; 2008. p. 630–6.
124
El-Shimy M, Ghaly N, Abdelhamed M. Parametric study for stability analysis of grid-
connected wind energy conversion technologies. In: 15th International middle-east power
conference (MEPCON). Alex., Egypt: IEEE; 2010. p. 1–8.
125
EL-shimy M, Ghaly B. Grid-connected wind energy conversion systems: transient response. In:
Encyclopedia of energy engineering and Technology. 2nd ed., vol. IV. CRC Press, Taylor &
Francis Group; 2014. p. 2162–83
126
Multi-mass models are not considered for DFIG because of its controls which
significantly cancel the effect of shaft stiffness by decoupling the mechanical and
electrical power
384
Fig. 9.1: The two-area system with wind power integration
For the same disturbance stated in section 5.5 (i.e. A temporary 3-cycle
three-phase fault started at t = 1 sec applied on bus 8), and the PSAT is used
as a simulation environment, the system equivalent inertia at the interface
buses and the equivalent inertia of the SMIB equivalence of the system are
determined based on the method of section 5.2. The results are shown in
Table 9.2. The impact of the wind power on the equivalent inertia constants
and the transient reactances is illustrated in Fig. 9.2.
Table 9.2: Equivalent inertia (M) in sec, p.u transient reactances, and p.u
transient emf (E’); all values are on 100 MVA base
Area 1 equivalence Area 2 equivalence SMIB equivalence
’ ’
Inertia Reactance E Inertia Reactance E Inertia Reactance
29.15 0.0237 1.034 47.61 0.0243 1.050 18.08 0.0120
It can be seen from these results that the wind power generators alters the
frequency, transient, and steady state responses of the system. This is
illustrated by the changes in the equivalent inertia constant and the equivalent
transient reactance. The impact of wind power is significantly appearing on
area 2 where the wind power is available.
Since the equivalent inertia constant and the equivalent transient reactance
of this area are reduced, then the stability of the system is degraded by the
presence of wind power. This is can be demonstrated by determining the
maximum drop in the power generated by the same way presented section
385
6.4. Without wind power, the maximum drop in the generation is found to be
5.6 p.u while it becomes 4.9 p.u with the presence of the wind power.
(a)
(b)
Fig. 9.2: Impact of wind power on the equivalency; (a) equivalent inertia
constant; (b) equivalent transient reactance
The difference between the two values demonstrates the stability

reduction caused by the wind power inclusion in the system. Due to the
nature of the wind as an input energy source, the wind power is inherently
variable and also suffers from some intermittency. Therefore, a power system
with large amount of wind power is expected to be subjected to sustained
changes in the power generation. The magnitude and sign of these changes
depends mainly on the changes in the wind speed. Therefore, without a
386
feasible method for bulk energy storage, integrated large amounts of variable
generation (such as wind and solar-PV) in power systems may results in
significant provoking of the stability, the security (static and dynamic), the
reliability (or the power supply continuity), and the energy security. The
hydrogen as an energy storage medium and an energy carrier (see chapter 1)
may be considered as a perfect solution for the future system with very large
amounts of variable generation.
9.3 Impact of wind power on the stability of power systems

The behavior of a power system is significantly determined by the
behavior and the interaction of the generators that are connected to it. The
grid itself consists mainly of passive elements, which hardly affect the
behavior of the system, and as for the loads, only those in which directly grid
coupled motors are applied have a significant impact on the behavior of the
system.
In wind turbines, the generating systems differ from the conventional
directly grid coupled synchronous generator which is traditionally used in
power plants. Due to their different characteristics, these generating systems
interact differently with the power system in comparison with the
conventional synchronous generators. The change in this interaction
determined mainly by the degree of penetration of wind power generation in
the system.
In the analysis of wind generation-system interaction a distinction is made
between local and system wide impacts of wind power127,128 ; this shown in
Fig. 9.3. Local impacts of wind power are the impacts that occur in the
(electrical) vicinity of a wind turbine or a wind park. These impacts can be
attributed to a specific turbine or park, i.e. of which the cause can be
localized. These effects occur at each turbine or park, independently of the
overall wind power penetration level in the system as a whole. Wind power
locally has an impact on the following aspects of a power system: branch
flows and node voltages, protection schemes, fault currents and switchgear
ratings, harmonics, and flicker. System wide impacts, on the other hand, are
127
J.G. Slootweg, W.L. Kling, “Modelling and analysing impacts of wind power on
transient stability of power systems”, Wind Engineering, v.26, n.1, 2002, pp.3-20.
128
J.G. Slootweg, W.L. Kling, “The impact of large scale wind power generation on power
system oscillations”, Electric Power Systems Research, v.67, n.1, 2003, pp. -20
387
impacts of which the cause cannot be localized. They are a consequence of

the application of wind power that can, however, not be attributed to
individual turbines or parks. Nevertheless, they are strongly related to the
penetration level in the system as a whole. However, in opposition to the
local effects, the level of geographical spreading of the wind turbines and the
applied wind turbine type are less important. Apart from the local impacts,
wind power also has a number of system wide impacts, because it affects:
dynamics and stability, reactive power generation/voltage control
possibilities, and system balancing: frequency control and dispatch of the
remaining conventional units.
Fig. 9.3: Impacts of wind power on power systems
This section presents a detailed analysis of the impact of large scale FS-
SCIG based wind farms on both the transient voltage stability (TVS) and the
transient stability (TS) of electric power systems. With the PSAT as a
simulation environment, the following problems have been analyzed:
 Different penetration of wind power impact on TS and TVS following
a major fault in the transmission system,
388
 Determination of the acceptable wind power penetration level in power

systems without deteriorating both transient stability and voltage stability of
the system.
 The effect of SVC on the system stability is studied.
The mathematical modeling of various is as presented in chapter 8. The
IEEE 9-bus system shown in Fig. 9.4 represents the interconnected power
system used for this study. This system is considered the “original system”
and consists of 9 buses, 3 synchronous generator, 6 lines, 3 transformers, and
3 load centers. Each generator is equipped with an AVR of the standard
IEEE model 1 type with its block diagram shown in Fig. 9.5. The original
system data on a 100 MVA base obtained from the Anderson & Fouad
textbook129.
Fig. 9.4: The Original 9-bus system
Fig. 9.5: Standard IEEE Model 1 AVR

129
P.M. Anderson & A.A. Fouad, “Power System Control and Stability”, Galgotia
publications, 1981
389
The FS_SCIG wind farm has been modeled as an equivalent induction

generator that is connected to the original system at bus 7 via the two
transformers and a line as shown in Fig. 9.6. The SCIG parameters are shown
in Table 9.1.
Fig. 9.5: The 9-bus system with a wind farm connected at bus 7 and an
SVC connected to bus 8
Fig. 9.7: SVC Regulator
An SVC at bus 8 has been also added to the original as the reactive
power a controllable source. The SVC is modeled in the simulation program
as a limited linear controlled susceptance. The regulator has an anti-windup
limiter, thus within the operating limits the SVC performs automatic voltage
control at the connected node and outside the operating limits the SVC
susceptance is locked. The block diagram of the SVC regulator is shown in
390
Fig. 9.7. The SVC regulator parameters on a 100 MVA base are: the
regulator time constant, Tr = 10 sec.; the regulator gain, Kr = 100; the
reference voltage, Vref = 1.00 p.u.; the maximum susceptance, bmax = +1 p.u.,
the minimum susceptance, bmin = -1 p.u.
This study system can be considered as a simple model for studying
impact of large scale wind power on power system stability. After the load
flow analysis, the focus is to perform dynamic analyses to verify that the
system remains transiently stable under different contingencies and different
levels of wind power production. The system loads are kept fixed at the
values of the original system.
For setting a base for comparison between the system transient response
with and without wind generation, a 3 cycle (i.e. 0.05 sec, for 60 Hz) duration
fault is simulated on line 5-7 very close to bus 7 (the fault is started at t = 0.5
sec) in the original IEEE 9-bus system (i.e. 0.0 % penetration level, and no
SVC). The power angles and bus voltage magnitude (shown for buses 7 and 8
only to avoid crowding of plots) variations are shown in Fig. 9.8 and Fig. 9.8.
Based on these figures, following the applied fault the system is dynamically
stable and the system is able to restore the voltage.
Fig. 9.8: Original System (0.0 % penet. level, and no SVC) Power Angle
Variations
391
Fig. 9.9: Original System (0.0 % penet. level, and no SVC) Bus Voltage
Magnitude Variations
As a matter of investigation, the simulation is performed with the same

fault location and fault duration but with the SVC connected at bus 8 (0.0 %
penetration level, and with SVC). The power angles, and bus voltage
magnitude variations are shown in Fig. 9.10 to Fig. 9.11. In addition, the
variations of the SVC susceptance, voltage magnitude at bus 8, and the net
reactive power injected to bus 8 are shown in Fig. 9.12.
Fig. 9.10: Power Angle Variations (0.0 % penet. level, and with SVC)
392
Fig. 9.11: Bus Voltage Magnitude Variations (0.0 % penet. level, and with
SVC)
Fig. 9.12: Variations of b, V8, Q8 (0.0 % penet. level, and with SVC)
Based on these simulation results, and considering the results associated

with absence of the SVC, following the applied fault the system is transient
stable and the system is able to restore the voltage; however, the results show
a reduction in the damping of the system as a result of the inclusion of the
SVC. This appears as less settling time and larger magnitudes of oscillations
in the power angles. This impact is in agreement with the model analysis
presented in chapter 7 and these problems can be avoided by proper FACTS-
POD controllers. Although, the damping of the system is reduced, the
transient voltage stability is enhanced at least in terms of the initial post fault
voltage magnitudes.
393
Now, the impact of wind power generation on the system transient and
voltage stability is considered. Different penetration levels of wind power
generation are considered such that a part of the power produced by the
synchronous generator at bus 2 is replaced by an equal amount of wind
power generation at bus 12 and the system is simulated for the previously
described fault location and fault duration. The question to be answered is:
what is the limit of wind power penetration level to keep the system transient
and voltage stability for the considered contingency? The following cases are
considered:
 Case 1: No SVC, 0.63 p.u wind power generation (19.7 % total

penetration level, 24.55% of the replaced conventional synchronous
generation)
 Case 2: SVC, 0.63 p.u wind power generation (19.7 % total penetration
level, 24.55% of the replaced conventional synchronous generation)
 Case 3: SVC, 1.40 p.u wind power generation (43.8% classical
penetration level, 77.94% of the replaced conventional synchronous
generation)
The power angles, and bus voltage magnitude variations for each of these
cases are shown in Fig. 9.13 to Fig. 9.18.
Fig. 9.13: Case 1 - power angle variations
394
Fig. 9.14: Case 1 - bus voltage magnitude variations
Fig. 9.16: Case 2 - bus voltage magnitude variation
395
Fig. 9.18: Case 3 - bus voltage magnitude variations
Based on the responses of cases 1 and 2, it is clear that both the transient
voltage stability of the system is enhanced with the SVC installed; however,
the transient response of the power angles shows reduction in the system
damping. In both cases the system is stable. With penetration level of
77.94% (case 3), the system becomes unstable both in transient stability and
voltage stability. The instability phenomena can be viewed as an
insufficiency of reactive power compensation; this can be detected form the
variations of the SVC susceptance (shown in Fig. 9.19 and Fig. 9.20 for cases
2, and 3 respectively). Based on Fig. 9.20, it is clear that the SVC
susceptance hits its upper limit and locked at that value which means that the
reactive power limits of the SVC are reached. Therefore, the system
instability in this situation is a voltage instability phenomenon in nature
which is clear from Fig. 9.18. Based on the presented study, the maximum
penetration level is less than 43.8%.
396
Fig. 9.19: Case 2 - variations of b, V8, Q8.
Fig. 9.20: Case 3 - variations of b, V8, Q8.
For the investigation of the maximum penetration level, case 4 is

considered. In this case, the wind generation is set to be 1.2 p.u and the
generation of bus 2 is 0.4 p.u with the wind total penetration level of 37.55%,
which represents 60.13% of replaced conventional synchronous generation.
The penetration level of case 4 is determined by a trial and error approach
such that the system becomes critically stable at the stated penetration level.
The power angle variations and the bus voltage magnitudes are shown in Fig
9.21 and Fig 9.22. The variations of the SVC susceptance and reactive power
397
Fig 9.21: Case 4 - power angle variations
Fig 9.22: Case 4 - bus voltage magnitude variations
Fig 9.23: Case 4 - variations of b, Q8

398
Recalling that the wind power integrated to the original system is not an
added power generation but it is a replacement of the conventional power by
the same amount of the added wind power. In addition, with the absence of
the wind power, the original system is stable for the same disturbance
regardless of the impact of the SVC on the system damping. Therefore, the
instability of the system at the maximum wind power penetration level is
attributed to the characteristics of the FS-SCIG. The squirrel-cage induction
machines consumes reactive power reactive power either during motoring or
generating modes. Therefore, the steady-state reactive power loading on the
system increases with the increase in the penetration level. In addition,
further increase of reactive power loading occurs during the fault conditions
due to the reduction of the SCIG terminal voltage. Consequently, the system
fails to survive due to the transient voltage instability that is mainly caused
by the insufficiency of the reactive power. Providing more reactive power
during the fault by an SVC with higher rating is a possible way for enhancing
the transient stability of the system while the associated reduction in the
damping can be overcome by the use of PODs. This is will be illustrated later
in this chapter.
It should be noted that the maximum penetration level is highly
dependent on the point of connection of the wind farm to the system, the
considered contingency, and the WTG technology. For example, if the point
of connection of the considered wind farm becomes bus 9 instead of bus 7,
then 100% of the power generation at bus 3 can be replaced by wind power
generation without causing system instability.
9.4 Response of WTGs in comparison with Conventional

Synchronous Generators
In this section a detailed analysis of the steady state as well as the
transient response of FS-SCIG and DFIG wind power technologies in
compassion with conventional synchronous generators will be presented. In
addition, the effect of static and dynamic load models will also be considered.
The study system is this section shown in Fig. 9.24 in which a power plant
is connected to the stiff-grid through a transmission system and a local load.
The transmission system consists of two transformers and two parallel lines.
The local load is connected to the system through a transmission line. The
399
generators of the power plant are made of conventional synchronous

generators (SGs), or FS-SCIGs, or DFIGs. Therefore, the system shown in
Fig. 9.24 may be considered as three separate systems where the only
difference between them is the type of the generating technology. The
topology of the system is popular in many industrial loads where the local
power generation is installed for several reasons. These reasons include but
not limited for,
1. Enhancing the reliability of the grid as a main power source. The local
power source can either operates in a backup source (i.e. supplying
power only when the grid power is unavailable) or continuously
working in parallel with the grid. In this study, continuous parallel
operation is considered.
2. Reducing the energy consumption from the grid and consequently
reducing the electricity tariff. In this case, the power sources operate
continuously in parallel with the grid such that it provides a large
portion of the local load.
3. Supplying the local load with the possibility of exporting power to
either nearby loads or to the grid. In this case, the local generation
should be carefully sized for fulfilling the local load requirements as
well as energy export contracts with the nearby loads or the grid. The
local generation can be considered as a sort of an independent power
producer (IPP) and the power export to the grid is subjected to the
national feed-in-tariff structure. The grid in this case, purchases the
surplus local energy while the possible local power shortage is
purchased from the grid.
4. Apart of the active power production, the local generator may be
mainly used for modifying the power factor (PF) of the local load for
meeting juristic contracts with the grid as a part of the interconnection
requirements. This is common in large industrial loads where their low
power factor may affect the voltage regulation and stability of the grid.
400
Fig. 9.24: The study system
Conventional generation or renewable generation technologies may be

used for the local power generator. The need of local energy storage is
determined by the objectives of installing it and the degrees of the variability
and intermittency of the primary source of energy. For example, a power
source for enhancing the power supply reliability of the local load should
itself be characterized by a high reliability. If the utilized power source is
variable or intermittent, then sufficient local energy storage should be
provided for fulfilling the target reliability level. The variability and
intermittency (see chapter 1) are not only characterizing some renewable
energy sources such as wind and solar resources but also characterize
conventional energy sources. For example, if the local generator utilized
petroleum oil and if the local load is located in remote areas relative to the
location of the oil supply. Therefore, oil periodic transportation and local oil
storage are required for feeding the generator. Consequently, the degree of
the availability of the oil defines the variability and intermittency of the local
power source.
The study system is modeled and simulated through the PSAT simulation
software tool. The models of the generating technologies and loads are
available in the previous chapter as well as the manual of the PSAT. On 100
MVA bases, Table 9.3 shows the p.u parameters of the synchronous
generator while Table 9.1 shows the parameters of the SCIG and the DGIG.
The lines and transformer data are shown in Table 9.4. The impressed load on
the system (connected to bus 5) is 4.0 +j 1.0 p.u. Various static and dynamic
401
load models are considered for evaluating the system performance, studying
the impact of load model variations, and comparing the generating
technologies. The parameters of the considered loads are shown in Table 9.5.
Table 9.3:Synchronous generator data.

Parameter Value
Machine dynamic order 4
Stator resistance (ra) 0.0000
Leakage reactance (xl) 0.0000
[0.8958, 0.1198,
Direct-axis steady-state reactance [xd, x'd, x''d]
0.0000]
Direct-axis open circuit time constants in sec. [T'do, T''d0] [6.00, 0.00]
[0.8645, 0.1969,
Quadrature-axis reactance [xq, x'q, x''q]
0.0000]
Quadrature-axis open circuit time constants in sec. [T'qo,
[0.5350, 0.0000]
T''qo]
Inertia in sec. (M=2H) and damping in p.u [12.80, 0.00]
Table 9.4: Data of lines and transformers

Value
Parameter
L2-3(1) and L2-3(2) L3-5 T1 T2
Length in km 10 3 - -
Reactance in p.u (H/km) 0.002 0.001 0.0625 0.0576
Resistance in p.u (ohm/km) 0.0002 0.0001 0 0
Susceptance in p.u (F/km) 0.0035 0.00175 - -
Bus 4 is modeled as a PV-bus with a unity voltage setting while the

maximum active power setting is selected for this bus. The maximum active
power injected to bus 4 is chosen such that the system will remain stable after
being subjected to a specified disturbance. The considered disturbance in this
case is a 3-cycle temporally three-phase fault at the midpoint of L2-3(2)
started at t = 0.5 sec. The maximum power values for each of the load models
shown in Table 9.5 are shown in Table 9.6 and Fig. 9.25.
In the following sections, both static and dynamic analysis of the system
will be presented. The static analysis includes load flow (LF) and static
voltage stability (SVS) while the dynamic analysis includes transient stability
(TS), dynamic voltage stability (DVS), and small-signal stability (SSS).
402
Table 9.5: Load models

Case ID
Static 1 Constant impedance model: α=2.0 , β= 2.0
models130 2 Composite load model: α=1.5 , β=2.5
70% composite static load model + 30% induction
3
Dynamic motors with the 3rd order model
models 30% composite static load model + 70% induction
4
motors with 3rd order model
It is clear from Table 9.6 and Fig. 9.25 that maximum power associated
with SGs and DFIGs is sensitive to the load models while the SCIG is not
sensitive. The presence of the induction motor load causes a decline in the
maximum power level of the SG and increase for DFIG. For SG, the
reduction in the maximum power level increases with the increase in the
amount of induction motor loads while the DFIG is not significantly sensitive
to that issue. The impact of the amount of induction motor loads on the
maximum power of the FS-SCIGs is negligible. It is also depicted that the
synchronous generator provides the highest maximum power followed by the
DFIG then the SCIG. The results also show that the maximum power setting
for all generators and all load models are 0.6 p.u which is limited by the
SCIG. Fixing this power setting to allow proper comparison between various
generation technologies in the following studies. This is because the only
difference between the systems is the technological structure of the
generators.
Table 9.6: Maximum power settings
Max. Power level
Case
SG DFIG SCIG
Case 1 3 1 0.6
Case 2 3 1 0.6
Case 3 2.7 1.3 0.6
Case 4 2.6 1.3 0.6
130
The static load is represented by the exponential model given by and
β
where P, and Q are respectively the load active and reactive power at a
load bus voltage V while Po, and Qo are respectively the load active and reactive power at
an initial voltage Vo (which is usually 1.0 p.u). For more details, see Appendix 3.
403
Fig. 9.25: Maximum power settings
A. Load flow analysis

The load flow analysis is an inspection of the static performance of power
systems. In the standard load flow analysis, the loads are represented by a
constant PQ-model while the generators are represented by PV-buses. The
stiff node is defined as a standard independent voltage source (i.e. as a
constant voltage magnitude and phase angle) for representing a very large
system in comparison with the considered system. In the load flow analysis,
the stiff node acts as a slack or a swing bus. The principal information of
power flow analysis is to find the magnitude and phase angle of voltage at
each bus and the active and reactive power flowing in each transmission line.
This is valuable in checking the intrinsic and operational limits of the system
(see chapter 1) under the given operating conditions. In this analysis the
maximum acceptable bus voltage deviation is 10%. Numerical iterative
techniques are usually used for solving the load flow problem. In this section,
the Newton-Raphson’s method is used for solving the load flow problem
while the trapezoidal rule is used for the numerical integration in the TD
analysis.
Fig. 9.26 shows the power flow results for all load models for various
generator types. The results show that acceptable voltage levels are obtained
for various conditions. Recalling that the local load impressed on the system
is 4.0 +j 1.0 p.u at unity voltage at bus 5; however, this value is changed
according to the actual bus voltage magnitude and the model of the load. The
results show that the magnitude of bus 5 voltage is highly dependent on the
404
load model. This is because the power consumption of various load models is
sensitive to the bus voltage by different relations. As can be seen, the
dynamic load models have higher impact on the bus voltage magnitudes in
comparison with the static load models. Consequently, the drop in the load
bus voltage is higher with the dynamic loads in comparison with the
considered static loads. Due to the presence induction motors, the reactive
power consumption is very high in comparison with the considered static
loads. Since the generator bus is represented as a PV-bus, the impact of the
technologies of the generators and their actual capability limits on the power
flow are absent.
(a)
(b)
Fig. 9.26: Some power flow results
B. Transient analysis
In this section, the impact of the fault Clearing Time (CT) on the TS and
the TVS of the system are considered. In addition, the fault Critical Clearing
405
Time (CCT) is determined. The considered disturbance is a temporarily 3-

phase fault at midpoint of the line L2-3(2); see Fig. 9.24. Various load
models and generation technologies are considered. The fault starts at t = 0.5
sec and the CT is 3-cycles (0.06 sec). Fig. 9.27 shows the effect of load
models with the same generation technology on the bus voltages while Fig
9.28 shows the effect of various generation technologies with the same load
model on the bus voltages.
(a) (d)
(b) (e)
(c) (f)
Fig. 9.27: Transient voltage response as affected by load models for the same generation
technology; a, b, and c: generator terminal voltage; d, e, and f: load terminal voltage
406
(d) (e)
(e) (f)
(f) (g)
(g)
(h)
Fig. 9.28: Transient voltage response as affected by generating technologies for the same load
model; a, b, c, and d: generator terminal voltage; e, f, g, and h: load terminal voltage
It can be seen from these results that the transient voltage responses
associated with the considered static loads models is nearly identical. The
transient voltage responses associated with the dynamic load models are
highly different than that of the static load models. In addition, the dynamic
load models cause significant degradation in the transient voltage responses
407
in terms of the transient deviations, and the settling time of the transient. It is
also shown that the negative impacts of the dynamic load models on the
transient voltage response increases with the increase in the percentage of the
induction motor load in the composite load structure.
For the considered disturbances, stable equilibrium points are reached in
the post-fault conditions. Therefore, the stated CT is less than the CCT of any
of the load and generation technological conditions. It is also clear that the
transient voltage response of the DFIG is better in comparison with the
conventional SG and the FS-SCIG. The worst transient voltage responses are
associated with the SCIG. This is can be attributed to the voltage control
capability of the DFIGs and the SGs. On the other side, the SCIGs cannot
provide either a controllable reactive power neither they have a reactive
power capability131,132,133. As these generators are equipped with fixed
capacitors at their terminals, their voltage support during steady state
conditions is limited while they did not provide contributions in enhancing
the transient voltage support. The fixed capacitors of the SCIGs are usually
sized based on the initial conditions for providing an acceptable terminal
voltage magnitude by reducing the reactive power loading on the system. As
stated in the previous chapter, controllable reactive power support (for
example, SVCs or STATCOMs) is recommended to be provided to the FS-
SCIG for enhancing the voltage control as well as the system stability.
The transient responses of the active and reactive power of various
generators as affected by load models are shown in Fig. 9.29 while Fig. 9.30
shows comparative transient power responses between various generating
technologies for the same load models.
The comments on the transient voltage responses are also applicable to the
transient responses of the active and reactive power; however, the impact of
the FS-SCIG is clearly illustrated. The results show that the SCIG presents an
initial high reactive power loading during the fault period and its remains
131
M. Ahmed, M. EL-Shimy, and M. Badr, “Advanced modeling and analysis of the
loading capability limits of doubly-fed induction generators,” Sustainable Energy
Technologies and Assessments, vol. 7, pp. 79-90, 2014.
132
EL-Shimy M. Reactive Power Management and Control of Distant Large-Scale Grid-
Connected Offshore Wind Power Farms. International Journal of Sustainable Energy
(IJSE), 2012. Available online: Mar 20, 2012. Volume 32, Issue 5, pp. 449 - 465, 2013.
133
M. EL-Shimy, “Modeling and analysis of reactive power in grid-connected onshore and
offshore DFIG-based wind farms”, Wind Energy, 17:27 – 295, 2014
408
performing as a reactive power loading during the post fault transients. This
is attributed, as previously stated, to the reactive power characteristics of the
induction machines. This also explains the degraded transient voltage
responses associated with the SCIGs.
(a) (e)
(f)
(b)
(c) (g)
(d) (h)
Fig. 9.29: Transient active power responses of the generators as affected by load models for
the same generation technology; a, b, c, and d: active power; e, f, g, and h: reactive power
409
On the other hand, both SGs, and DFIGs presents a momentary reactive
power loading on the system during the fault period. Afterward, the voltage
(or reactive power) control actions of these generators provide a reactive
power support to the grid. Consequently, the transient voltage responses
associated with these generators are better in comparison with the SCIG. In
addition, the voltage collapse with these generators is less likely in
comparison with the FS-SCIGs.
(a)
(d)
(b) (e)
(c) (f)
Fig. 9.30: Transient active power responses of the generators as affected by generation
technologies for the same load model; a, b, and c: active power; d, e, and f: reactive power
The determination of the CCT for the stated faults is determined by

increasing the CT till the instability is detected. Table 9.7 and 9.31 shows the
value of CCT for various load models and various generation technologies.
410
Table 9.7: CCT values

Case SG DFIG SCIG
Case 1 0.99 -----134 1.23
Case 2 0.99 ------ 1.23
Case 3 0.47 1.42 0.65
Case 4 0.21 0.22 0.1
1.60
1.40
SG
1.20
1.00 DFIG
CCT (s)
0.80 SCIG
0.60
0.40
0.20
0.00
case 1 case 2 case 3 case 4
Case ID
Fig. 9.31: Critical clearing time for 3 based system with the different load
models
The results show that the CCT is highly dependent on the generation
technologies and the load models. In addition, as shown in the previous
chapters, the CCT is also highly dependent on the initial conditions, the
system inertia, and the strength of the grid-coupling link. With the static load
models, the DFIG remains stable till the CT is set to 20 sec while the other
technologies show a specified CCT. The CCT of the SCIG is higher than the
CCT of the synchronous generator. This s can be attributed to the impact of
the AVR which for some parameters should be equipped with PSS for
stability enhancement of the SGs. The changes in the parameters of the static
load model did not show a significant impact on the CCT. With the dynamic
load models considered, the CCT is generally reduced with the increase of
the amount of induction motors in the load composition.
9.5 Design of PODs in the presence of wind power generation

The basic traditional methods for the design of PODs are presented and
implemented in chapter 7. In this chapter, the frequency response method will
134
Till CT = 20 sec, the system stable
411
be utilized for the POD design. The presence of wind power, as previously
explained, alters the dynamics of the systems and causes changes in the
eigenvalues. Therefore, no additional POD design techniques are required for
the design of POD considering the impact of wind power; however, the
impact of wind power on the system dynamics will demonstrated via the
modal analysis for more clarification of their impacts.
In this section, the impact of wind power on the dynamic stability of
weakly interconnected power systems and the design of the required PODs
are considered135,136,137. Two situations are covered. The first one is the
replacement of conventional power by wind power while the second one
includes the addition of wind power to an existing conventional power
generation system.
The considered system is composed of the two weakly interconnected
areas of chapter 7. In this system, the area 1 is exporting power to the area 2.
This situation defines the main objective of integrating the wind power. As
shown in Fig. 9.32, the wind power is integrated to area 1 for reducing the
fuel consumption and consequently reducing the GHG emissions. This
objective is based on the fact that area 1 has surplus power. On the other
hand, area 2 is importing power from area 1 over a weak link. This is due to
the shortage in the power generation in area 2 in comparison with its loads.
Therefore, the main objective of the wind power integration in this case is the
reduction of the dependency of area 2 on area 1. This is achieved also via
clean energy sources; however, the main problem associated with wind
power is the availability of the wind resource. In the following analysis, the
wind power integration to area 1 will be considered as first option of wind
power integration while the wind power integration to area 2 is considered as
the second option. One option will be considered at a time i.e. the analysis
135
M. Mandour, M. EL-Shimy, F. Bendary, and W.M. Mansour. Design of power
oscillation damping (POD) controllers in weakly interconnected power systems including
wind power technologies. Accepted for publications in the JEE, 2015
136
M. Mandour, M. EL-Shimy, F. Bendary, and W.M. Mansour. The Design of POD
Considering Conventional and Renewable Power Generation. International Electrical
Engineering Journal (IEEJ), Vol. 6 (2015) No.7, pp. 1962-1972
137
M. Mandour, M. EL-Shimy, F. Bendary, W.M. Mansour. Impact of Wind Power on
– 8, 2015, Cairo, Egypt.
412
and the POD design will consider a situation where only one option will be
implemented.
Fig. 9.32: The two-area system with wind power integrations
For the same reasons stated in chapters 7 and 8 as well as in this chapter,
the traditional fixed capacitors of the original system (Fig. 7.18) are replaced
by SVCs as shown in Fig. 7.13 and 9.32. This is also important for handling
the reactive power support needed with the presence of wind power
generators. Both FS-SCIG and DFIG technologies are considered. The
modeling of various technologies is presented in chapter 8 while the data of
various WTGs as well as the system are shown in Appendix 8.
A. Sizing of the wind farms

In this context, the maximum wind penetration (MWP) is defined as the
maximum value of the wind power after which the system will lose its
stability due to the impact of the wind power. Two methods can be applied
for the determination of the maximum wind penetration level. The first
method depends on the response of the system for a specific fault defined by
three parameters; location, duration, and type. In this context, this method
will be called the time-domain simulation (TDS) method. The main problem
associated with the TDS method is the high dependency of the determined
MWP level on the chosen disturbance as well as the parameters of the
413
disturbance. The second method overcomes this problem. In the second

method, the MWP level is determined based on the modal analysis of the
system. In this case, the MWP level is defined as the maximum amount of
integrated wind power before provoking the small-signal stability of the
power system. Although, the evaluation of the stability of this system in this
method is limited to small disturbances, the results are independent on a
specific disturbance. Of course, the MWP level is highly dependent on many
factors such as the WTG technology, the topology of the hosting system and
its parameters as well as the considered operating. Usually, the peak loading
level of the system is considered as the design operating point and also the
point at which the MWP level is determined. Fig. 9.33 shows the various
values of the MWP level considering the location of the wind farm, the WTG
technology, and the operation philosophy. In the conventional power
replacement philosophy, the amount of wind power added to area 1 is
reflected as a reduction in the conventional power generation in that area. In
the wind power addition philosophy, the wind power added to area 2 do not
affect the original convention power generation in that area. The results show
that the MWP level of the FS-SCIG is the same for the power replacement
and power addition options. On the other side, higher power in the power
addition option with DFIGs is achieved in comparison with the power
replacement option of the same WTG technology. For both options, the
MWP level of the DFIG is much higher than the FS-SCIG MWP level. This
attributed to the better control and stability characteristics of the DFIGs in
comparison with the FS-SCIGs.
600
MWP level (MW)
500
400
300
200
100
0
FS-SCIG DFIG FS-SCIG DFIG
“Power replacement” “Power addition”
MWP level for area 1 MWP level for area 2
Fig. 9.33: MWP level
414
B. Modal analysis
The system without any wind power integration is considered for
inspecting the impact of the wind power on the eigenvalues of the system.
The dominant eigenvalues as well as the frequency and the damping of the
dominant modes of the system without wind power integration are shown in
Table 7.12 and also shown here in Table 9.8 for the convenience of the
reader.
For the MWP level shown in Fig. 9.33 and for values slightly higher than
them, the dominant eigenvalues as well as the frequency and the damping of
the dominant modes are shown in Tables 9.9 and 9.10 for the power
replacement and power addition options of the FS-SCIG respectively. Tables
9.11 and 9.12 shows the same quantities for the power replacement and
power addition options of the DFIG.
Table 9.8: Dominant eigenvalues and participation factors

without wind power integration
f  Status of
Eigenvalues
-0.52663j6.6178 9.50.. 8% Unacceptable

-0.5382j6.8191 9.5114 7.92% Unacceptable
-0. 0.03599+j3.7071 5.01551 0.45% Critical
Table 9.9: Dominant eigenvalues and participation factors for power

replacement in area 1 by FS-SCIGs
f  Eigenvalue
Eigenvalues
(Hz) (%) Status
P1+P2=1260MW -0.59442j6.5443 9.5701 9.04% unacceptable
P12=140MW -0.15025j3.821 5..51. 3.9% Critical
-0.57982j6.5303 9.5717 8.88% unacceptable
P1+P2=1250MW -0.14699j3.8239 5..515 3.86% Critical
P12=150MW -0.42722j7.0026 9.99.. 6.08% unacceptable
0.12483+j0 0 ------- Unstable

addition in area 2 by FS-SCIGs
f  Eigenvalue
Eigenvalues
(Hz) (%) Status
P12=140 MW -0.59524j6.5765 9.509 %9.02 unacceptable
415
-0.17052j3.9459 5..11. %4.31 Critical

-0.60136j6.5699 9.50 %9 unacceptable
-0.16879j3.9522 5..1101 %4.3 Critical
P12=150 MW
-0.57935j6.9841 9.9907 %8.28 unacceptable
0.17508+j0 0 ------- Unstable

replacement in area 1 by DFIGs
f  Eigenvalue
Eigenvalues
(Hz) (%) Status
P1+P2=900MW -0.81087j6.2776 9.5547 12.8% acceptable
P12=500MW -0.20784j4.0177 5..7511 5.16% Critical
-0.85172j6.2167 5.111.. 13.5% acceptable
P1+P2=850MW -0.22005j4.0365 5..7111 5.4% Critical
P12=550MW -0.12055j1.7216 5.147.4 7.04% unacceptable
0.008 j0 0 ------ Unstable

addition in area 2 by DFIGs
f  Eigenvalue
Eigenvalues
(Hz) (%) Status
P12=350 -0.78117j6.3636 9.5157 %12.3 acceptable
MW -0.27893j4.182 5...45. %6.65 Critical
-0.83732j6.2965 9.5951 %13.06 acceptable
P12=400 -0.30421j4.2103 5..4911 %7.1 Critical
MW -0.22003j1.3624 5.191.7 %16.6 acceptable
0.000j0 --------- -------- Unstable
7
Damping of the critical mode (%)
6
5
4
3
2
1
0
Replacement
Replacement
Addition
Addition
Two SVCs
FCs; No SVCs
No Wind FS-SCIG DFIG
Fig. 9.34: The damping of the critical mode
416
The impact of various options of the wind power integration as well as the
technological impact of the WTGs on the damping of the critical mode of the
system are shown in Fig. 9.34. The figure also shows the damping of the
critical modes without wind power integration; in this case, the impact of the
SVCs (see chapter 7) is also demonstrated.
It can be seen from these tables and Fig. 9.34 that:
1. The SVCs causes a decline in the system damping in comparison with
the fixed capacitors (FCs). It should be noted that the flexibility in the
reactive power support and voltage control provided by SVCs is of
high importance in the secure operation of power systems. The
problem associated with the reduction of damping can be solved by the
POD as a supplementary SVC controller for enhancing the system
stability.
2. The wind power technologies enhance the system damping if their
penetration level is less than the MWP level. Increasing the wind
power penetration level to values higher than the MWP level causes
instability of the system. This is can be attributed to the reduction of
the system inertia caused by the wind power technologies (see section
9.2). Therefore, the integration of wind power causes two opposing
effects. The positive effect is the enhancement in the system damping
while the negative effect is the reduction in the system inertia.
Therefore, even if the total power generation is unchanged due to
replacing conventional power by wind power, the replacement level is
limited due to the opposing impacts of wind power on the system
stability. This is also demonstrated in section 9.2.
3. The impact of DFIGs on the system damping is better in comparison
with the FS-SCIGs. In addition, the power addition option has a better
impact on the system damping in comparison with the power
replacement option.
For the given study system, none of the wind power options could
enhance the minimum system damping to an acceptable level (i.e. 10% or
more). Therefore, PODs are designed for various options and system
417
structures. The design is based on the same algorithm provided in chapter 7

and acceptable minimum damping ratios are achieved 138,139.
138
M. Mandour, M. EL-Shimy, F. Bendary, and W.M. Mansour. The Design of POD
Considering Conventional and Renewable Power Generation. International Electrical
Engineering Journal (IEEJ), Vol. 6 (2015) No.7, pp. 1962-1972
139
. Mandour, M. EL-Shimy, F. Bendary, W.M. Mansour. Impact of Wind Power on
– 8, 2015, Cairo, Egypt
418
Chapter 10
Overview, Characterization, and Modeling of Solar-PV
Systems
10.1 Introduction
The solar energy is classified as one of the promising renewable-energy
worldwide. In the power and energy discipline, the sunlight energy can be
involved in energy conversion processes through three main evolving
technological categories. As shown in Fig. 10.1, these categories are
Photovoltaics (SOLAR-PV), Concentrated Solar Power (CSP), and Solar
Heating and Cooling (SHC). This classification is based on the energy
conversions processes involved in the solar energy conversion stream.
Fig. 10.1: Fundamental classification of solar energy technologies
The classification shown in Fig. 10.1 illustrates only the fundamental

categories of solar energy technologies; however, each of the shown classes
can be classified to a large number of technologies. These three ways of
harnessing the sun are complementary. Therefore, developers should
carefully assess their needs, project constraints, meteorological conditions,
and environmental impact when choosing a specific solar technology to
419
use140. This chapter focuses on the solar-PV technologies, which generates

electricity through the direct conversion of sunlight.
Over the last decade, solar-PV technologies have shown the potential to
become a major source of power generation in the world with a strong and
continuous growth even during times of financial and economic crisis. This
growth is expected to continue in the years ahead as worldwide awareness of
the advantages of solar-PV increases. Now, solar-PV is the third most
important renewable energy in terms of global installed capacity (after hydro
and wind power). At the end of 200 , the world’s SOLAR-PV cumulative
installed capacity was approaching 23 GW 141. In 2010, the global
photovoltaic market was almost doubled; it was exceeding 40 GW 142. In
2011, more than 70 GW are installed globally and could produce 85 TWh of
electricity every year. This energy volume is sufficient to cover the annual
power supply needs of over 20 million households143. The year 2012 was
another historic one for solar photovoltaic (solar-PV) technology, which has
experienced outstanding growth over the past decade. The world’s
cumulative solar-PV capacity is currently more than 102 GW 144. This
capacity can annually produce at least 110 TWh of electricity and save more
than 53 million tones of CO2.
Solar energy is expected to play a crucial role in meeting future energy
demand through clean energy resources. Existing studies expect the long-
term growth (e.g., until 2050) of solar energy vary widely based on a large
number of assumptions 145. For example, EPIA/Greenpeace produces the most
ambitious forecasts for future solar-PV installation146. The study assumes that
if sufficient market support mechanisms are provided, a dramatic growth of
140
El Chaar, L., & El Zein, N. (2011). Review of photovoltaic technologies. Renewable
and Sustainable Energy Reviews, 15(5), 2165-2175.
141
Ecogeneration, (2010). Soaring to new heights:Solar PV in Australia. In
ecogeneration pp. 1-53, Great Southern Press Pty Ltd, Melbourne.
142
Solarbuzz. Available from: www.solarbuzz.com
143
Masson, G., Latour, M., & Biancardi, D. (2012). Global market outlook for
photovoltaics until 2016. European Photovoltaic Industry Association, 5.
144
Masson, G., Latour, M., Rekinger, M., Theologitis, I. T., & Papoutsi, M. (2013).
Global market outlook for photovoltaics 2013-2017. European Photovoltaic Industry
Association, 12-32.
145
Timilsina, G. R., Kurdgelashvili, L., & Narbel, P. A. (2012). Solar energy: Markets,
economics and policies. Renewable and Sustainable Energy Reviews, 16(1), 449-465.
146
European Photovoltaic Industry Association (EPIA). (2011). Solar generation 6: Solar
photovoltaic electricity empowering the world. Retrieved February, 23, 2012. Greenpeace
International & European Photovoltaic Industry Association: Brussels, Belgium.
420
solar-PV would be possible, which will lead to worldwide PV installed

capacity rising from around 40 GW in 2010 to 737 GW in 2020 to 1845 GW
in 2030 to 3256 GW in 2040 to 4669 GW by 2050. The capacity would reach
over 1000 GW in 2030 even with a lower level of political commitment.
Another study by the International Energy Agency (IEA) 147 estimates
solar power development potential based on two scenarios that are
differentiated on the basis of global CO2 emission reduction targets. In the
first scenario, where global CO2 emission in 2050 are restricted at 2005
level. In this case, the global solar-PV capacity is estimated to increase from
11 GW in 2009 to 600 GW in 2050. In the second scenario, where global
CO2 emission is reduced by 50% from 2005 levels in 2050. This is estimated
to approach or exceed 1100 GW capacity in 2050.
Like solar-PV, predictions are available for CSP technology. The
International Energy Agency (IEA) predicts that CSP capacity could reach
380 GW to 630 GW, depending on global targets for GHG mitigation. The
lower value represents to the scenario of limiting global CO2 emissions in
2050 at 2005 level, whereas the upper value refers to the scenario to reduce
global CO2 emissions in 2050 by 50% from 2005 levels. A study by
Greenpeace, The European Solar Thermal Power Industry Association
(ESTIA), and the International Energy Agency (IEA) 148 expected that the
global CSP capacity would expand by one hundred-fold to 37 GW by 2025
and then skyrocket to 600 GW by 2040. On the same way another study by
the Greenpeace, and International and European Renewable Energy Council
(EREC) expected that global CSP capacity could reach 29 GW in 2020, 137
GW in 2030 and 405 GW by 2050 149.
Chapter 10, 11, and 12 are complementary to each other and can be used
as course handling solar-PV generators. Their main objective is to present
overview, modeling, simulation, control, and grid-interconnection
requirements of solar-PV systems. They may be considered as an extended
147
OECD/IEA (2008). Energy Technology Perspectives: Scenarios and Strategies to 2050;
International Energy Agency: Paris, France.
148
Aringhoff, R.; Brakmann, G.; Geyer, M.; Teske, S.; Baker, C (2005). Concentrated
Solar Thermal Power- Now! Greenpeace International, European Solar Thermal Power
Industry Association (ESTIA): Amsterdam, the Netherlands, IEA SolarPACES
Implementing Agreement.
149
Teske, S.; Zervos, A.; Scha¨fer, O (2007). Energy revolution: a sustainable world
energy outlook; Greenpeace International-European Renewable Energy Council (EREC):
Amsterdam, the Netherlands.
421
chapter (or part of this book). Therefore, sometimes the phrase ‘this chapter’
will be used for referring to a specific one of these chapters or all the three
chapters. Chapter 10 provides an overview of various solar-PV technologies.
In addition, detailed quantitative characterization of various technologies is
presented. Various models related to the solar-PV projects from the electrical
energy prospective are given. In addition, dynamic modeling and analysis of
grid-connected solar-PV power plants are provided in details in chapter 11.
An enhanced Maximum Power Point Tracking (MPPT) technique is also
presented in chapter 11. Analysis of the grid codes for interconnecting solar-
PV and wind power generators will be presented in chapter 12.
10.2 Qualitative Overview of solar-PV technologies

The history of solar-PV technologies starts in 1839 150, when
Alexander-Edmund Becquerel observed that electrical currents arose from
certain light induced chemical reactions. Later on in 1877, the photovoltaic
effect in solid Selenium was observed by Adams and Day. Fritz in 1883
developed the first photovoltaic cell and its efficiency was less than 1%.
Discontinuous evolution of PV technologies is observed since those ancient
times. Collapse and growth rates in the PV industry have been observed in
many periods. For example, the years immediately following the oil-shock in
the 1970s saw much interest in the development and commercialization of
solar energy technologies; however, this incipient solar energy industry of the
1970s and early 1980s has been collapsed due to the sharp decline in oil
prices and a lack of sustained policy support 151. Solar energy market growths
as well as technological evolution have regained momentum since the early
2000s, exhibiting phenomenal growth recently. The total installed capacity of
solar based electricity generation has increased to more than 40 GW by the
end of 2010 from almost negligible capacity in the early nineties 152.
150
Chapin, D. M., Fuller, C. S., & Pearson, G. L. (1954). A new silicon p‐n junction
photocell for converting solar radiation into electrical power. Journal of Applied Physics,
(25), 676-677.
151
Bradford, T. (2006). Solar revolution: the economic transformation of the global
energy industry. MIT Press Books, 1.
152
REN21. Global Status Report 2005–2011; Renewable Energy Policy Network for the
21st century (REN 21)
2005–2011: Paris, France.
422
Currently, there is a wide range of PV cell technologies on the market.

Depending mainly on the basic material used and their level of commercial
maturity, PV cell technologies are usually classified into three generations as
illustrated in Fig. 10.2; namely, first generation, second generation, and third
generation technologies.
Fig. 10.2: PV cell technologies
The first generation of PV technologies is mainly made of crystalline

structure which uses silicon (Si) to produce the solar cells. These cells are
connected together to make PV modules. The first generation technologies
are not vanished rather they are constantly being developed to improve their
capability and efficiency153. In general, the efficiency of crystalline silicon (c-
Si) modules range from 14% to 19% and they have a relatively high
production cost and subsequently high selling price154. The relatively high
costs result from the complex and numerous production steps involved in the
silicon wafer and cell manufacturing as well as the large amount of highly
purified silicon feedstock required. Continued cost reductions are possible
through improvements in materials and manufacturing processes, and from
economies of scale if the market continues to grow, enabling a number of
high volume manufacturers to emerge. As shown in Fig. 10.2, the first
153
IRENA (2012). Renewable energy technologies: Cost analysis series- solar
Photovoltaics. pp. 45, International Renewable Energy Agency, Abu Dhabi.
154
Poullikkas, A. (2010). Technology and market future prospects of photovoltaic systems.
International Journal of Energy and Environment, 1(4), 617-634.
423
generation technologies include the mono-crystalline silicon (mono-Si), poly-

crystalline silicon (poly-Si), and Gallium Arsenide (GaAS).
The Mono-Si technology has higher efficiency in compassion with
other crystalline PV materials. The efficiency claims from manufacturer are
normally lies between 15% and 17%; however, mono-crystalline solar cells
can have an efficiency higher that 20%. The Mono-Si cells are the most
commonly used; constitute about 80% of the market and will continue to be
the leader until a more efficient and cost effective PV technology is
developed. The development of the poly-Si is mainly for the reduction of the
flaws, and contamination as well as crystal structure improvement 155.
Therefore, multi-Si modules will continue to retain a large portion of the
market in the future. The GaAs is a compound semiconductor form by
Gallium (Ga) and Arsenic (As) that has similar crystal structure as silicon.
Compared to silicon based solar cells, GaAs has higher efficiency and less
thickness. The GaAs is normally used in concentrator PV module and for
space application since it has high heat resistance and high cell efficiency
reaching about 25 to 30%; however, GaAs material and manufacturing
process can be costly in comparison with silicon based solar cells 156.
In comparison with the first generation technologies, the second
generation technologies hold the promise of reducing the cost of PV array by
lowering material and manufacturing costs without affecting the lifetime of
the cells as well as their environmental hazards. Thin film technology is the
core of the second generation technologies. Unlike crystalline solar cells,
where pieces of semiconductors are sandwiched between glass panels to
create the modules, thin film panels are created by depositing thin layers of
certain materials on glass or stainless steel substrates using sputtering tools.
The advantage of this methodology lies on the fact that the thickness of the
deposited layers which are a few microns (smaller than 10 m) thick compare
to crystalline wafers which tend to be several hundred microns thick. In
addition, it is possible to film deposited on stainless steel sheets which allows
the creation of flexible PV modules. As a result the overall manufacturing
155
Manna, T. K., & Mahajan, S. M. (2007, May). Nanotechnology in the development of
photovoltaic cells. In Clean Electrical Power, 2007. ICCEP'07. International Conference
on (pp. 379-386). IEEE.
156
Mah, O. (1998). Fundamentals of photovoltaic materials. National Solar Power
Research Institute, 1-10.
424
costs are lowered. Technically, reducing the thickness of the PV layers

reduces the solar radiation absorption and consequently reduces the
efficiency of thin film cells in comparison with crystalline cells; however;
this can be bypassed by using appropriate alloys for enhancing the efficiency.
In addition, the material flexibility of thin film technologies facilitates the
building-integrated PVs (BIPV)157. These distinguishing characteristics of
thin film technologies along with their temperature withstand-ability increase
the market share of these technologies to about 15% - 20%. The main
commercially available types of thin-film solar cells are the Amorphous
Silicon (a-Si and a-Si/µc-Si), Cadmium Telluride (CdTe), Copper-Indium-
Selenide (CIS), and Copper-Indium-Gallium-Diselenide (CIGS).
The third-generation PV technologies are at the pre-commercial stage
and vary from the technologies under the demonstration (e.g. multi-junction
concentrating PV) to novel concepts still require basic R&D activities (e.g.
Quantum-dots (QD) PV cells, QD meta-materials are a special semiconductor
system that consists of a combination of periodic groups of materials molded
in a variety of different forms. They are on the nanometer scale and have an
adjustable band-gap of energy levels performing as a special class of
semiconductors). Some third-generation PV technologies are beginning to be
commercialized, but it remains to be seen how successful they will be in
taking market share from existing technologies. There are four types of third-
generation PV technologies: Concentrating PV (CPV), Dye-sensitized solar
cells (DSSC), Organic solar cells, and Novel and emerging solar cell
concepts. Due to lack of sufficient information, the third generation
technologies will not be considered in this paper.
10.3 Quantitative Characterization of solar-PV technologies

The previous section provided a qualitative overview of various solar-PV
technologies. In this section, various technologies are characterized through a
quantitative approach. The results are then very in selecting and matching a
specific technology and module to a desired application considering various
technical and economic constraints.
157
Sheikh, N. M. (2008, March). Efficient utilization of solar energy for domestic
applications. In Electrical Engineering, 2008. ICEE 2008. Second International
Conference on (pp. 1-3). IEEE.
425
With the available PV technological alternatives and the large number of

PV modules in each technological alternative, the design engineer faces a
challenge in selecting the best PV module under a specified project design
criteria and constraints. Therefore, this paper provides a decision aid through
site-independent quantitative characterization of various PV technologies
and modules. In this regard, two stages are performed 158,159.
 In the first stage, various PV technologies are characterized as shown

in Fig. 10.3 by their peak wattage (Wp) per unit area (Wp/area),
efficiency, lifetime, degradation rate, temperature coefficient, and cost.
These characterization parameters are used to arrange various
technologies according to various goals of technology selection. The
goals include Wp/area maximization, efficiency maximization, lifetime
maximization, degradation minimization, temperature coefficient
minimization, and cost minimization. The best PV technology is
determined based on the achievement of these selection goals. Within a
specific PV technology, the characteristics of commercially available
PV modules differ significantly.
Fig. 10.3: Parameters of technological characterization of PV modules
158
T. Abdo, M. EL-Shimy, “Quantitative Characterization and Selection of Photovoltaic
Technologies”, MEPCON 2014.
159
M. EL-Shimy, T. Abdo, “PV Technologies: History, Technological Advances, and
Characterization’, In Sohail Anwar (ed.) Encyclopedia of Energy Engineering and
Technology, Taylor & Francis - CRC Press, 2014.
426
 In the second stage, the best PV module for each of the considered
technologies is determined. The PV module selections goals are the
peak wattage per unit area (Watt-peak/area), and the efficiency. For
each PV technology, the best module is determined according to the
fulfillment of the objectives i.e. maximization of the Watt-peak/area
and efficiency. The global best PV module is then determined.
It is worthy to be mentioned that the presented selection and

characterization approach is site-independent. This is a point of strength in
comparison with the capacity factor (CF) based selection 160 which is a site-
dependent method and it is mainly used for optimal site matching of solar-
solar-PVs. Since, each solar-PV technology has its own features that limit its
practical application (for example, thin film technologies are significantly
suitable in building integration solar-PV in comparison to crystalline
technologies due to its flexibility), then the final selection of PV technology
and module should be based on the project model.
The characterization of PV technologies and modules requires a huge
amount of data. Therefore, the availably of the required data in an accurate
form is a challenge. The RETScreen product database161 is used herein as a
data source. This database provides an up-to-date information about the Wp,
area, and efficiency data of various PV technologies and modules. In
addition, the database includes almost all the commercially available PV
modules; 1210 PV modules where 332 modules are mono-Si, 769 modules
are poly-Si, 78 modules are a-Si, 14 modules are CdTe, and 17 modules are
CIS. The included PV technologies in the RETScreen database are mono-Si,
poly-Si, a-Si, CdTe, and CIS. Therefore, the first and second generation
technologies are available in the RETScreen database. The rest of the
required data for assessing the lifetime, degradation rate, and module cost of
160
Salameh, Z. M., Borowy, B. S., & Amin, A. R. (1995). Photovoltaic module-site
matching based on the capacity factors. Energy Conversion, IEEE Transactions on, 10(2),
326-332.
161
"Renewable energy project analysis software", in RETScreen International Clean
Energy Decision Support Centre, Minister of Natural Resources Canada.
427
various PV technologies is obtained from162,163,164 respectively. In this section

analysis of the available data, characterization , and selection criteria are
presented.
 First Stage: Technological Characterization

At this stage, various PV technologies are characterized considering
the characterization parameters shown in Fig. 10.3. The ranges (maximum,
average, minimum, and standard deviation) of each characterization
parameter are determined for each technology. The standard deviation ()
shows how much variation or spreading exists for the average, or expected
value. A low standard deviation indicates that the data points tend to be very
close to the mean value while high standard deviation indicates that the data
points are spread out over a large range of values 165.
A module efficiency depends primarily on the type of the PV cell;
however, within various types there are wide variations in module efficiency
from manufacturer to another, depending on the manufacturing processes
used. Based on analysis of the RETScreen PV database, Fig. 10.4 shows the
results of PV efficiency analysis. It is depicted from Fig. 10.4, that the
efficiency of the mono-Si PV technology is the highest followed by Poly-Si
then CIS then CdTe then a-Si. The standard deviation of the efficiency values
is less than 2% for all technologies; however, the lowest standard deviation is
associated with the a-Si technology and the highest is associated with the
poly-Si technology. This means that the efficiency of the first PV generation
is still the highest in comparison with the rest of technological generations.
This is attributed to the maturity of the manufacturing process. The modules
and their manufacturers that are associated with the maximum and minimum
efficiency values for each PV technology are shown in Table 10.1. It is
shown in Table 10.1 that manufacturers that produce PV modules with the
highest efficiency are Sunpower, Canadian solar, Q-cells, and abound.
162
Sherwani, A. F., & Usmani, J. A. (2010). Life cycle assessment of solar PV based
electricity generation systems: A review. Renewable and Sustainable Energy Reviews,
14(1), 540-544.
163
Jordan, D. C., & Kurtz, S. R. (2013). Photovoltaic degradation rates—an analytical
review. Progress in photovoltaics: Research and Applications, 21(1), 12-29.
164
PVXChange, http://www.pvxchange.com/
165
DasGupta, A. (2010). Fundamentals of probability: a first course. Springer Science &
Business Media.
428
Fig. 10.4: Efficiency analysis of various PV technologies
Table 10.1: PV modules with the lowest and highest efficiencies and their
manufacturers
PV Lowest efficiency Highest efficiency
tech.  (%) Model Manuf.  (%) Model Manuf.
STP005S- SPR-320E-
Mono 7.58 Suntech 19.62 Sunpower
12 WHT
ND-070 Canadian
Poly 6.07 Sharp 17.59 CS 6A 190W
ERU Solar
Uni-Pac
a-Si 1.81 Uni-Solar 8.16 SN2-145.0W Q-Cells
10W(24V)
CdTe 4.86 CX-35W Q-Cells 10.07 AB1-72 Abound
Centennial
CIS 6.60 CIGS 50W 11.33 SL 1-85W Q-Cells
Solar
The efficiency of solar PV cells varies with their operating temperature.

Most cell/module types display a decrease in efficiency as their temperature
increases. The PV temperature coefficient (µ expressed in % /°C), is defined
by:
where (T) is the efficiency of the solar cell at an operating temperature T,

and (Tref) is the efficiency of the cell at the reference temperature Tref
(usually 25oC).
429
The ideal value of the temperature coefficient is zero. With this value, the
efficiency is insensitive to temperature variations. It can be seen from
equation (10.1) that for a specific operating and reference temperature, the
efficiency decrease as the temperature coefficient increase. Therefore, the
sensitivity of the PV efficiency increases with the increase in the temperature
coefficient. The value of the temperature coefficient depends primarily on the
cell type and it is difficult to find this value on manufacturer data sheets.
Therefore, in the absence of this information the default values shown in Fig.
10.3 may be used as recommended by the RETSCreen. It is depicted from
Fig. 10.3 that the efficiency values of the thin film technologies expect the
CIS are less sensitive to temperature variations in comparison with the
crystalline silicon. The efficiency of the CIS technology has the highest
sensitivity to temperature which is significantly higher than the crystalline
silicon sensitivity.
Fig. 10.5: Default temperature coefficient values for various PV

Technologies
The Wp/A or the Watt-peak/area is the peak power capacity of a PV

module (in Wp) over its frame area. The high value of the Wp/A indicates
better land use and higher density of the energy production. The Wp/A can be
used as a selection index for PV modules 166. Based on analysis of the
RETScreen PV database, Fig. 10.6 shows the results of Watt-peak/area
analysis.
166
El-Shimy, M. (2009). Viability analysis of PV power plants in Egypt. Renewable
Energy, 34(10), 2187-2196.
430
Fig. 10.6: The WP/A of various PV technologies
It is depicted from Fig. 10.6, that the Watt-peak/area of the mono-Si PV

technology is the highest value followed by Poly-Si, CIS, CdTe and a-Si
technologies. This means that the first PV generation is still has the highest
Watt-peak/area in comparison with the other technological generations. The
modules and their manufacturers that are associated with the maximum and
minimum Watt-peak/area values for each PV technology are shown in Table
10.3. It is shown in Table 10.3 that manufacturers that produce PV modules
with the highest Watt-peak/area are Q-cells, and abound.
Table 10.3: The highest and lowest WP/A for PV technologies and their
manufacturers
PV Low Wp/A High Wp/A
tech Wp/A Model Manuf. Wp/A Model Manuf.
Mono 71.43 STP005S-12 Suntech 200.00 Q6LM-1680 Q-Cells
Poly 15.00 YL15(17)P Yingli Solar 200.00 Q6LTT-1640 Q-Cells
Uni-Pac
a-Si 18.10 Uni-Solar 81.46 SN2-145.0W Q-Cells
10W(24V)
CdTe 48.61 -CX-35W Q-Cells 100.00 AB1-72 Abound
Centennial
CIS 66.67 CIGS 50W 113.33 SL 1-85W Q-Cells
Solar
The ability to predict output power over the course of time is important in
the prediction of the long-term performance of various PV technologies.
431
Among the key drivers in the PV industry growth are the efficiency (with
which sunlight is converted into power) and how this relationship changes
over time (i.e. the degradation rate). The quantification of power production
decline or diminishing over a time course is known as a degradation
rate167,168,169. The degradation occurs due to chemical and material processes
associated with weathering, oxidation, corrosion, and thermal stresses.
Financially, the degradation of a PV module or system is equally important,
because a higher degradation rate translates directly into less power produced
and therefore, reduces future cash flows.
Solar-PV systems are often financed based on an assumed (0.5 – 1.0) %
per year degradation rate 170; however, 1% per year is used based on modules
warranties171. This rate is faster than some historical data given for silicon
PV. In a study172, more than 70% of 19–23 year-old c-Si modules had an
annual degradation rate of 0.75% which is still less than the 1% year
assumed. Technically, the degradation mechanisms are important to be
understood because they may for the long run lead to failure. The
classification of degradation mechanisms through experiments and modeling
can lead directly to lifetime improvements. Typically, a drop in the output
power production by 20% at standard conditions is considered as a failure
that requires partial of full replacement of the PV modules; however, this is
not an agreement with the definition of a failure from the reliability point of
view. This is because the modules are still able to produce electricity. In
167
M. EL-Shimy, “Analysis of Levelized Cost of Energy (LCOE) and grid parity for
utility-scale photovoltaic generation systems”, 15th International Middle East Power
Systems Conference (MEPCON’12), Dec. 23-25, 2012, Alexandria, Egypt, pp. 1- 7.
168
M. Saed, M. EL-Shimy, and M. Abdelraheem, “Photovoltaics Energy: Improved
modeling and analysis of the Levelized Cost of Energy (LCOE) and Grid Parity”,
Sustainable Energy Technologies and Assessments, vol. 9, pp. 37-48, 2015.
169
Jordan, D. C., & Kurtz, S. R. (2013). Photovoltaic degradation rates—an analytical
review. Progress in photovoltaics: Research and Applications, 21(1), 12-29.
170
Campbell, M., Aschenbrenner, P., Blunden, J., Smeloff, E., & Wright, S. (2008). The
drivers of the levelized cost of electricity for utility-scale photovoltaics. White Paper:
SunPower Corporation.
171
Zweibel, K., Mason, J., & Fthenakis, V. (2008). A solar grand plan. Scientific
American, 298(1), 64-73.
172
Skoczek, A., Sample, T., & Dunlop, E. D. (2009). The results of performance
measurements of field‐aged crystalline silicon photovoltaic modules. Progress in
Photovoltaics: Research and Applications, 17(4), 227-240.
432
addition, a high efficiency module degraded by 50% may still have a higher
efficiency than a non-degraded module from a less efficient technology.
Fig 10.7, shows the degradation rates of various solar-PV technologies.
These summarized rates are the long-term degradation rates and they do not
include short-term, light-induced degradation rates. The light-induced
degradation rate effect is generally ascribed to boron-oxygen (B-O) defects
in the cell itself which is a defect complex that is formed by prolonged
exposure of solar cells to light and is accompanied by a reduction in the
minority-carrier lifetime in the bulk of the cell. This degradation in the
lifetime of the wafer is thought to be fully recoverable upon annealing,
which causes the dissociation of the B-O complex173.
Fig 10.7: Degradation rates of various PV technology
In Fig 10.7, the denotations “pre” and “post” refer to a date of installation
prior to and after the year 2000, respectively. The selection of year 2000 was
chosen such that the number of data points in each category is approximately
equal. It is depicted from this figure that the crystalline Si technologies show
low degradation rates in comparison with other technologies for pre-2000 and
post-2000 categories; however, the figure shows that there are a significant
reduction in degradation rates of the thin-film technologies in the analysis the
post-2000 in comparison with the pre-2000 installations.
173
Sopori, B., Basnyat, P., Devayajanam, S., Shet, S., Mehta, V., Binns, J., & Appel, J.
(2012, June). Understanding light-induced degradation of c-Si solar cells. In Photovoltaic
Specialists Conference (PVSC), 2012 38th IEEE (pp. 001115-001120). IEEE.
433
Generally, the working life of an asset is the life for which it continues to
perform its tasks effectively. It is often true that the operation and
maintenance (O&M) costs rise with the age of the asset. Since annual capital
costs tend to decline and annual O&M costs rise, there is a minimum average
cost per year at which point it is considered the economic life of the asset.
This is called the economic lifetime. By the end of the economic life, the
asset is then replaced or refurbished, since thereafter it becomes more
expensive to run the asset174.
The financial life of a solar PV system is usually considered to be the
guarantee period offered by the manufacturer which is often 20–25 years;
however, researches show that the life of solar PV panels is well beyond 25
years even for the older technologies. The current technologies are likely to
improve lifetime further175. The O&M costs are due to replacing inverters
(usually every 5 or 10 years), occasional cleaning, and electrical system
repairs. It is expected that some of these costs, such the inverter, will
decrease with time.
It should also be noted that the life of many conventional power plants is
much longer than the rated lifetime since they tend to be theoretically
refurbished or re-commissioned indefinitely, then the same could be true of
solar-PV plants. Therefore, the economic life of the system depends on the
acceptable energy output which is mainly dependant on the degradation rate.
On the other hand, the predicted or expected lifetime of a photovoltaic
module is one of the main factors which define the levelized cost (LCOE) of
PV solar electricity176. Extensive effort has been made for more than 20 years
to define, predict and assure the lifetime of PV solar modules from the
manufacturers, the research laboratories and international standards bodies.
The effort to assure the functionality, quality and lifetime of PV modules is
introduced by the International Electro-technical Commission (IEC) through
accelerated test procedures called type approval tests. These test procedures
174
Leblanc, S., & Roman, P. (2005). Project Management: An Engineering Economics
Perspective. Pearson Custom Publishing.
175
Skoczek, A., Sample, T., & Dunlop, E. D. (2009). The results of performance
measurements of field‐aged crystalline silicon photovoltaic modules. Progress in
Photovoltaics: Research and Applications, 17(4), 227-240.
176
M. Saed, M. EL-Shimy, and M. Abdelraheem, “Photovoltaics Energy: Improved
modeling and analysis of the Levelized Cost of Energy (LCOE) and Grid Parity”,
Sustainable Energy Technologies and Assessments, vol. 9, pp. 37-48, 2015.
434
seek to assure, based on accelerated climatic testing, that the PV modules

which successfully pass these tests would reasonably be expected to survive
for 20 years in field operation. These tests however do not give a precise
measure of the lifetime 177. The commonly used three technology types of
solar-PV systems had been discussed based upon the lifetime assessment.
The lifetimes have been estimated for amorphous, mono-crystalline, poly-
crystalline PV systems. It is found that the average lifetime for each
technology is approximately to be the same (26 year).
A PV module warranty usually refers to two different issues. The first is
the warranty for materials and workmanship, which usually ranges from 1 to
5 years. The second is the warranty on the power produced by the PV module
which typically guarantees that after the first 10–12 years the output power of
the module will be at least 90% of its initial nominal power and that after 20–
25 years of operation the output power of the module will be at least 80% of
its initial nominal power178. Recently, the manufacturers have redefined the
concept of output power warranty which is now referred to as the minimum
nominal power taking into account module power tolerances, and not with
respect to the nominal power. The measurement tolerance is also quantified
as an extra 3% of uncertainty179. The average PV module power warranty has
increased from 5 years before 1987 to 25 years since 1999. Obviously, these
warranties are not simply the result of thorough tests in the field because the
modules have not been in the market long enough, but they are probably the
combined outcome of empirical approaches, field tests of limited durations
and a set of particular degradation tests that each manufacturer developed for
this purpose.
In this book, the effective lifetime (L.Teffective) of a PV module is defined as
the span time at which the module will be able to effectively produce power
such that by the end of the effective lifetime the energy production drop is
177
Dunlop, E. D., Halton, D., & Ossenbrink, H. (2005, January). 20 years of life and more:
where is the end of life of a PV module?. In Photovoltaic Specialists Conference, 2005.
Conference Record of the Thirty-first IEEE (pp. 1593-1596). IEEE.
178
Vázquez López, M., & Rey-Stolle Prado, I. (2008). Photovoltaic module reliability
model based on field degradation studies. Progress in Photovoltaics: Research and
Applications, 16(5), 419-433.
179
Cereghetti, N., Bura, E., Chianese, D., Friesen, G., Realini, A., & Rezzonico, S. (2003,
May). Power and energy production of PV modules statistical considerations of 10 years
activity. In Photovoltaic Energy Conversion, 2003. Proceedings of 3rd World Conference
on (Vol. 2, pp. 1919-1922). IEEE.
435
equal to 20%. In order to calculate the L.Teffective of a selected PV technology

module, the degradation rate of the PV technology should be known. Based
on the given definition, the effective lifetime (L.Teff) in years can be easily
determined based on Fig. 10.8 which shows the energy production decline
with time; where d is the percentage degradation rate of the PV module.
Considering the post-year 2000 values of degradation rates shown in Fig

10.7 and equation (10.2), the determined effective lifetime values are shown
in Table 10.4 and Fig. 10.9. This results also shows the percentage change of
the effective lifetime in comparison with an assumed financial lifetime of 25
years (L.Tfinancial). The percentage change in the lifetime (L.T) is calculated
using,
Fig. 10.8: The effective lifetime of a PV module
Table 10.4: Effective and percentage change of the lifetime of various PV

technology
L.Teff (years) ΔL.T (%)
PV Tech.
Minimum Average Maximum Minimum Average Maximum
Mono-Si 30 41 51 20 64 104
Poly-Si 23 30 41 -8 20 64
a-Si 15 19 26 -40 -24 4
CdTe 21 37 101 -16 48 304
CIS 12 18 34 -52 -28 36
436
It is depicted that the minimum effective lifetime of mono-Si is higher (30

years; L.T = 20%) than the financial 25 year lifetime while it is lower than
that value in the rest of the technologies. The lowest minimum effective
lifetime is associated with the CIS modules (12 years; L.T = -52%). The
minimum effective lifetime of poly-Si (23 years; L.T = -8%) is close to the
financial lifetime while it is significantly lower in the a-Si technology (15
years; L.T = -40%). From the average effective lifetime point of view, the
results show that all the considered technologies, except the a-Si and CIS,
have effective lifetimes higher than the financial lifetime. Highest average
effective lifetime is associated with the mono-Si (41 years; L.T = 64%) and
the lowest one are associated with the CIS (18 years; L.T = -28%) which is
still lower than the financial lifetime. It is also shown that the average
effective lifetime of the a-Si modules (19 years; L.T = -24%) is less than the
financial lifetime. From the maximum effective life time point of view, the
results show that all the considered technologies, except the a-Si, are having
effective lifetimes that are higher than the financial lifetime. The highest
maximum effective lifetime is associated with CdTe modules (101 years;
L.T = 304%) while the lowest maximum effective lifetime is associated
with a-Si modules (only 26 years; L.T = 4%).
120
L.Teff (years)
100
Min.
Effective lifetime (years)
L.Teff (years) Av.

80
60
40
20
0
Mono-Si Poly-Si a-Si CdTe CIS
L.Teff (years) Min. 30 23 15 21 12
L.Teff (years) Av. 41 30 19 37 18
L.Teff (years) Max. 51 41 26 101 34
Fig. 10.9: Effective lifetime of various solar-PV technologies
437
If the minimum effective lifetimes are considered pessimistic and the

maximum effective lifetimes are considered optimistic, then the average
effective lifetime is considered the probable one. Therefore, Table 10.4 and
Fig. 10.9 shows that both the a-Si and CIS modules will not survive to the
end of the financial lifetime while the rest of technologies will survive for
durations higher than the financial lifetime.
Driven by advances in technology and increases in manufacturing scale,
the cost of PV cells has declined steadily since the first solar cells were
manufactured180. Although the cost of electricity produced from PV systems
is still higher than the other competing technologies, this cost is expected to
continue to decrease steadily. The cost of PV installation was USD 2 per unit
of generating capacity in 2009 which came down to about USD 1.50 in 2011.
According to industry analysis, this price is expected to be lowered in the
future. These potential reductions in the cost, combined with the simplicity,
versatility, reliability, and low environmental impact of PV systems, should
help PV systems become highly utilized sources of economical, premium-
quality power over the next 20 – 30 years181.
Due to the high installation costs, the recent solar electric prices are at
approximately USD 0.30/kWh, or around 2–5 times the average residential
electricity tariffs (the actual ratio depends on the location of the installation
and local electricity rates). On the other hand, solar energy can provide the
basic energy needs for houses in remote and rural areas at a fraction of the
cost spent on traditional electricity. Thus, the cost of electricity from PV
systems is relatively cost effective in that and similar applications. As
expected, the demand for solar powered systems is very high in countries
with high electricity tariffs. On the other hand to get insight about the PV
module prices, the pvXchange website provides a closed trading platform that
provides spot market prices of various PV technologies. The prices of the
year 2012 are shown in Fig. 10.10.
It is clear from Fig. 10.10 that the prices of all technologies decline with
time even on a monthly basis. In addition, the prices of the crystalline
modules made in Germany and Japan are comparable but they are
180
Swanson, R. M. (2009). Photovoltaics power up. Science, 324(5929), 891-892.
181
Devabhaktuni, V., Alam, M., Depuru, S. S. S. R., Green, R. C., Nims, D., & Near, C.
(2013). Solar energy: Trends and enabling technologies. Renewable and Sustainable
Energy Reviews, 19, 555-564.
438
significantly higher than the prices of the crystalline modules made in China.
The prices of the a-Si modules are the lowest followed by the Cds/CdTe
modules then the a-Si/-Si modules and the China crystalline modules, then
Germany and Japan crystalline modules. Although the prices of the a-Si
modules are the lowest as shown in Fig. 10.10, they are also having a low
average effective lifetime as shown in Table 10.4. Fig. 10.10 also shows that
China crystalline modules offer a very low price in compassion to other
crystalline modules and at the same time they offer a high average lifetime as
shown in Table 10.4 (for the crystalline technologies). That is a probable
reason for the high potential of the China PV market and production.
Fig. 10.10: Module prices of various PV technologies for year 2012
B. Second stage: module selection

In this stage, the best PV module in each of the considered
technologies is determined. The PV module selections goals are the Watt-
peak/area, and the efficiency. For each PV technology, the best module is
determined according to the fulfillment of the objectives i.e. maximization of
the Watt-peak/area and the efficiency. For each of the considered PV
technologies, the top modules that fulfill each of the stated objectives are
shown in Tables 10.5. As expected, it is not necessary that the PV module
that has the maximum efficiency will also have the highest Watt-peak/area.
Of course high efficiency results in high energy production and high Watt-
439
peak/area results in lower land use. Therefore, the selection of an appropriate

module will be based on the design requirements and limitations. In addition,
since each PV technology has its own features that limit its practical
application, and then the final selection of PV technology and module should
be based on the project model. For example, thin-film technologies are the
best for BIPV while crystalline technologies as well as thin-film technologies
are suitable for other standalone and grid-connected applications.
Tables 10.5: The best PV modules in various technologies

Manufacturer Module Efficiency() Wp/A (Watt/m2)
Mono-crystalline
Sunpower SPR-320E-WHT 19.62 197.53
Q-Cells Q6LM-1680 16.80 200.00
Poly-Crystalline
Canadian Solar CS 6A 190W 17.59 146.15
Q-Cells Q6LTT-1640 16.40 200.00
a-Si
Q-Cells SN2-145.0W 8.16 81.46
CdTe
Abound AB1-72 10.07 100.00
CIS
Q-Cells SL 1-85W 11.33 113.33
In the mono-Si technological category, it is clear from Tables 10.5 that the
Sunpower mono-Si-SPR-320E-WHT offers the highest efficiency (19.62%)
with a Watt-peak/area of 197.53 Watt/m2 while the Q-cells mono-Si-Q6LM-
1680 module offers the highest Watt-peak/area (200 watt/m2) with an
efficiency of 16.8%. Since the Watt-peak/area gained from the Q-cells
module is insignificant in comparison with the Sunpower module and the
efficiency of the Sunpower module is significantly higher than the Q-cells
module, then the best found mono-Si module is the Sunpower mono-Si-SPR-
320E-WHT. In the poly-Si technological category, it is clear from Tables
10.5 that the Canadian Solar poly-Si-CS 6A 190W offers the highest
efficiency (17.59%) with a Watt-peak/area of 146.15 Watt/m2 while the Q-
cells poly-Si-Q6LTT-1640 module offers the highest Watt-peak/area (200
watt/m2) with an efficiency of 16.4%. Since the Watt-peak/area gained from
440
the Q-cells module is significant in comparison with the Candian Solar

module and the efficiency of the Canadian Solar module is not much higher
than the Q-cells module, then the best found poly-Si module is the Q-cells
poly-Si-Q6LTT-1640. As shown in Tables 10.5, the best found modules in
the a-Si, CdTe, and CIS technological categories respectively are the Q-cells
a-Si-SN2-145.0W, Abound CdTe-AB1-72, and Q-cells CIS-SL 1-85W.
A comparison between the efficiency and the Watt-peak/area values of the
best found modules in various technologies are illustrated in Fig. 10.11. In
order to show them clearly on the same scale with the Watt-peak/area, the
efficiency values are multiplied by 10. It is clear from Fig. 10.11 and Table
10.5 that the highest efficiency is associated with the mono-Si module
followed by the poly-Si module, then CIS module, then the CdTe module,
then the a-Si module. This is in conformance with the general results shown
in Fig. 10.4. In addition, it is clear that the Watt-peak/area of the mono and
poly crystalline modules are compared with a slight gain in the poly-Si
module. Therefore, the highest Watt-peak/area is associated with the
crystalline modules followed by the CIS module, then the CdTe module, then
the a-Si module. This is also in conformation with the general results shown
in Fig. 10.6.
Fig. 10.11: Comparison between the best found modules
441
10.4 Fundamentals and Applications of solar-PV systems

Fig. 10.12 shows a basic scheme of solar-PV electric power generators
and load types as well as examples of the energy storage options. Depending
on the system requirements, some of the shown components may be absent.
As shown in the figure, the main components of a solar-PV generator are: (1)
Solar-PV array; (2) Sun tracking mechanism; (3) Maximum Power Point
Tracking (MPPT) controller; and (4) Power Conditioning Unit (PCU).
10.4.1 Solar-PV arrays

The solar-PV array is made up with a number of solar-PV modules (or
panels). These modules are electrically connected for providing specific
current and voltage ratings. Based on the rating of one module, the number of
series connected modules defines the voltage rating of the solar-PV generator
while the number of parallel modules defines the current rating of the
generators. PV-modules are commercially available as a unit while the
connection between them is based on the system design and operational
requirements. In the manufacturing process, PV-modules are formed by
series/parallel connection of a number of solar-PV cells (see Fig. 10.13).
These internal connections and the current/voltage rating of the cell
technology define the rating of the PV-module. The main function of the
solar-PV generator is to transform the solar energy to DC electrical energy.
Details about the physics of this transformation are outside the scope of this
book; however, the fundamentals of this energy conversion process as well as
details of the technical requirements for connecting cells and modules are
available at many resources including the pveducation.org website 182.
As shown in Fig. Fig. 10.14(a), two main dependent characteristics can be
used for describing the performance of solar cells. These characteristics are
the I-V and P-V curves. The open-circuit voltage (Voc) and short-circuit
current (Isc) are illustrated at the figure. As shown in this figure, there is a
specific operating point at which the power extracted from the cell is the
maximum value. This point is called the maximum power point (MPP); Pmp.
The voltage and current values at the MPP are Vmp and Imp. In an ideal
theoretical solar cell, the Vmp = Voc and Imp = Isc; however, realistic cell do not
provide this relation. The fill factor (FF) defines the deviation of this relation
182
http://pveducation.org/
442
with respect to the ideal cell; FF = (Vmp Imp)/( Voc Isc). The efficiency () of
the cell is determined as Pmp/(Ht A) where the solar irradiance incident on the
plan of the cell (Ht) in unit power/m2 and A is the cell surface area exposed to
the solar radiation in m2. The value of Pmp equals to Vmp Imp.
(a)
(b)
Fig. 10.12: Solar-PV systems; (a) Generators and their possible load
types; (b) Energy storage options
The characterizing operating points (Voc, Isc, Vmp, Imp, Pmp) are usually
determined and provided by the manufacturer at specific conditions called
Standard Test Conditions (STC) which are mainly 1000 W/m2 solar
443
radiations incident on the plan of the solar-PV cell at 25oC (77oF) of average
air temperature and 1.5 air mass (or atmospheric density) 183,184.These
characterizing operating points (Voc, Isc, Vmp, Imp, Pmp) are sensitive to both the
incident solar radiation at the cell surface and the surface temperature of the
cell. The effects of the solar irradiance and the temperature of the cell
characteristics are illustrated in Fig. 10.14(b) to (d) respectively.
As shown in Fig. 10.14(b), the changes in the irradiance have a significant
impact on the short-circuit current and a minor impact on the open-circuit
voltage. Increasing the irradiance increases the short-circuit current and the
open-circuit voltage as well as the maximum power. The MPP voltage is
increased with the increase in the irradiance. It is clear from Fig. 10.14(c) that
the open-circuit voltage rapidly increases with the increase of the irradiance
from zero. Shortly after that rapid increase, the open circuit voltage becomes
less sensitive to the changes in the irradiance. The short-circuit current shows
a linear direct relation with the changes in the solar irradiance. On the other
hand, the changes in the cell temperature (Fig. 10.14(d)) have an insignificant
impact on the short-circuit current and a significant impact on the open-
circuit voltage. Increasing the temperature slightly increases the short-circuit
current and significantly reduces the open-circuit voltage. Consequently, the
maximum power is reduced with the increase in the temperature and the MPP
voltage is also reduced.
The characteristics of either modules or arrays can be easily derived from
the characteristics of a cell. For example, a module having Ncells cells
183
IEC 61215, IEC 61646 and UL 1703
184
The solar-PV manufacturers usually use the STC to determine the characteristics of
solar-PV modules. One the other hand, the PVUSA Test Conditions (PTC) which is
sometimes also called performance test conditions is developed to test and compare solar-
PV systems as part of the PVUSA (Photovoltaics for Utility Scale Applications) project,
which was undertaken by a partnership between the US Department of Energy and several
major utilities. PTC and STC are different test basis; however, the PTC is considered to be
a better reflection of real operational conditions in comparison with the STC. This is
because PTC ratings are based on conditions that are closer to typical outdoor operating
environments. Therefore, PTC ratings are more realistic than nameplate (i.e. STC) ratings.
Unlike the STC ratings which are measured values, the PTC ratings are calculated values.
The Normal Operating Cell Temperature (NOCT) ratings are adopted by power utilities
for more accurate estimation of the performance of solar-PV modules in the real world
conditions; 800 W/m2 of sunlight irradiance at the plan of the module, 20°C (68°F)
average air temperature, and 1.0 m/s (2.24 miles/hour) of wind speed with the back side of
the solar panel open to that wind breeze. A method of estimating the PTC from NOTC is
stated at http://goo.gl/ro1gX2  
444
connected in series and Mcells cells connected in parallel, will then

approximately have,
(a)
(b)
Fig. 10.13: Solar-PV. (a) Cell, module, and array; (b) Connection of cells
in a module; 36 cells are connected in series for producing a rated voltage of
12 V.
Similarly, an array consists of Nmod modules connected in series and Mmod

modules connected in parallel, will then approximately have,
445
(b)
(a)
(c) (d)
(e)
Fig. 10.14: Characteristics of solar-PV arrays; (a) Characteristics of a cell; (b)
Impacts of change in the solar irradiance; (c) Changes in Isc and Voc with the
changes in the irradiance; (d) Impacts of changes in the temperature; (e)
Characteristics of the Sanyo HIT 215W module at 25 oC.
446
Solar-PV cells can be represented by equivalent circuits such as that

shown in Fig. 10.15. This circuit includes a series resistance (Rs) and a diode
in parallel with a shunt resistance (Rsh). The letter V represents the voltage at
the load. This equivalent circuit can be used for an individual cell, for a
module consisting of several cells, or for an array consisting of several
modules.
(a)
(b) (c)
Fig. 10.15: Equivalent circuit of PV cells; (a) Circuit diagram; (b) Impact of
the series resistance; (c) Impact of the shunt resistance
The impacts of the series and the shunt resistances on the I-V
characteristics are also shown in the figure. It is clear that increasing the
series resistance decreases the MPP voltage and MMP current. Consequently,
the maximum power and the efficiency are reduced with the increase in the
series resistance. On the other hand, increasing the shunt resistance has an
opposite impact in comparison with the impact of the series resistance. The
figure also shows a way of determining these resistances from measured I-V
curves. The series resistance is determined as the negative of the slop of the I-
447
V curve at open-circuit conditions while the shunt resistance equals to the

negative of the slop at short-circuit conditions.
In an ideal cell, the shunt resistance is infinite while the series resistance is
zero. As shown in Fig. 10.14(a), the overall resistances of a cell can be
represented by the characteristic resistance (Rch) which is the negative slop of
the straight line connecting the origin and the MPP current and voltage. This
resistance is of major importance in the operational optimization of solar
cells. According to the maximum power transfer theory, a cell supplying a
resistive load of a resistance equals to the cell’s characteristics resistance
will be supplied by the maximum power. This is important in either matching
the load and the solar-PV array or the MPPT. It is also clear from the shown
characteristics that the characteristic resistance is sensitive to the irradiance
and the temperature as well as the series and shunt resistance of the cell.
There are two tracking control systems for optimizing the performance of
solar-PV generators. These tracking systems are the MPPT and the sun
tracking. The ideal MPPT 185 is a control system (see Fig. 10.16) that extract
the maximum power for all operating conditions. This is can be achieved by
controlling the load resistance at the terminals of the array to exactly match
the array characteristic resistance. In another way, the MPPT modifies the
terminal voltage of the array to be of the same value as the MPP voltage for
maximum power transfer. The sun tracking system changes the position of
the surface of the array for maximization of the irradiance incident at the plan
of the modules. Both tracking systems will be briefly explained in the
following.
Fig. 10.16: MPPT by DC-DC boost converters

185
The MPPT is an electronic DC to DC converter that optimizes the match between the
solar-PV array (or module) and the connected load or the grid
448
10.4.2 PCU and MPPT

The PCU is a functional block that handles one or more functions in
one physical unit or more units. The main functions of the PCU includes the
DC-AC inversion, harmonic filtering and possibly some other power quality
(PQ) enhancements, and surge protection. According to the load solar-PV
generator, DC and AC power are available; however, supplying AC loads
from the solar-PV generator requires a conversion of the DC produced power
to the required AC supply specifications. Therefore, inverters are needed for
supplying AC loads. As shown in Fig. 10.12, solar-PV generators may be
operated in the off-grid or grid-connected modes. In the off-grid mode, no
inverters are needed if the entire system is DC; however, inverters are needed
in the grid-connected operation. The MPPT function may also be included in
the PCU block.
The DC to AC inversion is provided by inverters. Currently, there are two
options in the power inversion process; either centralized or micro inverters.
With centralized inverters, large solar-PV power plants can be connected to
one inverter station for handling the DC-AC conversion. With micro
inverters, each module is integrated with a micro inverter. In comparison with
micro inverters, the centralized inverters offer reduction in both the power
conversion costs and the associated power losses; however, micro inverters
has some unique features that cannot be provided in centralized-based
inverter installations. These unique features include the capability of
operating each module at its MPP and also performing maintenance and
repair without interrupting the operation of the entire solar-PV plant.
Therefore, micro inverters also enhance the overall availability of the solar-
PV output energy.
There are many techniques for the MPPT 186,187 control. These techniques
include (Fig. 10.17(a)) the constant voltage, open circuit voltage, short
circuit current, perturb and observe (P&O), and incremental conductance
(IC) methods. The rank of an MPPT technique is usually presented as a
percentage of the maximum power extracted at various irradiance (and
temperature) conditions in comparison with the ideal MPPT algorithm. For
186
Dolara, A., Faranda, R., & Leva, S. (2009). Energy comparison of seven MPPT
techniques for PV systems. Journal of Electromagnetic Analysis and Applications, 2009.
187
Faranda, R., & Leva, S. (2008). Energy comparison of MPPT techniques for PV
Systems. WSEAS transactions on power systems, 3(6), 446-455.
449
example, an MPPT technique with a rank of 85% indicates that the extracted
maximum power is 85% of the actual available maximum power at various
input conditions to the solar-PV array. Fig. 10.17(b) also illustrates a
comparison between the approximate rankings of various MPPT techniques
considering the variations of the irradiance values around the STC irradiance
at a constant STC standard temperature. It should be notes that the dynamic
performance and the hardware requirements of an MPPT technique are not of
low importance in comparison with its rank. This issue will be considered
later in this chapter.
(a)
(b)
Fig. 10.17: MPPT; (a) Methods; (b) Approximate ranking at irradiance
levels close to the STC value and a constant STC temperature
450
In the constant voltage (C.V) method, the average MPP voltage is used for
the MPPT. As shown in Fig. 10.14 and 10.18, the changes in Vmp due to the
changes in the irradiance are not very large; however, the impact of the
temperature various on the MPP is significant. Consequently, in this method,
the average value of Vmp is determined and the solar-PV array terminal
voltage is fixed at this average value which is also called the reference
voltage Vref.
(a) (b)
10.18: Changes in the MPP; (a) Impact of irradiance; (b) impact of
temperature
The reference voltage may also be chosen to be equal to the Vmp at the
STC conditions. Fig. 10.19 shows the constant voltage MPPT algorithm
where k is the sampling rate of the measurements and D is the duty cycle of
the DC-DC boost converter188 (Fig. 10.16); the output voltage is proportional
to the duty cycle. The C.V method is the simplest MPPT technique; however,
it has low rank in comparison with the other techniques when the irradiance
is close to the STC value; however, the rank of the C.V method surpasses the
P&O as well as the IC methods at low irradiance levels. This is due to
variable sensitivity of the change of the MPP point with the irradiance.
188
https://en.wikipedia.org/wiki/Boost_converter
451
Fig. 10.19: Constant voltage MPPT algorithm
In the open circuit (O.C) voltage MPPT technique, the reference

voltage of the boost converter is set to an approximate MPP voltage. The
variations in the MPP voltage are related to the variations in the O.C voltage
approximately by Vmp = OC VOC. Therefore, Vref = OC VOC. Typical value of
OC is between 0.7 to 0.8. The main disadvantage of this method is the
necessary of interrupting the output power for determining the open circuit
voltage according the variations in the irradiance and temperature. Fig. 10.20
shows the O.C voltage based MPPT algorithm. As shown from Fig. 10.17,
the O.C voltage based MPPT is having a better ranking in comparison with
the constant voltage and the S.C current based MPPT methods.
In the S.C current, the MPPT is based on the current instead of the
voltage. The MPP current is determined as a fraction of the S.C current i.e.
Imp = SC ISC. Therefore, Iref = SC ISC while the range of SC is between 0.9 to
0.98. Fig. 10.21 shows the S.C current based MPPT algorithm.
452
Fig. 10.20: O.C voltage based MPPT algorithm
Fig. 10.21: S.C current based MPPT algorithm
453
The Perturb and Observe (P&O) MPPT method is a search algorithm. In

the P&O method searches for the maximum power point by perturbing the
solar-PV voltage (or Current) and detecting the consequent change in solar-
PV power output. The direction of the change is reversed when the solar-PV
power decreases. The main P&O MPPT concept is illustrated in Fig. 10.22;
however, there are many forms of the P&O MPPT algorithm 189. Although the
ranking of the P&O method is very close to the ideal MPPT (Fig. 10.17), the
operation of the algorithm depends on the continuous disturbance of the
solar-PV terminal voltage. In addition, these perturbations are not stopped
even if the MPP is achieved; the operating point at the MPP conditions
continuous to oscillate around the MPP as a result of the MPPT algorithm.
Therefore, the dynamic performance of the P&O method reduces the value of
the ranking of this method.
Fig. 10.22: Basic concepts of the P&O based MPPT algorithm
189
Dolara, A., Faranda, R., & Leva, S. (2009). Energy comparison of seven MPPT
techniques for PV systems. Journal of Electromagnetic Analysis and Applications, 2009
454
The P&O method also suffers from operational failure in the cases of
sudden changes in the irradiance or the temperature. This problem is
illustrated in Fig. 10.23. Suppose that the solar-PV array operates at the MPP
(a) at specific meteorological conditions. The P&O algorithm then perturbs
this operating point by a small change in the terminal voltage (V). With the
meteorological conditions unchanged, the operating point goes to point (b).
Therefore, the power is reduced and the MPP (a) is restored in the second
MPPT cycle. Now consider a situation where the meteorological conditions
are rapidly changed during the sampling period when the system operates at
point (a). The power curve moves from curve (1) to curve (2). The
perturbation of the operating point then causes point (a) to move to point (c).
Since, the power at (c) is higher than the power at (a), the MPPT then
considers (c) as the MPP while it is not the actual MPP at the new
meteorological conditions. This divergence action will be sustained if the
meteorological conditions remain changed rapidly.
Fig. 10.23: Failure of the P&O MPPT during rapid changes in the
meteorological conditions
The incremental conductance (IC) based MPPT algorithm is based on

the fact that dP/dV = 0 at the MPP (Fig. 10.18). Since P = VI, then dP/dV =
d(VI)/dV = I (dV/dV) + V(dI/dV) = I + V(dI/dV) = 0. Therefore at the MPP,
I/V = -(dI/dV) i.e. at the MPP, the conductance of the solar-PV equals to its
455
incremental conductance. In addition, to the left of the MPP, dP/dV >0 which
resulting in I/V + (dI/dV) > 0 while I/V + (dI/dV) < 0 holds if the operating
point is to the right of the MPP. Therefore, information about the
conductance and the incremental conductance leads to the identification of
the location of the operating point on the P-V curve as well as the correct
direction of change in the voltage for reaching the MPP. If I/V + (dI/dV) > 0,
then the voltage should be increased while with I/V + (dI/dV) < 0, the voltage
should be decreased. The ranking of this method is comparable with the P&O
method while each of them has different dynamic performance as will be
illustrated later. Fig. 10.23 illustrates the basic concepts of the IC based
MPPT.
Fig. 10.24: Basic concepts of the IC based MPPT algorithm
Later in this chapter the dynamics associated with the P&O and IC MPPT
methods (as the most popular MPPT techniques) will be presented. In
456
addition, the control algorithms of these methods will be enhanced for better
dynamic performance.
10.4.3 Earth-Sun Geometry and Observer Angles

The location of the sun is very important for solar energy project. The
location is a function of the time of the day, the day in the year, and the
location on the earth. Every day the sun moves from the east to the west
between sunrise and sunset. In addition, the sun also moves from north to
south throughout the course of the year. If the position of the sun is measured
every day at solar noon it would be at a different angle every day. Tracking
the sun position is important for more extraction of the solar energy incident
on a solar energy conversion machine (such as solar-PV or solar CSP). Sun
tracking mechanisms are the tools used to do so. In this section the
fundamental geometry of the sun-earth relation is presented while the next
section provides an overview of the sun tracking options and their impact on
the energy production.
Fig. 10.25: Daily rotation of the earth
457
The earth continuously rotates about its axis of daily rotation as shown in
Fig. 10.25. The line connecting the north and south poles is called the earth’s
axis. This axis is perpendicular to the plane of the equator; however, it is not
perpendicular to the plane of the earth’s orbit plan. The tilt or obliquity of the
earth’s axis to a line perpendicular to the plane of its orbit is currently about
23.5°.
The plane of the sun is then defined as the plane at which the center of the
sun is located and at the same time parallel to the equator plane. The elliptic
cycle is then the annual cycle at which the earth passes alternately above and
below the plane of the sun. Fig. 10.26 illustrates the plane of the sun and the
elliptic cycle as well as the earth’s motion through the space.
(a)
(b)
Fig. 10.26: Motion of the earth; (a) Plane of the sun; (b) Motion through
space
The sun declination angle (δ) is defined to be that angle between the
equator plane and the line connecting the center of the earth and the center of
458
the sun. This is illustrated in Fig. 10.27. as shown in the figure, the
declination angle various between 23.5o (at the summer solstice) and -23.5o
(at the winter solstice).
(a)
(b)
(d)
(c)
Fig. 10.27: Sun declination; (a) angle definition; (b) angle variations; (c) at
summer solstice; (d) at winter solstice
Consider Fig. 10.28, where a random location (P) on the earth’s surface is
shown. The geocentric latitude angle () of the random location P is the
angle between the lines joining that location to the center of the earth and the
equatorial plane. Therefore, all locations at the same latitude experience the
same geometric relationship with the sun. The semicircles along the surface
of the earth joining the north to the south poles are called lines of longitude.
The longitude line crossing the Greenwich city at England is called the prime
meridian. For the shown point (P) the latitude is +o north relative to the
equatorial plane while the longitude of the point is +o east relative to prime
459
meridian. Fig. 10.29 illustrates various latitude and longitude lines on the
earth’s surface.
Fig. 10.28: latitude (), longitude (), and prime meridian
Fig. 10.29: Latitude and longitude lines and their references
The solar noon is defined to be that time of day at which the Sun’s rays
are directed perpendicular to a given line of longitude. Therefore, the solar
noon occurs at the same instant for all locations along any common line of
460
longitude. Since one revolution of the earth is equivalent to 24 hours and

360o, then 15o of longitudes are passed during one hour. The solar noon
occurs one hour earlier for every 15 degrees of longitude to the east of a
given line and one hour later for every 15 degrees west of a given line.
The sun hour angle (ω) is the angular distance between the meridian of
the observer (P) and the meridian whose plane contains the sun’s rays (Fig.
10.30). Thus, the hour angle is zero at the local noon (i.e. when the sun
reaches its highest point in the sky). At this time the sun is said to be due
south at the northern hemisphere and due north in the southern hemisphere.
Since the meridian plane of the observer contains the sun, the hour angle
increases by 15 degrees every hour. At the local noon, the hour angle is then
zero.
Fig. 10.30: Sun hour angle

The previous angle shown in Fig. 10.30 (declination, latitude, and hour
angle) are said to be the earth-sun angles. There are some additional angles
required for the estimation of the solar radiation incident on the plan of a
solar-PV module. These angles are shown in Fig. 10.31 and described in the
following.
The zenith angle (θz) is the angle between the line that points to the sun
and the vertical — basically, this just defines the location of the sun is in the
461
sky. At sunrise and sunset this angle is 0º while at the solar noon this angle is
zero.
Fig. 10.31: Main observer angles
The solar altitude angle (αs) is the angle between the line that points to the
sun and the horizontal (i.e. the tangent of the earth’s surface). It is the
complement of the zenith angle. At sunrise and sunset this angle is 0º while at
the solar noon this angle is 90.
The solar azimuth angle (γs) is the angle between the line that points to the
sun and south. Angles to the east are negative. Angles to the west are
positive. This angle is 0º at solar noon. It is probably close to -90º at sunrise
and 90º at sunset, depending on the season. This angle is only measured in
the horizontal plane; in other words, the height of the sun is neglected
The angle of incidence (θ) is the angle between the line that points to the
sun and the line that points straight out of a solar-PV module (it is also called
the line that is normal to the surface of the module). This is the most
important angle. If the module orientation is controlled for minimization of
the incident angle, then the solar energy incident on the module will be
maximized. Consequently, the energy production will be maximized. This
462
control action is the basis of all sun tracking control systems. In some cases,
the modules are kept fixed at a specific angle. For this purpose, the tilt angle
() should optimized for the maximization of the energy production and the
seasonal load variations 190.
The surface azimuth angle (γ) is the angle between the line that points
straight out of the module’s surface and south. It is only measured in the
horizontal plane. Again, east is negative and west is positive. If a module
pointed directly south, this angle would be 0º.
The collector slope or tilt angle (β) is the angle between the plane of the
solar collector and the horizontal. If a module is placed horizontally, then it is
0º.
10.4.4 Sun tracking mechanisms

A sun tracking mechanism is defined as the way that used to orient a
solar-PV module towards the sun for increasing the PV system performance.
The normal tracking mode is the fixed mode in which the PV module is fixed
at a typical angle called the tilt angle and does not track the daily sun
movement. The optimal tilt angle of a fixed PV array is related to the local
climatic condition, geographic latitude and the period of its use 191. Many
studies have been performed to get the optimum tilt angle that relates to the
maximum energy in different places. Previous studies show that the average
yearly optimum tilt angle is close to the latitude of the location; however, the
tilt angle can be optimized for a specific load pattern and specific set of
objectives. For example, maximization of the energy production and
minimization of the size of the required solar-PV generator for a specified
load pattern are popular set of objectives.
Generally, there are two main sun tracking modes; the single-axis (also
called one-axis) and the dual-axis (also called two-axis) tracking shown in
Fig. 10.32.
190
El-Shimy, M. (2013). Sizing optimisation of stand-alone photovoltaic generators for
irrigation water pumping systems. International Journal of Sustainable Energy, 32(5),
333-350.
191
Ulgen, K. (2006). Optimum tilt angle for solar collectors. Energy Sources, Part A:
Recovery, Utilization, and Environmental Effects, 28(13), 1171-1180.
463
The single-axis tracker pivots on it an axis to track the sun, where the
solar-PV modules rotate from east to west to track the daily sun movement192.
The modules face the east in the morning and the west in the afternoon (Fig.
10.33). The tilt angle of this axis can be optimized in the same way of
optimizing the tilt angle of the fixed modules or can be set to be equal the
latitude angle of the location. In consequence for this type of sun tracking, a
seasonal adjustment of the tilt angle is recommended for achieving the
maximum energy production. The single-axis tracking can be implemented
by several methods. These include horizontal single-axis trackers (HSAT),
vertical single-axis trackers (VSAT), tilted single-axis trackers (TSAT) and
polar aligned single-axis trackers (PSAT)193.
(a) (b)
Fig. 10.32: Basic types of sun tracking systems; (a) Single-axis; (b) Dual-
axis
The two-axis trackers follow the sun in both azimuth and altitude (i.e.
elevation) by having two-axes of rotation. Consequently, the PV module
rotates from east to west to track daily sun movement and north or south to
track seasonal sun movement throughout the year. Therefore, the sun's rays
are kept normal to the module surface. Considering their ability to combine
two rotational motions described around perpendicular axis, they are able to
follow very precisely the sun path along an annual period. Therefore, the
192
System Advisor Model (SAM). SAM's Help system, National Renewable Energy
Laboratory (NREL), September 2013
193
http://www.pvresources.com/PVSystems/Trackers.aspx
464
dual-axis trackers are more efficient than the single one; however, the
selection of the sun tracking option is based on many factors. These factors
include the economic feasibility and the benefits of the tracking system in
terms on the energy production. As will be shown later, the impact of various
tracking options on the energy production and the LCOE is highly depended
on the location of the solar-PV project194. The dual-axis tracking has two
common implementations; tip-tilt dual-axis trackers (TTDAT) and azimuth-
altitude dual-axis trackers (AADAT). Recently, a combination between solar
trackers and mirrors is used to maximize the energy production. The most
common combination uses V-trough mirrors with trackers.
Fig. 10.33: Daily motion of the single-axis trackers
PV module tracking has the advantages of increasing the PV system’s

output by about 40%; however, tracking equipment requires additional costs
(cost of equipment and O&M costs) and it needs more land for tracking
equipment as it must be spaced out to avoid the problem of shading 195. In
comparison with the fixed PV modules, it is founds 196 that the one-axis
194
Said, M., EL-Shimy, M., & Abdelraheem, M. A. (2015). Photovoltaics energy:
Improved modeling and analysis of the levelized cost of energy (LCOE) and grid parity–
Egypt case study. Sustainable Energy Technologies and Assessments, 9, 37-48.
195
Kumar, A., Thakur, N., Makade, R., & Shivhare, M. K. (2011). Optimization of tilt
angle for photovoltaic array. International Journal of Engineering Science and
Technology, 3(4), 3153-3161.
196
Ong, S., Campbell, C., Denholm, P., Margolis, R., & Heath, G. (2013). Land-use
requirements for solar power plants in the United States. Golden, CO: National Renewable
Energy Laboratory.
465
trackers increases the land use by about 13%, while two-axis trackers
increases the land use by about 40%. Based the average market price is the
year 2014, the minimum, maximum, and the average costs in US$/W p of
single-axis trackers are 0.36, 0.4, and 0.38 while their values for dual-axis
trackers are 0.4, 3, and 0.96 US/Wp. The annual operating and maintenance
costs of the tracking systems are in the range of 12 US$/kWp-year to 25
US$/kWp-year197.
10.4.5 Modes of operation of solar-PV generators

Based on Fig. 10.12(a), solar-PV generators can be operated in either
the off-grid or grid-connected modes. In the off-grid operation, the loads may
be non-deferrable or deferrable. For non-deferrable loads the continuous
energy balance is a major challenge especially when the solar-PV arrays are
the only source of power production. In this case energy storage is essentially
required to supply the essential and critical portions of the load. The critical
loads should not be interrupted while the essential loads can be interrupted
for a short-duration while the non-essential loads can be interrupted for
longer durations. The classification of various loads dependents on the
importance on the load equipment in the overall function(s) of the non-
deferrable loads as well as the hazardous situations and safety issues that can
be resulted from the interruption of the load equipment. On the other hand,
non-deferrable loads can be shifted in time without affecting the main target
of the load. For example, water irrigation pumping is a typical deferrable load
where the irrigation process can be shifted in time without significant impacts
on the crops. Energy storage is not generally required for deferrable loads;
however, non-electrical energy storage may be used for enhancing the
utilization of the solar-PV generator. Energy storage in the form of pumped
water in storage tanks or pumped storage energy production systems is
examples of non-electrical storage options.
In grid connected applications, the size of the solar-PV generator
relative to the size of the rest of the grid is among the method of determining
the need of energy storage. The energy storage is basically need for load
following and energy balance requirements. Therefore, in isolated and small
197
http://solarprofessional.com/articles/products-equipment/racking/pv-
trackers
466
grids, the energy storage may be needed while with large grids (e.g. utility)
grid, the energy storage is not essentially required.
Fig. 10.12(b) shows various options of energy storage. The figure shows
two main classification of energy storage devices or systems. The
electrochemical energy storage is provided by various types of batteries while
the non-electrical energy storage can be provided by many options. These
include the pumped storage, compressed air energy storage (CAES), and the
hydrogen. Details about various classes are outside the scope of this book;
however, due to its importance, the hydrogen as energy storage and an energy
carrier will be explained in this section.
Among the main critical challenges of renewable energy systems are the
variability and intermittency of common renewable energy sources such as
wind and solar (see chapter 1). Conceptually, appropriate sizing of variable
resources accompanied sufficient energy storage can solve these operational
problems; however, it is neither economical nor technically possible to
construct bulk power energy storage by the use of traditional electrochemical
options. Therefore, very large scale integration of variable renewable energy
sources to power systems is not impossible from operation and security
points of view.
Fig. 10.34: Use of hydrogen as an energy storage and energy carrier for
enhancing the reliability of variable power sources
467
The use of variable and intermittent electrical energy sources can be used
for hydrogen production (as shown in Fig. 10.34). The produced hydrogen
can be then stored in tanks at a specific compression level 198. The availability
of this stored hydrogen facilitates the controllable use of the overall energy
produced from an originally non-controllable resource (such as wind speed or
solar irradiance). In addition, the hydrogen can be moved over long
geographical distance in the same way of moving the oil and natural gas.
Unlike the conventional electrical power transmission, the hydrogen energy
transmission is not associated with energy losses or geographical barriers.
The amazing issue here is the use of the hydrogen. It can be used directly for
thermal energy production (like the natural gas in home and industrial
applications as well as supplying vehicle). In addition, indirect use of
hydrogen for electrical energy production is possible either through fuel cells
or hydrogen gas turbines. Advances in the research and industry of hydrogen
related technology for energy storage, energy transmission, and electrical
energy production are needed for enhancing the security, safety, economy,
and feasibility of such systems.
Variable renewable energy sources can also be used for many other
objectives. For example, reduction of the utility grid energy tariffs is possible
by reducing the dependency on the utility grid through installing on-site, for
example, solar-PV generators (Fig. 10.35). In this case, a utility customer will
not only be capable of reducing the electricity tariff but also can export power
to the grid. Usually feed-in-tariff structures are available in world-wide
countries for this purpose. The structures allow independent power producers
to sell power to the utility grid. Usually, the prices of the renewable energy
exported by the customers are higher than the retail prices of the utility
energy. This attracts the load-side integration of renewable resources even in
the residential class.
With the use of energy storage, these renewable sources can also be used
for enhancing the load supply reliability as well as reducing the possible
energy interruptions cause by the utility grid. The size of the required energy
storage should be minimized. Therefore, the load should be classified
198
Afgan, N. H., & Carvalho, M. G. (2004). Sustainability assessment of hydrogen energy
systems. International Journal of Hydrogen Energy, 29(13), 1327-1342.
468
according to its importance and the sizing of the storage should be based on
the critical and essential load classes.
Fig. 10.35: Use of solar-PV supply for reducing the electricity tariff and
possibly for enhancing the main supply reliability
10.5 Modeling of grid-connected solar-PV generators

Based on the objectives of the mathematical modeling of solar-PV
systems, two modeling categories can be realized. The first category of
models is basically used for the solar resource assessment and a rough
estimate of the energy production models. In this context, the first category of
models is called Resource Assessment and Energy Production (RA&EP)
Models199. These models are useful in the initial stages of solar-PV projects
where the accuracy of the results is not of the major importance. For these
models, high details about the meteorological conditions of the project
location are required while the modeling of the solar-PV cells is highly
simplified; the PV modules are represented mainly by their efficiency and
temperature coefficient values. Techno-economical feasibility analysis
software tools such as the RETSCreen utilize the energy production models.
The second category of models simulates the PV-system in high details from
the prospective of the PV-cell and the time step of the meteorological date
requirements. The second category models are called in this context
199
333-350.
469
equivalent circuit based models200 (ECBM). The Solar Advisor Model201

(SAM) software tool uses such models for estimating the performance of
solar-PV power plants.
The time step of the meteorological data required for the RA&EP models
are usually long term averaged values on daily, or monthly, or even annual
basis. For the ECBMs, the time step of the meteorological data is usually on
hourly or less time basis. This is required for achieving a reliable accuracy
level. The performance analysis of solar-PV generator is performed using
historical long-term average meteorological data when using the EPMs while
the ECBMs can be used for simulating the generator using either the
historical or the existing (online) data.
The nomenclature is listed in Appendix 9. This nomenclature will be used
in the rest of this chapter and the next chapter.
10.5.1 Resource Assessment and Energy Production (RA&EP) Models

of solar-PV modules
As shown in Fig. 10.12(a), the energy available from a solar-PV
generator is the energy output from the PCU.
(10.4)
where S is the overall solar collector area of the PV-generator given by

. The number of modules m is defined by ,which
should be approximated to the nearest higher integer number.
The average module temperature is given by
(10.5)
where is a correction factor to account for the effect of off-optimal tilt

angle of the PV-generator on the temperature of the PV-modules.
200
201
https://sam.nrel.gov/
470
(10.6)
The approximate optimal tilt angle of a PV-generator and the declination

angle are defined by,
(10.7)
(10.8)
The total daily radiation in the plane of the PV-generator equals to the
summation of the hourly irradiance in the plane of the PV-generator i.e.
. The estimation of involves the determination of the hourly
beam (or direct) and diffuse irradiance on a horizontal surface for all
hours of an average day having the same daily global radiation as the
monthly average. The definitions of the global solar radiation of the
horizontal surface (GSR)202 and its components i.e. the direct and diffuse
radiations are illustrated in Fig. 10.36.
The monthly mean hourly global radiation on a horizontal surface H can
be correlated to monthly mean daily global radiation on a horizontal surface
using a proper mathematical correlation model rt i.e. .
Inappropriate solar correlation model selection for a specific design site is a
major cause for inaccurate system sizing of solar-PV generators. Therefore, it
is recommended to use the most appropriate correlation model for the design
site in sizing PV-generators203. Two accepted correlation models are
considered in this book. The best model is to be selected for the considered
study site. These correlation models are the Liu and Jordan (LJ) given by
equation (10.9) and Collares-Pereira and Rabl (CR) given by equation
(10.10). The coefficients of the CR model are defined in equations (10.11)
202
Abdo, T., & EL-Shimy, M. (2013). Estimating the global solar radiation for solar
energy projects–Egypt case study. International Journal of Sustainable Energy, 32(6),
682-712.
203
Yesilata, B., & Firatoglu, Z. A. (2008). Effect of solar radiation correlations on system
sizing: PV pumping case. Renewable Energy, 33(1), 155-161.
471
and (10.12). The solar hour angle and the sunset hour angle are given
by equations (10.13) and (10.14) respectively.
Fig. 10.36: GSR, direct, and diffuse radiations
(10.11)
(10.12)
(10.13)
(10.14)
472
The hourly beam (or direct) and diffuse irradiance, on a horizontal

surface, are given by equations (10.15) and (10.16) with being the Liu and
Jordan (LJ) correlation model. Equation (10.16) is valid for .
The hourly irradiance in the plane of the PV-generator is then defined by
equation (10.17).
(10.15)
(10.16)
(10.17)
where the tilt factors for beam Rb and diffuse Rd radiations are defined by
10.5.2 Equivalent Circuit based Models (ECBMs) of solar-PV

generators
To be familiar with the models of solar-PV cells, modules, and arrays,
it is recommended to read section 11.2.1 in chapter 11; however, this is not
essential if the reader knows the fundamental models of photovoltaics.
The Sandia PV Array Performance model204 consists of a set of equations
that provide values for five points on a module's I-V curve (which is the
relation between the output voltage and current curve from the PV module as
illustrated in Fig. 10.37 and a database of coefficients for these equations.
204
King, D. L., Boyson, W. E., & Kratochvil, J. A. (2004). Photovoltaic Array
Performance Model, Sandia National Labs.
473
The values of these coefficients are stored in the Sandia Modules library 205.
These coefficients have been empirically determined based on a set of
manufacturer specifications and measurements taken from modules installed
outdoors in real, operating systems. The five points include the three classical
points on the IV curve, i.e. short-circuit current (Isc), open-circuit voltage
(Voc), and the maximum-power point voltage and current (Vmp and Imp
respectively). The remaining two points are the current (Ix) at the half of the
open-circuit voltage and the current (Ixx) at a voltage midway between Vmp
and Voc.. Sandia Model is described by the following equations.
Fig. 10.37: PV module characteristics (IV curve: red; power curve: blue)
showing the five points required by the Sandia performance model
205
System Advisor Model (SAM). SAM's Help system. Available: http://sam.nrel.gov
474
The California Energy Commission (CEC) Performance Model uses the

University of Wisconsin-Madison Solar Energy Laboratory's five-parameter
model with a database of module parameters maintained by the CEC for the
California Solar Initiative. The database of this model tends to be more up-to-
date than Sandia model database. Therefore, this model used in this chapter.
The CEC model uses the equivalent circuit shown in Fig. 10.38 for modeling
PV cells. This circuit includes a series resistance and a diode in parallel with
a shunt resistance. The letter V represents the voltage at the load. As
previously explained, this circuit can be used for an individual cell, for a
module consisting of several cells, or for an array consisting of several
modules. The basic equations of the model are,
475
Fig. 10.38: Equivalent circuit representing the PV cell used in the ECE
model. Note: the light current (IL) is also called the photocurrent (Iph)
The flat plate simple efficiency model is a simple representation of

module performance that requires the user to provide the module area, a set
of conversion efficiency values, and temperature correction parameters. It is
476
recommended to use this model for the modules that do not exist in the CEC
module database. The model calculates the module's hourly DC output,
assuming that the module efficiency varies with radiation incident on the
module as defined by the radiation level and efficiency table.
10.6 Case Studies

For reducing the volume of the book, the case studies related to the
presented characteristics, and modeling of solar-PV systems will be presented
in some external references by the main author of this book.
The characteristics and tilt angle optimization of standalone solar PV-
generators can be found in reference206. In addition, this reference provides
an improved method for sizing minimization of solar-PV generators
supplying deferrable loads. The considered load is irrigation water pumping.
In reference207, many studies are presented. The most relevant study is the
detailed analysis of the effectiveness of various sun tracking options is
presented. The effectiveness is identified based on the impact of various sun
tracking options on the energy production and the LCOE considering a wide
range of latitudes.
Detailed analysis of the dynamic performance of grid-connected solar-
PV generator will be considered in the next chapter. This chapter as well as
the next two chapters can be considered as an extended chapter handling
most of the technical and economical issues associated with solar-PV
systems.
206
333-350.
207
477
478
Chapter 11
Modeling, Dynamic Analysis, and Control of Grid-
Connected Solar-PV generators
11.1 Introduction
The integration of large amounts of solar PV power in to the electricity
grid presents new challenges 208. These challenges are mainly related to the
modeling, control, performance, technical impacts, operational impacts, and
techno-economic feasibility. One of the main operational and control targets
of solar PV generators is the extraction of the maximum available active
power. This available power is not only depends on the solar-PV generator
but also highly related to various meteorological conditions. The solar
irradiance and temperature are the main environmental input variables that
affect the power production. The sun tracking and the MPPT are also among
the main factors that affects the energy extraction. In the previous chapter,
various MPPT techniques are presented while in this chapter the dynamic
performance of grid-connected solar-PV generator equipped with some
selected MPPT methods will be presented. In addition, enhanced control of
the generator MPPT for better dynamic performance will also be presented.
The modeling of solar PV cells and generators is considered in many
previous researches 209,210,211,212,213. These researches provide methods for
estimating the parameters of solar PV cells, modules, and arrays. In addition,
mathematical models for the simulation of the behavior and performance are
presented considering variations in the temperature and solar irradiance.
208
European Photovoltaic Industry Association. (2014). Global market outlook for
photovoltaics 2014–2018.[Online] Available: http://www. epia. org/filead
min/user_upload. Publications/44_epia_gmo_report_ver_ 17_mr. pdf.
209
De Soto, W., Klein, S. A., & Beckman, W. A. (2006). Improvement and validation of a
model for photovoltaic array performance. Solar energy, 80(1), 78-88.
210
Chenni, R., Makhlouf, M., Kerbache, T., & Bouzid, A. (2007). A detailed modeling
method for photovoltaic cells. Energy, 32(9), 1724-1730.
211
Tian, H., Mancilla-David, F., Ellis, K., Muljadi, E., & Jenkins, P. (2012). A cell-to-
module-to-array detailed model for photovoltaic panels. Solar Energy, 86(9), 2695-2706.
212
Chouder, A., Silvestre, S., Sadaoui, N., & Rahmani, L. (2012). Modeling and
simulation of a grid connected PV system based on the evaluation of main PV module
parameters. Simulation Modelling Practice and Theory, 20(1), 46-58.
213
Yang, Y., Chen, W., & Blaabjerg, F. (2014). Advanced Control of Photovoltaic and
Wind Turbines Power Systems. In Advanced and Intelligent Control in Power Electronics
and Drives (pp. 41-89). Springer International Publishing.
479
Generally, as shown in the previous chapter, the output power of PV

generators is nonlinear and varies according to the available level of
irradiance and temperature. In addition, at certain operating point, the PV
module produces its maximum output power; the MPP. The MPPT
controllers are developed to derive the maximum power from solar PV
generators214,215,216. In comparison with other MPPT techniques, the perturb
and observe (P&O) technique is found to be simpler in structure and easier in
implementation. In addition, the P&O-based MPPT is widely used in many
installations. The operation of P&O-based MPPT controllers is solely
dependent on the output DC voltage and current values at the solar PV
generator terminals. Therefore, no meteorological sensors are needed for its
operation and implementation.
Despite these advantages, the conventional P&O exhibits a trade-off
between the tracking speed and steady state oscillation. However in this
chapter a new method is introduced to solve the problem of output power
oscillation caused by perturb and observe technique.
After extracting the extracted maximum DC active power from the solar-
PV plant, it is required to inject this active power in its specified AC form
along with the demanded reactive power to the power grid. This task can be
done by the inverter. Stability analysis is required to identify the effect of
solar-PV plants on the grid, and the response of the solar-PV plant to the
sudden changes in grid parameters (voltage, frequency… etc).
In this chapter the dynamic performance of a grid connected PV generator
will be investigated. Various types of disturbances are considered. These
disturbances include changes in meteorological conditions i.e. the solar
irradiance and the temperature. Grid disturbances such as voltage sag, voltage
swill, frequency fluctuations and strength of the grid coupling are also
important stability studies.
214
Salam, Z., Ahmed, J., & Merugu, B. S. (2013). The application of soft computing
methods for MPPT of PV system: A technological and status review. Applied Energy, 107,
135-148.
215
Onat, N. (2010). Recent developments in maximum power point tracking technologies
for photovoltaic systems. International Journal of Photoenergy, 2010.
216
Ahmed, J., & Salam, Z. (2014). A Maximum Power Point Tracking (MPPT) for PV
system using Cuckoo Search with partial shading capability. Applied Energy, 119, 118-
130.
480
11.2 Modeling of solar-PV generators for dynamic analysis

Configurations for the grid-connected solar-PV-generator (SPVG) are
shown Fig.11.1. The SPVG is connected to the grid via a transformer at a bus
called the point of common coupling (PCC). For grid interconnection, the DC
power produced by the SPVG is transformed to a specified AC power
according to the grid specifications. This is performed by a power electronics
based grid interface. The grid interface performs three main functions: (1)
MPPT; (2) Control of the active and reactive power injections to the PCC; (3)
Filtration of harmonics and protection against current and voltage surges.
(a)
(b)
Fig. 11.1: Grid-connected solar-PV generator block diagram; (a) With DC-
DC converter for MPPT control; (b) Without DC-DC converter
481
SPVGs are usually composed of many components as shown in the figure.

The first component is a solar array that converts solar energy into DC
electricity. The sun trackers are providing sun tracking control for
maximization of the solar resource extraction. In this chapter, the dynamics
associated with the sun tracking systems are neglected. This assumption is
valid during the short period of study as no change in the sun position is
assumed during that period.
According to the characteristics and design of the system, the MPPT
control may be achieved either by controlling a DC-DC converter (Fig.
11.1(a)) or directly controlling the inverter which is the DC-AC converter
(Fig. 11.1(b)). In both situations, the output voltage (and/or the current) of the
SPVG is adjusted according to the MPPT algorithm (see chapter 10). The
DC-link provides a stabilization facility to the input DC voltage to the
inverter. The inverter is not only providing conversion from DC power to AC
power and possibly MPPT control but also provides control of the active and
reactive power injections to the grid. There are some controllers for achieving
the sun tracking, MPPT, and power control. The connection filter block
provides a tool for enhancing the power quality of the output power and also
provides some protection capabilities such as surge protection.
In this chapter, the configuration shown in Fig. 11.1(a) is considered. The
modeling of various components and the dynamic simulation of the system as
well as control enhancements will be presented. It is also assumed that the
array is controlled in a centralized manner.
11.2.1 Solar-PV array modeling

The I-V and P-V characteristics presented in the previous chapter can be
found using measurements. Mathematical modeling can also be used for
representing these characteristics and it will be considered in this section.
The solar-PV array model will be presented starting from the ideal solar-
PV cell model. An ideal solar- PV cell can be represented by a current source
connected in parallel with a diode; Fig. 11.2. This is because the cell
generates current when it is illuminated and acts as a diode when it is
not.
The physical structure of a solar cell is similar to that of the
popular P-N junction diode. When this junction is subjected to sunlight, the
photons may pass straight through the junction if the photon energy is lower
482
than the silicon energy band gap value 217 which is 1.1 eV or reflect off the
surface or absorbed by the silicon if the photon energy is higher than the
silicon band gap value 218. The absorbed photons generate an electron-hole
pair and sometimes heat depending on the band structure. For solar-PV cells,
the absorbed photons generate electron-hole pairs. As a result, an open-circuit
voltage appears at the terminals of the junction. If the junction is connected to
an external circuit, a current will flow on the external circuit causing
recombination of the electrons and holes. If the load resistance is zero, then
the current is called the short-circuit current. This process is called the light-
generated or the photo-generated current which is directly proportional to the
solar irradiance and it is demonstrated in Fig. 11.3 which is based on the
http://pveducation.com.
(b)
(a)
Fig. 11.2: Solar-PV cell; (a) Ideal equivalent circuit; (b) Circuit symbol
As shown in Fig. 11.2(a), the cell output current (I) takes the form,
217
https://en.wikipedia.org/wiki/Band_gap
218
In solid-state physics, a band gap, also called an energy gap or bandgap, is
an energy range in a solid where no electron states can exist. In graphs of the
electronic band structure of solids, the band gap generally refers to the energy
difference (in electron volts) between the top of the valence band and the
bottom of the conduction band in insulators and semiconductors. In solar-PV
cells, the optical band gap determines the portion of the solar spectrum a cell
can absorb. A semiconductor will not absorb photons of energy less than its
band gap. In addition, the energy of the electron-hole pair produced by a
photon is equal to the band gap energy. The band gap of silicon is 1.1 eV and
its 0.67 eV for Germanium.
483
where Iph is the light-generated current (A) which equals to the short-
circuit current and ID is the diode current (A). The diode current takes the
form,
Therefore, the cell’s output current takes the form,
The value is called the thermal voltage because of its units and
because it is only dependent on the temperature. Defining VT as,
Therefore, equation (11.3) can be written as,
where Io is the saturated reverse current or the leakage current of the diode
(A); q is the electron charge (= 1.602  10-19 C); k is the Boltzmann constant
(= 1.38  10-23 J/oK); V is the cell output voltage (V); Tc is the cell operating
temperature (oK); a is the diode ideality factor. The ideality factor (also called
the emissivity factor) is a fitting parameter that describes how closely the
diode's behavior matches that behavior predicted by theory.
In theory, the p-n junction of the diode is assumed to be an infinite plane
and consequently no recombination occurs within the space-charge region. A
perfect match to theory is indicated when a = 1. When recombination in the
484
space-charge region dominate other recombination, however, a = 2. Most

solar cells, which are quite large in comparison with the conventional diodes,
can be approximately considered as an infinite plane. Therefore, the cell
usually exhibits a near-ideal behavior under STC i.e. a  1 at STC. As
illustrated in Fig. 11.4, the value ideality factor is also technological
dependent219,220. The reverse saturation current (Io) can be determined at
open-circuit conditions where I = 0 and V = Voc. Consequently, the reverse
stautaion current can be determined based on the datasheet parameters at STC
conditions using,
Fig. 11.3: The light-generated current
219
Bellia, H., Youcef, R., & Fatima, M. (2014). A detailed modeling of photovoltaic
module using MATLAB. NRIAG Journal of Astronomy and Geophysics, 3(1), 53-61.
220
Tsai, H. L., Tu, C. S., & Su, Y. J. (2008, October). Development of generalized
photovoltaic model using MATLAB/SIMULINK. In Proceedings of the world congress
on engineering and computer science (Vol. 2008, pp. 1-6). Citeseer.
485
Fig. 11.4: Impact of solar-PV technologies on the diode ideality factor
or
Later, the impact of changes in the irradiance and the temperature on the
cell characteristics as well as the impact of the cell losses will be presented.
In the equivalent circuit of Fig. 11.2(a) which describes an ideal cell,
losses are neglected. In reality, the losses reduce the output power and affect
the overall I-V characteristics of the cell. The resistivity of the semiconductor
material and the contacts can be represented by a series resistance (Rs)
connected at the terminals of an ideal cell as shown in Fig. 11.5. In this case
equation (11.6) can be written as,
The voltage at the terminals of the diode (VD) is related to the cell’s
terminal voltage by . Therefore, (11.9) becomes,
486
Fig. 11.5: Equivalent circuit of a solar-PV cell with semiconductor and

contact losses (also called one-diode four-parameters model)
The I-V relation and the open-circuit voltage can be then defined by,
Around the STC, the value of the photocurrent is nearly equals to the
short-circuit currents i.e.
In the following, the impact of the variations in the solar irradiance and
temperature will be presented. Therefore, the values of various variables and
parameters at STC (1000 W/m2 and 25oC) will be illustrated by the suffix ref.
The reference values are usually given in the datasheets of Photovoltaics. The
short-circuit current at a temperature Tc is then,
487
where Ht is irradiance at the plane of the solar-cell which dependent on the

tilt angle of the cell. KTi is temperature coefficient of the short-circuit current
(/oK). The change in the cell temperature (K) is defined by,
The change in the short circuit current as affected by changes in the

irradiance and temperature is then,
The reverse saturation current at STC can be found at open-circuit

conditions (i.e. I = 0) using (11.13)
i.e.
The changes in the reverse situation current as affected by temeperature

changes is given by,
where Eg is the band-gap energy of the semiconductor material; 1.1 eV for

silicon. Based on (11.19), (11.18) is then
488
The output power is then takes the form,
The value of the -1 in (11.6) or (11.7) is small enough in comparsion

with the exponential term. The -1 can be then neglected if more
simplifications are desired. Under this assumption, the series resistance can
be determined as,
This is illustrated in Fig. 10.15 based on measured characteristics. The

value of this resistance is usually determined at STC; however, its value is
dependent on the temperature and the irradiance as depicted from the stated
models.
Better representation of the losses can be achieved by adding a shunt
resistance as shown in Fig. 11.6. This equivalent circuit is called the one-
diode five-parameters circuit model. Therefore, in this model, two parasite
resistances are considered; the series resistance (Rs) and the shunt resistance
(Rsh) such as Rsh >> Rs is a usual relation. Therefore, Isc  Iph.
Fig. 11.6: Equivalent circuit of a cell considering the losses
In this case, the current balance is stated as,
489
Consequently (11.10) becomes,
As shown in Fig. 10.15, the STC values of the series and shunt resistances
can be approximated by,
As shown in Fig. 10.13 and 11.7, a solar-PV module is constructed from a

number of identical cells. Series and parallel connections of the cells
respectively define the voltage and current ratings of the module. The
presented cell models can be used to describe the module characteristics by
modifying the cell characteristics according to the number of series and
parallel cells within a module. The module voltage and current are then
defined by,
Fig. 11.7: Solar-PV module with Ns series cells and Np parallel cells.
490
The equivalent circuits of the module considering various models are

shown in Table 11.1. This table illustrates the equivalent circuit of the ideal
model (a), the single-diode four-parameter model (b), and the single-diode
five-parameter model (c). The previous models of a cell are also valid for
representing a module provided that all the quantities are modified according
to the connections of the cells within the module. The characteristics of the
module are illustrated in Fig. 11.8.
Fig. 11.8: Characteristics of a solar-PV module consists of Ns series cells

and Np parallel cells
The solar-PV array model can be derived based on the same concepts.
Therefore, an array with Ms series modules and Mp parallel modules can be
represented also as shown in Table 11.1. For the five-parameter model,
491
Table 11.1: Solar-PV array models

Model
(a) Ideal model Representation
Note: , ,
(b) Four-parameter model
Note:
(c) Five-parameter model
Note:
492
If the impedance of the wiring connecting the modules is negligible, then
The accuracy of the one-diode five-parameter model is evaluated

considering the I-V characteristics of the Sanyo-HIP-225HDE1 module. The
module parameters are shown in Table 11.2. The simulation will be
performed at the STC temperature while the irradiance is varied from 200
W/m2 to 1000 m2. The results are then compared with the datasheets of the
considered module. This is shown in Fig. 11.9.
11.2.2 DC-DC Converters and MPPT control

Considering the configuration shown in Fig. 11.1(a) where a DC-DC
converter is controlled for providing the MPPT requirement. Generally, the
MPPT controls the DC-DC converter for regulating the input voltage at the
493
array’s MPP and also capable of providing load matching for the maximum
power transfer by controlling the input resistance of the array to match the
load resistance. Various MPPT algorithms are described in Chapter 10. In
this chapter two popular MPPT algorithms will be considered. These
algorithms are the P&O and the IC. In addition, these algorithms will be
enhanced for better dynamic performance of the SPVG.
Table 11.2: Technical data of the Sanyo-HIP-225HDE1 module

Quantity Symbol Value
Number of series cells per module Ns 60
Number of paralell cells per module Np 0
Maximum power (W) Pmp 225
Maximum power voltage (V) Vmp 33.9
Maximum power current (A) Imp 6.64
Open circuit voltage (V) Voc 41.8
Short circuit current (A) Isc 7.14
Temperature coefficient of Pmax (% / oC) KTPmax -0.3
Temperature coefficient of Vo.c (V / oC) KTv -0.105
Temperature coefficient of Is.c (mA / oC) KTi 2.14
(a) (b)
Fig. 11.9: I-V characteristics of the Sanyo-HIP-225HDE1 module; (a)
Simulated by the five-parameter model; (b) Manufacturer measured data
The comparison indicates the accuracy of the presented model.
494
Generally, DC to DC converters belong to a facility of converters

called switching mode DC-DC converters 221. In comparison with their input
voltage, DC-DC conveerters are capable of increasing or decreasing their
output voltage in a controlled manner. In addition, the DC-DC converters are
capable of providing impedance matching. DC-DC converters have a wide
range of applications in DC power supplies. The salient applications can be
found in battery equipped systems such as laptops, hybrid electric cars, and
UPSs. In addition, they have a significant role in SPVG for various modes of
SPVG operations (see chapter 10) as well as in SPVGs with battery storage.
The main function(s) of DC-DC converters is achieved by fast
switching of linear passive components such as inductors and capacitors;
however, the switching itself is provided by nonlinear controlled electronic
fast switches such as MOSFETs. In the MPPT, the duty cycle of the DC-DC
converter is controlled for achieving the MPP according to the MPPT
algorithm.
The MPPT feedback control can be based on two variables 222; the
terminal voltage of the array or the array output power. In the terminal
voltage MPPT control, the terminal voltage of the array is used as the
feedback control signal. The MPPT is achieved by regulating the array’s
terminal voltage for matching it to a desired level which should be the MPP
voltage. The main problem with this control approach is its insensitivity to
wide changes in the irradiance and the temperature as the desired voltage
level setting is fixed and valid for a narrow range for meteorological
variations. Therefore, the terminal voltage based control is not suitable for the
SPVG applications; however, it is very suitable in satellite systems where the
solar irradiance is nearly constant. The main tracking problems associated
with the terminal voltage based control is avoided with the use of the output
power as a feedback control signal. In this case, the MPPT control forces the
dP/dV to zero. Therefore, the MPPT control system will work satisfactory for
any solar-PV system as its operation is not mainly dependent on the
221
George, K. (2006). DC Power Production, Delivery and Utilization: An EPRI White
Paper. Available at: h ttp://www. epri. org [Accessed: 2.4. 12].
222
Hua, C., & Shen, C. (1998, May). Study of maximum power tracking techniques and
control of DC/DC converters for photovoltaic power system. In Power Electronics
Specialists Conference, 1998. PESC 98 Record. 29th Annual IEEE (Vol. 1, pp. 86-93).
IEEE.
495
characteristics of the array; however, this method maximizes the power to the
load rather than the power from the array.
The basics and various types of DC-DC converters will be presented in
the following. Three types of DC-DC converters will be considered; buck
(step down converter), boost (step up converter), and buck-boost converters.
The fundamental DC-DC converter for stepping down the DC input voltage
is shown in Fig. 11.10 where R is the equivalent load resistance. As shown
in Fig. 11.10(a), switching on and off of the switch in a periodic manner
causes as shown in Fig. 11.10(b) reduction in the output voltage in
comparison with the input voltage. The switch control is usually achieved by
pulse width modulation (PWM) in which the switch is closed for an on
duration (ton) and open for an off duration (toff) in a periodic way. The PWM
control signal is shown in Fig. 11.10(c).
With a switching periodic time Ts, the duty cycle (D) of the switch is
defined as,
The output voltage (vo(t)) and its average value (Vo) are then represented
by,
It is clear from Fig. 11.10(b) and (11.41) that the output DC voltage
contains harmonics due to the switching action (called switching harmonics)
as shown in Fig. 11.11. The switching frequency is related to the switching
periodic time by,
496
(a) (b)
(c)
Fig. 11.10: Fundamentals of DC-DC step down control; (a) conceptual

switching circuit; (b) Power voltage waveforms; (c) PWM and lowpass filter
Fig. 11.11: Switching harmonics in the output voltage
These harmonics may cause degradation or even malfunction of the

operation of the supplied DC load (in this case, the load is the inverter input
port). Therefore, either elimination or reduction of these harmonics to a safe
level defined by the load sensitivity is required (the reader may refer to the
497
IEEE 519223 standard for more details). This problem can be solved by the
use of lowpass filters as shown in Fig. 11.10(c). This filter passes the DC
component while attenuates the switching harmonics. The filter action is
governed by its L and C components. The series inductor (L) presents zero
impedance to the DC component of the output voltage while it presents
increasing impedance with the increase in the switching frequency.
Therefore, the series inductor attenuates significant amounts of the
harmonics. On the other hand, the shunt capacitor presents an infinite
impedance to the DC component while it traps the harmonics to the ground
due to the reduction of its impedance. The corner frequency (fc) of the filter
(see Fig. 11.12) is defined as,
To be effective in attenuating the switching harmonics, the corner

frequency of the filter should be less than the switching frequency (i.e. fc <<
fs).
Fig. 11.12: Selection of the corner frequency of the lowpass filter
Fig. 11.10(c) illustrate a feedback PWM control for adjusting the output
voltage of the converter to a desired reference level by controlling the duty
cycle of the switch. Although the PWM control is described for stepping
down the DC voltage (i.e. buck control action), all the other considered
converters are also based on the PWM control. In a realistic buck converter
as shown in Fig. 11.13, the diode is used for defining the current pass during
223
Duffey, C. K., & Stratford, R. P. (1989). Update of harmonic standard IEEE-519: IEEE
recommended practices and requirements for harmonic control in electric power systems.
Industry Applications, IEEE Transactions on, 25(6), 1025-1034.
498
the off periods of the switch. The switch may be a transistor such as BJT or
MOSFET.
(a)
(b) (c)
(d)
Fig. 11.13: Buck converter; (a) Circuit representation; (b) switch on; (c)
Switch off; (d) Inductor ripple current variations due to charging and
discharging of the capacitor
The analysis of switching circuits that contains inductors and capacitor

needs understanding of two main concepts; the inductor volt-second balance,
and the capacitor charge balance.
For understanding of the inductor volt-second balance, consider the
simple circuit shown in Fig. 11.14. In this circuit, the voltage is forced across
the inductor. Therefore, a current flow through the inductor such that,
499
(a)
(b) (c)
Fig. 11.14: Forcing voltage across an inductor
The k is the integration constant that presents the initial current. Fig.
11.14(b) shows the current as a function of time for k = 0. Now suppose that
the impressed voltage is a DC voltage of a value VL, then the solution of
(11.46) becomes,
Equation (11.47) and Fig. 11.14 (b) show that the inductor current will be
increases to infinity as the voltage is forced across its terminals and no steady
state value will be reached. If the inductor is not pure and having a parasite
small resistance (r), then a steady state very high current that is limited by the
value of this resistance will be reached. This current value may be high
enough to destroy the inductor. Therefore, whenever voltage is applied across
a pure, the reactor current will be a time dependent linearly increasing
500
function even if the voltage is DC. For the switching circuit shown in Fig.
11.13, the output voltage waveform takes the form shown in Fig. 11.15.
(a)
(b)
(c)
Fig. 11.15: Output voltage of a buck converter; (a) Circuit; (b) Output
voltage waveform; (c) Inductor voltage and current waveform
If the output voltage ripple is small enough to be neglected, then the

output voltage may be considered a constant value of magnitude Vo. When
the switch is close i.e. position 1 (Fig. 11.13(b)), the voltage across the
inductor is then Vdc – Vo while its value is –Vo when the switch is open
(position 2). Based on (11.47), the value of the inductor current takes the
form,
501
Based on (11.46) and over one complete switching period, the following
relation holds,
where the term presents the net variations of the current

over one switching cycle. During periodic steady state operation, these net
current variations should be zero i.e.
Equation (11.50) shows that the average voltage across the inductor is
zero. Equation (11.50) is known as the volt-second balance of inductors. For
the buck converter and neglecting the output voltage ripple, the volt-second
balance can be written as,
Therefore,
Equation (11.52) is in agreement with the fundamental switching concept

of Fig. 11.10. The capacitor charge balance is based on the relation,
Integrating (11.53) over a complete operation cycle results in,
502
During a steady state periodic operation, the net changes in the capacitor
voltage is zero i.e.
Equation (11.55) shows also that the average capacitor current is zero.
This is called the capacitor charge balance.
It should be noted that Fig. 11.13(d) shows the inductor current in the
continuous mode of operation where the inductor current do not drop to zero
during switching off periods. This is the normal mode during high loading
conditions; however, during light loading conditions, the inductor current
may be decrease to zero before the end of the next switching on. Therefore,
the load current becomes discontinuous. A situation which may lead to
problems in the operation of the supplied load.
Boost converters (Fig. 11.16) are performing the opposite function of
buck converters as they increase the output voltage in comparison with the
input voltage. During the on operation of the switch, the input voltage is
forced across the inductor. As a result, the inductor current increases. In this
case, the load current is supplied by the capacitor. When the switch is off, the
inductor voltage is added to the input voltage and supplies the capacitor by
appositive voltage. Therefore, the diode becomes forward biased and the
capacitor charges to a voltage that is higher than the input voltage. During the
switch off of the switch, the inductor current supplies both the load and the
capacitor. The inductor current waveform is similar to that shown in Fig.
11.13(d). With the switch on the inductor voltage is Vdc while its value is
approximately Vdc – Vo when the switch is closed. Therefore, the inductor
volt-second balance presented by (11.50) takes the form,
Therefore, average output voltage takes the form,
503
(a)
(b) (c)
Fig. 11.16: Boost converter; (a) Circuit representation; (b) Switch on; (c)
Switch off
(a)
(b) (c)
Fig. 11.17: Buck-Boost converter; (a) Circuit representation; (b) Switch on;
(c) Switch off
504
Unlike the buck and the boost converters, the buck-boost converter (Fig.
11.17) is capable of stepping down or stepping up the output voltage i.e. it is
capable of performing the boost and buck functions by controlling the duty
cycle of the switch. The Buck-Boost is supplied by a DC input voltage and
produces a DC output voltage that is in opposite polarity with respect to the
input voltage. The negative output voltage is controllable and it can be either
larger or smaller than the input voltage.
The inductor volt-second balance can be then stated as,
A summary of the considered DC-DC converters is shown in Table 11.3.
The presented models of these converters consider ideal elements where

the losses are neglected. Therefore, the presented models consider ideal
converters. Under ideal conditions, the input power equals to the output
power i.e. the efficiency is 100% and consequently, generally,
The MPPT by the use of DC-DC converters can be explained considering

Fig. 11.18. In Fig. 11.18(a), the PV array is supplying a resistive load of a
constant resistance R. The characteristics of the array at the operating
meteorological conditions are shown in Fig. 11.18(c). In addition, the load
line (V/R – V) is shown in the figure. Under these conditions, the system
operates at point (a) where the array current and voltage are respectively Ii
and Vi. The power consumed by the load is Pi which is less than the MPP
power (Pmp).
505
Table 11.3: Summary of DC-DC converters
Type
General
Buck
Boost
Buck-boost Configuration Characteristics
On the I-V characteristics of the array, the locus of various constant power
values is shown. It is clear that the load consumption will increase to the
506
MPP value if the load voltage is increased to . At this value of the voltage
across the load terminals, the load consumes the MPP power value; however,
this point (i.e. point b) is not available on the characteristics of the array. This
problem can be solved by inserting the DC-DC boost converter shown in Fig.
11.18(b) and operating it at the duty cycle D* at which the output voltage at
the load terminals is the desired voltage for MPPT i.e. . This operation and
control action is equivalent to controlling the load resistance by decreasing
the load resistance to R* which consumes the MPP power at the MPP voltage
(Vmp). Later in this section, some additional details about the MPPT will be
presented. The presented function of the DC-DC converter is stepping up the
voltage. Therefore, buck-boost converters are also suitable while buck
converters are not suitable as they only step down the voltage.
(a)
(b) (C)
Fig. 11.18: Operation and control concepts of DC-DC converters for MPPT
Recalling that the considered DC-DC converters are ideal lossless and
acts as ideal DC step-up and/or step-down DC electronic transformers. In
507
reality there are power losses associated with the inductor parasite resistance,
the switches, and other losses. The losses are highly dependent on the types
of used switches and the quality of components 224,225. The efficiency and
impact of these losses on the input – output voltage relation can be
represented by,
The efficiency curve of the converter can be obtained from either tests or
manufacturer datasheets. Fig. 11.19 illustrates the representation of the
converter losses in the equivalent circuit by an equivalent loss resistance req.
Therefore, in this representation, an ideal voltage transformation is
considered.
Fig. 11.19: Representation of the converter losses
11.2.3 DC-AC Converters (Inverters) and Power control

An inverter (Fig. 11.20) is equipment that converters a DC supply to a
controllable AC supply. During this energy conversion process, the DC
power is transferred to the AC side at specified frequency and voltage
magnitude. These specifications are determined by the AC side load or
system requirements.
224
Espinosa, C. A. L. (2014). State Variable Model for Considering the Parasitic Inductor
Resistance on the Open Loop Performance of DC to DC Converters. Journal of Computer
and Communications, 2(13), 41.
225
Hua, C., & Shen, C. (1998, May). Study of maximum power tracking techniques and
control of DC/DC converters for photovoltaic power system. In Power Electronics
Specialists Conference, 1998. PESC 98 Record. 29th Annual IEEE (Vol. 1, pp. 86-93).
IEEE.
508
The inverter also converts the input DC power to desired values of active
power and reactive power. In addition, the inverters are capable of controlling
the power factor of the AC side. Automatic voltage control of the AC side is
also possible. These functions are determined by the inverter control that
eventually controls the on/off actions of various electronic switches that
comprise the inverter.
Fig. 11.20: DC to AC conversion via an inverter
The fundamentals of the DC – AC power conversion and power control

will now be considered. Fig. 11.21 shows the fundamental single-phase
inverter. The DC source may be a battery or the output of a rectification
circuit or the output of a solar-PV array. The inverter consists of four
switches (S1 to S4) shown in the figure. Each couple of these switches is
paired together for providing a closed path when they are turned on. In the
circuit, S1 and S4 are coupled while S3 and S2 are coupled. Each group of
switches is turned on and off simultaneously. In the normal operation, each
group of switches are sequentially turned on and off in an opposite sequence
in comparison with the other group. As a result, an AC waveform is produced
at the inverter output.
For symmetrical output voltage, the on and off durations should be
identical. The frequency of output voltage is determined by the duration of a
complete on-off cycle (i.e. the periodic time) i.e. . The control of the
output voltage magnitude can be generally achieved by the PWM. In the
PWM, the on duration of each group of switches is divided into several
smaller periods of cyclic on/off action of the group that is supposed to be
509
turned on during that on duration. Consequently, instead completely turning

on each group for a duration T/2, each group is effectively switched on for a
reduced duration ( ). If the PWM is not applied, then and the
magnitude of the voltage is E. The magnitude of the output AC voltage can
be then generally determined as . The PWM control can also be
used for reducing the output harmonics.
(a)
(b) (c)
(d)
Fig. 11.21: Fundamental single-phase inverter; (a) Operation; (b) Output
frequency control by periodic time adjustment; (c) Control of the output
voltage magnitude by PWM; (d) Reduction of harmonics using PMW control
510
The conversion of a DC supply to three-phase AC is performed using the

same concept of the single phase inversion; however, balanced phase shift of
120o is adjusted between the phases as shown in Fig. 11.22.
Referring to Fig. 11.20, the equivalent circuit of the system and its
phasor diagram are shown in Fig. 11.23. In this system the inverter is
assumed to be connected to a large grid that can be considered as an infinite
bus. The point of connection is called the Point of Common Coupling (PCC).
With respect to a synchronized reference, the phase angle of the PCC voltage
is  while its magnitude is VPCC. The load angle of the inverter is  while the
inverter output voltage magnitude is Vac. Therefore, the phase between the
PCC voltage and the inverter output voltage is  while the phase angle of the
inverter output voltage with respect to the reference is +. The phase angle
between the inverter current and the inverter terminal voltage is the power
factor angle . The active and reactive power transfer between the inverter
and the grid are given by (see Appendix 1),
Fig. 11.22: Fundamental three-phase inverter
511
Fig. 11.23: Equivalent circuit of an inverter connected to an infinite bus

through reactive circuit and the associated phasor diagram
It is clear from (11.63) and (11.64), that the active power transfer is highly
sensitive to the load angle while the impact of the voltage magnitude on the
active power is minor. On the other hand the reactive power transfer is highly
sensitive to the voltage magnitude in comparison with the load angle.
Therefore, from the control point of view, the reactive power flow can be
effectively controlled by controlling the output voltage of the inverter while
the active power control can be effectively achieved by controlling the load
angle of the inverter. As will be shown later, these control actions can be
realized by controlling the direct and quadrature current components of the
voltage source inverter (VSI) output current.
The power control system of the VSI using the current control is briefly
illustrated in Fig. 11.24. The output interface contains an LC filter and
possibly as protection and control devices.
512
Fig. 11.24: Inverter current based power control of VSIs
The current control scheme is mainly a method of translating the desired

or reference output active and reactive power values (Pref, Qref) to a reference
current phasor that is for setting the converter switching actions by the
Sinusoidal Pulse Width Modulation (SPWM) scheme. Based on the
fundamental current component of the VSI output current and the
synchronized measurements of the PCC voltage phasor, the fundamental
reference current peak and power factor angle are,
513
Fig. 11.25: Active and reactive power control based on the synchronized
d-q control scheme
The d-q based current control scheme can be used instead of the presented
current control scheme especially for decoupling the active and reactive
power control. The d-q based control scheme is shown in Fig. 11.25. Detailed
modeling of this scheme is available in many publications; however, the L.
Wang (ed.) textbook226 is recommended as it uses the same reference
synchronized frame of this book. The implementation of the overall grid-
connected solar-PV generator on the MATLAB Simulink will be presented in
the following.
A. Transformations and PLL

One of the main functions of the Phase-Locked Loop (PLL) is the
tracking the angular velocity and the angle of the fundamental component of
the grid voltage, even if severe background harmonics are present. Thus, the
PLL can be regarded as a high-order band-pass filter, with zero phase
distortion. It should be noted that the grid voltage angle equals to the
integration of the grid angular velocity (= 2f). Therefore, the angle and the
226
Wang, L. (2012). Modeling and Control of Sustainable Power Systems. Springer.
514
angular velocity are not independent quantities. The PLL also is used for
setting the direct axis component of the grid voltage (vd) to zero. This is
required for decoupling the active and reactive power control functions.
These issues will be illustrated in the next section.
11.3 Matlab-Based Modeling and Simulation of the grid-

connected Solar-PV generator
In the following, modeling and simulation of the Grid-Connected
Solar-PV generator will be presented. In addition, improved MPPT control
will also be presented for improving the dynamic performance of the overall
system. The synchronous frame d-q control scheme (Fig. 11.25) is considered
in this section. The Matlab simulink will be used for the mentioned purposes.
The study system is shown in Fig. 11.26. The PV array consists of five
modules connected in series per string with 66 parallel strings. The module
type is SunPower SPR-305-WHT, the module parameters are as follows:
Open circuit voltage (VOC) = 64.2 V

Short circuit current (ISC) = 5.96 A
Maximum power voltage (Vmp) = 54.7 V
Maximum power current (Imp) = 5.58 A
Series resistance (Rs) = 0.037998 
Shunt resistance (Rsh) = 993.51 
Saturation current (Io) = 1.1753*10-8 A
Normalized ideality factor = 1.3
Temperature coefficient (Ki) = 0.003516
The PV array maximum power is 100 KW which will be considered as the

base power. The PV generator output power is 260 V (line to line) then it will
be stepped up through transformer to 25 KV. The base voltage will be taken
as 25 KV. The reference temperature and irradiance are 25 0C and 1000 W/m2
respectively. The parameters of the grid interconnection and the rest of the
system parameters can be found at the MATLAB central 227.
227
http://www.mathworks.com/matlabcentral/fileexchange/34752-grid-connected-pv-
array/content/PVarray/PVarray_Grid_PandO_avg.mdl
515
Fig. 11.26: Study system
11.3.1. Modified MPPT techniques

As explained in chapter 10, the P&O MPPY is among the popular
techniques for its simplicity and efficiency in MPP extraction; however, it
suffers from a poor dynamic performance. This poor dynamic performance is
highly attributed to the use of fixed step of the duty cycle for the DC-DC
converter for tracking the MPP. If the fixed step in the duty cycle is large, the
MPPT will accelerate the operating point toward the MPP; however, there
will be sustained oscillations around the MPP as the need precise value of the
516
duty cycle for operation at the MPP is not found. Reducing the duty cycle to
very small values will not cause sustained oscillations around the MPP;
however, the MPPT will not be effective in reasonably tracking the MPP.
Therefore, the idea of enhancing the traditional P&O algorithm is to the
design of variable duty cycle control. In the following, multiple steps of the
duty cycle will be considered. The choice of the duty cycle step depends on
the distance from the MPP. If the operating point is far from the MPP, large
duty cycle step will be used for accelerating the MPPT. On the other hand,
small duty cycle step will be commanded when the operating point is close to
the MPP. Consequently, the MPPT control will be shown to have a better
dynamic performance. The enhanced P&O algorithm is called the Improved
P&Q algorithm (IP&O). The same idea will be used for enhancing the IC
algorithm which will be called the Integral Regulator & Incremental
Conductance (I&IC) algorithm.
The enhancement approach is used depends on dividing the rate of change
of the duty cycle by two each time it crosses the MPP. A block called delta-D
calculator was added to the output of the P&O block to divide the duty cycle
by two each time it crosses the MPP, this block contains the following items:
a detect increase sub-block whose output will be 1 if the duty cycle is
increasing and zero otherwise, a detect decrease sub-block which output will
be 1 if the duty cycle is decreasing and zero otherwise, unite delay, logical
operator, and finally a block that divide the rate of change in duty cycle by 2
when the enable signal equals 1. The block simulation is shown in Fig. 11.26.
Fig. 11.27: The deltaD calculator block

517
Fig.11.27 shows the duty cycle variations with P&O and IP&O
techniques. It can be seen that with P&O technique the duty cycle is always
oscillating around the MPP, but with IP&O the duty cycle is fixed at the
MPP.
Fig.11.27: Duty cycle variations in the P&O and IP&O algorithms
The drawback of this technique is when it tracks the MPP at certain input
level (i.e. irradiance, temperature, voltage at the PCC ... etc) the rate of
change of duty cycle will be too small. Therefore, when one of the inputs
significantly change (e.g. the temperature), this technique will not be able to
track the new MPP. This problem can be solved by setting the rate of change
of duty cycle to a larger value when sensing changes in inputs. This will
explained in details later.
As an enhancement to the incremental conductance technique, an integral
regulator was added to it to minimize the error and optimize the duty cycle
correction factor. In incremental conductance technique the error equals to,
The duty cycle correction factor can be found by integrating the error; this
can be done as shown in Fig. 11.28.
518
Fig. 11.28: Integral regulator and incremental conductance (I&IC) block
The voltage source converter (VSC) is responsible for converting the DC

voltage signal into AC signal and also control the active and reactive power
injected to the grid. The VSC inputs are the DC-link voltage and the
reference signal which control the output voltage magnitude and phase shift
which in turn controls the active and reactive power injected to the grid. The
output voltage of the VSC can be calculated according to the following
equations,
(11.69)
(11.70)
Where is the fundamental component of the output voltage; vi is the

DC-link voltage; and is the amplitude modulation ratio which is defined
as the ratio of the reference signal ( ) to the carrier signal ( ) of the
sinusoidal pulse width modulation (SPWM).
The VSC control generates the amplitude modulation ratio which will be
converted to the input reference signal to the VSC. The VSC control (Fig.
11.25) consists mainly of four blocks PLL & measurement, VDC regulator,
current regulator, and reference signal generator as shown in Fig. 11.29. Each
of these blocks will be explained in details later in this section, the inputs to
this system are the measured voltage and current at point of common
coupling (PCC) and the DC-link voltage.
519
Fig. 11.29: VSC control simulation
The phase locked loop (PLL) has two functions the first one is to
determine the voltage frequency at PCC and determining the ramp (t) of
phase (a) voltage varying from (0) to (2) where  is the angular speed in
radian. The second function included in that block is the Park’s
transformation of the measured voltage and current (i.e. transform it from abc
to dq0 synchronously rotating reference frame) and then the PLL sets the
quadrature axis component of the measured voltage (Vq) to zero. This is
performed by setting the angle and the rotation speed of the dq0 reference
frame equals to the angle and rotation speed of phase (a). The PLL &
measurement block model is shown in Fig. 11.30.
Fig. 11.30: PLL & measurement block
520
The inputs to this block are the three phase measured voltage and current
at PCC. The outputs are the ramp (t) of phase (a) voltage varying from 0 to
2 and the per-unit voltages and currents in the dq0 reference frame.
The purpose of the DC regulator block is to adjust the DC-link voltage at
a pre-determined value to achieve the needed output voltage level according
to equation (11.69). The DC-link voltage can be determined according to the
following equation,
(11.71)
where Vdc is the DC-link voltage, Cdc is the DC-link capacitance, IBC is the
boost converter output current, and IVSC is the input current to the VSC, those
currents are shown in Fig. 11.31.
Fig. 11.31: DC-link current
In the VDC regulator, the measured DC link voltage is compared with the
reference signal, the error is sent to PID controller and the output will be the
reference direct axis current (Id,ref) signal which determine the output active
power as shown in Fig. 11.32.
Fig. 11.32: VDC regulator model
521
The error between the measured and reference voltage is divided by the
nominal DC-link voltage to make the output current signal in per-unit. The
inputs to VDC regulator block are the measured and reference DC-link
voltage.
In the dq0 reference frame, the active and reactive power values are given
by the following equations,
(11.72)
(11.73)
As explained before, Vq is set to zero by the PLL. Therefore, the

decoupling between the active and reactive power is achieved as,
(11.74)
(11.75)
Fig. 11.33: VSC control mask
522
The active and reactive power can be then controlled by controlling Id and
Iq respectively. The reference direct-axis current (Id_ref) has been calculated as
explained in Fig. 11.31. The quadrature axis current reference (Iq_ref) is
determined according to the inverter mode of operation. As shown in Fig.
11.33, three control modes are considered; unity power factor (Iq_ref = 0),
constant power factor (Iq_ref/Id_ref = constant), and constant non-zero reactive
power (Iq_ref  0).
The function of current regulator block is to make the inverter output
current rapidly tracks the reference current. This is performed by a PID
controller. The inverter output current in the dq0 reference frame can be
calculated according to the following equations,
where Vd,mes and Vq,mes are the measured voltage at PCC in the dq0
reference frame; Id and Iq are the inverter output current in the dq0 reference
frame; R is the total resistance of the filter and transformer; L is the total
reactance of the filter and transformer. All these quantities are in per-unit.
These equations will be the planet to be controlled by a PID controller where
the inputs are Vd and Vq and the outputs are Vd,conv and Vq,conv where the
controlled inputs Vd and Vq.
Fig. 11.34: The current regulator block
523
The model of the current regulator block is shown in Fig. 11.34. The
function of the inverter reference voltage generator block (Fig. 11.35) is to
generate the magnitude modulation ratio. Consequently, the required current
and voltage that achieve the demanded active and reactive power are
generated. The magnitude modulation ratio can be calculated from following
equation,
where md,q is the magnitude modulation ratio in the dq0 reference frame;
Vdc is the DC-link voltage; and Vd,q_conv is the converter voltage in the dq0
reference frame. The magnitude modulation ratio is then converted from the
dq0 to the abc reference frame and sent to the inverter.
Fig. 11.35: Reference voltage generator block
In the rest of this section, the system inputs (i.e. irradiance, temperature,
grid voltage … etc.) will be subjected to disturbance considering various
operation modes. The effect of those disturbances on the PV generator
performance will be investigated, studied and analyzed with the presented
MPPT technique.
524
In the first disturbance, the temperature of the PV generator is increased

from 10oC to 50oC at time t = 0.5 sec. With the PV generator operating at
unity power factor, the following results were obtained.
As the temperature increased, the PV array static power decreased and the
PV array output current and voltage at the MPP are decreased as shown in
Fig. 11.36. The dynamic performance (Fig. 11.37) shows that the I&IC
technique is able to find the new MPP while the IP&O technique fails. This is
due to the rate of change of duty cycle (delta-D) in IP&O technique will be
very small (almost zero) after tracking the MPP when the temperature was
stable at 100 C. Therefore, when the temperature increased to 50 0 C the duty
cycle was not able to change fast enough to track the new MPP.
Fig. 11.36: The array I-V and P-V curves for variable temperature
Fig. 11.37: Delivered active and reactive power for variable temperature
525
To solve this problem, the rate of change of duty cycle (delta-D) needs to
be set at a larger value when the system senses a change in temperature; this
can be done as shown in Fig. 11.38. The output of the derivative of the
temperature is compared to zero. If there is a change in temperature the
output of the compare-to-constant block will be 1 which will be used as an
enable signal to the delta-D reset block, otherwise the output will be 0. The
delta-D reset block will set the delta-D to a larger value when the enable
signal equal to 1. The same modification will be applied for irradiance and
voltage at the PCC.
Fig. 11.38: Reset delta-D when temperature changes
After adding the previous modification to delta-D calculator block and the
dynamic simulation of the IP&O algorithm is shown in Fig. 11.39 for the
same disturbance in the temperature. It is clear that, with the I&IC; however,
this requires the monitoring of the changes.
Fig. 11.39: Delivered active and reactive power for variable temperature
with delta-D reset to 5e-3
For the considered disturbance in the temperature, the changes in the d-q
currents and d-q reference currents as well as the duty cycle are shown in Fig.
526
11.40 and 11.41.These results show that as the temperature increased the PV
array power decreased and the PV array output current and voltage at the
MPP decreased (Fig. 11.36). This leads to changes in the reference direct axis
current with a final steady reduction in its value in comparison with its initial
value (Fig. 11.40). The maximum power point (MPP) is changed because of
the change in temperature. Consequently, the MPPT started to track the new
MPP by changing the duty cycle (Fig. 11.41). As the new MPP is achieved
by a lower voltage so the function f(D) increased.
Fig. 11.40: Measured and reference direct and quadrature axis currents for
variable temperature
Fig. 11.41: f(D) changes for variable temperature
If the mode of operation was chosen at constant reactive power or non-

unity power factor instead of unity power factor, the same scenario will
happen, but this time the quadrature axis current (Iq,ref and Iq,mes) will have a
527
certain value depending on the required reactive power or power factor.

Disturbances in other variables can be performed using the presented model
and the performance is found to be in conformation with the fundamental
theories of dynamic systems.
528
Chapter 12
Consistence of Wind Power Technologies with the Fault
Ride-through Capability Requirements
12.1 Introduction
In addition to their environmental benefits, the recent progress in wind
energy technologies leads to cost reduction to levels comparable, in many
cases, with conventional generation technologies. As a result the penetration
of wind power in power systems is increasing. This major change in the
energy mix of power systems raises a number of challenges regarding grid
stability, power quality and behavior during fault situations. Consequently,
elaboration of specific technical requirements or grid codes for the
connection of large wind farms has been constructed. The new grid codes
stipulate that wind farms should contribute to power system operation and
control in a similar way the conventional generating systems do 228,229,230.
Essential grid code requirements are related to frequency, voltage and
wind turbine behavior in case of grid faults. The most common requirements
include231 active power control, frequency control, frequency and voltage
acceptable ranges, voltage control, voltage quality, fault ride-through (FRT)
capability, wind farm modeling, and communication and external control.
The main aim of the requirements is to ensure that wind farms do not
adversely affect the power system operation and control with respect to
security of supply, reliability, and power quality.
The occurrence of a fault (short-circuit) at some point of a power network
inevitably results in voltage dips in one or more phases and possibly also a
voltage rise in the healthy phase. This depends on the type and location of the
fault. These voltage disturbances may be propagated to fairly remote
228
Tsili, M., & Papathanassiou, S. (2009). A review of grid code technical requirements
for wind farms. IET Renewable Power Generation, 3(3), 308-332.
229
onshore and offshore DFIG‐based wind farms. Wind Energy, 17(2), 279-295.
230
M. EL-Shimy, and N. Ghaly, “Grid-Connected Wind Energy Conversion Systems:
Transient Response”, In Sohail Anwar (ed.) Encyclopedia of Energy Engineering and
Technology, Taylor & Francis - CRC Press, 2014.
231
Holttinen, H., Meibom, P., Ensslin, C., Hofmann, L., Mccann, J., & Pierik, J. (2009).
Design and operation of power systems with large amounts of wind power. In VTT
Research Notes 2493.
529
locations on the network, especially in the case of weak grids. The duration
of the voltage dips is dependent on the protection system response time. It
may vary between 0.1s to several seconds; the most usual duration being in
the range of a few tenths of a second. In the event of such dips, generating
stations may encounter stability problems, depending on the type, magnitude
and duration of the dip, as well as on the type and technology of the power
station.
The large increase in the installed wind power in the transmission systems
necessitates that wind generation remains in operation in the event of network
disturbances. Consequently, current grid codes invariably demand that large
wind farms (especially those connected to the HV grids) must withstand
voltage dips down to a certain percentage of the nominal voltage (0% in some
cases) and for a specified duration. Such requirements are known as FRT or
low voltage ride through (LVRT) requirements and they are described by a
voltage against time characteristic, denoting the minimum required immunity
of the wind power station to dips of the system voltage. Based on popular
international grid codes, various FRT capability requirements will be
presented in the next section.
This chapter provides an overview of the requirements imposed by
international grid-codes for connecting large amounts of wind power to
various electric power systems. The main concern here is the fault ride-
through (FRT) capability requirements. The consistency of popular wind
power technologies with the FRT requirements is evaluated considering the
German E.ON Netz code. This is achieved through appropriate modeling and
simulation of grid-connected fixed-speed and variable-speed wind power
technologies. Two methods of evaluation are presented in the paper.
12.2 FRT Capability Requirements

This section presents the FRT capability requirements encountered in the
majority of grid codes concerning wind farm interconnection. Generally,
various FRT requirements take the form shown in Fig. 12.1. Above the FRT
requirements line, wind farms must not be disconnected to support the system
while below the line the wind farm can be disconnected; however, each grid
code may add some other constraints on the possibility of connection and
disconnection of wind farms. Four parameters can be used to define the FRT
530
requirements. As shown in Fig. 12.1, these parameters are the fault duration,
the minimum voltage during the fault (Vmin), the voltage restoration time, and
the acceptable minimum steady state voltage (Vss); however, the minimum
value of the acceptable steady state voltage is usually 90% of the nominal
voltage while the maximum value is usually 110%. Fig. 12.2 shows the
values of these parameters according to various international grid codes. It is
clear from that figure that the parameters of various are widely from country
to country. The differences are attributed to the characteristics of each power
system, and the employed protection.
Fig. 12.1: Generic FRT requirements
It is depicted from Fig. 12.2 that the requirements of the German, UK,
Nordic, Danish, Belgian, Hydro-Quebec, Swedish and New Zealand grid
codes demand that wind farms must remain connected during voltage dips
down to zero; however, it must be noted that these requirements apply to the
point of common coupling (PCC) with the network, which is generally at the
HV level. Taking into account the typical impedance values of the step-up
transformers and interconnecting lines, the corresponding voltage dip at
lower voltage levels, near the wind turbine terminals, are likely to be
somewhat around 15%. This facilitates compliance with the LVRT
requirements. As shown in Fig. 12.2, the FRT specifications may vary
according to the voltage level or the wind farm power. For example, wind
531
farms connected to the Danish grid at voltages below 100 KV are required to
withstand less severe voltage dips than the ones connected at higher voltages,
in terms of voltage dip magnitude and duration. Similar differences can be
observed in the regulation governing the connection of wind farms below and
above 100MW in the Swedish transmission system.
Fig. 12.2: Parameters of FRT requirements
Apart from the FRT curve, the Denmark and Hydro-Quebec codes define
specific kinds of faults (or sequences of faults, in the Danish code) that the
wind farm must withstand (including remote faults in the case of the Hydro-
Quebec code, cleared by slow protective devices). These more detailed
requirements could be attributed to the isolation of the Hydro-Quebec
transmission system, which has no synchronous link to neighboring
systems232. Another important difference lies in the active power restoration
rates specified by the German and British/Irish grid codes, whereas the
British code requires immediate restoration (at 90% in 0.5s after voltage
recovery), the German (E.ON Netz) code requires restoration with a rate at
232
Bernard, S., Beaulieu, D., & Trudel, G. (2005). Hydro-Quebec grid code for wind farm
interconnection. In Power Engineering Society General Meeting, 2005. IEEE (pp. 1248-
1252).
532
least equal to 20% of the nominal output power (reaching 100% in 5s after
voltage recovery). The less demanding requirement of the German code may
be attributed to the physical location of the German grid and its strong
interconnected system, as opposed to the weakly interconnected British
system, where the need for active power restoration to the pre-fault values is
more crucial for system stability.
Because of its popular use in the research and its demanding requirements
(see Fig. 12.2), the German E.ON Netz code will be considered in this paper
for the analysis of the compatibility of various Wind Energy Conversion
Technologies (WECT) with the grid code requirements. In addition, the
E.ON Betz code demanding FRT capability requirements that are very close
to the Egyptian wind farm grid connection code 233. The FRT requirements of
the E.ON Netz code are shown in Fig. 12.3.
Fig. 12.3: FRT requirements of the E.ON Netz code
This code requires a wind farm to be connected to the grid as long as the
voltage at the grid connection point is above the solid line (limit line 2)
shown in Fig. 12.3. The E.ON Netz code specifies the following:
233
Egyptian Electricity Transmission Company (EETC). Wind Farm Grid Connection
Code as a part of the Egyptian Transmission Grid Code - 1st Draft on January, 5, 2012.
EETC, Egypt, 2012
533
a) According to Fig. 12.3, wind farms must withstand voltage drops down
to zero at the connection point for durations up to 150 ms (7.5cycles)
for 50Hz systems (or 125 ms (7.5 cycles) for 60 Hz systems). In
addition, the E.ON Netz code requires a voltage restoration time of no
more than 1.5 s as shown in Fig. 12.2.
b) Three-phase short circuits or fault-related symmetrical voltage dips
must neither lead to instability above limit line 1 nor disconnection of
the wind farm. After fault clearance, the active power in-feed must
increase with a rate of 20% of the rated power per second.
c) Voltage drops within the area between limit line 1 and limit line 2
should not lead to disconnection, but in case of wind turbine
instability, short-time disconnection is allowed. The resynchronization
must take place within up to 2s and active power in-feed must increase
with a rate of 10% of the rated power per second after fault clearance.
1. Below limit line 2 disconnections of the wind turbines are allowed.
12.3 Numerical Example

12.3.1 The Study System234
Fig. 12.4 SMIB system with wind power generation
The study systems, as shown in Fig. 12.4, consist of a Wind Turbine

Generator (WTG) representing a wind farm connected to the grid through a
transmission link. In the first case, the generator is a SCIG while in the
second case the generator is a DFIG. For proper comparison and analysis,
identical parameters are used for both machines as well as the network
interconnecting the generator to the stiff grid. The study system shown in Fig.
12.4 is modeled and simulated through the classes of the PSAT with the
models presented in this book. The parameters of the connection to the grid
234
N. Ghaly, M. EL-Shimy, M. Abdelhamed (2014). Consistence of Wind Power
Technologies with the Fault Ride-through Capability Requirements. MEPCON 2014
international conference, Cairo, Egypt.
534
are shown on Fig. 12.4 while the wind turbine, generator, and control
parameters are shown in Table 12.1.
Table 12.1: Wind turbine and induction machine parameters

Value
Parameter
SCIG DFIG
Number of blades 3
Rated wind speed, u [m/s] 16
Blade length [m] 75 75
Turbine inertia constant (Hwr) [kWs/kVA] 2.5 2.5
Generator inertia constant (Hm)[kWs/kVA] 0.5 0.5
Shaft stiffness for FSWT[p.u.] 0.3 ------
Pitch control gain [p.u] 10
Pitch control time constant [s] 2
Voltage control gain Kv [p.u] ------ 10
Power control time constant Te [s] ------ 0.01
Gear box ratio [41/89]
Number of poles 4
Rated voltage [kV] 0.69
Frequency [Hz] 60
Resistance of the stator, rS [p.u.] 0.01
Resistance of the rotor, rr[p.u.] 0.01
Leakage inductance of the stator, xs [p.u.] 0.1
Leakage inductance of the rotor, xr [p.u.] 0.08
Mutual inductance, xm [p.u.] 3
12.3.2 Methodology
The main objective of this paper is to check the compatibility of the
transient response of the considered generators with the E.ON Netz grid code.
In addition, the paper provides a comparison between the responses of the
considered generators as affected by faults of various durations. To fulfill the
main objective, two methods are adopted.
In the first method, the wind farm is subjected to three-phase fault
close to its terminals and the transient response of the system determined by
simulation. The success criteria here is the success of the system to withstand
a three-phase fault for a duration of 7.5 cycles (150 ms on 50 Hz systems or
535
125 ms on 60 Hz systems) as required by the E.ON code. In addition, the

quality of the response responses is determined by calculating two indices.
The indices are the duration of the voltage restoration (i.e. the time required
to restore the post-fault voltage to 90% of the nominal voltage), and the
duration of the power restoration (time required to restore the generated
power to the pre-fault value), The calculation of the duration of the power
restoration is valid only if the system is capable of producing the same power
as the pre-fault value.
The second method is based on the determination of the Critical
Clearing Times (CCT) when the system is subjected to three-phase faults
each of which result in different voltage dips. The idea here is to construct the
CCT-voltage dip curve of the system. This CCT-voltage dip curve in
comparison with the limit line 2 shown in Fig. 12.3 gives a clear evaluation
of the compatibility of the system with the E.ON code.
12.3.3 Results and Discussions

Considering the first method, a three-phase fault is applied to the terminals
of the wind farms shown in Fig. 12.4. The faults duration is varied from 7.5
cycles to 36 cycles and the minimum voltage during the faults is
approximately zero for all values of the fault durations. These low voltages
dip at the generator terminals results in a very conservative result. This is
because the generators are subjected to the most severe fault conditions.
Table 12.2: Quality of response indices

Duration of voltage Duration of power
Fault duration
restoration (cycles) restoration (cycles)
(cycles)
SCIG DFIG SCIG DFIG
7.5 12 5.4 259.2 3.6
18 102 7.8 390 7.8
33 182.4 8.4 393 6.6
33.6 Inf. 8.4 Inf. 7.2
36 Inf. 8.7 Inf. 7.2
The wind farm terminal voltage, active power, and reactive power are
respectively shown in Fig. 12.10, Fig. 12.11, and Fig. 12.12. For comparison
536
between the responses of the considered systems, the duration of the voltage
restoration, and the duration of the power restoration indices are determined
and presented in Table 12.2.
(a)
(b)
Fig. 12.10 Terminal voltage: (a) SCIG based system; (b) DFIG based
system
The voltage transients shown in Fig. 12.10 can be explained based on the
reactive power of the machines shown Fig. 12.12 and the nature of the
available voltage controls. With the DFIG-based system, the voltage control
loop causes an initial overshoot in the terminal voltage by injecting a
controlled amount of reactive power from the machine. This is followed by a
transient period till the voltage settles to a steady state value. In the SCIG-
based system, such voltage control loop is absent. Therefore, as shown in Fig.
12.10 to 12.12 the machine consumes reactive power from the power system
537
at the instant of fault clearing to restore the voltage magnitude and generation
mode. Generally, it can be seen from Fig. 10 that the DFIG-based system is
capable of restoring the system stability faster and smoother in comparison
with the SCIG-based system. In addition, the results show that the DFIG-
based system can withstand longer fault durations in comparison with the
SCIG-based system.
(a)
(b)
Fig. 12.11 Active Power: (a) SCIG based system; (b) DFIG based system
Fig. 12.10 to 12.12 and Table 12.2 shows that both systems are capable of
fulfilling the zero voltage dip requirements of the E.ON code; however, it can
be easily depicted from Table 12.2 that the DFIG significantly surpasses the
SCIG from the points of view of voltage restoration time and duration of
power restoration. Therefore, power systems with large amounts of wind
power from DFIGs are less vulnerable to stability and security problems in
comparison with power systems with a large penetration level of SCIG. The
538
results show that the CCT of the SCIG based system is 33.6 cycles while the
DFIG is stable for a fault duration of 36 cycles or even more. For both
WECTs, the results show that the voltage restoration duration increases with
the increase in the fault duration.
(a)
(b)
Fig. 12.12 Reactive power: (a) SCIG based system; (b) DFIG based
system
In the second method, the CCTs of the system are determined for a set of
three phase faults located in the close proximity of the wind farm. The
simulation is conducted such that these faults results in voltage dips that
follow the limit line 2 of the E.ON grid code shown in Fig. 12.3. This is
controlled by controlling the fault impedance. Since the CCTs of the DFIG
based system are very long and could not be determined through the
simulation, this approach is applied only to the SCIG based system. The
results are shown in Fig. 12.13. In addition, the figure also shows the limit
line 2 of the code.
539
Fig. 12.13 Fault tolerance curve of the SCIG-based system and the FRT
E.ON code
It is depicted from Fig. 12.13 that the CCT increases with the decrease of
the voltage dip (or the increase of the remaining voltage). In addition, the
figure shows that for the same voltage dip, the CCT of the system is larger
than the E.ON code minimum withstand durations of faults. Therefore, the
results show the compatibility of the SCIG based system with the FRT
requirements of the E.ON grid code. It is should be mentioned that other code
requirements such as reactive power control, voltage control, and active
power control cannot be provided by the SCIG technology due to the
limitations of the SCIG and the available controls in the FSWT system.
Despite of these drawbacks in the SCIG, the results show that the already
operating SCIG wind farms may be able to withstand the FRT capability
required by grid codes such as the E.ON Netz code.
540
Chapter 13
Overview of the Operational Requirements of Grid
Connected Solar-PV Systems
13.1 Introduction
Currently many developing countries are facing a major power and energy
problems. These problems are attributed to many factors. These factors
include insufficient power generation, low network reliability and insufficient
maintenance, insufficient fossil fuels, and bad economy 235. Due to their
availability and competitive prices, renewable energy can contribute in the
national energy security. The whole world is expanding in using the
renewable energy, especially German, China and USA 236.
The nuclear power can be considered as a feasible alternative to renewable
energy. With nuclear power, huge amounts of energy can be generated using
small amounts of nuclear. In addition, the variability, intermittency, and
dispatchability of nuclear power sources are significantly better in
comparison with most renewable energy sources; however, there are these
two major disadvantages related to the nuclear power. These are the nuclear
waste and the safety. Although the waste is very small, it is more dangerous
and hazardous in comparison to fossil fuels; the nuclear waste needs to be
buried deep down to earth for thousands of years so the radioactivity can
diminish. It should also be kept safe from earthquakes, floods and terrorist
attack. The nuclear power is although reliable but maintaining the safety of
the plant is very expensive. In case of any accident, the nuclear power station
can result into a catastrophe. Another operational issue related to large
nuclear plants is the MW/sec rate limit capability. This rate limit is not large
enough to accommodate the variations of the system load, especially during
its peaking. Therefore, nuclear power plants may be considered as base-
loaders. On the other side, the rate limit of renewable energy sources such as
wind and solar power plants is much higher than that of the nuclear plants.
235
M. EL-Shimy, (2015). Wind Energy Conversion Systems: Reliability Prospective,”
Encyclopedia of Energy Engineering and Technology – Second Edition. Vol. 2, pp. 2184 –
2206.
236
REN21 (2014). "Renewables 2014: Global Status Report". pp. 15, 16. Available at
http://www.webcitation.org/6SKF06GAX
541
Therefore, renewable energy sources show a good fit to large load variations.
Problems such as the inherent intermittency and variability of renewable
energy sources are currently given high R&D focus.
Connecting a renewable energy plant to the utility grid is facing a lot of
operation and control challenges, also the renewable energy sources have to
comply with the grid code of the relevant country. Most grid codes mandate
that the operational capabilities of renewable energy sources be as close as
possible to conventional power plants. According to SANDIA 2012 report 237,
solar-PV installations are typically separated into three categories:
residential, non-residential, and large or utility scale. Nonresidential solar-
PV would include installations at government buildings and retail stores
ranging from tens of kilowatts (kW) to several MW, while residential
installation would be installed in the homeowner's premises, typically less
than 10 kW. These types of installations are typically on the customer’s side
of the meter and the energy produced is used predominantly on site.
Customer-side generation is under state jurisdiction, and their interconnection
is conducted pursuant to state-specific interconnection procedures.
The solar-PV plants may be connected to either low voltage or medium
voltage power grid depending on the plant size and the available output
power. The grid regulation required for connecting the solar-PV plant to the
medium voltage power grid may vary than that required for connecting the
SOLAR-PV plant to the low voltage grid. Both German grid code238 and
SANDIA 2012 report deals with the technical requirements for connecting
photovoltaic plant to the medium voltage power grid, while the IEEE 929-
2000239 deals mainly with the interconnection regulation of connecting solar-
PV plants to the low voltage power grid and the regulation modification
when the solar-PV plant is connected to the medium voltage power grid.
237
Ellis, A., Karlson, B., & Williams, J. (2012). Utility-Scale Photovoltaic Procedures and
Interconnection Requirements. Sandia National Laboratories SAND2012-2090.
238
Troester, E. (2009, March). New German grid codes for connecting PV systems to the
medium voltage power grid. In 2nd International workshop on concentrating photovoltaic
power plants: optical design, production, grid connection (pp. 9-10).
239
IEEE. (2000). IEEE Recommended Practice for Utility Interface of Photovoltaic (PV)
Systems. IEEE.
542
In this chapter240 the challenges and requirements related to the connection

of the photovoltaic generating stations to the utility grid will be investigated
considering various grid codes and international standards. The impact of the
connection voltage level is also considered in the survey. In addition,
operational and control requirements during normal operation and abnormal
conditions are presented. The SCADA systems applications in the solar-PV
power plants are also briefly presented.
13.2 Grid code and standard requirements

In this section grid code requirements for connecting solar-PV plants to
either the LV or MV power grid will be investigated. The grid code can be
divided into two main categories; normal operation and under grid
disturbance requirements. Both of these categories will be investigated in this
section.
13.2.1. Normal operation requirements
The normal operation requirements can be divided to frequency deviation,
voltage deviation, active power control,+ and reactive power control.
A.1. Frequency deviation
According to IEEE 929-2000, a small solar-PV system connected to the
LV grid side has to operate properly within a frequency range of 59.3 Hz
(98.83%) - 60.5 Hz (100.83%) based on nominal frequency of 60 Hz. This
means that the solar-PV plant has to trip when the frequency drops to 59.2 Hz
(98.66%) or increased to 60.6 Hz (101%). When the frequency lies outside
the allowable limits the inverter should cease to energize the utility lines
within 6 cycles. On the other hand, the IEC 61727 241 stated that the frequency
range is 49 Hz (98%) to 51 Hz (102%) based on a system frequency of 50
Hz, when the system frequency lies outside these limits the solar-PV system
must be disconnected within 0.2 sec (10 cycles). It is clear that the
international standards provide discrepancies in the connection
requirements. Therefore, the proper settings should be re-determined
according to the considered system operational practices and characteristics.
240
H. Khairy, M. EL-Shimy, G. Hashem. Overview of Grid Code and Operational Requirements of Grid-
connected Solar PV Power Plants. Industry Academia Collaboration (IAC) Conference, 2015, Energy and
sustainable development Track, Apr. 6 – 8, 2015, Cairo, Egypt. ‫‏‬
241
International Electrotechnical Commission, (2004). IEC 61727 - Photovoltaic (PV) systems –
Characteristics of the utility interface.
543
When the solar-PV system is connected to the MV grid side, the frequency
deviation is required to meet the requirements in Table 13.1 as stated in
Germany, France and Spain grid codes242.
Table 13.1: Frequency tolerance (in Hz) according to German, France, and
Spain grid codes for SOLAR-PV plants connected to the MV grid
Country Germany France Spain
Frequency tolerance 47.5 < f < 51.5 48 < f < 51 47.5 < f < 52
Over frequency trip time (sec.) 10 cycles n/a n/a
Under-frequency trip time(sec.) 10 cycles n/a n/a
According to SANDIA, the grid connected SOLAR-PV generator must

meet off-nominal frequency (ONF) tolerance requirements. For example,
large-scale solar-PV plants connected in the Western Electricity Coordinating
Council (WECC) footprint may need to comply with the existing WECC
ONF requirement. In addition, the proposed NERC PRC-024-1243
requirement addresses a generator frequency tolerance curve. The details of
both the WECC ONF and the NERC PRC-024-1 frequency ride through
requirements are shown in Fig. 13.1 and Table 13.2.
No trip
Zone
Fig. 13.1: NERC PRC-024 frequency ride-through curves (60 Hz system)
242
Crăciun, B. I., Kerekes, T., Séra, D., & Teodorescu, R. (2012, May). Overview of
recent grid codes for PV power integration. In Optimization of Electrical and Electronic
Equipment (OPTIM), 2012 13th International Conference on (pp. 959-965). IEEE.
243
Std, N.E.R.C. "PRC-024-1/Draft 6." Draft on Generator Frequency and Voltage
Protective Relay Settings (2013).
544
Table 13.2: WECC frequency ride-through requirements (60 Hz system)

WECC frequency ride-through requirement
Under-frequency limit Over-frequency limit Minimum time
> 59.4 60 to < 60.6 N/A (continuous)
< 59.4 > 60.6 3 min
< 58.4 > 61.6 30 sec
< 57.8 - 7.5 sec
< 57.3 - 45 cycle
< 57 > 61.7 Instantaneous
A.2. Voltage deviation

According to IEEE 929-2000, a solar-PV system connected to the LV grid
side must be able to operate healthy within the voltage window of 106 - 132
V at the PCC, that is 88% to 110% of nominal voltage which is 120 V. That
No trip
means the system will trip Zone when the voltage becomes outside these limits
with disconnection time as shown in Table 13.3. For system with line voltage
greater than 120 V, the same ratio of 88% - 110% of nominal voltage is
applied. It has to be noticed that if the system line voltage differ than 120 V,
the percentage value of nominal voltage in Table 13.3 should be followed.
According to IEC 61727 the voltage limit for LV grid connected solar-PV
plants is from 85% to 110% for nominal voltage with disconnection times as
shown in Table 13.4 when the voltage becomes outside these allowable
limits.
In countries with increased solar-PV penetration, in normal conditions the
voltage limits specified by the LV GCs should not exceed the limits
expresses in Table 13.5. The maximum allowed voltage rise caused by the
solar-PV systems should be less than 3% and is estimated in terms of short
circuit power of PCC and the apparent power of the solar-PV system.
TABLE 13.3: Response to voltage deviations for solar-PV systems connected

to the LV grid according to IEEE 929-2000
Voltage at PCC (120 V base) Maximum trip
V < 50% 6
time (Cycles)
50% < V < 88.33% 120
88.33% < V < 110% Normal operation
110% < V < 137.5% 120
V  137.5 % 2
545
TABLE 13.4: Response to voltage deviations for solar-PV systems connected

to the LV grid according to IEC61627
Voltage range Disconnection time (cycles)
V < 50% 5
50% < V < 85% 100
85% < V < 110% Normal operation
110% < V < 135% 100
V > 135% 2.5
Table 13.5: Voltage limits for high solar-PV penetration countries

Germany Spain France
80% < V < 110% 85% < V < 110% 90% < V < 110%
Fig. 13.2: Voltage flickers limit 244
Another point that attention has to be paid for is voltage flickers i.e. the
frequent variation of voltage which could lead to modulations of light
intensity of incandescent. Even more flickers can affect the operation of some
244
IEEE. (2000). IEEE Recommended Practice for Utility Interface of Photovoltaic (PV)
Systems. IEEE.
546
apparatus like computers, instrumentation and communication equipment.

Any voltage flicker resulting from the connection of the inverter to the utility
system must not exceed the permissible limits. The voltage flicker produced
by the connection of a solar-PV plant to the power grid must not exceed the
border line of irritation in Fig. 13.2 according to the IEEE 519 standard 245.
This is in order to minimize the noxious effects to other customers on the
utility system.
A.3. Active power control

Two modes of active power control are required when connecting large
solar-PV plant to a MV grid. The first one is when the plant is intended to
operate at constant output power while the second mode is when the plant is
required to participate in frequency control of the grid. The solar-PV plant
has to control the output power by reducing it in steps of 10% of the rated
power. A set point given by the utility grid operator has to be reachable at
any operation point of the plant; usually set points of 100%, 60%, 30% and
0% of the plant rated power are used.
When the system frequency is above or below the nominal value, it means
that there is too much or too less generated power in the system compared to
the load. For example, as in grid connected wind generators the system
operator prefers to have a plant capable of offering the so called spinning
reserve for security issues246, it means that the plant should be available and
ready to respond to the frequency changes by reducing or increasing their
output automatically as shown in Fig. 13.3. This requirement is also
applicable to solar-PV power plants. Fig. 13.4 shows how the plant output
power has to be controlled with respect to frequency; it is shown that the
plant output power has to be reduced with a gradient of 40%/Hz when
frequency increased than 100.4%. The plant output power is permitted to
increase again when the frequency is below 100.1%. When the frequency is
above 103% or below 95% the plant has to be disconnected from the grid.
245
IEEE 519 Working Group. (1992). IEEE recommended practices and requirements for
harmonic control in electrical power systems. IEEE STD, 519-1992.
246
M. EL-Shimy. “Renewable energy: Grid Codes and Integration”. Lecture notes
available at: http://shimymb.tripod.com/id12.html
547
Fig. 13.3: Active power and frequency control in - wind power plants
Fig. 13.4: Active power and frequency control of solar-PV power plants
A.4. Reactive power control

Solar-PV plants connected to the MV power grid has to be able to
supply reactive power to the grid at any point of operation to achieve a power
factor of between 0.95 lag and 0.95 lead to support grid voltage stability
under normal operation. The reactive power has to be supplied during the
feed-in operation, which means there is no need to supply reactive power
during night. The reactive power set point can be one of the following
operational modes:
 A fixed power factor.
548
 A variable power factor depending on the delivered active power

(Fig. 13.5).
 A fixed reactive power in MVAR.
 A variable reactive power depending on the voltage (Fig. 13.6 and
13.7).
Usually the plant has to be able to inject the required reactive power
within a few minutes and as often as required.
Fig. 13.5: Power factor – active power dependency requirements for solar-
PV plants connected to the MV grid
During a symmetrical fault or heavy load steady changes, the plant has
to support the utility voltage by injecting more reactive current to the
network, voltage control should comply to Fig. 13.6. In Germany, if the
voltage drops more than 10% of nominal voltage the control must inject
reactive current at the low voltage side of the plant transformer with a
contribution of at least of 2% of the rated current per percent of voltage drop.
The plant must be capable of injecting the required reactive current within 20
ms into the grid. If required, the plant has to be able to supply reactive current
at least 100% of the rated current. In case of an unsymmetrical fault, the
reactive current must not exceed values that cause voltages higher than 110%
of nominal voltage in the non-faulty phases.
549
Fig. 13.6: Required injected reactive current with respect to voltage drop,
according to German grid code for the MV grid
Fig. 13.7: Reactive current requirement during faults according to Spain

grid code for MV grid
In Spain, the amount of injected reactive current during voltage drop is

defined by the polygonal curve ABCD. In overvoltage conditions the reactive
current is defined by the curve D’-C’ which is the mirror of the polygonal
ABCD. Above voltage level of 1.3 p.u of nominal voltage, the solar-PV plant
550
must be disconnected by the means of protective relay 247 as shown in Fig.

13.7.
Finally, for a solar-PV plant connected to the LV power grid, the solar-
PV system should operate at a power factor higher than 0.85 (lagging or
leading) when output active power is higher than 10% of the rating. Most
solar-PV inverters designed for utility-interconnected service operates close
to unity power factor. Specially designed systems that provide reactive power
compensation may operate outside of this limit with utility approval.
A.5. Short circuit limits
The short circuit current may exceed the utility grid limit at the point of
connection of the solar-PV plant. The short circuit current of a synchronous
generator is typically eight times the rated current. For a solar-PV plant, the
short circuit current is typically the same as the rated current 248. Therefore,
there is no need to limit the solar-PV plant short circuit current by external
current limiter.
A.6. Harmonics
The harmonic component of the injected current has to be in
accordance with clause 10 of IEEE Std. 519-1992. This is to avoid the
adverse effect of harmonics on the other equipments connected to the grid.
The harmonic limits mentioned are summarized as follows:
 Total harmonic current distortion at the PCC should be less than 5%

of the fundamental frequency current at the rated inverter output.
 Each individual current harmonic should be limited to the
percentages listed in Table 13.6 when the voltage at the PCC is
ranging between 120V and 69KV. The limits in this table are a
percentage of the fundamental frequency current at full system
output. Even harmonics in these ranges should be less than 25% of
the odd harmonic limits listed.
247
Guozhen, H., Tao, C., Changsong, C., & Shanxu, D. (2009, May). Solutions for
SCADA system communication reliability in photovoltaic power plants. In Power
Electronics and Motion Control Conference, 2009. IPEMC'09. IEEE 6th International
(pp. 2482-2485). IEEE.
248
Troester, E. (2009, March). New German grid codes for connecting PV systems to the
medium voltage power grid. In 2nd International workshop on concentrating photovoltaic
power plants: optical design, production, grid connection (pp. 9-10).
551
These requirements are for six-pulse converters and general distortion

situations. IEEE standard [9] gives a conversion formula for converters with
pulse numbers greater than six, also gives different current harmonic limits
for different voltage levels at the PCC. The voltage harmonic distortions are
limited to the values in Table VII depending on the voltage level. The values
in this table are in percentage (%) of fundamental frequency voltage and for
conditions lasting for more than one hour. For shorter periods the values can
be exceeded by 50% [9].
Table 13.6: Current harmonic limits recommended by IEEE 929-1992 for

six-pulse converter
Odd harmonics Distortion limit
3rd – 9th < 4.0%
th th
11 - 15 < 2.0%
th th
17 - 21 < 1.5%
rd rd
23 - 33 < 0.6%
rd
Above the 33 < 0.3%
Table 13.7: Voltage harmonic limits recommended by IEEE 519-1992

Individual voltage Total voltage
Voltage at the PCC
distortion (%) distortion (THD %)
69 KV and below 3.0 5.0
69.001 KV up to 161 KV 1.5 2.5
More than 161.001 KV 1.0 1.5
13.2.2. Under grid disturbance requirements

The main goals are to ride through momentary network faults and at
the same time to provide grid support which is called fault ride through
capability (FRT). If a large solar-PV plant is immediately disconnected
instead of helping the system to regain a steady state operating point, the
electrical grid stability will be even more negatively affected. These
requirements apply to large solar-PV plant connected to the MV power grid.
The term fault ride through (FRT) is related to how the plant has to act
in the case of utility voltage drop because of faults to maintain grid stability,
reliability and operational security. The general form of FRT requirements is
552
depicted in Fig. 13.8249; above the solid line the plant must not be
disconnected from the utility grid while underneath the line there are no
requirements to stay connected. However, each code can add more
constraints on the connection and the disconnection of the plant. Four main
parameters can define the FRT requirements which are the minimum
acceptable voltage during the fault (Vmin), fault duration, voltage restoration
time and steady state voltage (Vss). These FRT requirements are applicable
for both wind and solar-PV plants.
Fig. 13.8: General form of FRT requirements
For a wind plant, the FRT parameters according to some different grid
codes are shown in Table 13.8. While in solar-PV grid connected systems,
Fig. 13.9 represent the FRT limits in the German E.ON Netz grid code. The
plant must not be disconnected above borderline 1, which means it must not
disconnect when voltage drops to 0% of the utility nominal voltage with a
duration not more than 150 ms (7.5 cycles based on 50Hz system). The
voltage restoration time must not exceed 1500 ms (75 cycles on 50Hz
system) with minimum acceptable steady state voltage equals to 90% of the
utility nominal voltage. Underneath border line 3 there are no requirements to
remain connected to the grid. In the area above borderline 2 and beneath
249
M. EL-Shimy, and N. Ghaly, (2015). Grid-Connected Wind Energy Conversion Systems:
Transient Response. Encyclopedia of Energy Engineering and Technology, second edition.
Vol. 2, pp. 2162 – 2183.
553
borderline 1, the following options are available according to an agreement

with the utility operator: feed-in of a short circuit current, or short-time
disconnection up to 2 seconds, or depending on the concept of grid
connection borderline 2 can be moved. Below borderline 2 a short-time
disconnection can be carried out in any cases, also a longer disconnection
time are possible.
Table 13.8: FRT requirements for wind generator according to different grid
codes
Grid Fault duration Fault duration Min Voltage
Germany
code 150
(ms) based 7.5
on 50Hz 0
voltage 1.5
restoration
Denmark
(E.On) 100 system 5(cycles) 25
level 10
(s)
Spain 500 25 20of
(% 1
Vnom)
Fig. 13.9: FRT limits according to German grid code
According to the SANDIA 2012 report, in the US the FRT requirements

for wind plants were first standardized in the Federal Energy Regulatory
Commission (FERC) Order 661A250. This requirement often applies to
transmission-connected solar-PV plants, even though the standard states that
it applies only to wind plants. FERC’s FRT requirement mandates that a
250
Order 661 - Interconnection with Wind Energy, issued by Federal Energy Regulatory
Commission (FERC) of United States, June 2, 2005
554
generator shall withstand zero voltage at the point of common coupling

(PCC) -typically the primary side of the station transformer- for up to 0.15
seconds (7.5 cycles based on frequency of 50Hz) and the ensuing voltage
recovery period. The FERC requirement is not specific about the requirement
to ride-through during the voltage recovery period.
The North American Electric Reliability Corporation (NERC) proposed
PRC-024-1 standard251 addresses voltage tolerance for all generators. If
approved, NERC’s voltage ride-through (VRT) standard will have to be
reconciled with FERC Order 661A and other LVRT regional standards that
may exist. Fig. 13.10 shows the FRT curve contained in the proposed NERC
PRC-024-1 requirement.
Fig. 13.10: FRT limit according to (NERC) PRC-024-1
13.2.3 SCADA Integration Requirements

The FERC Order 661A also contains Supervisory Control and Data
Acquisition (SCADA) requirements for wind plants. As mentioned in the
previous discussion, SCADA requirements contained in FERC Order 661A
are sometimes applied to large-scale solar-PV plants. The purpose of the
requirement was for the plant owner to be able to transmit data and receive
251
Std, N.E.R.C. "PRC-024-1/Draft 6." Draft on Generator Frequency and Voltage
Protective Relay Settings (2013).
555
instructions from the transmission provider in order to protect system

reliability. SCADA data to be shared are based on needs for:
 Real-time monitoring, operations and control (line switching,

generation dispatch, etc.),
 State estimation (to determine real-time stability),
 Remedial action schemes (planned response to contingencies),
 Remote communication, and
 Safety issues (confirming energized/de-energized components).
Local SCADA system in solar-PV plants is composed of data acquisition

unit, RTU, and a communications unit. The SCADA system could measure
and collect the solar-PV array temperature, solar irradiance, DC output
voltage and current, inverter output AC voltage and current relay switch state
and so on. Data acquisition unit consisted for current transformer (DCT and
ACT), voltage transformer (PT) and communication unit such as RS485 or
Ethernet ah shown in Fig. 13.11.
Fig. 13.11: SCADA in grid connected SOLAR-PV plant252
252
Guozhen, H., Tao, C., Changsong, C., & Shanxu, D. (2009, May). Solutions for
SCADA system communication reliability in photovoltaic power plants. In Power
556
Appendix 1
Steady state modeling of the SMIB considering the effect of the
excitation control
A1.1 The SMIB with controlled excitation

The Single Machine Infinite-Bus system (SMIB), as shown in Fig. A1.1,
consists of a generator with controlled excitation connected to a network
represented by reactance Xe to and Infinite Bus (IB) of constant frequency.
Fig. A1.1: SMIB system with controlled excitation
An excitation system and an excitation control system (Fig. A1.2) is a

combination of devices designed to generate a field current and control it by
means of Automatic Excitation Regulators (AER). Usually the devices used
for generating and controlling the field current are considered as single unit.
Hence, the main purpose of the excitation system is to feed the field winding
of the synchronous machine with direct current so that the main flux the rotor
is generated. Further, both the terminal voltage of the synchronous machine
and reactive power flow are controlled by the excitation system, which also
performs a number of protection and control tasks. The excitation control
system for a synchronous machine includes:
Electronics and Motion Control Conference, 2009. IPEMC'09. IEEE 6th International
(pp. 2482-2485). IEEE.
557
Fig. A1.2: Excitation control system for conventional synchronous machines
1. A field winding placed either in the slots of the rotor or on its poles in
the form of coils.
2. An exciter, which is a dc power, source with its auxiliary equipment. A
dc generator, an inverted ac generator or a rectifier can be used as the
exciter.
3. An automatic excitation regulator (AER) controlling the field current
of the synchronous machine within a specified field limits in
accordance with a desired law.
Currently, a large number of different types of exciter systems are used.

Three main types can be distinguished:
1. DC excitation system, where the exciter is a DC generator, often on the
same axis as the rotor of the synchronous machine (Fig. A1.3).
Fig. A1.3: DC excitation

558
2. AC excitation system, where the exciter is an AC machine with rectifier

(Fig. A1.4).
Fig. A1.4: AC brushless excitation
3. Static excitation system, where the exciting current is fed from a

controlled rectifier that gets its power either directly from the generator
terminals or from the power plant's auxiliary power system, normally
containing batteries. In the latter case, the synchronous machine can be
started against an un-energized network (black start). The batteries are
usually charged from the network (Fig. A1.5).
Fig. A1.5: Static excitation
The excitation regulators are divided into regulators with a dead zone
performs discontinuous control and regulators without a dead zone that
559
performs continues (linear) control. Large generators are equipped with linear
regulators. Theses regulators are subdivided into two basic groups:
1. Proportional-Action AER (PAER) that controlling the field current in

proportion to the deviation of an operating variable (Ex. in proportion
to the deviation of the generator terminal voltage deviation V). This
type of excitation system is a legacy one and it is currently rarely used
in modern excitation control systems.
2. Forced-Action AER (FAER) that responds not only to the deviation of
an operating variable (Ex. V) but also to the rate of change and its
derivative. FAERs are having fast response elements allowing for
shorter rise time of the field current.
According to the excitation system, three main types of characteristics of

generators can be distinguished as follows,
1. Steady-State Characteristic. This is the characteristic of uncontrolled

machine that is not subjected to rapid changes in the load. Here, the
open-circuit (or excitation) emf Eq, that is proportional to the excitation
current, is assumed constant for slow change in the operating
conditions.
2. Normal Transient Characteristic. This is the characteristic of a
machine equipped with PAER and it also applies at the initial stage of
a transient change caused by a sudden changes. Here, the quadrature
axis component (Eq’) of the transient emf (E’) that caused by the
resultant flux linkage of the field winding, remains constant at the first
instant of any sharp change in the operating conditions of the
generator.
3. Characteristics with Constant Terminal Voltage. This condition can
only be attained if the excitation is forced so strongly (FAER) such that
560
it provides complete compensation of the armature reactance. This

capability is not only a function of the control system speed but also
dependent on the rating of the excitation system.
Two types of conventional synchronous generators will be considered in

the following analysis; the non-salient pole (cylindrical pole or turbo
alternators) generators and salient pole (Hydrogenerator) generators. The
power angle characteristics of each type of generators in the following
conditions:
1. Uncontrolled machine.
2. Machine equipped with PAER (or subjected to a sudden change).
3. Machine equipped with FAER.
A1.2 Non-salient Pole Machine

The characteristics of interest can be obtained using the vector diagram
shown for non-salient pole machine in Fig. 12.
a. Uncontrolled non-salient pole machine
In this case the field current is constant and the generator is modeled
by a constant emf Eq (the excitation emf) behind the synchronous reactance
Xd as shown in Fig. A1.6. The current is defined as positive flowing out of
the machine.
Fig. A1.6: Simplified equivalent circuit of an uncontrolled non-salient pole

machine
561
Generally, the complex power is given by

S VI *  P  jQ (A1.1)
which can be written as

S  (Vd  jVq ) ( I d  jI q )*
(A1.2)
 (Vq I q  Vd I d )  j(Vq I d  Vd I q )
by comparing (A1.1) and (A1.2) we get:

P V q I q V d I d
(A1.3)
Q V q I d V d I q
Equations (A1.1) – (A1.3) are general independent on machine type and

excitation control. Moreover, these equations can be applied at different
buses of a power system. For example, consider the SMIB shown in Fig.
A1.7. Equation (A1.3) can be applied at the mentioned locations in the SEPS.
The main difference between these locations is (Vq, Vd) values as (Iq, Id) are
common to all locations. Therefore, the power received at the IB takes the
form:
Po V q I q V d I d
(A1.4)
Qo V q I d V d I q
Fig. A1.7: definition of various power values
562
The power at generator terminals is given by

Pg V gq I q V gd I d
(A1.5)
Q g V gq I d V gd I q
The electromagnetic (air gap) power is given by

Pem  E q I q
(A1.7)
Qem  E q I d
Based on the vector diagram,

V q V cos 
(A1.8)
V d V sin 
and
V sin 
Iq 
X d
(A1.9)
E q V cos 
Id 
X d
X d  X d  X e (A1.10)
Substituting (A1.8) and (A1.9) in (4 A1.) resulting in:
EqV
P sin (A1.11)
X d
EqV V2
Q cos   (A1.12)
X d X d
563
The power – angle characteristics for this case is shown in Fig. A1.8.
Fig. A1.8: Power-angle characteristics of uncontrolled non-salient pole

machine
The peak of this characteristic, Pmax defines the pull-out power or the
steady state stability limit (SSSL). The corresponding angle in this case is
equals to 90o. From (A1.11),
EqV
SSSL  Pmax  (A1.13)
X d
If the generator deliver power Po to the IB then the steady state stability
margin (SSSM) Ks is defined as:
Pmax  Po
Ks  (A1.14)
Pmax
564
Suppose that a certain minimum value of the steady state stability margin is
considered for secure steady-state operation of the system. Then the
maximum loading can be found as follows:
Pmax  Pomax
K smin  (A1.15)
Pmax
Hence,
Max. Loading  Pomax  (1  K smin )Pmax (A1.16)
Based on the vector diagram of Fig. 12 the emf as function of the power can
be found as follows:
Eq  ( I a X d ) 2  (V  I r X d ) 2 (A1.17)
where Ia and Ir are the active and reactive components of the stator current.
Also,
Po  VI cos  VI a  I a  Po / V (A1.18)
Qo  VI sin  VI r  I r  Qo / V (A1.19)
Substituting (A1.18) and (A1.19) in (A1.17) results in the desired function.
2 2
P   Q 
Eq   o X d   V  o X d  (A1.20)
V   V 
Similarly the power angle as function of the power can be found as follows:
I a X d
tan  (A1.21)
V  I r X d
565
Substituting (A1.18) and (A1.19) in (A1.21) results in the desired function.
( Po / V ) X d
tan  (A1.22)
V  (Qo / V ) X d
b. Non-salient pole machine equipped with PAER (or a machine in a

transient state)
In this case the generator is modeled by a constant emf E’q (the
transient emf) behind the transient reactance Xd as shown in Fig. A1.9.
Fig. A1.9: Simplified equivalent circuit of a non-salient pole machine with

PAER or a machine in a transient state
Based on the vector diagram of Fig. 12, the d- and q- components of the
current and the IB bus voltage are defined as:
V q V cos 
(A1.23)
V d V sin 
and
V sin 
Iq 
X d
(A1.24)
E q' V cos 
Id 
X d' 
566
Substituting (A1.23) and (A1.24) in (A1.4) resulting in:

Eq' V V2  1 1 
P sin   '   sin 2 (A1.25)
X d'  2  X d X d 
Eq' V V2 V2
Q cos   ' cos  
2
sin2  (A1.26)
X d'  X d X d
Equation (A1.25) cab be written as:
P  Asin  B sin 2 (A1.27)
Usually B << A. The power – angle characteristics for this case is shown in
Fig. 10.

machine
567
Although the PAER (or the transient condition) results in higher SSSL
that improves the steady state characteristics of the machine, the
synchronizing power is lower than that of uncontrolled machine that results
in less dynamic stability.
The SSSL and max can be found by differentiation of (A1.27) as follows:
P
0 (A1.28)
 P Pmax
Then using (A1.27) results in:
P
 Acos max  2 Bcos2δ max  0 (A1.29)
 P  Pmax
But cos 2 = 2 cos2 - 1;substitute in (A1.29)
4Bcos2 max  Acos  max  2B  0 (A1.30)

or
A 1
cos 2  max  cos  max   0 (A1.31)
4B 2
Equation (A1.30) or (A1.31) can be solved as a quadratic equation in max.

The correct solution satisfied Fig. 10 i.e. max > 90o.
1  A 
2
 A
cos  max      2 (A1.32)
2  4B  4B  
 
Then the SSSL takes the form:
568
Pmax  Asin max  B sin 2 max (A1.33)
The SSSM and maximum loading can be found by equation (A1.14) and
(A1.16) respectively.
Based on the vector diagram of Fig. 12, the transient emf can be found as
function of power as follows:
(A1.34)
(A1.35)
The transient power angle can also be found as:
(Po /V )X d' 
tan  
'
(A1.36)
V  (Qo /V )X d' 
c. Non-salient pole machine equipped with FAER

In this case the FAER forces strongly the excitation so that it provides
complete compensation of the armature reactance. Hence, the terminal
voltage of the machine Vg is considered constant. Based on the vector
diagram of Fig. 12, the d- and q- components of the IB voltage and current
are:
V q V cos 
(A1.37)
V d V sin 
569
and
V sin 
Iq 
X d
(A1.38)
V gq V cos 
Id 
Xe
Sub. (A1.37) and (A1.38) in (A1.4) resulting in:

V gqV V 2 1 1 
P sin      sin 2 (A1.39)
Xe 2  X e X d 
V gqV V2 V2
Q cos   cos  
2
sin 2  (A1.40)
Xe Xe X d
Equation (A1.39) takes the same form of (A1.27). Also, B << A and the
power – angle characteristics for this case is shown in Fig. A1.11.

machine
570
Similarly to the effect of the PAER, the FAER improves the steady
state characteristics as compared with both unregulated machines and
machines equipped with PAER; however, more reduction in the
synchronizing power occurs resulting in degrading of the dynamic
performance of the system. The SSSL and max can be found in the same
way described by equations (A1.28) – (A1.33). Based on the vector diagram
of Fig. 12, the terminal voltage of the machine and e as function of power
takes the form:
2 2
P   Q 
Vg   o X e   V  o X e  (A1.41)
V   V 
V gq V g cos(  e ) (A1.42)
(Po /V )X e
tane  (A1.43)
V  (Qo /V )X e
The effect of various types of AER on the SSSL is shown in Fig. A1.12.
Fig. A1.12: Effect of excitation system control on the characteristics of non-

salient pole machines
571
In comparison with the unregulated machine, the use of the AER

increases the SSSL and SSSM as well as the stability range of power angles.
As the speed of the AER increases, the steady-state stability is improved.
A1.3. Salient Pole Machine

The characteristics of interest can be obtained using the vector diagram
shown in Fig. 13.
a. Uncontrolled salient pole machine

In this case the field current is constant and the generator is modeled by a
constant emf Eq behind its direct and quadrature reactances xd and xq
receptively. Based on the vector diagram of Fig. 13, the d- and q-
components of the IB voltage and current are:
V q V cos 
(A1.44)
V d V sin 
and
V sin 
Iq 
X q
(A1.45)
E q V cos 
Id 
X d
where Xq = Xq + Xe.
EqV V 2  1 1 
P sin   sin 2 (A1.46)
X d 2  X q X d 
EqV V2 V2
Q cos   cos  
2
sin2  (A1.47)
X d X d X q
572
The power-angle characteristics in this case is shown in Fig. A1.13.
Fig. A1.13: Power-angle characteristics of uncontrolled salient pole machine
It is clear that in this case the power angle characteristic is not a pure
sinusoidal curve as in the case of non-salient pole machine. In addition, the
angle at the SSSL, max < 90o. Although, max < 90, the synchronizing
power of uncontrolled salient pole machine is larger than that of uncontrolled
non-salient pole machine. Hence, uncontrolled salient pole machines is
dynamically more stable than non-salient pole machines with any of the
considered excitation control conditions.
Equation (A1.46) takes the form
P  Asin  B sin 2 (A1.48)
In this case the term B sin () is known as the saliency power. Usually, B <<
A. Based on the vector diagram of Fig. 13, the excitation emf and the power
angle can be found as functions of the power Po and Qo as follows:
573
E q  EQ  I d (X d  X q ) (A1.49)
where EQ has no physical meaning and it is just a point on the vector diagram
helping in calculations.
2 2
P   Q 
EQ   o X q   V  o X q  (A1.50)
V   V 
E q V cos 
Id  (A1.51)
X d
X d Xd Xq 
Eq  EQ V   cos  (A1.52)
X q  X q 
and
( Po / V ) X q
tan  (A1.53)
V  (Qo / V ) X q
b. Salient pole machine equipped with PAER

In this case the generator is modeled by a constant emf E’q (the
transient emf) behind the transient reactance x’d.
Based on the vector diagram of Fig. 13, the d- and q- components of the
current and the IB bus voltage are defined as:
V q V cos 
(A1.54)
V d V sin 
574
and
V sin 
Iq 
X q
(A1.55)
E q' V cos 
Id 
X d' 
Sub. (A1.54), and (A1.55) in (A1.4) resulting in:
Eq' V V 2  1 1 
P sin   sin 2 (A1.56)
X d'  2  X d'  X q 
Eq' V V2 V2
Q cos   ' cos  
2
sin2  (A1.57)
X d'  X d X q
The power angle characteristics in this case is shown in Fig. A1.14
Fig. A1.14: Power-angle characteristics of salient pole machines with PAER
575
Although, the PAER results in higher SSSL, it hinders the high

synchronizing power of uncontrolled salient pole generator. Based on the
vector diagram of Fig. 13, the transient emf and angle ’ can be found as
functions of the power Po and Qo as follows:
2 2
P   Q 
E   o X d'    V  o X d'  
'
(A1.58)
V   V 
Eq'  E ' cos(   ' ) (A1.59)
( Po / V ) X d' 
tan  '
(A1.60)
V  (Qo / V ) X d' 
c. Salient pole machine equipped with FAER

In this case the terminal voltage of the machine Vg is considered
constant. Based on the vector diagram of Fig. 13, the d- and q- components
of the current and the IB bus voltage are defined as:
V q V cos 
(A1.61)
V d V sin 
and
V sin 
Iq 
X q
(A1.62)
V gq V cos 
Id 
Xe
Sub. (A1.61), and (A1.62) in (A1.4) resulting in:
576
VgqV V 2  1 1 
P sin   sin 2 (A1.63)
Xe 2  X e X q 
VgqV V2 V2
Q cos   cos  
2
sin2  (A1.64)
Xe Xe X q
The power angle characteristics in this case is shown in Fig. A1.15
Fig. A1.15: Power-angle characteristics of salient pole machines with PAER
Based on the vector diagram of Fig. 13, the terminal voltage and angle
e can be found as functions of the power Po and Qo as follows:
2 2
P   Q 
Vg   o X e   V  o X e  (A1.65)
V   V 
Vgq  Vg cos(   e ) (A1.66)
577
( Po / V ) X e
tan e  (A1.67)
V  (Qo / V ) X e
The effect of various types of AERs on the SSSL of SEPS with non-salient
pole generator is shown in Fig. A1.16.
Fig. A1.16: Effect of excitation system control on the characteristics of

salient pole machines
A1.4 Treatment of shunt reactive elements in the SMIB system

Power systems are usually containing shunt reactive elements for many
objectives including voltage control and stability enhancement. The presented
models in this appendix can be used to simulate such systems. For example,
consider the SMIB system shown in Fig. A1.17(a) in which a fixed shunt
reactive element is added at the terminals of the generator’s transformer. The
equivalent circuit of the system is shown in Fig. A1.17(b). Since the original
SMIB system (Fig. A1. 7) has a direct reactance connecting the internal node
578
of the generator (node 1) to the infinite bus node (node 2), the system of Fig.
A1.17(a) can be transformed to the original structure by the Y-
transformation. The validity condition of the transformation and the use of
the presented models is that both nodes (1 and 2) must be of a fixed voltage
magnitude during a transient process. Therefore, the additional shunt
reactances (X1o, and X2o) resulting from the transformation (Fig. A1.17(c))
will not significantly affect the transient performance of the system.
(a)
(b)
(c)
Fig. A1.17: SMIB system with fixed shunt reactive element; (a) system
topology; (b) equivalent circuit of the system; (c) reactance transformation
579
According to the definition of the infinite bus, it is a bus of constant

voltage phasor and constant frequency. According to the excitation control
type, the generator internal node (node 2) is assumed to have a fixed voltage
and the generator is represented by fixed reactance(s). Therefore, the fixed
values of the fixed voltage of node 2 and the corresponding machine
reactance(s), the values of Eg and Xg are determined. The uncontrolled salient
pole machine presents a special case as the machine has two acting
reactances Xd and Xq; however, the vector diagram of Fig. 13 shows that the
machine can be represented by the hypothetical emf EQ behind the steady-
state quadrature axis reactance (Xq) simultaneously with the equation relating
the actual internal emf of the machine (Eq) and the hypothetical internal emf
EQ. Equation (A1.49) or (A1.52) provides the required relation between these
emfs. The required Xd is determined in this case by performing the Y-
transformation with Xg = Xd. Table A1 summarizes the values of Eg and Xg
for various machine types and excitation control method.
Table A1.1: Equivalent representation of the machine in Fig. A1.17

Machine type Excitation control type Eg Xg X12
No control Eq Xd Xd
salient
PAER E’q X’d X’d

Non-
pole
FAER Vg -
No control EQ Xq Xq
Salient
PAER E’q X’d X’d 

pole
FAER Vg -
Another method can be used to treat shunt elements in the SMIB

system. The method is based on the Thevenin theorem as shown in Fig.
A1.18. In this method, the assumption of node 2 that must be of a fixed
voltage is not required. Therefore, the method provides a more general
approach for handling shunt elements; however, as shown in the figure, the
value of the infinite bus voltage phasor should be modified according to the
equivalent Thevenin’s value. The method provides an updated equivalent
value of the series external reactance connecting the generator terminals to
the infinite bus without shunt elements. Therefore, the generator model and
the generator’s terminal bus are preserved in that equivalence.
580
The updated external reactance or the Thevenin’s reactance (Xth) is

determined as the input reactance of the system at the terminal bus with the
infinite bus voltage deactivated. This method is utilized in 253 without
mentioning the necessary update of the value of the infinite bus voltage
phasor. Consequently, the resulting model is incorrect. The updated value of
the infinite bus voltage phasor is obtained as the input voltage at the terminal
bus with the generator disconnected. This shown in equations A1.68 and
A1.69. For pure reactive circuits, the Thevenin’s equivalent voltage is in-
phase with the infinite bus voltage; however, there will phase a phase
difference if resistances are not neglected.
Fig. A1.18: The use of the Thevenin theorem for treating shunt elements in
the SMIB system; (a) original system; (b) the equivalent Thevenin’s network
(A1.68)
253
Demello, F. P., & Concordia, C. (1969). Concepts of synchronous machine stability as
affected by excitation control. IEEE Transactions on power apparatus and systems, 88(4),
316-329.
581
A1.5 Numerical Examples and further analysis
Example 1: Steady-state characteristics of non-salient pole machine

A SMIB system consists of a thermal power station connected to an IB
through a transmission system is shown in Fig. A1.19. The system data, in
p.u., are: Xd = 1.5, Xd’= 0.55, XT1 = 0.1, XL = 0.4, XT2 = 0.1, V = 1, Po = 0.5,
and Qo = 0.4.
Fig. A1.19: SMIB system
It is required to determine the SSSL, SSSM (Ks), and maximum loading

assuming a minimum Ks value of 10%; for the following cases:
a. The machine is unregulated.

b. The machine is equipped with PAER.
c. The machine is equipped with FAER.
Explain the actual benefits of the AERs considering the physical limits of the
system. Consider that the generator apparent power rating is 1.7 p.u. The
thermal limit of the transmission system also equals to 1.7 p.u.
Solution
Xe = 0.1 +0.4 0.1 = 0.6 p.u.
Xd = 1.5 + 0.6 = 2.1 p.u.
X’d = 0.55 + 0.6 = 1.15 p.u.
a. Unregulated machine.
EqV
The P- relation is given by P  sin
X d
Consequently, max = 90o and Pmax = SSSL = VEq/Xd
2 2
P   Q 
Eq is calculated using Eq   o X d   V  o X d   Therefore, Eq =
V   V 
2.1185 p.u.
582
Hence, Pmax = SSSL = 1.0 p.u.

Pmax  Po
The steady state stability margin is represented by K s 
Pmax
Ks = (1.0 – 0.5)/1.0 = 0.5 p.u. = 50%
Max. Loading  Pomax  (1  K smin )Pmax
Assume that the minimum value of the SSSM = 10%, then the Max. Loading
= (1.0 – 0.1) x 1.0 = 0.9 p.u.
o = sin-1 (Po/Pmax) = 30o.
b. The machine is equipped with PAER.

The power-angle characteristics of the machine is defined by
Eq' V V2  1 1 
P sin   '   sin 2
X d'  2  X d X d 
It is then required to calculate E’q. E’, ’ and E’q can be determined using
(Po /V )X d' 
tan   '
V  (Qo /V )X d' 
The results are,

E’ = 1.569 p.u, ’ = 21.4 6o, and E’q = E’ cos( - ’) = 1.553 p.u. The P-
relation is then: P = 1.35 sin ) – 0.197 sin 
Generally, this equation takes the form P  Asin  B sin 2 where B << A
for typical values of power system parameters. The SSSL and max can be
found by the following condition,
P
0
 PPmax
Therefore,
P
 Acos max  2 Bcos2δ max  0
 P  Pmax
Since cos 2 = 2 cos2 - 1, then the following quadratic equation is obtained.

4Bcos2 max  Acos  max  2B  0
583
or
A 1
cos 2  max  cos  max   0
4B 2
The solution of this quadratic equation gives max. Two values are obtained;
however, the correct solution satisfies max > 90o. Therefore,
1  A 
2
 A 
cos  max      2

2 4B  4B  
 
Consequently, the SSSL takes the form,
Pmax  Asin max  B sin 2 max
For the given conditions and parameters, max = 104.74o, Pmax = SSSL = 1.4
p.u., Ks = 64.29%, and the max. loading = 1.26 p.u.
c. The machine is equipped with FAER

The power-angle characteristics and the required variables are defined by,
V gqV V 2 1 1 
P sin      sin 2
Xe 2  X e X d 
2 2
P   Q 
Vg   o X e   V  o X e 
V   V 
(Po /V )X e
tane 
V  (Qo /V )X e
V gq V g cos(  e )
Therefore, Vg = 1.276 p.u, e = 13.6o, Vgq = Vg cose) = 1.226 p.u, and

P = 2.04 sin ) – 0.65 sin (. Using the same previously described
procedure, the required variables can be determined. Therefore, max =
113.58o, Pmax = SSSL = 2.31 p.u, Ks = 78.35%, and the max. loading =
2.079 p.u.
The effect of various AER types on the SSSL is shown in the
following figure.
584
Fig. A1.20: Effect of excitation system control on the P- characteristics of

non-salient pole machines
In comparison with the unregulated machine, the use of the AER

increases the SSSL and SSSM as well as the stability range of power angles.
As the speed of the AER increases, the steady-state stability is improved. It
should be noted that the given characteristics describe the power transfer
capability of the system under steady state conditions; however, the physical
limits of the system are not considered in their derivation. These physical
limits include the rating of the prime mover, the generator, and the
transmission network. If the power transfer limit is set, for example, to 1.7
p.u, then the gain in the active power transfer limit obtained with the FAER
will be constrained by the physical limits as shown in Fig. A1.20. On that
figure, the dashed parts of the P- characteristics represent the unstable
steady state operating points.
It would be interesting to demonstrate the shape and operating
characteristics of some other relations. Fig. A1.21 – A1.25 show respectively
the Q-, S-, Q-P, PF-, and S-P characteristics; where PF is the power factor
and S is the apparent power. Many conclusions can be derived from these
relations; these conclusions and the practical use are left to the reader.
585
Fig. A1.21: Effect of excitation system control on the Q- characteristics of

Fig. A1.22: Effect of excitation system control on the S- characteristics of

586
Fig. A1.23: Effect of excitation system control on the Q-P characteristics of

Fig. A1.24: Effect of excitation system control on the Power factor (PF) - 
characteristics of non-salient pole machines
587
Fig. A1.25: Effect of excitation system control on the S-P characteristics of

Example 2: Characteristics of salient pole generators

A SMIB consists of a hydro-generator connected on an IB through a
transmission system shown in Fig. 1. The system data, in p.u., are: Xd = 1.0,
Xq = 0.65, Xd’= 0.18, XT1 = 0.15, XL = 0.3, XT2 = 0.15, V = 1, Po = 0.8 at 0.8
lagging PF. Determine the maximum power transmitted and corresponding
angle and the maximum loading in the following cases:
i. The machine is unregulated.

ii. The machine is equipped with PAER.
iii. The machine is equipped with FAER.
Solution
Preliminary calculations:
Xe = 0.15 + 0.3 + 0.15 = 0.6 p.u.;
Xd = 1.0 + 0.6 = 1.6 p.u.;
Xq = 0.65 + 0.6 = 1.25 p.u.;
X’d = 0.18 + 0.6 = 0.78 p.u.;
Qo = Po tan () = 0.8 tan (36.87o) = 0.6 p.u.
The excitation emf and its phase angle (i.e.the power angle) are determined
as follows,
( Po / V ) X q
tan    = 29.74o.
V  (Qo / V ) X q
588
2 2
P   Q 
EQ   o X q   V  o X q   EQ = 2.016 p.u.
V   V 
 
 
E q  E Q  I d (X d Xq) 
 
 E q V cos  
I d    Eq = 2.3373 p.u.
 X d 
 Xd Xq  
E  X d  E V   cos  
 q X q Q  X  
 q 
The transient emf and its phase angle can be determined as,
( Po / V ) X d' 
tan '
 o’ = 23.03o (< )
V  (Qo / V ) X d '
2 2
P   Q 
E   o X d'    V  o X d'    E’ = 1.60 p.u
'
V   V 
Eq'  E ' cos(   ' )  E’q = 1.589 p.u (< Eq)
The terminal voltage and e
( Po / V ) X e
tan e   e = 19.44o ((<  & < o’)
V  (Qo / V ) X e
2 2
P   Q 
Vg   o X e   V  o X e   Vg = 1.442 p.u.
V   V 
Vgq  Vg cos(   e )  Vgq = 1.419 p.u(< Eq & < E’q)
(i) Unregulated machine

EqV V 2  1 1 
P sin   sin 2  P = 1.461 sin  + 0.0875 sin 
X d 2  X q X d 
Therefore, max = 83.31o (< 90o) , Pmax = SSSL = 1.4713 p.u.

Assume that SSSMmin = 10%, the and Max. Loading = 0.9 Pmax = 1.3242 p.u.
589
(ii) The machine is equipped with PAER

Eq' V V 2  1 1 
P sin   sin 2  P = 2.037 sin  - 0.2413 sin 
X d'  2  X d'  X q 
Therefore, max = 102.4o, Pmax = SSSL = 2.10 p.u, Max. Loading = 0.9 Pmax =
1.89 p.u.
(iii) The machine is equipped with FAER

VgqV V 2  1 1 
P sin   sin 2  P = 2.365 sin  - 0.4333 sin 
Xe 2  X e X q 
Therefore, max = 107.5o., Pmax = SSSL = 2.50 p.u, Max. Loading = 0.9 Pmax =
2.25 p.u.
590
Appendix 2
Linearized transfer function of a simulink model, and root
locus based controller design
Extraction of the open-loop transfer function of an SISO system is

explained in Fig. A2.1. the shown codes may be typed on the workspace or
constructed as an m-file.
Fig. A2.1: Transfer function extraction from the simulink model of SISO
system
This process is applied to the system of Fig. 56 for the determination of

the transfer function between Efd and Vt . The results are shown in Fig.
A2.2. note that in these results the transfer function of the AER is included
with a unity gain. Therefore, the impact of the AER time constant is
considered in the determined transfer function which will be used for
determining the impact of the AER gain.
591
Fig. A2.2: Transfer function of the system in Fig. 56 between Efd and Vt
considering the time delay of the AER
Representation of the transfer function in zero-pole-gain form and the

poles of the open-loop transfer function can be obtained respectively by the
commands zpk(TF) and eig(TF). The results are,
Once the system transfer function is determined, the regulator gain can be
determined using, for example, the root locus method. The root locus is used
here for the analysis of the feedback gain (KA) loop considering the open-loop
transfer function of the system (Fig. A2.3). The impact of the value of the
592
gain on the system stability and dynamic as well as steady state performance
can be easily determined from the root locus plot.
Fig. A2.3: Block diagram of the closed loop system
593
594
Appendix 3
Modeling of individual and aggregate loads in power systems
A3.1 Introduction
In the analysis of power systems, a great deal of attention has been
given to the detailed representation of generating units and the transmission
system whereas load representation has received secondary attention. There
is a continues need for improvement of power system stability studies
requires an improvement of power system load modeling in practice and
theory. To develop a general-purpose load model, which would be flexible
enough to be utilized for all types of studies for system planning and
operation, would be difficult indeed.
When the modeling is done on a device or a component level, it needs
more detailed representation in comparison with the modeling on composite
basis as viewed from a supply point. Intended application of the model and
consideration of computational constraint must govern the determination of
the model detail.
In stability analysis, representation of effective power demand at high
voltage buses is necessary. This may include the aggregate effect of
numerous load devices that may be divided into two basic classes, static and
rotating. The static type includes heating and lighting equipment. The
rotating type of loads consists mainly of induction motors driving equipment
with variety of torque/speed characteristics.
In many cases, simplified aggregate load representation like constant
impedance model is used alongside the detailed generator models 254. This
seems related to three-research questions referred as model justification,
structure, determination, and identification respectively. These questions are:
 To what extent are accurate load models important in power system
stability analysis?
 What are appropriate mathematical structures to represent high
voltage effective load?, and
 How can the aggregate load models be determined in practice?
254
Wen-Shiow, “The Effect of Load Models on Unstable Low-Frequency Damping in
Taipower System Experience w wo Power System Stabilizers”, IEEE trans. on power
systems, Vol. 16, No. 3, Feb. 2001
595
Generally, the specific cases where modeling load dynamics are necessary
are:
 Study of industrial plant.
 An area having a large concentration of motors.
 Self-excitation and sub-synchronous resonance.
 Study of applying shunt-reactive power control devices as an aid to
stability.
In this appendix two basic tasks will be performed 255:

1. Modeling of induction motor loads for transient stability studies.
2. Presenting a developed approach for dynamic aggregation of
induction motor loads to enhance the accuracy of transient stability
studies.
A3.2 Dynamic modeling of induction motors

It is proved by T. Thiringer 256 that in transient stability analysis, a
third-order dynamic model of induction motors seems to be a good
compromise between simplicity and accuracy. The basic equations describing
the third-order dynamic model of induction motors are obtained by
neglecting the stator transients of the five order counterpart model. Therefore,
the third-order dynamic model represents the electromagnetic transients of
the rotor circuits and the rotor’s equation of motion or the swing equation.
This approximation is widely used in most techniques of analysis of power
systems including the effect of induction motors. The first-order dynamic
model of induction machines is the simplest dynamic model. In that model,
both stator and rotor electromagnetic transients are neglected. Therefore, the
machine, in this case, is represented by its steady state equivalent circuit and
the swing equation for representing the electromechanical transients.
Mathematically, the third-order dynamic model of induction machine
in the network (i.e. synchronously rotating or  = ot) reference frame (Fig.
255
Abuel-wafa AR, EL-Shimy M. Dynamic Aggregation of Induction Motor Loads for
Transient Stability Studies. Scientific Bulletin - Faculty of Engineering - Ain Shams Uni
2003 June 30;38(2):553 - 67.
256
Torbjorn Thiringer, “Comparison of Reduced-Order Dynamic Models of Induction
Machines”, IEEE, P.S., Vol. 16, No. 1, Feb. 2001
596
A3.1) is a reduced form of the fifth-order model based on the following

approximations.
a) Neglecting stator transients i.e.
pds = 0, pqs = 0 (A3.1)
b) The rotor winding terminals are short-circuited i.e.

vdr = 0, and vqr = 0 (A3.2)
Fig. A3.1: Reference frame axis 257
Considering the standard steady-state equivalent circuit parameters of

an induction machine (Fig. A3.2(a)), the following additional
parameters are defined; the induction motor transient reactance (X’)
and the rotor open circuit transient time constant (T’o). the transient
reactance is the input reactance of the first order model with
resistances neglected.
257
597
X mX r X m2
X Xs 
'
 X ss  (A3.3)
X rr X rr
X rr
T o'  (A3.4)
o rr
(a)
(b)
Fig. A3.2: Equivalent circuits of induction machines; (a) steady-state
equivalent circuit (first-order dynamic model); (b) Transient equivalent
circuit considering the rotor electromagnetic dynamics (third-order equivalent
circuit)
Substituting (A3.1) in (11) and (12) in Appendix 4, results in stator voltage

equations with stator transients neglected, as
v d V sin (A3.5)
v q V cos (A3.6)
The stator voltage equations take the form,
598
v d   rs X '  i d   E d 
'
v          (A3.7)
 q   X rs  i q   E q' 
'
It is clear that the stator voltage equations are algebraic as the stator transients
are neglected. Expressions for id and iq can be obtained from (A3.17). Also, a
simple equivalent circuit can represent these equations as shown in Fig.
A3.2(b). The rotor state equations take the form,
 pE d'   1/ To' s o   E d'   0 (X ss  X ' ) / T o'  i d 

 '   '       (A3.8)
 pE q   s o 1/ To   E q  -(X ss  X ) / T o  i q 
' ' '
0
The mechanical state equation of the rotor (i.e. the swing equation) is
obtained from the fundamental acceleration equation of rotating bodies and
takes the form
2Hps  Pm  (E d' i d  E q' i q ) (A3.9)
The term (E d' i d  E q' i q ) presents the electrical power or torque.
A3.3 Some useful models and data

The next section presents an efficient and simple method for dynamic
aggregation of power system loads. In this section some useful data about
power system loads are presented. Table A3.1 shows various models for
representing either static loads or the static representation of dynamic loads.
The static characteristics of various load components are shown in
Table A3.2 while the characteristics of aggregate loads in various load
classes are shown in Table A3.3.
The basic characteristics of various mechanical loads are shown in
Table A3.4. Typical parameters of induction motor loads considering various
applications and scales are shown in Tables A3.5, and A3.6.
599
Table A3.1: Static load models

Model Expression Field of applications
Exponential P = Po a (1 + KPF f) Traditional model used as follows:
model Q = Qo b (1 + KQF f) i) with a = b = 0 gives constant
power model
With:  = V/Vo , a  (dP/dV) at V = Vo, ii) with a = b = 1 gives constant
b  (dQ/dV) at V = Vo, f = f - fo, KPF current model
constant ranges from 0 to 3.0, KQF iii) with a = b = 2 gives constant
constant ranges from –2.0 to 0 impedance model
iv) For composite load model a = 0.5
to 1.8, b = 1.5 to 6.
ZIP or P = Po [p1 2 + p2  + p3] (1 + KPF f) For composite loads composed of
Polynomial Q = Qo [q1 2 + q2  + q3] (1 + KQF f) constant impedance, constant current
model With: p1, p2, p3, q1, q2, and q3 are & constant power portions
constants
Flexible P = Po [ PZIP + PEXP1 + PEXP2] Comprehensive static load model
model Q = Qo [ QZIP + QEXP1 + QEXP2] that offers the flexibility of
With: PZIP = p1 2 + p2  + p3, QZIP = q1 accommodating several forms of
2 + q2  + q3, PEXP1 = p4 a1 (1 + KPF1 load representations
f), QEXP1 = q4 b1 (1 + KQF1 f), PEXP2
= p5 a2 (1 + KPF2 f), QEXP2 = q5 b2 (1
+ KQF2 f), a1, a2, b1, b2, p4, p5, q4, q5,
KPF1, KPF2, KQF1, and KQF2 are constants
Firs static Pm = Pmo a Static representation of motors with
model Qm = Qmo b voltage dependency considered. For
With: a, b are constants example load flow and voltage
stability applications.
Second Pm = Pmo [p0 + p1  + p2 2 + p3 3] Comprehensive static representation
static Qm = Qmo [q0 + q1  + q2 2 + q3 3] of motors with voltage dependency
model With: considered. For example load flow
po, p1, p2, p3, qo, q1, q2, q3 are constants and voltage stability applications.
General P / (Pfrac Po) = [p1 2 + p2  + p3] Standardized static model for
static +p4 a1 (1 + KPF1 f) dynamic simulation
model + p5 a2 (1 + KPF2 f)
Q / (Qfrac Qo) = [q1 2 + q2  + q3]
+ q4 b1 (1 + KQF1 f)
+ q5 b2 (1 + KQF2 f)
With: Qo  0, p1 = 1 – (p2 + p3 + p4 +
p5), q1 = 1 – (q2 + q3 + q4 + q5), Pfrac
and Qfrac are the fractions of busload
represented by voltage and frequency
independence.
600
Table A3.2: Static characteristics of load component

Component P.F dP/dV dQ/dV dP/df dQ/df
Air conditioner
3-phase central 0.90 0.088 2.5 0.98 -1.3
1-phase central 0.96 0.202 2.3 0.90 -2.7
window type 0.82 0.468 2.5 0.56 -2.8
Water heaters,
Range top, oven, 1.0 2.0 0 0 0
Deep fryer
Dishwasher 0.99 1.8 3.6 0 -1.4
Clothes washer 0.65 0.08 1.6 3.0 1.8
Clothes dryer 0.99 2.0 3.2 0 -2.5
Refrigerator 0.8 0.77 2.5 0.53 -1.5
Television 0.8 2.0 5.1 0 -4.5
Incandescent lamp 1.0 1.55 0 0 0
Fluorescent lamp 0.9 0.96 7.4 1.0 -2.8
Industrial motors 0.88 0.07 0.5 2.5 1.2
Fan motors 0.87 0.08 1.6 2.9 1.7
Agricultural pumps 0.85 1.4 1.4 5.0 4.0
Arc furnace 0.70 2.3 1.6 -1.0 -1.0
Distribution transformers 0.64 3.4 11.5 0 -11.8
(no load)
Table A3.3: Sample characteristics of different load classes

Load class P.F dP/dV dQ/dV dP/df dQ/df
Residential
Summer 0.9 1.2 2.9 0.8 -2.2
Winter 0.99 1.5 3.2 1.0 -1.5
Commercial
Summer 0.85 0.99 3.5 1.2 -1.6
Winter 0.90 1.3 3.1 1.5 -1.1
Industrial 0.85 0.18 6.0 2.6 1.6
Power plant 0.8 0.1 1.6 2.9 1.8
auxiliaries
601
Table A3.4: Basic mechanical load representation
Mechanical load torque Representation (p.u or SI) Examples
Constant TL = TLo Conveyor systems
PL = r PLo
Proportional to the speed TL = r TLo Loads
PL = r2 PLo characterized by
friction
Proportional to the speed TL = r2 TLo Fans and
squared PL = r3 PLo centrifugal pumps
Second order polynomial TL = TLo[ A r2 + B r + C] General model
Where A, B and C are constants.

Modified second order TL = TLo[ A r2 + B r + C + D General model
polynomial ram]
Where A, B, C and D are

constants,
am = motor torque-speed
sensitivity exponent,
C = 1 – (A+B+D)
Table A3.5: Typical parameters of induction motors data (all values in p.u)
Type Rs Xso Xm Rr Xro A B D H L.F
1 0.031 0.1 3.2 0.018 0.18 1 0 0 0.7 0.6
2 0.013 0.067 3.8 0.009 0.17 1 0 0 1.5 0.8
3 0.013 0.14 2.4 0.009 0.12 1 0 0 0.8 0.7
4 0.013 0.14 2.4 0.009 0.12 1 0 0 1.5 0.7
5 0.077 0.107 2.22 0.079 0.098 1 0 0 0.74 0.46
6 0.035 0.094 2.8 0.048 0.163 1 0 0 0.93 0.6
7 0.064 0.091 2.23 0.059 0.071 0.2 0 0 0.34 0.8
Table A3.6: Description of terms in Table A3.5

Type Description
Type1 Small industrial motor
Type2 Large industrial motor
Type3 Water pump
Type4 Power plant auxiliaries.
Type5 Weight aggregate of residential motors
Type6 Weight aggregate of residential and industrial motors
Type7 Weight aggregate of motors dominated by air conditioning
Xso Unsaturated portion of stator leakage reactance
Xro Unsaturated portion of rotor leakage reactance
L.F The motor load factor
602
A3.4 Dynamic Aggregation of Composite Industrial Loads

The approach presented in this book overcomes most of the limitations
in previous techniques of aggregating industrial motor loads. The limitations
placed on the motors load level to be the rated258 or that the motors must be
connected at common bus259,260 or the motors must be identical261 are avoided
in the presented technique. The rotor electromechanical transients are
considered by aggregating the third-order dynamic model of induction
motors. The parameters of the aggregate motor are calculated to be valid
when using the steady state and transient equivalent circuits and all the
parameters of the aggregate load are calculated based on physical
assumptions without the need of further modification of the value of any of
them (slip and inertia constant) to fit the actual system response. In addition,
the effect of the interconnecting network of the load subsystem is taken into
consideration.
The load subsystem to be aggregated is considered to be radially
connected to the rest of the power system by a bus called the “interface bus”
or “response bus”. The load subsystem is assumed to take the form of an
arbitrary system with induction motors and static loads distributed within it
and interconnected by a network as shown in Fig. A3.3.
Fig. A3.3: Definition of the load subsystem
The proposed aggregate model of the load subsystem consists of four

components:
258
Graham J., others, “An Aggregate Motor Model for Industrial Plants”, PAS-103, No. 4,
April 1984
259
Farhad Nozari, M. David, “Aggregation of Induction Motors for Transient Stability
Load Modeling”, IEEE, P.S., PWRS-2, No.4, Nov. 1987
260
K.O. Abeldu, “Equivalent Load Model of Induction Machines Connected to a Common
Bus”, Ph.D. Thesis, Iowa State Uni., Ames, Iowa, 1 83
261
Aleksander M. Stankovic, Bernard C. Lesieutre, “A Probabilistic Approach to
Aggregate Induction Machine Modeling”, IEEE trans. on power systems, Vol. 11, No. 4,
Nov. 1996
603
i. A single induction motor aggregate that represents the induction

motors in the load subsystem.
ii. A single constant admittance to ground to represent the static loads in
the system.
iii. A transformer in series with the aggregate motor to connect this motor
to the high bus “interface bus” of the rest of the system.
iv. A short line model that represents the interconnections in the load
subsystem.
According to the available data, two proposed locations for static load
placement in the aggregate system model are shown in Fig. A3.4 and A3.5.
Fig. A3.4: First aggregate model
Fig. A3.5: Second aggregate model
The first aggregate model is used when the data of static load distribution
within the plant to be aggregated (load subsystem) is available. The second
aggregate model is used when the static loads are given as a percentage of the
total interface busload. From practical point of view, static loads represents
30% to 40% of industrial area load, the rest (60% to 70%) is mainly
induction motor loads. The equivalent circuits of the first and second
aggregate models are shown in Fig. A3.6 and A3.7 respectively.
604
Fig. A3.6: Equivalent circuit of the first aggregate load model
Fig. A3.7: Equivalent circuit of the first aggregate load model
A. Static Load Aggregation

Considering the first aggregate model, the static loads in the load
subsystem are aggregated as the sum of individual components Gk, Bk
representing admittance to ground of individual static load,
m m
Psi
G  G i   2
(A3.10)
i 1 i 1 |V Li |
m m
Qsi
B   Bi   2
(A3.11)
i 1 i 1 |V Li |
where m is the number of static load in the load subsystem, the voltage VLi is
the bus voltage of each static load bus and can be obtained from base-case
load flow with all motors represented by constant PQ loads.
For the second aggregate model, with the static load at the interface bus, the
static load is given as percentage of the total busload and it is represented as
constant admittance to ground at the interface bus. The components G, B
representing admittance to ground of this static load are calculated as
Ps
G (A3.12)
|V b |2
605
Qs
B (A3.13)
|V b |2
The interface bus voltage Vb can be obtained from base-case load flow with
all motors represented by constant PQ loads.
B. Aggregate of Induction Motors and Network

The static loads are now separated from the overall load subsystem. To
obtain the induction motor and network aggregate model, the parameters of
all remaining components are prepared in a common base which equals to the
aggregate motor base. The parameters of each induction motor in the load
subsystem are usually available in the individual machine base and these data
are the required input data. The transient impedance of the proposed
aggregate motor, the interconnection as well as the transformer models will
be calculated as the input impedance of the load subsystem at the interface
bus. This input impedance is calculated as follows,
 Write down the bus admittance matrix Ybus representing all the k buses
in the load subsystem and the interface bus with all parameters entered
on the aggregate motor base. This matrix will take the form
 I b  Y bb Y bk  V b 
 I   Y    (A3.14)
 k   kb Y kk  V k 
where [Ib] is the interface bus current vector = Ib,

I k   Bus current vector  I 1 I 2  I k  , and Vk   Bus voltage vector  V1 V2 Vk 
T T
The induction motors in the load subsystem are represented by

transient emf behind transient impedance each as shown in Fig. A3.2(b) (the
third order dynamic model). These individual motors transient impedances
and the motor’s transformer reactances in series with it are used to modify the
diagonal of Ybus to obtain the modified Ymod. This modified bus admittance
matrix represents the interface bus and the internal node of each motor in the
load subsystem. This modified matrix takes the form
606
 I b  Y bb Y bs  V b 
 I   Y   '  (A3.15)
 s   sb Y ss   E 
where  I s   The motors' stator current vector   Is 1 Is 2 Isn  , and

T
T
 E '   The motors' transient voltage vector   E1' E '2 E 'n 
  
In (A3.15) it is assumed that there is at least one induction motor connected

to each bus of the n buses in the load subsystem. It is clear that buses with no
induction motor loads will not be modified in the modified admittance
matrix. The input admittance of the load subsystem at the interface bus is
obtained by reducing (A3.15) by matrix form Kron elimination method by
setting [Is] = [0]. Then,
Y inp   Y bb   Y bs Y ss  Y sb  (A3.16)

1
The input impedance of the load subsystem at the interface bus is defined in
Fig. A3.8 and takes the form
1
Re  jX e   Yinp  (A3.17)
Fig. A3.8: parameters of the aggregate impedance at the load subsystem
The parameters involved in the value of the input impedance of the

aggregate load subsystem from the first image can be found using parameter
estimation (PE) techniques. The target of the PE process is to determine these
parameters for minimizing the error between the aggregate model response
and a set of response curves of the actual load subsystem. The new proposed
technique for determination of these parameters is based on using a
hypothetical bus named the aggregate motor bus in Fig. A3.8. To this
hypothetical bus the aggregate motor in the aggregate model is connected and
607
aggregating the individual motors in the load subsystem as if they are

connected to it in parallel. By this assumption the effect of transmission
network is considered in the aggregate model. The parameters of the
aggregate motor will be calculated based on weighting techniques 262 with
individual motor data prepared on individual motor base.The aggregate motor
inertia constant M and loading  are calculated from the formula of weighted
average impedance of individual motors parameters.
n
M  j M j (A3.18)
j 1
n
  j  j (A3.19)
j 1
where j is the weighting coefficient of individual motors and is defined as

the relative kVA rating of individual motors to the kVA rating of the
aggregate motor.
kV A j
j  n
(A3.20)
 kV A j
j 1
and
Prated , j
kVA j  (A3.21)
 j cos j
The rest of the aggregate induction motor parameters are calculated

using weighted average admittance method. The aggregate stator impedance
is calculated as
1 n j
 (A3.22)
R s  jX s j 1 R sj  jX sj
The aggregate magnetizing reactance is calculated as
262
William W. , others, “Load Modeling for Power Flow and Transient Stability
Computer Studies”, IEEE trans. on power systems, Vol. 3, No. 1, Feb. 1 88
608
n 
1
 j (A3.23)
X m j 1 X mj
The aggregate rotor impedance is calculated as
1 n j
 (A3.24)
R r / so  jX r j 1 R rj / s oj  jX rj
In calculating the rotor impedance, the running slip of individual motor

soj is required and it can be calculated using simplified steady state equivalent
circuit of induction motors shown in Fig. A3.9. This simplified equivalent
circuit263 is obtained by simplifying the normal steady state equivalent circuit
of induction motors by neglecting the stator resistance and combining the
stator and the rotor leakage reactances to be Xt after shifting the magnetizing
branch to the motor terminals.
Fig. A3.9: Simplified steady-state equivalent circuit
From the simplified equivalent circuit the normal rotor current Io is calculated
as
V so
Io  (A3.25)
(R r / s o )  X
2
t
2
Then the output electrical power of induction motor is equals to the motor
loading  in p.u, and takes the form
263
V. Venikov: “Transient Processes in Electrical Power Systems”, Mir Publishers–
Moscow 1985
609
Rr V so2 (R r / so )
 I o2  (A3.26)
s o (R r / so )2  X t2
Solution of (A3.26) for so have two values, the stable value of the operating
slip is the lower among the two solutions. This is because the stable part of
the steady-state torque-slip relation (Fig. A3.10) is characterized by dTe/ds >
0.Therefore, for individual motors in the load subsystem, so can be calculated
as
 2 
1 V so2 R r R r  V so2 
so      4  (A3.27)
2  X t2 X t  X t 
 
Fig. A3.10: Steady-state Torque-slip relation of an induction motor
The aggregate rotor impedance can be now calculated. In order to separate Rr

and so of the aggregate induction motor it is assumed that the critical slip scr
(slip at peak value of the induction motor torque/slip characteristics; also
called maximum slip sm) of the aggregate motor is the mean value of the
critical slips of individual motors. This assumption with the parameters
calculated above, ensures that the steady state performance and the stable
operating range of the aggregate motor to be the average of individual
motors. Therefore
610
n
s cr    j scrj (A3.28)
j 1
and
n R rj  j
Rr  X t  (A3.29)
j 1 X tj
The transient reactance of the aggregate motor can be calculated using

(A3.3). The aggregate line parameters is calculated as
Rc  jX c  Re  R s  j (X e  X T  X ' ) (A3.30)
For sake of simplicity, the aggregate transformer reactance XT is assumed to

be 9% on the aggregate motor base, which is an acceptable value 264. The
efficiency and the power factor of the aggregate motor are calculated as,
n
    j j (A3.31)
j 1
n
cos     j cos j (A3.31)
j 1
A3.5 Numerical Application and evaluation example

The presented dynamic aggregation technique is used to obtain an aggregate
model for the load subsystem shown in Fig. A3.11. The load subsystem
consists of 5 interconnected induction motors, representing the detailed
dynamic part of the load subsystem, and a static load given at the power
system bus. The load subsystem data are listed in Tables A3.7 to A3.10 while
the static load is: 6 MVA, 0.8 P.F lag. The induction motor loads represent
70% of the total load, which is an acceptable level in a normal situation in
large industrial loads.
264
P.M. Anderson & A.A. Fouad, “Power System Control and Stability”, Galgotia
publications, 1981
611
Fig. A3.11: Load subsystem

Table A3.7: Induction motors specifications
Parameter M1, M3, M4 M2, M5
Stator Voltage (V) 4000 4000
Output power (kW) 4178 1194
Efficiency (%) 96.25 94.33
Power factor (%) 91 90
Full load slip (%) 1.79 0.83
Sync speed (rad/sec) 314.29 104.76
2
Inertia (kg.m ) 101 78
Starting current (A) 4520 1366
Torque ratio (Starting/rated) 0.82 1.1
Table A3.8: Induction motors parameters on individual machine bases
M1, M3, M4 M2, M5
H 1.50521 0.43799
Rr 0.01946 0.00889
Rs 0.00973 0.03163
Xs 0.09588 0.09842
Xr 0.09588 0.09842
Xm 2.38500 2.21445
X’ 0.18805 0.19265
To’ 0.40560 0.82687
 0.8 0.7
612
Table A3.9: Lines Data on 100 MVA base

Line R X B
Lines L1-L4 0.0010 0.0055 0.0008
Line L5 0.0020 0.0110 0.0016
Table A3.10: Transformers Data

Transformer RT XT Tap
TA * 0.000 0.090 1.1
TB * 0.000 0.090 1.0
T1-T5 ** 0.000 0.090 1.0
* On 20 MVA base
** On the appropriate motor rating
The interface bus or the point of common coupling (PCC) with the rest
of the power system is defined as the bus at which the aggregate load model
will be attached. The parameters of the aggregate load as obtained using the
proposed algorithm and the results are shown in Table 3.11.
Table 3.11: Parameters of the aggregate load

Re 0.01346 X’ 0.1897
Xe 0.28161 so 0.0162
H (sec) 1.329  0.783
Rr 0.01787 sfl (%) 1.86
Rs 0.01298 cos  (%) 90.8
Xs 0.0969 Rc 0
Xr 0.0966 Xc 0.00161
Xm 2.356 G 0.24
XT 0.09 B 0.18
(All in p.u on aggregate motor base)
In order to validate the proposed technique of dynamic aggregation of

induction motor loads, transient studies are performed considering the IEEE
nine-bus system (refer to volume 1 of this text book) for the data of the
system. The system consists of 9 buses, 6 lines, 3 generator-buses, and 3
load-buses as shown in Fig. A3.12.
613
Fig. A3.12: The IEEE nine-bus system
Load B in the IEEE nine-bus system is replaced by the load subsystem

of Fig. A3.11. With this load represented in detail various types of
disturbances can be simulated and the power system responses are compared
to that obtained with the detailed load replaced by its determined aggregate
counterpart. System response to transformer switching is simulated.
The transformer switching is a classical method for presenting sudden
voltage changes in power system for simulation and on-line analysis
purposes. In this disturbance while transformer TA have tapping of 110% and
TB have tapping of 100%.When transformer TB is suddenly disconnected,
step change in the voltage of the interface bus. Consequently, the power
angle responses of generators G2 and G3 relative to G1 in the system with
actual dynamic load are compared with those responses obtained with
aggregated load. The results are shown in Fig. A3.13 and A3.14. The
correlation factors between the responses are determined and the results are
shown in Table A3.12. The speed and torque responses of the motors in the
actual system and the aggregate motor are shown in Fig. A3.15. The load
stability is obvious from detailed and aggregate responses. The results show
that both transient stability (TS) and transient voltage stability (TVS) are
preserved by the aggregate load. The comparison of response curves and
correlation factors shows an adequate match between the dynamic response
of actual system and its aggregate model.
614
Fig. A3.13: Interface bus voltage – dynamic or transient voltage stability
Fig. A3.14: Transient stability
Table A3.12: Correlation factors

Vb-act 21 31
Vb-agg 0.9996 - -
21 - 0.9828 -
31 - - 0.9926
615
(a) Speed (b) Torque

Fig. A3.15: Speed and torque responses
616
Appendix 4
Fundamentals of speed governors
Given that a prime mover is the source of mechanical power for supplying
a generator. Prime movers are mechanical power sources such as turbines and
engines. A governor is then the device which controllers the speed or output
power of a prime mover. A governor senses the speed or load of a prime
mover and consequently controls the primary input source (such as fuel,
steam, or water flow) to the prime mover. The target of the governor control
action is to maintain the speed (or load) of the prime mover at a specific
desired level.
As an example of governors is a person driving a car. The person in
this case acts as a governor to the car’s engine which is the prime mover. To
do the governing function, you have to read the speed meters and to know the
acceptable speed limit. Then you have to control the input fuel throttle to the
engine for controlling the speed of the car. Therefore, a governor should have
sensors of speed (or load) and tools/power to change the operating setting to
a desired one. The main components of a governor are then:
1. A way to set the desired speed or load,

2. A way to sense the actual speed or load,
3. A way to determine the difference between the actual and the desired
speed or load. This difference is called the error signal,
4. A way for the governor to control the primary input source (e.g. fuel or
steam),
5. A way to stabilize the prime mover after changes has been made.
Fig. A4.1 illustrates a hydro-mechanical based speed (or load)

governor for a steam turbine. The main illustrated governing concepts can be
used to explain the function of steam governors and even hydro-electric
modern digital governors.
617
Fig. A4.1: Conventional hydro-mechanical speed governor for steam

turbines. The shown operation is associated with a power increase command.
Changes in the input power to the turbine are determined by the changes
in the position of the control valve. The position of the control valve is
determined by the movement xE which control the position of point E and
consequently the position of the control valve. Downward movements of the
control valve (xE > 0) results in increase in the steam flow and consequently
increase to the turbine power input. The opposite action occurs with upward
movements in the control valve (xE < 0). In the following analysis of the
operation of the system, downward movements are considered positive while
upward movements are considered negative. The changes in the position of
the control valve are due to one of the following causes:
1. Indirect automatoc feedback actions of the prime mover due to

changes in the speed. Due to changes in the position of the connection
point B (xB), resulting from changes in speed; where
. Therefore, increase in the speed causes point B to move
upward while speed reductions cause downward movements in point
B. This is achieved by the action of the centrifugal forces of the
618
rotating flyballs as the governor rotates by the same speed of the

turbine. An increase in the speed, for example, of the turbine causes
the centrifugal forces of the flyballs to move upward. As a result, the
connection point B moves upward (xB < 0). Since the position of
points A and D are yet unchanged, the change in the position of B
causes a change in the position of point C in the opposite direction.
Therefore, point C moves downward (i.e. xC > 0). Yet, the position of
the control valve is unchanged. Therefore, a change in the position of
C causes a change in the position of D in the same direction (i.e. xD >
0). Consequently, the pilot valve of the hydraulic amplifier moves
downward and the resulting flow of pressurized oil cause the control
valve to move in the upward direction (xE < 0). Therefore, the steam
flow will decrease causing the input power to the turbine to decrease.
The speed of the turbine is the decreased and point B starts to move in
the downward direction which reduces the flow of the pressurized oil.
The operation continues till s balance (equilibrium point) is achieved.
2. Indirect automatic feedback actions due to variations of the steam
pressure.
3. Direct control action by the changes in the speed changer (or power
command) settings. This action may be manual or automatic.
Automatic actions are the reaction of the Automatic Generation
Control (AGC) where the governing system acts as the primary control
loop of the AGC. The movements shown in Fig. A4.1 are
corresponding to a raise power command in which the change position
of point A is forced by the supplementary control action of the speed
changer. The servomotor operates in a direction the results in
downward (power increase command) or upward (power reduction
command) of point A. The direction of rotation of the servomotor is
determined by the polarity of its DC input voltage and it speed of
motion is determined by the magnitude of its input voltage. This
voltage control action can be achieved locally at the generating unit or
from remote locations through communication channels.
619
In the governing system shown in Fig. A4.1, the interaction between the
governing system and the control valve is the cause of the droop. In the
isochronous governors (Fig. A4.2), these interactions are absent.
Fig. A4.2: Isochronous Governor
The droop in speed-governors is essential for stable operation.

Consider an isochronous governor; an increase in the load is associated with
a reduction in turbine speed. Therefore, the governor will respond by
increasing the steam (or any primary input source) in order to restore the
initial speed. Due to the inertial effect of the generating system, the changes
power production lags the changes in the steam flow. Therefore, the governor
actions will cause the control valve to open more than needed. Consequently,
the speed will increase above the original setting. Consequently, the governor
reacts by closing the control valve; however, the closing action will also be
more than needed due to inertia/power lag characteristics. The over-
correction of the speed in both directions is usually cause instability. This
instability problem can be eliminated with the speed-droop action. As the
load increases, the magnitude of the speed reduction will be reduced due to
the feedback between the control valve and the governor (see Fig. A4.1).
Therefore, when the governor moves to correct for the speed decrease caused
by the increased load, it will be correcting to a lower speed setting. This
lower speed setting prevents the speed from overshooting and causing
instabilities.
620
Appendix 5
Transient droop compensators for speed-droop governors of
hydroelectric units
A schematic of a hydroelectric generating unit is shown in Fig. A5.1.

The performance of a hydraulic turbine is influenced by the characteristics of
the water column feeding the turbine which include,
Fig. A5.1: Schematic of a hydroelectric power unit
1. Water inertia. The effect of water inertia is to cause changes in

turbine flow to lag behind changes in turbine gate opening.
2. Water compressibility.
3. Pipe wall elasticity in the penstock. The effect of elasticity is to
cause traveling waves of pressure and flow in the pipe. This
phenomenon is referred to as water hammer. Typically, the speed of
propagation of such waves is about 1200 meters/sec. Traveling
wave model is required only if penstock is very long.
The representation of the hydraulic turbine and water column in

stability studies usually assumes that: (a) the penstock is inelastic, (b) the
water is incompressible, and (c) hydraulic resistance is negligible. The
velocity of the water in the penstock is given by
621
where U is the water velocity, G is the gate position, H is the hydraulic

head at gate, and Ku is a constant of proportionality. The turbine mechanical
power is proportional to the product of pressure and flow. With Kp is a
proportionality constant, the mechanical power is given by,
The acceleration of water column due to a change in head at the turbine,

characterized by Newton's second law of motion, may be expressed as
where L is the length of the conduit, A is the pipe area, ρ is the mass density,
ag is the acceleration due to gravity. The term ρLA represents the mass of
water in the conduit while the term ρagH represents the incremental change
in the pressure at turbine gate. For small displacements about an initial
operating point (subscript "o") it can shown that,
where
Tw is referred to as the water starting time. It represents the time required

for a head Ho to accelerate the water in the penstock from standstill to the
velocity Uo. It should be noted that Tw varies with the load. Typically, Tw at
full load lies between 0.5 s and 4.0s and its average value is 1.0 sec. Equation
(A5.4) represents the classical transfer function of the turbine-penstock
system. It shows how the turbine power output changes in response to a
change in gate opening for an ideal lossless turbine. The TD response of the
transfer function can be obtained by the inverse Laplace transform applied to
eq. (A5.4); the TD function is,
622
This TD function is plotted as shown in Fig. A5.2 for a unit step change in
the gate position for a turbine with a time constant of 1.0 second. As shown
in the figure, an initial surge of power reduction of 2.0 p.u (i.e. opposite to
the direction of the gate direction and with double the magnitude of the gate
change) is shown at the instant of the unit step increase in the gate position.
This surge is due to the inertia of the water that prevents immediate change in
the water flow. Consequently, the sudden change in the gate position is
initially translated to a reduction in the pressure across the turbine. As a
result, the initial surge of power reduction appears. With time, the water
accelerates and the power starts to compensate the initial surge and finally the
power reaches the new steady state value which equals to + 1.0 p.u.
Fig. A5.2: Response of a hydro-turbine to a unit step increase in the gate

position
Based on the transient characteristics of hydro-turbines, the simple

speed-droop regulator shown in Fig. 80 will not be able to compensate the
inertial characteristics of water flow (Fig. A5.2). For stable control
performance, a large transient (temporary) droop with a long resetting time is
therefore required to compensate the inertial response of water. This is
623
accomplished by the provision of a rate feedback or transient gain reduction

compensation as shown in Figure A5.3.
Fig. A5.3: Transient droop compensator for hydro-turbines
624
Appendix 6
Interrelation between the dynamic and the static security
It is of critical importance to note that the dynamic security is mainly

guaranteed by the transient stability; however, the transient stability is not a
guarantee of the static security (or adequacy). The static security (see volume
1/chapter 1) concerns mainly with the capability of the system – within the
framework of selected N-k adequacy level - to provide acceptable steady-
state operating conditions while all of its components operating within their
operational and intrinsic limits.
As an illustrative example, consider the system of Fig. 5.24. If the
disconnection of one of the lines leads to transient stable conditions if A acc 
Apossdec; however, the new steady state operating point may violate the
loading capability limit of the remaining line. Therefore, for the same
disturbance (or contingency), the system may be dynamically stable but
statically insecure due to the insufficient rating of the post-disturbance
remaining circuits. In cases where the dynamic security (i.e. the transient
stability) is not satisfied, there are many methods for enhancing the transient
stability of power systems as will be presented in this book.
The static security is out of the scope of this book. Therefore, it is
recommended to read a textbook like the Allan J. Wood book 265; however,
there are many corrective actions that can be taken for handling static
insecurity problems. Fig. A6.1 and A6.2 summarizes some of these corrective
actions.
The corrective actions can be classified according to the implementation
cost and according to the implementation time. From cost point of view,
alternative corrective actions may be cost-free, with low-cost, and with high-
cost. From the implementation time point of view, corrective actions may be
classified to immediate, actions with small time-delay in the order of minutes
of hours, and actions with large time delay.
265
Wood, A. J., & Wollenberg, B. F. (2012). Power generation, operation, and control.
John Wiley & Sons
625
Fig. A6.1: Corrective actions for power flow overloads
Fig. A6.2: Corrective actions for voltage magnitude problems
626
Cost-free options mainly dependent on adjusting the operation of the

existing infrastructure while it has no impact of the optimal economic
dispatch (OED) or the security constrained optimal power flow (SCOPF) or
power transactions. Activation of the automatic AVR function of ULTC
transformers is an example of cost-free corrective actions. These are also
examples of immediate actions. Non-cost-free options usually include
changes in the system economic operation or addition of new infrastructure.
For example, generation shift and adjusting power transactions have a
negative impact on the system economy. Planning-based corrective actions
such as installing new line are usually associated with large investment and
time of implementation.
(a)
(b)
Fig. A6.3: OED-based static security analysis; (a) Pre-fault; (b) Steady-
state post-fault operating conditions
For a system that is dynamically secure, the static security may be

adjusted through proper and defensive operation of power systems. For
example, suppose that the system of Fig. 5.24 is subjected to a three-phase
fault at one of the transmission lines where the transient stability analysis
reveals that the system is stable. Now suppose that the system is supplying a
load of 1200 MW, the maximum line loading is 400 MW, and the generator
maximum active power is 800MW. Therefore, the parallel lines are not fully
redundant. Based on the OED, the power production allocated to the local
generator is 500 MW while the rest of the load is supplied from the rest of the
system as shown in Fig. A6.3. Now of one of the line is subjected to a fault
627
and consequently disconnected (i.e. forced outage) or disconnected for

maintenance purpose (i.e. selective outage), the remaining line becomes
overloaded as shown in Fig. A6.3. Therefore, the disconnection of the faulty
line causes disconnection of the other line due to overloading. In larger
systems, such situations can cause cascaded outages which may lead to
system blackout.
The SCOPF, on the other hand, allocates the generators for the system to
be statically secure and defensive. For example, the system of Fig. A6.3 is to
be operated based on the SCOPF under the N-1 security framework. This
shown in Fig. A6.4. In this case, the generator allocations is not only based
on cost minimization but also considering the N-1 requirements. Therefore,
the generator power production becomes 400 MW for overcoming the outage
of one of the lines. Consequently, the system becomes statically secure.
(a)
(b)
Fig. A6.4: SCOPF-based static security analysis; (a) Pre-fault; (b) Steady-
state post-fault operating conditions
The shown corrective action affects the economy due to the shift in the
power generation. Therefore, it is a non-cost free action; however, proper
action may include increasing the transfer capacity by adding new line(s)
such the local generator can operate at full load without violating the static
security requirements. In this case, the time frame of implementing the
corrective action is in the order of months or years and the associated costs
may to very large.
628
Appendix 7
Numerical integration methods
In this appendix the characteristics of various numerical integration

methods will be presented and their potentials and drawbacks will be
summarized.
As an initial step the differential equations to be solved are represented
as a set of first-order differential equations or in the state-space form. For
example, with damping power neglected, the swing equation of the SMIB
system can be represented as,
o
  Pm  Pe ( (t ))  (a )
(A 6.1)
2H
  (b )
For various conditions of a fault (see chapter 5), Pe(δ(t)) is given by the
power flow equations:
where i represented either the pre-fault (I), during-fault (II), or post-fault (III)
transfer reactance between the generator’s internal emf and the infinite bus.
Note that in multi-machine power systems (see volume 1 of this textbook),
each generator has a swing equation similar to that presented by (A6.1) while
the electrical power takes the form,
Pei  E i2G ii   E i E jY ij cos(ij   i   j ) (A 6.3)

j 1
j i
where Ei, Ej are the internal bus voltage magnitudes that are having angles of
δi and δj, respectively. Yij=Gij+jBij=Yij/_θij is the element in the ith row, jth
column of the appropriate Y-bus matrix. Depending on the particular time
interval we are integrating, the “appropriate Y-bus” will be YI, YII, or YIII,
corresponding to pre-fault, during-fault, and post-fault conditions,
629
respectively. This equation is also applicable for the SMIB system if the
losses are not neglected.
The solution of (A6.1) can be found by integrating both sides of each
equation. This action results in the form,
t t
o
 (t )    ( )d     Pm  Pe ( ( )) d  (a )
0 0
2H
t t
(A 6.4)
 (t )    ( )d     ( )d  (b )
0 0
Note that there is a cross-coupling between the parts of A6.4 i.e. δ appears in
the integrand of (a), and that ω appears in the integrand of (b). Since, the
required solutions are (t) and (t), then a general closed form solutions
cannot be obtained. Therefore, numerical integration is a suitable approach
for solving such a problem. Let x1=ω and x2=δ, then (A6.4) can be written as,
x 1  f 1 (x 1, x 2 ) 
  x  f (x ) (A 6.5)
x 2  f 2 (x 1 , x 2 ) 
It is important to note that, a power system with any scale and any
configuration can be represented by the form of equation (A6.5). in solving
(A6.5), the boundary conditions are the initial values (0) = oI and (0) =
o. Therefore, the problem is an initial value problem.
The fundamentals of the numerical integration for initial value problems
can be demonstrated considering a single-state-variable; however, the
treatment of many state-variable will be the same. For a single-state-variable,
the problem is represented by,
x  f (x (t )), x (0)  x 0 (A 6.6)
Hence, it is required to solve,
630
t t
x (t )   x ( )d    f (x ( ))d  (A 6.7)
0 0
Numerical integration methods utilize discrete time formulations rather than

continuous time models. Therefore, (7) is written as,
kT
x (kT )   f (x (k  ))d 
0
kT T kT
  f (x (k  ))d    f (x (k  ))d 
0 kT T
x ( kT T )
kT
 x (kT  T )   f (x (k  ))d  (A 6.8)
kT T
In (A6.8), k is an integer number. Hence, the approach is comprehended as

follows: “Given the value of x at t = kT-T i.e. the last known value of x, to
determine value of x at the next time sample t = kT, the integral of (A6.8)
must be determined”. This integral defines the change in x from the last step
to the next in an iterative manner, i.e.,
kT
x   f (x (k  ))d  (A 6.9)
kT T
Therefore, solving the considered problem requires only the ability to

compute the integral in (A6.9). The value of the discrete time step (or
sampling time) is T. Generally, the accuracy of the numerical integration is
better with the reduction in the sampling time. There are many numerical
methods for that purpose.
There are two main classes of numerical integration methods. The first
class is the explicit method such as the forward Euler, predictor-corrector,
and Runge-Kutta methods. They are so-called because the evaluate x(kT)
explicitly as a function of values at previous steps and their derivatives. All
631
explicit methods are numerically unstable, i.e., they have a bounded stability
domain. Explicit methods are therefore must utilize small step-sizes to work
correctly. The second class of integration methods that does not have this
problem is called implicit methods. The implicit methods include the
trapezoidal rule and the backward rule. Implicit methods require a value of
the function x(kT) at a future time step. Implicit methods are good for “stiff”
problems, where. Stiff problems are often characterized by large differences
between the real parts of system eigenvalues. Details and implementation
examples of these classes and methods are available at Hairer 266.
266
Hairer, E., & Wanner, G. (1996). Solving ordinary differential equations ii: Stiff and
differential-algebraic problems second revised edition with 137 figures. Springer series in
computational mathematics, 14.
632
Appendix 8
Data of the detailed two-area system
The system is shown in Fig. A8.1. The system consists of two similar
areas connected by a weak tie. Each area consists of two synchronous
generator units. The rating of each synchronous generator is 900 MVA and
20 kV. Each of the units is connected through transformers to the 230 kV
transmission line. The generator parameters in per unit on the rated MVA and
KV base are shown in Table A8.1. Each step up transformer has as
impedance of per unit on 900 MVA and 20/230 KV base. The
transmission system nominal voltage is 230 KV. The line lengths are
identified in Fig. A8.1. The parameters of the lines in per unit on 100 MVA,
230 KV base are r = 0.0001 pu/km, xL = 0.001 pu/km, and bc = 0.00175
pu/km. The operating loadings on various generators are listed in Table A8.2.
The loads and reactive power supplied by the shunt capacitors at buses
7 and 9 are shown in Table A8.3. The AVR data are shown in Table A8.4
while the PSS data are shown in Table A8.5. The parameters of the SVCs are
listed in Tables A8.6 and A8.7.
Fig. A8.1: Two-area power system with wind power integration
633
Table A8.1: Generator data
Table A8.2: Loading of generators

Machine (MW) (MVAr) (pu)
G1 700 185
G2 700 235
G3 719 176
G4 700 202
Table A8.3: Loads and reactive power compensations

ZIP Model
Bus (MW) (MVAr) (MVAr)
P Q
7 967 100 200 Constant Constant
9 1767 100 350 current impedance
Table A8.2: AVR data
20 0.055 0.36 0.125 1.8 0.0056 1.075 0.05
Table A8.2: PSS data
20 10 0.05 0.02 3 5.4
Table A8.6: Parameters of SVC at Bus 7

Regulator time constant Tr (sec) 0.1
Regulator gain Kr (pu) 75
Reference voltage (p.u.) 1
Bmax (pu) 5
Bmin (pu) -5
634
Table A8.7: Parameters of SVC at Bus 9:

Regulator time constant Tr (sec) 0.1
Regulator gain Kr (pu) 75
Reference voltage (p.u.) 1
αmax (rad) 3.14
αmin (rad) -3.14
Integral deviation Kd (pu) 0.001
Transient time constant T1 (sec) 0
Measurement gain Km (pu) 1
Time delay Tm (sec) 0.01
Inductive reactance Xl (pu) 0.2
capacitive reactance Xc (pu) 0.1
The parameters of the FS-SCIG and the DFIG integrated with the system
(chapter 9) are listed in the following tables.
Table A8.8: Parameters of SCIGs

Power rating Sn (MVA) 60
Voltage rating Vn (kV) 0.48
Frequency rating fn (Hz) 60
Stator resistance Rs (p.u.) 0.01
Stator reactance Xs (p.u.) 0.1
Rotor resistance Rr (p.u.) 0.01
Rotor reactance Xr (p.u.) 0.08
Magnetization reactance Xm (p.u.) 3
Initial constant Hm (kWs/kVA) 3
Number of poles p 4
Number of Blades nb 3
Gear box ratio ηGB 1/89
Table A8.9: Parameters of DFIGs

Power rating Sn (MVA) 900
Voltage rating Vn (kV) 20
Frequency rating fn (Hz) 60
Stator resistance Rs (p.u.) 0.01
Stator reactance Xs (p.u.) 0.1
Rotor resistance Rr (p.u.) 0.01
Rotor reactance Xr (p.u.) 0.08
Magnetization reactance Xm (p.u.) 3
Initial constant Hm (kWs/kVA) 3
Voltage Control Gain Kv (p.u) 10
Pitch Control Gain (p.u) 10
Time constants (s) 3
Power Control Time Constant (s) 0.01
Number of poles p 4
Number of Blades nb 3
Gear box ratio ηGB 1/89
635
Table A8.10: Wind speed model

Wind model type Mexican_hat
Average wind speed (m/s) 16
Air density ρ(kg m3) 1.225
Filter time constant τ(s) 4
Sample time for wind measurements t 0.1
Scale factor for Weibull distribution c 20
Shape factor for Weibull distribution k 2
636
Appendix 9
Nomenclature of the symbols used in Chapter 10 and 11
Cf Correction factor to account for the effect of off-optimal tilt angle of the PV-
generator on the temperature of the PV-modules.
EH Monthly mean daily hydraulic energy demand (kWh/day)
Es Monthly mean daily radiation on a horizontal plane (terrestrial radiation) in
(kWh/m2/day)
EA Actual monthly mean produced energy available for water pumping
(kWh/day)
Load matching factor to PV generator characteristics
g Acceleration due to gravity (m/s2) = 9.81 m/s2
H Monthly mean hourly global radiation on a horizontal surface (kW/m2)
Hb Beam or direct component of H
Hd Diffuse component of H
Ht Hourly irradiance in the plane of the PV-generator (kW/m2)
Total daily radiation in the plane of the PV-generator (kWh/m2/day)
Monthly mean daily global radiation on a horizontal surface (kW/m2/day)
Diffuse component of
HTE Total pumping head (m)
i The month of a year (i = 1 for January)
Monthly average clearance index (dimensionless)
m Number of PV modules
n The day of a year (n = 32 for February 1)
NOCT Nominal Operating Cell Temperature (oC)
Nominal capacity or peak Watt of a PV- generator (Wp)
Optimal capacity of a PV-generator (Wp)
PModule Rated power of a PV module (Wp)
Qdc Total daily demanded water flow required for irrigating the crops (plant
water requirements) in (m3/day)
Rb Tilt factor for beam radiation
Rd Tilt factor for diffuse radiation
rd Ratio of total hourly to total daily diffuse radiation
rt Ratio of hourly total to daily total values
of monthly mean global radiation on a
horizontal surface (dimensionless)
S Overall solar collector area of the PV-generator (m2)
t Local time on the 24-hour basis.
to Solar noon (hr)
Tc Temperature of the PV module (oC)
Tr Reference temperature of the PV module ( oC) = 25oC
Ta Mean monthly ambient temperature (oC)
 Diffuse reflectance (or ground albedo) of the ground
 Actual tilt angle (deg)
637
* Optimal tilt angle (deg)

p Temperature coefficient of the PV module ( oC-1)
 Declination angle (deg)
Motor–pump unit efficiency
Inverter efficiency
Miscellaneous PV-array losses (wiring losses)
Irrigation efficiency (shows to what extent the water enters an irrigation
system is exploited)
Irrigation system efficiency (account for energy losses due to friction of
water in the irrigation system)
Efficiency factor of the PV pumping system
Overall efficiency of the PV generator
PV module efficiency at reference temperature.
 Density of water (kg/m3 ) = 1000 kg/m3
 Latitude (deg)
 Solar hour angle (deg)
s Sunset hour angle (deg)
Sandia Model
Short Circuit Current (A)
Current at the maximum-power point (A)
Current at module voltage V = 0.5⋅ , defines 4th point on I-V curve for
modeling curve shape (A)
Current at module voltage V = 0.5⋅( + ), defines 5th point on I-V
curve for modeling curve shape (A)
Open-circuit voltage (V)
Voltage at maximum-power point (V)
Power at maximum-power point (W)
Fill Factor
Number of cell-strings in parallel in module
Boltzmann’s constant, 1.38066E-23 (J/K)
Elementary charge, 1.60218*10-19 (coulomb)
Cell temperature inside module (°C)
Reference cell temperature, typically 25°C
Reference solar irradiance, typically 1000 W/m2
‘Thermal voltage’ per cell at temperature .
The ‘effective’ solar irradiance. This value describes the fraction of the
total solar irradiance incident on the module to which the cells inside
actually respond.
, Empirically determined coefficients relating to effective irradiance,
. + = 1, (dimensionless)
. is dimensionless and has units of 1/V.
, Empirically determined coefficients relating to effective irradiance, .
+ = 1, (dimensionless)
. + = 1, (dimensionless)
638
Empirically determined ‘diode factor’ associated with individual cells in
the module, with a value typically near unity, (dimensionless).
Fraction of diffuse irradiance used by module
Empirically determined polynomial relating the solar spectral influence
on to air mass variation over the day, where:
Empirically determined polynomial relating optical influences on to

solar angle-of-incidence (AOI), where:
at ( 1000 W/m2, , o
C, = 0) (A)
at ( , ) (A)
at ( , ) (V)
at ( , ) (V)
at ( , ) (A)
at ( , ) (A)
Normalized temperature coefficient for , (1/°C)
Normalized temperature coefficient for , (1/°C)
2
Temperature coefficient for module at a 1000 W/m irradiance level,
(V/°C)
Coefficient providing the irradiance dependence for the temperature
coefficient, typically assumed to be zero, (V/°C)
Temperature coefficient for module maximum-power-voltage as a
function of effective irradiance, . Usually, the irradiance dependence
can be neglected and is assumed to be a constant value
2
Temperature coefficient for module at a 1000 W/m irradiance level,
(V/°C)
Coefficient providing the irradiance dependence for the temperature
coefficient, typically assumed to be zero, (V/°C)
CEC Model
Ideality factor (eV)
Absolute air mass
Ideality factor at SRC (eV)
Total irradiance on horizontal surface (W/m2)
Beam component of total irradiance on horizontal surface (W/m 2)
Diffuse component of total irradiance on horizontal surface (W/m2)
Effective irradiance incident on the tilted surface (W/m2)
Effective irradiance at SRC (W/m2)
Predicted current (A)
Light Current (A)
Light Current at SRC (A)
Current at maximum power point (A)
Current at maximum power point at SRC (A)
Diode reverse saturation current (A)
Diode reverse saturation current at SRC (A)
Incidence angle modifier at incidence angle θ
Incidence angle modifier at θ =58o
Air mass modifier
639
Air mass modifier at SRC (=1)
Number of cells in series
Predicted power (W)
Ratio of beam radiation on tilted surface to that on a horizontal surface
Series resistance (ohm)
Series resistance at SRC (ohm)
Shunt resistance (ohm)
Shunt resistance at SRC (ohm)
Cell temperature (oC)
Cell temperature at SRC(oC)
Assigned voltage (V)
Voltage at maximum power point (V)
Voltage at maximum power point at SRC (V)
Temperature coefficient for short circuit current (A/ oC)
Slope of the PV panel (Deg.)
Open circuit voltage temperature coefficient (V/ oC)
Material band gap energy (eV)
Incidence angle, angle between the beam of light and the normal to the
panel surface (Degree)
Zenith angle, angle between the vertical and the line to the sun (beam)
(Degree)
Ground reflectance factor
Other quantities
Life of the project (years)
Year number; t = 0, 1, 2…, T
Net annual cost of the project for year t ($)
The energy produced in year t (kWh)
Initial investment and cost of the system for year t ($)
Maintenance cost for year t ($)
Operational cost for year t ($)
Interest expenditure for year t ($)
Discount rate (%)
The initial rated energy output (kWh/year)
Degradation rate (%)
Depreciation rate (%)
Tax rate (%)
Residual value (%)
The net present value of the cost factors till the loan payment ending period ($)
The net present value of the cost factors after loan payment ending period to
the end of the effective lifetime of the project ($).
The total energy production till the end of the effective lifetime period (kWh).
Effective lifetime of the PV technology (years).
Financial lifetime of the PV project (years).
No. of inverter replacements.
Inverter capacity (W)
Inverter cost ($/W)
Per unit energy produced in year t (p.u)
640
List of references
[1] M. EL-Shimy. Dynamic Security of Interconnected Power Systems - Volume 1. Lap Lambert Academic Publishing /
Omniscriptum Gmbh & Company Kg; Germany; 978-3-659-71372-9; 117 pages; May, 2015. DOI:
10.13140/RG.2.2.19425.71520
[2] Kundur, P. (1994). Power system stability and control (Vol. 7). N. J. Balu, & M. G. Lauby (Eds.). New York: McGraw-hill.
[3] Venikov, V. A. (1977). Transient processes in electrical power systems. Mir Publishers.
[4] Machowski, J., Bialek, J., & Bumby, J. R. (1997). Power system dynamics and stability. John Wiley & Sons.
[5] Savulescu, S. C. (Ed.). (2014). Real-time stability in power systems: techniques for early detection of the risk of blackout.
Springer.
[6] Kundur, P., Paserba, J., Ajjarapu, V., Andersson, G., Bose, A., Canizares, C., & Vittal, V. (2004). Definition and classification of
power system stability IEEE/CIGRE joint task force on stability terms and definitions. Power Systems, IEEE Transactions on,
19(3), 1387-1401.
[7] Ravalli, P. (1986). Frequency Dynamics and Frequency Control of a Power System, Particularly Under Emergency Conditions
(Doctoral dissertation, University of Melbourne).
[8] Andersson, G. (2012). Dynamics and control of electric power systems. Lecture notes 227-0528-00, ITET ETH, EEH - Power
Systems Laboratory, ETH Z¨urich
[9] Eremia, M., & Shahidehpour, M. (Eds.). (2013). Handbook of electrical power system dynamics: modeling, stability, and control
(Vol. 92). John Wiley & Sons.
[10] Wall, P., Gonzalez-Longatt, F., & Terzija, V. (2012, July). Estimation of generator inertia available during a disturbance. In
Power and Energy Society General Meeting, 2012 IEEE (pp. 1-8). IEEE.
[11] Elgard, O. I. (1982). Electric energy systems theory. New YorkMc Graw-Hill.
[12] Elgerd, O. (1981). Control of electric power systems. IEEE Control Systems Magazine.
[13] Eremia, M., & Shahidehpour, M. (Eds.). (2013). Handbook of electrical power system dynamics: modeling, stability, and control
(Vol. 92). John Wiley & Sons.
[14] Machowski, J., Bialek, J., & Bumby, J. R. (1997). Power system dynamics and stability. John Wiley & Sons.
[15] Krause, P. C., Wasynczuk, O., Sudhoff, S. D., & Pekarek, S. (2013). Analysis of electric machinery and drive systems (Vol. 75).
John Wiley & Sons.
[16] M. Mandour, M. EL-Shimy, F. Bendary, and W. Ibrahim, “Damping of Power Systems Oscillations using FACTS Power
Oscillation Damper – Design and Performance Analysis”, 16th International Middle East Power Systems Conference
(MEPCON’14), Dec. 23-25, 2014, Cairo, Egypt, pp. 1- 8.
[17] M. Mandour, M. EL-Shimy, F. Bendary, W.M. Mansour. Impact of Wind Power on Power System Stability and Oscillation
Damping Controller Design. Industry Academia Collaboration (IAC) Conference, 2015, Energy and sustainable development
Track, Apr. 6 – 8, 2015, Cairo, Egypt
[18] Demello, F. P., & Concordia, C. (1969). Concepts of synchronous machine stability as affected by excitation control. IEEE
Transactions on power apparatus and systems, 88(4), 316-329.
[19] Rogers, G. (2012). Power system oscillations. Springer Science & Business Media.
[20] Padiyar KR. Power System Dynamics: Stability and Control: Anshan; 2004.
[21] http://mkpalconsulting.com/
[22] http://tutorial.math.lamar.edu/Classes/DE/InverseTransforms.aspx
[23] http://tutorial.math.lamar.edu/Classes/DE/Laplace_Table.aspx
[24] WECC Control Work Group. WECC Tutorial on Speed Governors. Une 20002. WECC.
[25] El-Shimy, M. (2015). Stability-based minimization of load shedding in weakly interconnected systems for real-time applications.
International Journal of Electrical Power & Energy Systems, 70, 99-107.
[26] Tsai, S. J., Zhang, L., Phadke, A. G., Liu, Y., Ingram, M. R., Bell, S. C., ... & Tang, L. (2007). Frequency sensitivity and
electromechanical propagation simulation study in large power systems. Circuits and Systems I: Regular Papers, IEEE
Transactions on, 54(8), 1819-1828.
[27] Wall, P., González-Longatt, F., & Terzija, V. (2010, August). Demonstration of an inertia constant estimation method through
simulation. In Universities Power Engineering Conference (UPEC), 2010 45th International (pp. 1-6). IEEE.
[28] Chassin, D. P., Huang, Z., Donnelly, M. K., Hassler, C., Ramirez, E., & Ray, C. (2005). Estimation of WECC system inertia
using observed frequency transients. Power Systems, IEEE Transactions on, 20(2), 1190-1192.
[29] Zhang, Y., Bank, J., Wan, Y. H., Muljadi, E., & Corbus, D. (2013). Synchrophasor Measurement-Based Wind Plant Inertia
Estimation: Preprint (No. NREL/CP-5500-57471). National Renewable Energy Laboratory (NREL), Golden, CO.
[30] Chow, J. H. (2013). Power system coherency and model reduction. London: Springer.
[31] Chow, J. H., Chakrabortty, A., Vanfretti, L., & Arcak, M. (2008). Estimation of radial power system transfer path dynamic
parameters using synchronized phasor data. Power Systems, IEEE Transactions on, 23(2), 564-571.
[32] Kundur, P., Paserba, J., Ajjarapu, V., Andersson, G., Bose, A., Canizares, C., ... & Vittal, V. (2004). Definition and classification
of power system stability IEEE/CIGRE joint task force on stability terms and definitions. Power Systems, IEEE Transactions on,
19(3), 1387-1401.
[33] Kimbark, E. W. (1995). Power system stability (Vol. 1). John Wiley & Sons.
[34] Milano, F., Vanfretti, L., & Morataya, J. C. (2008). An open source power system virtual laboratory: The PSAT case and
experience. Education, IEEE Transactions on, 51(1), 17-23.
[35] Grainger, J. J., & Stevenson, W. D. (1994). Power system analysis (Vol. 31). New York: McGraw-Hill.
[36] Nagrath, I. J. (2003). Modern power system analysis. Tata McGraw-Hill.
[37] Willems, J. (1969). Generalisation of the equal-area criterion for synchronous machines. Electrical Engineers, Proceedings of the
Institution of, 116(8), 1431-1432.
[38] Willems, J. L. (1971). Direct method for transient stability studies in power system analysis. Automatic Control, IEEE
Transactions on, 16(4), 332-341.
[39] Edwards, L., Gregory, J. D., Osborn, D. L., Doudna, J. H., Pasternack, B. M., & Thompson, W. G. (1986). Turbine fast valving to
aid system stability: benefits and other considerations. Power Systems, IEEE Transactions on, 1(1), 143-153.
[40] Karady, G. G. (2002). Improving transient stability using generator tripping based on tracking rotor-angle. In Power Engineering
Society Winter Meeting, 2002. IEEE (Vol. 2, pp. 1113-1118). IEEE.
[41] Wang, Y., Mohler, R. R., Spee, R., & Mittelstadt, W. (1992). Variable-structure FACTS controllers for power system transient
stability. Power Systems, IEEE Transactions on, 7(1), 307-313.
[42] Japan, I. E. E. (1999). State-of-the art and trends in fault current limiting technologies. IEEJ Report, (709).
[43] Youjie, Z. X. X. L. M., & Xiaoning, X. (2004). The Development and Application of Fault Current Limiting Device [J].
Transactions of China Electrotechnical Society, 11, 000.
[44] Noe, M., & Steurer, M. (2007). High-temperature superconductor fault current limiters: concepts, applications, and development
status. Superconductor Science and Technology, 20(3), R15.
[45] S. Yadav, G. K. Choudhary, and R. K. Mandal (2014).Review on Fault Current Limiters.IJERT 3(4), 1595 – 1603
[46] Kumar, R., Harada, A., Merkle, M., & Miri, A. M. (2003, June). Investigation of the influence of series compensation in AC
transmission systems on bus connected parallel generating units with respect to subsynchronous resonance (SSR). In Power Tech
Conference Proceedings, 2003 IEEE Bologna (Vol. 3, pp. 6-pp). IEEE.
[47] Rahim, A. H. M. A., & Al-Sammak, A. I. J. (1991, January). Optimal switching of dynamic braking resistor, reactor or capacitor
for transient stability of power systems. In Generation, Transmission and Distribution, IEE Proceedings C (Vol. 138, No. 1, pp.
89-93). IET.
[48] Balu, N. (1980). Fast turbine valving and independent pole tripping breaker applications for plant stability. IEEE Transactions on
Power Apparatus and Systems, 4(PAS-99), 1330-1342.
[49] Taylor, C. W., Mittelstadt, W. A., Lee, T. N., Hardy, J. E., Glavitsch, H., Stranne, G., & Hurley, J. D. (1986). Single-pole
switching for stability and reliability. IEEE transactions on power systems, 2(1), 25-36.
[50] Grewal, G. S., Konowalec, J. W., & Hakim, M. (1998). Optimization of a load shedding scheme. Industry Applications Magazine,
IEEE, 4(4), 25-30.
[51] Pilo, F., Celli, G., & Mocci, S. (2004, April). Improvement of reliability in active networks with intentional islanding. In Electric
Utility Deregulation, Restructuring and Power Technologies, 2004.(DRPT 2004). Proceedings of the 2004 IEEE International
[52] Rahman, H., & Khan, B. H. (2008). Stability improvement of power system by simultaneous AC–DC power transmission.
Electric Power Systems Research, 78(4), 756-764.
[53] Hafiz, H. M., & Wong, W. K. (2004, November). Static and dynamic under-frequency load shedding: a comparison. In Power
System Technology, 2004. PowerCon 2004. 2004 International Conference on (Vol. 1, pp. 941-945). IEEE
[54] Uhlen, K., Pálsson, M., Time, T. R., Kirkeluten, Ø., & Gjerde, J. O. (2002, June). Raising stability limits in the Nordic power
transmission system. In 14th Power Systems Computation Conference
[55] El‐Shimy, M. (2014). Modeling and analysis of reactive power in grid‐connected onshore and offshore DFIG‐based wind farms.
Wind Energy, 17(2), 279-295.
[56] www.epri.com
[57] Edris, A. A. (1997). Proposed terms and definitions for flexible AC transmission system (FACTS). IEEE Transactions on Power
Delivery, 12(4).
[58] EL-Shimy M. Multi-objective Placement of TCSC for Enhancement of Steady-State Performance of Power System. Scientific
Bulletin - Faculty of Engineering - Ain Shams Uni. 2007;42(3):935 - 50.
[59] M. Mandour, M. EL-Shimy, F. Bendary, and W. Ibrahim, “Damping of Power Systems Oscillations using FACTS Power
Oscillation Damper – Design and Performance Analysis”, MEPCON 2014.
[60] N. Martins and L. Lima, "Eigenvalue and Frequency Domain Analysis of Small-Signal Electromechanical Stability Problems,"
IEEE Symposium on Application of Eigenanalysis and Frequency Domain Method for System Dynamic Performance, 1989.
[61] B. C. Pal, "Robust pole placement versus root-locus approach in the context of damping interarea oscillations in power systems,"
IEE Proceedings on Generation, Transmission and Distribution, vol. 149, Nov 2002.
[62] R. Rouco and F. L. Pagola, "An eigenvalue sensitivity approach to location and controller design of controllable series capacitors
for damping power system oscillations," IEEE Transactions on Power Systems, vol. 12, Nov 1997.
[63] R. Sadikovic, et al., "Application of FACTS devices for damping of power system oscillations," presented at the IEEE Power
Tech, St. Petersburg, Russia, June 2005.
[64] NGHingorani, L. G. (2000). Understanding facts concepts & tech nology of flexible AC transmission system. NewYork,
IEEEPowerEngineering Society.
[65] H. M. Ayres, I. Kopcak, M. S. Castro, F. Milano and V. F. d. Costa, "A didactic procedure for designing power oscillation
damper of facts devices", Simulation Modelling Practice and Theory, vol.18, no.6 June 2010
[66] R. Sadikovic, et al., "Application of FACTS devices for damping of power system oscillations," presented at the IEEE Power
Tech, St. Petersburg, Russia, June 2005.
[67] D. Ravi Kumar, N. Ravi Kiran " Hybrid Controller Based UPFC for Damping of Oscillations in Multi Machine Power Systems,"
International Journal of Engineering Research & Technology (IJERT) , ISSN: 2278-0181, Vol. 2, Issue 10, October - 2013.
[68] Mozhgan Balavar,” Using Neural Network to Control STATCOM for Improving Transient Stability” Journal of Artificial
Intelligence in Electrical Engineering, Islamic Azad University, Ahar, Iran, Vol. 1, No. 1, June 2012.
[69] Dakka Obulesu, S. F. Kodad, B V. Sankar Ram, " Damping of oscillations in multi-machine integrated power systems using hybrid
fuzzy strategies," International Journal of Research and Reviews in Applied Sciences, Vol. 1, Issue 2, November 2009.
[70] Prasanna Kumar Inumpudi and Shiva Mallikarjuna Rao N,” Development of a Fuzzy Control Scheme with UPFC’s For Damping of
oscillations in multi machine integrated power systems” International Journal of Engineering Research and Applications (IJERA),
Vol. 1, Issue 2, pp.230-234.
[71] Aghazade, A.; Kazemi, A.”Simultaneous coordination of power system stabilizers and STATCOM in a multi-machine power system
for enhancing dynamic performance”4th International Power Engineering and Optimization Conference (PEOCO), IEEE, pp: 13 – 18,
Shah Alam, 23-24 June 2010
[72] M. Mandour, M. EL-Shimy, F. Bendary, and W.M. Mansour. Design of power oscillation damping (POD) controllers in weakly
interconnected power systems including wind power technologies. Accepted for publications in the JEE, Oct. 2015.
[73] A. Berizzi, C. Bovo, V. Ilea,” Optimal placement of FACTS to mitigate congestions and inter-area oscillations” PowerTech, 2011
IEEE Trondheim, pp: 1 – 8, Trondheim, 19-23 June 2011
[74] J. Mehmedalic, T. Knuppel and J. Ostergaard,” Using H∞ to design robust POD controllers for wind power plants” Universities
Power Engineering Conference (UPEC), 2012 47th International, IEEE, pp.1-6, London, 4-7 Sept. 2012
[75] Issarachai Ngamroo and Tossaporn Surinkaew,” Optimization of Robust Power Oscillation Dampers for DFIG Wind Turbines
Considering N-1 Outage Contingencies” Innovative Smart Grid Technologies Conference Europe (ISGT-Europe), 2014 IEEE PES,
Istanbul, October 12-15
[76] EL-Shimy M. Multi-objective Placement of TCSC for Enhancement of Steady-State Performance of Power System. Scientific
Bulletin - Faculty of Engineering - Ain Shams Uni. 2007;42(3):935 – 50.
[77] S.N. Singh, A.K. David, “Optimal Location of FACTS Devices for Congestion Management”, Electric Power System Research,
Vol. 58, 2001, pp. 71-79.
[78] G.C. Ejebe, and B.F. Wollenberg, “Automatic Contingency selection”, IEEE, PAS, Vol. 98, No. 1, 1979, pp. 92-104.
[79] G Shaoyun, and T S Chung, “Optimal Power Flow Incorporating FACTS Devices with Power Flow Control Constraints”,
Electrical Power & Energy Systems, Vol. 20, No. 5, 1998, pp. 321-326
[80] Maczulak, A. (2010). Renewable Energy: Sources and Methods. New York: Infobase Publishing, 2010
[81] Sovacool, B. K.; Brown, M. A. (2010). Competing dimensions of energy security: An international perspective. Annual Review
of Environment and Resources, 35, 77-108.
[82] Difference between renewable and sustainable energy, http://www.solarpowernotes.com/difference-between-renewable-and-
sustainable-energy.html#.UnAsrlPnURQ
[83] Hand, M.M.; Baldwin, S.; DeMeo, E.; Reilly, J.M.; Mai, T.; Arent, D.; Porro, G.; Meshek, M.; Sandor, D. (Eds. 4 Vols). (2012).
Renewable Electricity Futures Study NREL/TP-6A20-52409. Golden, CO: National Renewable Energy Laboratory.
[84] Tsili, M.; Papathanassiou, S. (2009). A review of grid code technical requirements for wind farms. IET Renewable Power
Generation, 3(3), 308-332.
[85] El‐Shimy, M. (2012). Modeling and analysis of reactive power in grid‐connected onshore and offshore DFIG‐based wind farms.
Wind Energy. http://dx.doi.org/10.1002/we.1575, Dec. 6, 2012
[86] Holttinen, H.; Bettina Lemström, V. T. T.; Meibom, F. P.; Bindner, H. (2007). Design and operation of power systems with large
amounts of wind power. State-of-the-art report. VTT (Espoo))(= VTT Working Papers, 82). URL: http://www. ieawind.
org/AnnexXXV/Publications W, 82
[87] Parfomak, P. W. (2012). Energy storage for power grids and electric transportation: A technology assessment. Congressional
Research Service, Tech. Rep. R, 42455
[88] EL-Shimy, M. (2013). Probable power production in optimally matched wind turbine generators. Sustainable Energy
Technologies and Assessments, 2, 55-66.
[89] Spera, D. A. (2009). Wind Turbine Technology: Fundamental Concepts of Wind Turbine Engineering. American Society of
Mechanical Engineers
[90] EL-Shimy, M. (2010). Alternative configurations for induction-generator based geared wind turbine systems for reliability and
availability improvement. Proc. of IEEE 14th international middle-east power conference MEPCON2010. Cairo, Egypt.(pp. 617-
623). IEEE
[91] Ciang, C. C.; Lee, J. R.; Bang, H. J. (2008). Structural health monitoring for a wind turbine system: a review of damage detection
methods. Measurement Science and Technology, 19(12), 122001
[92] Milano, F. (2010). Power system modelling and scripting. Springer.
[93] Milano, F. (2008). Documentation for PSAT version 2.0.0, 2008.
[94] EL-Shimy, M. (2010) Steady State Modeling and Analysis of DFIG for Variable-Speed Variable- Pitch Wind Turbines. Ain
Shams Journal of Electrical Engineering, 1, 179 – 89
[95] Slootweg, J. G.; Kling, W. L. (2002). Modelling and analysing impacts of wind power on transient stability of power systems.
Wind Engineering, 26(1), 3-20
[96] El-Shimy, M.; Badr, M. A. L.; Rassem, O. M. (2008, March). Impact of large scale wind power on power system stability. In
12th International Middle-East Power System Conference MEPCON 2008. Aswan, Egypt. (pp. 630-636). IEEE
[97] Ghaly, N.; EL-Shimy, M.; Abdelhamed, M. Parametric study for stability analysis of grid-connected wind energy conversion
technologies. In 15th International Middle-East Power System Conference MEPCON 2012. Alexandria, Egypt. (pp. 1-7). IEEE
[98] R. Pena, J.C.Clare, and G. M. As her, Doubly fed induction generator using back-to-back PWM converters and its application to
variable-speed wind-energy generation, IEE Proc.-Electr. Power Appl, Vo1 143, No. 3, May 1996, pp. 231-24
[99] M. EL-Shimy, “ Modeling and Control of Wind Turbines Including Aerodynamics”, Scientific Bulletin - Faculty of Engineering -
Ain Shams Uni. Vol. 41, No. 2, June 30, 2006
[100] Mustafa Kayikçi and J. V. Milanovic´, “Assessing Transient Response of DFIG-Based Wind Plants—The Influence of
Model Simplifications and Parameters”, IEEE Transactions On Power Systems, Vol. 23, No. 2, May 2008, pp. 545-554
[101] EL-Shimy M, Badr MA. Modal Based Analysis of Induction Generator Infinite-Bus System. ASJEE. 2008; 2:249 – 61
[102] El-Shimy M, Badr M, Rassem O. Impact of large scale wind power on power system stability. In: Power system conference,
12th international middle-east power conference (MEPCON). Aswan, Egypt: IEEE; 2008. p. 630–6.
[103] El-Shimy M, Ghaly N, Abdelhamed M. Parametric study for stability analysis of grid-connected wind energy conversion
technologies. In: 15th International middle-east power conference (MEPCON). Alex., Egypt: IEEE; 2010. p. 1–8.
[104] EL-shimy M, Ghaly B. Grid-connected wind energy conversion systems: transient response. In: Encyclopedia of energy
engineering and Technology. 2nd ed., vol. IV. CRC Press, Taylor & Francis Group; 2014. p. 2162–83
[105] Multi-mass models are not considered for DFIG because of its controls which significantly cancel the effect of shaft stiffness
by decoupling the mechanical and electrical power
[106] J.G. Slootweg, W.L. Kling, “Modelling and analysing impacts of wind power on transient stability of power systems”, Wind
Engineering, v.26, n.1, 2002, pp.3-20.
[107] J.G. Slootweg, W.L. Kling, “The impact of large scale wind power generation on power system oscillations”, Electric Power
Systems Research, v.67, n.1, 2003, pp.9-20
[108] M. Ahmed, M. EL-Shimy, and M. Badr, “Advanced modeling and analysis of the loading capability limits of doubly-fed
induction generators,” Sustainable Energy Technologies and Assessments, vol. 7, pp. 79-90, 2014.
[109] EL-Shimy M. Reactive Power Management and Control of Distant Large-Scale Grid-Connected Offshore Wind Power
Farms. International Journal of Sustainable Energy (IJSE), 2012. Available online: Mar 20, 2012. Volume 32, Issue 5, pp. 449 -
465, 2013.
[110] M. EL-Shimy, “Modeling and analysis of reactive power in grid-connected onshore and offshore DFIG-based wind farms”,
Wind Energy, 17:279 – 295, 2014
[111] M. Mandour, M. EL-Shimy, F. Bendary, and W.M. Mansour. Design of power oscillation damping (POD) controllers in
weakly interconnected power systems including wind power technologies. Accepted for publications in the JEE, 2015
[112] M. Mandour, M. EL-Shimy, F. Bendary, and W.M. Mansour. The Design of POD Considering Conventional and Renewable
Power Generation. International Electrical Engineering Journal (IEEJ), Vol. 6 (2015) No.7, pp. 1962-1972
[113] M. Mandour, M. EL-Shimy, F. Bendary, W.M. Mansour. Impact of Wind Power on Power System Stability and Oscillation
Damping Controller Design. Industry Academia Collaboration (IAC) Conference, 2015, Energy and sustainable development
Track, Apr. 6 – 8, 2015, Cairo, Egypt.
[114] El Chaar, L., & El Zein, N. (2011). Review of photovoltaic technologies. Renewable and Sustainable Energy Reviews, 15(5),
2165-2175.
[115] Ecogeneration, (2010). Soaring to new heights:Solar PV in Australia. In ecogeneration pp. 1-53, Great Southern Press
Pty Ltd, Melbourne.
[116] Solarbuzz. Available from: www.solarbuzz.com
[117] Masson, G., Latour, M., & Biancardi, D. (2012). Global market outlook for photovoltaics until 2016. European Photovoltaic
Industry Association, 5.
[118] Masson, G., Latour, M., Rekinger, M., Theologitis, I. T., & Papoutsi, M. (2013). Global market outlook for photovoltaics
2013-2017. European Photovoltaic Industry Association, 12-32.
[119] Timilsina, G. R., Kurdgelashvili, L., & Narbel, P. A. (2012). Solar energy: Markets, economics and policies. Renewable and
Sustainable Energy Reviews, 16(1), 449-465.
[120] European Photovoltaic Industry Association (EPIA). (2011). Solar generation 6: Solar photovoltaic electricity empowering
the world. Retrieved February, 23, 2012. Greenpeace International & European Photovoltaic Industry Association: Brussels,
Belgium.
[121] OECD/IEA (2008). Energy Technology Perspectives: Scenarios and Strategies to 2050; International Energy Agency: Paris,
France.
[122] Aringhoff, R.; Brakmann, G.; Geyer, M.; Teske, S.; Baker, C (2005). Concentrated Solar Thermal Power- Now! Greenpeace
International, European Solar Thermal Power Industry Association (ESTIA): Amsterdam, the Netherlands, IEA SolarPACES
Implementing Agreement.
[123] Teske, S.; Zervos, A.; Scha¨fer, O (2007). Energy revolution: a sustainable world energy outlook; Greenpeace International-
European Renewable Energy Council (EREC): Amsterdam, the Netherlands.
[124] Chapin, D. M., Fuller, C. S., & Pearson, G. L. (1954). A new silicon p‐n junction photocell for converting solar radiation
into electrical power. Journal of Applied Physics, (25), 676-677.
[125] Bradford, T. (2006). Solar revolution: the economic transformation of the global energy industry. MIT Press Books, 1.
[126] REN21. Global Status Report 2005–2011; Renewable Energy Policy Network for the 21st century (REN 21) 2005–2011:
Paris, France.
[127] IRENA (2012). Renewable energy technologies: Cost analysis series- solar Photovoltaics. pp. 45, International
Renewable Energy Agency, Abu Dhabi.
[128] Poullikkas, A. (2010). Technology and market future prospects of photovoltaic systems. International Journal of Energy and
Environment, 1(4), 617-634.
[129] Manna, T. K., & Mahajan, S. M. (2007, May). Nanotechnology in the development of photovoltaic cells. In Clean Electrical
Power, 2007. ICCEP'07. International Conference on (pp. 379-386). IEEE.
[130] Mah, O. (1998). Fundamentals of photovoltaic materials. National Solar Power Research Institute, 1-10.
[131] Sheikh, N. M. (2008, March). Efficient utilization of solar energy for domestic applications. In Electrical Engineering, 2008.
ICEE 2008. Second International Conference on (pp. 1-3). IEEE.
[132] T. Abdo, M. EL-Shimy, “Quantitative Characterization and Selection of Photovoltaic Technologies”, MEPCON 2014.
[133] M. EL-Shimy, T. Abdo, “PV Technologies: History, Technological Advances, and Characterization’, In Sohail Anwar (ed.)
Encyclopedia of Energy Engineering and Technology, Taylor & Francis - CRC Press, 2014.
[134] Salameh, Z. M., Borowy, B. S., & Amin, A. R. (1995). Photovoltaic module-site matching based on the capacity factors.
Energy Conversion, IEEE Transactions on, 10(2), 326-332.
[135] "Renewable energy project analysis software", in RETScreen International Clean Energy Decision Support Centre,
Minister of Natural Resources Canada.
[136] Sherwani, A. F., & Usmani, J. A. (2010). Life cycle assessment of solar PV based electricity generation systems: A review.
Renewable and Sustainable Energy Reviews, 14(1), 540-544.
[137] Jordan, D. C., & Kurtz, S. R. (2013). Photovoltaic degradation rates—an analytical review. Progress in photovoltaics:
Research and Applications, 21(1), 12-29.
[138] PVXChange, http://www.pvxchange.com/
[139] DasGupta, A. (2010). Fundamentals of probability: a first course. Springer Science & Business Media.
[140] El-Shimy, M. (2009). Viability analysis of PV power plants in Egypt. Renewable Energy, 34(10), 2187-2196.
[141] M. EL-Shimy, “Analysis of Levelized Cost of Energy (LCOE) and grid parity for utility-scale photovoltaic generation
systems”, 15th International Middle East Power Systems Conference (MEPCON’12), Dec. 23-25, 2012, Alexandria, Egypt, pp. 1-
7.
[142] M. Saed, M. EL-Shimy, and M. Abdelraheem, “Photovoltaics Energy: Improved modeling and analysis of the Levelized
Cost of Energy (LCOE) and Grid Parity”, Sustainable Energy Technologies and Assessments, vol. 9, pp. 37-48, 2015.
[143] Jordan, D. C., & Kurtz, S. R. (2013). Photovoltaic degradation rates—an analytical review. Progress in photovoltaics:
Research and Applications, 21(1), 12-29.
[144] Campbell, M., Aschenbrenner, P., Blunden, J., Smeloff, E., & Wright, S. (2008). The drivers of the levelized cost of
electricity for utility-scale photovoltaics. White Paper: SunPower Corporation.
[145] Zweibel, K., Mason, J., & Fthenakis, V. (2008). A solar grand plan. Scientific American, 298(1), 64-73.
[146] Skoczek, A., Sample, T., & Dunlop, E. D. (2009). The results of performance measurements of field‐aged crystalline silicon
photovoltaic modules. Progress in Photovoltaics: Research and Applications, 17(4), 227-240.
[147] Sopori, B., Basnyat, P., Devayajanam, S., Shet, S., Mehta, V., Binns, J., & Appel, J. (2012, June). Understanding light-
induced degradation of c-Si solar cells. In Photovoltaic Specialists Conference (PVSC), 2012 38th IEEE (pp. 001115-001120).
IEEE
[148] Leblanc, S., & Roman, P. (2005). Project Management: An Engineering Economics Perspective. Pearson Custom
Publishing.
[149] Skoczek, A., Sample, T., & Dunlop, E. D. (2009). The results of performance measurements of field‐aged crystalline silicon
photovoltaic modules. Progress in Photovoltaics: Research and Applications, 17(4), 227-240.
[150] Dunlop, E. D., Halton, D., & Ossenbrink, H. (2005, January). 20 years of life and more: where is the end of life of a PV
module?. In Photovoltaic Specialists Conference, 2005. Conference Record of the Thirty-first IEEE (pp. 1593-1596). IEEE.
[151] Vázquez López, M., & Rey-Stolle Prado, I. (2008). Photovoltaic module reliability model based on field degradation studies.
Progress in Photovoltaics: Research and Applications, 16(5), 419-433.
[152] Cereghetti, N., Bura, E., Chianese, D., Friesen, G., Realini, A., & Rezzonico, S. (2003, May). Power and energy production
of PV modules statistical considerations of 10 years activity. In Photovoltaic Energy Conversion, 2003. Proceedings of 3rd World
[153] Swanson, R. M. (2009). Photovoltaics power up. Science, 324(5929), 891-892.
[154] Devabhaktuni, V., Alam, M., Depuru, S. S. S. R., Green, R. C., Nims, D., & Near, C. (2013). Solar energy: Trends and
enabling technologies. Renewable and Sustainable Energy Reviews, 19, 555-564.
[155] http://pveducation.org/
[156] Dolara, A., Faranda, R., & Leva, S. (2009). Energy comparison of seven MPPT techniques for PV systems. Journal of
Electromagnetic Analysis and Applications, 2009.
[157] Faranda, R., & Leva, S. (2008). Energy comparison of MPPT techniques for PV Systems. WSEAS transactions on power
systems, 3(6), 446-455.
[158] https://en.wikipedia.org/wiki/Boost_converter
[159] Dolara, A., Faranda, R., & Leva, S. (2009). Energy comparison of seven MPPT techniques for PV systems. Journal of
Electromagnetic Analysis and Applications, 2009
[160] El-Shimy, M. (2013). Sizing optimisation of stand-alone photovoltaic generators for irrigation water pumping systems.
International Journal of Sustainable Energy, 32(5), 333-350.
[161] Ulgen, K. (2006). Optimum tilt angle for solar collectors. Energy Sources, Part A: Recovery, Utilization, and Environmental
Effects, 28(13), 1171-1180
[162] System Advisor Model (SAM). SAM's Help system, National Renewable Energy Laboratory (NREL), September 2013
[163] http://www.pvresources.com/PVSystems/Trackers.aspx
[164] Kumar, A., Thakur, N., Makade, R., & Shivhare, M. K. (2011). Optimization of tilt angle for photovoltaic array.
International Journal of Engineering Science and Technology, 3(4), 3153-3161.
[165] Ong, S., Campbell, C., Denholm, P., Margolis, R., & Heath, G. (2013). Land-use requirements for solar power plants in the
United States. Golden, CO: National Renewable Energy Laboratory.
[166] http://solarprofessional.com/articles/products-equipment/racking/pv-trackers
[167] Afgan, N. H., & Carvalho, M. G. (2004). Sustainability assessment of hydrogen energy systems. International Journal of
Hydrogen Energy, 29(13), 1327-1342
[168] Abdo, T., & EL-Shimy, M. (2013). Estimating the global solar radiation for solar energy projects–Egypt case study.
International Journal of Sustainable Energy, 32(6), 682-712.
[169] Yesilata, B., & Firatoglu, Z. A. (2008). Effect of solar radiation correlations on system sizing: PV pumping case. Renewable
Energy, 33(1), 155-161
[170] King, D. L., Boyson, W. E., & Kratochvil, J. A. (2004). Photovoltaic Array Performance Model, Sandia National Labs
[171] System Advisor Model (SAM). SAM's Help system. Available: http://sam.nrel.gov
[172] European Photovoltaic Industry Association. (2014). Global market outlook for photovoltaics 2014–2018.[Online]
Available: http://www. epia. org/filead min/user_upload. Publications/44_epia_gmo_report_ver_ 17_mr. pdf.
[173] De Soto, W., Klein, S. A., & Beckman, W. A. (2006). Improvement and validation of a model for photovoltaic array
performance. Solar energy, 80(1), 78-88.
[174] Chenni, R., Makhlouf, M., Kerbache, T., & Bouzid, A. (2007). A detailed modeling method for photovoltaic cells. Energy,
32(9), 1724-1730.
[175] Tian, H., Mancilla-David, F., Ellis, K., Muljadi, E., & Jenkins, P. (2012). A cell-to-module-to-array detailed model for
photovoltaic panels. Solar Energy, 86(9), 2695-2706.
[176] Chouder, A., Silvestre, S., Sadaoui, N., & Rahmani, L. (2012). Modeling and simulation of a grid connected PV system
based on the evaluation of main PV module parameters. Simulation Modelling Practice and Theory, 20(1), 46-58.
[177] Yang, Y., Chen, W., & Blaabjerg, F. (2014). Advanced Control of Photovoltaic and Wind Turbines Power Systems. In
Advanced and Intelligent Control in Power Electronics and Drives (pp. 41-89). Springer International Publishing.
[178] Salam, Z., Ahmed, J., & Merugu, B. S. (2013). The application of soft computing methods for MPPT of PV system: A
technological and status review. Applied Energy, 107, 135-148.
[179] Onat, N. (2010). Recent developments in maximum power point tracking technologies for photovoltaic systems.
International Journal of Photoenergy, 2010.
[180] Ahmed, J., & Salam, Z. (2014). A Maximum Power Point Tracking (MPPT) for PV system using Cuckoo Search with partial
shading capability. Applied Energy, 119, 118-130.
[181] Bellia, H., Youcef, R., & Fatima, M. (2014). A detailed modeling of photovoltaic module using MATLAB. NRIAG Journal
of Astronomy and Geophysics, 3(1), 53-61.
[182] Tsai, H. L., Tu, C. S., & Su, Y. J. (2008, October). Development of generalized photovoltaic model using
MATLAB/SIMULINK. In Proceedings of the world congress on engineering and computer science (Vol. 2008, pp. 1-6).
Citeseer.
[183] George, K. (2006). DC Power Production, Delivery and Utilization: An EPRI White Paper. Available at: h ttp://www. epri.
org [Accessed: 2.4. 12].
[184] Hua, C., & Shen, C. (1998, May). Study of maximum power tracking techniques and control of DC/DC converters for
photovoltaic power system. In Power Electronics Specialists Conference, 1998. PESC 98 Record. 29th Annual IEEE (Vol. 1, pp.
86-93). IEEE.
[185] Duffey, C. K., & Stratford, R. P. (1989). Update of harmonic standard IEEE-519: IEEE recommended practices and
requirements for harmonic control in electric power systems. Industry Applications, IEEE Transactions on, 25(6), 1025-1034.
[186] Espinosa, C. A. L. (2014). State Variable Model for Considering the Parasitic Inductor Resistance on the Open Loop
Performance of DC to DC Converters. Journal of Computer and Communications, 2(13), 41.
[187] Hua, C., & Shen, C. (1998, May). Study of maximum power tracking techniques and control of DC/DC converters for
photovoltaic power system. In Power Electronics Specialists Conference, 1998. PESC 98 Record. 29th Annual IEEE (Vol. 1, pp.
86-93). IEEE.
[188] Wang, L. (2012). Modeling and Control of Sustainable Power Systems. Springer
[189] http://www.mathworks.com/matlabcentral/fileexchange/34752-grid-connected-pv-
array/content/PVarray/PVarray_Grid_PandO_avg.mdl
[190] El‐Shimy, M. (2014). Modeling and analysis of reactive power in grid‐connected onshore and offshore DFIG‐based wind
farms. Wind Energy, 17(2), 279-295.
[191] Holttinen, H., Meibom, P., Ensslin, C., Hofmann, L., Mccann, J., & Pierik, J. (2009). Design and operation of power systems
with large amounts of wind power. In VTT Research Notes 2493.
[192] Bernard, S., Beaulieu, D., & Trudel, G. (2005). Hydro-Quebec grid code for wind farm interconnection. In Power
Engineering Society General Meeting, 2005. IEEE (pp. 1248-1252).
[193] Egyptian Electricity Transmission Company (EETC). Wind Farm Grid Connection Code as a part of the Egyptian
Transmission Grid Code - 1st Draft on January, 5, 2012. EETC, Egypt, 2012
[194] N. Ghaly, M. EL-Shimy, M. Abdelhamed (2014). Consistence of Wind Power Technologies with the Fault Ride-through
Capability Requirements. MEPCON 2014 international conference, Cairo, Egypt.
[195] M. EL-Shimy, (2015). Wind Energy Conversion Systems: Reliability Prospective,” Encyclopedia of Energy Engineering
and Technology – Second Edition. Vol. 2, pp. 2184 – 2206.
[196] REN21 (2014). "Renewables 2014: Global Status Report". pp. 15, 16. Available at
http://www.webcitation.org/6SKF06GAX
[197] Ellis, A., Karlson, B., & Williams, J. (2012). Utility-Scale Photovoltaic Procedures and Interconnection Requirements.
Sandia National Laboratories SAND2012-2090.
[198] Troester, E. (2009, March). New German grid codes for connecting PV systems to the medium voltage power grid. In 2nd
International workshop on concentrating photovoltaic power plants: optical design, production, grid connection (pp. 9-10).
[199] IEEE. (2000). IEEE Recommended Practice for Utility Interface of Photovoltaic (PV) Systems. IEEE.
[200] H. Khairy, M. EL-Shimy, G. Hashem. Overview of Grid Code and Operational Requirements of Grid-connected Solar PV
Power Plants. Industry Academia Collaboration (IAC) Conference, 2015, Energy and sustainable development Track, Apr. 6 – 8,
2015, Cairo, Egypt. ‫‏‬
[201] International Electrotechnical Commission, (2004). IEC 61727 - Photovoltaic (PV) systems – Characteristics of the utility
interface.
[202] Crăciun, B. I., Kerekes, T., Séra, D., & Teodorescu, R. (2012, May). Overview of recent grid codes for PV power
integration. In Optimization of Electrical and Electronic Equipment (OPTIM), 2012 13th International Conference on (pp. 959-
965). IEEE
[203] Std, N.E.R.C. "PRC-024-1/Draft 6." Draft on Generator Frequency and Voltage Protective Relay Settings (2013).
[204] IEEE. (2000). IEEE Recommended Practice for Utility Interface of Photovoltaic (PV) Systems. IEEE.
[205] IEEE 519 Working Group. (1992). IEEE recommended practices and requirements for harmonic control in electrical power
systems. IEEE STD, 519-1992.
[206] M. EL-Shimy. “Renewable energy: Grid Codes and Integration”. Lecture notes available at:
http://shimymb.tripod.com/id12.html
[207] Guozhen, H., Tao, C., Changsong, C., & Shanxu, D. (2009, May). Solutions for SCADA system communication reliability
in photovoltaic power plants. In Power Electronics and Motion Control Conference, 2009. IPEMC'09. IEEE 6th International
(pp. 2482-2485). IEEE.
[208] Troester, E. (2009, March). New German grid codes for connecting PV systems to the medium voltage power grid. In 2nd
International workshop on concentrating photovoltaic power plants: optical design, production, grid connection (pp. 9-10).
[209] M. EL-Shimy, and N. Ghaly, (2015). Grid-Connected Wind Energy Conversion Systems: Transient Response. Encyclopedia
of Energy Engineering and Technology, second edition. Vol. 2, pp. 2162 – 2183
[210] Order 661 - Interconnection with Wind Energy, issued by Federal Energy Regulatory Commission (FERC) of United States,
June 2, 2005
[211] Std, N.E.R.C. "PRC-024-1/Draft 6." Draft on Generator Frequency and Voltage Protective Relay Settings (2013).‫‏‬
[212] Guozhen, H., Tao, C., Changsong, C., & Shanxu, D. (2009, May). Solutions for SCADA system communication reliability
in photovoltaic power plants. In Power Electronics and Motion Control Conference, 2009. IPEMC'09. IEEE 6th International
(pp. 2482-2485). IEEE.
[213] Demello, F. P., & Concordia, C. (1969). Concepts of synchronous machine stability as affected by excitation control. IEEE
Transactions on power apparatus and systems, 88(4), 316-329
[214] Wen-Shiow, “The Effect of Load Models on Unstable Low-Frequency Damping in Taipower System Experience w/wo
Power System Stabilizers”, IEEE trans. on power systems, Vol. 16, No. 3, Feb. 2001
[215] Abuel-wafa AR, EL-Shimy M. Dynamic Aggregation of Induction Motor Loads for Transient Stability Studies. Scientific
Bulletin - Faculty of Engineering - Ain Shams Uni 2003 June 30;38(2):553 - 67.
[216] Torbjorn Thiringer, “Comparison of Reduced-Order Dynamic Models of Induction Machines”, IEEE, P.S., Vol. 16, No. 1,
Feb. 2001
[217] Graham J., others, “An Aggregate Motor Model for Industrial Plants”, PAS-103, No. 4, April 1984
[218] Farhad Nozari, M. David, “Aggregation of Induction Motors for Transient Stability Load Modeling”, IEEE, P.S., PWRS-2,
No.4, Nov. 1987
[219] K.O. Abeldu, “Equivalent Load Model of Induction Machines Connected to a Common Bus”, Ph.D. Thesis, Iowa State Uni.,
Ames, Iowa, 1983
[220] Aleksander M. Stankovic, Bernard C. Lesieutre, “A Probabilistic Approach to Aggregate Induction Machine Modeling”,
IEEE trans. on power systems, Vol. 11, No. 4, Nov. 1996
[221] William W. , others, “Load Modeling for Power Flow and Transient Stability Computer Studies”, IEEE trans. on power
systems, Vol. 3, No. 1, Feb. 1988
[222] V. Venikov: “Transient Processes in Electrical Power Systems”, Mir Publishers–Moscow 1980
[223] P.M. Anderson & A.A. Fouad, “Power System Control and Stability”, Galgotia publications, 1981
[224] Wood, A. J., & Wollenberg, B. F. (2012). Power generation, operation, and control. John Wiley & Sons

Dynamic Security of Interconnected Electric Power Systems - Volume 2: Dynamics and Stability of Conventional and Renewable Energy Systems

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Dynamic Security of Interconnected Electric Power Systems - Volume 2: Dynamics and Stability of Conventional and Renewable Energy Systems

Caricato da

Copyright:

Formati disponibili

Power System Dynamic Security - Volume 2

Dr. M. EL-Shimy is currently an Assoc. Prof. in the

To my wife Hala, and my children (Sara, Malak, Omar, and Aly)

measurements. Single-machine infinite-bus (SMIB) equivalences of power

- Chapter 9: dynamical impacts of wind power and damping

The following table summarizes the main subject(s) associated with

the literature review of solar-PV technologies is performed by Eng. Taha

Generally, nothing is either absolutely perfect or final. Therefore, please do

M. EL-Shimy, Nov. 2015

TABLE OF CONTENTS (VOLUME 2)

7.8.2. General form of static-sensitivity analysis 291

10.4.3 Earth-Sun Geometry and Observer Angles 457

12.2 FRT Capability Requirements 530

of the excitation control

4.2 Identification and characteristics of steady and dynamic

main target and reasonable accuracy as well as high simulation speed is

I. Slow changes with small magnitudes: the reaction of a power system to

It should be noted that the impact of active power mismatches is

4.3 Frequency dynamics of power systems

 The generator currents and rotor magnetic field create a constant

Assume now this equilibrium is suddenly upset by an electrical load

Fig. 4.2: Model for the ALFC loops 8

Fig. 4.3: Average frequency of the system and frequencies of individual

The frequency dynamics can be classified according to the strength of

Fig. 4.4: Typical small frequency variations 7

Fig. 4.5: Large frequency variations resulting from a generation trip of

As previously mentioned, the inertia of the generators, and the ALFC

4.4 Linearized classical modeling and simulation of the traditional SMIB

Fig. 4.7: The traditional SMIB system

In the following it is assumed that,

4.4.1 The swing equation

where J is the total moment of inertia in kg.m2, m is the rotor shaft

Fig. 4.9: Classical dynamic model of conventional synchronous generators

Fig. 4.10: the power angle 

where Dm is the damping torque coefficient in N.m.sec.

Since, the synchronous speed is a constant value, then

Since the power = torque x angular speed, then .

where is the damping torque coefficient and Mm = Jsm

The previous modeling of the swing includes mechanical quantities while

The swing equation can be represented in p.u by dividing (4.16) by Sn,

In this representation, the inertia constant in seconds as well

Although we assumed that the damping coefficient is a constant value

1. The induction motor action (i.e. the electromagnetic damping) due to

4.4.2 The transient power-angle characteristics

causes an increase of the maximum power. For simplification, it is assumed

Table 4.1: Power-angle characteristics of the SMIB system considering

E q'  E ' cos(   ' )  E ' (4.24)

 It is also assumed that the saliency effects on the characteristics are

Fig. 4.18 illustrates a comparison between exact and approximate power –

Fig. 4.18: Exact and approximate power-angle characteristics

4.4.3 Linearization and linearized analysis

The swing equation is then a nonlinear function of the power angle ;

where  is a differential operator i.e. d/dt operator.

Linearizing (4.27) about an initial operating condition represented by o as

Ps  Pmax cos  o (4.31)

Therefore, (4.29) becomes,

Equation (4.32) can be written matrix form:

Which takes the standard state-space form form:

It is clear that the state matrix A is dependent on the system parameters D,

Fig. 4.19: Blok-diagram of the SMIB system with classical model

As an example, the SMIB system shown in Fig. 4.7 consists of a 50-Hz

 Case #1: The response of the system as subjected to a sudden

with Power Oscillation Damping (POD) controllers. PODs can be designed