Power Systems
Naser Mahdavi Tabatabaei Ali Jafari Aghbolaghi Nicu Bizon Frede Blaabjerg Editors
Reactive Power Control in AC Power Systems Fundamentals and Current Issues
Power Systems
More information about this series at http://www.springer.com/series/4622
Naser Mahdavi Tabatabaei Ali Jafari Aghbolaghi Nicu Bizon Frede Blaabjerg •
Editors
Reactive Power Control in AC Power Systems Fundamentals and Current Issues
123
Editors Naser Mahdavi Tabatabaei Electrical Engineering Department Seraj Higher Education Institute Tabriz Iran
Nicu Bizon Faculty of Electronics, Communication and Computers University of Piteşti Piteşti Romania
Ali Jafari Aghbolaghi Zanjan Electric Energy Distribution Company Zanjan Iran
ISSN 1612-1287 Power Systems ISBN 978-3-319-51117-7 DOI 10.1007/978-3-319-51118-4
Frede Blaabjerg Department of Energy Technology Aalborg University Aalborg Denmark
ISSN 1860-4676
(electronic)
ISBN 978-3-319-51118-4
(eBook)
Library of Congress Control Number: 2017930608 © Springer International Publishing AG 2017 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, express or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Printed on acid-free paper This Springer imprint is published by Springer Nature The registered company is Springer International Publishing AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Dedicated to all our teachers and colleagues who enabled us to write this book, and our family and friends for supporting us all along
Foreword
Electric power systems will be operated in reliable and efficient situation considering reactive power control and voltage stability management. Reactive power margins are related to the voltage stability. The aspects are satisfied by designing and operating of right voltages limits, maximizing utilization of transmission systems and minimizing of reactive power flow. Therefore, controlling reactive power and voltage is one of the major challenges of power system engineering. Reactive power as the dissipated power is affected by capacitive and inductive phenomena that they drop voltage and draw current in the form of heat or waste energy. Reactive power is generated by the capacitors and generators, whereas it is consumed by the inductors and is essential in the parallel connection circuits as power factor controlling and power transmission lines. Reactive power control and voltage stability aspects are effective in reliability of electric power networks. Voltage instability commonly occurs as a result of reactive power deficiency. The trends are to reduce reactive power and increase voltage stability to improve efficiency and operation of power systems. There is a direct relation between reactive power and voltage behavior which serves the voltage collapse and rising effects in power systems. Regulating the reactive power and voltage control should be done according to flexible and fast controlled devices. Placement and adjustment of reactive power play important roles in operation of reactive power compensation and voltage control. Therefore, the operations of reactive power resources in the power systems such as automatic transformer tap changer, synchronous condenser, capacitor banks, capacitance of overhead lines and cables, static VAR compensators and FACTS devices are very significant. Reactive power control and voltage stability management are considered as regional challenges to meet, which otherwise can cause the scale of blackouts increase in the power systems. Theoretical and application issues in these areas help us to identify problems related to reliability and stability of the power systems and prevent the system degradation.
vii
viii
Foreword
The above aspects are illustrated in this book by the editors and authors, in the following topics: electrical power systems operation and control, reactive power and voltage stability in power systems, reactive power control in transmission lines, reactive power compensation and optimal placement, reactive power in renewable resources, reactive power optimization and software applications, optimal reactive power dispatch, induction generator operation and analysis, communication networks and standards in power systems, power systems SCADA applications, and geomagnetic storms effects in electric networks. The book chapters and materials are very efficient in theoretical and application issues and are highly recommended for studying and considering in educational and research fields. November 2016
Academician Arif M. Hashimov Institute of Physics Azerbaijan National Academy of Sciences Baku Azerbaijan
Preface
The modern electric power systems are more expanded worldwide and include more energy resources and critical parts based on the requirements of the twenty-first century. General parts of electric power systems as generation, transmission and consumption are important to be analyzed and well operated for the development of industry and life. The engineers and scientists need applicable and renewable methods for analyzing and controlling each part of the electric power systems and to overcome complicated actions which occur in the systems due to their operational and interconnection behaviors. The objective of the analysis is minimizing the losses of the networks and increasing the overall efficiency and economic advantages. The central and distributed generation of electric power networks connect to more loads, transmission lines, transformers and energy sources together including nonlinear equipment such as power semiconductor devices. The engineers and scientists are interested in analyzing the power systems operations to control and develop the AC/DC networks including high voltage transmission lines and equipment. Flexible and fast power flow control and transmission are expected to raise the network effective operation, power wheeling requirement and transmission capability as well as voltage stability. Computational intelligence methods are applied to electric power analysis to facilitate the effective analysis techniques and solve several power system problems especially in power transmission and voltage stability. Reactive Power Control in AC Power Systems: Fundamentals and Current Issues is a book aimed to highlight the reactive power control and voltage stability concepts and analysis to provide understanding on how they are affected by different criteria of available generations, transmissions and loads using different research methods. A large number of specialists joined as authors of the book chapters to provide their potentially innovative solutions and research related to reactive power control and voltage stability, in order to be useful in developing new ways in electric power analysis, design and operational strategies. Several theoretical researches, case ix
x
Preface
analysis, and practical implementation processes are put together in this book that aims to act as research and design guides to help graduates, postgraduates and researchers in electric power engineering and energy systems. The book, which presents significant results obtained by leading professionals from industries, research and academic fields, can be useful to a variety of groups in specific areas. All works contributed to this book are new, previously unpublished material or extended version of published papers in the proceedings of international conferences and transactions on international journals. The book consists of 16 chapters in three parts.
Part I Fundamentals of Reactive Power in AC Power Systems The six chapters in the first part of this book present the fundamentals of reactive power in AC power systems considering different operating cases. The topics in this part include the advanced methods and applications in electric power systems and networks related to the fields of fundamentals of reactive power in AC power systems, reactive power role in AC power transmission systems, reactive power compensation in energy transmission systems with sinusoidal and nonsinusoidal currents, reactive power importance in wind power plants, and fundamentals and contemporary issues of reactive power control in AC power systems. Chapter 1 describes the general overview of electric power systems including power generation, transmission and distribution systems, linear AC circuits in steady state conditions, flow of power between generator and customers is studied by using the active, reactive, apparent and complex power, electric power system quality, measurement and instrumentation methods of power systems parameters, and general standards in energy generation, transmission and marketing. The importance of reactive power in AC power systems and its various interpretations are also discussed in this chapter. The basic theory of AC circuits, behavior of two-port linear elements and analysis methods of AC circuits are given in Chap. 2. The physical interpretation of electric powers in AC power systems, fundamental problems of reactive power consumption automated management in power systems, equipment for power factor correction, designing simple systems for compensating of reactive power for different levels of installation, the overall harmonic distortion of voltage and current, and qualitative and quantitative aspects related to active and reactive power circulation in AC power systems including several examples and case studies referring to classical linear AC circuits under sinusoidal and nonsinusoidal conditions are also the topics of this chapter. Chapter 3 presents basic principles of power transmission operation, equipment for reactive power generation, shunt/series compensation, control of reactive power in power transmission system. The chapter describes the capacitive and inductive properties of power transmission lines and also reactive power consumption by transmission lines which increases with the square of current. The chapter states the
Preface
xi
sources, effects and limitations of the reactive power and flowing in transmission lines and transformers as well as control of reactive power should satisfy the bus voltages, system stability and network losses in the power systems. The definition of reactive power under nonsinusoidal conditions in nonlinear electric power systems is described in Chap. 4. This chapter discusses and simulates the reactive power compensation for sinusoidal and nonsinusoidal situations, where nonlinear circuit voltages and currents contain harmonics and also the control algorithms of automatic compensators. The main aim of the chapter is based on the dissipative systems and cyclodissipativity theories for calculation of compensation elements for reactive power compensation by minimizing line losses. The chapter is also including the examples and computer simulations to show the mathematical framework for analyzing and designing of compensators for reactive power compensation in general nonlinear loads. Chapter 5 deals with the rate of reactive power absorption or injected by the wind units and also the key role of reactive power generation and consumption in large-scale wind farms. The chapter describes requirements of reactive power compensation, voltage stability and also power quality improvement in the electric grid of wind turbine to reduce the power losses and control of voltage level. The units of wind turbines of types 1 to 4 are also categorized and discussed in the chapter considering their construction, generation, converters, reactive power and voltage control abilities. The coordination related to reactive power adjustment in the wind turbines is also discussed in this chapter. The concept of power quality and voltage stability improvement based on the reactive power control is introduced in Chap. 6. The chapter describes the impact of reactive power flow in the power system and defines the power components of electrical equipment that produces or absorbs reactive power. Then the reactive power control and relations between reactive power and voltage stability are presented. The chapter also contains reactive power control methods for voltage stability and presents voltage control management based on case studies.
Part II Compensation and Reactive Power Optimization in AC Power Systems The second part of this book tries to highlight in six chapters the concepts of reactive power optimization and compensation. The topics in this part include optimal reactive power control for voltage stability improvement, reactive power compensation, optimal placement of reactive power compensators, reactive power optimization in classic methods and also using MATLAB and DIgSILENT, and multi-objective optimal reactive power dispatch. Chapter 7 is entirely focused on the voltage stability control using three main techniques of reactive power management, active power re-dispatch, and load shedding. The chapter discusses about determining the location of VAR sources
xii
Preface
and their setting and installation, online and offline reactive power dispatch, and optimal reactive power flow (ORPF). The reactive power flow and voltage magnitudes of generator buses, shunt capacitors/reactors, output of static reactive power compensators, transformer tap-settings are considered as the control parameters and are used for minimizing the active power loss and improving of the voltage profile in ORPF. This chapter also confers the reactive power dispatch as a nonlinear and nonconvex problem with equality and inequality constraints. The reactive power compensators based on advanced industrial applications are highlighted in Chap. 8. The basic theoretical background of reactive power compensation as well as conventional compensators and improved FACTS are introduced in the chapter. The compensation devices including shunt, series and shunt-series configurations for transmission lines regarding their characteristics and also analytical expressions are presented in the chapter. The power flow control, voltage and current modifications as well as stability issues are also analyzed and compared for similar compensation devices and emerging technologies. Chapter 9 provides a framework and versatile approach to develop a multi-objective reactive power planning (RPP) strategy for coordinated handling of reactive power from FACTS devices and capacitor banks. This chapter deals with power system operators for determining the optimal placement of FACTS devices and capacitor banks should be injected in the network to improve simultaneously the voltage stability, active power losses and cost of VAR injection. A formulation and solution method for reactive power planning, and voltage stability based on cost functions are also presented in the chapter. Chapter 10 presents the reactive power optimization using artificial optimization algorithms as well as the formulations and constraints to implement reactive power optimization. The classic method of reactive power optimization and basic principles and problem formulation of reactive power optimization using artificial intelligent algorithms are discussed in the chapter. In addition, this chapter focuses on the particle swarm optimization algorithm and pattern search method application in reactive power optimization including the case studies. The efficient approach using parallel working of MATLAB and DIgSILENT software with the intention of reactive power optimization is discussed in Chap. 11. This chapter presents the toolboxes, functions and flexibility powers of MATLAB and DIgSILENT in electrical engineering calculation and implementation. Also it provides the advantages of parallel calculations of MATLAB and DIgSILENT and relation of two software to carry out the heuristic algorithms as fast, simple and accurate as possible to optimize reactive power in AC power systems. In Chap. 12, the reactive power compensation devices are modeled using deterministic multi-objective optimal reactive power dispatch (DMO-ORPD) and two-stage stochastic multi-objective optimal reactive power dispatch (SMO-ORPD) in discrete and continuous studies. They are formulated as mixed integer nonlinear program (MINLP) problems, and solved by general algebraic modeling system (GAMS). A case study for evaluation of the performance of different proposed MO-ORPD models is also shown in the chapter. This chapter presents the MO-ORPD problem taking into account different operational constraints such as
Preface
xiii
bus voltage limits, power flow limits of branches, limits of generators voltages, transformers tap ratios and the amount of available reactive power compensation at the weak buses.
Part III Challenges, Solutions and Applications in AC Power Systems The final part of this book consists of four chapters and considers some applications and case studies in AC power systems related to the issues of active and reactive power concepts. The topics in this part include self-excited induction generator, communications for electric power systems, SCADA applications for electric power systems and effect of geomagnetic storms on electrical networks. Chapter 13 discusses about a three-phase self-excited induction generator in an autonomous power generation mode. The chapter presents generator operating points and control strategies to maintain the frequency at quasi-constant values and to use it as power converter such as a simple dimmer to control the reactive power. The frequency analysis in steady state and transient cases is studied in this chapter using a single-phase equivalent circuit as well as theoretical and numerical results are also validated on a laboratory test bench. Chapter 14 describes communications applied for electric power systems including communication standards and infrastructure requirements for smart grids. The chapter presents three primary functions of smart grids to accomplish in real time requests of both consumers and suppliers based on communications technologies. The most usual communication systems including fiber optic communication, digital subscriber line/loop, power line communications, and wireless technologies for using the power system control for smart grids architecture are highlighted in the chapter. The case studies related to communication systems of electric power system are also carried out in this chapter. The SCADA systems and applications in electric power networks are studied in Chap. 15. The chapter explains the role and theory of SADA systems, security, real-time control and data exchange between remote units and central units. The SCADA systems are also applied for optimization and realization of reactive power in AC power systems. Some disadvantages of dispatching systems such as graphical information and interface are explained in the chapter and the rules of improving them are also carried out. The flexibility designing of the systems for small and large networks are also explained. Chapter 16 introduces the effect of geomagnetic fields called as storms on electric power systems. This chapter discusses about the physical nature of earth’s magnetic field and its measurements in geomagnetic observatories and shows that the variation of geomagnetic field affect the operation of various distracting electronic devices, such as electrical transmission systems. An algorithm for calculating
xiv
Preface
induced currents in the power transmission lines and also the violation of stability of the system considering the illustrative example are also derived in this chapter. The editors recommend this book as suitable for an audience professional in electric power systems, as well as researchers and developers in the field of energy and power engineering. It is anticipated that the readers have sufficient knowledge in electric power engineering and also advanced mathematical background. In total, the book includes theoretical background and case studies in reactive electric power and voltage stability concepts. The editors have made efforts to cover the essential topics of reactive electric power to balance theoretical and applicative aspects in the chapters of this book. The book has been written by a team of researchers from which use the dedicated intensive resources for achieving certain mental attitudes for interested readers. At the same time, the application and case studies are intended for real understanding and operation. Finally, the editors hope that this book will be useful to undergraduate and graduate students, researchers and engineers, trying to solve reactive electric power problems using modern technical and intelligent systems based on theoretical aspects and application case studies. Tabriz, Iran Zanjan, Iran Piteşti, Romania Aalborg, Denmark
Naser Mahdavi Tabatabaei Ali Jafari Aghbolaghi Nicu Bizon Frede Blaabjerg
Contents
Part I
Fundamentals of Reactive Power in AC Power Systems
1
Electrical Power Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Horia Andrei, Paul Cristian Andrei, Luminita M. Constantinescu, Robert Beloiu, Emil Cazacu and Marilena Stanculescu
3
2
Fundamentals of Reactive Power in AC Power Systems . . . . . . . . . Horia Andrei, Paul Cristian Andrei, Emil Cazacu and Marilena Stanculescu
49
3
Reactive Power Role and Its Controllability in AC Power Transmission Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 Esmaeil Ebrahimzadeh and Frede Blaabjerg
4
Reactive Power Compensation in Energy Transmission Systems with Sinusoidal and Nonsinusoidal Currents . . . . . . . . . . . 137 Milan Stork and Daniel Mayer
5
Reactive Power Control in Wind Power Plants . . . . . . . . . . . . . . . . . 191 Reza Effatnejad, Mahdi Akhlaghi, Hamed Aliyari, Hamed Modir Zareh and Mohammad Effatnejad
6
Reactive Power Control and Voltage Stability in Power Systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227 Mariana Iorgulescu and Doru Ursu
Part II 7
Compensation and Reactive Power Optimization in AC Power Systems
Optimal Reactive Power Control to Improve Stability of Voltage in Power Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251 Ali Ghasemi Marzbali, Milad Gheydi, Hossein Samadyar, Ruhollah Hoseyni Fashami, Mohammad Eslami and Mohammad Javad Golkar
xv
xvi
Contents
8
Reactive Power Compensation in AC Power Systems . . . . . . . . . . . 275 Ersan Kabalci
9
Optimal Placement of Reactive Power Compensators in AC Power Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317 Hossein Shayeghi and Yashar Hashemi
10 Reactive Power Optimization in AC Power Systems . . . . . . . . . . . . 345 Ali Jafari Aghbolaghi, Naser Mahdavi Tabatabaei, Narges Sadat Boushehri and Farid Hojjati Parast 11 Reactive Power Optimization Using MATLAB and DIgSILENT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 411 Naser Mahdavi Tabatabaei, Ali Jafari Aghbolaghi, Narges Sadat Boushehri and Farid Hojjati Parast 12 Multi-objective Optimal Reactive Power Dispatch Considering Uncertainties in the Wind Integrated Power Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 475 Seyed Masoud Mohseni-Bonab, Abbas Rabiee and Behnam Mohammadi-Ivatloo Part III
Challenges, Solutions and Applications in AC Power Systems
13 Self-excited Induction Generator in Remote Site. . . . . . . . . . . . . . . . 517 Ezzeddine Touti, Remus Pusca, J. Francois Brudny and Abdelkader Chaari 14 Communications for Electric Power System . . . . . . . . . . . . . . . . . . . 547 Maaruf Ali and Nicu Bizon 15 SCADA Applications for Electric Power System. . . . . . . . . . . . . . . . 561 Florentina Magda Enescu and Nicu Bizon 16 Effect of Geomagnetic Storms on Electric Networks. . . . . . . . . . . . . 611 Daniel Mayer and Milan Stork Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 631
List of Figures
Figure 1.1
Figure 1.2
Figure 1.3
Figure Figure Figure Figure Figure Figure Figure
1.4 1.5 1.6 1.7 1.8 1.9 1.10
Figure 1.11 Figure 1.12 Figure Figure Figure Figure Figure Figure Figure
1.13 1.14 1.15 1.16 1.17 1.18 1.19
Figure 1.20
Representation of three-phase symmetrical and positive phase-sequence system: a Time domain, b Cartesian coordinates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Representation of three-phase symmetrical and negative phase-sequence system: a Time domain, b Cartesian coordinates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Representation of three-phase symmetrical and zero phase-sequence system: a Time domain, b Cartesian coordinates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Star connection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Delta connection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Symmetrical and balanced three-phase system . . . . . . . . . Equivalent three-phase circuit . . . . . . . . . . . . . . . . . . . . . . Equivalence delta—star . . . . . . . . . . . . . . . . . . . . . . . . . . . Single-phase circuit of phase 1 . . . . . . . . . . . . . . . . . . . . . Decomposition of an unsymmetrical system in three symmetrical systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Decomposition of the a unsymmetrical system, b positive, c negative, and d zero-sequences. . . . . . . . . . . Single phases of a positive, b negative, and c zero-sequence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Voltage measurement in mono-phase AC circuits . . . . . . . Phase and line voltage measurements . . . . . . . . . . . . . . . . Current measurement in mono-phase circuit . . . . . . . . . . . Current measurement in three-phase circuit . . . . . . . . . . . . Current measurement for a three-phase balanced circuit . . Active power measurement in mono-phase circuits . . . . . . Active power measurement in three-phase circuits with neutral line load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Power measurement in three-phase circuits without neutral line load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
..
7
..
8
. . . . . . .
. . . . . . .
9 9 11 14 15 15 16
..
18
..
19
. . . . . . .
. . . . . . .
20 22 22 23 23 23 24
..
24
..
25 xvii
xviii
List of Figures
Figure 1.21 Figure 1.22 Figure 1.23 Figure 1.24
Figure Figure Figure Figure
1.25 1.26 1.27 1.28
Figure 1.29 Figure 2.1 Figure 2.2 Figure 2.3 Figure 2.4 Figure 2.5 Figure 2.6 Figure Figure Figure Figure Figure Figure Figure Figure Figure
2.7 2.8 2.9 2.10 2.11 2.12 2.13 2.14 2.15
Figure 2.16 Figure 2.17
Figure 2.18
Power measurement in three-phase circuits without neutral line load using two wattmeters . . . . . . . . . . . . . . . Power measurement for balanced load and symmetrical source voltage with natural neutral point. . . . . . . . . . . . . . Power measurement for balanced load and symmetrical source voltage with an artificial neutral point . . . . . . . . . . Exemplification of a voltage dip and a short supply interruption, classified according to EN 50160; Un—nominal voltage of the supply system (rms), UA—amplitude of the supply voltage, U (rms)—the actual rms value of the supply voltage [3] . . . . . . . . . . . . a Power Q-meter, b Measurement stand . . . . . . . . . . . . . . Wiring diagram for U, I, f measurements . . . . . . . . . . . . . Significance of phase-shifts u and w. . . . . . . . . . . . . . . . . Fourier series decomposition of a periodic distorted voltage signal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Current signals, harmonics and various PQ parameters measured for a low-voltage industrial load . . . . . . . . . . . . Signals and values characteristic for a sinusoidal variation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The phase-shift between two sinusoidal signals . . . . . . . . Complex representation of sinusoidal and complex signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Passive circuit’s elements . . . . . . . . . . . . . . . . . . . . . . . . . Ideal current ad voltage generators . . . . . . . . . . . . . . . . . . Passive linear two-port system magnetically not coupled to the exterior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Imittances’ triangles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Powers’ triangle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . AC circuit with 6 branches . . . . . . . . . . . . . . . . . . . . . . . . The equivalent AC circuit . . . . . . . . . . . . . . . . . . . . . . . . . AC circuit with two energy sources . . . . . . . . . . . . . . . . . Superposition principle . . . . . . . . . . . . . . . . . . . . . . . . . . . A simple installation for an AC electric drive . . . . . . . . . . The electric equipment in the mechanical work-place . . . . Convention of the directions for defining the power at terminals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The current and voltage signals of the instantaneous power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Power factor variation function of asynchronous motor’s loading and the reactive power consumption for small and big power motors function of the relative power voltage . . . . . . . . . . . . . . . . . . . . . . . . . . . . Examples for placing the capacitors bank . . . . . . . . . . . . .
..
25
..
26
..
26
. . . .
. . . .
29 32 33 37
..
39
..
45
.. ..
51 52
.. .. ..
53 54 54
. . . . . . . . .
. . . . . . . . .
55 56 60 62 63 65 67 68 70
..
74
..
74
.. ..
83 86
List of Figures
Figure 2.19 Figure 2.20 Figure 2.21 Figure 2.22 Figure 2.23 Figure 2.24 Figure 2.25 Figure 2.26 Figure 2.27 Figure Figure Figure Figure Figure Figure
2.28 2.29 2.30 2.31 2.32 2.33
Figure 2.34 Figure 3.1 Figure 3.2
Figure 3.3
Figure Figure Figure Figure Figure Figure
3.4 3.5 3.6 3.7 3.8 3.9
xix
Individual compensation for asynchronous motors and for transformers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Connections possibilities for capacitors bank . . . . . . . . . . . . The principle scheme for an automated controlled power factor in an installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Connecting the power factor correction system in a non-sinusoidal state system . . . . . . . . . . . . . . . . . . . . . . . . . . Selecting the reactive power compensation possibility function of installation’s nonlinear receivers’ weight . . . . . . Selection of reactive power compensation functions of installation’s nonlinear loads’ weight . . . . . . . . . . . . . . . . . . Connecting the detuned reactors in a D connection capacitors bank . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The energy quality parameters of the industrial consumer under investigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The current variation on the consumer’s most loaded phase taken on a monitor interval . . . . . . . . . . . . . . . . . . . . . Representation of an AC circuit branch . . . . . . . . . . . . . . . . AC circuit under non-sinusoidal conditions . . . . . . . . . . . . . . Resistive-inductive AC circuit . . . . . . . . . . . . . . . . . . . . . . . . Three-phase circuit with star connection . . . . . . . . . . . . . . . . The “splitter” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Multisim analysis of power absorbed by, a R01 in DC regime, b R001 in AC regime, c R002 in DC regime . . . . . . . . . . AC circuit with CEAPP . . . . . . . . . . . . . . . . . . . . . . . . . . . . Transmission line connecting two buses (i, j) presented by a PI equivalent model . . . . . . . . . . . . . . . . . . . . . . . . . . . Parallel compensators for reactive power control, a Thyristor-Controlled Reactors, b Thyristor-Switched Reactor, c Thyristor-Switched Capacitor, d Fixed Capacitor Thyristor-Controlled Reactor, e Thyristor-Switched Capacitor-Thyristor-Controlled Reactor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Series compensators for reactive power control, a Thyristor-Switched Series Capacitor (TSSC), b Thyristor-Controlled Series Capacitor (TCSC), c Thyristor-Controlled Series Reactor (TCSR) . . . . . . . . . . . STATic synchronous COMpensator (STATCOM) . . . . . . . . Static Synchronous Series Compensator (SSSC) . . . . . . . . . . Unified Power Flow Controller (UPFC) . . . . . . . . . . . . . . . . Interline Power Flow Controller (IPFC) . . . . . . . . . . . . . . . . Model of a lossless power transmission system . . . . . . . . . . Simplified model of a compensated transmission line by a shunt-connected capacitor . . . . . . . . . . . . . . . . . . . . . . . . . . .
86 88 89 90 91 92 92 94 95 97 101 104 106 108 110 110 118
124
126 127 127 128 128 129 130
xx
Figure 3.10 Figure 3.11 Figure 3.12 Figure 3.13 Figure 3.14 Figure 3.15
Figure 4.1 Figure 4.2 Figure 4.3 Figure 4.4 Figure 4.5 Figure 4.6 Figure 4.7
Figure 4.8 Figure 4.9 Figure 4.10
Figure 4.11
Figure 4.12
List of Figures
Shunt compensation based on the power electronic converters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V-I characteristic of the shunt compensator . . . . . . . . . . . . Simplified model of a series compensated transmission line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Series compensation based on the power electronic converter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Voltage control block diagram in dq reference frame . . . . Voltage control based on phasor estimation by series compensator, a vector diagram of the voltages and current during compensation, b block diagram of the control scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . Principle of typical reactive power compensation . . . . . . . Example of RP compensation for linear RLC load and nonharmonic source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Current iS versus value of compensation capacitor CCO (Example 4.1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Current iS versus value of compensation inductor LC (Example 4.1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Example of RP compensation for linear RL load and nonharmonic source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Current iS versus value of compensation capacitor CCO (Example 4.2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Time evolution of voltage (top), current from source (middle) and effective current from source for uncompensated (time 300 kV
The nominal voltage means the rms value of a line voltage between phases; its standardized values are recommended by the International Electrotechnical Commission (CEI). For example the standardization values in kV are: 3; 3.3; 6; 10; 11; 15; 20; 22; 30; 45; 47; 66; 69; 110; 115; 132; 138; 150; 161; 220; 230; 287; 330; 345; 380; 400; 500; 700, and each country can adopt specific several such values. The LV is used in indoor electrical installations to supply directly the low voltage customers as well as the small urban and industrial networks, with power up to tens of kVA. The MV is used in urban and industrial networks for supply transformers with powers between tens and hundreds of kVA and also can supply directly medium voltage equipment. Transmission and distribution lines for powers between tens of MVA and 1–2 hundreds of MVA are carried out with HV. The VHV is used for transmission lines of powers between hundreds and thousands of MVA. – criteria of functions classify EPS in usage (utilization), distribution and transmission networks. – criteria of topology classify EPS in: (i) radial networks that means each customer can be supplied from one side (source) only; (ii) meshed networks that means each customer can be supplied from two sides; (iii) complex meshed networks that means each customer can be supplied from of more than the two sides. – criteria of adopted power system classify EPS in AC respectively DC. Starting from historical scientific insights of M. Dolivo-Dobrovolski and N. Tesla now the widespread EPS system consists in 3 and 2-phases electrical AC networks. But from struggle between AC and DC transmission systems is possible in the near future that very high voltage DC systems to win.
1.3
Basic Concepts and Analysis Methods of Electrical Power Systems
From theoretical point of view EPS are considering as symmetrical three-phase systems, condition that is ensured through the symmetrization of transmissions and distribution lines, and transformers. Starting from this ideal assumption the symmetrical three-phase transmission and distribution networks, including equipment and transformers, can be analyzed by using symmetrical components method and the decomposition in single phase circuits. The equivalent circuits of EPS contain
1 Electrical Power Systems
7
non-linear passive elements, as resistances, inductances and capacitances. Nonlinearity of these elements are neglected in most frequent calculations, considering that their values change relatively small according to the low and expected limits of change of the voltage, current or frequency. For this reason the equivalent circuits of EPS are considered linear and are disposed longitudinal and transversal as in C, T or P scheme types [11–13]. A three-phase symmetrical and positive phase-sequence system in instantaneous values is expressed as [14, 15] pffiffiffi vd1 ðtÞ ¼ V 2 sinðxt þ cÞ pffiffiffi 2p vd2 ðtÞ ¼ V 2 sinðxt þ c Þ 3 pffiffiffi pffiffiffi 4p 2p vd3 ðtÞ ¼ V 2 sinðxt þ c Þ ¼ V 2 sinðxt þ c þ Þ 3 3
ð1:1Þ
respectively by the complex vectors
V d1 ¼ Vejc ¼ V 2p
2p
V d2 ¼ Vejðc 3 Þ ¼ Vej 3 ¼ a2 V V d3 ¼ Ve
jðc þ 2p 3Þ
¼ Ve
j2p 3
ð1:2Þ
¼ aV
pffiffi 2p 4p where a ¼ ej 3 ¼ ej 3 ¼ 12 þ j 23 is called phase operator. The representation of instantaneous values (1.1) in time domain and of complex vectors (1.2) in complex Cartesian coordinates are presented in Fig. 1.1a, b.
Fig. 1.1 Representation of three-phase symmetrical and positive phase-sequence system: a Time domain, b Cartesian coordinates
8
H. Andrei et al.
A three-phase symmetrical and negative phase-sequence system in instantaneous values is expressed as pffiffiffi vi1 ðtÞ ¼ V 2 sinðxt þ cÞ pffiffiffi 2p vi2 ðtÞ ¼ V 2 sinðxt þ c þ Þ 3 pffiffiffi 2p vi3 ðtÞ ¼ V 2 sinðxt þ c Þ 3
ð1:3Þ
respectively by the complex vectors
V i1 ¼ Vejc ¼ V 2p
2p
V i2 ¼ Vejðc þ 3 Þ ¼ Vej 3 ¼ aV V i3 ¼ Ve
jðc2p 3Þ
¼ Ve
j2p 3
2
ð1:4Þ
¼a V
which have the representations in time domain and complex Cartesian coordinates illustrated in Fig. 1.2a, b. A three-phase symmetrical and zero phase-sequence system in instantaneous values is expressed as pffiffiffi v01 ðtÞ ¼ v02 ðtÞ ¼ v03 ðtÞ ¼ V 2 sinðxt þ cÞ
ð1:5Þ
V o1 ¼ V o2 ¼ V o3 ¼ Vejc ¼ V
ð1:6Þ
and by the complex vectors
The representation of values (1.5) and (1.6) is shown in Fig. 1.3a, b.
Fig. 1.2 Representation of three-phase symmetrical and negative phase-sequence system: a Time domain, b Cartesian coordinates
1 Electrical Power Systems
9
Fig. 1.3 Representation of three-phase symmetrical and zero phase-sequence system: a Time domain, b Cartesian coordinates
Fig. 1.4 Star connection
Practically, two general connections are considered in EPS: star and delta [16–18]. (i) Star connection. Let us consider an elementary EPS in which the generator (G) and consumer (load, receptor—R) are arranged into a star connection. The transmission line (L) links G with R. The star connection illustrated in Fig. 1.4 is defined in that all the phase terminals are connected together to form the neutral point. The both neutral nodes of the generator “O” and of the receptor “N” is linked by neutral wire. The electromotive voltages of generators e1, e2, e3, the voltages across the terminals of generators u1G ; u2G ; u3G , the voltages between the transmission lines u12, u23, u31, the currents in the transmission lines i1, i2, i3 and the voltages across the terminals of receptor u1R ; u2R ; u3R there are three-phase systems. For an adequate systematization of knowledge following definitions are useful: • the triplets ðe1 ; u1G ; i1G Þ, ðe2 ; u2G ; i2G Þ and ðe3 ; u3G ; i3G Þ are called phase signals of generators;
10
H. Andrei et al.
• the doublets ðu1R ; i1R Þ, ðu2R ; i2R Þ and ðu3R ; i3R Þ are called phase signals of receptors (consumers); • the voltages (u12, u23, u31) between of the three line wires are called line voltages, also the line impedances have not been taken into account; • the currents (i1, i2, i3) across the line wires are called line currents; • the impedances Z1, Z2, Z3 are called phase impedances of receptor. Generally speaking the set signals ðe1 ; u1G ; i1G ; u1R ; i1R ; Z 1 Þ makes up phase 1 of star connection and similar definitions are used for the other two phases. For star connection the line currents are equal to the phase currents i1G ¼ i1 ¼ i1R ; i2G ¼ i2 ¼ i2R ; i3G ¼ i3 ¼ i3R
ð1:7Þ
respectively in complex values I 1G ¼ I 1 ¼ I 1R ; I 2G ¼ I 2 ¼ I 2R ; I 3G ¼ I 3 ¼ I 3R
ð1:8Þ
and in rms values Il = Iph. Also for the neutral wire are defined: i0 or in—the current, uNO or un—the voltage, and Z0 or ZN—the impedance. Considering the positive reference directions for line currents from generators to receptors and for current across the neutral wire from neutral point of the receptors to that of the generators, then by applying Kirchhoff laws following relations in complex values are described the star connection I1 þ I2 þ I3 ¼ I0
U 12 ¼ U 1R U 2R U 23 ¼ U 2R U 3R U 31 ¼ U 3R U 1R
U 12 þ U 23 þ U 31 ¼ 0
ð1:9Þ
U 1R þ U NO þ U 10 ¼ 0 U 2R þ U NO þ U 20 ¼ 0 U 3R þ U NO þ U 30 ¼ 0 Another set of relations express Ohm’s law for receptor phases and for neutral wire in complex values are expressed as U 1R ¼ Z 1 I 1 U 2R ¼ Z 2 I 2 U 3R ¼ Z 3 I 3
U NO ¼ Z 0 I 0
ð1:10Þ
1 Electrical Power Systems
11
Fig. 1.5 Delta connection
(ii) Delta connection. Let us consider an elementary EPS in which the generator and receptor (consumer) are arranged into a delta (mesh) connection (Fig. 1.5). The transmission line links generator with receptor. Symmetrical cycling between the pairs of impedances leads so that the start terminal of one phase impedance is connected to the finish terminal of another are used for delta connection. For an adequate systematization of knowledge following definitions are useful: • the triplets ðe12 ; u21G ; i21G Þ, ðe23 ; u23G ; i23G Þ and ðe31 ; u31G ; i31G Þ are called phase signals of generators; • the doublets ðu12R ; i12R Þ, ðu23R ; i23R Þ and ðu31R ; i31R Þ are called phase signals of receptors (consumers); • the voltages (u12, u23, u31) between of the three line wires are called line voltages, also the line impedances have not been taken into account; • the currents (i1, i2, i3) across the line wires are called line currents; • the impedances Z12, Z23, Z31 are called phase impedances of receptor. Generally speaking the set signals ðe12 ; u12G ; i12G ; u12R ; i12R ; Z 12 Þ makes up phase 1 of delta connection and similar definitions are used for the other two phases. If the line impedances are not considering, for star connection the line voltages of generators are equal to the line voltage of receptor u12G ¼ u12R ¼ u12 ; u23G ¼ u23R ¼ u23 ; u31G ¼ u31R ¼ u31
ð1:11Þ
and in rms values Vl = Vph. Based on Kirchhoff’s current law (KCL) the following relations are true I 1 ¼ I 12 I 31 I 2 ¼ I 23 I 12 I 3 ¼ I 31 I 23
ð1:12Þ
12
H. Andrei et al.
By summing (1.12) results I1 þ I2 þ I3 ¼ 0
ð1:13Þ
Also applying Ohm’s law on each phase of the receptor are obtained U 12 ¼ Z 12 I 12 U 23 ¼ Z 23 I 23 U 31 ¼ Z 31 I 31
ð1:14Þ
A three-phase system is called symmetrical if it is supplied by a symmetrical voltages system (1.1), (1.3) or (1.5). Otherwise it is considered unsymmetrical. If all the complex phase impedances of star or delta connection are equal the system is balanced, otherwise is non-balanced. A symmetrical and balanced system produces symmetrical lines and phases currents systems. In symmetrical and balanced systems all neutral points are the same potential and the current across the neutral wire is null. On the other hand in symmetrical and balanced conditions for star conpffiffiffi nection the rms value of line voltage is 3 times the rms value of phase voltage pffiffiffi pffiffiffi Vl ¼ 3Vph , respectively for delta connection the rms value of line current is 3 pffiffiffi times the rms value of phase current Il ¼ 3Iph . The overall complex powers absorbed by a receptor in star (including the neutral wire) and delta connections are given by SR;star ¼ U 1R I 1 þ U 2R I 2 þ U 3R I 3 þ U NO I 0 ¼ Z 1 I12 þ Z 2 I22 þ Z 3 I32 þ Z 0 I02 ¼ PR;star þ jQR;star
ð1:15Þ
respectively 2 2 2 SR;delta ¼ U 12R I 12 þ U 23R I 23 þ U 31R I 31 ¼ Z 12 I12 þ Z 23 I23 þ Z 31 I31 þ Z 0 I02
¼ PR;delta þ jQR;delta
ð1:16Þ
where the real part represents the overall active absorbed power (PR—dissipated in the resistors) and the imaginary part represents the overall absorbed reactive power (QR—dissipated in the inductors minus in the capacitors). When symmetrical and balanced conditions are verified then for any form of interlinkage (star or delta) the overall active and reactive absorbed powers are given by pffiffiffi 3Vl Il cos u pffiffiffi QR ¼ 3Vph Iph sin u ¼ 3Vl Il sin u
PR ¼ 3Vph Iph cos u ¼
ð1:17Þ
1 Electrical Power Systems
13
where the angle u is the phase displacement between the voltage and current phases, or in other word, is the phase angle of the complex phase impedance. Also the apparent (overall) power for star and delta connections is defined as SR;star ¼
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi P2R;star þ Q2R;star ;
SR;delta ¼
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi P2R;delta þ Q2R;delta
ð1:18Þ
According to the Eq. (1.17) in symmetrical and balanced conditions for both connections the following expression is true SR ¼ 3Vph Iph ¼
pffiffiffi 3V l I l
ð1:19Þ
The calculation methods of EPS are based on the properties of three-phase circuits with different types of supplying system’s voltages and on some properties of various phase connections [19, 20]. Two elementary methods are presented below. (a) Calculation of symmetrical and balanced three-phase circuits. In this case we assume that the voltage generators are symmetrical and know, and also the receptor is balanced. Symmetrical three-phase circuits consisting of many star or delta receptors are solved by the use of single-phase circuit, for example of phase 1. Based on the principles of this elementary method have been developed software programs for the analysis of three-phase circuits. This method contains following steps [21, 22]: (a1) The coupled inductances in phase 1 are replaced with equivalent inductance −Zm. Based on three-phase symmetrical currents property I1 + I2 + I3 = 0, then the voltage across the coupled inductance in phase 1 can be expressed as Um,1 = ZmI2 + ZmI3 = Zm(I2 + I3) = −ZmI1, where the complex mutual impedance is Zm = jxM, and M is the mutual inductance which can be considered in aiding (+) or opposite (−) direction. Analogous equivalences are applied to couple inductances of other two phases. (a2) All the delta connections are replaced with equivalent star connections, by using the relation Z Y ¼ Z D =3, where ZY and ZD are the phase impedances of star respectively delta connections. (a3) In a symmetrical and balanced three-phase star circuits all neutral points have the same potential and I1 + I2 + I3 = IN = 0. Therefore all neutral points can be connected each other through a null resistance wire without modifying the three-phase circuit voltages and currents. Such the single-phase circuit becomes a closed-loop. (a4) The voltages and currents of phase 1 are calculated by using the Kirchhoff’s laws in single-phase circuit. Finally on determine the other two phases voltages and currents through the use of phase operator a, according to the type of phase-sequences of the voltages generator system. (a5) By using Eqs. (1.17) and (1.19) the overall active, reactive, apparent and complex absorbed power are calculated.
14
H. Andrei et al.
Fig. 1.6 Symmetrical and balanced three-phase system
Example 1.1 Let us consider a symmetrical three-phase generator in star connection shown in Fig. 1.6. Its positive phase-sequence voltages system are pffiffiffi pffiffiffi 2p pffiffiffi 2p E 1 ¼ 400 2, E 2 ¼ 400 2ej 3 ; E 3 ¼ 400 2e þ j 3 supplying through a transmission line impedance Z1 = 40 + 80j first star connection receptor R1 = 40 X, XL1 = XC1 = XM1 = 40 X, neutral wire impedance ZON1 = 10(1−j) X, and through another transmission line impedance r1 = 10 X, the second delta connection receptor R2 = 90 X, XL2 = 150 X and XM2 = 30 X. In order to calculate the generator line currents (I11, I21, I31), the phase currents (I1Y, I2Y, I3Y) and voltages (U1Y, U2Y, U3Y) of star receptor, the phase currents (I12D, I23D, I31D) and voltages (U12D, U23D, U31D) of delta receptor and the overall absorbed powers, the above mentioned elementary method is applied: (a1) The equivalences of coupled inductances of star and delta connections are presented in Fig. 1.7; (a2) Replacement of delta connection with equivalent star connection whose neutral point is N2 is shown in Fig. 1.8. There the values of equivalent impedances are Z 1Y ¼ R1 þ jðXL1 XM1 XC1 Þ ¼ 40ð1 jÞX Z R2 þ jðXL2 XM2 Þ 90 þ 120j ¼ ¼ 30 þ 40jX Z 2Ye ¼ 2D ¼ 3 3 3
1 Electrical Power Systems
15
Fig. 1.7 Equivalent three-phase circuit
Fig. 1.8 Equivalence delta—star
(a3) A null resistance wire is introduced in order to connect all the neutral points O, N and N1 (dash line in Fig. 1.8); (a4) The single-phase circuit of phase 1 is shown in Fig. 1.9. By using Kirchhoff’s laws one obtains
16
H. Andrei et al.
Fig. 1.9 Single-phase circuit of phase 1
I 11 ¼
E1 Z1 þ
Z 1Y ðr1 þ Z 2Ye Þ Z 1Y þ r1 þ Z 2Ye
¼ 5e
jp4
A;
I 1Y
pffiffiffi re þ Z 1Ye 5 2 A ¼ ¼ I 11 2 Z 1Y þ re þ Z 2Ye
The currents of other two phases of first star receptor are obtained by using the phase operator I 2Y ¼ I 1Y e
j2p 3
pffiffiffi 5 2 j7p e 6 A; ¼ 2
I 3Y ¼ I 1Y e
j2p 3
pffiffiffi 5 2 jp e6 A ¼ 2
From KCL results the line 1 current of the second equivalent star pffiffi p ¼ I 1Ye ¼ 5 2 2 ej2 A. Then the other two phase currents are calculated as pffiffi pffiffi 2p 7p 2p p ¼ I 2Ye ¼ I 011 ej 3 ¼ 5 2 2 ej 6 A, I 031 ¼ I 3Ye ¼ I 011 ej 3 ¼ 5 2 2 ej6 A. If the properties of symmetrical systems are used thus the delta connection receptor is crossed by the phase currents
I 011 I 021
I 12D I 31D
pffiffiffi p 5 pffiffiffi j2p 6e 3 A; ¼ I 1Ye 3ej6 ¼ 2pffiffiffi pffiffiffi p 5 6 A ¼ I 3Ye 3ej6 ¼ 2
I 23D
pffiffiffi pffiffiffi jp 5 6 j4p e 3 A; ¼ I 2Ye 3e 6 ¼ 2
First star receptor has the phase voltages p
U 1Y ¼ Z 1Y I 1Y ¼ 200ej4 V; 2p
5p
U 3Y ¼ U 1Y ej 3 ¼ 200ej12 V
2p
11p
U 2Y ¼ U 1Y ej 3 ¼ 200ej 12 V;
1 Electrical Power Systems
17
and also the phase voltages of delta receptor are given by pffiffiffi 2p 4 U 12D ¼ Z 1D I 12D ¼ 375 6ejð 3 þ arctan3Þ V;
pffiffiffi 4p 4 2p U 23D ¼ U 12D ej 3 ¼ 375 6ejð 3 þ arctan3Þ
V pffiffiffi 4 2p U 31D ¼ U 12D ej 3 ¼ 375 6ejðarctan3Þ V
(a5) The overall absorbed active, reactive, apparent and complex powers by receptors and transmission lines are respectively Pabs ¼ 3Re½Z e I 211 þ 3Re Z 1y I 21y þ 3re I 211 þ 3Re½Z 2D I 212D ¼ 6000 W Qabs ¼ 3Im½Z 1 I 211 þ 3Im Z 1y I 21y þ 3Im½Z 2D I 212D ¼ 6000 VAr qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffi 2 Sabs ¼ 3jZ j1 I 211 þ 3jZ j1y I 21y þ 3re I 0 11 þ 3jZ j12D I 212D ¼ P2abs þ Q2abs ¼ 6000 2 VA Sabs ¼ P þ jQ ¼ 6000 þ 6000j
These powers are received from the generator whose overall complex power is Sgen ¼ E 1 I 11 þ E 2 I 21 þ E 3 I 31 ¼ 3E1 I 11 ¼ 6000 þ 6000j hence the conservation of active and reactive powers is proved h i h i Pgen ¼ Re Sgen ¼ 6000 W Pabs ; Qgen ¼ Im Sgen ¼ 6000 VAr Qabs Due to the losses in transmission lines, the active power transmission efficiency can be calculated as g¼
3Re½Z 1Y :I 21Y þ 3Re½Z 2D :I 212D ¼ 0:68 Pabs
Also the power factor (PF, also k or cos u) of the considered EPS is defined as k ¼ cos u ¼
P ¼ 0:706 S
(b) Calculation of asymmetrical three-phase circuits by using the method of symmetrical components. In EPS two kinds of asymmetry transverse and longitudinal may occur. Transverse asymmetry occurs when an unbalanced receptor is connected to a symmetrical three-phase network. Such an unbalanced load may take the form of asymmetrical short-circuits as line-to-line, one or two line-to-earth short-circuits. Longitudinal asymmetry occurs when the phases of a transmission line contain un-equal impedances (unsymmetrical section of transmission line) or when an open-circuit occurs in one or two phases. By the elementary method of symmetrical components an asymmetrical three-phase set of currents or voltages can be decomposed into three symmetrical systems positive, negative and zero
18
H. Andrei et al.
Fig. 1.10 Decomposition of an unsymmetrical system in three symmetrical systems
phase-sequence which are called symmetrical components [23]. Since it is based on superposition theorem the decomposition can be applied only to linear circuits. It is demonstrates that any asymmetrical three-phase system V 1 ; V 2 ; V 3 can be explained as the sum of three symmetrical systems: positive, negative and zero phase-sequences [24, 25], so V 1 ¼ V o1 þ V d1 þ V i1
ð1:20Þ
V 2 ¼ V o2 þ V d2 þ V i2 V 3 ¼ V o3 þ V d3 þ V i3 where 2p
2p
V d1 ¼ V d ; V d2 ¼ V d ej 3 ¼ a2 V d ; V d1 ¼ V d ej 3 ¼ aV d 2p
2p
V i1 ¼ V i ; V i2 ¼ V i ej 3 ¼ a V i ; V i1 ¼ V i ej 3 ¼ a2 V i V o1 ¼ V o2 ¼ V o3 ¼ V o Taking into account the relations (1.21)–(1.23), (1.21)—shown in Fig. 1.10—can be rewritten in the form V1 ¼ Vo þ Vd þ Vi
V 2 ¼ V o þ a2 V d þ a V i 2
V3 ¼ Vo þ a Vd þ a Vi
ð1:21Þ ð1:22Þ ð1:23Þ
the
decomposition
ð1:24Þ
If it is considered know the asymmetrical system V 1 ; V 2 ; V 3 , then zero ðV o Þ, positive ðV d Þ and negative-sequences ðV i Þ are calculated as
1 Electrical Power Systems
19
1 V o ¼ ðV 1 þ V 2 þ V 3 Þ 3 1 V d ¼ V 1 þ a V 2 þ a2 V 3 3 1 V i ¼ V 1 þ a2 V 2 þ a V 3 3
ð1:25Þ
In order to evaluate the state of asymmetry, are defined two dimensionless parameters: the coefficient of dissymmetry ed ¼ VVdi , and the coefficient of asym-
metry ea ¼ VVdo , where Vo , Vd , Vi are the rms values of zero, positive and negative-sequences. In real applications a system is considered symmetrical if ed and ea have values lower than 0.05. The calculation of three-phase systems operating under asymmetrical conditions can be made based on the superposition theorem: are calculated separately each symmetrical component, and finally these components gather. Example 1.2 Let us consider a balanced star receptor supplied by an asymmetrical system of phase voltages U 10 ; U 20 ; U 30 . The calculation of phase currents I 1 ; I 2 ; I 3 is made by method of symmetrical components. Asymmetrical system shown in Fig. 1.11a, is decomposed in three symmetrical systems positive (Fig. 1.11b), negative (Fig. 1.11c) and zero-sequences (Fig. 1.11d). Since the receiver is balanced, the current through the neutral wire is zero for positive and negative component, and for zero-sequence component is 3 times the
Fig. 1.11 Decomposition of the a unsymmetrical system, b positive, c negative, and d zero-sequences
20
H. Andrei et al.
Fig. 1.12 Single phases of a positive, b negative, and c zero-sequence
phase current because the phase currents are equal. The reducing of the system to a single phase is made by using the equivalent circuit shown in Fig. 1.12: single phase of positive (Fig. 1.12a), negative (Fig. 1.12a), and zero-sequence (Fig. 1.12c). Phase voltages corresponding to each circuit are respectively U d1 ¼ Z I d1 ;
U i1 ¼ Z I i1 ;
U o1 ¼ ðZ þ 3Z 0 Þ I o1
ð1:26Þ
The two other phase currents for positive and negative sequence are calculated by using the phase operator. Finally the phase currents of initial asymmetrical system are calculated by summing of symmetrical current components as U o1 U U þ d1 þ i1 Z þ 3Z 0 Z Z U o1 U U I 2 ¼ I o2 þ I d2 þ I i2 ¼ þ a2 d 1 þ a i 1 Z þ 3Z 0 Z Z U o1 U d1 U þ a2 i 1 þa I 3 ¼ I o3 þ I d3 þ I i3 ¼ Z þ 3Z 0 Z Z I 1 ¼ I o1 þ I d1 þ I i1 ¼
ð1:27Þ
If the initial asymmetrical system contains coupled inductances Zm then for positive and negative-sequence is used the equivalent circuit where the equivalent impedance is Z−Zm and the voltages are U d ¼ ðZ Z m Þ I d ; U i ¼ ðZ Z m Þ I i . Also for zero-sequence is obtained U o ¼ ðZ þ 2Z m Þ I o . The delta connections are replaced by equivalent star connections considering Z relation Z Y ¼ 3D . Equations (1.15) and (1.16) are used to calculate the overall P and Q powers.
1 Electrical Power Systems
21
In conclusion, the elementary method of symmetrical components comprises the following steps: (i) asymmetrical supplied voltage system is decomposed in three symmetrical components; (ii) after the equivalence of coupled inductances and of delta-star connections, are calculated the impedances of positive, negative and zero-sequences; (iii) symmetrical components of currents system are calculated by using single phase circuit and phase operator; (iv) asymmetrical currents of initial system are calculated by adding the symmetrical current components.
1.4
State of the Art: Measurement Methods of Power Systems Parameters
In EPS systems it is important to measure its electrical characteristics [26]. In this section are described and illustrated measurement methods of voltage, current, active power and frequency. Regardless of what parameter is measured, before the actual operation is done there are some precautions to be considered: • the level of the measured parameter • the scale of the measuring apparatus In order to have a safe measuring operation, the maximum value of the measured parameter should not be higher than the maximum indication of the measuring device. If this precaution is not carefully considered, it could lead to permanent damage of the measuring instrument.
1.4.1
Voltage Measurement
(i) Mono-phase circuits. In mono-phase circuits, the voltage measurement consists in connecting a voltmeter at the terminals of the voltage source [27] as indicated in Fig. 1.13. Voltmeter V indicates the rms value of the voltage source. (ii) Three-phase circuits contain two types of voltages, as are described in Sect. 1.3: line voltage—that is measured between two lines of the supply system, and phase voltage—that is measured between one line and the common point. In Fig. 1.14 is shown the measurement principle of line and phase voltages. There voltmeter V1 measures the line voltage between the phases U and V, while
22
H. Andrei et al.
Fig. 1.13 Voltage measurement in mono-phase AC circuits
Fig. 1.14 Phase and line voltage measurements
voltmeter V2 measures the phase voltage between the phases U and the common point.
1.4.2
Current Measurement
The current appears in an electrical circuit when it is closed on a load. If the circuit is open, so that no load is connected to the source’s terminals, there is no current flow. The current is measured by connecting an ammeter in series with the load [28]. (i) Mono-phase circuits. In mono-phase circuits the current in measured by connecting in series an ammeter with the load. In Fig. 1.15 is indicated the procedure to measure the rms value of current across the load Z. (ii) Three-phase circuits. In three-phase circuits the current is measured in the same way as in mono-phase circuits. The measuring of the current for each phase of the circuit there will be used three ammeters, one for each circuit phases. In Fig. 1.16 the ammeters A1, A2 and A3 measure the rms values of current through the loads Z1, Z2 and Z3, respectively. If the loads are identical on all the three phases, they constitute a balanced circuit, it is enough to connect in series only one ammeter on only one phase, as the other rms values of currents are equals. This situation is displayed in Fig. 1.17.
1 Electrical Power Systems
23
Fig. 1.15 Current measurement in mono-phase circuit
Fig. 1.16 Current measurement in three-phase circuit
Fig. 1.17 Current measurement for a three-phase balanced circuit
1.4.3
Active Power Measurement
(i) Active power measurement in mono-phase circuits can be done with the wattmeter. The wattmeter has two coils [29]: current and voltage. The voltage coil of the wattmeter is connected in parallel with the load, while the current coil is connected in series with the load, as displayed in Fig. 1.18. (ii) Active power measurement in three-phase circuits is quite similar regardless of load configuration as far as procedure [30]. This consists in connecting a wattmeter in the circuit with the current coil in series with the load and with
24
H. Andrei et al.
Fig. 1.18 Active power measurement in mono-phase circuits
Fig. 1.19 Active power measurement in three-phase circuits with neutral line load
the voltage coil in parallel with the load. The three-phase circuits have some particularities due to load configurations. If the load has a neutral line, such as star connected loads, then the power can be measured by using three wattmeters. The voltage coils of the wattmeters are connected between each phase and the neutral point of the load and voltage source [4]. The current coils are connected in series with the load on each phase, as displayed in Fig. 1.19. If the load does not have an accessible neutral point or is delta connected, in order to measure the power using three wattmeters, these are connected such as to construct an artificial neutral point for them. This situation is indicated in Fig. 1.20. Thus the active power absorbed by the load can be expressed by P ¼ P1 þ P2 þ P3
ð1:28Þ
If a neutral point is created, it is preferred that it has the same potential as one of the source phase as indicated in Fig. 1.21. In this situation the indication of the third wattmeter would be zero as its voltage coil would be connected at the same
1 Electrical Power Systems
25
Fig. 1.20 Power measurement in three-phase circuits without neutral line load
Fig. 1.21 Power measurement in three-phase circuits without neutral line load using two wattmeters
potential. Because of this, it is eliminated from the circuit as not being useful. The measured active power is the sum of the indications of each wattmeter, expressed as P ¼ P1 þ P2
ð1:29Þ
A different situation is the case of balanced loads and symmetrical voltages on each source phase. In this case it is enough to use one wattmeter as indicated in Fig. 1.22. In this situation, the total power is determined by multiplying the indication of the wattmeter by three, as in expressed in Eq. (1.30) P ¼ 3 P1
ð1:30Þ
The situation presented in Fig. 1.22 is valid if the load has an accessible neutral point. In case that the load does not have an accessible neutral point, or is delta connected, then it is created an artificial neutral point by using two additional
26
H. Andrei et al.
Fig. 1.22 Power measurement for balanced load and symmetrical source voltage with natural neutral point
Fig. 1.23 Power measurement for balanced load and symmetrical source voltage with an artificial neutral point
resistors as indicated in Fig. 1.23. The values of the additional resistors R1 and R2 are equal to the value of the resistance of the voltage circuit of the wattmeter.
1.5
State of the Art: Standards for Power Systems Parameters
Electrical energy needed to power industrial or household electric consumers, as all the products, should satisfy same quality demands. These demands, like good PQ ensure the electrical equipment is to operate without errors and this is a responsively of the supplier [17, 31]. On the other hand, an important part of the equipment in use today, in particular electronic and computer devices generates distortion of the voltage supply in the installation, because of its non-linear characteristics, i.e. it draws a non-sinusoidal
1 Electrical Power Systems
27
current with a sinusoidal supply voltage. In this case the PQ is a responsively of the electricity user. In consequence, maintaining satisfactory PQ is a joint responsibility for the supplier and the electricity user. Standard IEC 60038 distinguishes “two different voltages in electrical networks and installations” [32]: – supply voltage, which “is the line-to-line or line-to-neutral voltage at the point of common coupling (PCC), i.e. main supplying point of installation”; – utility voltage, which “is the line-to-line or line-to-neutral voltage at the plug or terminal of the electrical device”. The main document dealing with demands concerning the supplier’s side is standard EN 50160 [33] which characterize voltage parameters of electrical energy in public distribution systems. According with Standard EN 50160, “the main voltage characteristics of public distribution systems are”: Supply voltage—“the root mean square (rms) value of the voltage at a given moment at the PCC, measured over a given time interval”. Nominal voltage of the system (Un)—“the voltage by which a system is designated or identified and to which certain operating characteristics are referred”. Declared supply voltage (Uc)—“is normally the nominal voltage Un of the system. If, by agreement between the supplier and the user, a voltage different from the nominal voltage is applied to the terminal, then this voltage is the declared supply voltage Uc”. Normal operating condition—“the condition of meeting load demand, system switching and clearing faults by automatic system protection in the absence of exceptional conditions due to external influences or major events”. Voltage variation—“is an increase or decrease of voltage, due to variation of the total load of the distribution system or a part of it”. Flicker—“impression of unsteadiness of visual sensation induced by a light stimulus, the luminance or spectral distribution of which fluctuates with time”. Flicker severity—“intensity of flicker annoyance defined by the UIE-IEC flicker measuring method and evaluated by the following signals”: – Short term severity (Pst) “measured over a period of ten minutes”; – Long term severity (Plt) “calculated from a sequence of 12 Pst—values over a two-hour interval, according to the following expression”:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffi u 12 u X P3 3 sti Plt ¼ t 12 i¼1
ð1:31Þ
Supply voltage dip—“a sudden reduction of the supply voltage to a value between 90 and 1% of the declared voltage Uc, followed by a voltage recovery after
28
H. Andrei et al.
a short period of time”. Conventionally, the duration of a voltage dip is between “10 ms and 1 min”. The depth of a voltage dip is defined as the “difference between the minimum rms voltage during the voltage dip and the declared voltage”. Supply interruption—is a condition in which “the voltage at the supply terminals is lower than 1% of the declared voltage Uc”. A supply interruption is classified as: – “prearranged in order to allow the execution of scheduled works on the distribution system, when consumers are informed in advance”; – “accidental, caused by permanent (a long interruption) or transient (a short interruption) faults, mostly related to external events, equipment failures or interference”. Temporary power frequency over-voltages—“have relatively long duration, usually of a few power frequency periods”, and originate mainly from switching operations or faults, e.g. sudden load reduction, or disconnection of short circuits. Transient over-voltages—“are oscillatory or non-oscillatory, highly damped, short over-voltages with a duration of a few milliseconds or less, originating from lightning or some switching operations”, e.g. at switch-off of an inductive current. Harmonic voltage—“a sinusoidal voltage with a frequency equal to an integer multiple of the fundamental frequency of the supply voltage”. Harmonic voltages can be evaluated: – “individually by their relative amplitude Uh related to the fundamental voltage U1”, where h is the order of the harmonic; – “globally, usually by the total harmonic distortion factor THDu”, calculated using the following expression:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi , ffi u 40 uX THDu ¼ t ðUh Þ2 U1 h¼2
ð1:32Þ
Inter-harmonic voltage—“is a sinusoidal voltage with frequency between the harmonics”, i.e. the frequency is not an integer multiple of the fundamental. Voltage unbalance—“is a condition where rms value of the phase voltages or the phase angles between consecutive phases in a three-phase system is not equal”. Standard EN 50160 gives the main voltage parameters and their permissible deviation ranges at the customer’s PCC in public low voltage (LV) and medium voltage (MV) electricity distribution systems, under normal operating conditions (Fig. 1.24). In this circumstances, LV signify that the phase to phase nominal rms voltage does not exceed 1000 V and MV signify that the phase-to-phase nominal rms value is between 1 and 35 kV. The comparison of the EN 50160 requirements with those of the EMC standards EN 61000, listed in Table 1.2 show significant differences in various parameters [33–35].
1 Electrical Power Systems
29
Fig. 1.24 Exemplification of a voltage dip and a short supply interruption, classified according to EN 50160; Un—nominal voltage of the supply system (rms), UA—amplitude of the supply voltage, U (rms)—the actual rms value of the supply voltage [3]
Table 1.2 Comparison of supply voltage requirements according to EN 50160 and the EMC standards EN 61000 No
Parameter
Supply voltage characteristics according to EN 50160
Low voltage characteristics according to EMC standard EN 61000 EN 61000-2-2
1
Power frequency
2%
2
Voltage magnitude variations
3
Rapid voltage changes
LV, MV: mean value of fundamental measured over 10 s ±1% (49.5–50.5 Hz) for 99.5% of week –6%/+4% (47–52 Hz) for 100% of week LV, MV: ±10% for 95% of week, mean 10 min rms values (Fig. 1.1) LV: 5% normal 10% infrequently Plt 1 for 95% of week MV: 4% normal 6% infrequently Plt 1 for 95% of week
Other parts
±10% applied for 15 min
3% normal 8% infrequently Pst < 1.0 Plt < 0.8
3% normal 4% maximum Pst < 1.0 Plt < 0.65 (EN 61000-3-3) 3% (IEC 61000-2-12) (continued)
30
H. Andrei et al.
Table 1.2 (continued) No
Parameter
Supply voltage characteristics according to EN 50160
Low voltage characteristics according to EMC standard EN 61000 EN 61000-2-2
4
Supply voltage dips
Majority: duration
> > ; : fn ðx; uÞ s:t:
hðs; uÞ 0 gðs; uÞ ¼ 0
ð5:43Þ
where u is the vector of controlling or independent variables of grid including the generative powers apart from reference bass, generators voltages, transformer tap and injected reactive power by parallel elements and can be expressed in the following form. u ¼ ðPG ; VG ; T; Qsh Þ
ð5:44Þ
x ¼ ðPGref ; V; d; QG Þ
ð5:45Þ
Also, x is the vector of status or dependent variables of the grid including load bass voltage, base voltage phase, generators reactive power, generative power in reference bass and can be expressed as follows: G(x, u) represents the equal constraints which indicates the equations of system load distribution. With the adjustment of u as the controlling variable in each stage and solving the non-linear equations of load flow, the corresponding x quantities are calculated. H(x, u) indicates the unequal constraints and includes the following cases: Equal constraints: PGi
PDi ¼
NX Buses
Vi VJ Yij Cosðhij
di þ dj Þ 8i; j 2 NBuses
ð5:46Þ
j¼1
QGi
QDi ¼
NX Buses j¼1
Vi VJ Yij sinðhij
di þ dj Þ 8i; j 2 NLoad Buses
ð5:47Þ
218
R. Effatnejad et al. Ng X
PGi ¼ PD þ PLoss
ð5:48Þ
QGi ¼ QD þ QLoss
ð5:49Þ
i
Ng X i
Unequal constraints: (A) Capacity limits of generation units which includes the high and low limit of voltage rate, generation power of active and reactive. The output power of each generator should not be more than its nominal rate and also it should not be less than the quantity which is necessary for the durable use of steam boiler. So, the generation is limited such that it could place between the two predetermined limits of minimum and maximum: VGmin VGi VGmax i i
ð5:50Þ
max Pmin Gi PGi PGi
ð5:51Þ
max Qmin Gi QGi QGi
i ¼ 1; 2; ; . . .; Ng
ð5:52Þ
(B) Compensational power limits by parallel elements max Qmin shi Qshi Qshi
ð5:53Þ
timin tii timax
ð5:54Þ
(C) Limit of tap transformer
(D) Limit of equipment use which includes the acceptable scope for the voltage and loading rate: Vimin Vii Vimax
PLi Pmax Li ; i ¼ 1; 2; 3; . . .; NBranches QLi Qmax Li
ð5:55Þ ð5:56Þ
In this discussion, the optimization of reactive power is concerned which the objectives can be stated as follows [17–19].
5 Reactive Power Control in Wind Power Plants
219
5.10.1 Minimizing the Real Losses One of important goals of optimal use of reactive power is to minimize the losses of real power in the transfer grid which in this chapter has been considered as the function of the goal of optimization problem. The rate of losses in the transfer grid can be calculated as follows [18, 20, 21]: PLoss ¼
NL X
gk ½Vi2 þ Vj2
2Vi Vj cos ðdi
dj Þ
ð5:57Þ
k¼1
where Vi and Vj are the primary and last bus bar voltage, hij is the angle difference between bus bar i and j and gij is the conductive media between bus bar i and j.
5.10.2 Durability Indicator Calculation There are many indicators for the analysis of the voltage improvement in power systems including P–V curve analysis, Q–V curve analysis and L-index indicator. In this research, the L-index indicator is used to analysis the durability and voltage sensitivity. For this purpose, one system of n bus bar is divided into two groups of generative and load bus bars. The bus bars 1 to g are the generative bus bars and bus bars g + 1 are load bus bars.
Ig Il
Ygg ¼ Ylg
Ygl Yll
Vg Vl
ð5:58Þ
With regard to the admittance matrix, the L indicator for load bus bars is obtained from the following relation. L j ¼ 1
Ng X i¼1
Vi Fij ; Vj
j ¼ Ng þ 1; . . .; n
ð5:59Þ
Fij can be calculated with regard to the admittance matrix in the following form. Fij ¼ ½YLL 1 ½YLG ð5:60Þ L is an index between zero and one. To the extent that this index is closer to one, to the same extent it indicates the instability and disruption of voltage and the more this index is closer to zero, the more durability it has.
220
R. Effatnejad et al.
L ¼ maxðLj Þ;
j 2 aj
ð5:61Þ
5.10.3 Voltage Profile Indicator As voltage is one of the most important standards from the viewpoint of power quality in presenting services by the electricity companies, so that in the distribution networks, great attention has been made to study the impact of units on voltage. The optimization of voltage profile in power systems is of great significance. In the power grids, there is an effort to minimize the voltage profile. In Viref calculations, 1 per-unit is considered. DVL ¼
NPQ X Vi i 1
5.11
Viref
ð5:62Þ
IEEE Standard 30-Bus Test System
The single line 30-bus bar standard system is shown in Fig. 5.9. As it is obvious in Fig. 5.9 it has 6 generator buses bar. The specifications of these generator units are shown in Table 5.2. As it is clear, the limits of active power generation and cost coefficients and coefficients of pollution rate of each of the units have been pointed out.
Fig. 5.9 IEEE standard 30-bus test system
5 Reactive Power Control in Wind Power Plants
221
Table 5.2 Specifications of the system generator units [22] No.
k
n
c
b
a
c
b
a
PGmin
PGmax
PG1 PG2 PG5 PG8 PG11 PG13
2.857 3.333 8 2 8 6.667
2e-4 5e-4 e-6 2e-3 e-6 e-5
0.0649 0.05638 0.04586 0.0338 0.04586 0.0515
−0.05543 −0.06047 −0.05094 −0.0355 −0.05094 −0.05555
0.04091 0.02543 0.04258 0.05326 0.04258 0.06131
100 120 40 60 40 100
200 150 180 100 180 150
10 10 20 10 20 10
5 5 5 5 5 5
150 150 150 150 150 150
Table 5.3 N.R. results
Method
N.R.
Loss Voltage deviation Voltage stability
4.6174 0.6783 0.1171
The first method is to use the algorithm of N.R. whose results can be observed in Table 5.3. The next step is to calculate the system calculations with the smart algorithm of the second generation of genetics with non-dominated sorting. In this scenario, it is assumed that no wind power plant is placed in the grid and we have no limitation on this case and the problem is defined inform of a three-target problem to reduce the losses, voltage profile and voltage durability. It is clear that the reading of the problem in this state has three dimensions. With the implementation of algorithm of the best reading in the tridimensional space, they have formed a curve which totally none of the spots have a priority over the other one and the selection of the best response is done merely based on the indicators of losses, voltage profile and voltage durability and their preference is made with the employment of algorithm system. Here it is worth pointing that from the viewpoint of Pareto superiority, the extreme points of Pareto front in fact is the optimal solution to the optimization issue, when each of the objectives are studied exclusively. So, considering this issue, it is possible to study the quality of responses resulting from the suggested algorithm can be studied during the analysis of responses. Figure 5.10 shows the set of optimal Pareto. After optimization, it is observed that the best reading of the last repetition of Algorithm has N-member which none of them has priority over the other one. The lack of existing continuity is due to the disrupted nature of optimization process. The resulting optimal point based on algorithm output is a suggestion. The best responses which optimize each of the three objectives in this state exclusively are displayed in Table 5.4. The next step is to calculate the system with the smart algorithm of the second generation of genetics with the non-dominated sorting with the assumption of the presence of wind power plants with 10 MW active powers with power coefficient of 0.9.
222
R. Effatnejad et al.
Fig. 5.10 Curves of the first Pareto front without wind turbine
Table 5.4 Results without wind turbine Method
Best loss
Best voltage deviation
Best voltage stability
Loss Voltage deviation Voltage stability
1.5198 0.5392 0.0486
4.1485 5.6788e–06 0.0589
5.3467 0.5073 0.301
In this scenario, it is assumed that one power plant has been placed in the grid and the capacity generation limit of 10 MW active powers with power coefficient of 0.9 and the goal is to improve the reactive power. The problem is defined in for of a three-target problem to reduce the losses, voltage profile and voltage durability. It is clear that the reading of the problem in this state is three dimensional. With the implementation of the best algorithm of reading in tridimensional space, they have formed a curve which totally none of the points is preferred to the others and the selection of the most suitable response is merely done based on the indicators of loss indictors, voltage profile and voltage durability and their preference is done by employing the algorithm system. Figure 5.11 shows the optimal Pareto of this system. After optimization, it is observed that the best reading of the last repetition of Algorithm has N member and none of them has priority over the others. The resulting optimal point based on algorithm output is a proposal. The best response which optimizes exclusively each of the three objectives put forth in this state is displayed in Table 5.5.
5 Reactive Power Control in Wind Power Plants
223
Fig. 5.11 Curves of the first Pareto front with wind turbine
Table 5.5 Results with wind turbine Method
Best loss
Best voltage deviation
Best voltage stability
Loss Voltage deviation Voltage stability
1.4503 0.5167 0.0709
1.7768 9.7316e–05 0.0447
2.4338 0.3332 0.0305
Comparing the results, it is clearly learned that the presence of wind power plant is effective in improving the reactive power in the grid.
5.12
Conclusion
The role of wind power plant in today’s environmental and energy dependable development is so crucial. Lots of studying and researches have been performed on wind power technologies and numbers of wind farm have been utilized around the world. The performance of overall wind power plant depends on the subsystem such as reactive power compensation and energy storage to maintain stability. However, with the increasing capacity of the wind power plant the cost and benefit of these subsystems became unfeasible. By increasing the wind farm capacity the cost for reactive power and energy storage increases. In future works, it could be feasible to test wind power plant using combined capacitor and reactive power compensation which could low the cost. The overall development of these subsystems in wind power plant depends on their cost.
224
R. Effatnejad et al.
It is also concluded that a DFIG is a wound-rotor doubly-fed electric machine (similar to a Synchronous generator), and as its rotor circuit is controlled by a power electronics converter, the induction generator is able to control import and export reactive power. The control of the rotor voltages and currents enables the induction machine to remain Synchronized with the grid while the wind turbine speed varies. A variable speed wind turbine compared to a fixed speed wind turbine utilizes the available wind resource more efficiently, especially during light wind conditions. The converter cost is not as high as other variable speed solutions because only a fraction of the Mechanical Power is fed to the grid through the converter, the rest being fed to grid directly from the stator. The mechanical efficiency in a wind turbine is dependent of the power co-efficient. The power co-efficient of a rotating wind turbine is given by the pitch angle and the tip speed ratio. Adjustable speed will improved the system efficiency since the turbine speed can be adjusted as a function of wind speed to maximize output power. Performing simulations and by looking through real experiments it can be concluded that the wind turbine unit utilizing double fed induction generator is an important and effective tool from the voltage regulation point of view. So it plays a significant role in supplying the reactive power of the network.
References 1. A. Berizzi, C. Bovo, V. Ilea, M. Merlo, A. Miotti, F. Zanellini, Decentralized Reactive Power Control of Wind Power Plants, 2nd IEEE Energycon Conference & Exhibition, Future Energy Grids and Systems Symp, 2012. 2. D.F. Opila, A.M. Zeynu, I.A. Hiskens, Wind Farm Reactive Support and Voltage Control, IREP Symposium - Bulk Power System Dynamics and Control - VIII (IREP), Buzios, RJ, Brazil, August 1–6, 2010. 3. M.A. Nielsen, Power Quality and Grid Connection of Wind Turbines, Proc. of Solar ’97 Australian and New Zealand Solar Energy Society, Paper 154: Nielsen. 4. R. Jacobson, B. Gregory, Wind Power Quality Test for Comparison of Power Quality Standards, National Wind Technology Center, June 1999. 5. E.H. Camm, et al., Reactive Power Compensation for Wind Power Plants, IEEE PES Wind Plant Collector System Design Working Group, 2009. 6. A. Zare Bargabadi, Design of Power Oscillation Damping Controller Using Hybrid Fuzzy Logic and Computational Intelligence Base on Power System Stabilizer (PSS) and Supplementary Controller of Doubly Fed Induction Generator (DFIG) Wind Turbine, M.Sc. Thesis, Karaj Branch, Islamic Azad University, Summer 2014. 7. J.M. Garcia, M. Babazadeh, Control of Large Scale Wind Power Plants, IEEE Power and Energy Society General Meeting, 2012. 8. H. Modirzare, Analysis Impact of DG at Power System Planning Aspect of Reactive Power for Improvement Power Quality, M.Sc. Thesis, Karaj Branch, Islamic Azad University, Summer 2013. 9. H. Modirzare, P. Ramezanpour, R. Effatnejad, Optimal Allocation of DG Units and Var Compensators Suspect to GA Based Reactive Power for Power Losses Decreasing and Voltage Stability and Profile Improvements, International Journal on Technical and Physical Problems of Engineering (IJTPE), issue 18, vol. 6, no. 1, pp. 125–130, March 2014.
5 Reactive Power Control in Wind Power Plants
225
10. M. Akhlaghi, Improving Reliability Indices in HL2 System Through Reactive Power Management, M.Sc. Thesis, Karaj Branch, Islamic Azad University, February 2014. 11. M. Akhlaghi, P. Ramezanpour, R. Effatnejad, Weak Points Identification of HL2 Systems Using Contingency Analysis, The International Conference in New Research of Electrical Engineering and Computer Science, September 2015. 12. V. Karunakaran, R. Karthikeyan, Reactive Power Management for Wind Electric Generator, International Journal of Scientific & Engineering Research, vol. 2, issue 5, May 2011. 13. X. Chen, Reactive Power Compensation and Energy Storage in Wind Power Plant, A Major Qualifying Project Report Submitted to the Faculty of Worcester Polytechnic Institute in Partial Fulfillment of Requirements for the Degree of Bachelor of Science, January 2012. 14. H. Akagi, H. Sato., Control and Performance of a Doubly-Fed Induction Machine Intended for a Flywheel Energy Storage System, IEEE Trans. on Power Electronics, vol. 17, no. 1, pp. 109–116, 2002. 15. V. Akhmatov, Induction Generators for Wind Power, Multi-Science Publishing, Brentwood, 2007, ISBN 10: 0906522404. 16. Y. Mishra, S. Mishra, M. Tripathy, N. Senroy, Z. Dong, Improving Stability of a DFIG-Based Wind Power System with Tuned Damping Controller, IEEE Trans. on Energy Conversion, vol. 24, no. 3, pp. 650–660, 2009. 17. Z.X. Liang, J.D. Glover, A Zoom Feature for a Programming Solution to Economic Dispatch Including Transmission Losses, IEEE Trans. Power Syst., 7(2):544–50, 1992. 18. H. Aliyari, R. Effatnejad, A. Areyaei, Economic Load Dispatch with the Proposed GA Algorithm for Large Scale System, Journal of Energy and Natural Resources, 3(1):1–5, 2014. 19. R. Effatnejad, H. Aliyari, H. Tadayyoni, A. Abdollahshirazi, Novel Optimization Based on the Ant Colony for Economic Dispatch, International Journal on Technical and Physical Problems of Engineering (IJTPE), issue 15, vol. 5, no. 2, pp. 75–80, June 2013. 20. H. Shayeghi, A. Ghasemi, Application of MOPSO for Economic Load Dispatch Solution with Transmission Losses, International Journal on Technical and Physical Problems of Engineering (IJTPE), issue 10, vol. 4, no. 1, pp. 27–34, March 2012. 21. A. Khorsandi, S.H. Hosseinian, A. Ghazanfari, Modified Artificial Bee Colony Algorithm Based on Fuzzy Multi-Objective Technique for Optimal Power Flow Problem, Electric Power Systems Research, 95: 206–213, 2013. 22. H. Aliyari, Application of Meta-heuristic Algorithms for Multi-Objective Optimization of Reactive Power, Power Losses and Cost Function in Power System, M.Sc. Thesis, Science and Research Alborz Branch, Islamic Azad University, August 2014.
Chapter 6
Reactive Power Control and Voltage Stability in Power Systems Mariana Iorgulescu and Doru Ursu
Abstract Reactive power control is sometimes the best way to enhance power quality and voltage stability. In the first part of chapter we describe the reactive power flow impact in the system starting from the definitions of power components and presentation of the electrical equipment that produces or absorbs the reactive power. Then we present the reactive power control and the relations between voltage stability and reactive power. The third part of chapter contains reactive power control methods for voltage stability. In the end of chapter we present the management of voltage control based on case studies.
6.1
Introduction
It is known that the energy system is a great system, complex and continuous change. The electro-energetically system is a system made up of an ensemble of interconnected sub-systems. These sub-systems consist of generators, transformers, electric lines and many types of consumers, those that actually constitute the load of this system [1, 2]. Every moment of time the system operates in and out of its elements causes changes in system parameters. The quality of electricity delivered to consumers is a very important requirement in the power system operation. For every moment in time, for a system with a stable functioning, there must be a very good balance between the quantity of energy produced by hydro-generators, turbo-generators, nuclear reactors and the energy consumed by the loads at a certain moment. For an optimal functioning of the system and for reducing the energy losses in the system, it is necessary to lower the reactive power circulation [2]. M. Iorgulescu (&) Faculty of Electronics, Communications and Computers, University of Pitesti, Pitesti, Romania e-mail: [emailprotected] D. Ursu CEZ Distribute, Craiova, Romania e-mail: [emailprotected] © Springer International Publishing AG 2017 N. Mahdavi Tabatabaei et al. (eds.), Reactive Power Control in AC Power Systems, Power Systems, DOI 10.1007/978-3-319-51118-4_6
227
228
M. Iorgulescu and D. Ursu
Due to a permanent need for consuming electric energy, on the one hand, and the rise in the prices of the classic fuels, on the other, new sources of renewable energy have been developed [3, 4]. Their placement has led to the appearance of the problems related to the instability of the system and of the voltage in its nodes. The voltage stability is intimately related to the reactive power control, achieved by means of series capacitors, synchronous condensers, and the modern static compensator. Another extremely important problem in the energetic system is the quality of the delivered electric energy as it conditions the proper functioning of the electro-energetic system and of its consumers. Power quality parameters are: • • • • •
Voltage supply; Frequency; Total harmonic distortion of the voltage wave; Symmetry phase voltage system; Continuity of supply.
Voltage stability and frequency stability have lately become became the two important electric power quality parameters in the functioning of the power system. Equally important is the knowledge of how the power system elements that can cause system instability function. The voltage is a parameter of the electric energy which is different in the energy system nodes. This parameter depends on the values of the impedance of the elements which are in the network or exit the system and, implicitly, on the voltage drops on these impedances. Maintaining the imposed voltage level in a network node constitutes a problem of the area corresponding to that node and can be achieved by different means of adjusting the voltage. The ways the voltage can be adjusted in the power system is the focus of the specialists in the field of electric energy. The voltage stability within the nodes of the system is very important in coordinating and operating the system. Problems related to the voltage instability led to the fall of energy systems in countries like Japan in July 1987, on August 14th 2003 in US and Canada, on September 23rd 2003 in eastern Denmark and southern Sweden and days later in Italy and Central Europe following the cascade outage [5]. Considering the possibility of such blackouts appearing in energy systems, artificial intelligence based systems have been developed and tested, aimed at helping the adjustment of voltage and, implicitly, the system’s stable functioning [6]. Such an example of information system was called STABTEN and it was developed and tested by using the CIGRES 32 scheme [7]. The method of using such a device for controlling voltage in an energy system pilot, electronic device which is called ASVR—Automatic Secondary Voltage Regulation—as it is presented in the final of the chapter, is a successfully used method for adjusting voltage by means of optimal reactive power control.
6 Reactive Power Control and Voltage Stability in Power Systems
6.2
229
Impact of Reactive Power Flow in Power System
In order to study the reactive power impact in the system, it is important to start with the term—power—in the electrical system it refers to the energy-related quantities flowing in the transport and distribution network. The instantaneous power is the product of voltage and current with the following equation: pðtÞ ¼ uðtÞiðtÞ
ð6:1Þ
Voltage and current can be in phase or not. When these are not in phase there are two components of power: • active power P, that is measured in Watts P ¼ UI cos u ½W
ð6:2Þ
• reactive power Q (it is known as imaginary of apparent power), that is measured in VAr Q ¼ UI sin u ½VAr
ð6:3Þ
In the above equations, I and U are rms values of current and voltage, and u is the phase angle by which the current out phasing the voltage. The combination of active power with reactive is apparent power S, measured in VA. S ¼ P þ jQ½VA
ð6:4Þ
Reactive power—Q represents a part of apparent power—S. Reactive power is in opposition with active power—P. Reactive power is necessary to maintain voltage and to distribute active power through transmission lines. In this way different loads that use reactive power to convert the received power in mechanical, illumination and others will be in use. In case the presence of reactive power is under the acceptable level the voltage cracks down. The system can become unstable because the loads do not receive necessary power through the transmission line. It is important to study reactive power flow in system because some equipment depends on these few factors can absorb or produce this type of power. Depending on the phase angle between current and voltage, the electrical equipment will consume an amount of reactive power. In power system there are types of electrical equipment of the system that producing or absorbing reactive power, such as [10]:
230
M. Iorgulescu and D. Ursu
• Synchronous machines are electrical machines with three operating conditions: motor, generator and condenser. In generator mode the electrical machines depending on the excitation level can generate or absorb reactive power together with active power. When the machine is overexcited it supplies reactive power, and when is under excited it absorbs reactive power. • Loads are in a great number in the power system. The number of loads, their characteristics or power are dependent on the time of day, season or weather. In rated state every load absorbs active power and reactive power both leading to voltage variations. Some loads operate with low power factor (can be indeed 0.6), with important losses of active power in transmission network. In this case is mandatory for industrial consumers to improve the load power factor by natural means first and specialized devices second. • Transmission lines are parts of the transport and distribution system. Across the lines the voltage and current parameters change in every moment, proving the series and shunt parameters. Based on equivalent circuit it is settled that a transmission line is characterized by the following circuit parameters: – longitudinal or series parameters: resistance R and reactance X; – and transversal or shunt parameters: conductance G and susceptance B. Symbols of the parameters indicate their per-kilometer values. Depending on the load current lines absorb or supply reactive power. In case of loads below the natural load, the lines produce net reactive power, if the opposite situation applies, at loads above natural load, the lines absorb reactive power. The reactive power is given by [1]: Q½VAr ¼ U 2 ½VB½Siemens=km
ð6:5Þ
All parameters of the line are significantly different functions of the conductor size, material, spacing, height above ground, and temperature. • The distribution network is built sometimes in underground cables. These cables are always loaded below their natural loads. The series parameters— especially reactance is lower but shunt parameter—capacitance is higher, affects by charging reactive power under all operating conditions. Based on these, it is very important to design the line with the proper length in order to respect the cable thermal capability. Reactive power flow in electrical network has a negative impact on the power system. In practice almost always the specialists work to reduce the level of reactive power in order to improve the system efficiency. Effects of the reactive power flow in network are: • active power losses increase. In reactive power presence this losses will be: DP ¼ DPa þ DPr
ð6:6Þ
6 Reactive Power Control and Voltage Stability in Power Systems
231
• equipment oversize that increase the installation’s cost. It is known that the equipment’s proportion is according to: Sn ¼ Pn = cos un
ð6:7Þ
cos u \ cos un
ð6:8Þ
Sproportion [ Sn
ð6:9Þ
Operating mode with:
leads to
• decrease of supporting capacity of electrical networks: P ¼ S cos u \ Pn ¼ Sn cos un
ð6:10Þ
DU ¼ U1 U2 ¼ RI cos þ XI sin ¼ DUa þ DUr
ð6:11Þ
• increase of lost voltage:
6.3 6.3.1
Reactive Power Control in Electrical Networks Reasons and Nature of Voltage Variations in Electrical Networks
Reactive power control is fulfilled according to reactive power variations. In steady state of electrical network the voltage is a parameter variable in time and space. Voltage variation in space is given by the different voltage nodes induced by the drops in voltage. These drops in voltage are brought by active and reactive power flow in electrical networks and power transformers. Considering a power transformer or an electrical network with the following single phase equivalent diagram, Fig. 6.1—longitudinal impedance with equation [8, 9]: Z ¼ R þ jX Fig. 6.1 Single phase equivalent circuit
ð6:12Þ
232
M. Iorgulescu and D. Ursu
It establishes the complex drop in voltage with two components: – a longitudinal component DU, regularly named drop in voltage, with the equation:
DU ¼ U1 U2 ffi
PR þ QX PR þ QX ffi U2 Un
ð6:13Þ
– a transversal component dU, determination of angle h value between voltage in the extreme of network, with the equation:
dU ¼
PX QR PX QR ffi U2 Un
ð6:14Þ
Practical longitudinal impedances of transmission lines are characterized by the small value of resistance comparative with inductive reactance R X. Based on the phase diagram of drop in voltage, Fig. 6.2, and equations of drop in voltage, approximate equations of DU are DU ffi h ¼ arcsin
QX Un
dU PX ffi arcsin 2 U1 Un
ð6:15Þ ð6:16Þ
These equations point out an important characteristic of the system, the level of voltage is determined by the reactive power flow and the h value by the active power flow. Time variations of voltage in network nodes are tie-in with the voltage drops variations in time. These variations have the following causes: – active and reactive power flow variability in time as a consequence of variation of absorbed power by the consumers and produced power by the power plant, also. Fig. 6.2 Phasor diagram of drop in voltage on longitudinal impedance
6 Reactive Power Control and Voltage Stability in Power Systems
233
– modification of the network topology, longitudinal impedance and transversal admittance. This change is imposed by the necessity for optimizing the technical indicators of loaded transmission lines. As a result, in one node of the network the voltage is variable in time U(t). Voltage variation in time in the node is named voltage deviation, measured as per cent at one moment in time, with the equation: DU ðtÞ ¼
U ðtÞ Un 100½% Un
ð6:17Þ
These voltage variations can be hard growing or fast. Hard growing variations appear as a consequence of a flawed control or overloaded electric network. This variation can be automatically controlled through regulators. The regulators send an impulse to beakers in order to introduce the compensation equipment into the circuit, such as synchronous compensators, bank of capacitors or transversal coil. Fast growing of reactive power can be controlled automatically trough regulators with static commutation (GTO) in order to introduce quickly compensation equipment in the circuit. Synchronous machines, bank of capacitors, GTO and various types of other equipment are used to maintain voltages throughout the transmission system. Injecting reactive power in the system increases voltages, and absorbing reactive power decreases voltages.
6.3.2
Reactive Power Control Methods for Voltage Stability
The voltage is one parameter that fluctuates in time due to the modification of the total power absorbed by the receptor, together with operating specification of the power system (revisions, failures, etc.). Also, the voltage fluctuates in dimension because of lost voltage in network leads to different values of voltage in different nodes. Level of voltage is indissolubly linked to a stable operating of the power system. Also the voltage adjustment is interrelated with the reactive power control possibilities. Reactive power control methods for voltage stability are presented in the next section.
234
6.4 6.4.1
M. Iorgulescu and D. Ursu
Voltage Stability in Power System Voltage Control by Reactive Power Flow Adjustment
The voltage adjustment by reactive power flow control can be continuous, used like a primary means of voltage regulation, or discrete used like a secondary means of adjustment. The principle of voltage regulation by reactive power flow adjustment can be illustrated in Fig. 6.3. In Fig. 6.3, k is the installation of reactive power compensation linked to consumer bus, producing Qk reactive power. In this case reactive power transmitted on the line decreases to (Q2–Qk) value. As a consequence, dropping voltage decreases and the voltage value increases. In another case, with the k installation absorbing reactive power, having inductive consumer operating, dropping voltage increases and U2 decreases. The conclusion of this paragraph is that by modifying the reactive power, we can control voltage on consumer bus. This type of control is based on the reactive power sources of the power system (generators, synchronous compensators, capacitors). The same role is played by the coil mounted in the high voltage electrical network. These are consumers of reactive power. In the next section there are presented characteristics of important voltage control means: generators, condensers and inductors. Generators: The main function of electric-power generators is to convert different type of energy into electric power. The generators have significant control on their terminal voltage and reactive power output. This can produce or absorb reactive power depending on the magnetizing current value. The increasing of the magnetic field in the synchronous machine implies the raising of generator’s terminal voltage in order to produced reactive power. The magnetic field increasing requires current increasing in the rotating field winding. Absorption of reactive power is limited by the magnetic-flux design in the stator, which leads to over-heating of the stator-end iron. It is known that in every electrical machine there are the core-end heating limits [10]. Synchronous Compensators are synchronous motors used sometimes to provide dynamic voltage support to the power system as they provide mechanical
Fig. 6.3 Reactive power compensation on consumer bus
6 Reactive Power Control and Voltage Stability in Power Systems
235
Fig. 6.4 Reactive power variation versus magnetizing current
power to their load. In some power plants the hydro units are designed to permit the generator to function without its mechanical power source simply to provide the reactive power capability to the power system [10]. These hydro units work as a compensators when the active power generation is not necessary, Fig. 6.4. For a magnetizing current value Ie0,produced reactive power is zero. In case Ie [ Ie0 the synchronous condenser operates overexcited, with power factor capacitive conveying in electric network reactive power. When Ie \Ie0 the synchronous condenser operates under excited, with power factor inductive absorbing from electric network reactive power. Proportion of maximum reactive power s absorbed Qsb k and produced power Qk it is a very important characteristic of the condenser operation. Advantages of using synchronous condenser are: – operating possibilities as reactive power generator or consumer; – continuous regulation of produced or absorbed reactive power by magnetization current; – contribution to the system stability, with auto-regulation effect, once the decrease of voltage leads to an increase in reactive power supplied. A disadvantage can be the limited applicability of synchronous condensers on a large scale. The disadvantage is due to the consumption of active power around 3% of machine’s reactive power rating, or operating expenses. Capacitors are passive devices that generate reactive power, with some advantages: – significant real-power losses around 2–3% from the rated power; – low operating expense; – possibility to space out payment of investment by gradually developing the capacitor bank with new elements. Capacitor banks are composed of individual capacitor connected in series and/or parallel in order to obtain the desired capacitor-bank voltage and capacity rating
236
M. Iorgulescu and D. Ursu
Fig. 6.5 a Star connection of bank-capacitor, b triangle connection
[10], Fig. 6.5 [13]. Banks capacitor are discrete devices but they are often configured with several steps to provide a limited amount of variable control, and are spread in small power [10]. Produced reactive power is given by Qk ¼ mxCU 2 ½MVar
ð6:18Þ
where – U—network voltage, in kV; – C—bank capacity, in F; – m—coefficient with value 1 in star connection and 3 in triangle connection. The output of capacitors is proportional to U2. In fact this effect is bad in voltage control and in operating conditions of the system, also. To avoid this, the bank capacitors should be input by autotransformer. Inductors are passive devices that absorb reactive power. The inductors are shunt connected to high voltage transmission lines or to the tertiary autotransformer’s winding with purpose to absorb capacitive power generate by these lines in low load operation. In this case there were reduced possibilities to have in the system overvoltage produced by capacitive currents flow. Inductors are built in three phases or formed with three single phases. Frequently the principle of reactive power control consists in modifying the number of inductors in shunt connection. Static VAR Compensators (SVCs) By combining the banks capacitor with inductors we will obtain static VAR compensators. These compensators can be designed to absorb or produce reactive power. These devices have an operating principle like a synchronous condenser, the reactive power can be controlled continuously or step by step. In order to adjust in the continuous mode the power of installation is used for inductor’s control a bias
6 Reactive Power Control and Voltage Stability in Power Systems
237
Fig. 6.6 STATCOM schematic diagram [16]
winding or thyristors control scheme. This control is very useful for the receivers, whose operations produce power surges in the electrical network. Static Synchronous Compensators (STATCOMs) The STATCOM (STATic synchronous COMpensator) is a shunt-connected reactive compensation equipment which generates and absorbs reactive power. Its output can be varied so as to maintain control of the power system specific parameters [14–16, 19]. The STATCOM belongs to the family of devices known as flexible AC transmission system (FACTS). The schematic diagram is presented in Fig. 6.6. STATCOMs give the power system the following advantages: – using capacitors and inductors combined with fast switches, finally, STATCOM uses power electronics to synthesize the reactive power output [10]. – output capability is commonly symmetrical, providing as much capability for production as absorption [10]. – controls are designed to provide a fast and effective voltage control – STATCOMs capacity is not affected by degraded voltage. – STATCOMs are current limited so their MVAr capability responds linearly to voltage, increasing STATCOMs’ usefulness in preventing voltage collapse [10]. The voltage control in the medium voltage network can be made using a system called by its developers UPQC (Unified Power Quality Conditioner), Fig. 6.7 [17]. It has the role of compensating the voltage variations and of ensuring the user’s clamps receive a set voltage, practically constant. This installation was tested within the medium voltage network which a photo-voltaic plant was connected to. The installation has the role of correcting the ir electric current loop absorbed by the user as the parallel converter introduces a if current with such a synthesized form
238
M. Iorgulescu and D. Ursu
Fig. 6.7 Electrical network conditioner UPQC [17]
that the ir electric current to be sinusoidal. The Uf voltage introduced in series with the network voltage by the serial converter maintains a constant value of the user supply voltage.
6.4.2
Voltage Control by Power Network Parameters Adjustment
This process of voltage adjustment involves changes in electrical network parameters, either by connecting or disconnecting circuits operating in parallel, or by compensating the inductive reactance power lines. Some parts of the electrical network are built in parallel connection—two or three circuits, transmission lines or power transformers connected in parallel [8]. One of these can be activated or also stopped from operating, can modify longitudinal parameters (impedance) of network, leading to change voltage in consumption nodes. In the analyzed Fig. 6.8 under the minimum load operating conditions or empty load voltage U2 consumer bars may increase above the rated value. Disconnecting a circuit power line or a transformer in the network increase the longitudinal impedance of the circuit and voltage drop, respectively, so that the voltage level can lower the U2 consumer bars.
Fig. 6.8 Electrical network to a consumer through two parallel circuits
6 Reactive Power Control and Voltage Stability in Power Systems
239
Fig. 6.9 Capacitor mounted in cascade on transmissions line
Fig. 6.10 Single phase equivalent circuit of compensated electrical network
In practice, disconnecting transformers operating in parallel, with voltage control purpose in low load operating case is a measure recommended in terms of reducing power losses and active energy in electricity network. Because the power transformer is a component of the electrical networks that presents a high level of safety in operation, it is generally accepted to operate transformer stations with a single transformer. Regarding take off transmission lines, it is important to know that these are less reliable in operation and in practice, it is not recommended to disconnect transmission line circuits to control voltage in electrical networks. Another procedure used for voltage control is used cascade circuit compensation of an electrical network by inductive reactance. This is accomplished by the capacitors mounted in cascade with reactance, Fig. 6.9, obtaining a reduction longitudinal reactance of the line and thus a decrease of voltage drops. In Fig. 6.10 there is represented a single phase equivalent circuit of a compensated electrical network XL ¼ xL
ð6:19Þ
1 xC
ð6:20Þ
is transmissions line reactance, and XC ¼
is capacitive reactance series on line. Resulted reactance will be
240
M. Iorgulescu and D. Ursu
X ¼ XL XC ¼ xL
1 xC
ð6:21Þ
Based on the capacitive reactance’s value there are the following situations: – under-compensated line, when XC \XL ; – total compensated line, when XC ¼ XL ; – over compensated line, when XC [ XL . Longitudinal compensation with banks capacitor series connected with long lines of electricity transmission is done in order to increase the transport capacity of these lines and limits of stable operation of the power system. To control voltage from a practical standpoint, series compensation with banks capacitor is important for the medium voltage lines in the following cases: – overhead lines relatively long, slow voltage variations reduction to consumers; – electrical lines feed to consumers that produce shock load in the network, in order to reduce shock occurring in the voltage corresponding to power supply. On the other hand, it must be known that the compensation process causes some negative phenomenon: ferro-resonance, under-synchronous resonance or oscillation of the synchronous machines. For limiting them a research to develop possibilities and removing their appearance is required. Measures that can be taken to avoid these are: – restrict the compensation to between 1.5 and 2; – using series or parallel resistors with banks capacitors. Another drawback of the procedure- series bank capacitors for compensation of the inductive reactance of the power lines is the increase of the short-circuit current, symmetrical and non-symmetrical. The capacitors are protected by special circuits in the short circuit case within power system. This method is used in heavy growing and fast variation of reactive power in the system.
6.4.3
Node Voltage Set Up
The importance of voltage control in electrical network nodes is to change the level of voltage in admissible in them. This process seems to be the most natural, but it will not remove the cause of voltage fluctuations causing variations reactive loads transiting through electric transmission elements and distribution network. This voltage control method consists in injection of supplementary voltage in order to compensate longitudinal component of lost voltage. The injection voltage is realized by autotransformers or power transformer equipped with adjuster under longitudinal load or transversal. Often are used in power system voltage control by the automat voltage regulator.
6 Reactive Power Control and Voltage Stability in Power Systems
241
Base on introducing additional voltage in an electrical network, there are the following types of control methods [18]: • longitudinal adjustment, which is intended to compensate the longitudinal component of drop in voltage. Additional being introduced in phase with the line voltage at that point; • transverse adjustment when additional voltage is in with quadrature voltage network at that point, and serves mainly to modify the power flow in the mesh network topology; • adjustable long-transverse, when the longitudinal adjustment is used to adjust the transverse correlated. In terms of structural and functional features of the used means to introduce additional voltage, there are to be distinguished two types of control: – direct control, which is performed using power transformers and autotransformers fitted with sockets for adjusting the working windings, as well as with devices for switching them in the presence or absence of load; – indirect control, which is performed by means of special transformers and special autotransformer-overvoltage transformer. These constitute complexes devices that can be associated to regulating power transformers non-adjustable or can be used separately. These also include the regulators induction.
6.4.4
Management of Voltage Control
Base of voltage adjustment consists of generating units by using their capability to absorb and debit reactive power (to modify the magnetization current and implicit electromotive force linked by synchronous reactance), within diagram P-Q available. For basic voltage control use the following devices: – – – –
Synchronous condenser; Synchronous generators operating in thermo and hydro plants; Wind power plants; Photovoltaic power plants.
All other means of voltage control, maintained by the consumers/network operators have a secondary role regardless of their number and capability and they come in support of the generating sets without replacing the first. These are: – – – –
Banks capacitors; Inductors; Control by power transformer sockets; Connect/disconnect of electrical lines transmission.
242
6.5
M. Iorgulescu and D. Ursu
Case Studies
1. Q–V voltage regulation approach in the pilot node using an automatic voltage control (ASVR—Automatic Secondary Voltage Regulation) developed in Romania The goal is to maintain the required voltage in the pilot node within the desired voltage domain. ASVR equipment consists of an independent regulator working on the principle of regulatory with negative feedback response. ASVR are placed in the power plant or electrical station, and the desired voltage it is received by the voltage domain that must keep it in pilot node. ASVR = Automatic Secondary Voltage Regulation in pilot node of 400 kV— automation which controls the voltage in the 400 kV pilot node from a transformation station. Further on we will refer to such a system associated to a wind plant, but this system can be used for any kind of plant, photo-voltaic, for example. The wind plants management system—which is capable of controlling in real time depending on the imposed voltage level, meaning the decrease of increase of produced reactive power, by controlling the voltage in the medium voltage network (MT) and in the 110 kV, through the modification of the plots at the transformers (power transformer 110/MT from the power transformer transformation stations and power transformer 400/110 kV from the system station), according to the set voltage value from the 400 kV pilot node, but also to the park’s reactive power availability at that moment, increasing or decreasing the transit of active energy from the wind park, if the adjustment analysis takes into consideration the electric scheme from Fig. 6.11. The automatic installation for the secondary adjustment of the ASVR voltage is destined to the voltage adjustment at the 400 kV level in the 400/110/MT station, as the reference is given the voltage value on the station’s 400 kV bars. Thus, in order to achieve this objective, there are used both the wind generator capacity to produce/consume reactive power, but also switching the plots of the MT/110 kV transformers and the 400/110/MT transformers. The primary adjustment loop U–Q of the wind generator group has as reference the value of the voltage on the 110 kV bars so that the reactive power variation of the generators would follow the variation of this voltage, which, in turn—according to the ASVR logic—is proportional to the voltage variation on the 400 kV bars. Additionally, in order to extend the adjustment band, under the conditions of maintaining the inferior voltage of 110 kV, MT and 0.5 kV within the admitted limits, there are used the voltage regulators of the MT/110 kV transformers, respectively the 400/110/MT which act on the plot switches at the ASVR command. From a practical standpoint, compared to the initial (primary) function of adjusting the voltage to the 110 kV through the variation of the wind generators
6 Reactive Power Control and Voltage Stability in Power Systems
243
Fig. 6.11 The transformer station through which a central debits
reactive power, the ASVR installation is designed to perform the following supplementary functions: • The voltage adjustment by modifying the reactive energy produced by the wind generators from the wind power plants by means of the command-control system WFMS, having as reference the voltage at the level of 400 kV bars; • automatic switching of 400/110/MT transformer plots from the 400 kV station; • automatic switching of MT/110 kV transformer plots from all stations corresponding to the plant; • coordinating the three preious functions. • observing the condition to maintain the voltage levels inferior to 400 kV (meaning 110 kV, MT and 0.4 kV) within the admitted limits. • ASVR is composed of three adjustment loops, as follows: • adjustment loop voltage 400 kV; • adjustment loop voltage 110 kV; • adjustment loop voltage MT. The three adjustment loops function simultaneously, as they coordinated by the ASVR logic. Not including one of the loops in the ASVR scheme implies the ASVR functioning without the adjustment system corresponding to that loop (for example, not including the Mt adjustment loop determines the ASVR functioning without the MT/110 kV plot switching).
244
M. Iorgulescu and D. Ursu
Fig. 6.12 ASVR integration in electrical network
The coordinating function of the three adjustment loops leads to a hierarchy of their actions, so that the adjustment process records the following hierarchy (priority): 1. loading/unloading with reactive energy the wind generator ensemble from the electric plant; 2. plot switching at the MT/110 kV transformers of the producer (from the plant stations); 3. plot switching at the 400/110/MT transformers from the 400/110 kV where the pilot node is considered. The adjustment loop ensemble which ASVR consists of is schematically presented in Fig. 6.12. Release period/start: The interval between two control interventions are usually set on 18 20 s. Also every deviation must be eliminated by occurred adjusting 120 s of its release.
6 Reactive Power Control and Voltage Stability in Power Systems
245
• The ASVR will command the power transformers 400/110 kV on the basis of criterion of minimal voltage difference between the sockets; • In case of increase voltage 110 kV will be used by the transformer with the socket on lower voltage; • When voltage decreases 110 kV the transformer is used with socket to the higher voltage; • The criterion of reactive power transfer will be applied in voltage equality between the sockets. 2. Q–V voltage control in wind power plant, with imposed Q We have a wind power plant to which a ASVR is used in order to maintain voltage in the 110 kV, the consign the plant receives being of reactive power Qabs or debited Q, according to the ‘scale’ chart from Fig. 6.13 and the adjustment itself consists in loading/unloading the wind generator group from the wind plant with reactive energy. In the diagram presented in Fig. 6.13, imposed Q, as follows ASVR, Q absorbed or Q produced to be as close to the preset value. These values are indicated by Q measured which falls or how much it increases depending on the preset value Q. Also voltage of 110 kV aims at absorbing or dispensing Q, dropping respectively U with an increase for which we have in view the achievement of a certain preset Q. Acquisition and execution of preset reactive power provided by ASVR controllers will be using the entire reactive power reserves that actions wind power plant, including the existing compensation means to them. In case P = 0 for a wind plant will provide for the adjustment Q = 0 or Q 6¼ 0, and the controller must know the limits of maximum and minimum reactive power that wind plant led so it can provide active power when it’s generated or not.
Fig. 6.13 Q–V diagram with imposed Q
246
M. Iorgulescu and D. Ursu
3. Q–V voltage control in photovoltaic power plant, with imposed U We have a wind power plant to which a ASVR is used in order to maintain voltage in the 110 kV, the consign the plant receives being of voltage adjustment U, according to the ‘scale’ chart from Fig. 6.13 and the adjustment itself consists in the three loops’ coordinated functioning, thus leading to a hierarchy of their actions so that the adjustment process can be performed. In the diagram, in Fig. 6.14 can be seen as an imposed voltage, ASVR follows that voltage 110 kV to be as close to the preset value. These values are indicated by measured voltage that goes up and down depending on the value of voltage. Also instruction for reactive power Q is produced or absorbed in order to increase, decrease voltage respectively, as we have in mind to reaching any preset voltage. According to the short-circuit on bar power at the voltage level for which the voltage is aimed to be adjusted, there are the results ∆Q/∆U [MVAr/kV] for the voltage level we want to maintain, meaning, for example, that, at a variation of 2 kV we approximate ±16 MVAr (Fig. 6.14), respectively at a de ±8 MVAr variation, as the voltage varies with ±1 kV/8MVAr (the mentioned values are just for exemplification, as they depend on the voltage level where the adjustment is made). The dU/dT value shows the response speed, the decrease/increase of voltage in a minute, indicating the system’s capability to reach the values for U consign, according to the voltage rise.
Fig. 6.14 Q–V diagram photovoltaic power plant—U preset
6 Reactive Power Control and Voltage Stability in Power Systems
6.6
247
Conclusion
The wide development of renewable energy sources connected to the high medium voltage lines imposed the development of specific solutions for the voltage control in the point where these sources connect or in their vicinity. Today’s energy systems are facing the problem that, in some areas of the system there are is an exceeding production, while in other areas the production is lower, while energy is transited from the sources in order to balance the consumption. All these occur as result of the way the plants are located, as they, for different reasons, chosen for other criteria than the optimal location and, considering the degree to which the network in loaded, the power is transited to a consumption node which can be, in some situations, very far from the source, this leading to a decrease in the area’s level of voltage. The phenomenon can also appear in reverse, meaning an increase in voltage, in the situations when the plant connection is made in nodes with reduced short-circuit power, thus appearing the limitation of the power debited by the plant in accordance with the voltage maximally admitted at the final consumers, especially in the cases of plants recorded at the MT lines. For these reasons, it is important to control the voltage level, and this is why one must always analyze solutions to maintain the voltage within the limits which are adequate for the user’s system. For the transport network, one of the solutions is the ASVR system which maintains the voltage in pilot node chosen in the system, compensating the voltage variations in the system, with the three presented adjustment loops, thus using two of the adjustment methods, the reactive energy absorption and debit and the longitudinal and long-transversal voltage adjustment in the transformers by using the plots. The ASVR can be also used in the distribution network, with less control loops, two or even one. Considering the voltage variation at the final user clamps, from each voltage level from the system, depends in each moment on production, consumption, network loading, using an automated voltage regulator for that voltage level allow for the necessary conditions of ensuring the quality of the energy from the network and, consequently, for all users.
References 1. C.W. Taylor, Power System Voltage Stability, McGraw Hill, 1994. 2. T.J.E. Miller, Reactive Power Control in Electric System, 1982, John Wiley & Sons, Inc., USA. 3. P.W. Sauer, Chapter 2, Reactive Power and Voltage Control Issues in Electric Power Systems. 4. J. Hossain, H.R. Pota, Robust Control for Grid Voltage Stability: High Penetration of Renewable Energy, Power Systems, DOI:10.1007/978-981-287-116-9_2, Springer Science-Business Media, Singapore, 2014.
248
M. Iorgulescu and D. Ursu
5. IEEE Power Engineering Society Administrative Committee of the Power System Dynamic Performance Committee, Causes of the 2003 Major Grid Blackouts in North America and Europe, and Recommended Means to Improve System Dynamic Performance. 6. W. Nakawiro, Voltage Stability Assessment and Control of Power Systems Using Computational, Intelligence, http://duepublico.uniduisburgessen.de/servlets/DerivateServlet/ Derivate-26570/Nakawiro_Worawat_Diss.pdf, 2011. 7. C. Bulac, M. Eremia, A.I. Bulac, I. Tristiu, Voltage stability in SEE Phenomenon Assessment and Control through Artificial Intelligence Techniques, Power Journal 47-1999, no. 2. 8. A. Poeata, et al., Transport and Distribution of Electricity, Bucharest, pp. 267–310. 9. M. Eremia, P. Simon, D. Petricica, D. Gheorghiu, Some Aspects of Hierarchical Voltage Reactive Power Control, Power Engineering Society Summer Meeting, 2001, vol. 2, DOI: 10.1109/PESS.2001.970170, Publication Year: 2001, pp. 872–880, vol. 2. 10. http://electrical-engineering-portal.com/how-reactive-power-is-helpful-to-maintain-a-systemhealthy. 11. F.A. Viawan, D. Karlsson, Coordinated Voltage and Reactive Power Control in the Presence of Distributed Generation, Power and Energy Society General Meeting - Conversion and Delivery of Electrical Energy in the 21st Century, 2008 IEEE, DOI:10.1109/PES.2008. 4596855, Publication Year: 2008, pp. 1–6. 12. E. Demirok, P. Casado Gonzalez, K.H.B. Frederiksen, D. Sera, P. Rodriguez, R. Teodorescu, Local Reactive Power Control Methods for Overvoltage Prevention of Distributed Solar Inverters in Low-Voltage Grids, IEEE Journal of Photovoltaics, vol. 1, no. 2, December 2011. 13. S. Bisanovic, M. Hajro, M Samardzic, One Approach for Reactive Power Control of Capacitor Banks in Distribution and Industrial Networks, Electrical Power and Energy Systems, 60, pp. 67–73, 2014. 14. C.A. Canizares, Modeling and Implementation of TCR and VSI Based FACTS Controllers, Internal Report, ENEL and Politecnico di Milano, Milan. 15. H. Mori, Y. Goto, A Parallel Tabu Search Based Method for Determining Optimal Allocation of FACTS in Power Systems, Proc. of the International Conference on Power System Technology (PowerCon 2000), vol. 2, pp. 1077–1082, 2000. 16. S. Kamel, F. Jurado, D. Vera, A Simple Implementation of Power Mismatch STATCOM Model into Current Injection Newton-Raphson Power-Flow Method, Electrical Engineering, 96:135–144 DOI 10.1007/s00202-013-0288-4, Springer, 2014. 17. A. Marinescu, M. Radulescu, N. Golovanov, Control in Blood Medium Voltage Networks Are Connected Renewable Energy, VIIIth Edition of the ASTR Conference. 18. G. Georgescu, et al., Transport and Distribution of Electricity, Ghe. Asachi Publishing House, Iasi 2001. 19. E. Nasr Azadani, Optimal Placement of Multiple STATCOM for Voltage Stability Margin Enhancement Using Particles Warm Optimization, Electrical Engineering, September 2008.
Part II
Compensation and Reactive Power Optimization in AC Power Systems
Chapter 7
Optimal Reactive Power Control to Improve Stability of Voltage in Power Systems Ali Ghasemi Marzbali, Milad Gheydi, Hossein Samadyar, Ruhollah Hoseyni Fashami, Mohammad Eslami and Mohammad Javad Golkar
Abstract The current power systems have works near to the marginal voltage stability due to the market performance as well as their weightier operation loadings along with consideration of environmental constraints of transmission as well as generation capacity enlargement. In other words, at the present time wind power has confirmed to be one of the most efficient and competitive renewable resources and therefore, its use is indeed continually growing. Little wind power infiltration planes are generally contained in the current grid networks in view of that it is passively controlled and operated. On the other hand, this statement is no more suitable for immediately after the wind power energy infiltration commences
A. Ghasemi Marzbali (&) Technical Engineering Department, University of Mohaghegh Ardabili, Ardabil, Iran e-mail: [emailprotected] M. Gheydi H. Samadyar Young Researchers and Elites club, Science and Research Branch, Islamic Azad University, Tehran, Iran e-mail: [emailprotected] H. Samadyar e-mail: [emailprotected] R.H. Fashami Electrical Engineering Department, Damavand Branch, Islamic Azad University, Tehran, Iran e-mail: [emailprotected] M. Eslami Department of Electrical and Computer Engineering, Nikshahr Branch, Islamic Azad University, Zahedan Branch, Nikshahr, Iran e-mail: [emailprotected] M.J. Golkar Department of Electrical and Computer Engineering, Zahedan Branch, Islamic Azad University, Zahedan, Iran e-mail: [emailprotected] © Springer International Publishing AG 2017 N. Mahdavi Tabatabaei et al. (eds.), Reactive Power Control in AC Power Systems, Power Systems, DOI 10.1007/978-3-319-51118-4_7
251
252
A. Ghasemi Marzbali et al.
growing, a broad scope of scientific issues can come out, namely: voltage rise, bi-directional power flow, improved power quality issues as well as distorted voltage stability. The additional improvement of electricity construction from renewable resources in a trustworthy as well as consistent system performance is driving transmission as well as distribution control utilizers to employ novel working models that are not presently extant. A serious subject of the demanding status described in the foregoing is the reactive power managing that involves the planning as well as operation deeds that are asked for to be executed to get better voltage profile as well as stability in the power networks. For this reason, voltage stability is a major issue of current power systems. It signifies the capableness of a power system to keep voltage when the required load is boosted. Researches about this kind of instability fact proceed with its control as well as evaluation. The first one designates if a power system runs in the safe operational area, while the second one will carry out essential control actions if a power system gets close to unsafe operational zone. Diverse approaches put forth in the chapter deal with offline and online purposes. The center of attention of this chapter is the second part; it means control of voltage stability. Three major methods have been utilized for voltage stability which are reactive power management, load shedding and active power re-dispatch. Reactive power management signifies the ways designating the place of novel VAR sources and/or settings of the VAR sources that are installed currently and the settings of facilities including on-load tap changers (OLTCs). Reactive power sources ordinarily consist of synchronous generators/condensers, reactor/capacitor banks, as well as flexible AC transmission systems (FACTS) controllers. It can be classified into two subjects as reactive source programming as well as reactive power dispatch. For reactive programming, the concerned temporal duration is the coming few months or years, and besides considering the optimum milieu of facilities that are installed currently, installation of novel reactive power sources is contemplated. It is performed in offline and online ways. The main purposes of offline reactive dispatch can be found in the duration of the coming few days or weeks, while, another model is carried out in the coming few minutes or hours. Opposing the reactive planning, both online and offline reactive power dispatches only designate the optimum settings of extant facilities. Optimal reactive power flow (ORPF) which is a specific instance of the optimal power flow (OPF) issue is an utterly significant instrument with regard to assured and gainful utilization of power systems. The OPF’s control parameters have a proximate connection with the reactive power flow, including shunt capacitors/reactors, voltage magnitudes of generator buses, output of static reactive power compensators, transformer tap-settings. In the ORPF problem, the transmission power falloff is brought to a minimum and the voltage profile is modified and the operating and physical constraints are satisfied. Note that shunt capacitors/reactors and tap-settings of transformers are discrete variables while and except other variables are continuous. Hence, the reactive power dispatch issue is nonlinear, non-convex has equality and inequality limitations and has discrete and continuous variables.
7 Optimal Reactive Power Control to Improve …
7.1
253
Introduction
By the elevated exploitation as well as loading of the grid transmission system and besides because of refined enhanced operating conditions the issue of voltage stability and voltage collapse draws increasing consideration. A voltage drop could be taken in the power systems or subsystems and could emerge very suddenly [1]. Constant controlling of the system status is hence necessary. The reason of the 1977 New York blackout has been substantiated to be the reactive power issue. The 1987 Tokyo blackout was accepted to be because of reactive power deficit as well as a voltage drop at peak load in summer. These facts have strongly shown that reactive power play an important role in the security of power systems as view of voltage stability. An appropriate compensation of system voltage profiles will improve the system securities in the operation and will decrease system losses [2]. The essential purpose of voltage regulation in the distribution system performance is to maintain the status voltage in the power system steady in the suitable scope. The desirable voltages could be acquired by either directly manipulating the voltage or reactive power flow which in its own right will influence the voltage collapse. The reactive tools usually employed for the voltage and reactive power control are on-load tap-changer (OLTC) transformers, steps voltage regulator and switched shunt capacitors [3, 4]. Such reactive tools are generally utilized on the basis of a presumption that power runs in just one direction and the voltage diminishes along the feeder, from the substation to the remote end. An OLTC transformer is a transformer with automatically changeable taps. The OLTC is a section of most of HV/MV substation transformers [5]. A shunt capacitor produces reactive power to make up for the reactive power demand and hence increases the voltage. Shunt capacitors could be installed in the substation (hereinafter referred to as substation capacitors) or along the feeder (hereinafter referred to as feeder capacitors). A steps voltage regulator is an autotransformer with automatically adjustable taps that is ordinarily installed when the feeder is too long in such a manner which voltage regulation with OLTC and shunt capacitors is not enough. Voltage and reactive power control entails suitable coordination between the extant voltage and reactive power control equipment [6]. Many distribution network operators (DNOs) control these equipment locally via use of customary controllers to keep the voltages in the distribution system about approved range while bring to a minimum the voltage collapse and power falloffs. Various techniques have been presented in order to get better voltage and reactive power control in the distribution system for programming and operation stages. Within this time, many scholars have presented the trouble of voltage and reactive power control in distribution power systems through concentrating on automated distribution power system, with off-line setting control or real time control. The offline setting control intends to explore a dispatch program for OLTC movement and capacitor switching on the basis of day-ahead load prediction, in the meantime
254
A. Ghasemi Marzbali et al.
the real time control endeavors to control the capacitor and OLTC on the basis of real time surveying and trainings [7]. The major difficulty in utilization of the off-line setting control way is its affiliation to remote control and communication links to all capacitors. Nonetheless, a lot of DNOs do not communicate with links that are downstream to the feeder capacitor locations. In the other hand, the nature of modern power systems has changed due to a variety of factors: the increased demand for sustainability, rises in the price of oil and the need for the reduction of greenhouse gases, all of which have driven a large increase in the level of wind generation in the power system. The intergovernmental panel on climate change has cited that wind energy will be the primary source of renewable generation in the electricity sector [8]. Wind generation in both Europe and the United States is the dominant renewable resource currently present in power systems. In Europe, wind energy is set to triple in penetration by the year 2020, with 15.7% of the continent’s total energy supplied by wind generation [9]. In the United States, there is currently 42,432 MW of installed capacity providing 2.3% of the U.S. electricity mix, with the number set to rise to 25% by the year 2025 [10]. With wind generation set to become a significant generation resource in power systems around the world, it will become increasingly important to fully understand its impacts and interaction with the conventional elements in power systems. In fact, the real power unit output is generally restricted by radius (Vt Ia ), as follows P2G þ Q2G ðVt Ia Þ2
ð7:1Þ
Constrain of field is circular (VtEf/Xs) at (0, −V2t /Xs). It can be defined as P2G
2 Vt2 Vt Ef 2 þ QG þ Xs Xs
ð7:2Þ
where, PG, QG, Vt, Ia, Ef and Xs are active power, reactive power, terminal voltage, armature current, Excitation voltage and Synchronous reactance of the synchronous generator, respectively. Fundamentally, power systems have been designed and operated around the concept of generation delivery from large synchronous machines. These machines have high levels of reliability and complex control systems that allow the system to maintain high levels of operational security. The correct operation and control of these machines across the full spectrum of time-frames is critical for maintaining reliable power system operation and stability [11–13]. Maintaining voltage stability requires that the various components and elements of the system can interact without issue across all of the timeframes of the stability spectrum. Wind generation will have significant impact across the power system stability time-frame and as wind generation becomes a more common source of generation in the system, new mitigation techniques will be necessary to continue
7 Optimal Reactive Power Control to Improve …
255
operating the power system in an assured and stable way [14]. While the effect of reactive power reserves (RPRs) on system stability is widely acknowledged, few studies have been conducted to investigate how RPR levels could be used to indicate the amount of voltage stability margin (VSM) [15]. In contemporary years, certain papers with mathematical algorithms have been presented to think out the reactive power dispatch (RPD) problem [16–19]. These algorithms, including Non-linear Programming (NLP), Newton method, Gradient method, Linear Programming (LP), Jacobian matrix, Quadratic Programming (QP), interior point methods and so on, have been fruitfully used for thinking out the RPD problem. However, certain drawbacks are still linked to them. The RPD problem is non-linear, non-differential and non-convex problem with more than one local optimum, while these methods work based linearizing which make them less efficient in finding the global optimum. On the other hand, some of these methods suffer from special shortages, such as mathematical complexity and insecure convergence (NLP), piecewise quadratic cost approximation (QP), convergences to local optima (Newton method), a simplified piecewise linear estimate (LP), etc. Also, their optimization process mainly depends on the initial solution and can easily fall into local optima. To overcome these disadvantages, different heuristic-based techniques are developed for solving the RPD problem. Population-based optimization techniques inspired by nature may be classed in two significant categories that are swarm intelligence and evolutionary algorithms. Many methods rooted in these techniques including fuzzy Adaptive Particle Swarm Optimization (FAPSO) [20], Real Genetic Algorithms (RGA) [21], Tabu Search (TS) [22], Particle Swarm Optimization (PSO) [23], Improved PSO (IPSO) [24], stochastic method [25], Hybrid Stochastic Search (HSS) [26], Differential Evolution (DE) [27], Artificial Bee Colony (ABC) [28] as well as other methods have been broadly employed in the problem of RPD. However, these methods appear to be proper approaches for the unravelment of the RPD variable optimization problem. Attention to use optimization methods in distributed generation is increasing [29, 30]. In [31], an optimization method employed for wind power is offered, by a primal-dual predictor corrector interior point method, employed to explore the operating points of a single WT in a WF. The voltage stability based on reactive power control has been reviewed in rest of this chapter. The real transmission power loss minimization is consider as function, meantime the permitted transformer capacity, voltage range and conductor current capacity are added as the loading limitations. Moreover, OLTC operation and voltage fluctuation index are also analyzed. The reactive power in term of voltage stability is shown with reactive power control and local voltage in which timing of feeder capacitors are monitored and is still broadly employed.
256
7.2
A. Ghasemi Marzbali et al.
Voltage Stability Based RPD Model
The different purposes of power system are sum of voltage deviations on load busses, system transmission falloffs, voltage stability, security, etc. Such purposes are contradictory in their essence and couldn’t be dealt with by customary single purpose optimization techniques. Generally, the RPD model could be explained as follows in mathematical terms: Problem Purposes • Purpose 1: Bringing to a minimum the total real power losses Transmission falloffs in the network can be stated as economic losses procuring no advantages. Thus, transmission falloffs are understood as a falloff in proceeds via the utility. The intensity of each falloff requires accurate estimation and applicable moves made to bring them to a minimum. When the transmission falloffs are stated with regard to bus voltages and associated angles, the falloffs could be stated based on Newton–Raphson as follows J1 ¼ Ploss ðx; uÞ ¼
NL X
gk ½Vi2 þ Vj2
2Vi Vj cosðhi
hj Þ
ð7:3Þ
k¼1
i and j end respectively if gk is the conductance of the line i-j, Vi and Vj are line voltages and hi and hj are the line angles at the line. The kth network branch is k. It connects bus i to bus j. If ND is the set of numbers of power demand bus and j = 1,2, …, Nj where, Nj is the set of numbers of buses in adjacency with bus j, i = 1,2, …, ND. PG is the active power in line i and j. x and u are the dependent variables vectors and vector of control variables, respectively. • Objective 2: Bringing voltage deviation to a minimum Satisfying user’s demands with the smallest expense with a desirable continuity of supply and sufficiently little deviation in voltage is the second function of one RPD problem. The following is its expression J2 ¼ VDðx; uÞ ¼
Nd X
jVi
1:0j
ð7:4Þ
i¼1
where, Nd is number of load buses. • Objective 3: Minimization of L-index voltage stability Voltage stability and voltage drop problem draws increasing attention by increasing the power transmission system loading and exploitation, a voltage collapse could
7 Optimal Reactive Power Control to Improve …
257
happen in systems or subsystems. And it could emerge very suddenly [15, 34]. Lindex, Lj of the jth bus could be stated by means of below equation 8 < :
Lj ¼ 1
Fji ¼
N PV P
Fji VVij ; j
i¼1 ½Y1 1 ½Y2
¼ 1; 2; . . .; NPQ
ð7:5Þ
where, NPV and NPQ are numbers of PV and PQ buses, respectively. Y1 and Y2 are the YBUS system sub-matrices acquired following the segregation of PQ and PV bus parameters as shown in the below equation
Y1 Y3
Y2 Y4
VPQ VPV
I ¼ PQ IPV
ð7:6Þ
where L-index is calculated for the whole load buses. Lj shows there weren’t load case and voltage drop circumstances of bus j in a feasible numerical range of [0, 1]. Thus, a representative L delineating the fixedness of the complete system is formulated as follows L ¼ maxðLj Þ; j ¼ 1; 2; . . .; NPQ
ð7:7Þ
In the optimal RPD problem, incorrect tuning of continuous and discrete control variable settings might boost the L-index value, which may reduce the system voltage fixedness outskirt. Let the maximum value of L-index be denoted as Lmax. Therefore, to improve the voltage fixedness and to keep the system remote from the voltage drop margin, the formula for the succeeding purpose function will be as follows J3 ¼ VLðx; uÞ ¼ Lmax
ð7:8Þ
Objective Constraints • Limitations 1: Equality Limitations Power balance is equality limitations. To rephrase, the total power generation (PG) must cover the total demand (PD) as well as total real power losses in transmission lines. Equality constraints of real and reactive power in each bus can be expressed as below 8 > > > < PGi
> > > : QGi
PDi ¼ Vi
NB P
Vj ½Gij cosðhi
j¼1 NB P
QDi ¼ Vi
j¼1
hj Þ þ Bij sinðhi
hj Þ ð7:9Þ
Vj ½Gij sinðhi
hj Þ
Bij cosðhi
hj Þ
258
A. Ghasemi Marzbali et al.
where; NB and QGi are the numbers of buses and the reactive power produced for ith bus, respectively; PDi and QDi are real and reactive power at the ith load bus, respectively; Gij and Bij are the transfer conductance and susceptance between bus i and bus j, respectively; Vi and Vj are the voltage intensities at bus i and bus j, respectively; and hi and hj are the voltage angles at bus i and bus j, respectively. The equality limitations in (7.9) are nonlinear equations which could be thought out by employing Newton-Raphson method to create an answer to the load flow problem. Within the duration of answering, the real power output of one generator, titled slack generator, remains to fill in the real power losses and satisfy the equality limitation in Eq. (7.9). The load flow answer produces all bus voltage intensities and angles. Therefore, the real power losses in transmission lines could be obtained using Eq. (7.3). • Limitation 2: Generation Capacity Limitations For solid performance, the generator reactive power and bus voltage can be constrained through lower and upper limits as below Qmin Qi Qmax ; vmin vi vmax i i i i
ð7:10Þ
max , vmin and vmax are the minimal and maximal value for reactive where, Qmin i , Qi i i power and voltage magnitude of the ith transmission line, respectively. A clarified input/ output curve of the thermal unit understood as heat rate curve is indicated in Fig. 7.1.
• Limitation 3: Line flow Limitations A significant limitation of RPD problem is the line limitation. Since any line has a constrained capacity for current power, the constraint should be checked following the power system load flow. For that reason, this section argued the answer for RPD problem with line flow limitations. The following is the modeling of this constraint Fig. 7.1 Operating expenses curve for one generator
7 Optimal Reactive Power Control to Improve …
SLf ;k Smax ; k ¼ 1; 2; . . .; L Lf ;k
259
ð7:11Þ
where SLf,k is the real line k power flow; Smax Lf ;k is the upper limit of power flow of line k. L is the number of transmission lines [3]. • Constraints 4: Discrete control variables The shunt susceptance (Bsh) and transformer tap settings (Ti) values are taken as discrete values. These must be constrained by their lower and upper limits as below
Timin Ti Timax max Bmin shi Bshi Bshi
ð7:12Þ
Problem formulation Adding up the entire purpose performances and the equality and inequality limitations, a nonlinear limited multi-purpose optimization problem in mathematical terms could be the formula for the RPD problem, which can be represented by JFinal ¼ min½VLðx; uÞ; Ploss ðx; uÞ; VDðx; uÞ PG
s:t: : gðx; uÞ ¼ 0 hðx; uÞ 0
ð7:13Þ
xT ¼ ½½VL T ; ½SL T ; ½QG T uT ¼ ½½VG T ; ½QC T ; ½TT JFinal ¼ min½VLðx; uÞ; Ploss ðx; uÞ; VDðx; uÞ PG
where, g and h are the equality and inequality limitations, respectively. [VL], [QG] and [SL] are the vectors of load bus voltages, generator reactive power outputs and the transmission line loadings, respectively. [VG], [T] as well as [QC] are the vectors of generator bus voltages, transformer taps and reactive compensation instruments, respectively.
7.3
Reactive Power Capacity and Control Options in Wind Farms
The requisite of the membership of WFs in grid control matters has increased the inclusion of power electronics and the expansion of new WT generation concepts, leading to variable speed wind turbines [30]. The most popular wind generation technology employed among them today is the DFIG. In this study, the DFIG
260
A. Ghasemi Marzbali et al.
Fig. 7.2 Q property of an individual WT appertaining to G80-2.0 MW
Table 7.1 Properties of wind power Parameter
Value
Parameter
Value
Diameter Rotational speed Type
80 m 1680 rpm DFIG
PN Voltage Power factor
2.1 MW 0.68 kV 0.978 CAP–0.957 IND
technology is employed. The capacity of reactive power injection into the grid generally is associated to the used control approach, the active power generation and the converter size. The P-Q quality of the WTs employed in this study is given in Fig. 7.2. Specifically, it shows the business wind generator Gamesa WT G80-2.0 MW [26]; the useful data are given in Table 7.1. This WT acts with Power Factor (PF) 0.98 capacitive and 0.96 inductive. Consequently, the reactive power capability is bounded relying on the active power production. Its reactive power capacity is shown with the WT’s trait and the influences of lines as well as cables. Thus, the WFP-Q trait is analogous to the same for the WTs. However, it’s oriented to the capacitive side, as displayed in Fig. 7.3. The green line shows the PQ trait of WF for PF = 1. For a power production less than 10 MW, the cable influence is larger than the transformer influence, since the transformer influence is paramount for large output power from WF. Thus, novel consumption and production of WF areas are modeled, for a range in which the WF alters the reactive power requisites exists. If the WF gets a capacitive reactive power in the small active power production range, the WTs set point will alter to be inductive. Voltage control can be performed via reactive power injection as well as transformer taps, like the way it is given in Table 7.2.
7 Optimal Reactive Power Control to Improve …
261
Fig. 7.3 Q property of the tested WF comprised of twelve G80-2.0 MWWTs, with no compensation gear
Table 7.2 Voltage and reactive power control alternatives in wind farms Model
Control variable Voltage Reactive power
Capacitor STATCOM SVC DFIG Sync Tap
7.3.1
● ● ● ● ● ●
Discrete (Dis)/Continuous (Con)
Response time
Dis Con Con Con Con Dis
Slow Fast Fast Fast Fast Slow
Objective Function
As the WF has to be capable to acquire reactive power from the grid, the optimum handle or reactive power resources in the WF must be handled. For optimizing the RPD among WTs and the control of various equipment like STATCOM or capacitor bank, one HBMO algorithm can be used. This fitness is given to bring the active power drops to a minimum along the WF cables or lines min JðVarx ; Vary Þ ¼ min Plosses
ð7:14Þ
where, Vary denote the transformer tap location, the STATCOM reactive power setting and the capacitor bank state as independent variables, and Varx represent the dependent variables that are the single WT reactive power outputs. The control parameters are given by a j-dimensional vector, where j is the number of the enhanced variables. Meanwhile, any vector denotes one solution. i solutions exist and each one is a nominee answer.
262
7.3.2
A. Ghasemi Marzbali et al.
Objective Constraints
The reactive power of each WT, tab of transformers and the STATCOM reactive power are constrained by min max VarWT VarWTi VarWT ; i ¼ 1; 2; . . .; NG i i
ð7:15Þ
Timin Ti Timax
ð7:16Þ
min max VarStatcom VarStatcom VarStatcom
ð7:17Þ
Moreover, an additional equality limitation exists. The reactive power prerequisite in Point of Common Coupling (PCC) for the voltage control task can be modeled as an equality limitation by meas VarPCC ¼ VarPCC
ð7:18Þ
Within this study, the exploration of possible answers is employed to certificate an answer which meets the limitations that can be defined as follows SOki þ 1
8 < SOki þ vki þ 1 ; ¼ SOmax ; i : min SOi ;
SOmin SOki þ vik þ 1 SOmax i i k SOi þ vki þ 1 [ SOmax i SOmin [ SOki þ vki þ 1 i
ð7:19Þ
With Eq. (7.9), the inequality restraints are met, the equality restraint (7.18) still needs to be answered. Because in the WF the main elements influencing the reactive power flow are the reactive power consumption of transformers, a scheme for meeting the equality constraint is introduced. Yet, bearing in mind the goal to
Fig. 7.4 Flowchart for the possible answer exploration process
7 Optimal Reactive Power Control to Improve …
263
decrease the searching CPU time to explore a possible answer, the equality constrain is improved and an error value ɛ is modeled. meas Var VarPCC \e ð7:20Þ PCC This strategy is given in Fig. 7.4.
7.4
Voltage Stability Based RPP Model
Voltage deviation and stability limited VAr programming or Reactive Power Planning (RPP) is a significant demanding effect in power systems [32]. Investigations on voltage fixedness are fundamentally related to reactive power compensation sources. Having adequate reactive power compensation sources, principally in the shunt connection, the voltage stability boundary could be boosted a lot to make sure of system security [33]. Thus, appropriate programming of reactive power is a serious matter because of its specialized harness and the large expense of relocating a shunt compensator in economic terms when that is assembled. As an important demanding problem in power system research, RPP or VAr programming is a mixed integer nonlinear optimization programming with a huge scale of optimal variables [34]. The VAr programming is delineated in this part of applicable operational account limitations at various load levels. The mathematic formulation of optimization-based RPP model via accounting for voltage fixedness limitation could be stated as follows min a
ðkÞ
P
k¼1;...;NL N P
Ploss þ b
P
i¼1;...;N
yi ¼ N c
i¼1 ðkÞ PGi ðkÞ
ðkÞ
ðkÞ
Vi
PLi
ðkÞ
QGi þ QCi ðyi Þ ðkÞ
ðkÞ
P
ðkÞ
Vimin Vi Vimax ; ðkÞ
ðkÞ
ðkÞ
ðkÞ
ðkÞ
Vj
jwi ðkÞ Q Li
Qci yi þ c
Nd P
Vi
i¼1
Vi0 ; s:t::
ðkÞ
ðkÞ
ðGij cos hij þ Bij sin hij Þ ¼ 0 ðkÞ
Vi ðkÞ
P jwi
ðkÞ
Vj
ðkÞ
ðkÞ
ðGij sin hij
ðkÞ
Bij cos hij Þ ¼ 0
ð7:21Þ
ðkÞ
Sijmax Sij Sijmax
ðkÞ
PGi min PGi PGi max ; i 2 NG ðkÞ
QCi min QCi QCi max ; i 2 NG ðkÞ Ptieline
ðkÞ TTCðQC Þ
and, ðkÞ
Ploss ¼
X
ðkÞ
ðkÞ
gij ððVi Þ2 þ ðVj Þ2
ðkÞ
ðkÞ
ðkÞ
2Vi Vj cos hij Þ
ð7:22Þ
where, NG, N, NL and Nc, are number of generators, number of buses, number of operation load levels and number of VAr sources which must be installed, respectively.
264
A. Ghasemi Marzbali et al.
Subscript k indicates different load levels which k = 1,…, NL. T(k), a, b and yi are the time period of load level kth year, the energy expense per kWh, the calculated mean yearly preservation and assignment expense and binary variable (yi = 1” if the VAr tool is installed at bus i, or else, “0”), respectively. PGi, QGi, PLi, and QLi are the generator active outputs, reactive power outputs, the load active and reactive power demands at bus i, respectively. Skij is the line flow of line i-j where load level is k; other than for the value of zero, Qci is housed in a certain VAr capacity interval at bus i, Qkci is the VAr capacity needed at load level k, and Qci is the ultimate VAr size at bus i; Ptieline is the total active power flow with the tie lines from the source zone to sink zone; TTC(Q(k) c ) is a piecewise linear interpolation function employed as static voltage fixedness limitation to be delineated later in this research, and Qc is the Qci set at all nominee buses. The Eq. (7.3) brings the yearly expense of power system, voltage deviation and real falloffs plus VAr tool to a minimum. Nonetheless, additional purposes could be accounted for namely bringing generation expense and VAr installation expense, etc. to a minimum However, such points do not alter the point of focus in this study, and other purpose performances could be accounted for if it is needed in a certain power system. The mentioned VAr programming plane is static voltage stability (SVS) limited OPF model. The important cause for the computational demand of the voltage fixedness limited OPF model is the requisite of two constraints and variables sets associated to the regular performance and drop points [32]. The two variables sets offer a demand to answer the optimization model, particularly for a huge power system having several conditions. To depict this demand, TTC could be employed to roughly offer the SVS limitation. At a stable VAr compensation model by accounting for certain plausible conditions that are clarified in advance, the security limited TTC optimization model is given by 9 8 > > > > > > > > = < P ðkÞ min max ðPij Pij0 Þ ; s:t:: > k > > > i 2 Source Area > > > > ; : j 2 Sink Area N P yi ¼ N c i¼1 ðkÞ ðkÞ ðkÞ ðkÞ P ðkÞ ðkÞ Vj ðGij cos hij þ Bij sin hij Þ ¼ 0 PGi PLi Vi ð7:23Þ jwi ðkÞ P ðkÞ ðkÞ ðkÞ ðkÞ ðkÞ ðkÞ Vi Bij cos hij Þ ¼ 0 Vj ðGij sin hij QGi þ QCi ðyi Þ QLi jwi ðkÞ ðkÞ ðkÞ ðkÞ ðkÞ ðkÞ Vimin Vi Vimax ; Sijmax Sij Sijmax ðkÞ ðkÞ ðkÞ PGi min PGi PGi max ; i 2 Source Area ðkÞ ðkÞ ðkÞ QCi min QCi QCi max ; i 2 Source Area ðkÞ ðkÞ PLi QLi 0
PLi
¼
QLi
; i 2 Sink Area
7 Optimal Reactive Power Control to Improve …
265
where, Q0Li and P0Li are the base case reactive and real power demands at load bus i, respectively. Pij0 expresses the base case power flow between line i-j; and P(k) ij indicates the line flow of line i-j, having Var compensation. Meanwhile, the superscript k = 1,…, Ncntg, characterizes various performance statuses with the regular performance, and k > 0 characterizing the post-condition statuses for the kth condition phenomenon.
7.5
Simulation
A. Voltage stability in wind farm Wind Energy is one of the greatly encouraging renewable energy sources in Mongolia. WF within Inner Mongolia owns 200 wind turbines which are broken down into 20 sets, and there are 10 turbines in any set. The total installed capacity is as high as 300 MW. The wind turbine increases to 35 kV by box-type transformer T2, afterwards by 20 35 kV coupled transmission lines to the substation, which is comprised of two principal parts of the 220 kV transformer step-up substation. There are 403 nodes in the wind farm model, because of its sophistication and restricted area [35], only 42 nodes were selected as shown in Fig. 7.5. To account for cable lines losses, Static VAr compensation (SVC) is installed in the nodes for simulation. By computing the wind farm in the network, the optimum reactive compensation could be answered. Reference power SB is 100 MW, reference voltage is 220 kV, and the largest reactive investment is 500 million. Encrypt the SVC and on-load tap modifier. 50 is the highest code, and 100 is the highest number of disasters. Table 7.3 shows the compared of reactive power optimization between Traditional Genetic Algorithm (TGA), Improved Genetic Algorithm (IGA) and Honey Bee Mating Optimization (HBMO). The IGA reduces the network losses and financing of compensation gear. And it’s more appropriate than TGA. B. Voltage stability founded on RPP The IEEE 118-bus system [37, 38] is employed for case study, as indicated in Fig. 7.6. This power system is altered by decreasing maximal generator reactive power output and boosting reactive load, as indicated in Tables 7.4 and 7.5. This adjustment is employed in order to build the system stressed enough in the manner that the reactive power compensation is required. Therefore, the voltage intensity at bus 30 in the altered system is less than 0.944 p.u. in the case with heavy loading. The altered system information are employed as the “heavy-load” case. Afterwards, loads are moved down by 0.78 to construct the “medium-load” case, and moved down more by 0.76 to make the “light-load” case. In the total 8760 h yearly, heavy-load and light-load cases are presumed to be 1200 h each condition, and therefore, the medium-load case gets 6360 h. a, b and v are 40 $/MWh, 3600 $/year and 6400 kV for each VAr tool.
266
A. Ghasemi Marzbali et al.
Fig. 7.5 The model of the wind farm having 42 nodes
Table 7.3 Contrasted conclusions of reactive power optimization in wind farms Project
Financing of reactive power compensation (million Yuan)
The system falloff (kW) V = 4 m/s V = 8 m/s V = 12 m/s
TGA [36] IGA [36] HBMO [1]
338 336 311
1872 1731 1721
2480 2292 2282
3129 2892 2671
In this section, certain indices founded on voltage fixedness are indicated to mention new nominee VAr tools. Furthermore, we must employ the indices that could provide data at all load buses. Hence, three indices constituted voltage and voltages fixedness associated information of any bus to appear in the index dataset matrix for fuzzy grouping. They are L-index, H/H0 Index, and ultimately a voltage fluctuation index. (1) H/H0 Index: This index formulated in Eq. (7.24), is used to grant the test system delicate bus associated data.
7 Optimal Reactive Power Control to Improve …
267
Area
Fig. 7.6 IEEE 118-bus system
Table 7.4 Altered reactive load demand in the IEEE 118-bus system PQ bus no
Q MVar
PQ bus no
Q MVar
PQ bus no
Q MVar
2 3 5 7 9 11 13 14 16 17 20 21 22 23 28 29
24 43 6 24 5 75 39 19 30 17 23 19 15 13 21 28
30 33 35 37 38 39 41 43 44 45 47 48 50 51 52 53
5 26 36 5 4 36 45 21 21 58 32 24 21 21 23 28
57 58 60 63 64 67 68 71 75 78 79 81 82 83 84 86
17 17 83 5 5 33 5 5 51 72 41 5 55 25 16 26
268
A. Ghasemi Marzbali et al.
Table 7.5 Altered generator maximal reactive power output
Gen no
QGmax
Gen no
QGmax
Gen no
QGmax
1 4 6 8 10 12 15 18 19 24 25 26 27 31 32 34 36 40
12 295 46 295 195 115 25 45 20 296 136 996 296 296 38 21 21 295
42 46 49 54 55 56 59 61 62 65 66 69 70 72 73 74 76 77
295 95 205 295 18 10 175 295 15 195 195 295 27 97 97 5 17 67
80 85 87 89 90 91 92 99 100 103 104 105 107 110 111 112 113 116
277 19 997 297 297 97 5 96 151 36 19 19 197 19 997 997 197 997
H H1 Hk ; . . .; ¼ H0 H01 H0k
ð7:24Þ
where, H and H0 are base case voltage intensity vectors at each bus and the voltage intensity with all loads adjusted to 0 for the power system [39]. (2) L-Index: With the boosted loading and utilization of the power transmission system and also because of enhanced optimized performance, the issue of voltage fixedness and voltage drop draw increasing attention. A voltage drop could happen in systems or subsystems and could emerge very suddenly. L-index, Lj of the jth bus could be stated by using the equation below 8
miVdc, Q will be positive that causes STATCOM to absorb reactive power or vice versa [9]. The active power generated by the source to charge the capacitor is expressed as follows, P¼
jVS jmi Vdc sinðdS XS
aÞ
ð8:28Þ
The 24-pulse converter topology of STATCOM consists of six-pulse converters where it eliminates the 6n ± 1 ordered harmonics. The 23rd, 25th, 47th, and 49th harmonics in the ac output waveform are naturally eliminated since the n = 4 for 24-pulse converter. The 24-pulse STATCOM constituting two-level VSCs is illustrated in Fig. 8.21 where this topology is also known as quasi 24-pulse [9, 33]. Each VSC and transformer connection configures a stage that each one is connected to the transmission line by Y-Y and ∆-Y connections of transformers. The transformers may connected in zigzag that generates 15° phase shift according to each other with −7.5°, 7.5°, 22.5°, and 37.5° angles [9, 33, 34]. Another topology of quasi 24-pulse STATCOM is presented in Fig. 8.22 [35] where the VSCs are constituted according to neutral point clamping (NPC) topology that is previously shown in Fig. 8.18b. The conduction angle a defines the negative and positive cycles of intervals where a = 180 − 2b, where 2b stands for the dead-band angle. The on-off spectral transitions are ensured by the dead-band angle where the output voltages of a six-pulse three-level VSC is given by Eq. (8.29).
Fig. 8.21 Circuit diagram of a quasi two-level 24-pulse STATCOM
8 Reactive Power Compensation in AC Power Systems
301
Fig. 8.22 Circuit diagram of a three-level 24-pulse STATCOM
pffiffiffi 2 6 vdc 180 2b sin V1 ¼ 2 p 2
ð8:29Þ
On the other hand, output voltages of a 24-pulse three-level VSC is given as follows while the active and reactive powers are calculated as given below by considering XS as the reactance along with STATCOM and the transmission line by using Eqs. (8.31) and (8.32) [9, 35] pffiffiffi 2 6 vdc 180 2b p sin cos Vc1 ¼ np 2 24 6p 2 P¼ Q ¼ VS
VL VS sin dS XS
VS
VL sin dS XS
ð8:30Þ ð8:31Þ
ð8:32Þ
where the phase angle d between the ac bus voltage VS and STATCOM output voltage VL sustains the dc line voltage Vdc at a constant value.
302
8.4.1.2
E. Kabalci
Multilevel Converter Based STATCOM
In addition to their power conversion applications, the multilevel converters (MLC) are widely used in active filters, and FACTS. Although the primary applications are improved for medium power ranges, the recent applications have been increased the power range from a few kWs to multi MWs owing to their cascaded topologies. Almost all the three-level STATCOM topologies are implemented with conventional diode clamped (DC) topology where the switching devices were GTOs. This approach indicates that the control strategy of DC is the simplest one among other three-level MLCs in medium and high power application. The flying capacitor MLC (FC-MLI) that is one of the outstanding topology allows constructing low cost systems, particularly in high power applications, comparing to the DC. However, it involves pre-charge arrangement for the capacitors that are connected through the upper and lower arms. The cascaded H-bridge (CHB) topology is the most recent topology among others that extended its operation area since it provides higher power conversion and lower cost for a given output magnitude level. The CHB consists of several H-bridge cells connected in series to generate multilevel output voltage. One of the most significant features of this topology is higher switching frequency comparing the previous topologies, and increased power rate handled by the total device and equally shared to the each cell. Furthermore, the CHB based STATCOM is capable to eliminate harmonics and to compensate reactive power in superior comparing others. The main requirement of this configuration is its separated dc sources. However, CHB can be easily integrated to various type of energy sources to obtain higher output levels and energy storage applications [9, 36–39]. Figure 8.23 shows the circuit diagram of STACOM configured by using three-phase five-level CHB including series connection of three-level H-bridge cells at each phase leg. The output voltage synthesis of H-bridges generate a combined phase voltage in staircase waveform oscillating in the ranges of +Vdc, 0, −Vdc. Where Va1 and Va2 voltages of series connected H-bridge cells synthesize the phase voltage Van in a stepped waveform as seen in Fig. 8.24. The positive levels of output voltage are indicated with P1 and P2 while the negatives are P11 and P12. Equation (8.33) expresses the Fourier series expansion of the general multilevel stepped output voltage while the transform applied to the STATCOM in Fig. 8.23 is shown in Eq. (8.34), where n represents harmonic order of the output voltage. Van ðtÞ ¼
1 4Vdc X sinðnxtÞ ½cosðna1 Þ þ cosðna2 Þ þ . . . þ cosðna5 Þ n p n¼1;3;5;...
ð8:33Þ
The switching angles a1, a3,…, a5 in Eq. (7.33) are selected for the minimum voltage harmonics [36, 40].
8 Reactive Power Compensation in AC Power Systems
303
Fig. 8.23 Circuit diagram of a five-level CHB STATCOM
Van5 ðtÞ ¼
1 4Vdc X sinðnxtÞ ½cosðna1 Þ þ cosðna2 Þ n p n¼1;3;5;...
ð8:34Þ
Researchers have proposed novel MLCs based on hybrid topologies that are based on the existing devices [36, 40].
8.4.2
Static Synchronous Series Compensator (SSSC)
The series capacitor based compensation that brings some capabilities such as increasing the transient stability, reactance control, and load sharing is a conventional technique that was introduced in Sect. 8.3.3. Even though the series compensation is assumed to be used to decrease the reactive impedance of the transmission line, it is actually operated by increasing the line voltage, current, and thereby the delivered power from source to load. Although the mentioned advantages, series compensation has several drawbacks on SSR issues that large energy exchange is done by several sources under synchronous system frequency.
304
E. Kabalci
Fig. 8.24 Phase voltage generation of a five-level CHB STATCOM
There are several studies are performed to overcome SSR problem. The robust control methods and configurations of FACTS stand out of other methods proposed by researchers. The SSSC has been developed as an alternative to the TCSC devices where the SSR immunity is inherited from TCSC. In addition to regular advantages of series compensators, SSSC presents numerous capabilities to increase the reactive power control and stability issues. Furthermore, there are several studies are noted that SSSCs effectively damps the SSR comparing regular series compensators [6, 11, 14, 41–46]. Figure 8.25 illustrates basic circuit diagram of a SSSC based on VSC similar to STATCOM that is serially connected to the transmission line, and a capacitor on the dc side. The SSSC controller considers the line current and line voltage to compensate the transmission line by adjusting the reactance in steady-state operation. It should be noted that in case the current is flowing in the line, then SSSC generates compensation voltage VC that is quadrature to the current and regulates the reactive power demand. The compensation characteristic of the device is same with the one shown in Fig. 8.11 for 0 < d < 180° during series compensation that yields positive magnitude power delivered.
8 Reactive Power Compensation in AC Power Systems
305
Fig. 8.25 Static synchronous series compensator (SSSC)
The fundamental operating principles of SSSC is compared to conventional series capacitive compensator as shown in Fig. 8.26 where the circuit diagrams for only series compensation with only capacitive compensator (Fig. 8.26a) and with SSSC (Fig. 8.26c), and respective phasor diagrams are analyzed. Despite it is assumed to decrease the line impedance, a series capacitor increases the line voltage, line current, and consequently the power through the line impedance. This approach may be considered irrelevant. However, the voltage along the transmission line is the significant parameter while the capacitor is cared to increase the voltage in order to provide the required current level. The steady-state power transmission can be realized when a synchronous voltage source follows the series compensation as shown in Fig. 8.26c that is analytically presented as follows Vq ¼ VC ¼
jXC I
ð8:35Þ
It should be noted that there are particular differences between series capacitive compensator and synchronous voltage source type compensator. The operational characteristic of a series capacitor can be illustrated with an ideal voltage source. However, SSSC is significantly different from the series capacitor in terms of delivered power and angle characteristics, active power exchange, and immunity to resonance and SSR, and control ranges. Capacitor provides reactive impedance that causes proportional voltage to the line current when it is series connected to the line. The compensation voltage is changed regarding to the transmission angle d and line current. The delivered power PS is a function of the series compensation degree s where it is given by Pq ¼
V2 sin d XL ð1 sÞ
ð8:36Þ
where V = V1 = V2 and d = d1 − d2. The delivered power PS regarding to transmission angle d is shown in Fig. 8.27a as a function of s (s = 0, s = 1/3, and
306
E. Kabalci
Fig. 8.26 Symmetrical system under series compensating, a circuit diagram of series compensating capacitor, b phasor diagram of capacitor, c circuit diagram of SSSC, d phasor diagram of SSSC
s = 1/5). Additionally, SSSC is independent from line current while compensating the voltage. The delivered power Pq is a parametric j90 function of injected voltage Vq ¼ Vq series to line where V q ¼ Vq ½I= I e that is also given by Pq ¼
V2 V d sin d þ Vq cos XL 2 XL
ð8:37Þ
The delivered power Pq is shown in Fig. 8.27b as a parametric function of injected voltage Vq against transmission angle d. The normalized Vq voltage is selected to provide same maximum power with series capacitor configuration. The comparison of Fig. 8.27a, b expresses that the series capacitor increases the delivered power by a fixed rate of the delivered by the uncompensated line at a given transmission angle d. On the other hand, the SSSC increases the delivered power by a fixed fraction of the maximum power available by uncompensated line independent from d in the angle range of 0 d 90° [41, 45–47].
8.4.3
Unified Power Flow Controller (UPFC)
The UPFC is a VSC based shunt-series compensator that is constituted of STATCOM and SSSC combination with a common dc link as shown in Fig. 8.28.
8 Reactive Power Compensation in AC Power Systems
307
Fig. 8.27 Delivered power P and transmission angle d relation, a series capacitive compensation, b SSSC
Fig. 8.28 The block diagram of UPFC
Separate observers of each compensator perform the control of UPFC. The shunt and series compensators of UPFC exchange reactive power at the terminals while they are operating as independent FACTS compensators. In this case, SSSC injects a voltage quadrature to the line current, and imitates a series inductive or capacitive reactance to the transmission line. On the other hand, STATCOM injects reactive current in order to imitate shunt reactance to the line while operating stand-alone. In case of the combination as a UPFC), the series compensator voltage can be at any
308
E. Kabalci
phase corresponding to the predominant line current. Thereby, the exchanged power can be reactive as well as active power at the terminals of compensators. The active power exchanged by any of the compensators is transmitted to the other compensator across the common dc-link. However, exchanged reactive power is independent from each other compensator. The controller of STATCOM dynamically adjusts the phase angle between the VSC output voltage and line voltage to get the STATCOM generated or absorbed desired reactive power at the compensation connection. The SSSC can be operated in several modes such as voltage injection, regulation of phase angle, emulation of the line impedance, and automatic power flow control. In any of these applications, SSSC injects a voltage over its series connection to the line. The SSSC can be operated in the same way as examined in the previous section to control active and reactive current. The active power control would fluctuate capacitor voltage that should be observed by STATCOM [6, 11, 14, 26, 47]. The controller of STATCOM dynamically adjusts the phase angle between the VSC output voltage and line voltage to get the STATCOM generated or absorbed desired reactive power at the compensation connection. The maximum value of series voltage VC is limited by SSSC that is the most significant advantage comparing to thyristor-based compensators. In the case reactive power exchange, the UPFC increases the source voltage VS while generating reactive power, and decreases while absorbing the reactive power. The power flow of UPFC through the transmission line is analyzed as follows; PS ¼ Psh þ
imax OFimin
:
1
OFi OFimin min OFi \OFi \ OFimax OFi OFimax
OFimax
OFi
ð9:57Þ
for maximized target functions. Where OFimin and OFimax are the minimum and the maximum magnitude of ith target function among total non-dominated solutions, respectively. The membership equation v is changed between 0, 1. Where v ¼ 0 demonstrates the incompatibility of the solution with the set, while v ¼ 1 presents full compatibility. Figure 9.11 shows a typical structure of the membership equation. For each non-dominated solution k, the normalized membership equation vk is extracted as: PNob
vk v ¼ PM i¼1 PNobi k
k¼1
i¼1
vki
ð9:58Þ
where M and Nob are the number of non-dominated solutions and target functions. The function vk can be stated as a membership function of non-dominated solutions in a fuzzy set, where the solution having the maximum membership in the fuzzy collection is assumed as the best compromise solution.
9.5
Implementation
MOPSO-NTVE algorithm is applied to determine optimal reactive plan. A decision making method according to FDM algorithm is followed to determine the best solution from the collection of Pareto-solutions created by MOPSO-NTVE technique. Details of the solution synthesis are given in Figs. 9.12, 9.13 and 9.14. For this purpose, an initial population of MPSO-NTVE is randomly produced. For each particle, AC-OPF is directed and STC and VD are calculated. Then, the investment cost is gained. The calculation is repeated until the stopping condition is obtained. As shown in Fig. 9.15, in the RPM problem codification, each chromosome is
9 Optimal Placement of Reactive Power Compensators in AC Power …
START
A
Sub -algorithm for MOPSO -NTVE
Input network data
B
Sub -algorithm for fuzzy decision making
Report output of RPP as: 1) HFC size, setting and site 2) GUPFC size, setting and site 3) The rate of reactive power injection from slow VAR device
Encoding Verify objective functions related to RPP: a) Investment cost of RPP problem b) Voltage deviation c) STC-criterion
End Fig. 9.12 Flowchart of the proposed RPP
Generation i
A
l
x set
g
x set
Current population
Generate next population
Extended Pareto set Non-dominated sorting
Temporary set
Generation i+1
Non-dominated sorting
Updated x l set
Pareto set
Choose N random population from specified top part of first front
Next population
Updated x g set
Updated Pareto set
A Fig. 9.13 Sub-algorithm of A for design procedure delineated in Fig. 9.12
335
336
H. Shayeghi and Y. Hashemi
B Dominated solutions Non-Dominated solutions Pareto -front
f2
f2(x) Fuzzy mechanism
i f1
f1(x)
(1, OFi
min
)
1
OFi
B Fig. 9.14 Sub-algorithm of B for design procedure delineated in Fig. 9.12
0.12 11.1 18 0
1
0.7
19.3 3
121
12 1
0.5 323 22
0.62 18.4
0.23 Reactive power sources settings Reactive power sources sizes
26
Reactive power sources locations
1
Need or lack of need for reactive power sources Fig. 9.15 Simple codification for RPP problem
formed from four parts of number, location, size and settings of the candidate VAR devices to be installed. The proposed management method is implemented on a test system. The test grid employed in this work is a part of the American electric power system, AEP, applied in the Midwest in the early 1960s and is better identified as IEEE 57-node test system. The system data are available in MATPOWER toolbox [22]. The basic configuration of the test system is depicted in Fig. 9.16. The network as shown in Fig. 9.16 includes of 57 buses, 7 generators, and 80 lines. The generators are placed at nodes 1, 2, 3, 6, 8, 9, and 12. The voltage constraints are adjusted between 0.94 and 1.06 p.u.
9 Optimal Placement of Reactive Power Compensators in AC Power …
B5
337
B16
B4
B17
B20 B19
B18
B1
B2
B3
B15 B12
B21
B46
B6 B44
B14 B13
B47
B23
B48
B49
B50
B26 B45 B24
B57
B22 B25
B38 B36 B30
B27
B33
B39 B37
B56
B41 B11
B51
B40 B35
B28
B31
B32
B34
B7
B42
B43
B55 B29 B8
B53 B52
B9
B10
B54
Fig. 9.16 IEEE 57-bus system
The constraints have been enforced based on Eq. (9.59): Evaluation Function = Objective Function þ aðConstraints ViolationsÞ
ð9:59Þ
where a is a large value and constraint violations are analyzed as the aggregate of the absolute magnitudes of all deviations. The solution will end up with the least cost selection without violating limitations. Also, discrete variables have been encoded based on “ceil” operators. To analyze the effect of different FACTS tools on the power system operation, the following scenarios are studied: • Framework 1: Slow VAR device allocation. • Framework 2: Slow VAR device with HFC allocation. • Framework 3: Slow VAR device with GUPFC allocation. The performance of two frameworks is evaluated on the test system. For this purpose, the steps of the RPP problem (as given in Fig. 9.12) are performed. The Pareto-optimal archive obtained by MOPSO-NTVE algorithm in two-dimensional and three-dimensional target functions is depicted in Figs. 9.17, 9.18 and 9.19. It is observed that the gained solutions are distributed in the area, except some discontinuity, created by the discrete decision parameters. The trade-offs represented in
338
H. Shayeghi and Y. Hashemi
7 6 5 4 4.5
4.6
4.7
4.8
4.9
5
STC-Criterion (STC)
Voltage Deviation (VD)
Figs. 9.17, 9.18 and 9.19 can help the decision maker to choose appropriate reference membership magnitudes. A decision making approach by employing FDM theory is followed to determine the best solution as new lines from the collection of Pareto-solutions gained by MOPSO-NTVE method. FDM is used to choose size and site of the reactive power sources. Complete results related to two frameworks and four optimization methods, MOPSO, MOPSO-TVIW, MOPSO-TVAC and MOPSO-NTVE are tabled in Table 9.1.
0.84 0.82 0.8 0.78 0.76 4.5
4.6
4.7
6
0.82 0.8 0.78 4
4.5
5
5.5
6
6.5
7
STC-Criterion (STC)
STC-Criterion (STC)
0.84
0.76
4.8
4.9
5
INV ($*106/year)
INV ($*10 /year)
0.85 0.8 0.75 8
Volta
ge D
Voltage Deviation (VD)
6
eviati
on (V
4 4.4
D)
5
4.8
4.6
0 /year) INV ($*1 6
6 5 4 3 3.5
3.6
3.7
3.8
3.9
STC-Criterion (STC)
Voltage Deviation (VD )
Fig. 9.17 The Pareto archive in two-dimensional and three-dimensional objective area based on framework 1 0.4 0.38 0.36 0.34 3.5
3.6
6
0.38 0.36
3
3.5
4
4.5
5
Voltage Deviation (VD)
5.5
6
STC-Criterion (STC)
STC-Criterion (STC)
0.4
0.34
3.7
3.8
3.9
INV ($*106/year)
INV ($*10 /year)
0.4 0.35 6
Volta
ge D
4
eviati
on (V
2 3.4
D)
3.6
INV ($*1
3.8
4
0 /year) 6
Fig. 9.18 The Pareto archive in two-dimensional and three-dimensional objective area based on framework 2
5 4 3 2 1 1.9
2
2.1
2.2
2.3
2.4
2.5
STC-Criterion (STC)
Voltage Deviation (VD )
9 Optimal Placement of Reactive Power Compensators in AC Power … 0.18 0.16 0.14 0.12 0.1 1.9
2
2.1
6
0.16 0.14 0.12 2.5
3
3.5
4
4.5
STC-Criterion (STC)
STC-Criterion (STC)
0.18
2
2.2
2.3
2.4
2.5
INV ($*106/year)
INV ($*10 /year)
0.1 1.5
339
0.2 0.15 0.1 5
Volta
ge D
Voltage Deviation (VD)
eviati
D)
2.5
2
0 1.5
on (V
6 ear) /y V ($*10
IN
Fig. 9.19 The Pareto archive in two-dimensional and three-dimensional objective area based on framework 3 Table 9.1 Location, size and setting of reactive power sources added to network in frameworks under different solution methods
MOPSO
MOPSO-TVIW
MOPSO-TVAC
Framework 1 Slow VAR devices MVAr (bus)
Framework 2 Slow VAR devices MVAr (bus)
Framework 3 HFC Slow VAR MVAr devices (location) MVAr (bus)
Q16 = 423.57 Q4 = 355.56 Q46 = 436.56 Q21 = 365.56 Q49 = 367.97 Q39 = 314.56 Q22 = 431.56 Q11 = 478.03 Q32 = 434.67 Q41 = 414.76 Q57 = 432.45 Q39 = 421.45 Q43 = 431.56 Q44 = 476.67 Q14 = 365.76 Q50 = 432.67 Q12 = 391.67 Q18 = 409.65 Q46 = 434.74 Q50 = 446.76 Q49 = 346.76 Q39 = 476.65 Q41 = 421.76 Q56 = 365.74 Q31 = 418.76
Q14 Q19 Q35 Q55
= = = =
476.56 523.56 437.67 456.73
33.34 (13-9)
Q11 = 475.23 Q53 = 478.25
13.26 (24-25, 24-23)
Q48 Q56 Q43 Q30
= = = =
394.67 512.56 476.56 456.65
32.9788 (22-21)
Q27 = 487.67 Q10 = 421.78
12.6786 (45-44, 45-38)
Q11 = 535.87 Q35 = 465.76 Q32 = 421.85 Q7 = 365.87
32.6457 (23-22)
Q57 = 353.64 Q20 = 365.87
11.6543 (29-28, 29-52)
GUPFC MVAr (location)
(continued)
340
H. Shayeghi and Y. Hashemi
Table 9.1 (continued)
MOPSO-NTVE
Framework 1 Slow VAR devices MVAr (bus)
Framework 2 Slow VAR devices MVAr (bus)
Q4 = 529.32 Q20 = 440.84 Q46 = 377.67 Q50 = 457.34 Q38 = 401.83 Q24 = 277.65 Q43 = 357.67 Q53 = 401.86
Q21 Q14 Q11 Q33
= = = =
163.75 189.78 163.75 189.78
Framework 3 Slow VAR HFC devices MVAr MVAr (bus) (location) 27 (49-38)
Q = 229.32 Q40 = 340.84
GUPFC MVAr (location) 10 (15-13, 15-14)
Table 9.2 Frameworks results
Framework 1
Framework 2
Framework 3
Solution methods
RPM cost (INV) ($ 106/ year)
Voltage deviation (VD)
STC-criterion (STC)
MOPSO MOPSO-TVIW MOPSO-TVAC MOPSO-NTVE MOPSO MOPSO-TVIW MOPSO-TVAC MOPSO-NTVE MOPSO MOPSO-TVIW MOPSO-TVAC MOPSO-NTVE
4.9560 4.8598 4.7767 4.7257 4.0932 3.9517 3.8544 3.7713 2.9886 2.8361 2.3361 2.0085
5.3219 5.0032 4.9821 4.9532 4.3217 4.2345 4.1234 3.9865 2.6422 2.5432 2.4321 2.3456
0.7832 0.7755 0.7743 0.76432 0.3689 0.3578 0.3521 0.3456 0.1789 0.1675 0.1298 0.1234
To identify the advantage of the presented method, results of frameworks 1 and 2 in four solution methods are compared and tabled in Table 9.2. As shown in this table, RPP cost (INV) for framework 1 is: 4.9560, 4.8598, 4.7767 and 4.7257 and, for framework 2 it is: 4.0932, 3.9517, 3.8544 and 3.7713 and, for framework 3 it is: 2.9886, 2.8361, 2.3361 and 2.0085. RPP cost in framework 3 has 39.69, 41.64, 51.09 and 57.49% decrease compared to framework 1 and, 26.98, 28.23, 39.39 and 46.74% decrease compared to framework 2. This economic perspective of the presented way reveals priority of the obtained optimal configuration in framework 3 and the advantages of GUPFC than HFC and capacitor banks. From the viewpoint of voltage deviation (VD) and STC-criterion, these two indices in framework 2 are significantly improved than the framework 1 in each four scenarios as displayed in Table 9.2. Simulation test is applied for different discount rates to evaluate robustness of the proposed RPP problem. As shown in Table 9.3 twelve cases, C1 to C12, are studied. All the parameters for three discount rates (rT = 5, 10 and 20%) and three
9 Optimal Placement of Reactive Power Compensators in AC Power … Table 9.3 Considered cases
Case
Discount rate (%)
C.1 C.2 C.3 C.4 C.5 C.6 C.7 C.8 C.9 C.10 C.11 C.12
rT rT rT rT rT rT rT rT rT rT rT rT
Framework 1
= = = = = = = = = = = =
341
Solution methods
5 5 5 5 10 10 10 10 20 20 20 20
MOPSO MOPSO-TVIW MOPSO-TVAC MOPSO-NTVE MOPSO MOPSO-TVIW MOPSO-TVAC MOPSO-NTVE MOPSO MOPSO-TVIW MOPSO-TVAC MOPSO-NTVE
Framework 2
Framework 3
INV ($*10^6 /year)
6 5 4 3 2 1 0 C.1
C.2
C.3
C.4
C.5
C.6
C.7
C.8
C.9
C.10
C.11
C.12
C.1
C.2
C.3
C.4
C.5
C.6
C.7
C.8
C.9
C.10
C.11
C.12
C.1
C.2
C.3
C.4
Voltage Deviation (VD)
6 5 4 3 2 1
STC-Criterion (STC)
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 C.5
Fig. 9.20 Comparison of the performances
C.6
C.7
C.8
C.9
C.10
C.11
C.12
342
H. Shayeghi and Y. Hashemi
scenarios have been extracted, as depicted in Fig. 9.20. Results of both frameworks will change by increasing or decreasing the discount rate, but framework 3 has the best results and low changes than the nominal value of discount rate and achieves good robust performance.
9.6
Conclusion
An attempt has been made in this chapter to study solving RPP problem while HFC and GUPFC are involved in power system. A suitable model to compute reactive power that has to be injected by HFC and GUPFC is imported in the solution procedure. Multi-objective PSO-NTVE algorithm is employed and, from a decision maker perspective; the FDM method is used to define solutions by considering all attributes from the set of Pareto-solutions. The presented solution procedure is implemented on the IEEE 57-bus system as the first attempt for RPP. A comparative survey confirms that the presented management algorithm significantly improves the cost related to RPP. The results show that inclusion of the reactive power capability of HFC and GUPFC can improve voltage stability and voltage profile. The obtained results for various scenarios reveal that this planning method is a useful management tool for solving RPP issue. The presented planning method permits the power system designers to modify the structure of the system to obtain the best optimal program for the expanded system. Simultaneous planning of reactive power expansion with the presence of capacitor banks and FACTS devices instead of separate planning can be led to more economic, applicable and optimal scheme in a power network.
References 1. H. Shayeghi, M. Ghasemi, FACTS Devices Allocation Using a Novel Dedicated Improved PSO for Optimal Techno-Economic Operation of Power System, Journal of Operation and Automation in Power Engineering, 1 (2), 2013. 2. A. Rabiee, M. Vanouni, M. Parniani, Optimal Reactive Power Dispatch for Improving Voltage Stability Margin Using a Local Voltage Stability Index, Energy Conversion and Management, 59:66-73, 2012. 3. M. Mahfouz, M.A. El-Sayed, Static Synchronous Compensator Sizing for Enhancement of Fault Ride-through Capability and Voltage Stabilization of Fixed Speed Wind Farms, IET Renewable Power Generation, 8 (1):1-9, 2014. 4. R.A. Hooshmand, R. Hemmati, M. Parastegari, Combination of AC Transmission Expansion Planning and Reactive Power Planning in the Restructured Power System, Energy Conversion and Management, 55:26-35, 2012. 5. I.P. Abril, J.A.G. Quintero, VAR Compensation by Sequential Quadratic Programming. IEEE Transactions on Power Systems, 18 (1):36-41, 2003.
9 Optimal Placement of Reactive Power Compensators in AC Power …
343
6. T. Akbari, A. Rahimikian, A. Kazemi, A Multi-Stage Stochastic Transmission Expansion Planning Method, Energy Conversion and Management, 52 (8):2844-2853, 2011. 7. E. Ghahremani, I. Kamwa, Optimal Placement of Multiple-Type FACTS Devices to Maximize Power System Loadability Using a Generic Graphical User Interface, IEEE Transactions on Power Systems, 28 (2):764-778, 2013. 8. H. Shayeghi, M. Ghasemi, FACTS Devices Allocation Using a Novel Dedicated Improved PSO for Optimal Techno-Economic Operation of Power System, Journal of Operation and Automation in Power Engineering, 1 (2):124-135, 2013. 9. J. Pourhossein, G. Gharehpetian, S. Fathi, Static Inter-Phase Power Controller (SIPC) Modeling for Load Flow and Short Circuit Studies, Energy Conversion and Management, 64:145-151, 2012. 10. S.A.N. Niaki, R. Iravani, M. Noroozian, Power Flow Model and Steady-State Analysis of the Hybrid Flow Controller, IEEE Transactions on Power Delivery, 23 (4):2330-2338, 2008. 11. H. Shayeghi, Y. Hashemi, Technical-Economic Analysis of Including Wind Farms and HFC to Solve Hybrid TNEM-RPM Problem in the Deregulated Environment, Energy Conversion and Management, 80:477-490, 2014. 12. A. Lashkar Ara, A. Kazemi, S. Nabavi Niaki, Multiobjective Optimal Location of FACTS Shunt-Series Controllers for Power System Operation Planning, IEEE Transactions on Power Delivery, 27 (2):481-490, 2012. 13. Y. Hashemi, H. Shayeghi, B. Hashemi, Attuned Design of Demand Response Program and M-FACTS for Relieving Congestion in a Restructured Market Environment, Frontiers in Energy, 9:1-15, 2015. 14. B.S. Rao, K. Vaisakh, Multi-Objective Adaptive Clonal Selection Algorithm for Solving Optimal Power Flow Considering Multi-Type FACTS Devices and Load Uncertainty, Applied Soft Computing, 23:286-297, 2014. 15. Fardanesh B, Optimal utilization, sizing, and steady-state performance comparison of multiconverter VSC-based FACTS controllers. IEEE Transactions on Power Delivery, 19 (3):1321-1327. doi:10.1109/TPWRD.2004.829154, 2004. 16. K. Habur, D. OLeary, FACTS-Flexible Alternating Current Transmission Systems-for Cost Effective and Reliable Transmission of Electrical Energy, Siemens-World Bank Document-Final Draft Report, Erlangen, 2004. 17. M. Eghbal, N. Yorino, E. El-Araby, Y. Zoka, Multi-Load Level Reactive Power Planning Considering Slow and Fast VAR Devices by Means of Particle Swarm Optimization, IET Generation, Transmission & Distribution, 2 (5):743-751, 2008. 18. G. Yesuratnam, D. Thukaram, Congestion Management in Open Access Based on Relative Electrical Distances Using Voltage Stability Criteria, Electric Power Systems Research, 77 (12):1608-1618, 2007. 19. C. Coello, G. Lamont, D. Veldhuizen, Evolutionary Algorithms For solving Multi-Objective Problems, Springer, 2007. 20. F. Gunes, U. Ozkaya, Multiobjective FET Modeling Using Particle Swarm Optimization Based on Scattering Parameters with Pareto Optimal Analysis, Turk J. Elec Eng. & Comp. Sci., 20 (3):353-365, 2012. 21. A. Kavousi Fard, T. Niknam, Multi-Objective Probabilistic Distribution Feeder Reconfiguration Considering Wind Power Plants, International Journal of Electrical Power & Energy Systems, 55:680-691, 2014. 22. C. Ko, Y. Chang, C. Wu, A PSO Method with Nonlinear Time Varying Evolution for Optimal Design of Harmonic Filters, IEEE Transactions on Power Systems, 24 (1):437-444, 2009. 23. Y. Leung, Y. Wang, An Orthogonal Genetic Algorithm with Quantization for Global Numerical Optimization, IEEE Transactions on Evolutionary Computation, 5 (1):41-53, 2001.
344
H. Shayeghi and Y. Hashemi
24. A. Mazza, G. Chicco, A. Russo, Optimal Multi-Objective Distribution System Reconfiguration with Multi Criteria Decision Making-Based Solution Ranking and Enhanced Genetic Operators, International Journal of Electrical Power & Energy Systems, 54:255-267, 2014. 25. S. Christa, P. Venkatesh, Multi-Objective Optimization Problem for the Thyristor Controlled Series Compensators Placement with Multiple Decision-Making Approaches, European Transactions on Electrical Power, 23 (2):249–269, doi:10.1002/etep.658, 2011.
Chapter 10
Reactive Power Optimization in AC Power Systems Ali Jafari Aghbolaghi, Naser Mahdavi Tabatabaei, Narges Sadat Boushehri and Farid Hojjati Parast
10.1
Introduction
Reactive Power is one of the most important features in power networks so that its appropriate production and distribution among consumers can affect performance, efficiency, and reliability of the power networks positively. Creating competitive mechanism via establishing a market to present different services and changing the current rules necessitate that in the new condition programming and controlling of reactive power as well as the voltage to be considered more accurately. The purpose of reactive power optimization in AC power systems is to recognize the best value for control variables in order to optimize the target function considering the possible constraints. With current developments of power grids, it has been growing in popularity among researchers to probe into how to use existent reactive power compensators in order to reduce active power losses and improve voltage profile. The reactive power optimization is a complicated problem with a broad solution space, nonlinear and non-convex, in which there both continuous are and discrete variables. In general, reactive power optimization problem entails two separate branches as optimal placement of reactive power compensators and optimal operA. Jafari Aghbolaghi (&) Zanjan Electric Energy Distribution Company, Zanjan, Iran e-mail: [emailprotected] N. Mahdavi Tabatabaei Electrical Engineering Department, Seraj Higher Education Institute, Tabriz, Iran e-mail: [emailprotected] N.S. Boushehri Department of Management, Taba Elm International Institute, Tabriz, Iran e-mail: [emailprotected] F. Hojjati Parast Andishmand Shomal Gharb Engineering Consultancy, Zanjan, Iran e-mail: [emailprotected] © Springer International Publishing AG 2017 N. Mahdavi Tabatabaei et al. (eds.), Reactive Power Control in AC Power Systems, Power Systems, DOI 10.1007/978-3-319-51118-4_10
345
346
A. Jafari Aghbolaghi et al.
ation of existent reactive power compensators. Optimal placement of reactive power compensators problem tries to determine three parameters as the type of the compensator, the rate of the output power and installation location. However, optimal operation of existing reactive power compensators is about determining optimal reactive power output for the compensators that have already been installed. Although producing reactive power involves no fee in the operation step, it influences the final cost via affecting the active power losses in power transmission system significantly. Optimal distribution of reactive power is a sub-problem of Optimal Power Flow (OPF) and the parameters that should be controlled are actually the control variables such as output reactive power or the terminal voltage of generators, output reactive power of all reactive power compensators, and tap-changing transformers’ tap settings. Since flowing reactive power through power transmission system results in active power losses, the main goal of reactive power optimization is to reduce the active power losses via optimal controlling of the above mentioned parameters considering the security aspects, which must be fulfilled in order to have a reliable and stable power network with which consumers can be fed continuously. In addition, some other goals are pursuit beside the main goal, like improving voltage profile which involves the security measures of power systems. Therefore, the target function of reactive power optimization problem is active power losses equation, in which some constraints should be included as voltage rate boundaries, equipment’s power output limits, and transmission system margins for carrying power etc. In the reactive power optimization problem, there are both continuous and discrete types of variables, of which should be determined by an optimization algorithm that leads the power system to contain least possible active power losses. In one hand, discrete variables are transformers’ tap settings and capacitors’ output powers. In the other hand, continuous variables are reactive power outputs of generators as well as synchronous compensators’. In the recent years, wide variety of optimization algorithms have been recommended to solve the reactive power optimization problem, containing traditional algorithms such as quadratic programing (QP) and sequential quadratic programming (SQP) etc., and heuristic ones which are inspired by nature such as particle swarm optimization (PSO), genetic algorithm (GA), evolutionary programming (EP) and their derivatives etc. Traditional algorithms have too many defects encountering problems that have nonlinear essences, the ones which consist of discrete variables and have got many local optimum solutions. Eventually, traditional gradient based algorithms lose their capability to be efficiently solve such challenging optimization problems. Overall, the traditional optimization algorithms become nonfunctional confronting practical reactive power optimization in large power systems having huge dimensions. On the other hand, optimization algorithms inspired by nature have proved their high performance optimizing problems with so many variables covering discrete and continuous ones while having many local optimum points. This chapter is focusing on the reactive power optimization using artificial optimization algorithms and trying to give a comprehensive perspective of all
10
Reactive Power Optimization in AC Power Systems
347
formulations and constraints that are needed in order to implement reactive power optimization problem and use its practical implications effectively. First, brief fundamentals of reactive power optimization containing some relevancies which are of crucial importance is presented. Second, to enlighten how reactive power optimization works the classic method of reactive power optimization is presented, thanks to [1]. Third, basic principles and problem formulation of reactive power optimization using artificial intelligent algorithms is elaborated. Fourth, particle swarm optimization algorithm and pattern search method and how to use them in reactive power optimization problem have been expounded defining PSO’s parameters way back into reactive power optimization involving ones. Finally, the offered algorithms simulation result on two case studies have been presented.
10.2
Fundamentals of Reactive Power Optimization
There are two types of reactive power flow in power transmission systems; one for fulfilling transmission system needs and the other for feeding loads that naturally consume reactive power in order to work properly, that cause according to Eq. (10.1) active power losses and following that, the fuel consumption of the power plant raises [2]. Taking the Eq. (10.1) into consideration, it is obvious that the imaginary part of the current equation simply affects the absolute value of that which influences total active power losses in the next place. In addition, there is no any control possible on IR as it is corresponding current to the active power which should be supplied to the loads via transmission systems. The idealistic condition recommends compensating all the reactive power needs of the loads at the same place, just beside the loads, in order not to impose reactive current into transmission systems. The closer reactive power compensator to loads, the less reactive power flowing in power transmission systems. It is not feasible however, because of technical and economical restrictions upon that, such as the relevancy of voltage stability to the reactive power in AC power systems, of which was broadly investigated in [3, 4]. Reactive power is absorbed by different equipment as synchronous condensers being operated with lagging power factor, shunt reactors, transmission lines and transformers’ inductances, static reactive power compensators and loads etc., while being injected to transmission systems by generators, synchronous condensers operating with leading power factor, static capacitors, static compensators, and transmission lines’ capacitances etc. [2]. There is another type of equipment that avails power system operators to have a control on reactive power flow through transmission system and it is transformers with tap changing facility under the load condition. There is a straight connection between two neighboring buses’ voltages and reactive power flow in the corresponding transmission system which connects them together. In simple words, reactive power flows from the bus having higher voltage magnitude to the bus which has lower voltage magnitude in order to naturally raise the second bus’s voltage close to the same level. Take a simple network shown in Fig. 10.1 into consideration, reactive power will flow from the slack bus
348
A. Jafari Aghbolaghi et al. Generator A
Generator B
G
G
Slack Bus
Load A
PV Bus Line A
C
Li B
Li
ne
ne
PQ Bus
Load B
Fig. 10.1 Three-bus power system
to the PV bus if slack bus’s voltage magnitude is higher than PV bus’s. On the whole, there would not be any reactive power flow in transmission systems if the voltage magnitudes of all buses are equal.
Ploss ¼ RjI j2 I ¼ IR þ jIX
Iij ¼
Vi
ð10:1Þ
Vj
ð10:2Þ
Zij
Pij þ jQij ¼ Vi Iij
ð10:3Þ
(10.2) and (10.3) ) Pij þ jQij ¼ Vi
Vi
Vj Zij
8 Vi ¼ jVi jðcos di j sin di Þ > > < Vj ¼ Vj cos dj j sin dj : > > : Zij ¼ Zij cos hij j sin hij
ð10:4Þ
ð10:5Þ
where, |Vi| and |Vj| are voltage magnitudes of ith and jth busbars respectively. |Zij| is absolute value of the impedance of transmission line between ith and jth busbars. di and di are the voltage angles of ith and jth busbars respectively, while hij is angle of impedance between ith and jth busbars.
10
Reactive Power Optimization in AC Power Systems
349
Expanding Eq. (10.4) using the assumptions of Eq. (10.5) will be as following Pij þ jQij ¼ jVi jðcos di þ j sin di Þ
"
¼ jVi jðcos di þ j sin di Þ 0
"
jVi jðcos di
jVi j cos di
j sin di Þ Vj cos dj Zij cos hij j sin hij
j sin dj
# ð10:6Þ
# jjVi j sin di Vj cos dj þ jVj sin dj Zij cos hij j sin hij
1 jVi jVj cos di cos dj þ jjVi jVj sin dj cos di @ A þ jjVi j2 sin di cos di þ jVi j2 sin2 di jjVi jVj sin di cos dj jVi jVj sin dj sin di ¼ Zij cos hij j sin hij 0
¼
B @
jVi j2 cos2 di
þj
jjVi j2 sin di cos di
jVi j2 cos2 di
jVi jVj cos di cos dj þ jVi j2 sin2 di
jVi j2 sin di cos di þ jVi jVj sin dj cos di þ jVi j2 sin di cos di Zij cos hij j sin hij
1 jVi jVj sin dj sin di C A V sin d cos d V j ij j i j
jVi j2 jVi jVj cos di cos dj þ sin dj sin di þ j jVi jVj sin dj cos di ¼ Zij cos hij j sin hij 0
¼
2 Zij jVi j cos hij @ þ jZij jVi j2 sin hij
Pij þ jQij 0
sin di cos dj
Zij jVi jVj cosðdj di Þ cos hij þ jZij jVi jVj sinðdj jZij jVi jVj cosðdj di Þ sin hij Zij jVi jVj sinðdj 2 Zij
di Þ cos hij di Þ sin hij
1 A
2 1 Zij jVi j cos hij Zij jVi jVj cosðdj di Þ cos hij Zij jVi jVj sinðdj di Þ sin hij @ A þ j Zij jVi jVj sinðdj di Þ cos hij þ Zij jVi j2 sin hij Zij jVi jVj cosðdj di Þ sin hij ¼ 2 Zij
ð10:7Þ
jVi j2 sin hij þ jVi jVj sinðdj Qij ¼ Zij
di
hij Þ
ð10:8Þ
The term (dj − di) has a minute value and can be neglected in order to simplify Eq. (10.8), then
350
A. Jafari Aghbolaghi et al.
Fig. 10.2 The simplified electrical circuit of synchronous generator
Vt
Xs
Xe
V=|V|
|Ef |
Qij ¼
jVi j2 sin hij
jVi jVj sin hij Zij
! sin hij Qij ¼ jVi j jVi j Vj Zij |fflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflffl} |fflfflfflfflfflffl{zfflfflfflfflfflffl} Voltage Difference
ð10:9Þ
ð10:10Þ
Constant
Therefore, Eq. (10.10) shows that the value of the corresponding Qij flowing in transmission systems is directly dependent on the difference of voltage magnitudes between ith and jth busbars. According to the electrical rules which synchronous machines work based on, the reactive power injected or absorbed by a synchronous generator can be controlled by its excitation system that affects the generator’s terminal voltage straightforwardly [5]. Take Fig. 10.2 into consideration as a simplified electrical circuit of synchronous generators and power systems, then reactive power absorbing or producing by the generator will be as the following: 8 < V ¼ jV j\0 Ef ¼ Ef \d : X ¼ Xe þ Xs
I¼
ð10:11Þ
Ef \d
jV j\0 X\90
ð10:12Þ
The apparent power transmitted from the generator to the infinite bus is: S ¼ P þ jQ ¼ Ef I
ð10:13Þ
Expanding Eq. (10.13) using Eqs. (10.11) and (10.12) will be as following Ef \ d jV j Ef 2 \90 ¼ S ¼ Ef \d X\ 90
Ef jV j\90 þ d X
ð10:14Þ
10
Reactive Power Optimization in AC Power Systems
Fig. 10.3 The single-phase Thevenin equivalent circuit
351 E Is
Zs = Rs + jXs
V Il
Pl + jQl
Yl = Gl + jB l
Then by segregating Eq. (10.14), the reactive power absorbed or injected to the infinite busbar by the synchronous generator will be as following E f Ef jV j cos d Q¼ ð10:15Þ X
where Vt is the voltage of infinite busbar and d is the angle between infinite busbar and the generator’s terminal under no-load condition. If the voltage magnitude of the infinite busbar to be considered as constant, it is obvious that consuming or feeding reactive power by the generator is entirely correlated with the generator’s voltage magnitude which is controlled by the excitation system, while the active power being transmitted assumed to be constant over the handling period [5]. There is another issue as voltage regulation which is associated with the reactive power control in AC power systems. The voltage regulation is defined as a proportional (per-unit) change in the voltage magnitude of supply terminal in relation to defined change to the load current (e.g., from no-load to full-load). Taking Fig. 10.3 as Thevenin equivalent of supply system into consideration, the voltage drop in the transmission system in the absence of compensator is shown in Fig. 10.4 as DV, which is as the following DV ¼ E V ¼ Zs Il 8 < Zs ¼ Rs þ jXs : Il ¼ Pl jQl V
ð10:16Þ
ð10:17Þ
352
A. Jafari Aghbolaghi et al.
Fig. 10.4 The corresponding phasor diagram to Fig. 10.3, without compensation
E V
V RsIs
VX VR
Il
So that Pl DV ¼ ðRs þ jXs Þ ¼
jQl V
Rs Pl þ Xs Ql Xs Pl Rs Ql þj V V ¼ DVR þ jDVX
ð10:18Þ
ð10:19Þ ð10:20Þ
The voltage drop has two components as DVR in the same phase with V and DVX in quadrature with V which is illustrated in Fig. 10.4 elaborately. It is obvious that the magnitude and the phase of V are functions of the magnitude and the phase of load current so that the amount of voltage regulation depends straightly on the amount of real power as well as reactive power consuming of the load [2]. According to Eq. (10.10) reactive power flowing in transmission systems is a function of the difference of voltage magnitude between two neighboring busbars. Hence, the bigger the difference, the higher amount of reactive power flows toward busbar having lower voltage magnitude in order to raise it, and it causes additional active power losses in transmission systems. Therefore, it would be such a great asset if reactive power could be compensated at the same place as loads lie, which would help the power factor to reach near the unique value. However, a voltage drop that is relevant to active current flow in transmission systems would yet be consistent. Considering Eq. (10.18) while having loads’ reactive power consumption compensated on-site, the equation will be as the following Pl DV ¼ ðRs þ jXs Þ V Rs Pl Xs Pl ¼ þj V V ¼ DVR þ jDVX
ð10:21Þ
ð10:22Þ ð10:23Þ
In order to have a voltage regulation by the value of zero, the reactive power consumption by the load and the voltage drop being caused by active current flowing in the transmission system should be compensated. Referring to Fig. 10.5,
10
Reactive Power Optimization in AC Power Systems
Fig. 10.5 The corresponding phasor diagram to Fig. 10.3, with compensation
353
I
E
Is V s
R s Is
V
l
Il
it is possible to have |E| = V via producing reactive power and injecting to the load busbar in presence of the load which will be greater than the load’s absorption [2]. For sum up, all the components of power systems work together continuously as a united structure so that any single change in any part will influence all the other parameters directly, some of which are explained in the current section regarding that they are associated with reactive power’s role in AC power systems. Therefore, all the relations should be evaluated carefully whenever a change happens in a specific parameter, which necessitates having a comprehensive control system to fulfill all the needs of power systems in order to have a reliable, stable, and a cost-effective energy produced and dispatched.
10.3
Reactive Power Optimization Using Classic Methods
The main aim of reactive power optimization is to minimize reactive power flowing in transmission systems including transmission lines, transformers etc., which leads to less real power losses. This will be approachable via specifying the best reactive power output value of the reactive power compensators and the other controllable parameters under a number of constraints which should be met in order to have a reliable, stable and cost-effective power system [1]. At first glance, it might come into mind that if loads’ reactive power consumption need to be met just by the nearest reactive power source (in terms of electrical distance), the active power losses would be minimized, although it is not such a true interpretation. On one hand, while it is being talked about reactive power optimization, it should be considered that reactive power in AC power systems has plenty of associations with the other parameters of the system, most of
354
A. Jafari Aghbolaghi et al.
which are crucial in terms of power system security and reliability. On the other hand, reactive power optimization is not a straight optimization, however, the power flow plays a key role in the reactive power optimization problem. For instance, the voltage magnitude is one of the most important parameters which influences and also is influenced by reactive power of the system. According to Eq. (10.1), reactive power loss in AC power network can be considered as a function of load active and reactive power consumption as PL ¼ PL ðP1 ; P2 ; . . .; Pn ; Q1 ; Q2 ; . . .; Qn Þ
ð10:24Þ
Reactive power optimization is a constrained optimization problem that may be attacked formally using advanced calculus methods involving Lagrange function. In classic reactive power economic dispatch, the active power of all generators are already known and fixed during optimization procedure, except slack bus’s, so that any change in active power losses associated to reactive power optimization could be distinguished easily. One of the well-known techniques to implement optimization in power network, of which consists of some constraints to be fulfilled, is the Lagrange method. In reactive power optimization, one of the main constraints is reactive power balance as Eq. (10.25), which is shown in the following [1] M X
QGi ¼ QD þ QL
ð10:25Þ
i¼1
where QG includes all reactive power sources, QD is reactive power demands of the loads, and QL is reactive power losses in the transmission system. In order to establish the necessary conditions for an extreme value of the objective function, the constraint functions should be added to the objective function after the constraint functions has been multiplied by an undetermined multiplier. This is known as Lagrange function and is shown in Eq. (10.26). L ¼ FT þ k/
ð10:26Þ
where, FT is target function which is aimed to be optimized, k is an unspecified multiplier, and / contains all the probable constraint functions that must be met during the optimization procedure. Therefore, the Lagrange function formed to decrease active power losses, subjected to equality constraint of producing and consuming reactive power, constructed from Eqs. (10.24) and (10.25) will be as L ¼ PL
k
M X i¼1
QGi
QD
QL
!
ð10:27Þ
10
Reactive Power Optimization in AC Power Systems
355
Then, the necessary condition to have an extreme value for the Lagrange function is to set the first derivative of the function with respect to each independent variables (QG and k) equal to zero [1]. @L @PL ¼ @QGi @QGi
k 1
@L ¼ @k
@QL @QGi M X
QGi
¼0
i ¼ 1; 2; . . .; M
QD
i¼1
QL
!
¼0
ð10:28Þ
ð10:29Þ
From Eq. (10.28) @PL 1 ¼k @QGi 1 @QL
i ¼ 1; 2; . . .; N
ð10:30Þ
@QGi
Equation (10.30) is the formula of reactive power economic dispatch, where, is incremental rate of active power losses with respect to ith reactive power
@PL @QGi
@QL is incremental reactive power losses with respect to ith reactive power source, @Q Gi source. @QL @PL The terms @Q can be calculated with impedance matrix method which and @Q Gi Gi is depicted below. The real power losses in power transmission systems can be represented as [1]
PL þ jQL ¼ V T I ¼ ðZIÞT ^I ¼ I T Z T I
ð10:31Þ
I ¼ IP þ jIQ
ð10:32Þ
Z ¼ R þ jX
ð10:33Þ
where, I is the current vector in transmission lines, I* is the conjugate current vector in transmission lines, Z is the impedance matrix of transmission lines, and VT is the voltage vector of all busbars. Substituting Eqs. (10.32) and (10.33) into Eq. (10.31), PL and QL will be obtained as [1] PL ¼
n n X X
Rjk IPj IPk þ IQj IQk
ð10:34Þ
n n X X
Xjk IPj IPk þ IQj IQk
ð10:35Þ
j¼1 k¼1
QL ¼
j¼1 k¼1
356
A. Jafari Aghbolaghi et al.
The relation between injected power and current to the system is Pi þ jQi ¼ ðVi cos hi þ jVi sin hi ÞðIPi
jIQi Þ
ð10:36Þ
Then IPi and IQi will be as the following IPi ¼
Pi cos hi þ jQi sin hi Vi
ð10:37Þ
IQi ¼
Pi sin hi þ jQi cos hi Vi
ð10:38Þ
Substituting Eqs. (10.37) and (10.38) into Eqs. (10.34) and (10.35) PL and QL will be as [1] PL ¼
n n X X j¼1 k¼1
QL ¼
ajk Pj Pk þ Qj Qk þ bjk Qj Pk
n X n X djk Pj Pk þ Qj Qk þ cjk Qj Pk j¼1 k¼1
Pj Qk
ð10:39Þ
Pj Qk
ð10:40Þ
where ajk ¼ bjk ¼ djk ¼ ajk ¼
Rjk cos hj Vj Vk Rjk sin hj Vj Vk
Xjk cos hj Vj Vk Xjk sin hj Vj Vk
hk
ð10:41Þ
hk
ð10:42Þ
hk
ð10:43Þ
ð10:44Þ
hk
Then, taking the first derivative of Eq. (10.39) into consideration with respect to independent variables, the consequence will be as following [1] n X n @PL X @ ajk Pj Pk þ Qj Qk þ bjk Qj Pk Pj Qk ¼ @Pi @Pi j¼1 k¼1 n X n n X X @ajk bjk ðPk aik Qk bik Þ þ Pj Pk þ Qj Qk þ bjk Qj Pk Pj Qk ¼2 @Pi @Pi i¼1 k¼1 k¼1 |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl ffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} 0
ð10:45Þ
10
Reactive Power Optimization in AC Power Systems
357
The second term of the Eq. (10.45) is negligible, and then it will be simplified as [1] n X @PL 2 ðPk aik @Pi k¼1
Qk bik Þ
ð10:46Þ
In a high-voltage power network, (hj − hk) is infinitesimal, then sin (hj − hk) 0. Therefore, bjk can be ignored as well. n X @PL 2 Pk aik @Pi k¼1
ð10:47Þ
Similarly n X @PL 2 Qk aik @Qi k¼1 n X @QL 2 Pk dik @Pi k¼1 n X @QL 2 Qk dik @Qi k¼1
ð10:48Þ
ð10:49Þ
ð10:50Þ
Considering real and reactive power consumption of loads constant during the optimization procedure, two assumptions as the following can be made [1] dPi ¼ d ðPGi
PDi Þ ¼ dPGi
ð10:51Þ
dQi ¼ d ðQGi
QDi Þ ¼ dQGi
ð10:52Þ
Then, Eq. (10.47) to Eq. (10.50) can be written as n X @PL 2 Pk aik @PGi k¼1 n X @PL 2 Qk aik @QGi k¼1 n X @QL 2 Pk dik @PGi k¼1
ð10:53Þ
ð10:54Þ
ð10:55Þ
358
A. Jafari Aghbolaghi et al. n X @QL 2 Qk dik @QGi k¼1
ð10:56Þ
Therefore, if the power system which is being investigated has enough amount of reactive power sources, the steps of classic reactive power optimization using Lagrange function would be as following [1]: I. Power flow calculations should be carried out in order to have all generators’ active power output, then fix them all in the current value as active power consumption of loads has been considered constant, the only exception is the slack generator. The output power of slack generators would not remain constant during reactive power optimization. II. The value of k should be computed for each reactive power source using Eqs. (10.54) and (10.56). For the sources having k < 0, it means the active power losses of the system can be reduced by increasing the output amount of reactive power of the source. For the sources having k > 0, it will be vice versa. Therefore, in order to decrease active power losses of the system, the amount of reactive power output should be increased for the sources having k < 0 and also decreased for the sources having k > 0. Each time, the source with minimum value of k will be chosen to increase its output if k < 0, and the source with maximum value of k to decrease its output if k > 0. Eventually, power flow calculations should be computed to have the result of optimization. III. Using power flow results, the active power losses can be computed. Since the active power output of the reference unit was not fixed, whatever happening to the active power losses of the system can be sensed in the active power output of the slack generator. The reactive power process will be continued until active power losses cannot be reduced anymore. It should be noted that limitations of reactive power sources were not considered in the procedure above. There is a limitation for each reactive power source as QGi min QGi QGi max
ð10:57Þ
If they are supposed to be considered, the constraint in Eq. (10.57) should be checked in every iteration for each source. When it comes to choose an output amount for the power of the sources according to their k value, if k suggests to increase the output power of the ith source while it is exceeding either its above or its below margins, the amount of output reactive power of the corresponding source should be set to its margins accordingly. Thus, the source which its output power has been adjusted to either its maximum or its minimum values will not be considered any longer in the reactive power optimization procedure [1]. The above-mentioned structure of minimizing active power losses using reactive power economic dispatch was a simplified method depicted in order to represent the
10
Reactive Power Optimization in AC Power Systems
359
concept, and it is not much of a practical method to be used in a real system having large dimensions and plenty of constraints, as voltage profile, transmission lines bounds etc. There are a lot of conventional optimization methods which can be found in [6]. In the next section, reactive power optimization using artificial intelligence algorithms and the used model will be described elaborately, which is more applicable when it comes to a large system in which all the security constraints should be considered. The reason why heuristic methods are highly concerned in reactive power optimization is a few complexities in such problems’ nature, some of which known as non-convexity, having continuous and discrete variables, and having plenty of local and global optimums. Conventional methods and algorithms which were used to optimize such problems proved themselves rather unable to be applied on the problem satisfactorily, because conventional methods mostly use the gradient of the objective function, leading to a local minimum rather than a global one. Another problem is that conventional methods need the derivatives of objective functions which are not accessible in the ones consisting of discrete variables. Thus, there is a compelling need to some sort of algorithms, of which are able to redeem all defections of the conventional ones when applying to optimization problems in order to have as much performance as possible. Although pure mathematical methods like Lagrange function are very precise, the performance of intelligent algorithms has been proved in plenty of cases, and they happen to be more efficient than conventional methods in many of aspects.
10.4
Reactive Power Optimization Using Artificial Intelligent Algorithms
As it has been mentioned in the last part of previous section, the accuracy of conventional methods is their most redeeming feature, although they cause problems in terms of mathematical fulfilments. The conventional methods are not capable of finding the global optimum of the target function and they will get into trouble if the objective function consists of discrete variables. In addition, classic pure mathematical methods need the derivatives of target function, which will impose so many difficulties to the optimization procedure, such as the complexities that calculating derivatives of some functions possess. Heuristic algorithms use the target function itself during the optimization procedure instead of its derivatives and they are readily able to find the global optimum point. Fortunately, heuristic algorithms have solved many drawbacks that conventional algorithms had, and also they have proven their capabilities in the optimization respect. Therefore, it is worth to put some effort on using heuristic methods to optimize engineering problems like the reactive power optimization.
360
A. Jafari Aghbolaghi et al.
10.4.1 Basic Principles There are two types of variables concerning reactive power optimization as control and state variables. The terminal voltage of generators, tap setting of transformers and reactive power output of reactive power sources are control variables that are changeable within their bounds. Control variables are the tools that heuristic algorithms use in order to optimize the target function. The voltage magnitudes and voltage angels of PQ buses are state variables, of which the active power losses can be computed using Eq. (10.58) and the corresponding values. The general procedure of optimization by heuristic algorithms is shown in Fig. 10.6, whereas the flowchart of reactive power optimization using intelligent algorithms is shown in Fig. 10.7. Floss ¼
NL X k¼1
h 2 2 gk V1;k þ V2;k
2V1;k V2;k cos h1;k
h2;k
i
ð10:58Þ
where, gk is conductance value of the transmission line between starting and ending buses, V1,k and V2,k are voltage magnitudes of starting and ending buses, and, h1,k and h2,k are voltage angles of starting and ending buses, respectively.
Fig. 10.6 General procedure of optimization by heuristic algorithms
Start
Initiate values for control variables
Calculate the fitness function
Apply operators of heuristic algorithm to control variables
Has the function been optimized? Yes
Stop
No
10
Reactive Power Optimization in AC Power Systems
Fig. 10.7 General reactive power optimization trend using intelligent algorithms
361
Start
Initialize control variables using intelligent algorithms or use standard values
Apply the values of control variables to power network
Run power flow and obtain values of state variables
Apply the operators of heuristic algorithm to control variables in order to reach better position
Calculate fitness of objective function using state variables
No Optimization successful?
Yes Stop
As it can be inferred from Figs. 10.6 and 10.7, there is an important point in reactive power optimization problem to pay attention to, which is the calculation of objective function in order to examine whether it has improved or not. There is one extra step in reactive power optimization as the power flow calculation, shown with dashed rectangle in Fig. 10.7. In order to evaluate the objective function represented by Eq. (10.58) in which both state and control variables are involved, after initializing and/or determining control variables by heuristic algorithms, power flow should be run to make calculation of fitness function available. In other words, we have to run power flow calculations in order to achieve the values for the state variables which are mentioned before.
362
A. Jafari Aghbolaghi et al.
10.4.2 Problem Formulation of Reactive Power Optimization There are several methods of formulating and modeling reactive power optimization problem according to what exactly is expected as its practical implications, for instance, what constraints should be considered and what objectives are aspired will determine the approach. In this section, one general model will be presented which is useful when it turns to be used in heuristic algorithms to optimize reactive power in AC power systems. The objective function of the optimal reactive power flow (ORPF) includes technical and financial goals. The economic goal has mainly considered minimizing active power losses in transmission systems. The technical goals are to diminish voltage deviation of PQ buses (VD) from the ideal voltage setting and to increase voltage stability index (VSI). Therefore, objective functions for both the technical and monetary goals are considered in this chapter as following [7] f ðXÞ ¼ minðPL Þ ( ðVDÞ subject to: ðVSIÞ
10.4.2.1
Active Power Losses Objective Function
Reducing active power losses is the most crucial aim of reactive power optimization problem, which influences final cost of dispatched energy. The active power losses can be computed by Eq. (10.59) as follows Floss ¼
N X k¼1
h 2 2 gk V1;k þ V2;k
2V1;k V2;k cos h1;k
h2;k
i
ð10:59Þ
where, gk is conductance of the transmission line between starting and ending buses, V1,k and V2,k are voltage magnitudes of starting and ending buses, and h1,k and h2,k are voltage angles of starting and ending buses, respectively, and N is the number of transmission lines.
10.4.2.2
Voltage Deviation Constraint
Another significant goal of reactive power optimization in AC power systems is to shrink voltage deviation. Electrical equipment is designed for optimal operation at its nominal voltage. Any deviance from the nominal voltage can result in reducing the general effectiveness and decreasing longevity of electrical apparatus. Voltage
10
Reactive Power Optimization in AC Power Systems
363
deviance constraint is to enhance the voltage profile of power systems by minimizing the summation of voltage deviations at load buses. The voltage deviance constraint can be considered as the least possible amount of voltage deviation summation at each load bus. This function is defined as follows: VD ¼
M X Vj j¼1
Vjref
ð10:60Þ
where, Vj is the actual voltage of jth load bus, Vref j is the ideal voltage of jth load bus, and M is number of load buses [8].
10.4.2.3
System Voltage Stability Index
A rather simple voltage stability index to define is V/V0 ratio, where V is the voltage magnitude of all the PQ buses under load condition and V0 is the voltage magnitude of all the PQ buses under no-load condition, both identified from load flow calculations or state estimation studies of the system. The ratio V/V0 at each node offers a voltage stability diagram for the corresponding bus, providing power system operators with weak spots to be taken care of. There are wide range of indices for this purpose, while the current being chosen for the sake of simplicity. VSI ¼
T X 1 i¼1
Vi Vi0
ð10:61Þ
where, Vi is the voltage magnitude of ith PQ bus under load condition, Vi0 is the voltage magnitude of ith PQ bus under no-load condition, and T is number of load buses, respectively.
10.4.2.4
Constraints of Control and State Variables
The control variable constraints embrace tap changer settings of all transformers T, the output capacity of reactive power compensators C, and terminal voltage of all the generators V. The state variables consist of voltage magnitude of all PQ buses U, and reactive power output of all generators Q. Thus, the restriction expressions of control and state variables can be written as VGk;min \VGk \VGk;max
ð10:62Þ
Ti;min \Ti \Ti;max
ð10:63Þ
364
A. Jafari Aghbolaghi et al.
Cj;min \Cj \Cj;max
ð10:64Þ
QGk;min \QGk \QGk;max
ð10:65Þ
Vl;min \Ul \Vl;max
ð10:66Þ
where, VGk,min (VGk,max), Ti,min (Ti,max), Cj,min (Cj,max), QGk,min (QGk,max) and Vl,min (Vl,max) are lower (upper) boundary values of PV bus voltages, tap ratio of transformers, reactive power output of compensators, reactive power output of PV buses and voltage magnitude of load buses, respectively.
10.4.2.5
System Power Flow Constraint Equations
The reactive power optimization must fulfill the power flow balances, which are written as PGi
PLi
Vi
n X
Vj Gij cos dij þ Bij sin dij ¼ 0
ð10:67Þ
n X
Vj Gij sin dij
ð10:68Þ
j¼1
QGi
QLi
Vi
j¼1
Bij cos dij ¼ 0
where, n is number of buses, PGi and QGi are generator active and reactive power of the ith bus. PLi and QLi are load active and reactive power of ith bus. Vi and Vj are voltage magnitudes of ith and jth buses (two neighboring busbars). Gij, Bij, and dij are conductance parameters and voltage angle between ith and jth buses, respectively. It should be noted that the optimization algorithm that determines the control variables in relation to the optimization procedure, satisfies the corresponding constraints as well. The constraints related to state variables will be met by the standard power flow procedure and the algorithm being used, such as Newton or Gauss Sidle. In addition, violations happened in the state variables due to adjusting control variables, can be controlled using a penalty function added to the final objective function of reactive power optimization.
10.4.2.6
General Form of Objective Functions Used in Intelligent Algorithms
Most of the intelligent optimization algorithms tend to have an unconstrained objective function rather than a constrained one. However, different sorts of constraints such as linear and non-linear are typically inseparable part of optimization procedure. Therefore, there is a compelling need to be able to convert a constrained
10
Reactive Power Optimization in AC Power Systems
365
fitness function to an unconstrained one in order to use it in intelligent algorithms. The solution is the penalty function which exactly transforms both objective function and its constraints to a unique unconstrained function. Considering the optimization problem generally as Eq. (10.69) [6]
subject to:
(minimize f ðXÞ gj ðXÞ 0 j ¼ 1; . . .; m
ð10:69Þ
hi ðXÞ ¼ 0 i ¼ 1; . . .; n
It has been converted into an unconstrained optimization fitness function by constructing a function of the form: /ð X Þ ¼ f ð X Þ þ
m X
ri hgi ðXÞi2 þ
p X j¼1
i¼1
2 Rj hj ðXÞ
ð10:70Þ
where X is the vector of control variables, ri and Rj are penalty multipliers which are constant for all the constraints during the optimization procedure, gi(X) and Hj(X) are inequality and equality constraints, respectively. hgi ðXÞi is the bracket function which is defined as Eq. (10.71). hgi ðXÞi ¼
gi ðXÞ 0
gi ðXÞ [ 0 gi ðXÞ 0
ð10:71Þ
Equation (10.70) can be considered for maximization or minimization problems appropriate to which optimization algorithm is being used. If a specific algorithm is naturally good at minimizing functions and it is used to minimize a function, then it will be better to use F(X) = /(X) as the fitness function, Otherwise it will be better to use the form shown in Eq. (10.72) to define the maximization problem back to minimization one. For instance, Genetic Algorithm is naturally good at maximizing objective functions, thus the fitness function shown in (10.72) will be used if the considered problem aims to minimize the objective function. FðXÞ ¼
1 /ðXÞ
ð10:72Þ
As it can be seen in the Eq. (10.70), /(X) is a new fitness function consisting of constraints and the objective function itself. It is forced by an additional value, multiplying ri and Rj, which are allocated big values in case the optimization procedure violates the minimum and maximum bounds of constraints. Therefore, the new objective function will have a huge value when the limits are violated, having been influenced by the multipliers. It is the exact method optimization algorithms use to perceive whether the bounds are violated or not. The same method is highly appreciated when it comes to using intelligent algorithms in reactive power optimization problem.
366
A. Jafari Aghbolaghi et al.
Thus, the penalty function to use in intelligent algorithms for reactive power optimization will be as follows f ðV; hÞ ¼
N X k¼1
h 2 2 gk V1;k þ V2;k
2V1;k V2;k cos h1;k
h2;k
i
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
þ kv
Ploss
M X j¼1
Vj
Vi Vi0 i¼1 |fflfflfflfflfflfflfflfflfflffl ffl{zfflfflfflfflfflfflfflfflfflfflffl}
T X ref 1 Vj þ ks
|fflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflffl} Voltage Deviation
ð10:73Þ
Voltage Stability
where, f is an unconstrained objective function, kv and ks are penalty multipliers which are used to obligate optimization procedure not to violate the corresponding bounds. It should be noted that the constraints related to control variables are usually managed by optimization algorithms, while the constraints pertinent to power flow calculations represented by Eqs. (10.67) and (10.68) will be met by power flow. Moreover, the constraints containing state variables should be included in the penalty function to be fulfilled. There is another way to consider constraints of state variables and that is multi-objective optimization methods, in which each constraint is defined as an objective function separately and the procedure drives the main objective function as well as all the constraints simultaneously to be optimized and fulfilled. There is a major difference between using multi-objective optimization methods and single-objective ones, and that is, all the objectives including main fitness function are driven to be minimized while the multi-objective approaches are used. Using penalty function method (single-objective), however, tries to keep the constraints within their bounds.
10.5
Particle Swarm Optimization Algorithm
Particle swarm optimization (PSO) is one of the most successful optimization approaches, of which was inspired by nature. It has broadly been used in all sorts of optimization problems that possess large search spaces, without needing gradients of objective functions. Particle swarm optimization algorithm is based on the performance of a colony or swarm, a flock of birds, a school of fish and/or any kind of creatures that live in groups. The particle swarm optimization algorithm tries to simulate the behavior of these collective organisms which have evolved along centuries in order to enhance their performance of finding food, encountering danger and keeping themselves more competitive in a world that adaptability is a prerequisite feature to survive. The word ‘particle’ notes to an individual in a swarm which acts in a distributed way using its own intellect and also the cooperative intelligence of the crowd.
10
Reactive Power Optimization in AC Power Systems
367
Fig. 10.8 A fish swarm using their collective intelligence
Taking Fig. 10.8 into consideration, since there is sight limitation under the water, each individual can just see a near distance radially so that can put all of them into trouble all the time facing sharks or any offensive actions. Nevertheless, they have been capable of keeping their species alive in harsh nature of the underwater. As it can be seen in the photo, individuals try to move close to each other keeping precautionary distance so that whenever there is a danger the nearest individual will sense it and act, and since all of them follow the same rules, the farthest one in the group will sense the danger and act appropriately. That was just one of the advantages taken by swarms in the nature, using individualistic intelligence as well as the collective intelligence. Optimization methods founded on swarm intelligence are called behaviorally enthused processes which are called evolution-based procedures. The PSO algorithm was originally proposed by Kennedy and Eberhart in 1995 [1]. On the whole, each particle tries to observe three rules instinctively in swarms and these rules are the bases of the optimization algorithm as well. Rules are as follows [1]: 1. It tries not to come too near and not to go too far from other individuals simultaneously. 2. It directs toward the middling track of other individuals. 3. It tries to fit the “average position” among other individuals with no extensive gaps in the flock. Three rules mentioned above lead to the behavior of the swarm which is based on a mixture of three simple features as follows [1]: 1. Cohesion, which tries to keep the swarm altogether. 2. Separation, which causes the individuals not to came too close to each other. 3. Alignment, which causes the swarm to keep an eye on the general heading point of the flock.
368
A. Jafari Aghbolaghi et al.
In general, size of the swarm is assumed to be fixed, whereas each particle situated primarily at accidental positions in the multidimensional space of the optimization problem. Each individual has two sorts of data as position and velocity, both of which are stored and compared to each other continuously during the optimization procedure in order to work out the best position discovered by the particles. The velocity and position in the previous iteration are used to determine the new values of them in the next iteration. This process runs consecutively until it has found the best position possible or has reached one of the stopping criteria. Therefore, in simple words the procedure of particle swarm optimization algorithm is generally as follows [6]: 1. All the particles exchange their latest information with each other simultaneously to figure out which particle has found the best location so far. 2. All the particles considering the location and velocity of each one incline to the best point found in relation to their current parameters. 3. The past memory of each particle as well as its current position affects the next position where it will be. Overall, the particle swarm optimization algorithm quests for the optimum position via a group of individuals similar to other AI-based exploratory optimization methods. The presented model simulates a partly random search method that is armed with individualistic and collective artificial intelligence, leading the process to a global optimum point of the objective function.
10.5.1 Computational Implementation of PSO for Reactive Power Optimization Since the concept of particle swarm optimization algorithm has been elaborated in many articles and book chapters like [6, 9], this section focuses on the computational implementation of particle swarm algorithm in reactive power optimization problem instead of the algorithm itself. A reactive power optimization problem including the constraints using penalty function can be considered as: f ðV 0 ; h 0 Þ ¼
N X k¼1
þ kv
h 02 02 gk V1;k þ V2;k M X 0 V j j¼1
0 0 2V1;k V2;k cos h01;k
T X 1 Vj0ref þ ks i¼1
Vi0 Vi00
h02;k
i
ð10:74Þ
As particle swarm optimization algorithm is naturally good at maximizing objective functions, then the appropriate fitness function in order to minimize the target function in Eq. (10.74) will be as follows
10
Reactive Power Optimization in AC Power Systems
maximize FðV 0 ; h0 Þ ¼
369
1 f ðV 0 ; h0 Þ
ð10:75Þ
Subject to control variables, such as terminal voltages of generator buses, tap settings of transformers, and reactive power output of compensators X ¼ ½V
C
T
ð10:76Þ
V ¼ ½ V1
V2
...
Vn
ð10:77Þ
T ¼ ½ T1
T2
. . . Tm
ð10:78Þ
C ¼ ½ C1
C2
...
Cd
ð10:79Þ
X ðlÞ ¼ ½ Vmin
Tmin
Cmin
ð10:80Þ
X ðuÞ ¼ ½ Vmax
Tmax
Cmax
ð10:81Þ
where, X is the vector of control variables, while X(l) and X(u) are lower and upper bounds of them, V is a vector containing terminal voltages of generator buses, T is a vector consisting of tap settings of transformers, and C is the vector of reactive power compensators. n, m and d are the number of generator buses, transforms having tap changing facility, and compensators, respectively. The PSO procedure can be implemented through the following steps [1]: 1. Consider N as the size of the swarm which is mostly between 20 and 30. It is obvious that taking big numbers will raise the time of evaluation of the objective function and will influence the total calculation time. However, it should not be too small either as it can affect the performance of the PSO algorithm. In general, the size of the population is obtained using trial and error method for each optimization problem, although there as some approximate methods to apply in order to get appropriate numbers for them [6]. 2. Produce the primary population of X in between X(l) and X(u) randomly as X1, X2, …, XN. Henceforth, for the sake of convenience, the position of jth individual and its speed in ith iteration are signified as X(i) and V(i) j j , respectively. Accordingly, the particles created initially are indicated as X1(0), X2(0), …, XN(0). The vectors Xj(0) (j = 1, 2, …, N) are called particles or coordinate vectors of particles [6]. 3. Apply the generated control variables to the power network and run power flow in order to obtain the values of state variables, because the objective function in Eq. (10.75) needs the values of both state (voltage magnitudes of PQ buses) and control variables (voltage values of PV buses). After doing power flow, the bus voltages and voltage angles will be considered as follows [6]:
370
A. Jafari Aghbolaghi et al.
Yj ðiÞ ¼ V 0
h0
ð10:82Þ
where, V′ is the vector of voltage magnitudes and h′ is the vector of voltage angles of all the buses. 4. Work out the values of the objective function for each particle as F[(Y1(0)), F (Y2(0)), …, F(YN(0)]. 5. Determine the velocity values for all the particles. The velocity will help leading the particles through reaching the optimum point. Primarily, the speed value for all the particles are presumed zero. Then, set the iteration number to one (i = 1). 6. In the ith iteration, the two following significant parameters should be calculated using the data of jth particle: (a) Work out the best value of the objective function for Xj among all the iterations and allocate it to Pbest,j (personal best) which is the best value of the F[Yj(i)] found by jth individual so far. Find the best value of the objective function for all the particles (X) among all the iterations so far and assign it to Gbest (global best), which is the best value found for the objective function F[Yj(i)] j = 1, 2, 3, …, N [1]. (b) Find the velocity of jth particle in ith iteration by means of the following equation: Vj ðiÞ ¼ hVj ði
1Þ þ c1 r1 Pbest; j Xj ði 1Þ þ c2 r2 Gbest j ¼ 1; 2; . . .; N
Xj ði
1Þ
ð10:83Þ
where, c1 and c2 are the perceptive (individual) and collective (group) learning coefficients, respectively. r1 and r2 are uniformly distributed random numbers in the range of 0 and 1. The factors c1 and c2 signify the comparative rank of the memory (location) of the particles to the memory (location) of the swarm. c1 and c2 are commonly considered to have the amount of 2 in a lot of implementations. The inertia weight h is a constant which is used in order to decrease the velocities as time goes by (or iterations), facilitating the swarm to congregate more precisely and proficiently compared to the original PSO algorithm. Equation (10.83) represents a formulation for adjusting the velocity, which helps the accuracy increase. Equation (10.83) demonstrates that a greater amount of h supports global exploration, while a smaller value encourages a local exploration better. Hence, a great value of h marks the algorithm continually discover new areas deprived of much local examination, failing to find the true optimum point. To strike a balance between global and local search in order to speed up converging to the exact optimal location, an inertia coefficient whose value declines linearly in relation to the iteration number has been used as [6]
10
Reactive Power Optimization in AC Power Systems
hðiÞ ¼ hmax
371
hmax hmin i imax
ð10:84Þ
where, hmax and hmin are primary and final values of the inertia weight, respectively. imax is the maximum quantity of iterations used in PSO. Values of hmax = 0.9 and hmin = 0.4 are commonly used [6]. (c) Find the coordinate of the jth particle in ith iteration using [6] Xj ðiÞ ¼ Xj ði
1Þ þ Vj ðiÞ;
j ¼ 1; 2; . . .; N
ð10:85Þ
Evaluate the values of objective function for all the particles as F[Y1(i)], F [Y2(i)], …, F[YN(i)], in which the matrix Y is acquired from the power flow calculations after applying new control variables to the power network [6]. 7. Check if the algorithm has reached the optimal point or not, which is applicable via creating and checking a few stopping criteria for the optimization problem. If the points of all particles congregate to the similar set of values, the technique is considered to be converged. If the convergence criterion is not met, step 6 will be reiterated by updating the iteration number to i = i + 1, and calculating the new values for Pbest,j and Gbest. The reiterative procedure will continue until almost all of the particles congregate to the same optimal point. The flow chart of PSO algorithm founded on above declared process is shown in Fig. 10.9 [6]. There are plenty of stopping criteria to use, of which can be chosen in relation to what the algorithm is expected to do or reach. The first one can be the iteration number so that the process will stop after a specified number of iterations whether the algorithm has reached the optimum point or not. The second one can be considered as the time duration of the process running, which completely depends on how heavy the process might be. The third one can be to use a pre-specified fitness value and compare the last overcome of the optimization procedure to it, and then if the appropriateness of the solution is met, the process can be stopped. The fourth one can be taken as whether there is any substantial progression in the result or not, and stop the process if not. Usually a combination of several criteria is used in practice appropriate to what the expectation is from the algorithm and the procedure. Mostly, the criteria that consider just the calculation cost such as the time duration and the number of iterations can be used with some other criteria that consider the fitness of the result.
372
A. Jafari Aghbolaghi et al.
Fig. 10.9 The flowchart of reactive power optimization using PSO
Start
Create an initial swarm of control variables
Apply the values of control variable to the power network
Run power flow and obtain values for the state variables i=i+1 Evaluate the fitness of each particle
Check and update the individual and global best Update positions of the individuals Update the velocity value of each individual
Dissatisfied Check stopping criteria
Satisfied Stop
10.6
Pattern Search Optimization Algorithm
Pattern search (PS) algorithm is one of the arithmetical optimization approaches that do not need the gradient of the objective function which is meant to be optimized. Therefore, PS can be used on discrete functions that are not differentiable as well. Such optimization means are also recognized as direct-search, derivative-free, or black-box methods. They are founded on producing search patterns which positively sweep the search space in order to detect the minimum points. The procedure of optimizing starts from a random starting point, though being highly dependent on what location they start from (initial point) is their biggest defect. This drawback can cause the optimization procedure to get stuck in one of the local minimums which is not favorable. However, because of their flexibility, they can be
10
Reactive Power Optimization in AC Power Systems
373
integrated with heuristic algorithms (global optimizers) for global search, which is a mode pattern search technique receives some of the properties of the imported global optimization technique (global optimization), without risking the convergence and being trapped in one of the local optimum points [10–12].
10.6.1 Mathematical Description of Pattern Search Optimization Algorithm Pattern search algorithm can be counted as a direct search optimization algorithm in which the central notion is on the positive spanning sets. This section will present the PS algorithm in search/poll framework which is the best choice when it comes to cooperate with heuristic algorithms. One of the simplest positive spanning sets is formed by the vectors of the canonical foundation and also their negatives [10, 12] D ¼ fe1 ; . . .; en ; e1 ; . . .;
en g
ð10:86Þ
The set D is also a (highest) positive basis. The basic straight search technique based on this progressive spanning set is recognized as coordinate or compass search. Bearing in mind a progressive spanning set as D and the recent iteration y(t), two groups of points are defined as the net Mt and the election set Pt. The net Mt is known as [10] n o jDj Mt ¼ yðtÞ þ aðtÞDz ; z 2 Z þ
ð10:87Þ
where a(t) > 0 is the mesh dimension factor and also identified as the step-length controller, and Z þ is the set containing nonnegative integer numbers. The mesh has to fulfil some integrality necessities for the technique to attain global convergence to static points from random initial points. The matrix D has to be in the formula of ^ where G 2 Rnn is a nonsingular producing matrix and G 2 Rnn . The proGZ, gressive foundation D meets the prerequisites when G is entitled the identity matrix. The pursuit step conducts a limited exploration in the net Mt. The poll step is performed only if the examination step fails to discover a position for which f is lesser than f(y(t)). The poll step assesses the function at the positions in the poll set Pt in order to discover a location where f is minor than f(y(t)) [10, 12] Pt ¼ fyðtÞ þ aðtÞd; d 2 Dg
ð10:88Þ
It should be noted that Pt is a subdivision of Mt. If f is constantly differentiable at y(t), the poll step is assured to succeed if a(t) is appropriately small, since the progressive spanning set D comprises at least one pattern of descent which sorts an acute angle with rf ðyðtÞÞ. Consequently, if the poll step flops to discover a coordination better than the former one, the net size factor must be made smaller. The poll step which is the key tool of pattern search approach to explore the optimal point guarantees the global convergence.
374
A. Jafari Aghbolaghi et al.
In order to extrapolate pattern search optimization procedure for bound constrained problems, it is indispensable to use a practicable primary guess y(0)2X and to keep feasibility of the iterates safe by declining any trial position that is out of the acceptable region. Rejecting unviable test locations can be achieved by applying a pattern search algorithm to the subsequent penalty function [10, 12]. if z 2 X ^f ðzÞ ¼ f ðzÞ ð10:89Þ þ 1 otherwise There is no big dissimilarity between constrained and unconstrained pattern search optimization algorithm excluding it is applied to the minimization of f (z) subject to simple bounds and to the refusal of impractical test points. It is also essential to embrace in the exploration directions D those patterns that warranty the existence of a practicable descent track at any nonstationary location of the bound constrained problem [10]. In order to have an elaborate depiction of the basic pattern search algorithm, it is necessary to state in what way to expand and contract the net size or step-length control factor a(t). The growths and reductions use the parameter /(t) and h(t), respectively, which must observe the subsequent rules: /ðtÞ ¼ T lt ; for some lt 2 f0; . . .; lmax g if t is successful hðtÞ ¼ T mt ; for some mt 2 fmmin ; . . .; 1g if t is unsuccessful
ð10:90Þ
where, T [ 1 is a positive rational, lmax is a nonnegative integer, and mmin is a negative integer, selected at the commencement of the procedure and unaffected with t. For instance, it can be considered h(t) = 0.5 for unproductive iterations and /(t) = 1 or /(t) = 2 for up-and-coming iterations [10, 12]. The process of basic pattern search method has been offered in follows [10]: 1. Select a positive rational T and the tolerance atol > 0 as the stopping criterion. Pick the positive spanning set D ¼ D . 2. Set t = 0. Choose an primary practical guess y(0). Pick a(0) > 0. 3. [Search Step], Assess f at a limited number of points in Mt. If a position z(t) 2Mt is discovered for which ^f ðzðtÞÞ\^f ðyðtÞÞ then set y(t + 1) = z(t), a(t + 1) = /(t)a (t) (optionally increasing the net size factor), and tag both the exploration step and the present iteration as successful. 4. [Poll Step], Avoid the poll step if the examination step was successful. • If there exists d(t) 2 D so that ^f ðyðtÞ þ aðtÞdðtÞÞ\^f ðyðtÞÞ, then: Set y(t + 1) = y(t) + a(t)d(t) (poll step and iteration successful). Set a(t + 1) = /(t)a(t) (optionally increase the net size factor). • Otherwise, ^f ðyðtÞ þ aðtÞdðtÞÞ ^f ðyðtÞÞ for all d(t) 2 D, and: Set y(t + 1) = y(t) (iteration and poll step unsuccessful). Set a(t + 1) = h(t)a(t) (contract the net size factor). 5. If a(t + 1) < atol then break, where atol is the least value which is defined as the mesh dimension factor. Otherwise, increase t by one and go to Step 3.
10
Reactive Power Optimization in AC Power Systems
375
10.6.2 Pattern Search Algorithm in Simple Words The flowchart of pattern search algorithm is shown in Fig. 10.10 in the simplest way possible. At first an initial point as well as initial step size are produced randomly as x0 2 Rn (for one dimensional problem) and D0 [ 0, respectively. Then, the fitness function will be calculated in neighboring points x0 ; x0 þ D; x0 D. If f ðx0 þ DÞ\f ðx0 Þ then the point x0 þ D will be considered as the center point and D will be added to it, then new fitness values will be computed. This procedure will be continued until the fitness value does not get better, then the value of D will be decreased and added to center point, afterwards new fitness values will be calculated again. This procedure will go on until the stopping criteria stop the algorithm. A simple graphical example has shown in the Fig. 10.11 in order to illustrate how the algorithm works.
10.7
Particle Swarm Pattern Search Algorithm
The most considerable feature that distinguishes heuristic optimization techniques from traditional ones is their capability to find the global optimum point, not needing the gradient of the objective function as a great asset, so that the cost of the calculation (considering time as a resource) will be decreased. Nevertheless, their local minimization is not as efficient as their global optimization, or in another words, is more time consuming in comparison with to the extent the function gets minimized. Moreover, pattern search algorithm is such a great direct local minimizer, although in some cases it has proved its power finding even global optimum point. The more problem space gets non-convex, however, the more pattern search algorithm seems to fail finding the global optimum point, because of the so many local optimum points which exist. Therefore, the pattern search algorithm will be more practical finding local optimum points rather than global ones. In addition, pattern search algorithm is highly dependent on the starting point which is normally chosen randomly so that different starting points can lead the algorithm to different solutions not probably being a global optimum point. The idea is to take the advantages of both types using heuristic algorithms as the global optimizers and direct search methods as local optimizers, in this case pattern search algorithm. Since pattern search algorithm needs a starting point and the final answer is directly dependent on it, considering the global minimum point found by intelligent algorithms like PSO that is mostly near the exact solution as a starting point to pattern search algorithm will fasten the optimization procedure finding the more accurate point. Likewise, if there is a time limitation to calculation, using the particle swarm pattern search algorithm will not be empty of favor [10]. There are some different strategies combining an intelligent optimization algorithm and a direct search method, each of them has its own advantages and disadvantages. For instance, one can be considered as a series one where a heuristic algorithm is chosen to implement a global optimization, after reaching near the global optimum point the direct search method is used in order to pinpoint the local one of the neighboring search space in which the best optimum point lies. The first
376
A. Jafari Aghbolaghi et al.
Start
Specify expansion and contraction multipliers
Produce the initial point randomly
Calculate fitness function
Produce points having specified structure of reticulation around the initial point
Calculate fitness function for the new points created
Has fitness value got better?
Use contraction multiplier
No
Yes Use expansion multiplier
Select the best candidate as the central point
Dissatisfied Check stopping criteria
Satisfied Stop
Fig. 10.10 Flowchart of pattern search optimization algorithm
10
Reactive Power Optimization in AC Power Systems
Fig. 10.11 Graphical show of how pattern search optimization algorithm works [12]
377
378 Fig. 10.12 Flowchart of the first strategy using heuristic algorithms incorporated with direct search method
A. Jafari Aghbolaghi et al.
Start
Use an intelligent algorithm to implement global optimization, such as PSO
Use the final result of the global optimizer as the starting point for the direct search algorithm to pinpoint the optimum value
Stop
strategy is the simplest and fastest one which can be taken into practice. The flowchart of the first strategy has been shown in Fig. 10.12. The second strategy can be like using a direct search method for each point found by particles (PSO) or genes (genetic algorithm) in order to find the exact optimum point using positions of particles of each iteration as starting points for direct search methods. This approach can be more time consuming than the first one, as the direct search process should be applied to each particle. Although, It sounds to be more accurate and also more reliable than waiting to an intelligent algorithm to find a global optimum point, in each iteration an exact optimum point (probably not global one at the beginning) will be found so that in online implications would be at work. The flowchart of second strategy has been shown in Fig. 10.13 [10, 12]. In order to avoid implementation difficulties, the first strategy has been chosen to be presented in this section. In addition, it can be turned to the other strategy readily considering a few changes in the procedure. Therefore, the first strategy as particle swarm pattern search algorithm can be coincided as follows: The particle swarm pattern search algorithm procedure can be implemented through the following steps [6]: 1. Consider N as the size of the swarm which is mostly between 20 and 30. It is obvious that taking big numbers will raise the time of evaluation of the objective function and will influence the total calculation cost. However, it should not be too small as it can affect the performance of the PSO algorithm. In general, the number of population is obtained using trial and error method for each optimization problem, although there are some approximate methods to apply in order to get the appropriate numbers. Choose appropriate stopping criteria for PSO algorithm, considering the fact that optimization procedure will go on after getting to the neighborhood of global optimum point with pattern search algorithm which has its own stopping criteria [6].
10
Reactive Power Optimization in AC Power Systems
Fig. 10.13 The flowchart of the second strategy using heuristic algorithms incorporated with direct search methods
379
Start
Create an initial swarm (PSO)
Use coordinates of each particle as a starting point to the direct search algorithm
Apply direct search method to each particle (PS) i=i+1 Consider new points found by direct search method as new positions of particles
Check and update the individual and global best (PSO)
Update the position of all individuals
Update the velocity value for each individual (PSO)
Dissatisfied Check stopping criteria
Satisfied Stop
2. Produce the primary population of X in the assortments X(l) and X(u) randomly as X1, X2, …, XN. Henceforth, for the sake of convenience, the position of and (i) velocity of jth individual in ith iteration are signified as X(i) j and Vj , respectively. Accordingly, the particles created at the beginning are indicated as X1(0), X2(0), …, XN(0). The vectors Xj(0) (j = 1, 2, …, N) are called particles or vectors of coordinates of particles [6]. 3. Apply the generated control variables to the power network and run power flow in order to obtain state variables as they are needed in the objective function represented by Eq. (10.75). Therefore, after doing power flow the bus voltages and voltage angels will be considered as follows [6]
380
A. Jafari Aghbolaghi et al.
Yj ðiÞ ¼ V 0
h0
ð10:91Þ
where, V′ is the vector of voltage magnitudes of all buses and h′ is the bus voltage angles. 4. Calculate the objective function values for all the particles as F(Y1(0)), F (Y2(0)), …, F(YN(0)]. 5. Find the velocities of particles, which will help particles reach the optimum point. Primarily, the velocities of all particles are presumed zero. Set the iteration number to i = 1. 6. In the ith iteration, the two following significant parameters should be calculated using the data of jth particle: (a) Work out the best value of the objective function for Xj among all the iterations and allocate it to Pbest,j (personal best) which is the best value of the F[Yj(i)] found by jth individual so far. Find the best value of the objective function for all the particles (X) among all the iterations so far and assign it to Gbest (global best), which is the best value found for the objective function F[Yj(i)] j = 1, 2, 3, …, N [1]. (b) Find the velocity of jth particle in ith iteration by means of the following equation Vj ðiÞ ¼ hVj ði
1Þ þ c1 r1 Pbest; j Xj ði 1Þ þ c2 r2 Gbest j ¼ 1; 2; . . .; N
Xj ði
1Þ
ð10:92Þ
where, c1 and c2 are the perceptive (individual) and collective (group) learning coefficients, respectively. r1 and r2 are uniformly distributed random numbers in the range of 0 and 1. The factors c1 and c2 signify the comparative rank of the memory (location) of the particles to the memory (location) of the swarm. c1 and c2 are commonly considered to have the amount of 2 in a lot of implementations. The inertia weight h is a constant which is used in order to decrease the velocities as time goes by (or iterations). hðiÞ ¼ hmax
hmax hmin i imax
ð10:93Þ
where, hmax and hmin are primary and final values of the inertia weight, respectively. imax is the maximum quantity of iterations used in PSO. Values of hmax = 0.9 and hmin = 0.4 are commonly used [1].
10
Reactive Power Optimization in AC Power Systems
381
(c) Calculate the location or coordinate of the jth particle in ith iteration as [6] Xj ðiÞ ¼ Xj ði
1Þ þ Vj ðiÞ;
j ¼ 1; 2; . . .; N
ð10:94Þ
Evaluate the values of objective function for all the particles as F[Y1(i)], F [Y2(i)], …, F[YN(i)], in which the matrix Y is acquired from the power flow calculations after applying new control variables to the power network [1]. 7. Check if the algorithm has reached the optimal point or not, which is applicable via creating and checking a few stopping criteria for the optimization problem. If the points of all particles congregate to the similar set of values, the technique is considered to be converged. If the convergence criterion is not met, step 6 will be reiterated by updating the iteration number to i = i + 1, and calculating the new values for Pbest,j and Gbest. The reiterative procedure will continue until almost all of the particles congregate to the same optimal point. The flow chart of PSO algorithm founded on above declared process is shown in Fig. 10.9 [6]. 8. Select a positive rational T and the tolerance atol > 0 as the stopping criterion. Pick the positive spanning set D ¼ D . 9. Set t = 0. Choose a primary practical guess y(0) and pick a(0) > 0. The best position found by PSO algorithm will be considered as the initial point in this step. 10. Apply control variables to the power network and implement power flow to calculate the state values as
Y ¼ V0
h0
ð10:95Þ
C
ð10:96Þ
11. [Search Step], Assess f using Eq. (10.74) at a limited number of positions in Mt. If a coordinate Z 0 ðtÞ 2 Mt is discovered for which f ðYZ 0 ðtÞ Þ\f ðYZðtÞ Þ, [where, ðYZ 0 ðtÞ Þ is a matrix containing the values for the state variables after doing power flow for Z 0 ðtÞ], then consider ZðtÞ ¼ Z 0 ðtÞ and a(t + 1) = /(t)a(t) (optionally increasing the net size factor), and state successful both the exploration step and the present iteration. Z ¼ ½V
T
V ¼ ½ V1
V2
...
Vn
ð10:97Þ
T ¼ ½ T1
T2
. . . Tm
ð10:98Þ
C ¼ ½ C1
C2
...
Cd
ð10:99Þ
a ¼ ½ a1
a2
. . . an
ð10:100Þ
12. [Poll Step], Avoid the poll step if the exploration step was successful. • If there exists [Z(t) + a(t)d(t)] 2 D so that f ðYZðtÞ þ aðtÞdðtÞ Þ\f ðYZðtÞ Þ, then
382
A. Jafari Aghbolaghi et al.
Consider Z(t + 1) = Z(t) + a(t)d(t) (poll stage and iteration successful). Adjust a(t + 1) = /(t)a(t) (optionally increase the net size factor). • Otherwise, f ðYZðtÞ þ aðtÞdðtÞ Þ f ðYZðtÞ Þ for all d(t) 2 D, and Consider Z(t + 1) = Z(t) (iteration and poll stage unsuccessful). Adjust a(t + 1) = h(t)a(t) (contract the net size factor). 13. If a(t + 1) < atol then stop, where atol is the least value which is defined as the mesh dimension factor. Otherwise, increase t by one and go to step 10. The flowchart of particle swarm pattern search algorithm which was mentioned above is shown in Fig. 10.14. In the next section practical implementation of reactive power optimization using the proposed particle swarm pattern search algorithm will be presented.
10.8
Simulation Results of Reactive Power Optimization
Two algorithms as particle swarm pattern search and genetic pattern search algorithms have been implemented on two standard systems as IEEE 6-bus and IEEE 14-bus and the results have been compared. In addition, the simulation results for IEEE 39-bus New England power network using particle swarm pattern search algorithm has been presented in this section. Besides, how to implement such optimization procedures using MATLAB and DIgSILENT has been presented in Chap. 11 step by step.
10.8.1 Case Study 1—IEEE 6-Bus Power Network The corresponding data for IEEE 6-bus power grid has been presented in Appendix 1. The results for reactive power optimization on the related network will be offered in this section.
10.8.1.1
Reactive Power Optimization Using Particle Swarm Pattern Search Algorithm
The initial conditions and power flow results for 6-bus power network are presented in Table 10.19. In addition, the corresponding data after the optimization procedure on the same power grid are presented as below Figs. 10.15, 10.16 and 10.17. Referring to Tables 10.1, 10.2 and 10.3, it is obvious that operational parameters have been improved considerably. Active power losses have decreased from
10
Reactive Power Optimization in AC Power Systems
383
Fig. 10.14 Flowchart of particle swarm pattern search optimization algorithm
Start
Create an initial swarm of control variables
Apply the values of control variables to power network
Run power flow and obtain the values for state variables i = i+ 1 Evaluate the fitness for each particle
Check and update the individual and global best Update individuals’ position Update the velocity of each individual
Dissatisfied Check stopping criteria
Satisfied Choose the best particle as the starting point for pattern search algorithm
Produce points having specified structure of reticulation around the initial point
Apply the control variables’ values to the power network
Run power flow and obtain the state variables Use contraction multiplier Calculate fitness function for the new points created
Has fitness value got better?
Yes Use expansion multiplier
Select the best candidate as the central value
Dissatisfied Check stopping criteria
Satisfied Stop
No
384
A. Jafari Aghbolaghi et al.
Fig. 10.15 Reactive power optimization trend for 6-bus power system using particle swarm pattern search algorithm
Fig. 10.16 Voltage stability index for 6-bus power system using particle swarm pattern search algorithm
10
Reactive Power Optimization in AC Power Systems
385
Fig. 10.17 Voltage deviation for 6-bus power system using particle swarm pattern search algorithm
Table 10.1 Power flow results after optimization for 6-bus power system using particle swarm pattern search algorithm Bus no.
Voltage magnitudes and angles V (kV) h (degree)
Load consumption
Injection power
Pl (MW)
Ql (MVar)
PG (MW)
QG (MVar)
1 2 3 4 5 6
62.97 72.45 62.98 63.86 61.82 69.30
55 0 50 0 0.3 0
13 0 5 0 18 0
0 50 0 0 0 93.96
0 22.65 5 0 5 33.13
−11.69 −3.59 −10.97 −8.79 −11.23 0
Table 10.2 Active power losses for 6-bus power system using particle swarm pattern search algorithm
Active power losses (initial condition) Active power losses (after optimization) Reduction percentage
10,778,370 (W) 8,950,232 (W) 16.96121%
10,778,370 to 8,950,232 (W). Likewise, the total value of voltage stability index has increased from 2.665148 to 2.715127, and the total value of voltage deviation of load buses has been reduced considerably from 15.65904 to 1.185288 kV.
386
A. Jafari Aghbolaghi et al.
Table 10.3 Voltage deviation and voltage stability data for 6-bus power system using particle swarm pattern search algorithm Voltage deviation of load buses
Bus no.
Voltage stability index value V/V0 ideal value = 1 Initial status After optimization
Initial status kV
After optimization kV
1 3 5 Total
0.9009016 0.8845708 0.8796755 2.665148
4.881641 4.934923 5.842471 15.65904
0.03538300 0.02064700 1.129258 1.185288
0.9161407 0.9021702 0.8968166 2.715127
Fig. 10.18 Reactive power optimization trend for 6-bus power system using genetic pattern search algorithm
10.8.1.2
Reactive Power Optimization Using Genetic Pattern Search Algorithm
The initial conditions and power flow results for 6-bus power network are presented in Table 10.18. In addition, the corresponding data after the optimization procedure on the same power grid are presented as below Figs. 10.18, 10.19 and 10.20. Considering Tables 10.4, 10.5 and 10.6, a marked improvement is obvious in the operational parameters. Active power losses have decreased from 10,778,370 to 8,950,550 (W). Similarly, the total value of voltage stability index has increased from 2.665148 to 2.712409, and the total value voltage deviation of load buses has reduced from 15.65904 to 2.127879 kV.
10
Reactive Power Optimization in AC Power Systems
387
Fig. 10.19 Voltage stability index for 6-bus power system using genetic pattern search algorithm
Fig. 10.20 Voltage deviation for 6-bus power system using genetic pattern search algorithm
388
A. Jafari Aghbolaghi et al.
Table 10.4 Power flow results after optimization for 6-bus power system using genetic pattern search algorithm Bus no.
Voltage magnitudes and angles V (kV) h (degree)
Load consumption
Injection power
Pl (MW)
Ql (MVar)
PG (MW)
QG (MVar)
1 2 3 4 5 6
62.96 72.45 62.98 63.86 61.87 69.30
55 0 50 0 0.3 0
13 0 5 0 18 0
0 50 0 0 0 93.96
0 22.65 5 0 5 33.13
−11.69 −3.59 −10.97 −8.79 −11.23 0
Table 10.5 Active power losses data for 6-bus power system using genetic pattern search algorithm
Active power losses (initial condition) Active power losses (after optimization) Reduction percentage
10,778,370 (W) 8,950,550 (W) 16.95826%
Table 10.6 Voltage deviation and voltage stability data for 6-bus power system using genetic pattern search algorithm Voltage deviation of load buses
Bus no.
Voltage stability index value V/V0 Ideal value = 1 Initial status After optimization
Initial status kV
After optimization kV
1 3 5 Total
0.9009016 0.8845708 0.8796755 2.665148
4.881641 4.934923 5.842471 15.65904
0.4280500 0.3137400 1.386089 2.127879
0.9152744 0.9011948 0.8959399 2.712409
10.8.2 Case Study 2—IEEE 14-Bus Power Network The corresponding data to IEEE 14-bus power grid are accessible in Appendix 2. The results for reactive power optimization on the related network will be presented in this section.
10.8.2.1
Reactive Power Optimization Using Particle Swarm Pattern Search Algorithm
The initial circumstances and power flow results for 14-bus power network are presented in Table 10.22. In addition, the corresponding data after the optimization procedure on the same power grid are presented as below Figs. 10.21, 10.22 and 10.23.
10
Reactive Power Optimization in AC Power Systems
389
Fig. 10.21 Reactive power optimization trend for 14-bus power system using particle swarm pattern search algorithm
Fig. 10.22 Voltage stability index for 14-bus power system using particle swarm pattern search algorithm
Regarding Tables 10.7, 10.8 and 10.9, active power losses have decreased from 16,710,700 to 9,823,200 (W). Likewise, the total value of voltage stability index has increased from 8.953467 to 9.708002 and total value of voltage deviation of load buses has reduced considerably from 219.7092 to 56.17259 kV.
390
A. Jafari Aghbolaghi et al.
Fig. 10.23 Voltage deviation for 14-bus power system using particle swarm pattern search algorithm
Table 10.7 Power flow results after optimization for 14-bus power system using particle swarm pattern search algorithm Bus no.
Voltage magnitudes and angles V (kV) h (degree)
Load consumption
Injection power
Pl (MW)
Ql (MVar)
PG (MW)
QG (MVar)
1 2 3 4 5 6 7 8 9 10 11 12 13 14
125.376906 121.000004 114.908254 116.456311 117.737189 116.381916 114.224630 114.805564 112.753677 112.559232 114.056146 114.573162 113.897577 111.226337
0 21.7 94.2 47.8 7.6 11.2 0 0 29.5 9 3.5 6.1 13.5 14.9
0 12.7 19 0 1.6 7.5 0 0 16.6 5.8 1.8 1.6 5.8 5
231.36 40 0 0 0 0 0 0 0 0 0 0 0 0
45.67 16.73 7.436791 0 0 2.399997 0 3.129134 0 0 0 0 0 0
0 −3.957964 −10.725941 −8.576776 −7.335156 −13.008229 −11.632678 −11.632674 −13.290581 −13.530266 −13.391523 −13.866247 −13.900069 −14.602206
10
Reactive Power Optimization in AC Power Systems
Table 10.8 Active power losses data for 14-bus power system using particle swarm pattern search algorithm
391
Active power losses (initial condition) Active power losses (after optimization) Reduction percentage
16710700 (W) 9823200 (W) 41.21612%
Table 10.9 Voltage deviation and voltage stability data for 14-bus power system using particle swarm pattern search algorithm Bus no.
2 3 4 5 6 9 10 11 12 13 14 Total
10.8.2.2
Voltage deviation of load buses
Voltage stability index value V/V0 ideal value = 1 Initial status After optimization
Initial status kV
After optimization kV
9.409176 8.398882 8.563136 8.728775 7.977896 7.835140 7.761442 7.822556 7.780087 7.718829 7.538748 8.953467
6.499059 16.96615 15.24372 13.45913 21.72454 23.25914 24.08241 23.42380 23.91052 24.58441 26.55627 219.7092
11 4.975350 6.534758 7.815445 6.421239 2.831756 2.630989 4.112191 4.615991 3.943326 1.291542 56.17259
0.9552817 0.8916201 0.9047917 0.9160301 0.8778006 0.8636542 0.8598163 0.8658360 0.8651542 0.8609605 0.8470567 9.708002
Reactive Power Optimization Using Genetic Pattern Search Algorithm
The initial circumstances and power flow results for 14-bus power network are presented in Table 10.22. In addition, the corresponding data after the optimization procedure on the same power grid are presented as below Figs. 10.24, 10.25 and 10.26. Referring to Tables 10.10, 10.11 and 10.12, active power losses have decreased remarkably from 16,710,700 to 9,862,619 (W). Similarly, the total value of voltage stability index increased from 8.953467 to 9.655891 and the total value of voltage deviation of load buses has reduced considerably from 219.7092 to 57.82206 kV.
10.8.3 Case Study 3—IEEE 39-Bus New England Power Network The corresponding data to IEEE 39-bus New England power grid are accessible in Appendix 3. The results for reactive power optimization on the related network will be presented in this section.
392
A. Jafari Aghbolaghi et al.
Fig. 10.24 Reactive power optimization trend for 14-bus power system using genetic pattern search algorithm
Fig. 10.25 Voltage stability index for 14-bus power system using genetic pattern search algorithm
10
Reactive Power Optimization in AC Power Systems
393
Fig. 10.26 Voltage deviation for 14-bus power system using genetic pattern search algorithm
Table 10.10 Power flow results after optimization for 14-bus power system using genetic pattern search algorithm Bus no.
Voltage magnitudes and angles V (kV) h (degree)
Load consumption
Injection power
Pl (MW)
Ql (MVar)
PG (MW)
QG (MVar)
1 2 3 4 5 6 7 8 9 10 11 12 13 14
125.803591 120.793094 115.351591 116.493973 117.785497 116.251125 114.143428 114.561700 112.686674 112.480325 113.951417 114.445424 113.773793 111.132858
0 21.7 94.2 47.8 7.6 11.2 0 0 29.5 9 3.5 6.1 13.5 14.9
0 12.7 19 0 1.6 7.5 0 0 16.6 5.8 1.8 1.6 5.8 5
231.42 40 0 0 0 0 0 0 0 0 0 0 0 0
58.56 −0.15 12.74691 0 0 23.72837 0 2.373186 0 0 0 0 0 0
0 −3.853065 −10.701274 −8.503816 −7.265805 −12.931138 −11.564184 −11.564181 −13.225312 −13.463648 −13.320288 −13.791402 −13.826640 −14.535142
394
A. Jafari Aghbolaghi et al.
Table 10.11 Active power losses data for 14-bus power system using genetic pattern search algorithm
Active power losses (Initial condition) Active power losses (After optimization) Reduction percentage
16710700 (W) 9862619 (W) 40.98023%
Table 10.12 Voltage deviation and voltage stability data for 14-bus power system using genetic pattern search algorithm Bus no.
2 3 4 5 6 9 10 11 12 13 14 Total
10.8.3.1
Voltage deviation of load buses
Voltage stability index value V/V0 ideal value = 1 Initial status After optimization
Initial status kV
After optimization kV
9.409176 8.398882 8.563136 8.728775 7.977896 7.835140 7.761442 7.822556 7.780087 7.718829 7.538748 8.953467
6.499059 16.96615 15.24372 13.45913 21.72454 23.25914 24.08241 23.42380 23.91052 24.58441 26.55627 219.7092
10.59916 4.580411 6.543957 7.844678 6.543719 3.448204 3.164167 4.445315 4.777684 4.141013 1.733747 57.82206
0.9476370 0.8850283 0.8994890 0.9111127 0.8734883 0.8599726 0.8561004 0.8618414 0.8609751 0.8568609 0.8433849 9.655891
Reactive Power Optimization Using Particle Swarm Pattern Search Algorithm
The initial circumstances and power flow results for 39-bus New England power network are presented in Table 10.13. In addition, the corresponding data after the optimization procedure on the same power grid are presented as below Fig. 10.27. One of the best criteria that can be taken into consideration as the calculation cost is Number of Function Evaluation (NFE) when it comes to comparing the performance of two or more intelligent algorithms, especially the ones which do not use the gradient of objective functions but the function itself. As it can be inferred from the name, each time the algorithm refers to the objective function to work its value out the amount of NFE increases by one. Eventually, the best optimum point found by every algorithm can be compared to one another in relation to their NFE. In addition, NFE itself can be considered as a stopping criterion as well based on the requirements of the corresponding study. As mentioned before, one of the benefits of using MATLAB and DIgSILENT together is to take the advantages of built-in toolboxes and functions in both. After the optimization procedure small signal analysis (Eigenvalue or Modal Analysis) has been carried out for the grid in PowerFactory without experiencing
10
Reactive Power Optimization in AC Power Systems
395
Fig. 10.27 Reactive power optimization trend for 39-bus New England power system using particle swarm pattern search algorithm
Fig. 10.28 Eigenvalues of 39-bus New England power system without power system stabilizers after optimization
programming difficulties and the result has been shown in Fig. 10.28. The steps how to calculate eigenvalues has been presented in Chap. 11. According to Tables 10.14, 10.15 and 10.16, active power losses have decreased from 28,194,720 to 26,266,740 (W) via adjusting the terminal voltages of
396
A. Jafari Aghbolaghi et al.
Table 10.13 Active power losses data of IEEE 39-bus New England power system using particle swarm pattern search algorithm Bus name
Voltage magnitudes p.u.
Voltage angles deg.
Generators P (MW)
Q (Mvar)
Bus01 Bus02 Bus03 Bus04 Bus05 Bus06 Bus07 Bus08 Bus09 Bus10 Bus11 Bus12 Bus13 Bus14 Bus15 Bus16 Bus17 Bus18 Bus19 Bus20 Bus21 Bus22 Bus23 Bus24 Bus25 Bus26 Bus27 Bus28 Bus29 Bus30 Bus31 Bus32 Bus33 Bus34 Bus35 Bus36 Bus37 Bus38 Bus39
1.0474 1.0487 1.0302 1.0039 1.0053 1.0077 0.9970 0.9960 1.0282 1.0172 1.0127 1.0002 1.0143 1.0117 1.0154 1.0318 1.0336 1.0309 1.0499 0.9912 1.0318 1.0498 1.0448 1.0373 1.0576 1.0521 1.0377 1.0501 1.0499 1.0475 0.9820 0.9831 0.9972 1.0123 1.0493 1.0635 1.0278 1.0265 1.0300
−8.44 −5.75 −8.60 −9.61 −8.61 −7.95 −10.12 −10.62 −10.32 −5.43 −6.28 −6.24 −6.10 −7.66 −7.74 −6.19 −7.30 −8.22 −1.02 −2.01 −3.78 0.67 0.47 −6.07 −4.36 −5.53 −7.50 −2.01 0.74 −3.33 0.00 2.57 4.19 3.17 5.63 8.32 2.42 7.81 −10.05
– – – – – – – – – – – – – – – – – – – – – – – – – – – – – 250.00 520.81 650.00 632.00 508.00 650.00 560.00 540.00 830.00 1000.00
– – – – – – – – – – – – – – – – – – – – – – – – – – – – – 146.16 198.25 205.14 109.91 165.76 212.41 101.18 0.44 22.84 88.28
10
Reactive Power Optimization in AC Power Systems
397
Table 10.14 Power flow results after optimization for 39-bus New England power system using particle swarm pattern search algorithm Active power losses (Initial condition) Active power losses (After optimization) Reduction percentage
28194720 (W) 26266740 (W) 6.84%
Table 10.15 The tap setting of transformers of IEEE 39-bus New England power system after optimization Bus no.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Voltage magnitudes and angles V (kV) h (degree) 364.810659 370.798179 372.902266 374.141517 374.772457 375.577162 372.536129 371.586660 366.270135 377.511546 376.869300 150.734523 377.168446 376.168387 375.663420 376.505858 375.344286 374.016041 379.190499 237.505552 375.538973 378.827835 377.526406 377.641053 375.556902 376.153057 374.094746 371.978374 370.455789 17.130497
−6.452421 −3.981883 −6.776672 −7.781993 −6.949538 −6.385371 −8.253005 −8.667435 −8.296282 −4.188488 −4.942417 −4.933962 −4.770793 −6.099718 −6.121481 −4.680447 −5.652905 −6.467971 0.018998 −0.769076 −2.509637 1.548230 1.366011 −4.568311 −2.870753 −3.965808 −5.811787 −0.616475 2.031981 −1.599642
Load consumption
Injection power
Pl (MW)
Ql (Mvar)
PG (MW)
QG (Mvar)
– – 322.0 500.0 – – 233.8 522.0 – – – 7.5 – – 320.0 329.0 – 158.0 – 628.0 274.0 – 247.5 308.6 224.0 139.0 281.0 206.0 283.5 –
– – 2.4 184.0 – – 84.0 176.0 – – – 88.0 – – 153.0 32.3 – 30.0 – 103.0 115.0 – 84.6 −92.2 47.2 17.0 75.5 27.6 26.9 –
– – – – – – – – – – – – – – – – – – – – – – – – – – – – – 250.0
– – – – – – – – – – – – – – – – – – – – – – – – – – – – – −54.17687004 (continued)
398
A. Jafari Aghbolaghi et al.
Table 10.15 (continued) Bus no.
31 32 33 34 35 36 37 38 39
Voltage magnitudes and angles V (kV) h (degree)
Load consumption Pl (MW)
Ql (Mvar)
PG (MW)
QG (Mvar)
17.926045 17.584661 17.681268 16.557073 17.899414 18.144109 17.536826 17.563459 355.349990
9.2 – – – – – – – 1104.0
4.6 – – – – – – – 250.0
517.55180622 650.0 632.0 508.0 650.0 560.0 540.0 830.0 1000
179.62643204 82.93753618 328.35318619 −164.8434324 128.77506507 46.86633642 19.78359724 −58.73820034 −332.33380789
0.000000 2.436932 4.575524 4.375941 6.135797 8.626973 3.495544 8.560209 −7.996433
Injection power
Table 10.16 The data for transmission lines and transformers of IEEE 6-bus standard power system From bus
To bus
Transformers tap magnitude (p.u.)
Transformers tap tap position
12 12 6 10 19 20 22 23 25 2 29 19
11 13 31 32 33 34 35 36 37 30 38 20
1.0000 1.0000 1.0350 1.0350 1.0700 1.0000 1.0250 1.0000 1.0250 1.0250 1.0000 1.0300
0 0 1 1 2 0 1 – 1 1 0 1
generators and the tap settings of transformers considering the voltage limitations for all the busbars.
10.9
Summary
Investigating active power losses caused in transmission system is of crucial importance in designing power systems as well as their operation and development, of which so many methods have been carried out practically on. Plenty of approaches have been recommended in order to decrease active power losses in
10
Reactive Power Optimization in AC Power Systems
399
power grid in which reactive power optimization is one of the most influential ones. Reactive power optimization investigates the operational condition of reactive power sources in AC power systems in order to have a minimum reactive power current flowing in transmission systems to lessen active power losses associated with it. There are three main parameters on which power system operators have control, as the voltage magnitude of PV busbars, reactive power output of compensators and tap setting of transformers which have under-load tap changer facility. It seems quite obvious that the number of controlling parameters can be so many in a real power network so as there are eight dimensions in a 14-bus power grid. In addition, since there is a tough association between plenty of parameters in electric power systems, reactive power optimization is a very nonlinear and nonconvex problem comprising discrete and continuous variables simultaneously, which has a lot of local optimum points so that traditional optimization algorithms, most of which are based on the gradient of objective function, lose their performance. Heuristic optimization methods have proven their performance in such complex optimization problems and it will be efficient to be able to use them in reactive power optimization problem as well. Two heuristic algorithms in combination with direct search optimization methods have been applied to three standard power grids and the results have been presented. Genetic and particle swarm optimization algorithms are great methods in terms of global optimization, whereas pattern search optimization method is remarkable in the respect of local optimization. Therefore, using a combination of a global optimizer and a local one will have so many benefits in its favor. There are plenty of articles, book chapters, books etc. in this respect so that each ones’ advantages can be taken. This chapter tries to focus on the fundamentals and implementation of reactive power optimization problem using MATLAB and DIgSILENT and creating an effectual link between them. It will avail engineers of using the professional tools of PowerFactory DIgSILENT in the respect of electrical engineering as stability analysis, power flow calculations etc. Besides, MATLAB also has got a lot of toolboxes and flexibility while optimizing problems, using artificial intelligence. The method has been presented for a simple standard power grid step by step using particle swarm pattern search algorithm in Chap. 11, while both MATLAB and DIgSILENT files for both approaches have been presented in book attachment as particle swarm pattern search and genetic pattern search for IEEE 6- and 14-bus power grids. In addition, reactive power optimization using built-in particle swarm and pattern search algorithm on IEEE 39-bus New England power system has been depicted in Chap. 11.
400
A. Jafari Aghbolaghi et al.
Appendices Appendix 1: IEEE 6-Bus Standard Power System The single-line diagram of IEEE 6-bus power network has been presented in Fig. 10.29 and the corresponding data are given in Tables 10.17, 10.18 and 10.19.
Appendix 2: IEEE 14-Bus Standard Power System The single-line diagram of IEEE 14-bus power network has been presented in Fig. 10.30 and the corresponding data are given in Tables 10.20, 10.21 and 10.22.
G
P = 55 MW Q = 13 MVAR
1.05 0
Q = 5 MVAR Bus 6
Bus 4
Bus 1 1:0.91
1:0.9725 Bus 3
Bus 5
Bus 2
Q = 5 MVAR P = 50 MW Q = 5 MVAR
G
P = 30 MW Q = 18 MVAR
P = 50 MW V = 1.1
Fig. 10.29 Single-line diagram of IEEE 6-bus standard power system Table 10.17 The corresponding limitations for control and state variables of IEEE 6-bus standard power system Line no.
Starting busbar
Ending busbar
Line impedances R (X) X (X)
Line admittances G (S) B (S)
Transformer tap settings
1 2 3 4 5 6 7
6 6 4 5 2 3 4
3 4 3 2 1 5 1
4.88187 3.1752 3.84993 11.19258 28.69587 0 0
0.0109 0.0141 0.0140 0.0145 0.0112 0 0
– – – – – 0.9725 0.9100
20.55942 14.6853 16.15383 25.4016 41.6745 11.907 5.27877
−0.0460 −0.0651 −0.0586 −0.0330 −0.0163 −0.0840 −0.1894
10
Reactive Power Optimization in AC Power Systems
401
Table 10.18 Initial statues and power flow results for IEEE 6-bus standard power system
Min Max
Transformer tap settings
Voltage magnitudes of PV busbars
T35, T41
V6
0.910 1.110
63 69.3
Load bus voltage magnitudes
Reactive power output of PV busbars
V2
Output capacity of reactive power sources Q3 Q4
V1, V3, V5
Q2, Q6
69.3 72.45
0.0 5.0
56.7 69.3
−20 100
0.0 5.0
Table 10.19 The data for transmission lines and transformers of IEEE 14-bus standard power system Bus no.
Voltage magnitudes and angles V (kV) h (degree)
Load consumption
Injection power
Pl (MW)
Ql (MVar)
PG (MW)
QG (MVar)
1 2 3 4 5 6
58.12 69.3 58.07 59.11 57.16 66.15
55 0 50 0 0.3 0
13 0 5 0 18 0
0 50 0 0 0 95.79
0 27.63 0 0 0 44.37
−13.24 −4.78 −12.44 −9.85 −12.87 0
Fig. 10.30 Single-line diagram of IEEE 14-bus standard power system
402
Table 10.20 The corresponding limitations to control variables of IEEE 14-bus standard power system Starting busbar
Ending busbar
Line impedances R (X) X (X)
Line admittances G (S) B (S)
Line capacitances C (F)
Transformer tap settings
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
1 1 2 2 2 3 4 4 4 5 6 6 6 7 7 9 9 10 12 13
2 5 3 4 5 4 5 7 9 6 11 12 13 8 9 10 14 11 13 14
2.34498 6.53763 5.68579 7.03131 6.89095 8.10821 1.61535 0 0 0 11.49258 14.87211 8.00415 0 0 3.84901 15.38031 9.92805 26.73132 20.68253
0.041315 0.0085 0.0094 0.0139 0.0141 0.0164 0.0565 0 0 0 0.0162 0.0126 0.0256 0 0 0.0322 0.0118 0.0155 0.0206 0.0094
4.3636e–4 4.066e–4 3.61983e–4 3.0909e–4 2.80991e–4 2.8595e–4 1.05785e–4 0 0 0 0 0 0 0 0 0 0 0 0 0
– – – – – – – 0.978 0.969 0.932 – – – – – – – – – –
7.15957 26.98784 23.95437 21.33472 21.03948 20.69463 5.09531 25.30352 67.29778 30.49442 24.0669 30.95301 15.76267 21.31415 13.31121 10.2245 32.71598 23.24047 24.18548 42.11042
−0.12614 −0.0350 −0.0395 −0.0423 −0.0429 −0.0419 −0.1783 −0.0395 −0.0149 −0.0328 −0.0338 −0.0262 −0.0504 −0.0469 −0.0751 −0.0857 −0.0250 −0.0364 −0.0186 −0.0191
A. Jafari Aghbolaghi et al.
Line no.
10
Reactive Power Optimization in AC Power Systems
Table 10.21 Initial statues and power flow results for IEEE 14-bus standard power system
Output capacity of reactive power sources
T47, T49, T56
Voltage magnitudes of PV busbars V1 V2
Q3
Q6
Q8
0.910 1.110
110 126.5
0.0 40
−6 24
−6 24
Transformer tap settings
Min Max
403
110 121
Table 10.22 Data of lines of IEEE 39-bus New England power system (100 MVA, 60 Hz) [13] Bus no.
Voltage magnitudes and angles V (kV) h (°)
Load consumption
Injection power
Pl (MW)
Ql (MVar)
PG (MW)
QG (MVar)
1 2 3 4 5 6 7 8 9 10 11 12 13 14
110 102.39 91.83 93.67 95.53 87.10 88.14 88.14 85.51 84.68 85.36 84.88 84.19 82.17
0 21.7 94.2 47.8 7.6 11.2 0 0 29.5 9 3.5 6.1 13.5 14.9
0 12.7 19 0 1.6 7.5 0 0 16.6 5.8 1.8 1.6 5.8 5
242.05 40 0 0 0 0 0 0 0 0 0 0 0 0
102.32 50 0 0 0 0 0 0 0 0 0 0 0 0
0 −6.54 −16.11 −12.60 −10.60 −19.66 −17.72 −17.72 −20.66 −21.02 −20.58 −21.21 −21.35 −22.85
Appendix 3: IEEE 39-Bus New England Power System The Single-line diagram of IEEE 39-bus New England power system has been shown in Fig. 10.31 and the corresponding data are given in Tables 10.23, 10.24, 10.25, 10.26, 10.27 and 10.28. The nominal frequency of the New England transmission system is 60 Hz and the main voltage level is 345 kV (nominal voltage). For nodes at a different voltage level, following nominal voltages have been assumed for the PowerFactory model: Bus 12–138 kV, Bus 20–230 kV, Bus 30 and Bus 38–16.5 kV [13].
404
A. Jafari Aghbolaghi et al.
Fig. 10.31 Single-line diagram of IEEE 39-bus New England power system [13]
Table 10.23 Load demands of IEEE 39-bus New England power system [13] Line Line Line Line Line Line Line Line Line Line Line Line Line Line Line Line Line Line Line
01–02 01–39 02–03 02–25 03–04 03–18 04–05 04–14 05–06 05–08 06–07 06–11 07–08 08–09 09–39 10–11 10–13 13–14
From bus
To bus
R in p.u.
X in p.u.
B in p.u.
Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus
Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus
0.0035 0.0010 0.0013 0.0070 0.0013 0.0011 0.0008 0.0008 0.0002 0.0008 0.0006 0.0007 0.0004 0.0023 0.0010 0.0004 0.0004 0.0009
0.0411 0.0250 0.0151 0.0086 0.0213 0.0133 0.0128 0.0129 0.0026 0.0112 0.0092 0.0082 0.0046 0.0363 0.0250 0.0043 0.0043 0.0101
0.6987 0.7500 0.2572 0.1460 0.2214 0.2138 0.1342 0.1382 0.0434 0.1476 0.1130 0.1389 0.0780 0.3804 1.2000 0.0729 0.0729 0.1723 (continued)
01 01 02 02 03 03 04 04 05 05 06 06 07 08 09 10 10 13
02 39 03 25 04 18 05 14 06 08 07 11 08 09 39 11 13 14
10
Reactive Power Optimization in AC Power Systems
405
Table 10.23 (continued) Line Line Line Line Line Line Line Line Line Line Line Line Line Line Line Line Line
14–15 15–16 16–17 16–19 16–21 16–24 17–18 17–27 21–22 22–23 23–24 25–26 26–27 26–28 26–29 28–29
From bus
To bus
R in p.u.
X in p.u.
B in p.u.
Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus
Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus
0.0018 0.0009 0.0007 0.0016 0.0008 0.0003 0.0007 0.0013 0.0008 0.0006 0.0022 0.0032 0.0014 0.0043 0.0057 0.0014
0.0217 0.0094 0.0089 0.0195 0.0135 0.0059 0.0082 0.0173 0.0140 0.0096 0.0350 0.0323 0.0147 0.0474 0.0625 0.0151
0.3660 0.1710 0.1342 0.3040 0.2548 0.0680 0.1319 0.3216 0.2565 0.1846 0.3610 0.5130 0.2396 0.7802 1.0290 0.2490
14 15 16 16 16 16 17 17 21 22 23 25 26 26 26 28
Table 10.24 Generator dispatch of IEEE 39-bus New England power system [13]
15 16 17 19 21 24 18 27 22 23 24 26 27 28 29 29
No.
Load
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Load Load Load Load Load Load Load Load Load Load Load Load Load Load Load Load Load Load Load
Bus 03 04 07 08 12 15 16 18 20 21 23 24 25 26 27 28 29 31 39
Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus
03 04 07 08 12 15 16 18 20 21 23 24 25 26 27 28 29 31 39
P (MW)
Q (Mvar)
322.0 500.0 233.8 522.0 7.5 320.0 329.0 158.0 628.0 274.0 247.5 308.6 224.0 139.0 281.0 206.0 283.5 9.2 1104.0
2.4 184.0 84.0 176.0 88.0 153.0 32.3 30.0 103.0 115.0 84.6 −92.2 47.2 17.0 75.5 27.6 26.9 4.6 250.0
406
A. Jafari Aghbolaghi et al.
Table 10.25 Data of transformers (100 MVA) of IEEE 39-bus New England power system [13]
Generator
Bus
G G G G G G G G G G
Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus
01 02 03 04 05 06 07 08 09 10
39 31 32 33 34 35 36 37 38 30
Bus type
P in MW
V in p.u.
PV Slack PV PV PV PV PV PV PV PV
1000.0 N.A. 650.0 632.0 508.0 650.0 560.0 540.0 830.0 250.0
1.0300 0.9820 0.9831 0.9972 1.0123 1.0493 1.0635 1.0278 1.0265 1.0475
Table 10.26 Data of generators (100 MVA) of IEEE 39-bus New England power system [13] From bus
To bus
R (p.u.)
X (p.u.)
Transformers tap magnitude (p.u.)
Transformers tap angle (degree)
Lower and upper limits of taps
12 12 6 10 19 20 22 23 25 2 29 19
11 13 31 32 33 34 35 36 37 30 38 20
0.0016 0.0016 0.0000 0.0000 0.0007 0.0009 0.0000 0.0005 0.0006 0.0000 0.0008 0.0007
0.0435 0.0435 0.0250 0.0200 0.0142 0.0180 0.0143 0.0272 0.0232 0.0181 0.0156 0.0138
1.0060 1.0060 1.0700 1.0700 1.0700 1.0090 1.0250 1.0000 1.0250 1.0250 1.0250 1.0600
0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
1 1 1 1 1 1 1 – 1 1 1 1
± ± ± ± ± ± ±
(1 (1 (2 (2 (2 (1 (1
0.006) 0.006) 0.035) 0.035) 0.035) 0.009) 0.025)
± ± ± ±
(1 (1 (1 (2
0.025) 0.025) 0.025) 0.03)
10
Unit no.
H (s)
Ra (p.u.)
x’d (p.u.)
x’q (p.u.)
xd (p.u.)
xq (p.u.)
Tʹd0 (s)
Tʹq0 (s)
xl (p.u.)
x″ (p.u.)
T″ d0 (s)
T″ q0 (s)
1 2 3 4 5 6 7 8 9 10
500.0 30.3 35.8 28.6 26.0 34.8 26.4 24.3 34.5 42.0
0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
0.0060 0.0697 0.0531 0.0436 0.1320 0.0500 0.0490 0.0570 0.0570 0.0310
0.0080 0.1700 0.0876 0.1660 0.1660 0.0814 0.1860 0.0911 0.0587 0.0500
0.0200 0.2950 0.2495 0.2620 0.6700 0.2540 0.2950 0.2900 0.2106 0.1000
0.0190 0.2820 0.2370 0.2580 0.6200 0.2410 0.2920 0.2800 0.2050 0.0690
7.000 6.560 5.700 5.690 5.400 7.300 5.660 6.700 4.790 10.200
0.700 1.5000 1.5000 1.5000 0.4400 0.4000 1.5000 0.4100 1.9600 0.0000
0.0030 0.0350 0.0304 0.0295 0.0540 0.0224 0.0322 0.0280 0.0298 0.0125
0.0040 0.0500 0.0450 0.0350 0.0890 0.0400 0.0440 0.0450 0.0450 0.0250
0.050 0.050 0.050 0.050 0.050 0.050 0.050 0.050 0.050 0.050
0.035 0.035 0.035 0.035 0.035 0.035 0.035 0.035 0.035 0.035
Reactive Power Optimization in AC Power Systems
Table 10.27 Data of AVRs of IEEE 39-bus New England power system [13]
407
408
Table 10.28 Initial statues and power flow results for IEEE 39-bus New England power system [13] Unit no.
Ka = KA
Ta = TA
Vrmin = VRmin
Vrmax = VRmax
Ke = KE
Te = TE
Kf = KF
Tf = TF
Se1 = C1
Se2 = C2
E1 = EX1
E2 = EX2
2 3 4 5 6 7 8 9 10
6.2 5.0 5.0 40.0 5.0 40.0 5.0 40.0 5.0
0.05 0.06 0.06 0.02 0.02 0.02 0.02 0.02 0.06
−1.0 −1.0 −1.0 −10.0 −1.0 −6.5 −1.0 −10.5 −1.0
1.0 1.0 1.0 10.0 1.0 6.5 1.0 10.5 1.0
−0.6330 −0.0198 −0.0525 1.0000 −0.0419 1.0000 −0.0470 1.0000 −0.0485
0.405 0.500 0.500 0.785 0.471 0.730 0.528 1.400 0.250
0.0570 0.0800 0.0800 0.0300 0.0754 0.0300 0.0854 0.0300 0.0400
0.500 1.000 1.000 1.000 1.246 1.000 1.260 1.000 1.000
0.660 0.130 0.080 0.070 0.064 0.530 0.072 0.620 0.080
0.880 0.340 0.314 0.910 0.251 0.740 0.282 0.850 0.260
3.036437 2.342286 2.868069 3.926702 3.586801 2.801724 3.191489 4.256757 3.546099
4.048583 3.123048 3.824092 5.235602 4.782401 3.735632 4.255319 5.675676 4.728132 A. Jafari Aghbolaghi et al.
10
Reactive Power Optimization in AC Power Systems
409
References 1. S.R. Singiresu, Engineering Optimization - Theory and Practice, Hoboken, New Jersey: John Wiley & Sons, Inc., 2009. 2. J. Zhu, Reactive Power Optimization, in Optimization of Power System Operation, New Jersey, John Wiley & Sons, pp. 409–454, 2009. 3. T.J. Miller, Reactive Power Coordination, in Reactive Power Control in Electric Systems, New York, John Wiley & Sons, pp. 353–363, 1982. 4. P. Kundur, Power System Stability and Control, New York: McGraw-Hill, 1994. 5. J. Machowski, J.W. Bialek, J.R. Bumby, Power System Dynamics, John Wiley & Sons, 2008. 6. H. Saadat, Power System Analysis, New York: McGraw-Hill, 1999. 7. X. Hugang, C. Haozhong, L. Haiyu, Optimal Reactive Power Flow Incorporating Static Voltage Stability Based on Multi-Objective Adaptive Immune Algorithm, Energy Conversion and Management, vol. 49, pp. 1175–1181, 2008. 8. T. Niknam, M.R. Narimani, R. Azizipanah Abarghooee, B. Bahmani Firouzi, Multi-Objective Optimal Reactive Power Dispatch and Voltage Control: A New Opposition-Based Self-Adaptive Modified Gravitational Search Algorithm, IEEE Systems Journal, vol. 7, pp. 742–753, 2013. 9. S.A.H. Soliman, H.A.A. Mantawy, Modern Optimization Techniques with Applications in Electric Power Systems, New York Heidelberg Dordrecht London: Springer, 2012. 10. A.I.F. Vaz, L. Vicente, A Particle Swarm Pattern Search Method for Bound Constrained Global Optimization, Department of Production and Systems, School of Engineering, University of Minho, Campus of Gualtar, Braga, Portugal, 2006. https://en.wikipedia.org/wiki/Pattern_search_ 11. Wikipedia, the Free Encyclopedia, (optimization). 12. M.A. Abramson, C. Audet, J. Dennis, Generalized Pattern Search Algorithms - Unconstrained and Constrained Cases, IMA Workshop - Optimization in Simulation Based Models. 13. DIgSILENT PowerFactory Version 15 User Manual, Gomaringen, Germany: DIgSILENT GmbH, 2014.
Chapter 11
Reactive Power Optimization Using MATLAB and DIgSILENT Naser Mahdavi Tabatabaei, Ali Jafari Aghbolaghi, Narges Sadat Boushehri and Farid Hojjati Parast
Abstract There are plenty of methods to implement reactive power optimization, such as coding in MATLAB on its own or DIgSILENT, both of which are powerful pieces of software in the electrical engineering respect. However, on one hand MATLAB has a lot of toolboxes, functions, flexibility etc. when it comes to be used in implementation of artificial intelligence and heuristic algorithms. On the other hand, DIgSILENT is such a powerful tool while studying power network stability, power flow calculations, ready-to-use equipment blocks etc., which provides operators with the simplicity of carrying out their studies. Therefore, it would be very efficient to be available of using both pieces of software in parallel in order to take each one’s advantages. In this method, calculations related to power network are executed in DIgSILENT, the ones related to heuristic algorithms are carried out in MATLAB. Although both of them are able to put the whole procedure into practice solely, the point is to take the redeeming features of both to do the process as fast, simple, accurate as possible.
N. Mahdavi Tabatabaei (&) Electrical Engineering Department, Seraj Higher Education Institute, Tabriz, Iran e-mail: [emailprotected] A. Jafari Aghbolaghi F. Hojjati Parast Andishmand Shomal Gharb Engineering Consultancy, Zanjan, Iran e-mail: [emailprotected] F. Hojjati Parast e-mail: [emailprotected] N.S. Boushehri Department of Management, Taba Elm International Institute, Tabriz, Iran e-mail: [emailprotected] © Springer International Publishing AG 2017 N. Mahdavi Tabatabaei et al. (eds.), Reactive Power Control in AC Power Systems, Power Systems, DOI 10.1007/978-3-319-51118-4_11
411
412
11.1
N. Mahdavi Tabatabaei et al.
Introduction
There are plenty of methods to implement reactive power optimization, such as coding in MATLAB on its own or DIgSILENT, both of which are powerful pieces of software in the electrical engineering respect. However, on one hand MATLAB has a lot of toolboxes, functions, flexibility etc. when it comes to be used in implementation of artificial intelligence and heuristic algorithms. On the other hand, DIgSILENT is such a powerful tool while studying power network stability, power flow calculations, ready-to-use equipment blocks etc., which provides operators with the simplicity of carrying out their studies. Therefore, it would be very efficient to be available of using both pieces of software in parallel in order to take each one’s advantages. In this method, calculations related to power network are executed in DIgSILENT, the ones related to heuristic algorithms are carried out in MATLAB. Although both of them are able to put the whole procedure into practice solely, the point is to take the redeeming features of both to do the process as fast, simple, accurate as possible. The reason why it is beneficial to employ MATLAB and DIgSILENT both together to optimize reactive power in AC power systems is to simplify some basic approaches which should be done, although they are not the very case on which need to be concentrated. Therefore, using some ready-to-use tools and techniques will avail researchers of skipping the affairs that do not need to be spent much time on. For instance, in the reactive power optimization problem power flow calculations, stability analysis, modelling equipment etc. are not inevitable to use, whereas they are readily accessible without programming difficulties in DIgSILENT software. Thus, avoiding some routine and basic affairs, researchers could focus their energy on the further studies. This section will present an efficient approach using both MATLAB and DIgSILENT software in order to put reactive power optimization into practice so that both pieces of software will work in parallel until the optimum point has been found.
11.2
How to Implement Reactive Power Optimization Using MATLAB and DIgSILENT
The theoretical aspect of reactive power optimization has been presented elaborately in Chap. 10. In this section, DIgSILENT will do power flow calculations and other processes related to optimization procedure will be carried out by MATLAB. One of the best ways to couple MATLAB and DIgSILENT is to use text files to exporting data, importing data and also to issue orders to the both applications. It should be noted that it is assumed the reader has an intermediate level of knowledge of using DIgSILENT and MATLAB.
11
Reactive Power Optimization Using MATLAB and DIgSILENT
413
Having traced the power network in DIgSILENT and coded the optimization algorithm in MATLAB, the general procedure of reactive power optimization coupling DIgSILENT and MATLAB will be as follows: 1. At the beginning, control variables are initiated by optimization algorithm within feasible bounds and then are written in a text file named “ToDIg.txt”, while DIgSILENT is already waiting to a command from MATLAB to do the power flow calculations. 2. MATLAB sends a command to run the power flow to DIgSILENT via a text file named “Couple.txt”, which contains either “0” or “1” as stop or start order and/or “2” noting end of the procedure (completely stop DIgSILENT). Since the commands will be commuted between DIgSILENT and MATLAB steadily in a text file as “Couple.txt”, if “0” is stop command to DIgSILENT, it is start command to MATLAB and/or vice versa. In this chapter, 1 is used to start DIgSILENT and 0 to stop it. 3. After getting run command, DIgSILENT reads the new control variables created and written by MATLAB into “ToDIg.txt” and applies them to the power network and implements power flow calculations. After finishing power flow, DIgSILENT writes new state values as voltage magnitudes and angles of all busbars into another text file named “LoadFlow.txt”. Then it writes “0” in “Couple.txt” to send an order as to start MATLAB to continue the procedure. 4. MATLAB starts to calculate and investigate the fitness function. If the stopping criteria have been reached, the optimization procedure gets over and the best coordinates found so far will be chosen as the best values for the control variables, if not, the optimization algorithm produces new set of control variables and writes in “ToDIg.txt” to send to DIgSILENT. Then the procedure continues from step 2 again. The simple flowchart of above-presented approach has been shown in Fig. 11.1 to elaborate the trend. It will be presented step by step in the next section as elaborate as possible considering the fact that reader has intermediate level of knowledge using MATLAB and DIgSILENT. Hence, optimizing reactive power in order to reduce active power losses in AC power systems based on the approach presented in this chapter has two different stages. One is preparing a power network in DIgSILENT software and the other is coding the optimization algorithm in MATLAB software, both of which will be explained step by step in the next phases of the chapter.
11.2.1 First Step—Tracing a 6-Bus IEEE Standard Power Network in DIgSILENT The first step toward putting reactive power optimization into practice using MATLAB and DIgSILENT is to have a network ready in DIgSILENT to apply the
414 Fig. 11.1 The flowchart of reactive power optimization using DIgSILENT and MATLAB
N. Mahdavi Tabatabaei et al.
Start
Initialize control variables using intelligent algorithms in MATLAB
Write information in a txt file in a specified location to send to DIgSILENT (“ToDIg.txt”)
Send run signal to DIgSILENT (“Couple.txt”)
Run Power flow using DIgSILENT and write state values to (“LoadFlow.txt”) Apply the operators of optimization algorithm to the control variables
Send run signal to MATLAB (“Couple.txt”)
Calculate fitness function using state values which are obtained from power flow calculations
Is stopping criterion satisfied?
No
Yes Stop
optimization algorithm on. The IEEE 6-bus standard system given in the appendices section of Chap. 10 is used in order to elaborate the approach. The power grid has already traced for you and the corresponding file has been presented in the book attachments, although a brief explanation how to create a project is presented. The base values are considered as follows:
11
Reactive Power Optimization Using MATLAB and DIgSILENT
11.2.1.1
415
Vbase ¼ 63 kV
ð11:1Þ
Sbase ¼ 100 MVA
ð11:2Þ
Creating a New Project in DIgSILENT
To build a new project, the instruction below should be followed: Open DIgSILENT software, go to File Menu ! New ! Project (Fig. 11.2), then enter the project name in the matching field (Fig. 11.3), then leave all other settings as default and click on OK. Start tracing the 6-bus power grid (Fig. 11.4) with the given data in appendices section of Chap. 10 (use the software’s user manual for more information). Figure 11.5 shows the 6-bus power grid after being traced in DIgSILENT, its file has been presented in the book attachments as well and it can be imported to the software in order to avoid drawing it again. After tracing the basic network, there are some extra steps which should be done.
11.2.1.2
How to Enter Operational Limitations of Generators
To enter operational limitations of generators, double click on bus-6 generator (Fig. 11.6) ! go to “Load Flow” tab and tick “Use Limits Specified in Type” in Reactive Power Operational Limits section (Fig. 11.7). Then go to “Basic Data” tab ! “Right Arrow” in front of Type (Fig. 11.8), then enter the values of Nominal
Fig. 11.2 DIgSILENT PowerFactory—creating a new project [1]
416
N. Mahdavi Tabatabaei et al.
Fig. 11.3 DIgSILENT PowerFactory—assigning a name to the project [1]
Fig. 11.4 DIgSILENT PowerFactory—tracing sheet [1]
Apparent Power, Nominal Voltage etc. ! then go to Power Flow tab on the same page (Fig. 11.9) and enter Reactive Power Limits. The same procedure as generators should be done to the synchronous condensers for which the parameter is “qgini” (Fig. 11.10). It should be noted that every filed and variable in DIgSILENT has a separate parameter which will be at work when it comes to be used in DIgSILENT
11
Reactive Power Optimization Using MATLAB and DIgSILENT
417
Fig. 11.5 DIgSILENT PowerFactory—IEEE 6-bus power grid [1]
Fig. 11.6 DIgSILENT PowerFactory—the main page of synchronous generators [1]
Programming Language (DPL) to assign and/or read the values of different parameters. To work out what the parameter for a specific field is, just hold mouse pointer on the field then the parameter will be shown, for example, the voltage magnitude parameter for generators is “usetp” (Fig. 11.11).
418
N. Mahdavi Tabatabaei et al.
Fig. 11.7 DIgSILENT PowerFactory—the main page of synchronous generators, “Load Flow” tab [1]
Fig. 11.8 DIgSILENT PowerFactory—adjusting nominal apparent power of synchronous generators [1]
11.2.1.3
How to Adjust Tap Settings for Transformers
In order to set initial tap settings for the transformers to be in an appropriate form for controlling by DPL, double click on a transformer ! click on “Right Arrow” in front of “Type” (Fig. 11.12) ! enter the values for “Rated Power”, “Rated Voltage” etc. (Fig. 11.13) ! then go to VDE/IEC “Short-Circuit” tab (Fig. 11.14)
11
Reactive Power Optimization Using MATLAB and DIgSILENT
419
Fig. 11.9 DIgSILENT PowerFactory—adjusting reactive power limitations of synchronous generators [1]
Fig. 11.10 DIgSILENT PowerFactory—adjusting voltage magnitudes of synchronous generators [1]
and enter the values for “Additional Voltage Per Tap” = 0.0001, “Neutral Position” = 10,000, “Minimum Position” = 9100 and “Maximum Position” = 11,100 according to 6-bus IEEE power grid then click OK ! go to “Power Flow” tab (Fig. 11.15), you will see the tap changer data that you have already assigned. The parameter for tap setting is “nntap” which is shown when holding mouse pointer on “Tap Position” field.
420
N. Mahdavi Tabatabaei et al.
Fig. 11.11 DIgSILENT PowerFactory—the main page for transformers [1]
Fig. 11.12 DIgSILENT PowerFactory—the main page for transformers, “Basic Data” tab [1]
11.2.2 Second Step—How to Use DIgSILENT Scripting Facility There are some prerequisite steps while using DIgSILENT scripting facility, such as creating a DPL command set, creating a general set and introducing external variables to the DPL command set, each of which are representatives of parameters that are needed in order to control or extract values of parameters from equipment
11
Reactive Power Optimization Using MATLAB and DIgSILENT
421
Fig. 11.13 DIgSILENT PowerFactory—adjusting tap settings of transformers [1]
Fig. 11.14 DIgSILENT PowerFactory—the main page for transformers, “Load Flow” tab [1]
during the optimization procedure. All the steps have been represented in the following sections.
11.2.2.1
How to Create a DPL Command Set
In order to have an efficient communication between MATLAB and DIgSILENT scripting is inevitable. To write a script in DIgSILENT click on “Open Data
422
N. Mahdavi Tabatabaei et al.
Fig. 11.15 DIgSILENT PowerFactory—setting up operational limitations for synchronous condensers [1]
Fig. 11.16 DIgSILENT PowerFactory—“data manager” [1]
Manager” on the main toolbar (Fig. 11.16) ! click on “New Object” on the “Data Manager” window and choose “DPL Command and more” (Fig. 11.17), then click on OK ! write the script in the “Scrip” tab of “DPL Command” window (Fig. 11.18). The corresponding script to reactive power optimization for IEEE 6-bus power network has been given in the attachments.
11
Reactive Power Optimization Using MATLAB and DIgSILENT
423
Fig. 11.17 DIgSILENT PowerFactory—creating DPL command [1]
Fig. 11.18 DIgSILENT PowerFactory—“DPL command” window [1]
11.2.2.2
How to Create a “General Set” and Introduce It to DPL
There are a few other steps toward being able to use DPL dynamically in order to assign and export data, for instance, defining control and state variables to the written script in DPL. First, select all the equipment and right click on ! “Define” ! “General Set” (Fig. 11.19). To introduce the control variables of the network to the corresponding DPL, click on “Open Data Manager” on the main toolbar (Fig. 11.16) ! right click on the script you have already created and
424
N. Mahdavi Tabatabaei et al.
Fig. 11.19 DIgSILENT PowerFactory—creating general set [1]
Fig. 11.20 DIgSILENT PowerFactory—introducing “General Set” to DPL command [1]
then click on “Edit” (Fig. 11.20) ! in the “Basic Options” tab, select the general set to your script (Figs. 11.21 and 11.22). Please note that the settings related to the “General Set” section may be necessary to be done again after exporting the network in order to use it elsewhere, otherwise the DPL would not be able to access the equipment.
11
Reactive Power Optimization Using MATLAB and DIgSILENT
425
Fig. 11.21 DIgSILENT PowerFactory—introducing “General Set” to DPL command [1]
Fig. 11.22 DIgSILENT PowerFactory—introducing “General Set” to DPL command [1]
11.2.2.3
How to Introduce External Variables to DPL
In the “DPL Command” window and in the “External Objects” section the state variables should be defined. To do this, click on “Name” filed and assign a variable name like “G2” as the voltage magnitude variable of the generator of bus-2 ! double click in the “Object” filed and select Generator-2 from the list (Fig. 11.23). In order to be available of using power flow calculations in the script, it should be defined just like control variables (Fig. 11.24). The final phase is to write the script in the scripting window. The corresponding script has written and explained step by step in the next stages.
426
N. Mahdavi Tabatabaei et al.
Fig. 11.23 DIgSILENT PowerFactory—defining external variables to the DPL [1]
Fig. 11.24 DIgSILENT PowerFactory—introducing “Load Flow Calculation” function to DPL [1]
11.2.2.4
How to Write Scripts for DIgSILENT and MATLAB Step by Step
This section will present an expressive explanation of the written script for DIgSILENT and MATLAB step by step. It would be better to describe the written script for reactive power optimization using presented method dynamically switching from one piece of software to the other, so the procedure of the written
11
Reactive Power Optimization Using MATLAB and DIgSILENT
427
program in both pieces of software will be easy to comprehend. The procedure has been explained in following just as it will run and switch between two pieces of software practically.
Initializing MATLAB Software Since the suggested procedure of optimizing reactive power using DIgSILENT and MATLAB has programmed the way to consider MATLAB as the main calculator and DIgSILENT as the implicit one, the procedure turns out to be started by MATLAB [2].
Initializing
MATLAB
clear clc close all; hold on; CostFunction=@(x) OF_GA_PSO_PS(x);
% Defining Cost Function for PSO Algorithm
tic %% Global Parameters nVar=6; VarMin=[1.1 1 9100 9100 0 0]; VarMax=[1.15 1.1 11100 11100 5 5];
% Lower Bounds of Control Variables % Upper Bounds of Control Variables
%Creating 'ToDIg.txt' file to export the control variables values to be imported in DIgSILENT t=fopen('ToDIg.txt','w'); fprintf(t,'%d %d %d %d %d %d',1.1,1.05,9725,9100,0,0); fclose(t); % Making a blank "Couple.txt" file to order the switching process between MATLAB and DIgSILENT t=fopen('Couple.txt','w'); fprintf(t,'%d',0); fclose(t); % Waiting for DIgSILENT to Carry Out Power Flow Calculation a=0; while a==0 pause(0.01); a=load('Couple.txt'); end
428
N. Mahdavi Tabatabaei et al.
In the codes above, ‘clear’ was used to clear all possible existent variables in workplace, ‘clc’ to clear the workplace itself, ‘close all’ to close all possible open figures, and ‘hold on’ to retain the current graph and to add another graph to it. MATLAB adjusts the axes ranges to display full range of the added graphs. ‘CostFunction = @(x) OF_GA_PSO_PS(x);’ is used to define ‘OF_GA_PSO_PS (x)’ as the ‘CostFunction’ which is used within the PSO codes that will be presented in the next steps. ‘tic’ in the beginning and ‘toc’ at the end of the entire code are used to measure calculation time. In addition, ‘VarMin’ and ‘VarMax’ consist of lower and upper bounds of the control variables which were presented in appendices section of Chap. 10, and ‘nVar’ is the number of control variables, respectively [2]. In order to enter the values for control variables into DIgSILENT a ‘.txt’ file as ‘ToDIg.txt’ is created using the codes presented above. In the same way, it is needed to have a ‘.txt’ file as ‘Couple.txt’ to command both pieces of software one to proceed and the other to stop. It is noted that, in ‘Couple.txt’, 0 is written to mandate DIgSILENT to carry on and 1 to MATLAB, whenever one works the other stops. Thus, in general when ‘Couple.txt’ in the scripting section emerges, it means that process is being switched from one piece of software to the other. Therefore, at the end of codes above ‘Couple.txt’ switches the rest of the process to be continued in DIgSILENT.
Initializing DIgSILENT Software Like any other programming language, there are some primary steps to take in DPL, such as defining and initiating the variables that are used in the script [1].
Initializing int a; object o; double V,U,Phi,VL; double VG2,VG6,TT35,TT41,QC3,QC4; set s1,s2; string str1; a=0;
DIgSILENT
11
Reactive Power Optimization Using MATLAB and DIgSILENT
429
where, ‘double’ and ‘int’ are used to define real and integer variables respectively, ‘object’ is a reference to apparatus in DIgSILENT, ‘set’ is also used to address an array of components, and ‘string’ is used to define string class variables. DPL uses the following internal parameter types [1]: – – – – –
double, a 15 digits real number int, an integer number string, a string object, a reference to a PowerFactory objects set, a container of objects
Vectors and Matrices are available as external objects. The syntax for defining variables is as follows: – [VARDEF] = [TYPE] varname, varname, …, varname; – [TYPE] = double | int | object | set All the parameters should be defined in the first lines of the DPL script. The semicolon is obligatory. Examples [1]: – – – – –
double Losses, Length, Pgen; int NrOfBreakers, i, j; string txt1, nm1, nm2; object O1, O2, BestSwitchToOpen; set AllSwitches, AllBars;
Running Power Flow in DIgSILENT Based on Initial Condition Equations (10.59) and (10.61) will be used to calculate active power losses and voltage stability index in 6-bus power network, of which contain state variables that should be taken from power flow calculations. Thus, in this step DIgSILENT will run power flow on the given power grid [1].
430
N. Mahdavi Tabatabaei et al.
Running Power Flow in DIgSILENT Based on Initial Condition
DIgSILENT
! IEEE Standard Circumstances fopen('d:\Project\ToDIg.txt','r',0); fscanf(0,'%d %d %d %d %d %d',VG2,VG6,TT35,TT41,QC3,QC4); fclose(0); G2:usetp=VG2; G6:usetp=VG6; T35:nntap=TT35; T41:nntap=TT41; Q3:qgini=QC3; Q4:qgini=QC4; ! Load Flow Calculation ClearOutput(); ResetCalculation(); Ldf.Execute(); ! Exporting Load Flow Calculation’s Result s1=SEL.AllBars(); s1.SortToName(0); fopen('d:\Project\LoadFlow.txt','w',0); o=s1.First(); while (o) { V=o:m:Ul; U=o:m:u1; Phi=o:m:phiu; str1=o:loc_name; fprintf(0,'%f\t %f\t %f\t',V,U,Phi); o=s1.Next(); } fclose(0);
In the codes above, “! IEEE Standard Circumstances” section is used to import and apply control variables written by MATLAB in “ToDIg.txt” to the power grid in DIgSILENT. “! Load Flow Calculation” sector is used to carry load flow out on the power grid. Finally, “! Exporting Load Flow Calculation’s Result” step is used to write the corresponding state variables in a file as “LoadFlow.txt” to be imported in MATLAB for further calculations. In order to find a parameter name for each kind of equipment follow the instruction below: Do right click on the “Result Box” of the equipment ! “Edit Format for Nodes” (Fig. 11.25) ! Right click on “Variable” column ! “Insert Row(s)” (Fig. 11.26) ! Double click on the newly created row ! a new window entitled
11
Reactive Power Optimization Using MATLAB and DIgSILENT
431
Fig. 11.25 DIgSILENT PowerFactory—“Edit Format for Nodes” [1]
Fig. 11.26 DIgSILENT PowerFactory—“Edit Format for Nodes” “Insert Row(s)” [1]
“Variable Selections” will be opened (Fig. 11.27) ! Select “Load Flow” tab in the left side of the window ! In the “Available Variables” subsection choose the set of variables which is needed ! In the “Selected Variables” subsection see the parameters which can be directly used in the DPL.
432
N. Mahdavi Tabatabaei et al.
Fig. 11.27 DIgSILENT PowerFactory—“Variable Selection” window [1]
Voltage Stability Index Process Before Reactive Power Optimization According to the method explained in Sect. 10.4.2.3, the voltage magnitudes for PQ buses should be obtained after and before reactive power optimization under no-load and loaded circumstances. Another text file as “LoadBusVoltagesFLBO. txt” has been considered to embody voltage magnitudes of PQ buses under loaded condition before optimization, and “LoadBusVoltagesNLBO.txt” for under no-load condition before optimization which will be used in MATLAB for further processes of voltage stability index [1].
11
Reactive Power Optimization Using MATLAB and DIgSILENT
The Needed Calculations for Voltage Stability Index !! Start of Voltage Stability Index Process Before Optimization fopen('d:\Project\LoadBusVoltagesFLBO.txt','w',0); o=s1.First(); while (o) { VL=o:m:Ul; str1=o:loc_name; fprintf(0,'%f\t',VL); o=s1.Next(); } fclose(0); ! Making All Loads OutofService s2=SEL.AllLoads(); o=s2.First(); while(o){ o:outserv=1; o=s2.Next(); } ! Load Flow calculation without Load ClearOutput(); ResetCalculation(); s1=SEL.AllBars(); s1.SortToName(0); Ldf.Execute(); fopen('d:\Project\LoadBusVoltagesNLBO.txt','w',0); o=s1.First(); while (o) { VL=o:m:Ul; str1=o:loc_name; fprintf(0,'%f\t',VL); o=s1.Next(); } fclose(0); ! Making All Loads Back to Service s2=SEL.AllLoads(); o=s2.First(); while(o){ o:outserv=0; o=s2.Next(); } !! End of Voltage Stability Index Process Before Optimization
433
DIgSILENT
434
N. Mahdavi Tabatabaei et al.
! Send Run Message to MATLAB fopen('d:\Project\Couple.txt','w',1); fprinƞ(1,'%d',1); fclose(1); ! WaiƟng to Get Run Signal from MATLAB a=1; while(a=1){ fopen('d:\Project\Couple.txt','r',0); fscanf(0,'%d',a); fclose(0); }
There is a parameter to make a component of the network out of or back to service by the name “outserve”. As it has shown in the “! Making All Loads Out of Service” section, at first all the loads should be selected as a set, then allocating “outserv = 1” for them one by one to make them out of service. After having all the data gained which is needed to calculate voltage stability index before optimization for further comparisons with the data after optimization procedure, DIgSILENT will send a run signal to MATLAB using “Couple.txt” to continue the procedure.
Active Power Losses Calculation Based on Initial Circumstances In order to calculate active power losses based on initial conditions to figure out how much active power losses has decreased in the reactive power optimization procedure the following code will be proceeded in MATLAB [2].
11
Reactive Power Optimization Using MATLAB and DIgSILENT
Initial Active Power Losses Calculation
435
MATLAB
% Waiting for DIgSILENT to Carry Out Power Flow Calculation a=0; while a==0 pause(0.01); a=load('Couple.txt'); end % Loading Power Flow Results From DIgSILENT BUSData=load('LoadFlow.txt'); BUSData(:,1)=BUSData(:,1)*1000; V G=[0.0109 0.0141 0.0140 0.0145 0.0112 0 0];
%Transferring Voltage from KV to %The line's Conductance
PL1=abs(G(1)*(BUSData(6,1)^2+BUSData(3,1)^22*BUSData(6,1)*BUSData(3,1)*cos(deg2rad(BUSData(6,3)-BUSData(3,3))))); PL2=abs(G(2)*(BUSData(6,1)^2+BUSData(4,1)^22*BUSData(6,1)*BUSData(4,1)*cos(deg2rad(BUSData(6,3)-BUSData(4,3))))); PL3=abs(G(3)*(BUSData(4,1)^2+BUSData(3,1)^22*BUSData(4,1)*BUSData(3,1)*cos(deg2rad(BUSData(4,3)-BUSData(3,3))))); PL4=abs(G(4)*(BUSData(5,1)^2+BUSData(2,1)^22*BUSData(5,1)*BUSData(2,1)*cos(deg2rad(BUSData(5,3)-BUSData(2,3))))); PL5=abs(G(5)*(BUSData(2,1)^2+BUSData(1,1)^22*BUSData(2,1)*BUSData(1,1)*cos(deg2rad(BUSData(2,3)-BUSData(1,3))))); OF_IEEE=PL1+PL2+PL3+PL4+PL5; GlobalBest.Cost=OF_IEEE;
Scripting Particle Swarm Optimization Scripting PSO algorithm has done based on the PSO theory which was presented in Sect. 10.5.1. Some further explanations have been given in the script to enlighten what is going on. The “Particle Swarm Optimization Algorithm” is as follows [2].
436
N. Mahdavi Tabatabaei et al.
Particle Swarm Optimization Algorithm %% Using PSO as Optimization Algorithm VarSize=[1 nVar]; % Size of Decision Variables Matrix %PSO Parameters MaxIt=30; nPop=20;
% Maximum Number of Iterations % Population Size (Swarm Size)
% Constriction Coefficients phi1=2.05; phi2=2.05; phi=phi1+phi2; chi=2/(phi-2+sqrt(phi^2-4*phi)); w=chi; % Inertia Weight wdamp=1; % Inertia Weight Damping Ratio c1=chi*phi1; % Personal Learning Coefficient c2=chi*phi2; % Global Learning Coefficient % Velocity Limits VelMax=0.1*(VarMax-VarMin); VelMin=-VelMax; % Initialization empty_particle.Position=[]; empty_particle.Cost=[]; empty_particle.Velocity=[]; empty_particle.Best.Position=[]; empty_particle.Best.Cost=[]; particle=repmat(empty_particle,nPop,1); for i=1:nPop % Initialize Position particle(i).Position=unifrnd(VarMin,VarMax,VarSize); % Initialize Velocity particle(i).Velocity=zeros(VarSize); % Evaluation [particle(i).Cost particle(i).Sol]=CostFunction(particle(i).Position); % Update Personal Best particle(i).Best.Position=particle(i).Position; particle(i).Best.Cost=particle(i).Cost;
MATLAB
11
Reactive Power Optimization Using MATLAB and DIgSILENT
particle(i).Best.Sol=particle(i).Sol; % Update Global Best if particle(i).Best.Cost0.05)*(abs(BUSData(5,2)-Vref))); %VD=VD1+VD3+VD5; %F=PL+VD; F=PL; end
The PSO algorithm will be proceeded until the given stopping criteria have been exceeded. After having the optimization procedure by PSO finished, it is time to use Patter Search algorithm to pinpoint the exact optimum location [1, 2].
Using Pattern Search Algorithm as Supplementary Optimization Algorithm As immediately as PSO algorithm finishes its procedure, the Pattern Search algorithm starts its local optimizing. The best result discovered by PSO is considered as the starting point for pattern search algorithm to pinpoint the optimum point. There is a very complete toolbox for pattern search algorithm as a built-in function in MATLAB, of which is going to be used in series with PSO algorithm. In order to use pattern search combined with PSO algorithm as a collective one, it is needed to have the codes of pattern search. Fortunately, MATLAB has given this availability
11
Reactive Power Optimization Using MATLAB and DIgSILENT
443
Fig. 11.28 MATLAB—Pattern Search toolbox [2]
to its users to extract the codes form the toolbox. It is readily accessible clicking on “File” menu then “Generate Code” after adjusting all the corresponding parameters in the toolbox window (Fig. 11.28). Referring to help documentation of MATLAB and following the written codes will reveal what has been done. After generating the script of pattern search via the related toolbox, a general structure of the algorithm will be given which has been formed to optimize reactive power as follows [2].
Pattern Search Algorithm - Main Code
MATLAB
%% Using Pattern Search Algorithm as Second Optimization Algorithm x0=Var_PSO; % Using Best Control Variable Obtained from PSO to Start Point of Pattern Search Algorithm MaxIterPS=60; % Maximum Number of Iteration for Pattern Search Algorithm [Var_PS,OF_PS,exitflag,output] = PatternSearchTool(x0,VarMin,VarMax,MaxIterPS); LossReductionPercent=((OF_IEEE-OF_PS)/OF_IEEE)*100;
In order to adjust corresponding parameters for Pattern Search algorithm in the scripts, another function as “psoptimset” is needed. The related function has been implemented as follows for reactive power optimization. MATLAB software itself has an elaborate guide on different adjustments of the Pattern Search algorithm which is the best resource to pursue for more information [2].
444
N. Mahdavi Tabatabaei et al.
Pattern Search Algorithm - Adjusting psoptimset
MATLAB
%% Pattern Search Optimization Tool function [x,fval,exitflag,output] = PatternSearchTool(x0,VarMin,VarMax,MaxIterPS) options = psoptimset; % Modify Options Setting options = psoptimset(options,'MaxIter', MaxIterPS); options = psoptimset(options,'MeshAccelerator', 'on'); options = psoptimset(options,'CompletePoll', 'on'); options = psoptimset(options,'SearchMethod', @GPSPositiveBasis2N); options = psoptimset(options,'CompleteSearch', 'on'); options = psoptimset(options,'Display', 'iter'); options = psoptimset(options,'OutputFcns', { @PatternSearchOutputFcn }); [x,fval,exitflag,output] = patternsearch(@OF_GA_PSO_PS,x0,[],[],[],[],VarMin,VarMax,[],options);
Finishing the Optimization Procedure After having the optimization procedure finished and the optimum point found by pattern search algorithm, MATLAB sends a stop signal to the main power flow calculation loop of DIgSILENT. In addition, “toc” is used to stop measuring of how much time has been spent during optimization procedure. The codes are as follows [2].
Finishing the Optimization Process
MATLAB
%% Finishing Process % Send a Signal to Stop DIgSILENT t=fopen('Couple.txt','w'); fprintf(t,'%d',2); fclose(t); hold off; toc % Wait to DIgSILENT Runs Power Flow Calculation a=2; while a==0 pause(0.01); a=load('Couple.txt'); end
Voltage Stability Index Process After Reactive Power Optimization After having the optimization procedure finished, DIgSILENT runs a power flow considering the best optimum point found under loaded and no-load circumstances,
11
Reactive Power Optimization Using MATLAB and DIgSILENT
445
then writes the results in “LoadBusVoltagesFLAO.txt” and “LoadBusVoltagesNLAO.txt”, respectively. Afterwards, all the loads should be set back to service in order to have the power network in normal condition. DIgSILENT has finished its duty in this section and the procedure is switched to MATLAB to further calculations [1]. The Needed Calculations for Voltage Stability Index !!! Start Voltage Stability Index Process After Optimization ClearOutput(); ResetCalculation(); s1=SEL.AllBars(); s1.SortToName(0); Ldf.Execute(); fopen('d:\Project\LoadBusVoltagesFLAO.txt','w',0); o=s1.First(); while (o) { VL=o:m:Ul; str1=o:loc_name; fprintf(0,'%f\t',VL); o=s1.Next(); } fclose(0); ! Making All Loads Out of Service s2=SEL.AllLoads(); o=s2.First(); while(o){ o:outserv=1; o=s2.Next(); } ! Load Flow calculation without Load ClearOutput(); ResetCalculation(); s1=SEL.AllBars(); s1.SortToName(0); Ldf.Execute(); fopen('d:\Project\LoadBusVoltagesNLAO.txt','w',0); o=s1.First(); while (o) { VL=o:m:Ul; str1=o:loc_name; fprintf(0,'%f\t',VL); o=s1.Next(); } fclose(0); ! Make All Loads Back to Service s2=SEL.AllLoads(); o=s2.First();
DIgSILENT
446
N. Mahdavi Tabatabaei et al.
while(o){ o:outserv=0; o=s2.Next(); } !! End Voltage Stability Index Process After Optimization ! Send Run Message to MATLAB fopen('d:\Project\Couple.txt','w',1); fprintf(1,'%d',1); fclose(1);
Final Calculations of Voltage Stability Index After having all the data obtained via power flow calculations in DIgSILENT that is initially needed thorough processing the voltage stability index before and after optimization procedure, MATLAB gets the data and carries further calculations on them to have kind of tangible information of how much the index has been improved. The related codes are as follows [2]. Voltage Stability Index Final Calculations
MATLAB
%% Voltage Stability Index Calculation VLNLBF=1000*load('LoadBusVoltagesNLBO.txt'); % Voltage-Load Bus-No load-BeforeOptimization VLFLBF=1000*load('LoadBusVoltagesFLBO.txt'); % Voltage-Load Bus-Full load-BeforeOptimization VLNLAF=1000*load('LoadBusVoltagesNLAO.txt'); % Voltage-Load Bus-No load-BeforeOptimization VLFLAF=1000*load('LoadBusVoltagesFLAO.txt'); % Voltage-Load Bus-Full load-BeforeOptimization VSI.BO(1)=VLFLBF(1)/VLNLBF(1); VSI.BO(2)=VLFLBF(3)/VLNLBF(3); VSI.BO(3)=VLFLBF(5)/VLNLBF(5); VSI.AO(1)=VLFLAF(1)/VLNLAF(1); VSI.AO(2)=VLFLAF(3)/VLNLAF(3); VSI.AO(3)=VLFLAF(5)/VLNLAF(5); hold on; figure; BarSort=[VSI.BO(1),VSI.AO(1);0,0;VSI.BO(2),VSI.AO(2);0,0;VSI.BO(3),VSI.AO(3)]; bar(BarSort); xlabel('Load Bus No.'); ylabel('Voltage Stability Index (V/V0)'); legend('Befor Optimization','After Optimization'); hold off;
11
Reactive Power Optimization Using MATLAB and DIgSILENT
447
Voltage Deviation Calculations Before and After Optimization Another index as voltage deviation before and after reactive power optimization has been considered in order to work out the impact of the optimization on the voltage deviation of PQ busbars. The related data have been put in the bar figures to have an easy comparison between them. The codes are as following [2].
Voltage Deviation Calculations
MATLAB
%% Deviation of LoadBus's Voltages Before and After Optimization Vref=63000; VD.BO(1)=abs(VLFLBF(1)-Vref); VD.BO(2)=abs(VLFLBF(3)-Vref); VD.BO(3)=abs(VLFLBF(5)-Vref); VD.AO(1)=abs(VLFLAF(1)-Vref); VD.AO(2)=abs(VLFLAF(3)-Vref); VD.AO(3)=abs(VLFLAF(5)-Vref); figure; BarSort=[VD.BO(1),VD.AO(1);0,0;VD.BO(2),VD.AO(2);0,0;VD.BO(3),VD.AO(3)]; bar(BarSort); xlabel('Load Bus No.'); ylabel('Voltage Deviation, Unit (V)'); legend('Befor Optimization','After Optimization'); hold off;
Reduction Figures of Active Power Losses A bar type figure has been considered to indicated how much active power losses have reduced due to reactive power optimization using particle swarm algorithm as the main and global optimizer with the association of Pattern Search as the supplementary one, which is rather a local optimizer [2].
448
N. Mahdavi Tabatabaei et al.
Active Power Losses Reduction
MATLAB
%% Plot Bars of Loss Reduction figure; BarSort=[OF_IEEE,OF_PS]; bar(BarSort,0.1); xlabel('1-IEEE-Standard 2-PSO+PS Algorithm'); ylabel('Ploss, Unit (W)'); hold off;
Representing All Data in MATLAB Command Window In addition to the figures which are planned to indicate the data that provide us with having an easy comparison of the power network state before and after optimization, the additional and accurate data will be presented in the command window of MATLAB which are produced using the codes below [2].
11
Reactive Power Optimization Using MATLAB and DIgSILENT
Specifying All the Data Related to Reactive Power Optimization
449
MATLAB
%% Indicating Results disp('======================================================='); str = sprintf('Total Real Power Losses (W) on IEEE-Standard Circumstance = %d .',OF_IEEE); disp(str); disp('======================================================='); str = sprintf('Total Real Power Losses (W) After Optimization Using PSO+PS = %d .',OF_PS); disp(str); disp('======================================================='); str = sprintf('The Percentage of Power Loss Reduction = %d .',LossReductionPercent); disp(str); disp('======================================================='); disp('The Voltage Stability Index for Load Buses Before Optimization:'); disp('----------'); str = sprintf('VSI_Bus_1 = %d .', VSI.BO(1)); disp(str); str = sprintf('VSI_Bus_3 = %d .', VSI.BO(2)); disp(str); str = sprintf('VSI_Bus_5 = %d .', VSI.BO(3)); disp(str); disp('----------'); disp('The Voltage Stability Index for Load Buses After Optimization:'); disp('----------'); str = sprintf('VSI_Bus_1 = %d .', VSI.AO(1)); disp(str); str = sprintf('VSI_Bus_3 = %d .', VSI.AO(2)); disp(str); str = sprintf('VSI_Bus_5 = %d .', VSI.AO(3)); disp(str); disp('----------'); disp('======================================================='); disp('The Sum of Voltage Stability Index for all Load Buses Before and After Optimization are as follow:');
450
N. Mahdavi Tabatabaei et al.
disp('Considering the Ideal Value is 3'); disp('----------'); VSI_Total_BO=VSI.BO(1)+VSI.BO(2)+VSI.BO(3); VSI_Total_AO=VSI.AO(1)+VSI.AO(2)+VSI.AO(3); str = sprintf('Befor Optimization = %d .', VSI_Total_BO); disp(str); str = sprintf('After Optimization = %d .', VSI_Total_AO); disp(str); disp('======================================================='); disp('The Voltage Deviation for Load Buses Before Optimization:'); disp('----------'); str = sprintf('VD_Bus_1 = %d .', VD.BO(1)); disp(str); str = sprintf('VD_Bus_3 = %d .', VD.BO(2)); disp(str); str = sprintf('VD_Bus_5 = %d .', VD.BO(3)); disp(str); disp('----------'); disp('The Voltage Deviation for Load Buses After Optimization:'); disp('----------'); str = sprintf('VD_Bus_1 = %d .', VD.AO(1)); disp(str); str = sprintf('VD_Bus_3 = %d .', VD.AO(2)); disp(str); str = sprintf('VD_Bus_5 = %d .', VD.AO(3)); disp(str); disp('----------'); disp('======================================================='); disp('The Sum of Voltage Deviations for all Load Buses Before and After Optimization are as follow:'); disp('----------'); VD_Total_BO=VD.BO(1)+VD.BO(2)+VD.BO(3); VD_Total_AO=VD.AO(1)+VD.AO(2)+VD.AO(3); str = sprintf('Before Optimization = %d .', VD_Total_BO); disp(str); str = sprintf('After Optimization = %d .', VD_Total_AO); disp(str);
11
Reactive Power Optimization Using MATLAB and DIgSILENT
451
disp('======================================================='); disp('The Control Variables After Optimization are as follow:'); disp('----------'); VG_Bus2 = Var_PS(1,1); VG_Bus6 = Var_PS(1,2); TT_Bus35=Var_PS(1,3); TT_Bus14=Var_PS(1,4); QC_Bus3=Var_PS(1,5); QC_Bus4=Var_PS(1,6); str = sprintf('Generator Voltage in Bus_2 Before and After Optimization are %d and %d, respectively.',1.1,VG_Bus2); disp(str); str = sprintf('Generator Voltage in Bus_6 Before and After Optimization are %d and %d, respectively.',1.05,VG_Bus6); disp(str); str = sprintf('Tap Position of transformer 3 to 5 Before and After Optimization are %d and %d, respectively.',9725,TT_Bus35); disp(str); str = sprintf('Tap Position of Transformer 1 to 4 Before and After Optimization are %d and %d, respectively.',9100,TT_Bus14); disp(str); str = sprintf('Reactive Power Output of Synchronize Condenser on Bus_3 Before and After Optimization are %d and %d, respectively.',0,QC_Bus3); disp(str); str = sprintf('Reactive Power Output of Synchronize Condenser on Bus_4 Before and After Optimization are %d and %d, respectively.',0,QC_Bus4); disp(str); disp('=======================================================');
11.2.3 Third Step—How to Run MATLAB and DIgSILENT to Start Optimizing Reactive Power After having the mfiles of MATLAB and scripts of DIgSILENT ready, it is the final step to run the optimization process. It should be noted that, according to the structure used in the script files enclosed with this chapter, the mfile of MATLAB is the first one which should be run and then DIgSILENT. To run the optimization procedure via MATLAB by “RPObyPSOplusPS.m” simply click on “Run” icon in
452
N. Mahdavi Tabatabaei et al.
Fig. 11.29 DIgSILENT and MATLAB—running the optimization procedure [1], [2]
Fig. 11.30 MATLAB—the figures of results after reactive power optimization [2]
the mfile toolbar, and to run the script written in DIgSILENT right click on the script and click on “Execute” (Fig. 11.29) [1, 2]. Figures 11.30, 11.31 and 11.32 show the result of optimization after the process is done. The elaborate report of reactive power optimization for the same power grid has been presented in the Simulation Results (Sect. 10.8) in Chap. 10.
11
Reactive Power Optimization Using MATLAB and DIgSILENT
453
Fig. 11.31 MATLAB—the results which are represented in command window [2]
Fig. 11.32 DIgSILENT PowerFactory—the power network after reactive power optimization [1]
11.3
How to Use Built-in Optimization Functions to Implement Reactive Power Optimization
This section will present the simplest and more efficient way of carrying out reactive power optimization on 39-bus New England power network using built-in functions of MATLAB software and the corresponding ready-to-use network of DIgSILENT. The optimization algorithm will be particle swarm pattern search algorithm as its efficiency has been investigated before. Fortunately, both the
454
N. Mahdavi Tabatabaei et al.
optimization algorithms are available in MATLAB without suffering programming difficulties so much. In addition, the power network of IEEE 39-bus New England is available in DIgSILENT 15.1 which can be used as a reliable case study in terms of the correctness of the data.
11.3.1 How to Prepare the Network in DIgSILENT To do this there will be different approaches. IEEE 9-, 14- and 39-bus networks are readily available as examples of DIgSILENT 15.1 software. If they will satisfy the studies that are being done, they can be simply imported from the software’s data without bothering tracing and implementing them from scratch, otherwise the steps related to creating a new project presented in Sect. 11.2.1 will be at work. In addition, the related file has been enclosed with the chapter and can be imported easily in DIgSILENT.
11.3.1.1
How to Import IEEE 39-Bus New England Power Network from the Examples or a File
In order to import the power network from the examples open DIgSILENT PowerFactory ! go to the File Menu and press Examples (Fig. 11.33) ! in “Additional Examples” tab choose “39 Bus System” and press Folder Icon (Import and activate Example Project) (Fig. 11.34). Then the corresponding power network will be opened and activated and is ready to put a few changes on it (Fig. 11.35).
Fig. 11.33 DIgSILENT PowerFactory—File Menu ! Examples
11
Reactive Power Optimization Using MATLAB and DIgSILENT
455
Fig. 11.34 DIgSILENT PowerFactory—Examples ! 39 Bus System
Fig. 11.35 DIgSILENT PowerFactory—39 Bus System ready to use
11.3.1.2
Putting Finishing Touches on the Power Network in DIgSILENT
The next stage will be preparing the prerequisite conditions toward using DIgSILENT scripting facility. The first step will be “Creating a DPL Command Set”, refer to the Sect. 11.2.2.1 to find out how to do so. The next step will be creating a general set and introducing it to the DPL which can be easily found in Sect. 11.2.2.2. The third step is introducing external variables to DPL which is
456
N. Mahdavi Tabatabaei et al.
Fig. 11.36 DIgSILENT PowerFactory—DPL command set
already represented in Sect. 11.2.2.3. According to IEEE 39-bus New England Power network given in Chap. 10, there are 20 control variables as the voltage magnitude of 9 generators and the tap setting for 11 transformers, all of which should be defined in the “External Objects” section (Fig. 11.36). The final result has been shown in Fig. 11.36 and the file is enclosed with the chapter ready to be imported, containing all the necessary changes. Eventually, the script related to Reactive Power Optimization will be written in the “Script” tab in the “DPL Command” window.
11.3.2 How to Write Scripts for DIgSILENT and MATLAB Step by Step This section will represent the procedure of the reactive power optimization using built-in functions of MATLAB step by step. All the functions will be introduced where necessary in order to make it easy to comprehend. The optimization algorithm has considered to be particle swarm optimization algorithm, both of which have got their own efficient functions in MATLAB.
11.3.2.1
Initializing and Defining the Problem in MATLAB
As presented in the codes below, global variables have been considered to be used. Global variables are at work when a variable comes to be used in several functions and each function makes its own changes on the variable, all of which are vital in the procedure. Their values should be set to zero right at the beginning in order to
11
Reactive Power Optimization Using MATLAB and DIgSILENT
457
avoid any unexpected interference during several implementations of the process. Cost function has been written as a function itself and defined to the procedure using “Cost function” section in the scripts below. The rest of the codes represents the definition of the upper and lower boundaries of the control variables. Initializing and Defining the Problem
MATLAB RPONewEngland.m
clc; clear; close all; %% Problem Definition global nfe NFE BVMin BVMax BestFunValuePSO IterationPSO BestFunValuePS IterationPS; NFE = 0; nfe = 0; BestFunValuePSO = 0; IterationPSO = 0; BestFunValuePS = 0; IterationPS = 0; CostFunction = @(OCV) RPO_Function(OCV);
% Cost Function
% Decision Variables % % VG02,VG03,VG04,VG05,VG06,VG07,VG08,VG09,VG10 % TT1211,TT1213,TT0631,TT1032,TT1933,TT2034,TT2235,TT2537,TT0230 % TT2938,TT1920 % Busbars Voltage Limits P.U. % % Acceptable Voltage Limitations, +-10% for 16.5, 138, and 230 kV Busbars, % +-5% for 345 kV Busbars % BVMin = [0.95 0.95 0.95 0.95 0.95 0.95 0.95 0.95 0.95 0.95 0.95 0.90... % 0.95 0.95 0.95 0.95 0.95 0.95 0.95 0.90 0.95 0.95 0.95 0.95 0.95... % 0.95 0.95 0.95 0.95 0.9 0.9 0.9 0.9 0.9 0.9 0.9 0.9 0.9 0.9]; % BVMax = [1.05 1.05 1.05 1.05 1.05 1.05 1.05 1.05 1.05 1.05 1.05 1.10... % 1.05 1.05 1.05 1.05 1.05 1.05 1.05 1.10 1.05 1.05 1.05 1.05 1.05... % 1.05 1.05 1.05 1.05 1.10 1.10 1.10 1.10 1.10 1.10 1.10 1.10 1.10 1.10]; % Busbars Voltage Limits P.U. % % Acceptable Voltage Limitations, +-10% for 16.5, 138, 230 and 345 kV Busbars BVMin = 0.9*ones(1,39); BVMax = 1.1*ones(1,39); % Generators Voltage Limits P.U. % GVMin = [0.9 0.9 0.9 0.9 0.9 0.9 0.9 0.9 0.9]; GVMax = [1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1]; % Transformators Tap Limits %
458
N. Mahdavi Tabatabaei et al.
TTMin = [-1 -1 -2 -2 -2 -1 -1 -1 -1 -1 -2]; TTMax = [1 1 2 2 2 1 1 1 1 1 2]; VarMin = [GVMin TTMin]; VarMax = [GVMax TTMax];
11.3.2.2
% Lower Bound of Variables % Upper Bound of Variables
Initial Power Losses According to New England 39-Bus Network
In order to calculate initial value of active power losses according to IEEE 39-bus New England it is needed to apply initial condition to the network and after running power flow calculation of losses will be feasible. The matrix “SVCV” represents Standard Values for Control Variables which are taken from the data given in the appendix section of Chap. 10. Then they will be exported in a “.txt” file as “OCV. txt” to be imported in DIgSILENT. Finally, MATLAB will make a “Couple.txt” file as a means of exchanging commands between two pieces of software, waiting for DIgSILENT to implement its duty.
Initial Power Losses
MATLAB RPONewEngland.m
%% Initial Power Losses According to New England 39-Bus Network % Standard Values for Control Variables % SVCV = [0.982 0.9831 0.9972 1.0123 1.0493 1.0635 1.0278 1.0265 1.0475... 1 1 2 2 2 1 1 1 1 1 2]; t = fopen('OCV.txt','w'); % Optimal Control Variables fprintf(t,'%d %d %d %d %d %d %d %d %d %d %d %d %d %d %d %d %d %d %d %d',SVCV); fclose(t); % Making a Blank "Couple.txt" File, 0 => DIgSILENT, 1 => MATLAB % t = fopen('Couple.txt','w'); fprintf(t,'%d',0); fclose(t); % Waiting for DIgSILENT to Complete the PowerFlow Process % a = 0; while a == 0 pause(0.02); a = load('Couple.txt'); end
11
Reactive Power Optimization Using MATLAB and DIgSILENT
11.3.2.3
459
Initializing DIgSILENT Software
Initializing DIgSILENT consists in defining variables, which will be used throughout the procedure of optimization. Integer variables for storing and using integer values, object variables for storing equipment used in the power network, double variables for storing values for control variables, set variables for storing a group of objects, and string variables for storing string values.
Initializing DIgSILENT Software
DIgSILENT
int a; object o; double V,U,Phi,Ploss,Qloss; double VG02,VG03,VG04,VG05,VG06,VG07,VG08,VG09,VG10,TT1211,TT1213, TT0631,TT1032, TT1933,TT2034,TT2235,TT2537,TT0230,TT2938,TT1920; set s1; string BusName,LineName; a=1;
11.3.2.4
How to Apply Control Variables to the Network
The main loop of reactive power optimization in DIgSILENT consist in running power flow and exporting the results. Each time MATLAB wants DIgSILENT run, the procedure in DIgSILENT starts from this session, and if MATLAB sends command signal as “2”, the procedure in DIgSILENT will be over. Referring to the codes below, DIgSILENT first reads the “OCV.txt” file written by MATLAB and then allocated the values to the control variables “VG02” to “TT1920”. Then applying the values using the external objects defined in Sect. 11.3.1.2 will be feasible as done below. It means that now the values considered to be control variables are set to the values created by MATLAB. The next step will be running the power flow according to new values.
460
N. Mahdavi Tabatabaei et al.
Applying Control Variables to the Network
DIgSILENT
!! The Main Loop of Reactive Power Optimization while(a2) ! Checking to Finish the Process { fopen('C:\Users\ALI JAFARI\Documents\MATLAB\ IEEE 39-Bus New England Power System\Reactive Power Optimization\OCV.txt','r',0); fscanf(0,'%d %d %d %d %d %d %d %d %d %d %d %d %d %d %d %d %d %d %d %d',VG02,VG03,VG04,VG05,VG06,VG07,VG08,VG09,VG10,TT1211,TT1213, TT0631,TT1032,TT1933,TT2034,TT2235,TT2537,TT0230,TT2938,TT1920); fclose(0); G02:usetp = VG02; G03:usetp = VG03; G04:usetp = VG04; G05:usetp = VG05; G06:usetp = VG06; G07:usetp = VG07; G08:usetp = VG08; G09:usetp = VG09; G10:usetp = VG10; T1211:nntap = TT1211; T1213:nntap = TT1213; T0631:nntap = TT0631; T1032:nntap = TT1032; T1933:nntap = TT1933; T2034:nntap = TT2034; T2235:nntap = TT2235; T2537:nntap = TT2537; T0230:nntap = TT0230; T2938:nntap = TT2938; T1920:nntap = TT1920;
11.3.2.5
Running Load Flow in the Script Section in DIgSILENT
After applying values for control variables, it is time to carry the load flow calculations out. To do so, load flow should be introduced to the DPL command set as an external object, which is illustrated in Sect. 11.2.2.3. The procedure is as the following.
11
Reactive Power Optimization Using MATLAB and DIgSILENT
Running Load Flow
461
DIgSILENT
! Running Load Flow Calculation ClearOutput(); ResetCalculation(); Ldf.Execute();
11.3.2.6
Exporting Load Flow Results from DIgSILENT
The results of power flow will be exported using “fprintf” function via “LoadFlow. txt” as a comma separated value (CSV) file. The data being exported completely depends on what type of information is exactly needed within the calculations in MATLAB pertaining to the study being done. In this section voltage magnitudes, p.u. values for the voltages and voltages degrees have been exported.
Exporting Load Flow Results
DIgSILENT
! Exporting Load Flow Results fopen('C:\Users\ALI JAFARI\Documents\MATLAB\IEEE 39-Bus New England Power System\Reactive Power Optimization\LoadFlow.txt','w',0); fprintf(0,'%s,%s,%s,%s','BusName','Voltage_V','Voltage_PU','VoltageAngle_Deg'); s1 = SEL.AllBars(); s1.SortToName(0); o = s1.First(); while (o){ V = o:m:Ul; U = o:m:u1; Phi = o:m:phiu; BusName = o:loc_name; fprintf(0,'%s,%f,%f,%f',BusName,V,U,Phi); o = s1.Next(); } fclose(0);
11.3.2.7
Exporting Transmission Lines Power Losses from DIgSILENT
The transmission lines power losses can also be exported from DIgSILENT instead of calculating them in MATLAB using the active power losses of transmission lines equation expressed in Chap. 10. The only thing which should be done is to find the appropriate parameter name for active power losses to be able to export the corresponding value for all the transmission lines. DIgSILENT is able to calculate wide range of parameters related to power flow without suffering programming difficulties. How to find the parameter name which is used in the scripting session to export the related value
462
N. Mahdavi Tabatabaei et al.
has been presented in Section “Running Power Flow in DIgSILENT Based on Initial Condition”. Following the same instruction, “Ploss_bus1” is representative of active power losses in transmission lines and can be used in the scripting segment to export the results. The following code exports power losses for transmission lines in “TLPowerLosses.txt” as a CSV file. Eventually, DIgSILENT sends a signal to MATLAB via “Couple.txt” to let it know that power flow is done, results are exported and ready to use, then it waits to get another set of values for control variables and another command from MATLAB to run power flow based on new condition again.
Exporting Transmission Lines Power Losses
DIgSILENT
! Exporting Transmission Lines Power Losses fopen('C:\Users\ALI JAFARI\Documents\MATLAB\IEEE 39-Bus New England Power System\Reactive Power Optimization\TLPowerLosses.txt','w',0); fprintf(0,'%s,%s,%s','LineName','ActivePowerLoss_W','ReactivePowerLoss_VAR'); s1 = SEL.AllLines(); s1.SortToName(0); o = s1.First(); while (o){ Ploss = o:m:Ploss:bus1; Qloss = o:m:Qloss:bus1; LineName = o:loc_name; fprintf(0,'%s,%f,%f',LineName,Ploss,Qloss); o = s1.Next(); } fclose(0); ! Sending Run Signal to MATLAB fopen('C:\Users\ALI JAFARI\Documents\MATLAB\IEEE 39-Bus New England Power System\Reactive Power Optimization\Couple.txt','w',1); fprintf(1,'%d',1); fclose(1); ! Waitign for MATLAB to Finish the Corresponding Calculations a = 1; while(a = 1){ fopen('C:\Users\ALI JAFARI\Documents\MATLAB\IEEE 39-Bus New England Power System\Reactive Power Optimization\Couple.txt','r',0); fscanf(0,'%d',a); fclose(0); } }
11.3.2.8
Importing Active Power Losses from Text File in MATLAB
In order to import power losses results into MATLAB from the “TLPowerLosses. txt” file, a built-in function called “readtable(‘TLPowerLosses’)” can be used,
11
Reactive Power Optimization Using MATLAB and DIgSILENT
463
which creates a table by reading column oriented data from a file. “readtable” creates one variable in “TLPL” for each column in the file and reads variable names from the first row of the file. In the following code MATLAB imports the data written by DIgSILENT in the “TLPowerLosses.txt” file and then sums up the values in the “ActivePowerLoss_W” column to work out the total value of active power losses in transmission lines and stores it in “PowerLosses.Initial.APL”. In addition, it saves the values for control variables in “PowerLosses.Initial.Position”.
Importing Active Power Losses from “TLPowerLosses.txt”
MATLAB RPONewEngland.m
% Loading Active Power Losses from DIgSILENT Calculation % TLPL = readtable('TLPowerLosses'); % Transmission Lines Power Losses Data PowerLosses.Initial.APL = sum(TLPL.ActivePowerLoss_W); PowerLosses.Initial.Position = SVCV;
11.3.2.9
Reactive Power Optimization Using Built-in Particle Swarm Function
MATLAB provides a built-in function for particle swarm optimization algorithm in its recent versions. It can be effortlessly used in optimization algorithms adjusting some parameters without programming the algorithm itself. A user defined function has been created as “ParticleSwarmTool” in order to gather all the corresponding parameters and the Particle Swarm built-in function in a unique mfile. The inputs to this function is just lower and upper bounds for control variables and the initial position which is optional.
Reactive Power Optimization Using PSO Algorithm
MATLAB RPONewEngland.m
%% Reactive Power Optimization Using PSO Algorithm [PowerLosses.PSO.Position,PowerLosses.PSO.APL,exitflagPSO] = ... ParticleSwarmTool(VarMin,VarMax); LNFE = length(nfe);
The main function which carries out the optimization procedure is “particleswarm”, while there are a few prerequisite adjustments to be made before using the function itself. In order to access the details of the particle swarm optimization function MATLAB provides another function as “optimoptions”. The detailed information about how to use it is available either online or MATLAB software local help. In this section just the needed adjustments have been presented. “options = optimoptions (options,‘InitialSwarm’, [0.982 0.9831 0.9972, 1.0123 1.0493 1.0635 1.0278 1.0265 1.0475 1 1 2 2 2 1 1 1 1 1 2]);” has been used to define initial point to PSO algorithm
464
N. Mahdavi Tabatabaei et al.
which is optional. Self-adjustment and social-adjustment rates have been specified to the function using “options = optimoptions(options,‘SelfAdjustment’,1.49);” and “options = optimoptions(options,‘SocialAdjustment’,1.49);”. In addition, the member of members in the swarm has been defined using “options = optimoptions (options,‘SwarmSize’,20);” and so forth, to see more information please refer to MATLAB software help.
Reactive Power Optimization Using PSO Algorithm
MATLAB ParticleSwarmTool.m
function [x,fval,exitflagPSO] = ParticleSwarmTool(VarMin,VarMax) nVar = 20;
% Number of Decision Variables
options = optimoptions('particleswarm'); options = optimoptions(options,'CreationFcn',@pswcreationuniform); options = optimoptions(options,'Display','Iter'); options = optimoptions(options,'DisplayInterval',1); options = optimoptions(options,'FunValCheck','off'); options = optimoptions(options,'HybridFcn',[]); options = optimoptions(options,'InertiaRange',[0.1 1.1]); options = optimoptions(options,'InitialSwarm',[0.982 0.9831 0.9972... 1.0123 1.0493 1.0635 1.0278 1.0265 1.0475 1 1 2 2 2 1 1 1 1 1 2]); options = optimoptions(options,'InitialSwarmSpan',2000); options = optimoptions(options,'MaxIter',Inf); options = optimoptions(options,'MaxTime',Inf); options = optimoptions(options,'MinFractionNeighbors',0.25); options = optimoptions(options,'ObjectiveLimit',-Inf); options = optimoptions(options,'OutputFcns',@ParticleSwarmOutputFcn); options = optimoptions(options,'PlotFcns',[]); options = optimoptions(options,'SelfAdjustment',1.49); options = optimoptions(options,'SocialAdjustment',1.49); options = optimoptions(options,'StallIterLimit',20); options = optimoptions(options,'StallTimeLimit',Inf); options = optimoptions(options,'SwarmSize',20); options = optimoptions(options,'TolFun',1e-2); options = optimoptions(options,'UseParallel',0); options = optimoptions(options,'Vectorized','off'); % The Main Function of Particle Swarm Optimization Algorithm % [x,fval,exitflagPSO] = ... particleswarm(@RPO_Function,nVar,VarMin,VarMax,options);
The general form to use the built-in function of particle swarm is like “[x,fval,exitflag, output] = particleswarm(fun,nvars,lb,ub,options)” in which “x” returns the position of control variables, “fval” represents the best cost, “exitflag” describes the exit condition, “output” containing information about the optimization process, “fun” is a function containing objective function, “nvars” refers to number of control variables, “lb” and “lu” embodies lower and upper bounds for control variables and “options” is a variables
11
Reactive Power Optimization Using MATLAB and DIgSILENT
465
consisting of all the setting parameters of PSO function. The objective function will be defined as the following codes.
Reactive Power Optimization Objective Function
MATLAB RPO_Function.m
function [OF] = RPO_Function(OCV) global NFE BVMin BVMax; NFE = NFE+1; k = 1e15; % Penalty Factor % Calculation of Objective Function (Ploss) % t = fopen('OCV.txt','w'); % Consisting of Optimal Control Variables fprintf(t,'%d %d %d %d %d %d %d %d %d %d %d %d %d %d %d %d %d %d %d %d',OCV); fclose(t); % Sending Run Signal to DIgSILENT; 0 => DIgSILENT, 1 => MATLAB % t = fopen('Couple.txt','w'); fprintf(t,'%d',0); fclose(t); % Waiting for DIgSILENT to Complete the PowerFlow Process % a = 0; while a == 0 pause(0.03); a = load('Couple.txt'); end % Loading Power Losses from DIgSILENT Calculation % TLPL = readtable('TLPowerLosses'); % Transmission Lines Power Losses Data FAPL = sum(TLPL.ActivePowerLoss_W); % Transmission Systems Active Power Losses % FRPL = sum(TLPL.ReactivePowerLoss_VAR); % Transmission Systems Reactive Power Losses % Applying Penalty to the Cost Function for Keeping Voltages of PQ Busbars Within the Limitations % PF = readtable('LoadFlow'); % Load Flow Results PQBV = (PF{1:29,3})'; % Extracting PQ Busbars Voltatges PV = k*sum(BVMax(1:29)-PQBV0); % To Apply a Penalty to the Cost Function % if There is a Voltage Violation OF = FAPL + PV; end
466
N. Mahdavi Tabatabaei et al.
In the reactive power optimization function above “NFE” has been considered to work the number of function evaluation out. “% Calculation of Objective Function (Ploss) %” section tries to apply the new set of control variables to the power network in DIgSILENT and carries out power flow and the calculates active power losses accordingly. Section “% Applying Penalty to the Cost Function for Keeping Voltages of PQ Busbars Within the Limitations %” has been considered to apply a penalty to the cost value if there is a violation in the voltages of PQ busbars. The particle swarm built-in function refers to the objective function several times to figure out if the objective function has been optimized or not. In order to have access to the detailed data throughout the procedure a user defined function “ParticleSwarmOutputFcn” has been used, which is a facility that built-in PSO function provides. It also can be used in order to stop the function using a specific criterion, to see more information refer to MATLAB software help documentation. The code is as following.
Particle Swarm Output Function
MATLAB ParticleSwarmOutputFcn.m
function stop = ParticleSwarmOutputFcn(optimValues,state) global nfe NFE BestFunValuePSO IterationPSO; stop = false;
% This function does not stop the solver
switch state case 'init' case 'iter' BestFunValuePSO(optimValues.iteration) = optimValues.bestfval; nfe(optimValues.iteration) = NFE; case 'done' IterationPSO = optimValues.iteration; end
After finishing the optimization procedure of the PSO function, the optimum positions for control variables are stored in “PowerLosses.PSO.Position” and the cost value for the values of control variables in “PowerLosses.PSO.APL”. The next step will be using Pattern Search as the supplementary optimization algorithm to pinpoint the exact optimum control variables.
11
Reactive Power Optimization Using MATLAB and DIgSILENT
11.3.2.10
467
Reactive Power Optimization Using Built-in Pattern Search Function
Pattern search optimization algorithm has been used as a supplementary optimization algorithm in case PSO is not able to pinpoint the optimum control variables. User defined “PatternSearchTool” has been considered to gather all the parameters and settings which are needed to adjust and run pattern search algorithm all together. The codes below call the function.
Reactive Power Optimization Using Pattern Search
MATLAB RPONewEngland.m
%% Reactive Power Optimization Using PatternSearch Algorithm [PowerLosses.PS.Position,PowerLosses.PS.APL,exitflagPS] = ... PatternSearchTool(PowerLosses.PSO.Position,VarMin,VarMax);
The main function which carries out the optimization procedure is “patternsearch”, while there are a few requirements needed to be met before using the function itself. In order to access the details of the pattern search function, MATLAB provides another function as “psoptimset”. The detailed information about how to use it is available either online or MATLAB software local help. In this section just the needed adjustments have been presented. “options = psoptimset (options,‘TolFun’,1e-4);” is the tolerance on the objective function which stops the optimization procedure if both the change in function value and the mesh size are less than “TolFun”. The general form to use the built-in function of pattern search is like “[x,fval, exitflag,output] = patternsearch(fun,x0,A,b,Aeq,beq,LB,UB)” in which “x” returns the position of control variables, “fval” represents the best cost, “exitflag” describes the exit condition, “output” containing information about the optimization process, “fun” is a function containing objective function, “x0” is the starting point to the pattern search algorithm, “A”, “b”, “Aeq” and “beq” are parameters of inequality and equality constraints, “LB” and “UB” lower bounds and upper bounds of the control variables, and “options” is a variables consisting of all the setting parameters of pattern search function.
468
N. Mahdavi Tabatabaei et al.
Reactive Power Optimization Using Pattern Search
MATLAB PatternSearchTool.m
function [x,fval,exitflagPS] = PatternSearchTool(x0,VarMin,VarMax) options = psoptimset; options = psoptimset(options,'TolMesh',[]); options = psoptimset(options,'TolCon',[]); options = psoptimset(options,'TolX',[]); options = psoptimset(options,'TolFun',1e-4); options = psoptimset(options,'TolBind',[]); options = psoptimset(options,'MaxIter',Inf); options = psoptimset(options,'MaxFunEvals',[]); options = psoptimset(options,'TimeLimit',[]); options = psoptimset(options,'MeshContraction',[]); options = psoptimset(options,'MeshExpansion',[]); options = psoptimset(options,'MeshAccelerator','on'); options = psoptimset(options,'MeshRotate',[]); options = psoptimset(options,'InitialMeshSize',[]); options = psoptimset(options,'ScaleMesh',[]); options = psoptimset(options,'MaxMeshSize',[]); options = psoptimset(options,'InitialPenalty',[]); options = psoptimset(options,'PenaltyFactor',[]); options = psoptimset(options,'PollMethod',[]); options = psoptimset(options,'CompletePoll',[]); options = psoptimset(options,'PollingOrder','Success'); options = psoptimset(options,'SearchMethod',@GPSPositiveBasis2N); options = psoptimset(options,'CompleteSearch',[]); options = psoptimset(options,'Display','iter'); options = psoptimset(options,'OutputFcns',@PatternSearchOutputFcn); options = psoptimset(options,'PlotFcns',[]); options = psoptimset(options,'PlotInterval',[]); options = psoptimset(options,'Cache',[]); options = psoptimset(options,'CacheSize',[]); options = psoptimset(options,'CacheTol',[]); options = psoptimset(options,'Vectorized',[]); options = psoptimset(options,'UseParallel',[]); % The Main Function of Pattern Search Optimization Algorithm % [x,fval,exitflagPS] = ... patternsearch(@RPO_Function,x0,[],[],[],[],VarMin,VarMax,[],options);
“options = psoptimset(options,‘OutputFcns’,@PatternSearchOutputFcn);” is used to access the detailed information during the optimization procedure in every iteration as presented for PSO function in Sect. 11.3.2.9. See the help documentation of MATLAB for more information how to use it.
11
Reactive Power Optimization Using MATLAB and DIgSILENT
Pattern Search Output Function
469
MATLAB PatternSearchOutputFcn.m
function [stop,options,optchanged] = PatternSearchOutputFcn(optimValues,options,state) global nfe NFE BestFunValuePS IterationPS IterationPSO; stop = false; % This function does not stop the solver optchanged = false; switch state case 'init' case 'iter' BestFunValuePS(optimValues.iteration) = optimValues.fval; nfe(optimValues.iteration + IterationPSO) = NFE; case 'done' IterationPS = optimValues.iteration; end
After finishing the optimization procedure by Pattern Search function, the optimum positions for control variables are stored in “PowerLosses.PS.Position” which will be best point found, and the best cost value will be saved in “PowerLosses.PS.APL” which is the best minimum cost.
11.3.2.11
Finishing the Optimization Procedure and Exporting Results
After finishing optimization procedure, the final step will be sending a stop signal to DIgSILENT and printing the data in relation to what information was meant to be exported. The following codes are an example of such processes.
470
N. Mahdavi Tabatabaei et al.
Finishing the Optimization Procedure
MATLAB RPONewEngland.m
%% Sending Stop Signal to DIgSILENT t = fopen('Couple.txt','w'); fprintf(t,'%d',2); fclose(t); %% Results clc hold on plot(nfe(1:LNFE),BestFunValuePSO,'-o','LineWidth',1); plot(nfe(LNFE+1:end),BestFunValuePS,'-.*','LineWidth',1); xlabel('Number of Function Evaluation (NFE)'); ylabel('Active Power Losses (W)'); LR = PowerLosses.Initial.APL - PowerLosses.PS.APL; % Losses Reduction LRP = 100*((PowerLosses.Initial.APL-PowerLosses.PS.APL)... /PowerLosses.Initial.APL); % Losses Reduction Percentage, Total disp('------------------------------------------------------------------'); str = sprintf('Initial Active Power Losses (W) = %d',... PowerLosses.Initial.APL); disp(str); disp('------------------------------------------------------------------'); str = sprintf('Active Power Losses After Optimization (W) = %d',... PowerLosses.PS.APL); disp(str); disp('------------------------------------------------------------------'); str = sprintf('Losses Reduction (W), Total = %d',LR); disp(str); disp('------------------------------------------------------------------'); str = sprintf('Losses Reduction Percentage , Total = %d',LRP); disp(str);
11
Reactive Power Optimization Using MATLAB and DIgSILENT
471
11.3.3 How to Run the Optimization Procedure After having the mfiles of MATLAB and scripts of DIgSILENT ready, it is the final step to run the optimization process. It should be noted that, according to the structure used in the script files enclosed with this chapter, the mfile of MATLAB is the first one which should be run and then DIgSILENT. To run the optimization procedure via MATLAB by “RPO_NewEngland.m” simply click on “Run” icon in the mfile toolbar, and to run the script written in DIgSILENT right click on the script and click on “Execute” (Fig. 11.37). Figure 11.38 shows the result of optimization after the process is done for one run. The elaborate report of reactive power optimization for the same power grid has been presented in the Simulation Results in Chap. 10.
11.3.4 How to Run Small Signal Analysis in DIgSILENT There are a few steps through running small signal analysis in DIgSILENT. The IEEE 39-bus New England power system is already prepared for carrying out such analysis. To run the process, activate Small Signal Analysis in the Study Case section (Fig. 11.39), then choose Modal Analysis toolbar (Fig. 11.40). The equation system has to be initialized (Calculate Initial Conditions) (Fig. 11.41), then
Fig. 11.37 DIgSILENT and MATLAB—how to run the optimization procedure
472
N. Mahdavi Tabatabaei et al.
Fig. 11.38 MATLAB—the result after the optimization
Fig. 11.39 DIgSILENT PowerFactory—activating Small Signal Analysis study case
calculation of the eigenvalues can be executed by pressing the button Modal Analysis (Fig. 11.41). The study case “Small Signal Analysis (Eigenvalues)” contains an eigenvalue plot which displays the results (Fig. 11.41) [1].
11
Reactive Power Optimization Using MATLAB and DIgSILENT
473
Fig. 11.40 DIgSILENT PowerFactory—Modal Analysis tool
Fig. 11.41 DIgSILENT PowerFactory—Eigenvalue Plot
11.4
Summary
The main objective of reactive power optimization is to reduce active power losses in the power network and improve voltage profile. Reactive power optimization problem is a multidimensional problem which has plenty of local optimum points that causes the traditional optimization methods, which mostly are based on gradient method, to have difficulty finding the global point. Heuristic algorithms have
474
N. Mahdavi Tabatabaei et al.
proved their high performance in optimizing these kind of complex problems, one of which is the particle swarm optimization algorithm that is highly appreciated in terms of global optimization. In addition, there are some direct search methods which do not use target function’s gradient but itself, of which are very performable when it comes to local optimization. It seems, then, applicable to use a local optimization algorithm in association with a global optimizer to pinpoint the optimum point. In this chapter, PSO algorithm has been considered as the main and global optimizer and pattern search algorithm as supplementary and the local one through optimizing reactive power in the power grid. This chapter represents a particle swarm pattern search algorithm to optimize reactive power to decrease active power losses in an IEEE 6-bus power network. The optimization procedure has done by an approach consisting of two software MATLAB and DIgSILENT which is explained step by step. In addition, another method to optimize reactive power using ready to use IEEE 39-bus New England in DIgSILENT and built-in optimization functions in MATLAB has been presented while considering the fact that the reader has intermediate knowledge how to use both pieces of software. The corresponding files have been attached with the book to use. Additional approaches containing reactive power optimization using genetic pattern search algorithm for IEEE 6-bus power grid, particle swarm pattern search algorithm for IEEE 14-bus and genetic pattern search algorithm for IEEE 14-bus have been enclosed with the book. The simulations results have been presented in Chap. 10 (10.8—Reactive Power Optimization Simulation Results) and further comparisons have been put there. Overall, the aim of this chapter was to present an introductory method through using MATLAB and DIgSILENT working together in order to avoid a few unimportant steps towards further studies in the AC power network.
References 1. DIgSILENT PowerFactory Version 15 User Manual, Gomaringen, Germany: DIgSILENT GmbH, 2014. 2. MATLAB Help, The MathWorks, Inc., 2014.
Chapter 12
Multi-objective Optimal Reactive Power Dispatch Considering Uncertainties in the Wind Integrated Power Systems Seyed Masoud Mohseni-Bonab, Abbas Rabiee and Behnam Mohammadi-Ivatloo Abstract One of the most principle optimization problems which gained the attention of power system operators around the world is optimal power flow (OPF). The OPF basically performs an intelligent power flow and optimizes the system operation condition by optimally determination of control variables. It also considers a specific set of operational constraints and technical limits for this aim, which guaranties both feasibility and optimality of the scheduled operation condition. Generally, this problem can be categorized into two main sub-problems, i.e. optimal reactive power dispatch (ORPD) and optimal real power dispatch, which are differ in their aims and control variables. This chapter deals with the first one, ORPD, which has significant impact on power system security. ORPD is modeled as an optimization problem with nonlinear functions and mixed continuous/discrete variables. Thus, it is a complicated optimization problem. The multi-objective ORPD (MO-ORPD) problem is studied, taking into account different operational constraints such as bus voltage limits, power flow limits of branches, limits of generators voltages, transformers tap ratios and the amount of available reactive power compensation at the weak buses. Three different objective functions are considered in the proposed MO-ORPD framework, which are minimizing total active power losses, minimizing voltage variations and minimizing voltage stability index (L-index). These conflicting objectives are optimized via e-constraint method. In order to model the stochastic behavior of demand and wind power generation, it is necessary to modify the MO-ORPD problem, and develop a probabilistic approach to handle the uncertainties in MO-ORPD problem. Hence, a two-stage stochastic MO-ORPD (SMO-ORPD) is suggested to handle the load and wind power uncertainties in the MO-ORPD problem. In the proposed two-stage S.M. Mohseni-Bonab A. Rabiee (&) Department of Electrical Engineering, University of Zanjan, Zanjan, Iran e-mail: [emailprotected] S.M. Mohseni-Bonab e-mail: [emailprotected]; [emailprotected] B. Mohammadi-Ivatloo Faculty of Electrical and Computer Engineering, University of Tabriz, Tabriz, Iran e-mail: [emailprotected] © Springer International Publishing AG 2017 N. Mahdavi Tabatabaei et al. (eds.), Reactive Power Control in AC Power Systems, Power Systems, DOI 10.1007/978-3-319-51118-4_12
475
476
S.M. Mohseni-Bonab et al.
stochastic optimization model, the decision variables are classified into two categories, namely, “here and now” and “wait and see” variables. The optimal values of “here and now” variables should be known before realization of scenarios, and therefore, their values are the same for all scenarios while the optimal values of “wait and see” variables are based on the realized scenario, and hence their values are scenario dependent. Moreover, in order to examine performance of the proposed SMO-ORPD and the impact of wind power generation on the results of SMO-ORPD, deterministic ORPD (DMO-ORPD) has also been solved in two modes: DMO-ORPD without wind farms (WFs) and any uncertainty, for the sake of comparison with the available methods in recent literature, and DMO-ORPD with WFs. In this chapter the reactive power compensation devices are modeled as discrete/continuous control variables. DMO-ORPD and SMO-ORPD are formulated as mixed integer non-linear program (MINLP) problems, and solved by General Algebraic Modeling System (GAMS). Also, the IEEE 30-bus standard system is utilized for evaluation of the proposed MO-ORPD models.
12.1
Introduction and Problem Statement
Optimal reactive power dispatch (ORPD) is an important problem for power utilities from the viewpoints of system security and energy losses. ORPD is a specific form of optimal power flow (OPF) problem, in which various objective functions such as transmission losses or voltage stability enhancement indices are optimized by adjusting the generator voltages set-points, reactive power compensation in weak buses and optimal setting of transformers tap ratios.
12.1.1 Background and Review of the Recent Literature Active power losses are the most important objective function used in the classical OPRD problem. Many heuristic algorithms such as multi-agent-based particle swarm optimization approach (MA-PSO) [1], seeker optimization algorithm (SOA) [2], shuffled frog leaping algorithm (SFLA) and Nelder–Mead SFLA (NM-SFLA) [3], combination of hybrid modified imperialist competitive algorithm (MICA) and invasive weed optimization (IWO) named as (MICA-IWO) [4], combination of modified teaching learning algorithm (MTLA) and double differential evolution (DDE) algorithm [5] have been developed to obtain optimal solutions of ORPD while satisfying technical constraints. In [6], active power losses are minimized in wind farms (WFs) by PSO algorithm taking into consideration the reactive power requirement at the point of common coupling (PCC). In some papers, active power losses are considered with two other common objective functions of ORPD. At the first category active power losses are considered with voltage deviation, while in the second category active power losses are
12
Multi-objective Optimal Reactive Power Dispatch …
477
optimized along with a voltage stability enhancement index. In the third category active power losses are considered with both of the other objectives. In order to handle these categories, various intelligent algorithms are employed in the existing literature, such as feasible solutions differential evolution (SF-DE), self-adaptive penalty DE (SP-DE), e-constraint DE (EC-DE) and stochastic ranking DE (SR-DE) in [7], evolutionary programming (EP), PSO, DE and the hybrid differential evolution (HDE) in [8], combination of NM and Firefly Algorithm (FA) in [9], PSO with an aging leader and challengers (ALC-PSO) in [10] and gray wolf optimizer (GWO) in [11], linear programming (LP) based method using FACTS devices in [12], nonlinear programming (NLP) based method for optimizing local voltage stability index in [13], SOA in [14], DE in [15], PSO, simple genetic algorithm (SGA) and harmony search algorithm (HSA) in [16], gravitational search algorithm (GSA) in [17]. A different research carried out on the single objective ORPD problem, in which steady state voltage stability analysis has been carried out in [18] considering initial conditions for transient stability (TS), small disturbance (SD), and continuation power flow (CPF) for minimizing real power losses. Also, penalty function based method has been presented in [19] to convert discrete ORPD model to a continuous and differentiable model. The multi-objective ORPD (MO-ORPD) has attracted the attention researchers in recent years. Similar to single objective OPRD, many algorithms utilized to deal with MO-ORPD by considering two or three objective conflicting functions, simultaneously. Most popular objective function of MO-ORPD problem is real power losses which is considered with voltage deviation (as the conflicting objective function) in [20], and solved by strength Pareto evolutionary algorithm (SPEA). Also, it is solved in [21] by non-dominated sorting genetic algorithm-II (NSGA-II). MO-ORPD problem is modeled as a mixed integer nonlinear program (MINLP) and solved in some literature with considering real power losses and Lmax index. It is worth to note that, Lmax is a voltage stability index, where it varies from 0 at no-load condition to 1 at voltage collapse point. In [22], NSGA-II and modified NSGA-II (MNSGA-II), In [23] hybrid fuzzy multi-objective evolutionary algorithm (HFMOEA) and in [24], chaotic improved PSO based multi-objective optimization (MOCIPSO) and improved PSO-based multi-objective optimization (MOIPSO) are used for this aim. In more advanced MO-ORPD problem, three objective functions are considered simultaneously. For instance, teaching learning based optimization (TLBO) and quasi-oppositional TLBO (QOTLBO) in [25], chaotic parallel vector evaluated interactive honey bee mating optimization (CPVEIHBMO) in [26], and strength Pareto multi-group search optimizer (SPMGSO) [27] are used to solve MO-ORPD problem. Moreover, in recent literature ORPD or MO-ORPD problems are solved considering the effects of technical uncertainties raised from the power system restructuring, probable disturbances, or the integration of renewable energies. In [28], chance constrained programming technique is proposed to solve ORPD problem with the aim of minimizing active power losses. Nodal power injections and random branch outages are considered as uncertainty sources. In [29, 30]
478
S.M. Mohseni-Bonab et al.
the load uncertainty is included in MO-ORPD problem, considering different objective functions. Monte Carlo simulations (MCS) are used for handling the load uncertainties in the nonlinear constrained multi objective ORPD problem.
12.1.2 Chapter Contributions The main focus of this chapter is to solve the MO-ORPD problem in a wind integrated power system considering the uncertainties of load demand and wind power generation. Even these two sources of uncertainty are considered here, but the proposed method is generic and other uncertainty sources could be included via the proposed scenario based approach. The normal probability distribution function (PDF) and Rayleigh (PDF) are used for modeling the load and wind speed uncertainties, respectively. Three different objective functions, namely active power losses, voltage stability index (Lmax), and voltage deviations at load buses are considered. The multi-objective problem is handled using e-constraint technique and optimal Pareto sets are obtained for each pair of the above objective functions. In this chapter, for the sake of comparison with existing methods, the reactive power compensation by shunt VAR compensators is modeled as continuous variable in deterministic MO-ORPD (DMO-ORPD), while in real world problems discrete model is employed for these devices in the proposed stochastic MO-ORPD (SMO-ORPD). Thus, the DMO-ORPD is a NLP problem, while the SMO-ORPD is a MINLP problem. The proposed optimization problems are implemented in GAMS [31], and solved by SNOPT [32] and SBB [33] solvers, for NLP and MINLP problems, respectively. Hence, the main contributions of this chapter are outlined as follows: • Modeling and including stochastic nature of loads and wind power generations in the MO-ORPD problem (i.e. development of SMO-ORPD problem). • Investigation of the impact of renewable power generation on the results obtained by ORPD problem. • Utilizing discrete model for shunt VAR compensation devices in the proposed SMO-ORPD problem, since most of the pervious literature used continuous modeling for capacitor banks. • Implementation of e-constraint technique and fuzzy satisfying criteria to solve the MO-ORPD problem, and for selection of the best compromise solution, respectively. • Providing comprehensive illustrative studies for different types of ORPD problem, such as DMO-ORPD with and without wind integration and SMO-ORPD. • Comparison of the obtained results for DMO-ORPD case with previously published methods in literature, which confirms the efficiency of the proposed method.
12
Multi-objective Optimal Reactive Power Dispatch …
12.2
479
Uncertainty Modeling
Uncertain parameters in power systems are classified to technical and economical parameters. Technical parameters consist of operational (like demand and generation) and topological parameters, whereas economical parameters include macroeconomic and microeconomic parameters [34]. There are different methods for modeling these uncertain parameters which are summarized in Fig. 12.1. In this chapter, scenario based probabilistic method is utilized to handle to uncertainties. At the following, the detailed description of considered scenarios is given.
12.2.1 Demand Uncertainty Characterization via Scenario Based Modeling On account of stochastic nature of the load demand in electric power systems, it is required to model the load uncertainty in operation and planning of power systems. In the general manner load uncertainty can be modeled using the normal or Gaussian PDF [35]. In this chapter, it is presumed that the mean and standard deviation of the load PDF, µD and rD are known. Probability of d-th load scenario is represented by pd (probability of demand scenario d) and calculated using (12.1). max Figure 12.2 shows the load levels. It is worth to note that Pmin Dd and PDd (minimum/maximum value of real power demand at d-th load scenario) are the boundaries of d-th interval (or d-th load scenario), as shown in Fig. 12.2. Pmax
pd ¼
ZDd
Pmin D d
Uncertainty modeling approaches
1 pffiffiffiffiffiffiffiffiffiffi exp 2pr2
"
# ð P D lD Þ 2 dPD 2r2
ð12:1Þ
IGDT
Monte Carlo simulation (MCS)
Probabilistic
Point estimate method (PEM)
Interval based analysis
Scenario-based modeling
Robust optimization Fuzzy-Scenario
Hybrid probabilistic-possibilistic Fuzzy-Monte Carlo
Possibilistic Fig. 12.1 Uncertainty modeling approaches [34]
480
S.M. Mohseni-Bonab et al.
(a)
(b)
Fig. 12.2 The load PDF and load scenarios, a Normal PDF, b considered scenarios
Fig. 12.3 Rayleigh PDF for wind speed characterization
Wind Speed (m/sec)
max
PDd
1 ¼ pd
ZPD;d
Pmin D;d
1 PD pffiffiffiffiffiffiffiffiffiffi exp 2pr2
"
ðPD lD Þ2 2r2
#!
dPD
ð12:2Þ
12.2.2 Wind Power Generation Uncertainty Modeling Principally the wind speed uncertainty is modeled using the Rayleigh or Weibul PDF [36, 37]. It should be noticed that the Weibull distribution is a generalized form of the Rayleigh PDF. The Rayleigh PDF of the wind speed which is depicted in Fig. 12.3, can be exhibited as follows PDFðvÞ ¼ where v is wind speed in m/s.
v c2
exp
"
v pffiffiffi 2c
2 #
ð12:3Þ
12
Multi-objective Optimal Reactive Power Dispatch …
481
The wind speed variation range is classified into intervals, which is named wind scenarios. The probability of each scenario can be calculated from the following equation. The probability of scenario s and the corresponding wind speed vs is calculated using the following equations. " # Zvf ;w v v 2 pffiffiffi pw ¼ dv exp c2 2c
ð12:4Þ
vi;w
1 vw ¼ pw
Zvf ;w
v
vi;w
v c2
" #! v 2 pffiffiffi dv exp 2c
ð12:5Þ
where, vw is the wind speed at w-th wind scenario, and vi;w ; vf ;w are the starting and the last points of wind speed’s interval at w-th scenario, respectively. Also, c is scaling parameter which is acquired by historical wind data. The characteristics curve of a wind turbine determines the correspondence between the available wind speed and generated wind power. A linearized characteristics curve is presented in Fig. 12.4 [38]. Using this curve, the predicted production power of the wind turbine for various wind speeds can be obtained using the following equation.
Pavl w
¼
8
Fkmin Fkmax
:
Fk Fkmin Fkmin Fk Fkmax Fk Fkmax
ð12:21Þ
where, k is index of objective functions, Fk is individual value of k-th conflicting ^k is normalized value of k-th objective function. objective function and F For the obtained Pareto optimal set, the best compromise solution can be selected using the min-max method described in [45]. In this method, for r-th Pareto optimal solution the minimum membership number ð^ lr Þ is obtained as follows: ^r ¼ minðlk Þ; 8k 2 XOF l
ð12:22Þ
where, XOF is set of conflicting objective functions. Now, the best compromise solution is that which has the maximum value of minimum membership number, as follows: _
l ¼ maxð^ lr Þ; 8r 2 XPOS
ð12:23Þ
where, XPOS is set of all optimal Pareto solutions of a multi-objective optimization problem.
12.3.4 Constraints 12.3.4.1
Equality Constraints (AC Power Balance Equations)
The feasible solution should assure the power flow equations in each scenario, which are represented mathematically in the following.
488
S.M. Mohseni-Bonab et al.
8 > < PGi ;s þ PWi ;s
PDi ;s ¼
> : QGi ;s þ QWi ;s þ QCi ;s
Vi;s Vj;s Gij cos hi;s hj;s þ Bij sin hi;s hj;s j2XBi P Vi;s Vj;s Gij sin hi;s hj;s QDi ;s ¼ Bij cos hi;s P
j2XBi
hj;s
ð12:24Þ
where, PGi ;s ; QGi ;s are active and reactive powers production of generator at bus i in scenario s, PWi ;s ; QWi ;s active and reactive power generations of wind farm at scenario s, PDi ;s ; QDi ;s real and reactive power demand of i-th bus in scenario s, and Gij ; Bij are real and imaginary parts of ij-th element of YBUS matrix (pu/radian).
12.3.4.2
Inequality Constraints
The active and reactive power output of generators and voltage magnitudes of all buses should be kept in the predefined ranges as follows. max Pmin Gi PGsl ;s PGi ; 8i ¼ sb; 8s 2 XS
ð12:25Þ
max Qmin Gi QGi ;s QGi ; 8i 2 XG ; 8s 2 XS
ð12:26Þ
Vimin Vi;s Vimax ; 8i 2 XB ; 8s 2 XS
ð12:27Þ
max min max where, Pmin Gi ; PGi are minimum and maximum value for active power, QGi ; QGi are minimum and maximum value for reactive power of generator at bus i, and Vimin =Vimax are minimum and maximum value for voltage magnitude of i-th bus. The power flowing from the branches is constrained to its maximum value as follows.
S‘;s Smax ; 8‘ 2 XL ; 8s 2 XS
ð12:28Þ
tmmin tm tmmax ; 8m 2 XT
ð12:29Þ
‘
where, ‘ is index of transmission lines, S‘;s is power flow of ‘ - th branch in scenario s, and Smax is maximum transfer capacity of line ‘. ‘ The tap amounts of tap changers are also restricted as follows.
where, m is index of tap changing transformers, and tmmin ; tmmax are minimum/maximum value for m-th tap changer. It is noteworthy that the reactive power output of VAR compensation devices are modeled as a multi-step compensation, i.e. a discrete variable is utilized for each VAR compensation node as follows, which determines the required steps for VAR injections.
12
Multi-objective Optimal Reactive Power Dispatch …
QCi ;s ¼ QbCi uCi ;s ; 8i 2 XC ; 8s 2 XS
489
ð12:30Þ
where, QCi ;s is reactive power compensation at bus i in scenario s, QbCi is VAR compensation capacity in each step at bus i, and uCi ;s is reactive power compensation step at bus i in scenario s. The reactive power compensation stages are restricted as follows max umin Ci uCi ;s uCi
8i 2 XC ; 8s 2 XS
ð12:31Þ
max where, umin Ci ; uCi are minimum and maximum value for reactive power compensation at bus i. Also, for the available active and reactive power outputs of wind farms, the following constraints should be satisfied
0 PWi ;s fWi;s PrWi ; 8i 2 XW ; 8s 2 XS
ð12:32Þ
max Qmin Wi QWi ;s QWi ; 8i 2 XW ; 8s 2 XS
ð12:33Þ
where, PrWi is wind farm rated capacity installed in bus i, fWi;s is percentage of wind max power rated capacity realized at scenario s in bus i, and Qmin are Wi ; QWi minimum/maximum value of reactive power produced by wind farm. In this chapter in line with references [35–37], the reactive power output of wind farms are related to the active power output as follows. (
1 Qmax Wi ¼ tan cos ðPFlg;i Þ PWi ;s Qmin tan cos 1 ðPFld;i Þ PWi ;s Wi ¼
ð12:34Þ
where, PFlg;i ; PFld;i are lag/lead power factor limits of the wind farms located at node i.
12.4
Scenario Generation and Two-Stage Stochastic Programming
In this chapter two-stage stochastic programming method is utilized for decision making in an uncertain environment. In this method, the decision variables are classified as “here and now” and “wait and see” variables [46]. The optimal values of “here and now” or “first stage” variables should be recognized before realization of scenarios. In other words, their values are scenario independent and are similar for all scenarios. In other words, the optimal values of “wait and see” or “second stage” variables should be considered after realization of the scenarios. In other words, their values are scenario dependent and may be different for different
490
S.M. Mohseni-Bonab et al.
scenarios. In the suggested SMO-ORPD problem the decision variables (DVs) are generator voltages, tap values of tap changing transformers and reactive power output of VAR compensators in the weak buses. As it is mentioned before, the set of control variables is categorized into two separated subsets, i.e. here and now and wait and see control variables. The set of here and now decision variables ðDVHN Þ are as follows: DVHN ¼
Vi ; 8i 2 XG tm ; 8m 2 XT
ð12:35Þ
where, tm is value of m-th tap changer setting. Also, the set of wait and see decision variables ðDVWS Þ are as follows. DWS
8 < QCi ;s ; ¼ PWi ;s ; : QWi ;s;
9 8i 2 XC ; 8s 2 XS = 8i 2 XW ; 8s 2 XS ; 8i 2 XW ; 8s 2 XS
ð12:36Þ
where, QCi ;s is reactive power compensation at bus i in scenario s.
12.5
Simulations on a Standard Test System
Simulations are carried out on the IEEE 30-bus test system. In order to show the effectiveness of the presented approach, several cases are studied as follows. (A) Deterministic optimization without wind farms (by neglecting the uncertainties of load and wind farms) (B) Deterministic optimization with the expected value of wind farms power output and mean value of load (by neglecting uncertainty) (C) Stochastic optimization with load and wind farms power generation uncertainties (uncertainty representation using scenario based approach). For the sake of comparison with available methods in literature, the VAR compensation devices are modeled as continuous control variables in case (A). While in cases (B) and (C) the VAR compensations are modeled with discrete steps as defined in previous section. In cases (B) and (C), it is assumed that a wind farm with 56 MW rated capacity is installed at bus 20.
12.5.1 Test System IEEE 30-bus system [47] includes 30 buses with 6 generator buses as represented in Fig. 12.6. Bus 1 is the slack bus and buses 2, 5, 8, 11 and 13 are PV buses whereas
12
Multi-objective Optimal Reactive Power Dispatch …
491
C 29
28
27 25
30 26
C
C
23
24 19
C 12
13 G
18
15
C
C 17
20 C
14 C 21
System description 30 Buses 41 Branches 21 Load sides 6 Thermal units 9 VAR compensator 4 Tab changer 1 Wind farms
16 22
11
9 C
G
10
1
3
4
G C
8 G
6 5
2 G
7
G
G
Fig. 12.6 One-line diagram of IEEE 30-bus test system
the remaining 24 buses are PQ buses. The network consists of 41 branches, 4 transformers and 9 capacitor banks. Four branches, 6–9, 6–10, 4–12 and 28–27 are under load tap changing transformers. The tap ratios are within the interval [0.9, 1.1]. Additionally, buses 10, 12, 15, 17, 20, 21, 23, 24 and 29 are selected as shunt VAR compensation buses. In each study we have considered two cases regarding the objective functions. As mentioned in Sect. 12.2, in MO-ORPD problem F1 with F2 and F1 with F3 are conflicting. Minimization is classified in two part PL/NPL and VD/NVD (Case I) and PL/ and LM/NLM (Case II). Figure 12.7 shows the case studies conducted in this chapter. For each case, at first the DMO-ORPD is solved for the sake of comparison with the existing methods, and then SMO-ORPD is solved to investigate the impact of uncertainties on the obtained results.
492
S.M. Mohseni-Bonab et al.
Fig. 12.7 Illustration of the studied cases
MO-ORPD
SMO-ORPD with WFs
DMO-ORPD with expected WFs
Case-II Objective functions: PL - LM
DMO-ORPD without WFs
SMO-ORPD with WFs
DMO-ORPD with expected WFs
DMO-ORPD without WFs
Case-I Objective functions: PL - VD
12.5.2 Case-I: Active Power Loss Versus Voltage Deviations (PL-VD) 12.5.2.1
Solving DMO-ORPD Without WFs
In this case, the Pareto front is attained for IEEE 30-bus test system without considering any uncertainty and without wind power integration. The VAR compensation devices are modeled by continuous variables for the sake of comparison with the previously published works in literature. The data of compensation limits are available in Appendix (Table 12.15). Table 12.1 summarizes the acquired Pareto solutions for this case. By using min-max fuzzy satisfying criterion, it is observed from Table 12.1 that the best compromise solution is Solution#16, with the maximum weakest membership function of 0.7734. The corresponding PL and VD are equal to 4.4438 MW and 0.0092, respectively. It is also noticeable that Solution#1 corresponds to the loss minimization case in Solution#1, only PL is minimized, and the minimum value of PL is obtained 4.2875 MW. The Pareto optimal front of the two objective functions is depicted in Fig. 12.8. For this solution, the optimal values of control variables are given in Table 12.2.
12.5.2.2
Solving DMO-ORPD with Expected WFs
According to Fig. 12.6, IEEE-30 bus is modified and it is presupposed that wind turbine is located in bus 20. In this part, by using Fig. 12.5, the expected value of wind turbine generation is taken account in simulation. Wind power capacity
12
Multi-objective Optimal Reactive Power Dispatch …
Table 12.1 Pareto optimal solutions for DMO-ORPD without WFs (Case-I)
493
#
PL (MW)
VD
l1
l2
^r l
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
4.2875 4.2882 4.2904 4.2944 4.3005 4.3092 4.3190 4.3293 4.3401 4.3515 4.3637 4.3767 4.3909 4.4064 4.4238 4.4438 4.4680 4.4999 4.5544 4.9774
0.0387 0.0367 0.0348 0.0328 0.0308 0.0288 0.0269 0.0249 0.0229 0.0210 0.0190 0.0170 0.0151 0.0131 0.0111 0.0092 0.0072 0.0052 0.0033 0.0013
1 0.999 0.9958 0.990 0.9811 0.9686 0.9543 0.9395 0.9238 0.9072 0.8896 0.8706 0.8501 0.8276 0.8024 0.7734 0.7384 0.6921 0.6132 0
0 0.0521 0.1047 0.1574 0.2101 0.2627 0.3154 0.3681 0.4207 0.4734 0.5260 0.5787 0.6314 0.6840 0.7367 0.7894 0.8420 0.8947 0.9473 1
0 0.0521 0.1047 0.1574 0.2101 0.2627 0.3154 0.3681 0.4207 0.4734 0.5260 0.5787 0.6314 0.6840 0.7367 0.7734 0.7384 0.6921 0.6132 0
Fig. 12.8 Pareto optimal front for DMO-ORPD without WFs (Case-I)
5 4.9
PL (MW)
4.8 4.7 4.6 X: 0.00916 Y: 4.444
4.5 4.4 4.3 4.2 0
0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
VD
amounted to twenty percent of the total load on the network is considered. Expected value of wind farm generation is calculated as: Pw;exp ¼
X ps fWi;s PrWi
s2XS
ð12:37Þ
494
S.M. Mohseni-Bonab et al.
Table 12.2 Optimal control variables for the best compromise solution (Solution#16) in (Case-I) Control parameters
Parameter
DMO-ORPD (without WF)
Generator parameters
Vg1 (pu) Vg2 (pu) Vg5 (pu) Vg8 (pu) Vg11 (pu) Vg13 (pu) Pg1 (MW) Qc10 (MVar) Qc12 (MVar) Qc15 (MVar) Qc17 (MVar) Qc20 (MVar) Qc21 (MVar) Qc23 (MVar) Qc24 (MVar) Qc29 (MVar) t6-9 t6-10 t4-12 t28-27
1.0295 1.0204 1.0093 1.0030 1.0148 1.0258 62.8438 0 0 3.8495 3.6873 2.6983 6.6481 1.7635 3.0855 2.2464 0.9706 1.1000 0.9812 0.9970
Compensation
Transformer tap changer
It is explained that, discrete steps for compensation devices is used during this section based on the data provided in appendix (Table 12.15). Various Pareto solutions for this part are provided in Table 12.3. In order to solve the multi-objective ORPD problem by e-constraint method, maximum and minimum values of the real power loss ðF1 Þ and voltage deviation ðF2 Þ are considered, which are equal to 3.8771 MW, 3.1651 MW, 0.0428 pu and 0.0012 pu, respectively. These extreme values are reached by maximizing and minimizing the objective functions of MO-ORPD individually. It means that in Solution#1 and #20 the objectives are minimizing the PL and VD, respectively. Among these optimal solutions, Solution#16 is minimizing both objectives, with the equal to 3.2983 MW PL and the VD of 0.0099. Pareto front of answers and control variables for BCS of this case are reported in Fig. 12.9 and Table 12.4, respectively.
12.5.2.3
Solving SMO-ORPD with WFs
In this case the load and wind power uncertainties are considered in the MO-ORPD using the previously described two stage stochastic programming approach.
12
Multi-objective Optimal Reactive Power Dispatch …
Table 12.3 Pareto optimal solutions for DMO-ORPD with expected WFs (Case-I)
495
#
PL (MW)
VD
l1
l2
^r l
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
3.1651 3.166 3.1696 3.1759 3.183 3.1904 3.1982 3.2064 3.215 3.2241 3.234 3.2445 3.2556 3.2681 3.2823 3.2983 3.3182 3.3453 3.3928 3.8771
0.0428 0.0406 0.0384 0.0362 0.0340 0.0318 0.0296 0.0275 0.0253 0.0231 0.0209 0.0187 0.0165 0.0143 0.0121 0.0099 0.0077 0.0055 0.0033 0.0012
1 0.9987 0.9937 0.9849 0.9749 0.9645 0.9535 0.9421 0.9300 0.9172 0.9032 0.8886 0.8729 0.8553 0.8355 0.8129 0.7850 0.7470 0.6801 0
0 0.0526 0.1052 0.1579 0.2105 0.2631 0.3158 0.3684 0.4210 0.4737 0.5263 0.5789 0.6316 0.6842 0.7368 0.7895 0.8421 0.8947 0.9474 1
0 0.0526 0.1052 0.1579 0.2105 0.2631 0.3158 0.3684 0.4210 0.4737 0.5263 0.5789 0.6316 0.6842 0.7368 0.7895 0.7850 0.7470 0.6801 0
Fig. 12.9 Pareto front of DMO-ORPD with WFs (Case-I)
4 3.9 3.8
PL (MW)
3.7 3.6 3.5 3.4
X: 0.00992 Y: 3.298
3.3 3.2 3.1 0
0.01
0.02
0.03
0.04
0.05
VD
The attained Pareto optimal solutions in this case are presented in Table 12.5. It is observed from this table that NPL varies from 3.4003 to 4.1891 MW, while NVD varies from 0.0431 to 0.0013, respectively. The Solution#1 corresponds to the NPL minimization case, where the minimum value of 3.4003 MW is obtained for NPL ,
496
S.M. Mohseni-Bonab et al.
Table 12.4 Optimal control variables for the best compromise solution (Solution#16) in DMO-ORPD with expected wind-(Case-I) Control parameters
Parameter
DMO-ORPD (with WF)
Generator parameters
Vg1 (pu) Vg2 (pu) Vg5 (pu) Vg8 (pu) Vg11 (pu) Vg13 (pu) Pg1 (MW) uc10 uc12 uc15 uc17 uc20 uc21 uc23 uc24 uc29 Pw19 (MW) Qw19 (MVar) t6-9 t6-10 t4-12 t28-27
1.0236 1.0161 1.0062 1.0015 1.0156 1.0248 48.2003 0 0 1 2 1 3 1 3 1 29.222 2.2632 0.9711 1.1000 0.9840 0.9936
Compensation (switching steps)
Wind farms parameters Transformer tap changer
whereas Solution#20 deals with the case of NVD minimization, in which the minimum value of NVD is 0.0431. It is observed from Table 12.5 that Solution#16 is the best compromise solution, with NPL equals to 3.5475 MW and NVD equals to 0.0101. Also, Fig. 12.10 depicts the obtained optimal Pareto front in this case. Table 12.6 summarizes the obtained optimal here and now control variables for the best compromise solution. Also, the optimal values of wait and see control variables are depicted in Figs. 12.11, 12.12 and 12.13, in all possible scenarios. Figure 12.11 shows the active power generation at the slack bus (bus 1) in all 15 scenarios. Figure 12.12 represents the active/reactive power output of the wind farm in all scenarios. The optimal amount of reactive power compensation steps are also given in Fig. 12.13.
12
Multi-objective Optimal Reactive Power Dispatch …
Table 12.5 Pareto optimal solutions for SMO-ORPD (Case-I)
497
#
NPL (MW)
NLM
l1
l2
^r l
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
3.4003 3.4188 3.4194 3.4213 3.4253 3.4322 3.4405 3.4492 3.4584 3.4681 3.4785 3.4897 3.5018 3.5152 3.5302 3.5475 3.5687 3.5949 3.6396 4.1891
0.0431 0.0409 0.0387 0.0365 0.0343 0.0321 0.0299 0.0277 0.0255 0.0233 0.0211 0.0189 0.0167 0.0145 0.0123 0.0101 0.0079 0.0057 0.0035 0.0013
1 0.9765 0.9758 0.9734 0.9683 0.9595 0.9490 0.9380 0.9264 0.914 0.9008 0.8867 0.8713 0.8543 0.8353 0.8133 0.7865 0.7533 0.6967 0
0 0.0526 0.1053 0.1579 0.2105 0.2632 0.3158 0.3684 0.4211 0.4737 0.5263 0.5790 0.6316 0.6842 0.7368 0.7895 0.8421 0.8947 0.9474 1
0 0.0526 0.1053 0.1579 0.2105 0.2632 0.3158 0.3684 0.4211 0.4737 0.5263 0.5790 0.6316 0.6842 0.7368 0.7895 0.7865 0.7533 0.6967 0
Fig. 12.10 Pareto front of SMO-ORPD (Case-I)
4.2 4.1
EPL (MW)
4 3.9 3.8 3.7 X: 0.01007 Y: 3.548
3.6 3.5 3.4 0
0.01
0.02
0.03
EVD
0.04
0.05
498
S.M. Mohseni-Bonab et al.
Table 12.6 Optimal values for here and now control variables at the best compromise solution (Solution#16) in Case-I Control parameters
Parameter
DMO-ORPD(without WF)
Generator parameters
Vg1 (pu) Vg2 (pu) Vg5 (pu) Vg8 (pu) Vg11 (pu) Vg13 (pu) t6-9 t6-10 t4-12 t28-27
1.0270 1.0193 1.0089 1.0043 1.0165 1.0267 0.9694 1.1000 0.9841 0.9933
Transformer tap changer
Active Power Generation (MW)
90 80 70 60 50 40 30 20 10 0 s1
s2
s3
s4
s5
s6
s7
s8
s9
s10 s11 s12 s13 s14 s15
Scenario Number
Fig. 12.11 Active power generation in slack bus (bus 1) in all scenarios (in MW)-(Case-I)
12.5.3 Case-II: Active Power Loss Versus Lmax, (PL-LM) 12.5.3.1
Solving DMO-ORPD Without WFs
In this case, PL and LM are considered as the two considered conflicted objectives. The Pareto front is obtained without considering any uncertainty. Table 12.7 summarizes the obtained Pareto solutions for this case. By using min-max fuzzy
12
Multi-objective Optimal Reactive Power Dispatch …
499
Wind Farm Active/ Reactive power output
40 35 30 25 20 Pw
15
Qw
10 5 0 s1 s2 s2 s3 s4 s5 s6 s7 s8 s9 s10 s11 s12 s13 s14 s15 -5
Scenario Number
Fig. 12.12 Active/reactive power output of wind farm (located at bus 20) in all scenarios (in MW and MVAR) - (Case-I)
5
4
Switching Step
Bus 10 Bus 12
3
Bus 15 Bus 17 Bus 20
2
Bus 21 Bus 23
1
Bus 24 Bus 29
0 s1
s2
s3
s4
s5
s6
s7
s8
s9 s10 s11 s12 s13 s14 s15
Scenario Number
Fig. 12.13 Switching steps in VAR compensation buses at different scenarios (Case-I)
satisfying method, it is evident from Table 12.7 that the best compromise solution is Solution#17, with the maximum weakest membership function of 0.7988. The corresponding PL and LM are equal to 4.6106 MW and 0.1199 pu, respectively. Pareto front of this case is available in Fig. 12.14. Table 12.8 summarizes the obtained control variables for the best compromise solution, Solution#17, of this case.
500
S.M. Mohseni-Bonab et al.
Table 12.7 Pareto optimal solutions for DMO-ORPD without WFs (Case-II)
#
PL (MW)
LM
l1
l2
^r l
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
4.2875 4.2879 4.2893 4.2918 4.2956 4.3012 4.3087 4.3187 4.3318 4.3475 4.3659 4.3869 4.4105 4.4403 4.4835 4.5413 4.6106 4.7219 5.0212 5.8934
0.1323 0.1315 0.1308 0.1300 0.1292 0.1284 0.1277 0.1269 0.1261 0.1253 0.1246 0.1238 0.1230 0.1222 0.1215 0.1207 0.1199 0.1191 0.1184 0.1176
1 0.9997 0.9989 0.9973 0.9949 0.9915 0.9868 0.9806 0.9724 0.9626 0.9512 0.9381 0.9234 0.9048 0.8779 0.8420 0.7988 0.7295 0.5431 0
0 0.0526 0.1053 0.1579 0.2105 0.2632 0.3158 0.3684 0.4210 0.4737 0.5263 0.5789 0.6316 0.6842 0.7368 0.7895 0.8421 0.8947 0.9474 1
0 0.0526 0.1053 0.1579 0.2105 0.2632 0.3158 0.3684 0.4210 0.4737 0.5263 0.5789 0.6316 0.6842 0.7368 0.7895 0.7988 0.7295 0.5431 0
Fig. 12.14 Pareto front of DMO-ORPD without WFs (Case-II)
6 5.8 5.6
PL (MW)
5.4 5.2 5 X: 0.1199 Y: 4.611
4.8 4.6 4.4 4.2 0.118
0.12
0.122 0.124 0.126 0.128
0.13
0.132
Lmax
12.5.3.2
Solving DMO-ORPD with Expected WFs
Similar to Case-I, in this case, expected value of wind power as an input power is calculated by Eq. 12.37. Table 12.9 shows the obtained 20 selected Pareto solutions for this case. Among this answers, Solution#17 has minimized both objective simultaneously with PL = 3.2751 and LM = 0.1146. Figure 12.15 depicts the
12
Multi-objective Optimal Reactive Power Dispatch …
501
Table 12.8 Optimal control variables for the best compromise solution (Solution#17) in DMO-ORPD without wind-Case-II Control parameters
Parameter
DMO-ORPD (without WF)
Generator parameters
Vg1 (pu) Vg2 (pu) Vg5 (pu) Vg8 (pu) Vg11 (pu) Vg13 (pu) Pg1 (MW) Qc10 (MVar) Qc12 (MVar) Qc15 (MVar) Qc17 (MVar) Qc20 (MVar) Qc21 (MVar) Qc23 (MVar) Qc24 (MVar) Qc29 (MVar) t6-9 t6-10 t4-12 t28-27
1.0366 1.0286 1.0182 1.0134 1.0110 1.0434 63.0106 8.2754 0 0 7.0185 4.0128 14.2677 2.9065 7.7563 0 1.0706 0.9000 0.9956 0.9719
Compensation
Transformer tap changer
Pareto front of this case. According to Fig. 12.15, it is clear that when wind power injected into the network, it substantially reduces network losses and increases network stability. Table 12.10 presents the obtained control variables for best solution of this case (Solution#17).
12.5.3.3
Solving SMO-ORPD with WFs
The uncertainty is modeled using scenario based approach for Case-II and results of Pareto solutions are presented in Table 12.11. Figure 12.16 shows the Pareto front for Case-II with considering load and wind uncertainties using scenario based approach. Similar to Case-I, here and now control variables are shown in Table 12.12 and wait and see control variables are shown in Figs. 12.17, 12.18 and 12.19. It should be noticed that, these variables are presented for BCS (Solution#16).
502 Table 12.9 Pareto optimal solutions for DMO-ORPD with expected WFs (Case-II)
S.M. Mohseni-Bonab et al. #
PL (MW)
LM
l1
l2
^r l
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
3.1651 3.1654 3.1662 3.1682 3.1707 3.1742 3.1786 3.1841 3.1905 3.1977 3.2056 3.2144 3.2242 3.235 3.2468 3.2597 3.2751 3.2944 3.3189 3.7027
0.1247 0.1241 0.1234 0.1228 0.1222 0.1215 0.1209 0.1203 0.1196 0.1190 0.1184 0.1178 0.1171 0.1165 0.1159 0.1152 0.1146 0.1140 0.1133 0.1127
1 0.9994 0.9979 0.9943 0.9895 0.9832 0.9749 0.9647 0.9528 0.9395 0.9247 0.9084 0.8901 0.8700 0.8481 0.8241 0.7955 0.7594 0.7139 0
0 0.0526 0.1053 0.1579 0.2105 0.2632 0.3158 0.3684 0.4210 0.4737 0.5263 0.5789 0.6316 0.6842 0.7368 0.7895 0.8421 0.8947 0.9473 1
0 0.0526 0.1053 0.1579 0.2105 0.2632 0.3158 0.3684 0.4210 0.4737 0.5263 0.5789 0.6316 0.6842 0.7368 0.7895 0.7955 0.7594 0.7139 0
Fig. 12.15 Pareto front of DMO-ORPD with WFs (Case-II)
3.7
PL (MW)
3.6 3.5 3.4 X: 0.1146 Y: 3.275
3.3 3.2 3.1 0.112
0.114
0.116
0.118
Lmax
0.12
0.122
0.124
12
Multi-objective Optimal Reactive Power Dispatch …
503
Table 12.10 Optimal control variables for the best compromise solution (Solution#17) in Case-II Control parameters
Parameter
DMO-ORPD (with WF)
Generator parameters
Vg1 (pu) Vg2 (pu) Vg5 (pu) Vg8 (pu) Vg11 (pu) Vg13 (pu) Pg1 (MW) uc10 uc12 uc15 uc17 uc20 uc21 uc23 uc24 uc29 Pw20 (MW) Qw20 (MVar) t6-9 t6-10 t4-12 t28-27
1.0371 1.0301 1.0204 1.0161 1.0136 1.0461 49.3914 0 0 1 2 1 4 1 3 1 29.2220 2.4846 1.0550 0.9000 0.9930 0.9761
Compensation (switching steps)
Wind Farms parameters Transformer tap changer
12.6
Discussions on the Results
12.6.1 Comparison of DMO-ORPD Performance with Pervious Literature In Tables 12.13 and 12.14 the obtained PL, VD and LM are compared with the results reported by some recently published algorithms. Table 12.13 shows the obtained PL and VD for Solution#1 (PL minimization), Solution#20 (VD minimization), and Solution#16 (compromise solution), whereas Table 12.14 give a comparison for PL and LM. Solution#1 (PL minimization), Solution#20 (LM minimization), and Solution#17 (compromise solution) are compared with recently published works. According to these tables, it can be observed that the obtained
504 Table 12.11 Pareto optimal solutions for SMO-ORPD) (Case-II)
S.M. Mohseni-Bonab et al. #
NPL (MW)
NLM
l1
l2
^r l
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
3.4815 3.4865 3.4881 3.4914 3.4918 3.4942 3.4989 3.5048 3.5118 3.5196 3.5279 3.5375 3.5485 3.5602 3.573 3.5873 3.6039 3.6214 3.6399 4.0646
0.1258 0.1251 0.1245 0.1239 0.1233 0.1227 0.1221 0.1215 0.1209 0.1204 0.1198 0.1193 0.1188 0.1183 0.1179 0.1175 0.1174 0.1172 0.1171 0.1161
1 0.9915 0.9887 0.9831 0.9824 0.9783 0.9702 0.9601 0.9481 0.9347 0.9204 0.9040 0.8851 0.8650 0.8431 0.8187 0.7901 0.7600 0.7283 0
0 0.0635 0.1264 0.1890 0.2513 0.3131 0.3745 0.4352 0.4952 0.5541 0.6117 0.6675 0.7208 0.7703 0.8143 0.8488 0.8658 0.8863 0.8942 1
0 0.0635 0.1264 0.1890 0.2513 0.3131 0.3745 0.4352 0.4952 0.5541 0.6117 0.6675 0.7208 0.7703 0.8143 0.8187 0.7901 0.7600 0.7283 0
Fig. 12.16 Pareto front of SMO-ORPD (Case-II)
4.2 4.1
EPL (MW)
4 3.9 3.8 3.7
X: 0.1175 Y: 3.587
3.6 3.5 3.4 0.116
0.118
0.12
0.122
ELmax
0.124
0.126
12
Multi-objective Optimal Reactive Power Dispatch …
505
Table 12.12 Optimal values for here and now control variables at the best compromise solution (Solution#16) in Case-II Control parameters
Parameter
DMO-ORPD (without WF)
Generator parameters
Vg1 (pu) Vg2 (pu) Vg5 (pu) Vg8 (pu) Vg11 (pu) Vg13 (pu) t6-9 t6-10 t4-12 t28-27
1.0151 0.9976 1.0083 1.0092 1.0246 1.0398 1.0455 0.9000 0.9821 0.9698
Transformer tap changer
Active Power Generation (MW)
200 180 160 140 120 100 80 60 40 20 0 s1
s2
s3
s4
s5
s6
s7
s8
s9
s10 s11 s12 s13 s14 s15
Scenario Number
Fig. 12.17 Active power generation in slack bus (bus 1) in all scenarios (in MW)-(Case-II)
solutions are superior to the previously reported ones like as gravitational search algorithm (GSA) [17], differential evolutionary algorithm [15], quasi-oppositional teaching learning based optimization (TLBO) algorithm [25] and chaotic parallel vector evaluated interactive honey bee mating optimization (CPVEIHBMO) [26].
506
S.M. Mohseni-Bonab et al.
Wind Farm Active/ Reactive power output
60 50 40 30 Pw
20
Qw
10 0 s1
s2
s3
s4
s5
s6
s7
s8
s9 s10 s11 s12 s13 s14 s15
-10 -20
Scenario Number
Fig. 12.18 Active and reactive power output of wind farm (located at bus 20) in all scenarios (in MW and MVAR)—(Case-II)
5
4
Switching Step
Bus 10 Bus 12
3
Bus 15 Bus 17 Bus 20
2
Bus 21 Bus 23 Bus 24
1
Bus 29
0 s1
s2
s3
s4
s5
s6
s7
s8
s9 s10 s11 s12 s13 s14 s15
Scenario Number
Fig. 12.19 Switching steps in VAR compensation buses at different scenarios (Case-II)
12
Multi-objective Optimal Reactive Power Dispatch …
507
Table 12.13 Comparison of the obtained PL and VD in DMO-ORPD (without WFs), with the published methods PL Minimization
VD Minimization
PL (MW) Compromise solution
Proposed CPVEIHBMO [26] HBMO [26] QOTLBO [25] NSGA-II [27] DE [15] SPMGSO [27] GSA [17]
4.2875 4.37831 4.40867 4.5594 5.137 4.555 5.123 4.5143 PL minimization
4.9774 4.994831 5.2092 6.4962 5.686 6.4755 5.96 4.9752 VD minimization
4.4438 5.3243 5.535 5.2594 – – – – VD (pu) Compromise solution
Proposed CPVEIHBMO [26] HBMO [26] QOTLBO [25] NSGA-II [27] DE [15] SPMGSO [27] GSA [17]
0.0387 0.67352 0.87364 1.9057 0.6443 1.9989 0.73986 0.8752
0.0013 0.198756 0.2106 0.0856 0.1789 0.0911 0.1438 0.2157
0.0092 0.7397 0.87664 0.121 – – – –
12.6.2 Impact of Wind Energy on MO-ORPD Problem Figures 12.20 and 12.21 summarize the obtained results of DMO-ORPD with and without WFs and SMO-ORPD for both Case-I and Case-II. According to Fig. 12.20a, it is clear that by installing wind farms on the system, real power losses (PL in deterministic and EPL in stochastic model) are reduced considerably. It is also observed from this figure that in the case of SMO-ORPD, the EPL is higher than the case of DMO-ORPD with expected wind. Besides, Fig. 12.20b depicts the obtained values for VD and EVD in both Case-I and II. It is evidently observed from this figure that, installation of wind farm leads to deterioration of voltage deviations, especially in the case of SMO-OPRD. This is mainly because of fluctuations of wind farm output power, in different scenarios. Also, Fig. 12.21 shows the obtained values of PL (and EPL) along with Lmax (and ELM) in Case-II. According to Fig. 12.21a active power losses are decreased when the wind farm is considered, both in DMO-ORPD (with wind farm) and SMO-ORPD. Besides, according to Fig. 12.21b the voltage stability of system increases in the presence of wind farm, since the Lmax and ELM reduced in both DMO-ORPD (with wind farm) and SMO-ORPD cases.
508
S.M. Mohseni-Bonab et al.
Table 12.14 Comparison of the obtained PL and LM in DMO-ORPD (without WFs), with the published methods PL minimization
LM minimization
PL (MW) Compromise solution
Proposed CPVEIHBMO [26] HBMO [26] QOTLBO [25] MOCIPSO [24] MOPSO [24] DE [15] RGA [22] CMAES [22] GSA [17]
4.2875 4.7831 4.40867 4.5594 5.174 5.233 4.555 4.951 4.945 4.5143 PL minimization
5.8934 6.6501 6.66 5.2554 5.419 5.528 7.0733 5.0912 5.129 6.6602 LM minimization
4.6106 5.3243 5.5352 5.2594 5.232 5.308 – – – – LM (pu) Compromise solution
Proposed CPVEIHBMO [26] HBMO [26] QOTLBO [25] MOCIPSO [24] MOPSO [24] DE [15] RGA [22] CMAES [22] GSA [17]
0.1323 0.141 0.12101 0.1263 0.1273 0.12664 0.1317 0.5513 0.13965 0.13944
0.1176 0.116 0.111 0.1147 0.1242 0.1141 0.1192 0.1246 0.1386 0.1382
0.1199 0.1163 0.1203 0.1254 11.821 0.12191 – – –
The obtained results imply the positive impact of wind power generation on the voltage stability enhancement and decreasing system real power losses. Also, it can be observed that the installed wind farm has little effects on the voltage deviations.
12
Multi-objective Optimal Reactive Power Dispatch …
Fig. 12.20 Comparison of the obtained results of Case-I in different conditions, a PL and EPL (MW), b VD and EVD (pu)
509
6
DMO-ORPD (without WFs)
(a)
DMO-ORPD (with WFs)
5
SMO-ORPD (with WFs)
4 3 2 1 0 PL minimization 0.05 0.045
VD minimization
(b)
Compromise solution DMO-ORPD (without WFs) DMO-ORPD (with WFs)
0.04
SMO-ORPD (with WFs)
0.035 0.03 0.025 0.02 0.015 0.01 0.005 0 PL minimization
Fig. 12.21 Comparison of the obtained results of Case-II in different conditions, a PL and EPL (MW), b LM and ELmax
7
VD minimization
Compromise solution
(a) DMO-ORPD (without WFs) DMO-ORPD (with WFs)
6
SMO-ORPD (with WFs)
5 4 3 2 1 0 PL minimization
LMax minimization
Compromise solution
(b) 0.135
DMO-ORPD (without WFs) DMO-ORPD (with WFs)
0.13
SMO-ORPD (with WFs)
0.125 0.12 0.115 0.11 0.105 0.1 PL minimization
LMax minimization
Compromise solution
510
12.7
S.M. Mohseni-Bonab et al.
Conclusions
In this chapter the stochastic multi-objective optimal reactive power dispatch (SMO-ORPD) problem in a wind integrated power system is studied by considering the uncertainties of system load and wind power generations. For decision making under the above uncertainties two-stage stochastic optimization model is utilized. In the multi-objective optimization framework real power losses, voltage deviation and voltage stability improvement index (L-index) are optimized simultaneously. The e-constraint method is utilized to solve multi-objective optimization problem. The obtained results in the deterministic case are compared with the recently applied intelligent search-based algorithms and it is found that the suggested method can find better solutions for both objective functions in this case. In the stochastic case, a comprehensive set of decision variables including here and now and wait and see control variables are obtained. The offered SMO-ORPD model is verified using the IEEE 30-bus test system. The numerical results substantiate that in the presence of wind power generation, the expected value of active power losses and L-index are reduced in comparison with the deterministic case. This confirms the positive influence of wind power generation on the reduction of system losses.
Appendix See Table 12.15.
Table 12.15 The data of VAR Compensation devices Bus No.
10 12 15 17 20 21 23 24 29
DMO-ORPD without WFs [26, 47] Qmin Ci (MVAR)
Qmax Ci (MVAR)
DMO-ORPDwith WFs SMO-ORPD with WFs ICmax QCi (MVAR) ICmin i i
0 0 0 0 0 0 0 0 0
36 36 36 36 36 36 36 36 36
0 0 0 0 0 0 0 0 0
1 3 1 2 1 4 1 3 1
1 1.5 1.5 2 0.5 2 1 1.5 0.5
12
Multi-objective Optimal Reactive Power Dispatch …
511
References 1. B. Zhao, C. Guo, Y. Cao, A Multiagent-Based Particle Swarm Optimization Approach for Optimal Reactive Power Dispatch, IEEE Transactions on Power Systems, vol. 20, pp. 1070–1078, 2005. 2. C. Dai, W. Chen, Y. Zhu, X. Zhang, Seeker Optimization Algorithm for Optimal Reactive Power Dispatch, IEEE Transactions on Power Systems, vol. 24, pp. 1218–1231, 2009. 3. A. Khorsandi, A. Alimardani, B. Vahidi, S. Hosseinian, Hybrid Shuffled Frog Leaping Algorithm and Nelder-Mead Simplex Search for Optimal Reactive Power Dispatch, IET Generation, Transmission & Distribution, vol. 5, pp. 249–256, 2011. 4. M. Ghasemi, S. Ghavidel, M.M. Ghanbarian, A. Habibi, A New Hybrid Algorithm for Optimal Reactive Power Dispatch Problem with Discrete and Continuous Control Variables, Applied Soft Computing vol. 22, pp. 126–40, 2014. 5. M. Ghasemi, M.M. Ghanbarian, S. Ghavidel, S. Rahmani, E. Mahboubi Moghaddam, Modified Teaching Learning Algorithm and Double Differential Evolution Algorithm for Optimal Reactive Power Dispatch Problem: A Comparative Study, Information Sciences, vol. 278, pp. 231–249, 2014. 6. M. Martinez Rojas, A. Sumper, O. Gomis Bellmunt, A. Sudria Andreu, Reactive Power Dispatch in Wind Farms Using Particle Swarm Optimization Technique and Feasible Solutions Search, Applied Energy, vol. 88, pp. 4678–4686, 2011. 7. R. Mallipeddi, S. Jeyadevi, P.N. Suganthan, S. Baskar, Efficient Constraint Handling for Optimal Reactive Power Dispatch Problems, Swarm and Evolutionary Computation, vol. 5, pp. 28–36, 2012. 8. C.M. Huang, S.J. Chen, Y.C. Huang, H.T. Yang, Comparative Study of Evolutionary Computation Methods for Active-Reactive Power Dispatch, IET Generation, Transmission & Distribution, vol. 6, pp. 636–645, 2012. 9. A. Rajan, T. Malakar, Optimal Reactive Power Dispatch Using Hybrid Nelder-Mead Simplex Based Firefly Algorithm, International Journal of Electrical Power & Energy Systems, vol. 66, pp. 9–24, 2015. 10. R.P. Singh, V. Mukherjee, S. Ghoshal, Optimal Reactive Power Dispatch by Particle Swarm Optimization with an Aging Leader and Challengers, Applied Soft Computing, vol. 29, pp. 298–309, 2015. 11. M.H. Sulaiman, Z. Mustaffa, M.R. Mohamed, O. Aliman, Using the Gray Wolf Optimizer for Solving Optimal Reactive Power Dispatch Problem, Applied Soft Computing, vol. 32, pp. 286–292, 2015. 12. D. Thukaram, G. Yesuratnam, Optimal Reactive Power Dispatch in a Large Power System with AC-DC and FACTS Controllers, Generation, Transmission & Distribution, IET, vol. 2, pp. 71–81, 2008. 13. A. Rabiee, M. Vanouni, M. Parniani, Optimal Reactive Power Dispatch for Improving Voltage Stability Margin Using a Local Voltage Stability Index, Energy Conversion and Management, vol. 59, pp. 66–73, 2012. 14. C. Dai, W. Chen, Y. Zhu, and X. Zhang, Reactive Power Dispatch Considering Voltage Stability with Seeker Optimization Algorithm, Electric Power Systems Research, vol. 79, pp. 1462–1471, 2009. 15. A. Ela, M. Abido, S. Spea, Differential Evolution Algorithm for Optimal Reactive Power Dispatch, Electric Power Systems Research, vol. 81, pp. 458–464, 2011. 16. A. Khazali, M. Kalantar, Optimal Reactive Power Dispatch Based on Harmony Search Algorithm, International Journal of Electrical Power & Energy Systems, vol. 33, pp. 684–692, 2011. 17. S. Duman, Y. Sonmez, U. Guvenc, N. Yorukeren, Optimal Reactive Power Dispatch Using a Gravitational Search Algorithm, IET Generation, Transmission & Distribution, vol. 6, pp. 563–576, 2012.
512
S.M. Mohseni-Bonab et al.
18. J.M. Ramirez, J.M. Gonzalez, T.O. Ruben, An Investigation about the Impact of the Optimal Reactive Power Dispatch Solved by DE, International Journal of Electrical Power & Energy Systems, vol. 33, pp. 236–244, 2011. 19. E.M. Soler, E.N. Asada, G.R. Da Costa, Penalty-Based Nonlinear Solver for Optimal Reactive Power Dispatch with Discrete Controls, IEEE Transactions on Power Systems, vol. 28, pp. 2174–2182, 2013. 20. M. Abido, J. Bakhashwain, Optimal VAR Dispatch Using a Multi-Objective Evolutionary Algorithm, International Journal of Electrical Power & Energy Systems, vol. 27, pp. 13–20, 2005. 21. L. Zhihuan, L. Yinhong, D. Xianzhong, Non-Dominated Sorting Genetic Algorithm-II for Robust Multi-Objective Optimal Reactive Power Dispatch, Generation, Transmission & Distribution, IET, vol. 4, pp. 1000–1008, 2010. 22. S. Jeyadevi, S. Baskar, C. Babulal, M. Willjuice Iruthayarajan, Solving Multi-Objective Optimal Reactive Power Dispatch Using Modified NSGA-II, International Journal of Electrical Power & Energy Systems, vol. 33, pp. 219–228, 2011. 23. A. Saraswat, A. Saini, Multi-Objective Optimal Reactive Power Dispatch Considering Voltage Stability in Power Systems Using HFMOEA, Engineering Applications of Artificial Intelligence, vol. 26, pp. 390–404, 2013. 24. G. Chen, L. Liu, P. Song, Y. Du, Chaotic Improved PSO-Based Multi-Objective Optimization for Minimization of Power Losses and L Index in Power Systems, Energy Conversion and Management, vol. 86, pp. 548–560, 2014. 25. B. Mandal, P.K. Roy, Optimal Reactive Power Dispatch Using Quasi-Oppositional Teaching Learning Based Optimization, International Journal of Electrical Power & Energy Systems, vol. 53, pp. 123–134, 2013. 26. A. Ghasemi, K. Valipour, A. Tohidi, Multi-Objective Optimal Reactive Power Dispatch Using a New Multi-Objective Strategy, International Journal of Electrical Power & Energy Systems, vol. 57, pp. 318–334, 2014. 27. B. Zhou, K. Chan, T. Yu, H. Wei, J. Tang, Strength Pareto Multi-Group Search Optimizer for Multi-Objective Optimal VAR Dispatch, IEEE Transactions on Industrial Informatics, vol. 10, pp. 1012–1022, 2014. 28. Z. Hu, X. Wang, G. Taylor, Stochastic Optimal Reactive Power Dispatch: Formulation and Solution Method, International Journal of Electrical Power & Energy Systems, vol. 32, pp. 615–621, 2010. 29. S. M. Mohseni Bonab, A. Rabiee, S. Jalilzadeh, B. Mohammadi Ivatloo, and S. Nojavan, “Probabilistic Multi Objective Optimal Reactive Power Dispatch Considering Load Uncertainties Using Monte Carlo Simulations, Journal of Operation and Automation in Power Engineering, vol. 3(1), pp. 83–93, 2015. 30. S. M. Mohseni Bonab, A. Rabiee, B. Mohammadi Ivatloo, Load Uncertainty Analysis in Multi Objective Optimal Reactive Power Dispatch Considering Voltage Stability, Majlesi Journal of Energy Management, vol. 4(2), pp. 23–30, 2015. 31. A. Brooke, D. Kendrick, A. Meeraus, GAMS Release 2.25: A User’s Guide: GAMS Development Corporation Washington, DC, 1996. 32. P.E. Gill, W. Murray, M.A. Saunders, SNOPT: An SQP Algorithm for Large-Scale Constrained Optimization, SIAM Journal on Optimization, vol. 12, pp. 979–1006, 2002. 33. The GAMS Software Website, http://www.gams.com/dd/docs/solvers/sbb.pdf, 2013. 34. A. Soroudi, T. Amraee, Decision Making under Uncertainty in Energy Systems: State of the Art, Renewable and Sustainable Energy Reviews, vol. 28, pp. 376–384, 2013. 35. A. Rabiee, A. Soroudi, B. Mohammadi Ivatloo, M. Parniani, Corrective Voltage Control Scheme Considering Demand Response and Stochastic Wind Power, IEEE Transactions on Power Systems, vol. 29, pp. 2965–2973, 2014. 36. A. Soroudi, B. Mohammadi Ivatloo, A. Rabiee, Energy Hub Management with Intermittent Wind Power, in Large Scale Renewable Power Generation, Springer, pp. 413–438, 2014.
12
Multi-objective Optimal Reactive Power Dispatch …
513
37. A. Rabiee, A. Soroudi, Stochastic Multiperiod OPF Model of Power Systems with HVDC-Connected Intermittent Wind Power Generation, IEEE Transactions on Power Delivery, vol. 29, pp. 336–344, 2014. 38. A. Soroudi, A. Rabiee, A. Keane, Stochastic Real-Time Scheduling of Wind-Thermal Generation Units in an Electric Utility, IEEE System Journal, 2014. 39. R.D. Zimmerman, C.E. Murillo Sanchez, D. Gan, A MATLAB Power System Simulation Package, 2005. 40. S. Wen, H. Lan, Q. Fu, D. Yu, L. Zhang, Economic Allocation for Energy Storage System Considering Wind Power Distribution, IEEE Transactions on power Systems, vol. 30(2), pp. 644–52, 2015. 41. C.A. Canizares, Voltage Stability Assessment: Concepts, Practices and Tools, Power System Stability Subcommittee Special Publication IEEE/PES, thunderbox.uwaterloo. ca/*claudio/claudio.html#VSWG, 2002. 42. F. Jabbari, B. Mohammadi Ivatloo, Static Voltage Stability Assessment Using Probabilistic Power Flow to Determine the Critical PQ Buses, Majlesi Journal of Electrical Engineering, vol. 8, pp. 17–25, 2014. 43. H. Xiong, H. Cheng, H. Li, Optimal Reactive Power Flow Incorporating Static Voltage Stability Based on Multi-Objective Adaptive Immune Algorithm, Energy Conversion and Management, vol. 49, pp. 1175–1181, 2008. 44. R. Raghunatha, R. Ramanujam, K. Parthasarathy, D. Thukaram, Optimal Static Voltage Stability Improvement Using a Numerically Stable SLP Algorithm, for Real Time Applications, International Journal of Electrical Power & Energy Systems, vol. 21, pp. 289–297, 1999. 45. X. Wang, Y. Gong, C. Jiang, Regional Carbon Emission Management Based on Probabilistic Power Flow with Correlated Stochastic Variables, IEEE Transactions on Power Systems, vol. 30, pp. 1094–1103, 2014. 46. M. Alipour, B. Mohammadi Ivatloo, K. Zare, Stochastic Risk-Constrained Short-Term Scheduling of Industrial Cogeneration Systems in the Presence of Demand Response Programs, Applied Energy, vol. 136, pp. 393–404, 2014. 47. R.D. Zimmerman, C.E. Murillo Sanchez, R.J. Thomas, MATPOWER: Steady-State Operations, Planning, and Analysis Tools for Power Systems Research and Education, IEEE Transactions on Power Systems, vol. 26, pp. 12–19, 2011.
Part III
Challenges, Solutions and Applications in AC Power Systems
Chapter 13
Self-excited Induction Generator in Remote Site Ezzeddine Touti, Remus Pusca, J. Francois Brudny and Abdelkader Chaari
Abstract This chapter is devoted to the analysis of a self-excited asynchronous generator working in autonomous generation mode. It proposes a method of detection of the generator operating points and the control procedures which makes it possible to maintain quasi constant the frequency values. These strategies prevent the disengagement of the generator and use a Thyristor Controlled Reactor to regulate the reactive power. The control strategies present the study of the frequency in the steady case by the use of an appropriate single phase equivalent circuit. The space vector formalism is also used to study the transitory and steady output signals. At the end of the chapter, one will discuss the experimental validation of theoretical and numerical results.
13.1
Introduction
During the last decades, due to technological and industrial developments which impose the continuous demand of energy and when the cost of different types of energy increase, the efforts were continued to look for other solutions to diversify alternative energy sources. In this context, the wind energy seems to be a promising alternative source which can accompany the classical energy sources [1, 2]. The use
E. Touti (&) University of Northern Border, Arar, Saudi Arabia e-mail: [emailprotected] R. Pusca J.F. Brudny LSEE, University of Artois, Bethune, France e-mail: [emailprotected] J.F. Brudny e-mail: [emailprotected] A. Chaari Taha Hussein, University of Tunis, Tunis, Tunisia e-mail: [emailprotected] © Springer International Publishing AG 2017 N. Mahdavi Tabatabaei et al. (eds.), Reactive Power Control in AC Power Systems, Power Systems, DOI 10.1007/978-3-319-51118-4_13
517
518
E. Touti et al.
of wind turbines is an attractive solution because they can provide energy at a competitive price and mainly it does not cause environmental contamination [3]. The work presented in this chapter treats the study of a three-phase Self-excited Induction Generator (SEIG) generating power in remote site. The strategy presented allows the regulation of the reactive power. Consequently it enables the maintaining of the frequency at quasi constant values. Including Induction Machine (IM) in the wind generator, in the cases of limited electric power demand, becomes more and more expanded which explain their increased use in the electrical energy generation [3–6]. The IM is widely used because of its robustness, reliability, low cost and little maintenance [7–9]. Nevertheless, these types of generators require an external supply on reactive power needed for the machine magnetization and to establish the remnant rotating magnetic flux wave [10, 11]. In the case of Induction Generator (IG) connected at the grid, the IM use current drawn from the grid to creates the magnetic field. In remote site, the IG can operate in autonomous mode only when capacitors are connected at its stator terminals to produce the necessary reactive power for machine but also for its load. The Self-excited Induction Generator (SEIG) has been discovered for a few decades and was experienced at several remote sites, but the wave quality is not optimized. Also, one encounters difficulties such as frequency and voltage control [12, 13]. These difficulties are due to the inverse nature of the problem when no theory is available to describe the system. Indeed in this working mode the voltage, the frequency and the rotor sleep are unknown. In order to overcome these difficulties and to increase robustness of regulation system this chapter presents a control law established for a resistive-inductive load supplied by a SEIG and which can be considered as a preliminary study for an electrical motors load. When a ‘real’ load is presented the difficulty consist in adapting the control low, because during the regulation an important transient state appears. In this regard, it is assumed that L will be maintained constant and that the voltage regulation will be realized acting only on R load. This chapter presents the theoretical background of output voltage and frequency regulation for autonomous induction generators. It covers the analysis, modeling and simulation of an isolated self-excited induction generator working in remote site. Indeed, we are interested in the studying of a system able to vary the C capacitor when the SEIG supplies a resistive-inductive load (R–L). It suffices to parallel connect a fixed capacitor CM and a regulating inductive load (LR) connected to the isolated network via a dimmer. This system will control the reactive power consumption variation. The total reactive power is delivered by fixed capacitor CM and can be variable and so adjustable continuously by means of the TCR and the regulating inductive load. One treats, in the first section, the steady state analysis of IM. This study is based on a single phase equivalent circuit definition while considering the concept of voltage source. One will discuss subsequently the developed frequency control law. The operating points of the SEIG are fixed such that they give rated output at the rated conditions of voltage, current and speed. The objective of the various
13
Self-excited Induction Generator in Remote Site
519
developments presented is to determine the exact operating points, and for that it is proposed to use as mathematical solution an iterative (Newton-Raphson) method or a graphical approach. The self-excited induction generator modeling based on space phasor formalism is treated in the second section. Subsequently, one presents the analysis of the control law effectiveness in the steady and transient state cases. In a standalone SEIG, the capacitors connected across the SEIG terminals can supply the reactive power requested by the SEIG as well as the load. In the next part of this chapter, is presented the reactive power control strategy and the operating intervals imposed on the dimmer to stabilize the frequency. The dimmer allows the capacitors value variation across the machine terminals.
13.2
Proposed Autonomous Configuration
Figure 13.1 presents the global SEIG system configuration proposed for operating in remote site. Considering the wind as a prime mover, this asynchronous generator, which works in autonomous mode, will provide electrical power to supply a R–L load. In this configuration the capacitor bank allows to provide the reactive power needed for generator magnetization [14–17] as well as the load. This section starts by studying the induction machine and its modeling. To establish the model of the asynchronous machine, we adopt the following simplifying hypotheses: – The air gap has a uniform thickness by neglecting the slotting effects and saturation effects – The winding resistances does not vary with temperature,
Fig. 13.1 Global diagram configuration of an isolated SEIG
520
E. Touti et al.
Fig. 13.2 Representation of the machine statoric and rotoric windings
– The iron losses are neglected. – It is admitted that the magnetomotive force, which is created by each phase of the stator and rotor, has a sinusoidal spatial distribution along the air gap. These assumptions lead to consider a constant inductance. However, the mutual inductances between the stator and rotor windings are variable according to a sinusoidal law and depend on the electrical angle of their magnetic axes. Figure 13.2 represents the asynchronous machine in the electrical space (real variables reference). This figure shows a three phase n a fictitious winding. They are distributed as the rotor windings of the reel coil in the objective of taking account of the self-excitation phenomenon.
13.2.1 Space Vector Concept Space vector is used in order to overcome the difficulties of the working with three variables by simplifying the mathematical formulation through only two variables [18]. Hence, a view of the rotational dynamics is improved using the space vector formalism. Consequently, one joins at three variables x1 , x2 and x3 which belong to the assembly of real numbers, a complex number called vector of direct and inverse components. In a fixed reference, one can express this vector, denoted x, using the given relationship
13
Self-excited Induction Generator in Remote Site
521
Fig. 13.3 Changing of reference frame
x ¼ xd þ jxq ¼ ð2=3Þ 1 þ a þ a2 ½x1 x2 x3 t
ð13:1Þ
where a ¼ ej2p=3 is a rotational operator. One will consider a moving reference with rotational angle h. This vector presented in Fig. 13.3 is given using the following equation denoted X ¼ XD þ jXQ ¼ x e X
jh
ð13:2Þ
In the case of the study of a time derivative, following equation must be used e
jh
dx þ dX ¼ ðjx0 ÞX dt dt
ð13:3Þ
13.2.2 Induction Generator Model The real machines, with their different kinds of windings and geometries, are too complex to lend itself to an analysis taking into account their exact configurations, hence the need to develop a model whose behavior is as close as possible to that of the original [19, 20]. In this context, it is essential to use a simple model to facilitate digital implementation and minimize the simulator computation time. Through a numerical implementation of the SEIG model based on the space vector, one can study its steady and transient state. This study concerns a three-phase induction machine with p pole pair cage rotor. One will confuse the rotor and stator spatial reference (d r and d s ) with the rotor and stator phase 1 axes. The asynchronous generator must have a rotor residual magnetic field [21, 22], to produce the stator voltage. Let us consider a fictitious n a winding to create this
522
E. Touti et al.
magnetic field during the SEIG start-up step. These coils of n a turns are considered supplied by iaq DC currents (q = 1, 2 or 3). These all currents are null after the start-up step. The stator space vector variables referred to d s are denoted xs . Also, one will denote xr the rotor space vector variables defined relatively to d r . The stator space vector variables defined relatively to d r are labelled x0s . One can deduce x0s from xs using variable change: x0 s ¼ xs expð jhÞ where: h ¼ h0 þ x0 t represents the spatial angular gap between d r and d s , also x0 ¼ pX0 where X0 is the machine rotor angular speed. The asynchronous generator is characterized by the voltage equations written in the rotor reference frame which are expressed by the following system 8 0s 0s þ jx0 u 0s < v ¼ r si0s þ dtd u r vr ¼ r ri r þ dtd u ð13:4Þ : a a v ¼ r ai a þ dtd u The per phase winding resistances, in system (13.4), are represented by r. Whereas the fluxes linked by the rotor, the stator and the fictitious windings are 0s , u r and u a . They are given by labeled u 8 r ¼ Lr ð1 þ kr Þir þ M rsi0s þ M raia
s2 x4 RLCA0 þ s2 x2 ½LE00 RA0 þ sx2 LLs2 r 0r x2 RLCD0 > > > < þ RD0 þ ðR þ r s ÞLr 0r2 ¼ 0 ð13:18Þ 2 4 00 4 2 2 s 0 > s x LA sx LB þ s x r E þ sx2 Ls2 Rr0r > > > : x2 LD00 þ Rr s r 0r2 ¼ 0 The constants A00 , B, D00 , E00 are expressed by
8 A00 ¼ Ls2 ð1 þ kr ÞfLs ½1 ð1 þ ks Þð1 þ kr Þ > > < B ¼ Ls2 RCr 0r D00 ¼ r 0r2 fLs ð1 þ ks Þ þ r s RC g > > : 00 E ¼ Ls2 ð1 þ kr Þ2 ½R þ r s
r s RCð1 þ kr Þg ð13:19Þ
To obtain the values of the quantities A′, D′ and E′, one will considers that r s ¼ 0 in the expression of A00 , B and E00 . The electromagnetic torque Te of IM can be expressed from the cross product !s !s Te ¼ k E ðsÞ E ðrÞ
ð13:20Þ
where k ¼ 3p=xX s . The RMS voltage at the output of the generator is given by RLx V s ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi I s ðLxÞ2 þ R2 ðLCx2 1Þ2
ð13:21Þ
Hence the expression of the electromagnetic torque Ce ¼
3P SX s3 r 0r ½ðLxÞ2 þ R2 ðLCx2 1Þ2 V s2 xX s ðr 0r2 þ S2 Ls2 ð1 þ kr Þ2 x2 ÞRLx2
which allows to express V s versus Te
ð13:22Þ
528
E. Touti et al.
RL Vs ¼ s L
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Te x r 0r2 þ s2 Ls2 ð1 þ kr Þ2 x2 3p s½L2 x2 þ R2 ðLCx2 1Þ2 r 0r
ð13:23Þ
13.3.2 Identification of IM Operating Points The analysis of the system (13.18) can conclude that x and s depend only on the load (R, L), the capacitor C and the machine parameters. Therefore, if one imposes these parameters, the self-excited induction generator will be locked on fixed x and s independently of Pw values. In the objective of developments illustration, the studied generator is characterized by rated values given in appendix. When necessary, one uses the lower index “rat”. For practical tests, one connects the stator windings in star, to prevent the machine from high destructive transients which occurs during Pw and load variations, but also to evade magnetic saturation. The R, C and L shall be chosen such a way that the electrical reference of the Operating Point (OP), indicated with the lower index “ref”, may decreases from the rated pffiffiffi values by a 3 ratio. To determine the xex and sex exact values, one must take into account the parasitic elements. In order to solve system (13.18), two methods are used: an iterative (Newton-Raphson) and graphical methods.
13.3.2.1
Graphical Operating Point Determination
Starting with the first relationship of system (13.18), one can obtain M1 x4 þ N1 x2 þ P1 ¼ 0
ð13:24Þ
where M1 , N1 and P1 are expressed by 8 < M1 ¼ LRCA0 s2 N ¼ ðLE00 RA0 Þs2 þ LLs2 r 0r s : 1 P1 ¼ RD0 þ LðR þ r s Þr 0r2
RLCD0
ð13:25Þ
Since M1 is negative, one will consider only the positive x value because the quantity under the radical is greater than unity. The root of (13.24) can be given by x1 ¼
1=2 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi N1 1 þ 1 4M1 P1 =N12 =2M1
ð13:26Þ
13
Self-excited Induction Generator in Remote Site
529
Similarly, solutions of the second equation of system (13.18) x22ð1or2Þ are given by x2ð1or2Þ ¼
N2
1=2 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 1 1 4M2 P2 =N2 =2M2
ð13:27Þ
where 8 < M2 ¼ LsðA00 s BÞ 0 N ¼ r s E s2 þ Ls2 Rr0r s : 2 s 0r2 P2 ¼ Rr r
LD00
ð13:28Þ
In Eq. (13.27) one will associate the sign + before the radical with the lower index 1 between brackets and sign − with the lower index 2. One indicates that M2 has a positive value only for 0 [ s [ s , given that s ¼ B=A. Consequently, x2ð1Þ exists only in this margin of s changes however x2ð2Þ exists for s range between −l and 0 as it is shown in Fig. 13.7 plotted for R0 ,C0 and L0 (s ¼ 35:8%). As the OPs are given by the intersections of x1 with x2ð1Þ and x2ð2Þ , Fig. 13.7 shows that x2ð2Þ do not intervene in the OP determination. There are only two OPs, tied to x2ð1Þ which will be taken into account. The first OP ‘A’ with coordinates sex ¼ 6:03 % and xex ¼ 313:2 rad:s 1 , is placed in the x2ð1Þ characteristic flattened part situated around xmin where the changes of x are limited [24]. When the slip values tend to smin then x corresponds to xmin given by x2min
Fig. 13.7 Operating points for R0 ,C0 and L0 load
¼
2 s 0 smin r E þ smin Ls2 Rr 0r LD00 smin L½smin A00 B
ð13:29Þ
530
E. Touti et al.
Now we will consider the second equation with unknown s to find smin , we will have
s1;2 ¼
x4 LB
RLs2 r0r x2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðRLs2 r0r x2 x4 LBÞ2 4½LA00 x4 þ r s E 0 x2 :½ x2 LD00 þ r s r 0r2 R 2LA00 x4 þ 2x2 E 0 r s
ð13:30Þ
For the s existence, it is necessary that the quantity under the radical is positive or zero. So we obtain smin ¼
LBx2min RLs2 r 0r 2LA00 x2min þ 2r s E 0
ð13:31Þ
which gives s2min Ls2 Rr 0r LA00 þ r s E0 LB þ smin 2A00 D00 L2 þ D00 BL2 ¼ 0
ð13:32Þ
where xmin is considered the minimum angular frequency. The roots of this equation are given by smin
" A00 D00 L ¼ s2 0r 00 1 L Rr A þ BE 0 r s
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi# B½Ls2 Rr 0r A00 þ r s E0 B 1 A002 D00 L
ð13:33Þ
The area, considered as stable operating zone, is limited on the left taking into account the machine energetic performances. Whereas the OP ‘B’ is placed far from the stable zone and possesses high x and s values which can destroy the SEIG energetic performance.
13.3.2.2
Evolution of Operating Point for Variation of the L Load
To distinguish the L effect on the OPs, one has presented in Fig. 13.8 various cases for L values such as 0.17 H (L0 ), 1, 10 and 100 H. The case of L ¼ 100 H may be considered practically as a resistive load. From system (13.18), one deduces that R ¼ R0 , however C should have respectively the values C0 , 32:52 lF, 22:4 lF and 21:37 lF. As already indicated that, irrespective of L value, x2ð2Þ never intervenes in the OP determination. Therefore, one does not take into account x2ð2Þ in the next few analyses. Furthermore, it is noted that xex increases with L however sex possesses quasi constant values. Also, one can note that the margin for x nearly constant increases as L increases. This behavior can be benefit if there are no constraints on the energetic performance of the machine.
13
Self-excited Induction Generator in Remote Site
531
Fig. 13.8 Impact of L on the operating points positions, a L = 0.17 H, b L = 0.55 H, c L = 1 H, d L = 100 H
13.3.2.3
Newton-Raphson Numerical Solution
The numerical analysis is performed using the equations system (13.18) denoted by g1 and g2 respectively for the first and the second equation where the unknown are x and s. 8 > g1 ¼ s2 x4 RLCA0 þ s2 x2 ½LE00 RA0 þ sx2 LLs2 r 0r > > > < x2 RLCD0 þ RD0 þ ðR þ r s ÞLr 0r2 ð13:34Þ 2 4 00 > g2 ¼ s x LA sx4 LB þ s2 x2 r s E0 > > > : þ sx2 Ls2 Rr 0r x2 LD00 þ Rr s r 0r2
Using the iterative Newton-Raphson method based on the successive use of the Jacobian for the regulation of the variables adjustment pitch, it is possible to find a numerical solution of the nonlinear static system (13.34) [25–27]. The approximation of the gi (i = 1, 2) functions by a Taylor development stopped at the first order around the initial condition x0 ðx0 ¼ 314; s0 ¼ 1Þ to a zero target for these functions, allows to deduce the necessary variation of the variables x and s.
532
E. Touti et al.
Fig. 13.9 Flowchart of the iterative method of Newton-Raphson
The calculation is redone as g1 and g2 continue to exceed calculation accuracy fixed in advance. The algorithm of the Newton-Raphson method is summarized by the flowchart presented by Fig. 13.9. To stop the algorithm, a convergence test which usually comprises a double control is established: The first concerns a limited number of iterations and the second concerns the accuracy. This second test could be expressed as
13
Self-excited Induction Generator in Remote Site
533
maxðjgi jÞ er
ð13:35Þ
where er is a tolerance computation fixed by the operator (er = 0.001).This approach allowed to find xex ¼ 315:6 rad/s and sex ¼ 6%.
13.4
Impact of the C Variation on the Frequency
Figure 13.7 revealed the possibility of stabilizing the SEIG output frequency for an operating point defined at the bottom of the curve. Therefore, it is necessary to choose the load R–L and the capacitor C required for the wind turbine operating at this point. Given that the curve is flattened around this point, limited variations of the R load around 19% will keep quasi constant the frequency by changing the capacitor C. Table 13.1 gives calculation and measurement example. There are five cases presented in Table 13.1. The first was carried out for R0 = 111 X and C0 = 87.5 lF. This is the central SEIG operating point (Fig. 13.7). In the following, some variations in the R load will be tested around this central operating point « A » which is characterized by s = –5.7% and x = 312.7 rad/s (f = 49.8 Hz). These changes move the operating point to right and left of the central operating point. Regarding the second case, we choose R = 132 X (variation of 19%) that corresponds to the machine discharge characterized by the operating point whose cordinates are s = –4.8% and x = 315.8 rad/s (a displacement to the right, f = 50.3 Hz, Df = 0.5 Hz). The capacitor value should be 83.12 lF, but considering our capacitor bank we used C = 83 lF. For comparison, we have also tested (test 3) a variation of R value (R = 132 X) without changing C (C = C0 = 87.5 lF). We notice a slightly greater change in the stator angular frequency (x = 307 rad/s, f = 48.8 Hz, Df = 1 Hz). A fourth pair (R, C) was considered (a displacement to the left) such as R = 86 X (variation of 22.5%) and C = 95.5 lF which corresponds to an increase in the load. The change in C provides a smaller variation of x (x = 311.5 rad/s, f = 49.6 Hz, Df = 0.2 Hz). This case will be compared to another fifth case where C is kept constant when R varies (R2 = 86 X and C2 = C0 = 87.5 lF). It is also noted a Table 13.1 Changes in angular frequency, slip and voltage versus load at constant power L = 170 mH Pw = 1884 W
Experimental values f x C R rad/s Hz lF X
Theoretical values x s% V volts rad/s
f Hz
s%
C variable
111 132 86 132 86
−5.7 −4.8 −7.7 −4.9 −7.6
49.8 50.5 49.09 49.1 51.6
−6.03 −5.08 −7.8 −5.08 −7.8
C = 87.5 lF = constant
87.5 83 95.5 87.5 87.5
312.7 315.8 311.5 307 322.3
49.8 50.3 49.6 48.8 51.3
224.5 245.4 192.3 244.2 192.8
313.2 317.4 308.3 308.7 324.2
534
E. Touti et al.
greater variation of the stator angular frequency x (x = 322.3 rad/s, f = 51.3 Hz, Df = 1.5 Hz). According to the results predicted by the developed model, Table 13.1 confirms that when considering a R–L load we can improve the stator frequency stability at the wind generator output by changing C. In fact, we obtain a frequency change near zero (1%) for R variation about 19%. This change is more important (3%) if C has not changed. When the load is purely resistive, this variation may become much more important (11%). Other tests have been carried out for other operating points. They confirmed identically the test already presented. The theoretical values shown in Table 13.1 are obtained from graphical method (paragraph II) or numerically computed based on the analytical study already developed. It is noted that these values are close to those experimentally obtained. By maintaining constant tan q, one can obtain an operating while keeping the frequency at quasi constant values without need of any control loop of this variable. Here q is the argument of the whole load and capacitor which are placed in parallel and connected to the SEIG stator.
13.4.1 Load Variation at Constant C The angular frequencies variations for a load R change about 22% are shown in Fig. 13.10a, b. The reduction of R values from reference point R0 ¼ 111 X can displace the OPs to the curved area in the direction of increasing speed. This can cause a significant increase in the stator voltage frequency. This is due to the fact that C value is kept constant while varying R. In a second test we will increase the R value while keeping C constant. We can note a significant frequency change. By against, the
Fig. 13.10 a Decrease in R value at constant C = 87.5 lF, b Increase in R value at constant C = 87.5 lF
13
Self-excited Induction Generator in Remote Site
535
increases in R values are unlimited unless the machine energetic performance and voltage stability are lost.
13.4.2 R Load Variation at Variable C In this case we will accompany to any R variation a change in the value of C. Figure 13.11 shows the considered study with two tests of load tilting around the initial operating point (R0 and C0). The strategy proposed for controlling the frequency needs the definition of pairs (R, C) allowing to reach the OP in the controlled zone. With the considered machine parameters, we lead to the point A (Fig. 13.7) corresponding to R0 = 111 X and C0 = 87.5 lF. The next stage is to act on R and therefore C while respecting the following law which is deduced from tan q ¼ constant. C¼
ðR þ R0 ðLC0 x2rat LRx2rat
1ÞÞ
ð13:36Þ
In Fig. 13.11 is noted that the change of the pair (R, C) while respecting the Eq. (13.36), the frequency is kept quasi constant during the wind generator operating. We also note a small change in frequency (Df 1.5%) if R varies about 20% from R0. Through against, if C is kept constant while changing R close to 20%, the frequency variation can exceed 3.5%. Figure 13.10a, b show the changes of the OP
Fig. 13.11 Tilting from R0 = 111 X and C0 = 87.5 lF to (R2 = 86 X, C2 = 95.5 lF) and (R1 = 132 X, C1 = 83 lF)
536
E. Touti et al.
positions for only R load variations (L and C kept constant). Figure 13.11 presents the results of the R variations with variable C. It is noted that the three characteristics are confused in the stable region shown in Fig. 13.7 when applying the f frequency control law.
13.5
Reactive Power Control
We consider the SEIG operation in remote site. By mean of the analytical method developed in Sect. 13.3, it is possible to find a couple R, C adapted for a given inductive load L. This study leads to some operating points around the point A within the stable zone (Fig. 13.7). When the power Pw varies, the voltage regulation may be ensured by changing a resistive R load. This change concerns only the consumers who are not priority and they can be disconnected or introduced in accord with the wind power changes. The R load changes must be accompanied by a modification of the reactive power provided by the capacitor CM in order to stabilize the frequency. One can assure the reactive power changes using thyristor controlled reactor. This dimmer mounting is controlled by the firing angle w as in Fig. 13.12 [28]. Let us consider that regulating inductive load LR have also an internal resistance RLR . We talk about a load Z R . By means of this system, a reactive power consumer can be disconnected or introduced in the isolated grid. According to Eq. (13.36), the R load variation is accompanied by a capacitor change C ¼ Ceq ðwÞ through changing the firing angle w. In this case, the equivalent reactive power available on the isolated grid is defined by Qeq ¼
U 2 xCeq ðwÞ
ð13:37Þ
In practice, the evolution of w depends on the previously calculated values in order to ensure a relative voltage stability obtained by coupling or decoupling the non-priority loads with maximal variation about DR ±20%. This change keeps quasi constant the voltage and the frequency without the need of electronic power converters. Fig. 13.12 Regulating load in parallel with the capacitor CM
13
Self-excited Induction Generator in Remote Site
537
13.5.1 Variation of the Capacity To regulate the reactive power, a three-phase system using a balanced load Z R is considered. The CM capacitor modification is realized using balanced regulating impedance Z R which is constituted by a resistance RLR in series with an inductance LR . This load is supplied through a three-phase dimmer as it is shown in Fig. 13.13a. The dimmer choice is due to its low cost and its ability to compensate reactive power. Let us denote / the regulating impedance argument. To simplify the analytical study, it has been presented the v01 ðhÞ voltage curve and that of i01 ðhÞ corresponding current in Fig. 13.13b. It is assumed that the i03 ðhÞ third phase current will be null for an angle h ¼ h1 . The thyristor firing angle will be noted w. We will start the analytical study of the system, cited above, by considering a single phase dimmer model (Fig. 13.14a) which feed an inductive load Z R . In order to characterize the voltage Vq0 ðhÞ (q = 1, 2, 3) applied to the regulating load, a Fourier analysis is carried out. This development provides the RMS fun0 damental voltage value Vq1 given by [29]
Fig. 13.13 a Dimmer connected to regulating load, b Voltage and current characteristics: 1st mode
Fig. 13.14 a Single-phase dimmer, b current and voltage waveforms
538
E. Touti et al.
0 Vq1
2 1 Vq 4 2 ¼ p
1 cos 2ðb wÞ þ ðb 2 þ ðb wÞðsin 2w
312 wÞ2 þ 5 sin 2bÞ
ð13:38Þ
where Vq is the RMS value of phase to neutral voltage applied to the dimmer input and b is the angle for which the current becomes zero in phase q. For xLR RLR the reactive power absorbed by LR is determined by 0
QLR ¼ Vq1 Iq1 sin /1
ð13:39Þ
h i12 0 0 0 Iq1 = R2LR þ ðLR xÞ2 ¼ Vq1 =jZ R j ¼ Vq1
ð13:40Þ
with
0 where Iq1 is the RMS value of the fundamental current and /n ¼ a tanðn:LR x=RLR Þ with n = 1. Let us assume that LR consumes a part of reactive power which is produced by the total capacitor CM , the capacitance of capacitor removed CðwÞ will be expressed as
2 1 1 4 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 C ðwÞ ¼ xp2 R2LR þ ðLR xÞ2
1 cos 2ðb wÞ þ ðb 2 þ ðb wÞðsin 2w
3 wÞ2 þ 5 sin /1 sin 2bÞ
ð13:41Þ
Taking into account the Eq. (13.41), the total capacitor CM is modified by means of the dimmer. The capacitance of the equivalent capacitor Ceq at the SEIG terminals is expressed as Ceq ðwÞ ¼ CM
ð13:42Þ
CðwÞ
In the three phase operating case, the reactive power absorbed can be monitored by the dimmer control while / < w < 5p/6. Depending on the load parameters ðLR ; RLR Þ, two operating modes of the dimmer are possible [30]. – Mode 1: 2 or 3 thyristors conduct simultaneously (u < w < wlim) – Mode 3: 2 or 0 thyristors conduct simultaneously (wlim < w < 5p/6) The transition from mode 1 to mode 3 occurs for a w > wlim. The wlim value can be obtained from the following relationship sinðwlim
/
4p Þ¼ 3
sinðwlim
/Þ
1 2e p=3q 2 e p=3q
ð13:43Þ
13
Self-excited Induction Generator in Remote Site
539
where q ¼ LR x=RLR . Therefore, it is possible that the dimmer can operates in one of two modes. Nevertheless, the mode 3 corresponds to a small variation in the current RMS value which causes a discontinuity for the current weak value. For a given reactive power variation at the SEIG output, it is preferred that the dimmer operates in the mode 1 because there is a possible larger variation range of CðwÞ and more stability was noticed. The Eq. (13.41) enables to compute this capacitance if b is well known. Considering q = 1 (Fig. 13.14b) and taking into account the system symmetry for the mode 1, the current cancellation angle b can be expressed as b ¼ pþw
a¼
2p þ h1 3
ð13:44Þ
where a ¼ w þ p=3 h1 (Fig. 13.14b) and h1 represents the current cancellation angle in the third phase i03 ðhÞ obtained from sinðh1
/
4p ðh1 Þe 3
wÞ=q
¼
sinðw
/Þ
1 2e p=3q 2 e p=3q
ð13:45Þ
0 However it is noted that for h 2 ½w b the Eq. (13.41) considers only Vq1 applied to the load. Nevertheless, during this interval, the signals are not sinusoidal at the dimmer output. In the case of three-phase system, the RMS fundamental voltage value V10 is given by the following relationship
V10 ¼
3V p 1 2p 1 ððh1 w þ sin 2ðh1 þ Þ þ sin 2wÞ2 2p 3 2 3 2 1 2p 2 1=2 þ ðcos 2w cos 2ðh1 þ ÞÞ Þ 4 3
ð13:46Þ
where V is the phase to neutral voltage RMS value applied to the dimmer input. The h1 angle is determined by the numerical resolution of the Eq. (13.45). The reactive power consumed by the regulating inductive load is defined by QLR ¼ 3VI10 sin /1
ð13:47Þ
/1 ¼ a tanðLR x=RLR Þ
ð13:48Þ
where /1 is given by
and the I10 current RMS value is deduced from (13.46) according to the relationship I10 ¼ V10 =
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
R2LR þ ðLR xÞ2
ð13:49Þ
540
E. Touti et al.
Fig. 13.15 Variation of C versus the firing angle w
This reactive power will be subtracted from that provided by the CM capacitor which is computed by the following equation Q CM ¼
3V 2 xCM
ð13:50Þ
Consequently, one deduces the C capacitor (removed) expression depending on the thyristor firing angle CðwÞ ¼
3 sin /1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2xp R2LR þ ðLR xÞ2
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi p 1 2p 1 1 2p sin 2ðh1 þ Þ þ sin 2wÞ2 þ ðcos 2w cos 2ðh1 þ ÞÞ2 ðh1 w þ 3 2 3 2 4 3
ð13:51Þ
Using Eq. (13.51), the C variation curve versus the angle w is plotted in Fig. 13.15. In this figure, the C theoretical values are plotted with solid line. They are compared with those experimental which are represented by dashed line. The curve plotted in continuous thin line represents the linearization of the characteristic C ¼ f ðwÞ. It is noted that there is a variation interval where the three characteristics are confused. That makes it possible a linear variation of C (regulating area) using a dimmer.
13.5.2 Dimmer Operating Interval Considering a three-phase operation, the reactive power absorbed by Z R can be controlled by the dimmer as soon as w is between /1 and 5p/6. For a considered Z R ðRLR xLR Þ to increase the system robustness, mode 1 was chosen because it allows a better operating stability and enables an important variation range of CðwÞ given by Eq. (13.51). The voltage RMS value at the Z R terminals is given by the following expression
13
Self-excited Induction Generator in Remote Site
541
Fig. 13.16 Voltage at the dimmer output versus the w firing angle
V0 ¼ V
1 3 h þ 2ðh1 2 4p
pi wÞ þ sin 2w þ sinð2h1 þ Þ 3
1=2
ð13:52Þ
Figure 13.16 shows the voltage evolution versus the firing angle w for the operation first mode. This evolution given by Eq. (13.52) (solid line) is experimentally validated (dashed line) for V ¼ 224:5 V. It shows a voltage linear variation for a clearly defined interval of w called “regulating area”. The analytical study leads to wlim ¼ 119:5 but to avoid a significant drop in voltage, it is necessary to limit the w variations although wlim is not reached. Taking into account the Eqs. (13.51) and (13.52), constrained by the already mentioned regulating zone in Figs. 13.15 and 13.16, the implemented firing angle in the dSPACE system corresponds to the linear interval of the two characteristics. For this operating interval, it is possible to compensate the R variations respecting the low given by the Eq. (13.36).
13.6
Voltage Collapse and Self-excitation Practical Procedure
The self-excitation of IM is generally realized by increasing the rotor speed in presence of a remnant magnetization and it is easier to be obtained without R– L load connected. After self-excitation and obtained of the stable state of the generator output voltage, the connection of the load must be progressively introduced. However at instant of load connection or during a sudden load change a voltage collapse can appear. This case is shown in Fig. 13.17. Here the load requiring more active power is varied without regulation of the reactive power. We note in simulation case (Fig. 13.17a) a quickly decrease of the generator voltage. In experimental test (Fig. 13.17b) the transient state is accompanied by more severe oscillations around of a new decreased voltage reference value before the
542
E. Touti et al.
Fig. 13.17 Voltage collapse during sudden load variation, a simulation result, b experimental result
voltage collapse and the machine dropping out. These voltage oscillations are dangerousness for IM or for the load and their amplitude must be controlled. However, this relatively long transitory interval which is visible in experimental tests show that the system has a good stability and the SEIG voltage collapse appear more letter. So it can be easier to readjust the load in order to avoid the machine dropping out. Therefore in order to decrease the voltage oscillation amplitude, the load variation must be accompanied by capacitors with appropriate values. Generally voltage collapse is accompanied by a dropping out of the machine and the self-excitation of IM can be assured only in the presence of a residual magnetic field. If IG work in remote site, a specific procedure is necessary for return it at the self-excitation step. A specific procedure is proposed in [31] where the reactive energy is given by a capacitor banc charged by a DC source. To make a success of the self-excited procedure the IG rotational speed must be increased at estimated calculated value taking in consideration the load. Another procedure which we propose is shown in Fig. 13.18. Here the load is disconnected by breaker 1 and a DC supply is applied at the stator windings in two steps: the first by closing breakers 1–2, 3–4 and the second by changing 3–4 with
Fig. 13.18 Practical startup procedure
13
Self-excited Induction Generator in Remote Site
543
4–5. The current must be limited at the generator nominal value In. It is also possible after its disengagement to supply the machine with AC power system in order to operate as an asynchronous motor mode.
13.7
Conclusion
The work presented in this chapter treated a numerical model of a three phase Self Excited Induction Generator which operates in remote site. This wind generator supplies an inductive-resistive load connected in parallel with capacitor banks. This latter provides the reactive power required by the SEIG as well as the load. A control strategy, based on the reactive power change was studied. This study was conducted while respecting the low tan q ¼ constant from which, any change of C is realized by adjusting the thyristors firing angle of the dimmer which feed a regulating load. The first part treats the use of space vector formalism in order to develop the SEIG model. The second part deals with the steady state studies by means of a single phase equivalent circuit. Another section of this chapter discusses the frequency control law. The operating points of the SEIG are fixed such that they give rated output at the rated conditions of voltage, current and speed. In the last part of this chapter, is presented the reactive power control strategy and the operating intervals imposed on the dimmer for stabilizing the frequency. The dimmer allows obtaining a variable capacitor across the machine terminals.
Appendix The parameters of Induction Machine used in this study are: 380 V=660 V, pffiffiffi 7:3 A=4:2 A, cos u ¼ 0:8, p ¼ 2, f=50 Hz, 1420 rpm, 3 kW, ns =nr ¼ 2, r r ¼ 3 X, Lr ¼ 267 mH, M sr ¼ 377 mH, r a ¼ 10 X, M sa ¼ 100 mH; la ¼ 5 mH, p ffiffiffiffiffiffiffiffiffi a La ¼ 18:7 mH, k ¼ 0:26, M ra ¼ La Lr ¼ 70:6 mH, Ls ¼ 534 mH, r s ¼ 8:66 X, r 0r ¼ 6 X, ls ¼ 24:24 mH, l0r ¼ 36:36 mH, ks ¼ 0:0454, kr ¼ 0:068, Tf ¼ 1:3 Nm, s LR ¼ 215 mH, CM ¼ 113 lF, 380 V=660 V, 7:3 A=4:2 A, Vref ¼ 223 V, s 1 sref ¼ 5:4%, Iref ¼ 2:42 A, xref ¼ 314:2 rad s .
References 1. G.M. Joselin Herbert, S. Iniyan, E. Sreevalsan, S. Rajapandian, A Review of Wind Energy Technologies, Renewable and Sustainable Energy Reviews, 11, pp. 1117–1145, 2007. 2. A.H.M.A. Rahim, M. Ahsanul Alam, M.F. Kandlawala, Dynamic Performance Improvement of an Isolated Wind Turbine Induction Generator, Computers and Electrical Engineering, 35, pp. 594–607, 2009.
544
E. Touti et al.
3. G.K. Kasal, B. Singh, Voltage and Frequency Controllers for an Asynchronous Generator-Based Isolated Wind Energy Conversion System, IEEE Trans. Energy Convers., vol. 26, no. 2, pp. 402–416, 2011. 4. Y.Y. Deng, K. Blok, K. van der Leun, Transition to a Fully Sustainable Global Energy System, Energy Strategy Reviews, 1, pp. 109–121, 2012. 5. M. Godoy Simoes, F.A. Farret, Renewable Energy Systems: Design and Analysis with Induction Generators, CRC PRESS, USA, 2004. 6. P. Gipe, Evaluating the Technology - What Works and What Doesn’t, Chelsea Green Publishing Company, Wind Power for Home & Business, 1993. 7. S. Kumar, R. Narayan, Effect of Capacitive VAR on Performance of Three-Phase Self-Excited Induction Generator, International Journal of Emerging Technology and Advanced Engineering, vol. 2, issue 12, pp. 253–258, December 2012. 8. Sh. Vadhera, K.S. Sandhu, Constant Voltage Operation of Self-Excited Induction Generator Using Optimization Tools, International Journal of Energy and Environment, issue 4, vol. 2, pp. 191–198, 2008. 9. H. Kumar, N. Kamal, Steady State Analysis of Self-Excited Induction Generator, International Journal of Soft Computing and Engineering, vol. 1, issue 5, pp. 248–253, November 2011. 10. Sh. Boora, Analysis of Self-Excited Induction Generator under Balanced or Unbalanced Conditions, Int. J. on Electrical and Power Engineering, vol. 1, no. 3, pp. 59–63, 2010. 11. A.L. Alolah, M.A. Alkanthal, Optimization Based Steady State Analysis of Three Phase SEIG, IEEE Trans. Energy Convers., vol. 15, no. 1, pp. 61–65, 2000. 12. T. Ahmed, O.H. Noro, E.M. Nakaoka, Terminal Voltage Regulation Characteristics by Static Var Compensator for a Three Phase Self-Excited Induction Generator, IEEE Trans. Ind. Appl., vol. 40, no. 4, pp. 978–988, 2004. 13. A.L. Alolah, M.A. Alkanthal, Optimization Based Steady State Analysis of Three Phase SEIG, IEEE Trans. Energy Convers., vol. 15, no. 1, pp. 61–65, 2000. 14. A.E. Kalas, M.H. Elfar, S.M. Sharaf, Particle Swarm Algorithm-Based Self-Excited Induction Generator Steady State Analysis, The Online Journal on Electronics and Electrical Engineering (OJEEE), vol. 3, no. 1, pp. 369–373, 2013. 15. A. Nesba, R. Ibtiouen, O. Touhami, Dynamic Performances of Self-Excited Induction Generator Feeding Different Static Loads, Serbian Journal of Electrical Engineering, vol. 3, no. 1, pp. 63–76, June 2006. 16. S. Devabhaktuni, S.V. Jayaram Kumar, Performance Analysis of Wind Turbine Driven Self-Excited Induction Generator with External Rotor Capacitance, International Journal of Advanced Engineering Sciences and Technologies, vol. 10, no. 1, pp. 1–6, 2011. 17. L. Louze, A.L. Nemmour, A. Khezzar, M.E. Hacil, M. Boucherma, Cascade Sliding Mode Controller for Self-Excited Induction Generator, Renewable Energies Revue, vol. 12, no. 4, pp. 617–626, 2009. 18. R. Romary, J.F. Brudny, Chapter on: Harmonic Torques of Electrical AC Electrical Drives, Second Edition of the Industrial Electronics Handbook, CRC Press & IEEE Press, Part II, Chapter 11, Editors: B.M. Wimamowski, J.D. Irwin, ISBN: 9781439802854, ISBN 10: 1439802858, pp. 10.1–10.27, February 2011. 19. M. Radic, Z. Stajic, N. Floranovic, Performance Characteristics of a Three-Phase Self-Excited Induction Generator Driven by Regulated Constant Speed Turbine, Automatic Control and Robotics, vol. 11, no. 1, pp. 57–67, 2012. 20. R.J. Harrington, F.M.M. Bassiouny, New Approach to Determine the Critical Capacitance for Self-Excited Induction Generators, IEEE Transactions on Energy Conversion, vol. 13, no. 3, pp. 244–249, 1998. 21. I. Boldea, Variable Speed Generator, The Electric Generator Handbook, Edition Tailor & Francis Group, USA, pp. 1–30, 2006. 22. A. Kheldoun, L. Refoufi, Dj. Eddine Khodja, Analysis of the Self-Excited Induction Generator Steady State Performance Using a new Efficient Algorithm, Electric Power Syst. Res., 86, pp. 61–67, 2012.
13
Self-excited Induction Generator in Remote Site
545
23. J.F. Brudny, R. Pusca, H. Roisse, Wind Turbines Using Self-Excited Three-Phase Induction Generators: An Innovative Solution for Voltage-Frequency Control, Eur. Phys. J. Appl. Phys., vol. 43, pp. 173–187, 2008. 24. E. Touti, R. Pusca, J.P. Manata, J.F. Brudny, A. Chaari, Enhancement of the Voltage and Frequency at the Self-Excited Induction Generator Outputs by Adjusting Terminal Capacitor, 9th International Conference on Industrial Power Engineering, Bacau, Romania, pp. 199–208, 22–24 May 2014. 25. S.P. Singh, B. Singh, M.P. Jain, Comparative Study on the Performance of a Commercially Designed Induction Generator with Induction Motors Operating as Self-Excited Induction Generators, IEE Proceedings - C, vol. 140, no. 5, pp. 374–380, September 1993. 26. S.S. Murthy, O.P. Malik, A.K. Tandon, Analysis of Self-Excited Induction Generator, Proc. Inst. Elect. Eng. C, vol. 129, no. 6, pp. 260–265, November 1982. 27. J.J. Grainger, W.D. Stevenson, Power System Analysis, ISBN 0-07-061293-5, McGraw-Hill, Inc., USA, pp. 342–356, 1994. 28. E. Touti, R. Pusca, A. Chaari, Dimmer Control for Voltage Transient Performance Improvement in Isolated Wind Turbine, Wulfenia Journal, vol. 20, no. 4, Klagenfurt, Austria, pp. 108–117, April 2013. 29. E. Touti, R. Pusca, J.P. Manata, J.F. Brudny, A. Chaari, Active and Reactive Power Regulation in Remote Site Equipped with Self-Excited Induction Generator, 12th International Conference on Science and Techniques of Automatic Control and Computer Engineering, STA’2011, Tunisia, pp. 593–607, 2011. 30. Ch. Rombaut, G. Seguier, R. Bausiere, Power Converters - Alternative Conversion, vol. 2, TEC& DOC, ISBN 2-85206-316-6, 1986. 31. J.F. Brudny, H. Roisse, G. Peset, Practical Approaches for the Study of Wind Machine Self-Excited Induction Generators, International Conference Wind Energy and Remote Regions, Magdalen Islands, pp. 1–10, 19–21 October 2005.
Chapter 14
Communications for Electric Power System Maaruf Ali and Nicu Bizon
Abstract This chapter is an overview on Communications applied for the Electric Power Systems. Thus, in the first section of this chapter, the Standards for Electric Power Systems Communications are briefly shown in order to understand the communication infrastructure requirements for the Smart Grids. The layers of the Smart Grid Network are (1) Power Grid, (2) Smart Grid and (3) Application. This chapter is focused on the Smart Grid layer, which has three primary functions to accomplish in real-time the requests of both consumers and suppliers based on Communications Technologies for Smart Grid Metering. So, Smart Grid Metering is an important interface to be implemented and integrated with Smart Grids based on technologies such as Fibre Optic Communication, x Digital Subscriber Line/Loop, Power Line Communications, and Wireless Technologies. The Wireless Technologies for the Smart Grid Architecture of Communication System used for Power System Control is approached here as well. Some examples of Communication Systems for the Electric Power System based on IEEE standard (such as IEEE 802.11 Mesh Networking, IEEE 802.15.4 Wireless Sensor Networks and so on) are presented at the end of this chapter. Last section concludes the chapter.
M. Ali Department of Science and Technology, University of Suffolk Neptune Quay, Ipswich, UK e-mail: [emailprotected] N. Bizon (&) Faculty of Electronics, Communication and Computers, University of Piteşti, Piteşti, Romania e-mail: [emailprotected] © Springer International Publishing AG 2017 N. Mahdavi Tabatabaei et al. (eds.), Reactive Power Control in AC Power Systems, Power Systems, DOI 10.1007/978-3-319-51118-4_14
547
548
14.1
M. Ali and N. Bizon
Introduction
With the need to increase autonomous and automated monitoring of the electrical power system, various communication technologies are being considered for exploitation. This is an evolving process and the global market still has not reached a mature stable point in its adoption of any one particular technology. One major consideration is the maintenance of the power grid in a stable state, with minimal power outages. The implementation of a smart grid requires the essential components of an effective communication technology coupled with an intelligent low maintenance and low energy drain device. The system must require the absolute minimal consumer interaction, preferably a set and forget device. Studies have focused on using both wireless and wired technologies. These are outlined in this chapter. However, they all have their limitations, especially for wireless coverage using frequencies higher than 2 GHz. This is due to propagation characteristics and signal attenuation through building materials. A definition of a smart grid is provided by the European Commission: “Smart grids are energy networks that can automatically monitor energy flows and adjust to changes in energy supply and demand accordingly. When coupled with smart metering systems, smart grids reach consumers and suppliers by providing information on real-time consumption.” [1]. It has to be stressed that smart grid communication is bidirectional, not necessarily at the same time, between the customer and the utility company. The smart grid is normally considered to consist of the parts of: generation, distribution and consumption of the energy [2]. The transmission network itself is not usually considered, as it can use any existing networking technology. This concept can be summarized in Fig. 14.1. The use of a well-integrated smart grid is of immense benefit to both the consumer and the energy producer, these include [2]:
Fig. 14.1 The layer view of the smart grid network
14
Communications for Electric Power System
549
• A better power distribution management policy to handle peak demands, minimizing load shedding; • Efficient integration of renewable power sources into the grid; • Consumers being able to utilize cheap power; • Consumers being able to sell excess power during peak demand by the grid— for those generating their own, for example using a CHP [Combined Heat and Power] unit and • Maintaining a stable power grid. The smart grid layer has three primary functions to fulfill, these being of providing [3, 4]: • Smart Network Management—such as data acquisition, transmission, switching and monitoring of the energy flow between the customer and the supplier, ensuring an acceptable quality of service; • Smart Integrated Generation—encompassing energy distribution, storage, fusion from diverse sources integration with provision of future grid scalability. The need to provide charging points to EV (electric vehicle) should also be provisioned as they become more commonly accepted by the general populace; • Smart Market—two-way communication functionality must be implemented between the meter and the utility company in order to support advanced features such as dynamic pricing and load control. The meter in itself must offer advanced functionalities beyond just registering energy usage. Thus the smart meter must be coupled with inbuilt networking technologies linked to network gateways, much like how modern multimedia services are internetworked. An EU Commission report also estimates that by 2020, about 245 million smart meters will be deployed just in the EU alone. Whatever technology is employed, data protection, privacy and security are of the utmost importance to protect everyone in the revenue stream. Thus some form of encryption, preferably end-to-end from the consumer premises to the utility company should be implemented over the data channel.
14.2
Standards for Electric Power Systems Communications
For any implementation of a smart grid system, it must adhere to these communication infrastructure requirements [5]: (a) Scalability The number of consumers requiring connections to the energy grid is predicted to grow exponentially. Fortunately the reliance upon the network communication technology requirements in terms of protocols and signaling are expected to be the same. Thus scalability is an important factor to be dimensioned into a smart grid network for future growth. This will entail the use of a myriad of M2M
550
M. Ali and N. Bizon
(machine-to-machine) devices being linked to the low voltage electricity grid (11 kV level). The addressing task of each network device or devices will be helped by the utilization of IPv6 (Internetwork Protocol version 6) with its 128 bit addressing space. (b) Network Performance Under this term are specific network performance parameters such as the data rate, the quality of service, the network delay. Fortunately these parameters are not so stringent as those required for conveying multimedia real-time network traffic though the Internet. Smart grids require only data transfer rates of 10 Kb or less with latencies less than 100 ms, for network originating control and signaling information. Customer premises meter-originating data are even less demanding, being able to function with higher latencies of up to 72 h! (c) Availability Any power utility company must ensure they can deliver power ideally with 100% uptime or at best with the absolute minimum downtime or power outage of say at most a few hours. Even if an outage is unavoidable, then power must be maintained by rerouting to such critical facilities like hospitals. Load shedding must be avoided wherever possible. For smart grids, availability also means being in active communication link, this is less of a problem for power being provided by fixed network connections. However, radio links must be maintained for smart metering being provided by wireless radio transmission links. Radio interference must be combatted and error correction must be built into the transmission protocol to survive in a noisy radio transmission environment. To maintain network efficiency, FEC (Forward Error Correction) coding should be employed, avoiding the need for any ARQ (Automatic Repeat reQuest) of the data packets. Some smart grid services, especially at the industrial level require an SLA (Service Level Agreement) of between 99.5 and 99.9%. This can be made possible by service reservation and prioritization of queues. Again, at the consumer premises level, the requirements are less stringent and can be relaxed by having the metering data being stored locally before being uploaded when the smart grid network becomes online or less congested again. Hence, an SLA of less than 90% may be tolerable. (d) Costs When a smart meter is to be installed, two cost factors need to be taken into account, these being the price of the smart meter itself and the actual installation cost of the meter into the consumer premise or the home. The smart meter needs to offer several functionalities and also need to be made appealing to the user whilst keeping the cost to a minimum for the power utility company. Most consumers expect their smart meters to be installed at zero cost to them, with the entire cost being borne by the power utility company themselves. The installation of the smart meter should be as simple as possible with automatic self-configuration and
14
Communications for Electric Power System
551
diagnostics. This will minimize retraining of the installers. A third factor that also needs to be considered is the actual transmission cost of each reading, in either direction. These exchanges of data need to be reliable and cost effective, without overwhelming the network capacity. This is of particular concern when expecting readings from a large city from millions of potential customers. Thus some form of data scheduling may be necessary. (e) Security Protection of the smart grid network is extremely important, not only to protect the personal billing information of the consumer, but also to protect the network from spoofing and infiltration by hackers attempting to defraud and bring down the entire network. Thus, the user data must be protected from: misuse, the network from hackers and the messages must be conveyed securely to arrive uncorrupted. The technology for secure transmission of sensitive data already exist in conveying data through the internet and this can be applied in the case for the future advanced fully integrated smart grid network. (f) Smart meter grid/system life-cycle The smart grid community would like to see a life-cycle of at least 15 years with standardization and interoperability as key features. This is to ensure longevity of the network and ease of maintenance with minimal cost. Implementing a new smart gird network is quite an expensive endeavor and these considerations of using standardized, COTS (Customised-of-The-Shelf) smart meter components should allow a cost effective implementation, deployment, maintenance and safe realization of the national and international smart grid network. The utility companies’ goal is to minimize their maintenance of their smart meters, thus achieving considerable cost savings. They also do not like to be locked into purchasing from one specific vendor for the entire lifetime of their smart grid network. (g) Control To enable a smart grid to be under the effective control of the power generating utility, it may become necessary for the utility to use their own network - totally under their control. This is especially relevant when the network demands may exceed previously agreed SLAs with their third party data network capacity providers. The network must be always made available to the utility, this can only be guaranteed if the data network is under the total control of the utility. This will thus avoid complications of legal interpretations of contract laws and what to do when the network services beyond the initially agreed service levels are required. A solution may be is to use power line data transmission technology or to set-up a dedicated wireless network, for example like the GSM-R (Global System for Mobile Communications—Railway) network. This is the GSM mobile communications network, specifically employed by the railway operators, using their own frequencies.
552
14.3
M. Ali and N. Bizon
Communications Technologies for Smart Grid Metering
Whatever technology is chosen, they must adhere to the above requirements. It may also be noted that some situations may require the use of more than one type of technology. The types of technologies that are currently being utilized in the smart grid can be categorized into two, these being wired or wireless [6–8]. For most practical implementation both are required together.
14.3.1 Wired Technologies for Smart Grid [3] a. Fibre Optic Communication Without doubt, the use of fibre optic cabling to network the smart grid offers the advantage of the availability of a high bandwidth, compared to using copper cables or twisted pair wires. However, the penetration of optical fibre to the home is still at a low level. The main reason for this is the high cost of deployment. This situation is slowly changing as the uptake of ‘Fibre to the Home’ (FTTH) continues to accelerate. New regional developments and new build towns, villages and cities are far easier to network cost effectively using fibre cabling technology to the local electricity generating substation, however. b. xDSL (x Digital Subscriber Line/Loop) For most of the industrialized nations, broadband services are provided by a form of DSL technology, hence the ‘x’ in the acronym. The cities are well served in this regard. It would appear that this form of technology is ideally suited to implement the smart grid. However, there are many challenges, these being the regulatory issues and the cost of the equipment, to actually interface into the DSL line itself, along with being dependent on the customer. It must also be noted that the smart grid is at the mercy of the customer paying for the DSL line and not terminating the contract. Despite these potential problems, smart meters are already deployed that use WiFi connections and the DSL line to carry the data back to the power generating companies. c. Power Line Communications [7, 8] Using the actual power line infrastructure itself to both carry electricity and data makes much commercial sense as the network is already there and the company has full control over their grid. Such technologies are already being utilized. However, the actual transmission is functional and possible between the transformers only and not through the transformers. The data is transmitted by carrier modulation. The use of multiple carriers, much like DSL technology, now enables broadband communication to also take place. For power line communications to be effective, the
14
Communications for Electric Power System
553
signal must operate in a very electrically noisy environment. The technology currently offers two modes of operation, this being narrow band and broadband. Wideband communication suffers from greater attenuation and restricted range. Depending on the scenario, a mixture of both narrow band and broadband power line communication is generally utilized. In the event of a break in the power line due to say a storm or from solar radiation, then complete breakdown in smart metering will occur. Thus some form of a redundant backup network must be in place. This will add a further cost to the power line smart grid network. The implementation of power line communication is still in its infancy and no internationally agreed standard currently exists. Pilot schemes are in operation and performance data continues to be gathered.
14.3.2 Wireless Technologies for Smart Grid [3, 6] There are two ways of implementing a smart grid using wireless technology. One is to use the currently available wireless technologies and mobile cellular networks. The other is to set up a private wireless localized network. Clearly the first one has the advantage of a lower cost as the infrastructure is already in place. The major disadvantage is the lack of control by the utility itself. This is not a problem, however, when a private wireless network is used. The major disadvantage here would be the high initial cost in setting up the private wireless network. Using any shared or unlicensed part of the electromagnetic spectrum would not be an advantage, as the utility company would have to compete with other users and the consequent radio noise, both affecting the effective data throughput. For in-building use, frequencies up to 1 GHz may be employed before attenuation becomes a severe problem. This would mean that the lower GSM and GPRS (General Packet Radio Service) frequencies of 450, 800 and 900 MHz should only be utilized. a. GSM and GPRS services at 900 MHz The bandwidth is adequate and the technology is mature to implement a smart grid solution using GSM/GPRS at 900 MHz currently. Indoor coverage also exists with the use of microcells and femtocells. However, in dense urban areas, network congestion and availability is a serious problem and alternative, more secure solution is needed. However, latency is a problem and the network is being upgraded to other services. Thus the long stability of the technology is of major concern to the utilities. b. 800 MHz and LTE (Long Term Evolution) The use of this fourth generation mobile technology with its broadband data capacity offers clearly a superior alternative for implementing the smart grid. The technology is also available over different frequency bands. The lower band of 800 MHz is of particular interest for implementing the smart grid. As LTE
554
M. Ali and N. Bizon
continues to be rolled out and deployed, it will make implementing a smart grid using this solution an economic viability. Though this is not currently the state and not all regions offer LTE services at 800 MHz. Primarily it is designed for voice and internet traffic and not for M2M (Machine-to-Machine) data traffic. The utility also has to rely on the commercial third party provider, thus losing control of the network. c. 450 MHz and CDMA (Code Division Multiple Access) With its wider cellular coverage requiring fewer base stations and greater in-building penetration depth and hence lower cost, this under-utilized technology can be effectively employed to form regional smart grids. The smart grids will be regional due to the regional availability of these cellular technologies, as they are not deployed globally. Growth in areas like Brazil is possible. Compared with the other cellular technologies, the use of CDMA over 450 MHz is technically the most viable solution, where it exists geographically.
14.4
Architecture of Communication System Used for Power System Control
The complexity of the information flow in a smart grid framework is shown in Fig. 14.2. The quantity of monitored data will vastly exceed that of command and control data, this is so, because of the myriad of devices that have to be monitored [9]. Clearly the control, command and billing channels need to be highly secure and the information conveyed in a timely manner to be acted upon when demand for
Fig. 14.2 NIST Smart Grid Framework 3.0 [9]
14
Communications for Electric Power System
555
action requires it. In some parts of the world, effective monitoring of “stolen energy” is of paramount importance. Figure 14.2, also identifies five areas of a power generation system: the residence, the distribution system, the substation, the transmission system and the generating station.
14.5
Examples of Communication Systems for Electric Power System
A summary of the main technologies is given in Table 14.1 [3]. These technologies should not be seen as competitive, but rather as complementary depending on the scale, cost and geographical region of the intended smart grid implementation.
Table 14.1 Properties of the Communication Systems for the Smart Grid [3] Parameters
GSM 900 MHz
LTE 800 MHz
CDMA 450 MHz
Fibre optic
DSL
Power line
Scalability— competition on resources
Yes
Yes
No
No
No
No
Low latency
No
Yes
Yes
Yes
Yes
Narrow band—no
Data rate sufficiency
Problematic
Yes
Yes
Yes
Yes
Narrow band—no
Enhanced resiliency
Not available
Not available
Available
Available
Only limited SLAs
Not available
Indoor penetration
Fair
Fair
Good
Good
Good
Good
System availability
Constrained
Constrained
Yes
Yes
Yes
Constrained
M2M optimized
No
No
Yes
No
No
No
Interference expectability
No
No
No
No
No
Expected
Economic nationwide coverage
Yes
Yes
Yes
Very limited
Partially limited
Limited
Installation/rollout
Simple
Simple
Simple
Difficult
Difficult
Simple
Security
Public grid
Public grid
Closed network
Public grid
Public grid
Closed network
Long-term availability
No
Yes
Yes
Yes
Yes
Yes
Customer behaviour dependent
No
No
No
Yes
Yes
No
Exposure to broadband market developments
No
Yes
No
No
Yes
No
556
M. Ali and N. Bizon
Seven important parameters need to be carefully evaluated for the communication network, these being: latency, bandwidth, resilience, security, scalability, coverage and the life expectancy [10]. The latency is the round trip delay between the transmitter and the receiver, which includes the media access time, the propagation delay and queuing but not the processing time at the destination. When considering the bandwidth of the communication link, the effect of channel coding and error coding in reducing the effective goodput of the information transfer capacity must be calculated. The ability of the communication link to recover from noise and natural disasters such as earthquakes is in the domain of the resilience of the link. Security is of ever increasing importance when deciding to adopt the appropriate wireless technology in our uncertain world. Unauthorized access must be prevented and confidentiality of data must be protected. Attacks from ‘vampire’ nodes can be prevented by using techniques such as mutual authentication. Clearly low powered wireless technologies are only suitable for small-scale applications, thus the coverage of the wireless technology has to be appropriate to the application. With a rapidly expanding market, scalability of the technology must be seamless and implementable with ease. For example it must be relatively easy to poll a rapidly increasing number of measuring devices at the correct rate. The life expectancy of the communication network must match that of the distribution generation system, otherwise it does not make any economic sense
Fig. 14.3 The Integration of WiMAX for Smart Grid Applications [11]
14
Communications for Electric Power System
557
to select the communication network in the first place. This is not an easy decision to make with technologies having a short life lifetime often considerably less than the electrical distributed generation system. When updating the communication network, fault tolerance will become more of a critical factor to prevent power outages in times of excessive or peak demands, especially when handling the signaling to implement an efficient load shedding algorithm. A practical implementation of a smart grid network, as shown in Fig. 14.3, is to combine several technologies together, both wired and wireless including public and private networks.
14.6
Overview of IEEE 802.11 Mesh Networking
Whatever incarnation a smart grid takes, it will likely consist of a mesh network as show in Fig. 14.4, applied to a smart grid. The main difference with this type of network is that not all the nodes have access to the wired network, instead access is gained via anchor access points. Having key access points means that congestion management is a priority to free up bottlenecks. Latency and jitter throughout the mesh must also be taken into account. To enhance radio resource planning and security monitoring, a smart grid should employ a controller-based WLAN architecture.
Fig. 14.4 Mesh Networking in Smart Grid Application [12]
558
14.7
M. Ali and N. Bizon
IEEE 802.15.4 Wireless Sensor Networks
At the heart of the smart grid will be the sensors that are used to monitor the smart grid and to implement it. Clearly a wireless sensor network will afford the greatest flexibility in its placement. The use of an already existing standard, such as the IEEE 802.15.4 is judicious. This IEEE standard is mainly concerned with specifying the requirements for power efficient low data rate wireless networks. The ZigBee protocol is the main driver behind this protocol. Other competing standards such as the WirelessHART and ISA100.11a use the physical layer of this standard but have their own respective different layer two MAC (Media Access Control) definitions. Clearly using the lower frequency bands operation below 1 GHz, suffering less attenuation will give longer radio propagation paths than using the 2.4 GHz microwave band.
14.8
Conclusion
Those technologies that are resilient to network limitations should be considered for implementing the future smart grid. An overview of the various technologies, both wired and wireless, have shown that they have their respective advantages and disadvantages. The conclusion of this chapter author is that of a hybrid solution utilizing both wired and wireless technologies. The initial fast deployment should be to utilize the underlying cellular technologies to gain a market share. Then the future would be to gradually wean away from commercial networks over to the use of a private network set up by the utilities themselves. The use of the 450 MHz frequency is ideally suited for smart grid use and the cellular technology of CDMA in particular. The deployment any particular technology is also dependent on the level of urbanization of the environment. Rural areas will tend to favour a wireless smart grid solution, whereas heavily urbanized regions will favour a more commercial cellular solution, using WiFi or WiMax for example. For the interim phase, redundancy must be seriously considered in order not to be over dependent on any third party commercial network [13]. Finally the future smart grid should ideally be a “self-healing grid”. A self healing grid would be able to identify any problems and fix itself without any human intervention. “Not only can a self-healing grid avoid or minimize blackouts and associated costs, it can minimize the impacts of deliberate attempts by terrorists or others to sabotage the power grid” [14].
14
Communications for Electric Power System
559
References 1. https://ec.europa.eu/energy/en/topics/markets-and-consumers/smart-grids-and-meters. 2. Nicu Bizon, N.M. Tabatabaei, H. Shayeghi (Ed.), Analysis, Control and Optimal Operations in Hybrid Power Systems - Advanced Techniques and Applications for Linear and Nonlinear Systems, Springer Verlag London Limited, London, UK, 2013. 3. B. Sorries, https://www.cdg.org/resources/files/white_papers/CDG450SIG_Communication% 20_Technologies_Networks_Smart_Grid_Smart_Metering_SEPT2013.Pdf. 4. M. Shahidehpour, Y. Wang, Communication and Control in Electric Power Systems: Applications of Parallel and Distributed Processing, Wiley-IEEE Press, London, UK, 2003. 5. M. Uslar, M. Specht, C. Danekas, J. Trefke, S. Rohjans, J.M. Gonzalez, C. Rosinger, R. Bleiker, Standardization in Smart Grids: Introduction to IT-Related Methodologies, Architectures and Standards, Verlag London Limited, London, UK, 2013. 6. S. Borlase, Smart Grids: Infrastructure, Technology, and Solutions, Wiley-IEEE Press, London, UK, 2014. 7. S.F. Bush, Smart Grid: Communication-Enabled Intelligence for the Electric Power Grid, Wiley-IEEE Press, London, UK, 2014. 8. N. Bizon, N.M. Tabatabaei (Ed.), Advances in Energy Research: Energy and Power Engineering, Nova Science Publishers Inc., USA, 2013. 9. http://www.nist.gov/smartgrid/images/FrameworkGraphic_1_1.jpg. 10. M.S. Thomas, John Douglas McDonald, Power System SCADA and Smart Grids, 2015, CRC Press. 11. M. Paolini, Empowering the smart grid with WiMAX, http://www.senzafiliconsulting.com/ Blog/tabid/64/articleType/ArticleView/articleId/32/Empowering-the-smart-grid-withWiMAX.aspx. 12. http://www.trilliantinc.com/solutions/multi-tier-architecture/. 13. M. Shahidehpour, H. Yamin, Z. Li, Market Operations in Electric Power Systems: Forecasting, Scheduling, and Risk Management, Wiley-IEEE Press, 2002. 14. M. Amin, The Self Healing Grid: A Concept Two Decades in the Making, IEEE Smart Grid Newsletter Compendium, 2015, http://www.qmags.com/qmct.asp?q=382&qt=3872236&u=/ OLHit.asp?pub=SGNC&upid=19496&s=ML&url=/SGNC2015.
Chapter 15
SCADA Applications for Electric Power System Florentina Magda Enescu and Nicu Bizon
Abstract Main objective of this chapter is to present the Supervisory Control and Data Acquisition (SCADA) technology applied in the energy sector which requires distributed control and monitoring at different levels. If the process is distributed, then the advantages of SCADA system will be seen through low costs related to movements of the equipment to improve the performance tracking. Processes that need to be monitored on a large area, and request frequent and immediate interventions, can be solved more efficiently through a SCADA system. For this, the connection between the master station and remote units must be done via a communication system which can use different communication technologies such as cable, radio, mobile phones, and even satellites. The optimum data transfer may be obtained by using a specific communication protocol. It is worth to mention that such SCADA-based communications are developed using the optical fiber communications technology via the Internet. The information flow between remote and central SCADA units could be designed to be bidirectional for high performance and reliability of the distributed control system, but note that both digital and analog signals are involved in such systems. Besides, an important issue is the information security related to such systems. It is known that the automation and real-time control are used via the Internet and wireless technology, but these technologies have also brought some security problems, having a strong impact both in the business and to the users. The SCADA applications analyzed in this chapter is focused on Electrical Power Systems (EPS). The stepwise step design is shown using the programming environment named VIJEO CITECT SURVEILLANCE SOFTWARES —version 7.40®. The main objective is to show how can be optimized the real-time control to obtain affordable solutions for the EPS based on Renewable Energy F.M. Enescu (&) Department of Electronics, Computers and Electrical Engineering, Faculty of Electronics, Communications and Computers, University of Piteşti, Piteşti, Romania e-mail: [emailprotected] N. Bizon Faculty of Electronics, Communications and Computers, University of Piteşti, Piteşti, Romania e-mail: [emailprotected] © Springer International Publishing AG 2017 N. Mahdavi Tabatabaei et al. (eds.), Reactive Power Control in AC Power Systems, Power Systems, DOI 10.1007/978-3-319-51118-4_15
561
562
F.M. Enescu and N. Bizon
Sources (RESs). The both current and optimized solutions are presented, and the role of the reactive power is highlighted in the comparative solutions shown, which are implemented in practice as well. This was achieved by optimizing the SCADA solution of the operation, transmission and overseeing the execution of programs for the operation of power plant trough: (1) the description of the existing solution; (2) the description of the proposed solution with redundant SCADA servers; (3) the disadvantages of the proposed solution with Remote Terminal Unit 32 (RTU32). The optimization of the applications has been carried out because the graphical interface is poor, as long as the only information displayed is from the distribution stations, the states of the switches in medium-voltage lines and of the equipment in the stations. Therefore the dispatcher has not displayed the graphical information about the status of the line equipment, and does not have a quick overview to state of the medium-voltage line as well. Thus, the dispatcher must consult the printed diagrams, which means time consumed and such complicated maneuvers will occupy the most time their activity during a day. Thus, here it will be shown the steps to optimize the graphical interface in order to quickly see the status of the EPS. Furthermore, by improving the graphic interface, the efficiency in carrying out the appropriate maneuvers will increase, as well as the degree of safety. The applications have been designed with high flexibility and can be used either for small applications, either for large size systems. Also, the graphic interface was designed to display online the page’s status and the trends in state variables.
15.1
Introduction
In last decades the automation devices have evolved from simple electronics circuits used to monitor and control the parameters of the industrial processes [1], to smart devices that communicate with each other in automation networks [2]. The evolution of the automation devices has led to the development of technique for visualization of control parameters and state variables from the automation process [3]. The companies producing automation equipment are those that have achieved the first visualization systems. In time, they have developed the generic software named Viewer, which is adaptable to various situations possible in distributed processes [4, 5]. The Supervisory Control and Data Acquisition (SCADA) system collects data from the distributed processes through sensors. The central computer will store this big data after a preliminary processing. The system can receive commands automatically or manual commands based on the processed data by the human operator [6]. The first SCADA systems, which allowed acquisition and real-time data analysis, have appeared in the 1960s, but then were rarely used in installations. The term SCADA was referred to a comprehensive system of measurement and control. The Programmable Logic Controllers (PLCs), which were designed and built by Gould Modicon in 1971, have changed the philosophy to use the SCADA. The PLCs in EPS have been introduced in 1977 by Allen-Bradley, and this was easily
15
SCADA Applications for Electric Power System
563
accepted by electricians. The term of SCADA has been used for the first time in the late 1980s. This term was not widely used until the 1990s when the technology evolved. In 1998, most of the PLC manufacturers integrate the Human Machine Interface (HMI) in/SCADA systems using communication systems and open protocols, nonproprietary. The HMI/SCADA systems offer total compatibility with PLC-s by including the following components [7]: the HMI, the controllers, the input-output devices, the networks, the appropriate software etc. A generic SCADA system must to implement a distributed database that contains elements called endpoints, which can be either hardware or software. Thus, the distributed database is composed by hardware and software items. The system SCADA controls and monitors the input or output endpoints. The record of each endpoint, including its dynamic, is stored in the database to have the history of all endpoints. This will help in predictive maintenance and the security of the EPS as well [8, 9]. So, the SCADA system has a multi-layered structure composed from basic functions and graphical user interfaces which are hardware and software supervised in real-time [10, 11]. From the point of view of the design environment, a SCADA system contains some specific instruments such as: text editing applications, graphics editors, tools to import/export from/to SCADA libraries to achieve configuration parameters, drivers & tools with advanced features for HMI interfaces, etc. [12]. The carrying-out of a project in the SCADA system, the following aspects must be taken into consideration [10, 11]: – the needed components to be easily accessible from the point of view of visibility and how these are collaborating; – timing sequence; – scalability; – flexibility to extend the system by the addition of new subsystems; – redundancy; – the initial statuses that should be determined; – the operation diagram of the system. The functions of the SCADA system are the following [1, 2, 8]: • acquires data collected from the process; • manages alarms; • allows the needed actions for automation: – a technological preset action; – an initiating action of the events; – a monitoring of the pre-defined sequences. • stores and archived data;
564
F.M. Enescu and N. Bizon
• generates and pursues reports: – a default graphics; – a multiple graphics scaling; – a history of graphs. • allows the dispatcher to control the process via the HMI; • allows the communication with user interface via the HMI using: – – – – – –
a libraries with symbols; a connection between process and graphic elements; a link between screens; a collection of command’s operators: an animation based on multimedia features; a possibility of connection with other platforms [5].
SCADA systems may be extended to Large Scale System by architecture, maintenance, post-processing, decision support systems, and economic planning [4, 7]. The architecture of hydro energy system is proposed in [12] and extended here to the architecture shown in Fig. 15.1.
Fig. 15.1 Extended SCADA architecture with application in hydro-energetics
15
SCADA Applications for Electric Power System
565
A number of recent studies in the scientific literature have highlighted the importance of the SCADA systems for RES hybrid EPS [1, 13–15], and in particular, their use for hydropower plants as below: – the SCADA system applied for hydropower plants connected to national EPS; – the use of metering systems for the electric power and water from hydropower plants; – monitoring and control of the hydro power plants; – monitoring and control of the hydropower plants and RESs used for hybrid EPS. The SCADA architecture for a hydropower plant is presented in this chapter at level of hardware and software. Five levels of SCADA hierarchy are essentially in complex system [1, 9–11]: • • • • •
the the the the the
devices level for instrumentation and control operations; Remote Terminal Unit (RTU) terminals level; communications system level; master station level; data processing level.
The evolution of SCADA systems is also shown as system generations for cyber security. The cyber security risk assessment for SCADA includes: – Causes which can lead to catastrophic situations from common technical incidents; – Cyber Security concepts; – How incidents are treated; – The tolerance which is acceptable; A survey is conducted for a specific hydropower plant from Romania based on SCADA system. Also, an overview of the programming in VIJEO CITECT SURVEILLANCE SOFTWARES 7.40® is shown, and the projects to monitor the medium power subsystem of the EPS based on RESs is described step-by-step [10, 16–19]. SCADA system allows the command, control and operation of installations both from the control room of the station and from the center point. The main operative functions of the management system are as follows: – – – – – – – – –
acquisition, processing and exchange of data; instant recording of data; transmission by the higher command center (subsequently implemented); remote control (control room) and the local branch of the circuit breakers, traps, etc.; indication of the position of switching equipment; measurement of analog quantities; metering; sequential recording of data; the processing and management of alarms;
566
– – – – – – – –
F.M. Enescu and N. Bizon
voltage control circuits and plots; remote control and adjustment in plants; marking; long-term information archiving; registration of damage; interlock of primary equipment; switching sequences with checking of signals’ timing; surveillance system.
All these features are available in new SCADA user interface with VIJEO CITECT SURVEILLANCE SOFTWARES. Today, a SCADA system operates on large distances, up to the level of a country or continent, if we consider that the national EPSs will be integrated based on smart grids and these technologies. Examples of such processes are given below [20–24]: – the groups of small hydro stations that are switched on and off in response to energy demand of customers and which are located, in general, in remote places. These systems can be controlled by opening or closing the turbine valves, but must be pursued continuously in order to quickly respond to the dispatcher’s energy demands; – the electrical transmission systems, which cover thousands of km2, can be controlled by closing or opening the grid breakers, but these must respond almost instantly to the load demand on the line; – the areas of oil and gas extraction, including sensors, collection systems, measurement equipment and pumps. These systems are generally distributed over wide areas, require relatively simple controls, such as starting or stopping motors, and metrological information on a regular basis, but must quickly respond to the exploitation conditions; – the distribution network of gas, petroleum, chemicals, water, which have elements that are located at different distances from a central point of control. These can be controlled by closing or opening valves or starting and stopping the pumps, but must rapidly respond to market conditions and alarms due to the losses from toxic or dangerous materials; – irrigation systems covering hundreds of km2 can be controlled by closing or opening of simple valves, but also require a relative complex measurement of water supplied to consumers based on centralized system. In conclusion, the SCADA system is useful because it helps to [7, 10]: – – – – – – –
optimize the process; take a decision objectively; increase the efficiency of use for the exploitation equipment; decrease energy costs; reduce the capital in the future; improve the services for the future; obtain competitive advantages on the market;
15
SCADA Applications for Electric Power System
567
– decrease the number of staff; – solve the problems of the environment; – use of a new system instead of the old ones which are exceeded.
15.2
Research Extracts from Literature
The study shown in this section is linked to: – – – – –
The Electric Power Systems (EPS) as a critical infrastructure; Advanced SCADA architectures; SCADA system for optimal energy management; Cyber security risk assessment for SCADA; New SCADA applications.
A. SCADA system for driving the stock records from the hydropower plants developed in Valcea region The SCADA system presented in this section is used in Valcea region, Romania, for driving the hydropower plants from the dispatcher station. The driving system from the Hydro-energetic Dispatcher (HD) contains (Fig. 15.2):
Fig. 15.2 Research extracts from literature
568
F.M. Enescu and N. Bizon
– SCADA systems in central dispatcher; – SCADA systems in local dispatcher. The HD is connected with the Dispatcher in Local Center (DLC) by modems on wire or using the Global System for Mobile communications (GSM), in last case a large communication network dedicated to this process is obtained [12, 25, 26]. This system allows the control of the operative functions, the automatic acquisition of data, and generation of the databases needed. All these actions and data records are necessary for rational utilization of RESs. Note that this SCADA architecture is of open system type, which ensures integration of the regional power plants into national system and flexibility to further develop this system. The SCADA system from HD station provides information for water flow planning in relation with energy demand from dispatcher, which is of following type [12]: – for monitoring and ensuring the service continuity for power stations and power substations subordinated to them. – for implementation of the governmental program to monitor the operation of the main power stations; – for energy management and operative management in each power stations; – for operative exploitation of water accumulations and optimal distribution of the water flow to power stations; – for operation in normal and hazard situations based on the remote control units, switching equipment and primary outlet installations (Fig. 15.2). The SCADA system from DLC provides information for: – primary processing of data; – storage of data during a limited period; – increasing the reliability and the maintenance capacity of system. Because this architecture is of distributed type, the functions of the system must to show in real time: – the retrieval numeric signals for the status of switches and splitters, the operational equipment, the power cells, the protection loops, and timing cells; – the acquired and processed signals (of analogue and digital type) form the hydroelectric power stations monitored; – the commands to hydroelectric power plants, both in the local and central interfaces of the dispatcher from each hydropower plants; – the alarm signals from the dispatchers in the event of the appearance of some defects; – the record of the number of operations for each equipment, which is used in the management of revisions; – the schematic diagrams of the hydropower plants monitored, the status of state variables from these diagrams, the wiring diagrams and the synoptic chart of the installations used [13];
15
SCADA Applications for Electric Power System
569
– the stored records using the graphical or tabular configurations, features which are selectable by the operator [17]. B. Metering systems of electricity and water for a residential house The metering of electricity and water flow requested by a residential house is also a usual application of SCADA system. The SCADA system are responsible for automatic reading of the meters installed and storing the data in SQL database as a record which contains the index of consumption, the location, the time, the state of protection circuits etc. These data generate some alarms and reports related to monthly consumptions, which are used for the allocation of costs facilities for each location and meter. Data are provided directly in Comma Separated Values (CSV) format to be imported directly into the billing software. The network architecture is standard [10, 27, 28], and contains the following levels: – monitoring software: it is composed from separate applications, which are running on different servers; it can be accessed directly on the web through monitoring portal, which displays graphical reports in real-time based on the data received; – data concentrators, which are used to link the software that is run on the server and the meters installed in each home; – metering level, which provides support for remote transmission of the index of consuming or other monitored parameters. – Such mattering systems allow the following automatic operations [29–33]: – reading the consumption and other parameters in the days and at the times programmed by the user; note that data supplied by the meters installed can also be read out manually, at the request of a user, in order to determine current consumption or to verify the proper operation of equipment; – disconnection of the consumers can be remote controlled using the SCADA software and flow’s distributors installed in the network; – graphical analysis of the daily profile for the load demand or the carbon footprint reported to a day, week, or month, or other parameters available directly in web portal; – detection and location of the faults arising in operation of the equipment (meters, concentrators, etc.) or the software modules, and generation of alarms via email to the authorized users; – computing of daily consumption and monthly consumption (which is used for the forecast of the energy sources) by using the software management based on the difference between the automatic readings; – exporting of the various query reports generated from the database to Excel or other formats. C. Monitoring and control of the renewable energy sources The RESs (solar energy and wind power especially) are used in hybrid power systems based on distributed generation because:
570
F.M. Enescu and N. Bizon
Table 15.1 The evolution of energy systems Year The future
Energy based on Hydrogen—combustion chef
2010
Solar
2003
Wind
1969
Electric Power Plant
1948
Petroleum
1900
Steam
– the actual consumption increased and the trend for the energy; price is to rise even higher in years to come; – the classical resources begin to run out; – the importance of RESs in climate change [13–15]. The evolution of energy systems can be monitored in the Table 15.1. Two examples of RES hybrid power systems are given below: 1. Diesel-wind turbine hybrid EPS to supply the industrial consumers (Fig. 15.3). The application is submitted in under (Fig. 15.3) Sect. 15.7.2 2. Home applications based on solar power (Fig. 15.4) Supervision and monitoring of the small photovoltaic parks The dispatchers of small photovoltaic parks make the surveillance with dedicated appliances that enable measurements to be processed via Bluetooth from inverters. Supervision and monitoring in parks large photovoltaic Supervision and monitoring of the large photovoltaic parks The dispatchers of large photovoltaic parks use SCADA systems to monitoring the energy generated based on compact reconfigurable items which may avoid the fall
15
SCADA Applications for Electric Power System
571
Fig. 15.3 SCADA system for hybrid EPS based on RESs
Fig. 15.4 Exploiting the RESs for usual home applications
of some subsystems. These items can be easily changed, reconfigured and maintained [10]. The optimal energy management strategies may be the following [20]: – to control the balance between supply and demand with a lesser impact on carbon emissions and high profitability in maintaining the business; – to distribute the energy generated for different energy sources (in particular from RESs); – to offer clean energy at reasonable costs and under a control of the impact to the ambient.
572
15.3
F.M. Enescu and N. Bizon
Hardware and Software for SCADA Systems
As it is known, the SCADA systems are able to measure and monitor different parameters and variables, such as: the level of liquids, the volume of the gases, the volumetric flow rate, the mass of liquids and gases, the pressure, the temperature, the humidity, the viscosity, the position and the moving, etc. For this, some signals must be acquired based on the appropriate sensors and transducers from different electrical and electronic devices, before to be processed and analyzed by SCADA system. These data can be used for supervision of remote processes. For this, the data are processed and analyzed to generate reports which will be transmitted to control the processes parameters within certain limits. The SCADA applications have two basic levels: – the customer level, which will be performed the human-machine interaction; – the data server level which will control the entire process. Thus, the data server sends these data to the customer level via remote equipment such as PLCs (connected to the database server either directly or via a communications network, using various communication protocols). The developed protocols can be property of a company (e.g. Siemens H1) or of open type (Modbus, Profibus etc.). Database servers can be fully connected to each other and also to the customer stations through Local Area Network (LAN) networks based on Ethernet technology [21–23]. The SCADA systems may be relatively simple or very complex, depending of the dimension and complexity of the process monitored. These systems operate in real-time using a database system named Real Time Data Base (RTDB), which can be found on one of servers [3].
15.3.1 Hardware Architectures The servers are responsible for the acquisition and management of data for a set of parameters. Is it possible to have dedicated servers for certain tasks (for example, if the objective is to develop the process servers), servers dealing with the handling of alarms, or file servers [23]. A SCADA server will be connected directly to the PLC via an Open Platform Communications (OPC) or a RTU, which both collects and consolidates data from and from PLC-uri (Fig. 15.5). The SCADA takes on request the data from indicator reading appliances and equipment condition. The OPC and RTU work as a data concentrator. The data is structured using a man-machine interface (HMI) in a convenient format for the operator in order to be able to take any decisions that would optimize the communication between the equipment.
15
SCADA Applications for Electric Power System
573
Fig. 15.5 SCADA hardware
RTU performs: the connection with supervised equipment; reading of equipment status (such as open/closed position of the valve or relay); acquisition of measured signals, such as the pressure, flow, voltage or current; the control of equipment by sending command signals, such as the closing of a valve or relay or setting the speed of a pump; – reading the digital or analog signals, and sending the commands using both digital or analog signals.
– – – –
The PLC contains a microprocessor used for the automatization of processes. A PLC is provided with input/output sensors and relays. PLC is less expensive alter to old systems which used tens or hundreds of relays and timers. The PLC is programmed using the structured programming languages and elementary logic operations. The term “Supervisory Station” refers to the servers and software responsible for the communication with the equipment (RTUs, PLCs, etc.) and HMI software which run on master-station in the control room or elsewhere. The master station may be composed of one PC in small SCADA system. In large SCADA systems, the master station may include multiple servers, distributed software applications, and disaster recovery strategies. To increase the integrity of the system, the multiple servers will be often configured in a dual-redundant or hot-standby topology, providing control and monitoring even in the event of a server failure [9].
574
F.M. Enescu and N. Bizon
15.3.2 Generations of SCADA Systems In this section will be presented briefly the well-known generations of SCADA systems. • First generation is of “monolithic” type based on mainframe computers. The SCADA system was developed under conditions in which there were no computer networks. So, the SCADA systems were dependent designed to number of remote terminal units because the RTU was developed later (Fig. 15.6). An additional mainframe computer is added in the event of failure of the master system. • Second generation is of “distributed” type because the information is shared in real time in the LAN. The data processing is done on several stations connected to the LAN (Fig. 15.7). Responsibilities are divided to each station, which enables high speed data processing at reduced cost in comparison with first generation. • Third generation is of “network” type because the SCADA system uses the open system architecture, where can be easily connected peripherals such as printers, hard drives etc. The communication between the SCADA master and networked remote terminal unit is based on Wide Area Network (WAN) protocols such Internet Protocol (IP) address (Fig. 15.8). But the use of IP address standard via Internet makes the SCADA systems vulnerable to cyber-attacks. Consequently, new network protocols are developed by companies, but still remain important security issues to be solved.
Fig. 15.6 First generation of SCADA systems
15
SCADA Applications for Electric Power System
Fig. 15.7
Second generation of SCADA systems
Fig. 15.8 Third Generation of SCADA systems
575
576
F.M. Enescu and N. Bizon
15.3.3 Software Architectures The SCADA applications are implemented based on the software architecture (Fig. 15.9). SCADA software architecture has at least two components: • the SCADA server application; • the SCADA client application; The SCADA server application; is multi-tasking being responsible for storing the SCADA client applications in a database. The SCADA client application receives data via serial port using RS232 and MODBUS RTU protocols. In many cases the SCADA applications shall be addressed to and remove the users who have not an Internet connection. In this case it is necessary a new software component on a Web server. A Web server is used for generating dynamic Web pages, which are supplied to customers decrease. The Web SCADA clients can benefit to real-time access (in a manner similar to regular SCADA clients) to the lists of parameters, the lists of events and the list of alarms via a simple browser. But, the Web SCADA clients do not receive the same level of graphical user interfaces and have the same HMI features as the regular SCADA clients. The regular SCADA clients are accustomed to running specific applications as opposed to The Web SCADA clients who view the Web pages provided by Web SCADA server. In order to reduce as much difference between the regular SCADA clients
Fig. 15.9 SCADA software architecture
15
SCADA Applications for Electric Power System
577
Fig. 15.10 SCADA systems
and Web SCADA clients some drives, services and virtual instruments are implemented in HMI. Visualization software used for monitoring and control of industrial process is called “software” HMI or “SCADA software”. HMI as it is called generically, means any switch, interface through which you can control a device. The term “automation” refers to the graphical interfaces implemented through interaction with the mouse and/or the keyboard or by touch. HMI/SCADA appeared as a terminal dedicated for users from the system implemented with PLC units. A PLC is programmed to automatically control a process. Because the PLC units are distributed in a large system, the data from PLC are stored and then transmitted. HMI/SCADA has the role to gather and combine the data from the PLC using a specific protocol. HMI software can be connected to a database in order: – – – – –
to to to to to
fulfill the charts and diagrams in real time; analyze the PLC data; treat the planned maintenance; drill the schemes for a particular sensor or machine; apply the troubleshooting methods of the system.
The basic SCADA protocol is MODBUS, which is designed to send data to the master station even when the master interrogates the RTU station. MODBUS is based on a master-slave architecture known also as client-server architecture. The protocol is designed to be used by Modicon in own PLCs. Now, it became a standard of communication for the industry and is currently the most widely used to
578
F.M. Enescu and N. Bizon
Fig. 15.11 SCADA energy management system
connect all the industrial devices. The protocols contain extensions to operate over Transmission Control Protocol (TCP/IP). In practice, it is suggested to avoid the security of Internet connection in order to reduce risks of cyber-attacks. This protocol has the following features: 1. it is an open protocol with available documentation; 2. it can be quickly implemented (in few days, not months); 3. it works with bits and bytes, so no particular requirements on implementation. MODBUS allows the management of the network devices, which gives new features to the SCADA system related to flexibility, reliability and safety in operation at high performance.
15
SCADA Applications for Electric Power System
15.4
579
Assessment of Cyber Security Risk for SCADA
The developments in Information Technology (IT) systems regarding the safety of SCADA systems are shown in Table 15.2 [9–11]. Catastrophic situations which may occur due to incidental technical at level of the automation systems, Energy Management System (EMS), SCADA system, communications system, and support infrastructure. Besides these, other causes may be the incidents of cyber security and problems in the management of emergency situations. New risks which occurred “The Stuxnet incident has underlined what many ICS security experts had assumed for a long time: A sophisticated cyber-induced attack against an industrial facility would try to attack control systems rather than IT systems”—CIGRE WG B5-D2.46 – – – – –
June 2010, Stuxnet was discovered; September 1, 2011, Duqu was discovered; May 2012, Flame Retardants was discovered; June 2012, Gauss was discovered; etc.
What is the concept of Cyber Security? Security of those systems cannot be treated as a whole, and nobody can guarantee perfect security, because there is any system of defense perfect. Measures to protect and prevention are not enough, because emergencies and viruses could act in time using hidden forms or different rules. It is important as process to be defined from the point of view of management and responses to the incidents [8]. Incidents may be treated as shown in Fig. 15.12. Safety in the field of energy must be monitored on the basis of production, transport and distribution, until to the end-consumer. In general, the current
Table 15.2 Systems developments in IT and SCADA security Past
Present and future
SCADA (Fig. 15.10)
SCADA + Energy management system (EMS) + protection to electromagnetic pulse (EMP) (Fig. 15.11) Partially open systems Bi-directional communication and high cyber security (CIGRE—WG D2.34) More information Standard OS + commercial OS
Closed systems Push-to-talk communications and cyber safe Few information Dedicated operating system (OS) Few security requirements
Consistent data requirements
580
F.M. Enescu and N. Bizon
Fig. 15.12 Procedure for treating the incidents
problems are treated without taking into account new risks which can appear and could make huge damage, if will not be considered. Risk assessment management involves the following stages: – the analysis by which risks are identified in order to be assessed; – the risks assessment from the point of view of the impact and likelihood of occurrence within tolerance imposed; – the future decision by identifying and understanding risk [24]. Risk assessment can be done using several tools, one of the most popular being the probability-impact matrix which encompasses: – the assessment of the probability that a risk to materialize and measured the degree of certainty: – the assessment of the impact as the consequence or effect, where the risk would materialize. In the impact assessment, account shall be taken of the fact that the risks that have a strong impact on acute misconduct, while those which manifests itself frequently and have a low impact signals a chronic problem; – the evaluation of risk exposure as a combination of probability and impact can be felt where the risk would be materialize. In case of the materialization, the risk exposure is, in fact, an impact [24]. As a result of completing the stages of identifying and assessing of possible risk, finally, a synthetic and clear representation of the risks must be drawn, mentioning their ranking in accordance with the level of exposure. The prioritization of risks is a very important tool in risk management. Tolerance is estimated by determining the level of risk accepted. The level of risk accepted is established after the adoption of the response. The following should be taken into account: – inherent risks; – residual risks.
15
SCADA Applications for Electric Power System
581
Fig. 15.13 Assessment of exposure to risks
Risk tolerance involves some subjectivity because it depends to a large extent by the way in which this is perceived as risk. Risks which have a high level of the exposure is located on top of the tolerance and should be dealt with by means of measures by which exposure to residual risks must be made under this tolerance [24] (Fig. 15.13).
15.5
SCADA Applications
The HMI/SCADA software can be installed on personal computers or embedded computers with touch-screen, distributed through the house or into the industrial process. It runs a real time operating systems because it must respond in the shortest time to changes in the parameters of the process monitored. Consequently, the HMI/SCADA software must meet the following requirements: – – – – – – –
to be easy to use; to have an intuitive interface; to provide increased security; to enable the remote access (if it is the case); to be easily to access and use the information; to be flexible in use; to have the possibility of communication through the various services (sms, e-mail, fax).
The HMI/SCADA software must be capable of performing certain functions, such as: – – – – – – –
control and supervision of process; alarming the user; acquisition of the data needed; storing of the events; keeping the databases updated; developing of graphical evolution of the process; coupling with Enterprise Resource Planning (ERP) software.
582
F.M. Enescu and N. Bizon
The most important element of an integrated HMI/SCADA is the interface that connects the command and the system that is monitored. This interface is used by the human operator to access: – the graphic elements that symbolize the automated technological process in a systemic, appropriate and intuitive manner; – the most important functions in order to enable all functionalities; – the state of alarm variables and the alarms history; – and many other information displayed in a form easy to be followed by the human operator to see the evolution over time of the process monitored. Optimization through SCADA solutions Currently, in the world and in our country, the energy industry is characterized by profound structural transformations. Besides these requests for structural changes, the function in a competitive market leads to the amplification of the current efforts made by the power companies to increase efficiency and quality of their service. One of the ways of achieving this aim is the orientation towards the smart grids based on a new strategy for energy management. Thus, information of the EP system is one of the basic prerequisites for increasing the effectiveness and safety of the operation of the national energy system. An example of optimization of the electric power stations along a river is shown in next section.
15.5.1 SCADA System Existing in a Hydropower Plant The existing SCADA system (Fig. 15.14) consists of: – the SCADA systems associated with each Hydro Electric central Computer (HEC) (HEC1, HEC2, HECi); – the SCADA System for the Hydro Energy Dispatching (HED) from based Hydroelectric Station (HS). – At the level of the HECs arrangement, the SCADA system is composed of: – Distributed Antenna System (DAS) which represents the equipment for the data transmitted between the SCADA processes in HEDs from the Hydroelectric Station (HS). The DAS has available four RS232 l ports and an Ethernet port. The communication protocols supported are: (1) the DAS Serial, (2) the Modbus RTU, (3) the International Electrotechnical Commission (IEC) 1107 Mode C for serial ports, (4) the DAS TCP/IP, and (5) the Modbus TCP for the Ethernet port. Note that the communication with HEDs is made through the DAS serial ports on two redundant paths: (1) using the telephone line and the communication protocol; (2) using the radio communication protocol. – Functional Assemblies (FAs 1–4) may represent Hydro-aggregate automation systems, automation systems, General Services (GS), and HEC barrage.
15 SCADA Applications for Electric Power System 583
Fig. 15.14 Architecture of the existing SCADA System
584
F.M. Enescu and N. Bizon
The communication protocols implemented in the FA are the OPC industrial communication protocol (available in Wincc, Cimplicity). Communication with DAS is via Ethernet using the Modbus TCP protocol, which has the role of protocol gateway between OPC industrial communication protocol (available in Wincc, Cimplicity) and Modbus TCP protocol (available in DAS-900). – The operating Console of the SCADA HEC is based on the OPC industrial communication protocol available in SCADA iFix (Intellution). Programmable thermometry in Hydro-agregate Automation (HA1, HA2) systems is made based on (1) Modbus RTU communication protocols, which assure the communication with DAS via RS485 (or RS232) with the energy meters. CewePrometer is a family of high-precision four-quadrant energy meters which has the following communication protocols available: (1) Digital Library Management System (DLMS) (used by the energy metering application) and (2) IEC 1107 serial C Mode. Communication with DAS is via RS-422 (or RS232) using the communication protocol IEC 1107. The HED of the SCADA system is composed of: – The Serial Router (8 + 8 ports) is composed by two industrial computers with QNX operating system. Communication protocols supported by the application are the DAS Serial and DAS TCP/IP Ethernet. Communication with DAS HECs is done in two paths as well. – The radio Router System assures the communications with 16 hydroelectric power plants through radio-relay, using the same communication protocols as the Serial Router. – The SCADA application is installed on the SCADA iFix (Intellution) server and the OPC/DAS TCP/IP gateway. The second SCADA server will operates as the main SCADA server if will be the case. – The Data Base (DB) SCADA server use the iFix application based on the Microsoft SQL Server (MS SQL). MS SQL server is the data source for the “Technical Management” application. – Technical Management server is a computer on which is installed and runs the “Technical Management” application based on the installed MS SQL server and WEB server. Information flow diagram From the point of view of information flow, the existing SCADA is presented in Figs. 15.15 and 15.16. Thus, at the level of DAS-900, the HEC data concentrator acquires specific data for HECs SCADA from the FAs. FA1 and FA2 read the data from the automation systems HA1 and HA2. The General Services (GS) and the Barrage application use the OPC protocol (existing graphic stations as base driver) via the Modbus-TCP from the DAS-900. This conversion from OPC to Modbus-TCP has the following disadvantages:
SCADA Applications for Electric Power System
585
Fig. 15.15 SCADA system information feeds
15
586
F.M. Enescu and N. Bizon
Fig. 15.16 Streams and SCADA system architecture for the river hydro-arrangement
– the software application is not documented and it is very hard to be interfaced based on standard protocols with existing SCADA architectures. It appeared as a necessity to interface DAS-900, which doesn’t support for the OPC protocol, with the process charts from the HEC stations, which have locally implemented the OPC protocol; – it requires automatic start and stop upon launching the application of monitoring and control on each processor of the workstation graphics or restarting the operating system. So, it must be individually monitored for any restart in the event of a communication blockage;
15
SCADA Applications for Electric Power System
587
– as was seen in operation, there were frequent blockages of this application, which means that data and commands are not passed to and from the hierarchical level; – the application for operation relies on two text files with strict rules of editing (spaces, TABs, bookmarks, etc.), which must be configured identically for each workstation graphics (HA1, HA2, General Services, and Barrage) and DAS-900. Note that the existing SCADA system has the possibility of transmission the commands, but at the present time these are not used. Thermometry system for the HA1 and HA2 will read the temperatures monitored and the CewePrometer meters will give the main electric values: active power, reactive power, voltage, current etc. The data concentrated in DAS900 are transmitted by the SCADA system to the HED through the two redundant paths mentioned above. DAS900 system has the option of receiving SCADA commands and records from HED, but at the present time these are not used. The SCADA data acquired and processed by the HED are stored in the MS SQL database in order to be used by the “Technical Management” application.
15.5.2 SCADA System Proposed at the River Hydro-Arrangement The proposed architecture (Figs. 15.17 and 15.18) was chosen to obtain an increased security by physical separation of the LANs HEC, SCADA LAN, LAN DSZ, and LAN CORPORATE using Router/Firewall equipment. The implementation of a fiber optic network (of redundant ring type) will facilitate the communication between the HEDs related to East River and HEC. The IEC 60870-5-104 communication protocol is proposed, which is an industrial protocol suitable for communication between servers and optimized SCADA network of HED and HEC WAN. Note that the existing radio network will remain functional as a backup plan for fiber-optic (FO) network. Thus, the IEC 60870-5-101 radio protocol will be used as a hot backup (with automatic switching) for the IEC 60870-5-104 main protocol based on FO network. Consequently, the HED communication with other HECs will not be affected. The managers of the Hydro-electrica (HE) Company aim to implement a central SCADA system data of “real time” type on all hydro-stations. For reasons of security and reliability, it is not recommended to send the data directly from HE SCADA LAN (which is a closed and secure network), but it is possible to transmit data from the isolated nodes. The Technical Management server has all the necessary facilities for data transmission to HE Company. The proposal to modernize the HEC architecture by replacing the existing equipment with new DAS-900 equipment and two SCADA redundant servers (with
588 F.M. Enescu and N. Bizon
Fig. 15.17 Architecture of the proposed SCADA system
15
SCADA Applications for Electric Power System
589
Fig. 15.18 Streams and SCADA system architecture proposed for river hydro-arrangement
hot redundancy) to eliminate the existing malfunctions and in addition to implement new communication functions needed for operation in HED such as • Bidirectional communication between the automation systems (HA1, HA2, Barrage, GS) using the OPC industrial protocol, without the need of other applications for data conversion; • Communication (in terms of realtime monitoring of the main electrical variables) with existing CewePrometer meters using the IEC1107 and Ethernet TCP-IP communication protocols;
590
F.M. Enescu and N. Bizon
• Bidirectional communication (in terms of data monitoring and transmission of commands) of HED SCADA system using two paths communication: IEC 60870-5-104 communication protocol (primary path) and IEC 60870-5-101 (backup path). The SCADA applications dedicated servers hold a large number of standardized protocols for direct communication with automation devices used in monitoring and management of industrial processes. The advantages for data acquisition and remote management of the hydropower plants retrofitted with redundant SCADA servers are: – Simplification of the structure by removing the SCADA DAS-900 equipment (or other similar equipment) and use of the redundant servers for sending commands and data from the central dispatcher and vice versa; this allows easy capacity expansion and further development of software, no constraints; – The use of a SCADA software installed on the server, ensuring direct connection via OPC industrial communication protocol to the existing equipment for the management and supervision of the processes, without intermediate software; servers will provide the following functions: acquisition, operation of data, storage and archiving, printing and reporting to the higher and lower levels of communication, supervision of automated processes. This will also assure: (1) communication with existing equipment installed (meters, protection relays etc.), (2) high speed communication through the FO network of the each HEC and HED, (3) improved security provided by the two levels SCADA servers (two network cards and firewalls on the routers), (4) local drivers and OPC connectivity with inferior equipment levels, (5) storage and upgrade of existing licenses for iFix and SQL Server, (6) uninterrupted operation through hot-plug functions of the redundant servers, both in terms of data (change of the HDD without stopping the server) and to server maintenance (change of the power supply without stopping the server), and (7) easy interfacing with HED Technical Management server.
15.5.3 Issues Related to Data Acquisition and Remote Management Issues related to data acquisition and remote management applications of the power plants, which running on hydropower facilities with RTU, are the following: – RTU32 in the proposed SCADA architecture is not hot redundant at equipment level; – Unable to communicate with graphic stations of the RTU process and SCADA servers via the OPC industry standard protocol;
15
SCADA Applications for Electric Power System
591
– The need to create a new application (named “Date Gateway”) at the level of SCADA servers, as well as the “Technical Management” server, for bidirectional communication between the servers; – Managing under Windows XP of the Setup software in RTU with Straton; – Bringing of electrical signals directly through wires to the level 1 (RTU 32 level) is not indicated, but the solution to bring them to level 0 (PLC and RTU) it is recommended.
15.6
Overview of the Programming Environment VIJEO CITECT 7.40
CitectSCADA system is considered one of the most complex and complete products for the SCADA systems [34, 35], being part of the VIJEO CITECT SURVEILLANCE SOFTWARES 7.40® of to the Schneider company. This software product includes over 150 drivers for communication with the input-output devices and over 350 models for PLCs, RTUs, controllers and regulators, barcode readers etc. For any type of SCADA application, this system offers the needed flexibility in implementation and execution of the project, in terms of speed, efficiency and accessibility. More detailed, the CitectSCADA system offers the following opportunities to user: – – – – – – – – – – – – –
Use of the HMI graphic pages for central or local control stations; Realization of operation stages by creating the graphical control buttons; Display the status of the actuators via graphical charts; Display the alarm messages and their evolution; Language selectable to display the messages; Achieving of universal commands through the keystrokes; Controlling, storing, monitoring and displaying of the alarms; View the graphs in real time or in “history record”; Efficient and effective monitoring of the system; Creating of reports; Supervision of product quality by Statistical Process Control (SPC); Acceptance of transferring data with other applications; Security system can be configured on the priority levels. VIJEO CITECT SURVEILLANCE SOFTWARES has the following modules:
– – – –
Citect Explorer; Project Editor; Graphics Builder; Cicode Editor.
592
F.M. Enescu and N. Bizon
Citect Explorer The Citect Explorer module is used to create and manage a CitectSCADA project. This controls the configuration permitting enforcement of project in the rest of modules. Citect Explorer screen is shown in Fig. 15.19a: The launch of the Citect Explorer module will automatically run the Project Editor and Graphics Builder modules, but these are minimized in Fig. 15.19a. The close of Citect Explorer module will close all CitectSCADA applications. Project Editor The Project Editor module is used to create and manage the configuration information of CitectSCADA project. The Project Editor window can be viewed by all projects. Special commands are accessible from the buttons or submenus (Fig. 15.19b). Citect Graphics Builder The Graphics Builder module is used to achieve pages and editing graphical objects used on other new graphics (Fig. 15.19c). Graphics Builder editor is activated by a double click on a graphic object in Citect Explorer. Cicode Editor The programs are edited in the window of Cicode Editor. The Cicode Editor menu contains a list of functions. The support information for functions is obtained by a right click on the name of the function (Fig. 15.19d). Configuring CitectSCADA Projects The schematic of a CitectSCADA project looks like in Fig. 15.19e. As can be seen in Fig. 15.19e, the CitectSCADA project includes the following items: – Graphics; – Databases; – Cicode programs. Graphics mean a graphic page which allows the monitor to display the graphical interface with control buttons. Databases allow the storage of process information collected for monitoring and controlling the system. These may be linked to the graphics page if desired. Cicode programs allow functionality and contain a number of useful functions stored. In all projects carried out, the user can: – – – – – –
Create; Select; Open; Shut down; Delete; Etc.
15
SCADA Applications for Electric Power System
593
Fig. 15.19 a Screen of citect explorer. b Screen of project editor. c Screen of graphics builder d Screen of cicode editor. e Scheme of the CitectSCADA project
594
Fig. 15.19 (continued)
F.M. Enescu and N. Bizon
15
SCADA Applications for Electric Power System
15.7
595
Developing Step by Step a SCADA Application for Electric Power System
Application 1
15.7.1 Implementing a Medium Voltage EPS Based on RESs The CitectSCADA environment is used here to develop step by step a SCADA application for the Electric Power System based on RESs. The both interfaces will be shown in this section in order to shown the advantages of the new SCADA interface, which is a friendly and intuitive interface, with a high degree of integration schemes, facilitating for more dispatchers to work simultaneously. The old interface As can be seen in Fig. 15.20, the old interface used is poor. In case of maneuvers in power installations, the dispatcher has a limited overview on the system. The new interface A SCADA application is developed step-by-step starting with the identification of needed elements and objects, which then will be set for data acquisition, command and control equipment etc. The software structure will be identified correspondingly in the hardware structure. A new project must be opened and created, and finally must be saved. The project will define the appropriate hardware to achieve a simulation structure using the graphical page. For the simulation process, the structure must be monitored to be identified and set properly in the graphic pages as associations of hardware items
Fig. 15.20 Medium-voltage EPS based on RESs represented in the existing SCADA system
596
F.M. Enescu and N. Bizon
and their appropriate functions. An important step will be dedicated to security, namely: – setting application users; – setting their rights. After implementation of the simulation structure, each item must be check using the software to test all functional features (Fig. 15.21a). The steps for creating a new project are: – From the Citect Explorer ! File ! New Project, a new project called “Micro_energy” is open (see Fig. 15.21b); – From the Citect Explorer ! gen ! Communications ! Clusters, define a cluster of communication called “sch_el_cluster” (Fig. 15.21c); – From the Citect Project Editor ! Servers ! Network Address, define the TCP/IP address 127.0.0.1 called “sch_el_adr”; Address “sch_el_adr” is the TCP/IT data acquisition system which are allocated to the system (Fig. 15.21d). – Then, the further facilities will be configured as below: 1. “Alarm”: From the Citect Project Editor ! Servers ! Alarm Server (Fig. 15.21e). 2. “Trending”: From the Citect Project Editor ! Servers ! Server Report (Fig. 15.21f). 3. “Report”: From the Citect Project Editor ! Servers ! Server Trend (Fig. 15.21g). 4. “I/O Server”: From the Citect Project Editor ! Servers ! I/O Server (Fig. 15.21h). 5. “Express Communications Wizard”: From the Citect Explorer ! gen ! Communications ! Express I/O Device Setup ! Next ! Use an existing I/O ! Disk I/O Device ! Citect generic Protocol ! Automatic refresh of tags ! Finish (Fig. 15.21i).
15.7.1.1
Making the Graphic Page
To this end, in this section the libraries of the associated elements, which are required or which must be designed or imported if there are no such elements, are shown below (Fig. 15.21j). The following symbols and window is used for the transformer and to set the voltage ratio, and to obtain the animation symbol in order to switch the power on or off the power for transformer (Fig. 15.21k). Each of the symbols used in application are briefly shown for further identification. Also a suggestive name, a label and a serial number will be associated to
15
SCADA Applications for Electric Power System
Fig. 15.21 a Concept diagram of the SCADA application. b Window “New Project”. c Window “Cluster”. d Window “Network Address”. e Window “Alarm Server”. f Window “Report Server”. g Window “Trend Server” h Window “I/O Server”. i Window “Express communications wizard”. j Imported images. k “Transformer” symbol and window “Symbol Set Properties”. l “Electrical splitter” symbol and window “Symbol Set Properties” m “Switch” symbol and window “Symbol Set Properties”. n Animated symbols. o Window “Variable Tags”. p Window “Symbol Set Properties”. q Window for proposed process. r Graphic window of the active process. s The sequence of code in the graphical user interface (GUI)
597
598
Fig. 15.21 (continued)
F.M. Enescu and N. Bizon
15
SCADA Applications for Electric Power System
Fig. 15.21 (continued)
599
600
F.M. Enescu and N. Bizon
(j) Industrial consumers
Households
Photovoltaic boiler
Wind
(k)
Transformer
(i)
Electrical splitter
Fig. 15.21 (continued)
each symbol (Fig. 15.21l). See below the symbols for the electrical splitter (Fig. 15.21l) and the switch (Fig. 15.21m). In the same manner are designed the next symbols used in application (Fig. 15.21n).
15
SCADA Applications for Electric Power System
601
(m)
Switch
(n) Biomass power plant
Power station
Hydropower
Power level
(o)
Fig. 15.21 (continued)
15.7.1.2
Configuring the Variables
All SCADA application are based on the setting and using of the tag-type variables. Tags are the interface between the user and the monitor. There are several types of tag-type variables:
602
F.M. Enescu and N. Bizon
Fig. 15.21 (continued)
• Tag Variables are variables that are directly related to the process, being modified directly by the acquisition and control system, and similarly, they also have direct access to the SCADA application; • Local Variables—required in the SCADA application development; • Trends Variables—required for plotting the variables acquired from the process monitored. To add a tag variable for a switch or a splitter remote controlled, proceed as follows, using Citect Project Editor ! Tags ! Variable Tags (Fig. 15.21o).
15
SCADA Applications for Electric Power System
603
Fig. 15.21 (continued)
As shown, the variable called “s1” will be associated with the separator s1. Similarly, the rest of the variables were introduced and associated with each switching element or existing splitter shown in the diagram from Fig. 15.20 To associate an element to a symbol variable, the Properties window must be open for that element, as shown in the following figure for “Up command”. Here, access the list of variables using the right button. From the list that appears, select the setting required for an ease of identification which containing a serial number associated with its name (Fig. 15.21p).
15.7.1.3
Method of Implementation
A new graphic page with the name “L 20 kV SCADA” will be created from the Citect Explorer ! gen ! Graphics ! Pages ! Next ! Create new page ! (Fig. 15.21q). Place the elements in the graphic page using the library and then configure the related functions as indicated above. One can start the graphical page in accordance with the operating mode of each one, and then the variables will be set; or vice versa, the design program being very flexible. The simulation diagram is tested if running proper. Operation for each item is tested with the left button of the mouse, in different combinations of the tabs. Also,
604
F.M. Enescu and N. Bizon
the operation of the power level indicators is checked. Adjustment is possible through the existing cursor for both consumers and energy producers. If it appears that there are inconsistencies, errors and changes are required. Finally, the application will be saved. The whole process of the distribution system is monitored and controlled permanently by SCADA. The interface used is improved, showing the new features implemented at level of the dispatcher, which has an overview of the EPS and can perform maneuvers in the system in order to control the process. Besides the information about the state of switches at distribution stations, distribution lines appear illustrated to both household and industrial consumers. Depending on environmental conditions and the load demands are enabled the available energy sources: power plants, small hydro, wind turbines, biomass power plants and photovoltaic plants. These sources have the level indicators for the active power and the reactive power that can be modified and tracked at the consumer and the producer (Fig. 15.21s). The most important resource to increase the energy efficiency is to reduce own energy consumption. An effective tool for studies and planning of the development of electricity distribution networks, and to reduce the domestic electricity consumption, is the analysis of the load profiles. This verification process is repeated as many times as necessary, until the application will work correctly and as it is specified (Fig. 15.21r). The application contains circuit breakers and disconnectors that will act as in the “General Regulation” of the maneuvers in the electrical installations and EPS (Fig. 15.21s). Application 2
15.7.2 Diesel—Wind Turbine Hybrid EPS to Supply the Industrial Consumers 15.7.2.1
Concept Diagram of the Application
The concept diagram of this application is shown in Fig. 15.22.
15.7.2.2
Description of the Operation Diagram
The scheme shown in Fig. 15.23 considers the case of an industrial consumer powered from three different energy sources: wind, diesel, national EPS. Energy management strategy implemented is simple: the industrial consumer will be feed from the RES available, but not from two or more sources.
15
SCADA Applications for Electric Power System
Fig. 15.22 Concept diagram of the application
Fig. 15.23 EPS based on RESs
605
606
F.M. Enescu and N. Bizon
In this conception, the wind system is a priority, following on same level diesel generator and EPS. If the wind starts and can rotate the wind turbine, then the electric consumers will switch to energy produced by the wind turbine. If the wind stops, then the consumer will be automatically feed from the other two sources, whichever is convenient based on the specification: if the chosen second source will be interrupted, then it will automatically switch to the third variant. The energy management strategy will switch the consumer to energy line of the wind turbine if the wind will appear again. In the scheme shown in Fig. 15.3 were also introduced three buttons as follows: – “WIND ON” AND “WIND OFF” to start/stop voluntary the wind farm; – “APN ON” and “APN OFF” to start/stop the industrial consumers; – “GENERAL_ON” “GENERAL_OFF” to start/stop automatically the power flow between the national grid and diesel generator. If wind turbine works, then the network between “SWITCH GEN/NETWORK” and “APN” switch is not active for reasons related to consumer protection. The panel with cursor simulates wind speed which is established in a certain margin and which can turn on/off the wind turbine.
15.8
Conclusion
The focus on new IT strategies represents one of the basic premises of increasing the efficiency and safety of the national energy system. Advantages and disadvantages of using a SCADA system were illustrated by applications shown in this chapter. The first application proposed for River Hydro arrangement presents a complex SCADA architecture. For the existing architecture, it was proposed some improvements in order to increase its performance: – Implementing a network ring-type redundant with fiber-optic which will facilitate communication between HED and HEC network, instead of the radio communication (which will remain operational as standby solution); – Replacing of the existing equipment (DAS900 in Figs. 15.14, 15.15, and 15.16) with a new equipment (2 redundant servers), which will eliminate the existing DAS dysfunctions and in addition will implement new functions that are necessary in exploitation of the hydro facilities: (1) a bi-directional communication (for the purpose of monitoring and transmission of commands) with the automation systems (as HA1, HA2, SG, and Barrage) using the Open Platform Communications (OPC) industrial protocol; (2) a dedicated server for SCADA applications which has a very large number of standardized protocols for direct communication with devices of automation used in monitoring and driving of the industrial processes.
15
SCADA Applications for Electric Power System
607
– Simplification of the SCADA architecture (Fig. 15.17) by (1) using the DROP software on the servers, (2) ensuring the direct connection between equipment without intermediate software, (3) ensuring a high speed communication with existing equipment based on the standard protocols, and implementing of high security levels for servers; – Assuring of local drivers and OPC connectivity (Fig. 15.18) will permit (1) the functioning without interruption using the hot-plug redundant servers, (2) a future expansion and easy software development, without constraints, and (3) a simple interfacing of the equipment starting from the top-level to the bottom-level. The second application proposed in this chapter has shown the stepwise implementation of a medium-voltage EPS based on RESs. The application is implemented using the HMI interfaces of the programming environment VIJEO CITECT®. This application provides an improved and friendly interface, with a high degree of integration of the schemes that facilitates easy operation of the system. Compared to the initial application, this new application enables: – Monitoring and handling of the active and reactive power level indicators; the power factor is possible to be evaluated and improved in real-time; – Monitoring is done both for consumers and producers using visualization in both value and graphic representations. – The application will allow a quick view of the irregularities occurring in the system, so it will be possible to eliminate them (Fig. 15.21q, r, s). Another application shows a diesel—wind turbine hybrid EPS used to supply the industrial consumers. This system has the advantage that the wind speed is permanently monitored and adjusted if is the case. The advantages of this supply system are the low costs and the flexibility in choosing of best variants to use the RESs in relation with the environment conditions. Even if the initial costs related to SCADA implementation are relatively high, these will be recovered in short time due to the advantages offered by the SCADA: – improved operation of the installation and of the process, whence a number of savings will appear as a result of the optimization operation; – increased productivity; – full operational control of the process being monitored; – safety, robustness, minimum expenditure in exploitation; – improved system safety due to a better information of the dispatchers and improved supervision; – energy savings due to optimization of the manufacturing processes; – improving the access to information, and their quality and visualization in real-time; – extremely easy to extend the system; – the possibility to be connected with different systems.
608
15.9
F.M. Enescu and N. Bizon
Trends in the Evolution of SCADA Systems
The trends in the evolution of SCADA systems refer to communications, RTU and MTU as follow: – The development of new communication technology to optimize the equipment used as size and energy consumption, making it possible to integrate them even in the RTU; – The upgrade of the radio equipment with numerical subsystems to perform the auto-calibration, which reduce the starting time of the transmitters up to seconds; – The use of the geostationary satellites for SCADA systems using portable broadcasting stations which have the price comparable with that of mobile phones; – The use of communications on fiber-optic, which offers the advantages of high speed transmission, increased safety and confidentiality; this technology is well adapted to the needs of communications in field of energy as well; – The development in the RTU units based on embedded computers allowed more flexibility related to its functionality as controllers, meters, energy management units etc. The later developments of the MTU units are focused along three levels: (1) improved operator interface (graphical user interfaces, windows, objects, graphics-oriented etc.), (2) increased autonomy of the intelligent self-trained systems, and (3) improved car-to-car communication based on LAN networks.
References 1. J. Figueiredo, J. Sa da Costa, A SCADA System for Energy Management in Intelligent Buildings, Energy and Buildings, vol. 49, pp. 85–98, 2012. 2. J. Tomic, et al., Smart SCADA System for Urban Air Pollution Monitoring, Measurement, vol. 58, pp. 138–146, 2014. 3. O. Barana, et al., Comparison between Commercial and Open Source SCADA Packages - A Case Study, Fusion Engineering and Design, vol. 85, pp. 491–495, 2010. 4. A. Gligor, T. Turc, Development of a Service Oriented SCADA System, Procedia Economics and Finance, vol. 3, pp. 256–261, 2012. 5. Z. Vale, et al., Distribution System Operation Supported by Contextual Energy Resource Management Based on Intelligent SCADA, Renewable Energy, vol. 52, pp. 143–153, 2013. 6. P. Novak, et al., Integration Framework for Simulations and SCADA Systems, Simulation Modelling Practice and Theory, vol. 47, pp. 121–140, 2014. 7. A. Daneels, What is SCADA, International Conference on Accelerator and Large Experimental Physics Control Systems, Trieste, Italy, pp. 339–343, 1999. 8. A. Rezai, et al., Secure SCADA Communication by Using a Modified Key Management Scheme, ISA Transactions, vol. 52, pp. 517–524, 2013. 9. B. Genge, C. Siaterlis, Physical Process Resilience-Aware Network Design for SCADA Systems, Computers and Electrical Engineering, vol. 40, pp. 142–157, 2014.
15
SCADA Applications for Electric Power System
609
10. F.M. Enescu, Start to Design the HMI/SCADA - Applications (in Romanian), ISBN 978-606-560-425-4, Ed. Univ. Pitesti, 2015. 11. A. Fahad, et al., PPFSCADA: Privacy Preserving Framework for SCADA Data Publishing, Future Generation Computer Systems, vol. 37, pp. 496–511, 2014. 12. E. Ozdemir, M. Karacor, Mobile Phone Based SCADA for Industrial Automation, ISA Transactions, vol. 45, pp. 67–75, 2006. 13. M. Schlechtingen, et al., Wind Turbine Condition Monitoring Based on SCADA Data Using Normal Behavior Models, Part 1: System Description, Applied Soft Computing, vol. 13, pp. 259–270, 2013. 14. M. Schlechtingen, I.F. Santos, Wind Turbine Condition Monitoring Based on SCADA Data Using Normal Behavior Models, Part 2: Application Examples, Applied Soft Computing, vol. 14, pp. 447–460, 2014. 15. C.D. Dumitru, A. Gligor, A Management Application for the Small Distributed Generation Systems of Electric Power Based on Renewable Energy, Procedia Economics and Finance, vol. 15, pp. 1428–1437, 2014. 16. R.H. McClanahan, The Benefits of Networked SCADA Systems Utilizing IP Enabled Networks, Rural Electric Power Conference, 2002 IEEE, pp. C5- C5_7, 5–7 May 2002. 17. http://www.7t.dk/igss/default.asp - IGSS SCADA System - 2009. 18. http://www.7t.dk/igss/default.asp?showid=374 - IGSS Online SCADA Training - 2009. 19. http://www.7t.dk/free-scada-software/index.html - IGSS Free SCADA Software - 2009. 20. C.P. Nicolaou, et al., Measurements and Predictions of Electric and Magnetic Fields from Power Lines, Electric Power Systems Research, vol. 81, pp. 1107–1116, 2011. 21. I. Morsi, L.M. El-Din, SCADA System for Oil Refinery Control, Measurement, vol. 47, pp. 5–13, 2014. 22. A. Salihbegovic, et al., Web Based Multilayered Distributed SCADA/HMI System in Refinery Application, Computer Standards & Interfaces, vol. 31, pp. 599–612, 2009. 23. R. Ahiska, H. Mamur, A Test System and Supervisory Control and Data Acquisition Application with Programmable Logic Controller for Thermoelectric Generators, Energy Conversion and Management, vol. 64, pp. 15–22, 2012. 24. C. Damian, et al., Analysis of an Event in an Electric Transformer Station by Means of the SCADA System, Procedia Technology, vol. 12, pp. 740–746, 2014. 25. http://www.automation.ro/pdf/fisa_prez_SCADA_somes.pdf. 26. G. Clarke, D. Reynders, Practical Modern SCADA Protocols: DNP3, 60870.5 and Related Systems, Elsevier Ltd, 2003. 27. http://www.scada.ro/index.php%3Foption%3Dcom_content%26id%3D116%26lang%3Dro. 28. D. Bordea, Smart Grid or Secure Grid - New Opportunities and New Risks (in Romanian), http://cnrcme.ro/foren2014/presentations/RTF%202/pdf/02%20Dan%20Bordea%20foren% 202014.pdf. 29. K. Jamuna, K.S. Swarup, Optimal Placement of PMU and SCADA Measurements for Security Constrained State Estimation, Electrical Power and Energy Systems, vol. 33, pp. 1658–1665, 2011. 30. G.N. Korres, N.M. Manousakis, State Estimation and Bad Data Processing for Systems Including PMU and SCADA Measurements, Electric Power Systems Research, vol. 81, pp. 1514–1524, 2011. 31. W. Yang, et al., Wind Turbine Condition Monitoring by the Approach of SCADA Data Analysis, Renewable Energy, vol. 53, pp. 365–376, 2013. 32. P. Cross, X. Ma, Nonlinear System Identification for Model-Based Condition Monitoring of Wind Turbines, Renewable Energy, vol. 71, pp. 166–175, 2014. 33. E Luiijf, et al., Assessing and Improving SCADA Security in the Dutch Drinking Water Sector, International Journal of Critical Infrastructure Protection, vol. 4, no. 3–4, pp. 124–134, 2011. 34. http://www.citect.com/ - CITECTSCADA -2009. 35. http://www.free-scada.org/ - Free SCADA - 2009.
Chapter 16
Effect of Geomagnetic Storms on Electric Networks Daniel Mayer and Milan Stork
Abstract The beginning of this chapter summarizes the findings of the physical nature of Earth’s magnetic field and its measurements in geomagnetic observatories. Experience shows that the variation of the geomagnetic field may affect the operation of various distracting electronic devices, such as systems for the transmission of electrical energy. It is derived an algorithm for calculating the currents that can be induced in the power lines and can lead to the violation of stability of the system. The algorithm is demonstrated in the illustrative example.
16.1
Introduction
Solar activity causes changes (variation) of the magnetic field on the surface of the Earth. If these changes are insignificant, affect life on Earth, or the technical equipment. Exceptionally, however, a large and rapid variations of the geomagnetic field, the so-called magnetic storms, which have the character of shock waves, may adversely affect living organisms (including humans), and the technical systems that people use. There are also pessimistic forecast of extremely strong magnetic storms, according to which may be threatened the modern civilization. The sophisticated technology you use, the more this technique is vulnerable and thus the threat of strong magnetic storms becomes actual. Geomagnetic variations, one
D. Mayer (&) Department of Theory of Electrical Engineering, University of West Bohemia, Plzen, Czech Republic e-mail: [emailprotected] M. Stork Department of Applied Electronics and Telecommunications, University of West Bohemia, Plzen, Czech Republic e-mail: [emailprotected] © Springer International Publishing AG 2017 N. Mahdavi Tabatabaei et al. (eds.), Reactive Power Control in AC Power Systems, Power Systems, DOI 10.1007/978-3-319-51118-4_16
611
612
D. Mayer and M. Stork
Fig. 16.1 The course of the vertical component of the geomagnetic field on 22 March 2013, as measured by certified geomagnetic observatory at Budkov, Czech [10]
cannot avoid, but they can be a threat to some approximation of the short-term predictor. Magnetic storms can cause serious disorders in technical systems. Reduce the quality of the radio connections, may damage the cable telecommunication networks and satellite communications systems and thus block the data transmission, can interfere with navigational systems and thus jeopardizing the safety of air transport. Induced currents accelerate corrosion of pipelines transporting oil and gas, on railways may disrupting signaling devices and, in particular, threatens the stability of large electrification systems which can cause widespread blackouts and in limit case can cause a total collapse of the system and global chaos, with disastrous consequences [1–8]. Geomagnetic activity is monitored by the global network of regional centers [9], one of which is in Prague, Czech, with geomagnetic observatories in the Budkov near Prachatic. Geomagnetic measurements carried out here continuously, at intervals of one minute, or every second, with an accuracy of 0, 1 [nT] and the results are forwarded to the World Data Center in Boulder (USA) and to the International Centre in Edinburgh (Scotland). Geophysical Institute issues from 1994 forecasts of geomagnetic activity for Central Europe, Fig. 16.1, [10]. The issue of the geomagnetic field is also successfully engaged in work teams Institute of Atmospheric Physics of AS CR and the Astronomical Institute of Academy of Science, Czech [11]. This chapter presents a conceptual design that allows prediction of electrical power system emergency conditions by magnetic storms.
16
Effect of Geomagnetic Storms on Electric Networks
16.2
613
The Emergence of Geomagnetism and Geomagnetic Storms
16.2.1 Earth Magnetic Field Internal geomagnetic field: W. Gilbert is more than 400 years ago believed that at the center of the Earth is a permanent magnet—a magnetic dipole which generates a magnetic field on the surface of the Earth. In the northern hemisphere is the south magnetic pole and the axis of magnetic dipole is inclined by 11.8° of the Earth’s axis (Fig. 16.2). For dipole configuration performed in 1828, C.F. Gauss mathematical solution of the geomagnetic field using the spherical harmonic analysis. C. F. Gauss also developed a method of measuring absolute geomagnetic field and together with A. Humboldt and E. Weber established a worldwide network of observatories for continuous measurement of the geomagnetic field. The first geomagnetic observatory in Central and Eastern Europe was founded in 1839, K. Kreil in Prague. Measurement of the geomagnetic field and plotting global magnetic maps in different epochs showed that the geomagnetic field is not stationary, but with time changes—we are talking about secular variations. These are changes that are caused by physical processes in the Earth’s core. These changes are very slow, are detected on the scale of decades, and our problem geomagnetic field effect on the electricity system are completely irrelevant [12, 13]. Fig. 16.2 Geomagnetic field undisturbed with the solar wind
614
D. Mayer and M. Stork
The idea of geomagnetism with the permanent dipole magnet was abandoned after finding that the Earth’s core there is a temperature above the Curie temperature; the ferromagnetic magnetic domains are randomly oriented. For example, at a depth of 100 km below the surface temperature is of the order of 1000 °C, which is significantly higher than the Curie temperature of the ferromagnetic material. In the first period at least century model was designed to explain geomagnetism—called geodynamo [12]. Its physical nature is the rotational flow liquid outer core of the Earth, which is composed mainly of iron and nickel. Since it is an electrically conductive environment that in a magnetic field (maybe a weak magnetic field such as the Sun), induce in it a very high electric current (about 109 A), which generates the geomagnetic field. Model of geodynamo and his subsequent mathematical description underwent a complex evolution, in which participated many leading astrophysicists. Soon he had shown that simply cannot be represented geodynamo, with an axially symmetrical, stationary flow, as initially anticipated. The most accurate model, developed by Glatzmaier and Roberts [14], has a much more complex configuration: there are the three-dimensional, turbulent, unsteady magneto-hydrodynamic process, in which the applied laws of electromagnetic induction, the laws of fluid flow, electromagnetic and gravitational forces, Coriolis force, temperature convection and electric currents in the ionosphere. Although contemporary computers allow only an approximate solution of this complex dynamic process, computer simulation can be explained by secular variation and even inversion (i.e. change of polarity) geomagnetic field in the distant geological past of the Earth. The development of the theory of geodynamo deals e.g. the work of Chicago astrophysicist professor Parker [15, 16]. In order to understand the mechanism of the Earth’s geomagnetic field generation including its chaotic reversals several attempts to create an adequate enough, but sufficiently simple theory of mathematical models have been undertaken [14]. Some models in form of physically motivated sets of nonlinear ordinary differential equations were derived. Also a variety of different dynamo models based on a physical analogy, manifesting strongly nonlinear phenomena, has often been used for explanation of the origin and chaotic-like behavior related to stellar and planet magnetic fields [17]. The irregular changes in the Earth’s magnetic field polarity occurring in irregular intervals of order hundred thousands to millions of years have been widely investigated by means a concept of geodynamo in the second half of the last century. Results obtained by computer simulation are compared with measured values obtained by geomagnetic observatories, and also using satellites. About secular variations of the geomagnetic field in the past centuries, resp. millennia inform archaeomagnetic studies based on the determination of remanent magnetism burned archaeological objects of known age. For example, the samples derived from prehistoric kilns or hearths to determine direction and magnitude of the magnetic intensity of the geomagnetic field, in which there was examined object at the time of burning. Information about the geomagnetic field of the Earth before tens of
16
Effect of Geomagnetic Storms on Electric Networks
615
Fig. 16.3 Simplified view of the Earth’s interior
thousands to millions of years of research provide paleomagnetic remanent magnetization Quaternary of volcanic and sedimentary rocks, geological and radiocarbon dating methods, based on the decay of atmospheric carbon 14C. This will also find information on the geomagnetic poles wandering around the geographic poles in various geological epochs and then on multiple geomagnetic polarity inversion [18–24]. Intuitively, any real-world system which is able to convert mechanical power into its electric or electromagnetic form can be considered as a dynamo. When stable equilibrium between the mechanical input and the generated output electrical current is reached we get a self-sustaining magnetic dynamo. In theory of the Earth’s geomagnetic field generation the physical concept of geodynamo has been introduced. The simplified view of the Earth’s interior, which structure is known to play a crucial role in this context, is displayed in the Fig. 16.3. The key idea of [25], the most popular Larmor’s explanation for planetary magnetism is very simple. Any real-world dynamo generates a current. Every current produces magnetic field accumulating an energy. Therefore input power, both mechanical and magnetic, must be put into the system in order to maintain its motion. This is the reason why an input torque has to be applied, because otherwise the dynamo would run down due to Ohmic heating. Now comes a new idea concerning internal power-informational interactions and well known destabilizing effects of the positive feedback. Let’s assume existence of a current-carrying wire around the rotation axle of a rotating electromechanical system. It is easy to imagine that under certain conditions the resulting magnetic field reinforces the external field already present. The stronger field increases the Lorenz force and thus increase of the current follows. The growing current in turn creates an even stronger field, which increases the current again, and so forth. It can be demonstrated that under proper conditions the induced magnetic field will completely supplant the external field and as a result only mechanical energy needs be supplied. At about the same time as the Larmor’s theory of planetary magnetism was published, evidence for chaotic-like reversals in the Earth’s magnetic field appeared and began to accrue. This evidence came from paleomagnetic record, particularly from lava flows. This record indicates the presence of magnetic field over some billions of years [26], necessitating a self-regenerative energy source like
616
D. Mayer and M. Stork
a dynamo. The paleomagnetic record also indicates numerous reversals over geological history. The geophysics community has long conjected that a geodynamo in the outer core (Fig. 16.1) is the source of the Earth’s geomagnetic field. Now, it became clear that the cause of Earth’s main magnetic field is the turbulent flow of conducting material deep in the interior, which can only mean the outer core. It means that Larmor’s hypothesis was correct. About the question of the energy source that stimulates this turbulent flow, or equivalently about the physical nature of energy to drive the geodynamo with chaotic behavior, still remains considerable uncertainty. The problem is subject of further structurally oriented experimental as well as theoretical physical research. In order to understand the mechanism of the magnetic field generation in the Earth’s outer core the so called “dynamo effect” plays a crucial role and requires further research efforts. An alternative fundamentally different research direction, totally ignoring any information about the actual physical system structure seem to be possible [25, 27–31]. One such an approach is based on a very simple structure of the electromechanical system illustrated by the Fig. 16.4. The physical analogy of the geodynamo with different modifications of the one-disc, as well as the two-disc unipolar dynamos including the Rikitake system depicted in the Fig. 16.4 is obvious [29]. The structure of the double-disc dynamo, in which the field of one dynamo is applied to a second dynamo and vice versa in such a way that a destabilizing nonlinear positive feedback arises can exhibit sudden reversals and chaotic-like field changes. However, it is fair to notice that in the Earth’s interior context such a model is structurally highly unrealistic. It follows
Fig. 16.4 Physical structure of the Rikitake dynamo
16
Effect of Geomagnetic Storms on Electric Networks
617
Fig. 16.5 Vector of the geomagnetic field and its components: B = iBx + jBy + kBn
that the paleomagnetic record, particularly from lava flow, as well as observed data produced by complicated magneto-hydro-dynamical effects including geomagnetic inversion, cannot be expected to be completely explained by such simple models. For instance, Hide showed in [28] that neglecting effects of mechanical friction is unwarranted for the two-disc dynamo. Moreover, in case of a symmetric mechanical friction the conventional model of the two-disc dynamo was due effects of a positive feedback proven to be structurally unstable and hence incapable to produce any chaotic oscillations. In [30] a further natural question was discussed, whether the last statements concerning instability and chaotic oscillations remain valid when the azimuthal current distribution will be considered. In such a way interesting positive answers for modified two-disc Rikitake as well as one-disc dynamo have been gained. Geomagnetic field generates the Earth’s magnetosphere. Magnetic induction vector B has components Bx and By lying on the surface of the Earth (the resultant of the horizontal component Bh) and the vertical component Bn, which has a direction normal to the surface of the earth (Fig. 16.5). In geophysics usually determines the vector B from angles a (magnetic declination) and b (magnetic inclination) and from the normal component Bn. The vector magnetic field on the surface of the Earth is at the poles B * 60 lT, the magnetic equator B * 30 lT and in Czech is B * 44 lT [10].
16.2.2 Solar Activity External geomagnetic field: The magnetic field effects contribute their electric currents in the ionosphere and magnetic field of the Sun and other planets. Over a hundred years ago it was found that there is a fairly rapid geomagnetic field variations that correlate with solar activity and show some periodicity—we are talking about variations of daily, annual, eleven year and lunar. For our problem are crucial
618
D. Mayer and M. Stork
Fig. 16.6 Magnetosphere deformed with onslaught of the solar wind
variations, which have their origin in solar flares, when there is a surge of coronal mass containing electrons and high-energy protons and heavier ions, which then spread with great speed in interplanetary space. This product solar activity is known as a CME (Coronal Mass Ejection) and its flow is called the solar wind. The delay between the occurrence of solar flares and geomagnetic field variations usually lasts one to two days. The process flow of the solar wind in the interplanetary space undertook the marking space weather. Input electrically charged particles of the solar wind around the Earth to prevent the geomagnetic field [12, 31–34]. Solar wind interaction with the magnetosphere is reflected by the fact that the action of the Lorentz force is electrically charged particles moving in the direction of the magnetic field lines and flow around the magnetosphere. Thus the solar wind distorts the shape of the magnetosphere (Fig. 16.6): the front (daily, windward) side of the magnetosphere is compressed to about ten Earth radii (about 60,000 km) and dark (night, leeward) side is extended into a long tail of magnetosphere which extends far beyond the orbit of the moon [12]. Geomagnetism thus shields Earth from solar wind and thus protects the Earth’s biosphere. Without geomagnetism would not exist on our planet, the existing forms of life. Just around the geomagnetic poles, where the magnetic induction vector approximately perpendicular to the surface of the Earth, created channels through which the solar wind in particles can partially penetrate into the deeper layers of the magnetosphere and endanger passengers and crew transoceanic years. If CME has a sufficiently high energy of solar wind hits the Earth’s magnetosphere, the geomagnetic activity reflected not only in the polar regions, but some of this energy is transferred to the magnetosphere, resulting in rapid variations of the geomagnetic field. Their intensity is different: from minor defects, which occur several times a day, to strong variations, caused by the solar wind in shock wave— called magnetic storms [19, 35, 36]. There are accompanied by polar lights. Aurora is luminous phenomena in Earth’s atmosphere. They are caused by the excitation of atoms of the atmosphere in heights between 70 and 1000 km with
16
Effect of Geomagnetic Storms on Electric Networks
619
particles of the solar wind. They can watch them as green glowing walls and columns that changes in seconds to minutes. During increased solar activity constitutes the entire auroral ring around the northern and southern magnetic field. Minor variations in the time of day quiet geomagnetic field are 20–30 nT and during the day, resp. year show some periodicity. On the other hand, for big magnetic storms geomagnetic field reaching up to hundreds nT. In contrast, the changes of the internal magnetic field is changing in units nT per year. In order to perform these precise measurements are supplemented by continuous variation measurements with accurate absolute measurements. Note that the link between solar and geomagnetic activity is not as a close as previously assumed. The dynamics of the magnetosphere CME activity is complicated and can be described mathematically by a system of partial differential equations of Navier-Stokes, Maxwell’s equations and the equations for heat conduction. This is called a coupled nonlinear problem that has so far been found only an approximate solution. Any contact with the Earth’s magnetosphere significant object CME with consequent development of geomagnetic storms beyond our control, but we try to construct mathematical models whose simulations of this natural phenomenon can predict the time for us to perform technical measures to prevent greater damage.
16.3
Definition of the Problem
In this work we calculate parasitic currents during geomagnetic storms can induce electricity wires in the network. The question of the influence of variations of the geomagnetic field on the electricity supply system, caused by solar activity, became current at the end of the last century, in connection with several big accidents. Examples include massive power supply outage on the east coast of North America in October 1989. The largest solar flare observed followed by a strong magnetic storm (B * 3000 lT) registered British astronomer R. Carrington in August 1859. Since the introduction of electricity into practice at that time was at the very beginning, although magnetic storm knocked out telegraph network in the entire northern hemisphere, but did not cause more damage. At a time when the entire human civilization dependent electrical and electronics might causes incalculable similar storm damage. In recent times there has been a strong magnetic storm, coupled with the rapid changes in the magnetic field 13 March 1989, in the Canadian province of Quebec. Total mains power failures hit Ontario, British Columbia and parts of North America. After a period of nine hours or more were 6 million people without electricity. After 24 h have been taken out of operation US navigation satellites. That event began systematically monitor space weather. List of magnetic storms, which damaged electrical equipment in different parts of the country, is given in [1]. Research into this phenomenon is discussed by many authors, among whom a significant proportion by Finnish authors [1, 2, 37–41]. In addition to electrical
620
D. Mayer and M. Stork
systems were also considered telecommunications networks, network for transport oil and gas [16] and the rail track. Despite great efforts has been devoted to resolving this issue, no matter reliable prediction of the effects of magnetic storms closed. Physical causes harm of electricity networks are parasitic currents induced by geomagnetic variations in the branches of the transmission network—the abbreviation indicate the GIC (Geomagnetically Induced Currents) [42]. Geomagnetic variations have the character of shock waves, which are changes in seconds. GIC are superimposed together with the operating currents in the lines electricity system that transmits active and reactive power. In terms of electrical networks, these changes are very slow and therefore GIC denoted as quasi-DC currents. It follows that it is sufficient to count only with the resistance of the transmission lines, while the effect of inductance, capacity coupling and skin effect are negligible. GIC causing more load of conductor by current, increase the transmission losses, increase the voltage drop in the lines and cause oversaturation magnetic circuits of transformers and thus may cause a violation of stability and eventually black-out of system.
16.4
Geomagnetic Field on the Surface of the Earth and Its Rapid Variations
For electromagnetic induction GIC in the branches of the network is important the rate of time of normal (vertical) components of Bn(t) of vector magnetic induction B. The magnetic storm on Earth is delayed eruption on the Sun for about 25–60 h. Drawing on many years of observations of solar and geomagnetic variations subsequent measurements can thus be predicted with a certain probability what will be Bn(t). Estimated process Bn(t) will be for us starting point for a further specified size calculation algorithm GIC.
16.5
Calculation of Geomagnetically Induced Currents
Investigated power system will be replaced by network N distributed in the plane (x, y), which consists from n irregular polygons having m branches and p nodes etc. (Fig. 16.7). We know: The topological structure of the network N, i.e. the graph G, the metric structure of network N, i.e. the geometric dimensions of the network N which are given by the coordinates of the vertices of each polygon and its physical structure, which are given by the resistance Ri (i = 1, …, m) of each branch of network. On the network N operates time-varying geomagnetic field; we know its axial (vertical) component Bn(t) and we assume that in the field (x, y) of networks
16
Effect of Geomagnetic Storms on Electric Networks
621
Fig. 16.7 Polygonal network N—general model of the transmission system
N, the value of Bn(t) does not change. Using knowledge of matrix analysis of electrical circuits [43, 44], let the ith polygon containing a loop ci is irregular and his vertices have coordinates (x1, y1), ….., (xp, yp). Polygons graph G is a set of n independent loops {c1, …, cn}. According to the second Maxwell’s equations apply to the ith loop ci. I
dUi ; i ¼ 1; . . .; n dt
Edl ¼
ð16:1Þ
ci
where E is the intensity of electric field and Z Ui ¼ Bn dS ¼ Si Bn
ð16:2Þ
Si
is the magnetic flux coupled with loop ci. Equation (16.1) can be written as X k
uk ¼ Si
dBn i dt
ð16:3Þ
where uk is the voltage of the kth branch, which is incident with ith loop ci and Si is the content of the ith polygon. It is calculated from the formula 1 Si ¼ x1 ðy2 yp Þ þ x2 ðy3 2 1 y1 ðx2 xp 1 Þ þ y2 ðx3 ¼ 2
y1 Þ þ . . . þ x1 ðy1 x1 Þ þ . . . þ yn ðx1
yp 1 Þ
xp 1 Þ
ð16:4Þ
622
D. Mayer and M. Stork
The Eq. (16.3) can be written in matrix form Cu ¼ FðtÞ
ð16:5Þ
where C(n, m) is the second incidence matrix, which is determined by the topology of network N. It is a rectangular matrix whose rows (columns) correspond to loops (branches) of G. Her elements are: +1 (or –1) if kth branch is incident with ith loop consistently (or vice versa), or 0 if the kth branch is not incident with ith loop (i = 1, …, n, k = 1, …, m), u is the vector of voltage branch and F(t) is the vector of induced voltages in the loops. 2
3 S1 dBn F(t)¼4 . . .:: 5 dt Sn
2
3 u1 u¼4 . . .:: 5; um
ð16:6Þ
We introduce the vector of currents in the branches i and the vector of current in the loop iʹ 2
3 i1 iðtÞ¼4 . . .:: 5; im
3 i01 i0 ðtÞ¼4 . . .:: 5 i0n 2
ð16:7Þ
Both matrices i and iʹ are linked together with a second incidence matrix i ¼ C T i0
ð16:8Þ
where the upper index “T” denotes the matrix transpose. Between vector of branches voltage u and a vector of branch currents i is true the Ohm’s Law, which written by matrix is u ¼ Ri
ð16:9Þ
where R is a diagonal matrix of branch resistances: R = diag [R1, ….. Rm]. We multiplying Eq. (16.9) by the left matrix C and substituting the vector i from Eq. (16.8) Cu ¼ CRi ¼ CRC T i0
ð16:10Þ
with help of Eq. (16.5), we determine FðtÞ ¼ CRC T i0
ð16:11Þ
Thence calculate the vector of loop currents iʹ and using Eqs. (16.8), (16.11) and (16.5) we find vector of instantaneous values of GIC in the branches of the network N
16
Effect of Geomagnetic Storms on Electric Networks
dBn iðtÞ ¼ C T CRC T S dt
623
ð16:12Þ
Using the vector i(t) can be easily determined the vector of rms current in the branches Ief. dBn ¼ Af ðtÞ iðtÞ ¼ C T CRC T S dt
16.6
ð16:13Þ
Illustrative Example
The network N in the plane (x, y) has the topological structure characterized by graph G in Fig. 16.8, his metric structure is determined by the coordinates of nodes of the graph G: (0, 0), (220, 40), (180, 180), (30, 140) and (80, 70) km, and the physical structure of the network N is determined by the resistance of branches: R1 = 32.1 X, R2 = 50.2 X, R3 = 32.1 X, R4 = 34.6 X, R5 = 19 X, R6 = 24.1 X, R7 = 31.7 X, R8 = 33 X. Measurements were found the normal component of the magnetic induction of the geomagnetic field with a time step of 1 s; graph of Bn(t) is the upper Fig. 16.7, determine the immediate and effective values GIC in branches of network N.
Fig. 16.8 Network N, which is solved
624
D. Mayer and M. Stork
0.8 0.6 0.4
f(t)
0.2 0 -0.2 -0.4 -0.6 -0.8 -1 0
20
40
60
80
100
120
time [s]
Fig. 16.9 Course of the normal component of the magnetic induction Bn(t) depending on time (top), derivative of Bn(t) (middle) and f(t) (bottom)
16
Effect of Geomagnetic Storms on Electric Networks
625
Using the spline interpolation and subsequent numerical derivative of the function Bn(t) can be found dBn/dt. This function was normalized by its maximum size, max[|dBn/dt|] = 60 [nT]. 1 dBn n Graph of standardized function f ðtÞ ¼ maxj1dBn =dtj : dB dt ¼ 60 : dt is in the bottom part of Fig. 16.9. Oriented graph G (arbitrarily). The second incidence matrix, matrix of resistances of branches and vector of loops of polygon surfaces of network are 2 3 1 0 0 0 1 1 0 0 60 1 0 0 0 1 1 0 7 6 7 C¼6 7 40 0 1 0 0 0 1 15 0
1
1
1
R ¼ diag½ 32:150:232:134:61924:131:733 ½X
ð16:14Þ
3 4:5 6 6:1 7 7 6 S¼6 7 109 m2 4 9:2 5 2
6:25
Substituting of matrix (16.14) into Eq. (16.13) get the vector of instantaneous values of currents in the branches of the network i(t). Can easily calculate the rms currents in the branches of network Ief, in the time interval t 2 〈0, 130〉 [s]. 3 3 2 2 2:9 9:5 6 2:8 7 6 9:4 7 7 7 6 6 6 3:8 7 6 12:6 7 7 7 6 6 6 3:4 7 6 11:2 7 7 7 6 6 ð16:15Þ iðtÞ ¼ Af ðtÞ ¼ 6 7f ðtÞ ; Ief ¼ 6 0:53 7 ½A 7 6 6 1:7 7 6 0:03 7 6 0:09 7 7 7 6 6 4 0:98 5 4 3:2 5 0:42 1:4 The time evolution of currents i(t) [from i1(t) to i8(t)] are shown in Fig. 16.10.
16.7
Effect of Changes in the Earth’s Magnetic Field on the Reliability of the Distribution Power Network
In the previous section we explain the physical causes of time-variable Earth’s magnetic field and subsequent we found the solution of relatively slow variable phenomenon in the electricity network. We formulate the algorithm of calculation of the values of the quasi-direct current that induce in the branches of the electricity
626
D. Mayer and M. Stork
Fig. 16.10 Currents i1 (top) to i8 (bottom) for network according Fig. 16.8
network. These currents generate electric power which is completely converted to electric heat in conductors of transmission lines and in the winding of the transformers in the electricity system. These currents are purely parasitic and do not participate on the transmission of power from generators to appliances, as well as reactive power during normal operation of network. During magnetic storms (i.e. in the relatively quick changes in the Earth’s magnetic field), this can cause the dissipation of heat damage to wires of transmission lines and windings of the transformers. This will avoid using the protection network, but at the cost of the interruption of power supply, which leads to a collapse of the electricity system. Let in the ith transmission line is the current pffiffiffi iik ðtÞ ¼ Ioi þ 2fIi k sinðkxt þ ui k g;
k ¼ 1; 2; . . .; 1
ð16:16Þ
where I0i is the value of quasi-direct current components induced by changing the magnetic field of the Earth, and the Iik and uik are the effective value of the kth harmonic currents and their phase shift, both in the ith wire of transmission line. Then the dissipated power in the ith wire, according to the Parseval identity, may be expressed as a Joule’s power Pi ¼ Ri
1 X k¼0
Iik2
ð16:17Þ
16
Effect of Geomagnetic Storms on Electric Networks
627
A more complicated case occurs in the transformer winding. A quasi-direct current cause not only increases Joule losses in the windings, but also causes reduction of magnetization of magnetic circuit of the transformer. Decreases permeability of the magnetic circuit and also reduction of the inductance of the windings can leads to a decrease the stability of the electricity system and cause also reactive power increasing.
16.8
Conclusion
The input value when calculating the GIC is the time course of the vertical component of the magnetic induction Bn(t), which can only be estimated based on a larger number of long-term measurements. It is therefore a typical task with uncertain input data. These tasks are resolved in principle, two different approaches: stochastic (probabilistic), or the method of reliable solutions (i.e. using worst-case scenario), we assume the “worst” possible over Bn(t), even if the probability of its occurrence is small. The second of the two approaches is therefore pessimistic, but it is “safer”. The size of the GIC is not crucial size of the normal component of the magnetic intensity Bn, but the size of its derivative, i.e. dBn/dt. Based on the observation of the geomagnetic field it can be assumed that strong geomagnetic shock waves, which entitle to predict the state of emergency power system, events are very rare. On presented example it was shown, that for electrical networks are strong magnetic storms dangerous, because of induction of quasi-direct current. This leads to current overload of lines. Most dangerous is it this quasi-dc part for transformers especially for magnetic saturation [45, 46]. These phenomena can be avoided by means of protections which excludes some parts of networks from work. Should such a quite exceptional magnetic storms occurred, can the system operator to prevent destructive situations only intervention in the topological structure of the transmission system. Example, it can transform the transmission system into smaller units, or may interrupt galvanic connection some exposed line of system included with isolating transformers. Acknowledgements Milan Stork’s participation was supported by Department of Applied Electronics and Telecommunications, University of West Bohemia, Plzen, Czech Republic and by the European Regional Development Fund and the Ministry of Education, Youth and Sports of the Czech Republic under the Regional Innovation Centre for Electrical Engineering (RICE), project No. No. LO1607, the Internal Grant Agency of University of West Bohemia in Pilsen, the project SGS-2015-002 and GA15-22712S.
628
D. Mayer and M. Stork
References 1. D.H. Boteler, et al., The Effects of Geomagnetic Disturbances on Electrical Systems at the Earth’s Surface, Adv. Space Res., vol. 22, no. 1, pp. 17–27, 1998. 2. T. Makinen, Geomagnetically Induced Currents in the Finish Power Transmission System, Finnish Meteorological Institute Geophysical Publications, no. 32, pp. 101, 1993. 3. V.D. Albertson, J.N. Thorson, S.A. Miske, The Effects of Geomagnetic Storms on Electrical Power Systems, IEEE Trans., PAS-94, pp. 1931, 1974. 4. R.A. Gummow, P. Eng, GIC Effects on Pipeline Corrosion and Corrosion Control Systems, Journal of Atmospheric and Solar-Terrestrial Physics, vol. 64, pp. 1755–1764, 2002. 5. D.H. Boteler, R.J. Pirjola, H. Nevanlinna, The Effects of Geomagnetic Disturbances on Electrical Systems at the Earth Surface, Adv. Space Res., vol. 22, no. 1, pp. 17–27, 1998. 6. E.A. Eroshenko, et al., Effects of Strong Geomagnetic Storms on Northern Railways in Russia, Adv. Space Res., vol. 46, issue 9, pp. 1102–1110, 2010. 7. R.A. Gummow, GIC Effects on Pipeline Corrosion and Corrosion Control Systems, Journal of Atmospheric and Solar-Terrestrial Physics, vol. 64, pp. 1755–1764, 2002. 8. Y. Gallet, et al., Possible Impact of the Earth’s Magnetic Field on the History of Ancient Civilizations, Earth Planet, Scientific Letters, vol. 246, no. 1–2, pp. 17–26, 2006. 9. V. Haak, S. Maus, M. Koret, H. Luhr, The Earth’s Magnetic Field Screening and Monitoring, Physics in our Time, vol. 35, no. 5, pp. 218–224, 2003. 10. http://bdv.ig.cas.cz/index.html, http://rwcprague.ufa.cas.cz. 11. Z. Nemecek, et al., Dynamics of Magnetosphere and Ionization Processes and its Correlation with Solar Activity, Bull. GACR, vol. 14/4, no. 5–10, 2006 (in Czech). 12. J. Klezcek, Great Encyclopedia of the Universe, Academia, Prague, ISBN 80-200-0906-X, 2002 (in Czech). 13. R. Lanza, A. Meloni, The Earth’s Magnetism, Springer, Berlin, ISBN 3-540-27979-2, 2006. 14. G.A. Glatsmaier, P.H. Roberts, Phys. Earth Planet, Inter., vol. 91, pp. 63, 1995. 15. E.N. Parker, Cosmically Magnetic Fields, Clarendon Press, Oxford, 1979. 16. E.N. Parker, Interplanetary Dynamical Processes, Wiley-Interscience, New York, 1963. 17. D. Stevenson, Planetary Magnetic Fields, Earth and Planetary Science Letter, 6523, pp. 1–11, 2002. 18. P. Ertepinar, et al., Archaeomagnetic Study of Five Mounds from Upper Mesopotamia between 2500 and 700 BCE, Earth and Plane. Sci. Letters, vol. 357–358, pp. 84–98, December 2012. 19. J.P. Legrand, P.A. Simon, Geomagnetic Storms and their Associated Forecasts, Solar-Terrestrial Predictions - IV, Ed. by J. Hruska, et al., NOAA, Boulder, vol. 3, pp. 191, 1993. 20. G. Verbanac, et al., Four Decades of Geomagnetic and Solar Activity: 1960–2001, Journal of Atmospheric and Solar-Terrestrial Physics, vol. 72, issue 7–8, pp. 607–616, 2010. 21. P.N. Mayaut, Analysis of Storm Sudden Comment Cements for the Years 1868–1967, Journal of Geophysics, Res. 111, 1975. 22. P. Ertepinas, et al., Archaeomatic Study of Five Mounds from Upper Meoptamia between 2500 and 700 BCE: Further Evidence for an Extremely Strong Geomagnetic field from 3000 Years Ago, Earth and Planetary Science Letters, vol. 357–358, pp. 89–98, December 2012. 23. Y. Gallet, et al., On the Use of Archeology in Geomagnetism and vice-versa: Recent Developments in Archaeomagnetism, C.R. Physic, vol. 10, no. 7, pp. 530–648, 2009. 24. T. Nagata, et al., Secular Variation of Geomagnetic Total Force during the Last 5000 Years, Journal of Geophysics Research, vol. 68, pp. 5277–5281, 1963. 25. J. Larmor, Possible Rotational Origin of Magnetic Fields of Sun and Earth, Elect. Rev. 85, 1919. 26. D. Sisan, W. Shew, D. Lathrop, Lorenz Force Effects in Magneto-Turbulence, Physics of Earth and Planetary Interiors, vol. 135, pp. 137–159, 2003.
16
Effect of Geomagnetic Storms on Electric Networks
629
27. B. Skala, J. Hrusak, D. Mayer, M. Stork, On Strongly Nonlinear Phenomena in Electrical Machines, New Aspects of Systems, Proceedings of 12th WSEAS International Conference on Systems, Heraklion, Greece, pp. 120–125, July 22–24, 2008. 28. R. Hide, Structural Instability of Rikitake Disc Dynamo, Geoph. Res. Letter, vol. 22, pp. 1057–1059, 1995. 29. T. Rikitake, Electromagnetism and the Earth’s Interior, Elsevier, Amsterdam, 1966. 30. F. Plunian, P. Marty, A. Alemany, Chaotic Behavior of the Rikitake Dynamo with Symmetric Mechanical Friction and Azimuthal Currents, Proc. Roy. Soc. Lond., vol. 454, pp. 1835– 1842, 1998. 31. D.F. Webb, Solar and Geomagnetic Disturbances during the Declining Phase of Recent Solar Cycles, Adv. Space Res., vol. 16, no. 9, pp. 57–69, 1995. 32. E. Correia, R.V. de Souza, Identification of Solar Sources of Major Geomagnatic Storms, Journal of Atmospheric and Solar-Terrestrial Physics, vol. 67, issue 17–18, pp. 1702–1705, 2005. 33. J.R. Wait, Theory of Magneto-Telluric Fields, Journal Res. National Bureau of Standards, vol. 66D, no. 5, pp. 509–541, Sept./Oct. 1962. 34. S.P. Plun Keit, et al., Solar Source Regions of Coronal Man Ejections and their Geomagnetic Efforts, Journal of Atmospheric and Solar-Terrestrial Physics, vol. 63, issue 5, pp. 398–407, March 2001. 35. P.N. Mayaud, Analysis of Storm Sudden Commencements for the Years 1868–1967, Journal Geophysics Res., vol. 80, pp. 111, 1975. 36. R. Wang, Large Geomagnetic Storms of Extreme Solar Event Periods in Solar Cycle, Adv. Space Res., vol. 40, issue 12, pp. 1835–1841, 2007. 37. J.P. Elovaara, et al., Geometrically Induced Currents in the Nordic Power Systems, Proceedings of the CIGRE, Paper 36–301, Paris, pp. 10, 30 August - 5 September 1992. 38. A.A. Viljanen, R. Pirjola, Geomagnetically Induced Currents in the Finnish High Voltage Power System - A Geophysical Review, Surv. Geophys, vol. 15, pp. 383–408, 1994. 39. R. Pirjola, M. Lehtinen, Currents Produced in the Finnish 400 kV Power Transmission Grid and in the Finnish Natural Gas Pipeline by Geomagnetically-Induced Electric Fields, Ann. Geophys., vol. 3, pp. 485–491, 1985. 40. R. Pirjola, et al., Prediction of Geomagnetically Induced Currents in Power Transmission Systems, Adv. Space Res., vol. 26, no. 1, pp. 1–14, 2000. 41. I.A. Erinmez, et al., Management of the Geomagnetically Induced Current Risks on this National Grid Company’s Electric Power Transmission Systems, Journal of Atm. and SolarTerrestrial Physics, vol. 64, pp. 743–756, March-April 2002. 42. T.S. Molinski, Why Utilities Respect Geomagnetically Induced Currents, Journal of Atmospheric and Solar-Terrestrial Physics, vol. 64, issue 16, pp. 1765–1778, 2002. 43. D. Mayer, Analysis of Electrical Circuit with Matrix Calculus, Academia, Prague, 1966 (in Czech). 44. D. Mayer, Introduction to the Theory of Electric Circuits, 2nd Ed., SNTL/ALFA, Prague, 1981 (in Czech). 45. L. Bolduc, J. Aubin, Effects of Direct Currents in Power Transformers, Part I: A General Theoretical Approach, Electric Power Systems Research, vol. 1, issue 4, pp. 291–298, 1978. 46. W.B. Gish, W.E. Feero, G.D. Rockefeller, Rotor Heating Effects from Geomagnatic Induced Currents, IEEE Trans. Power Delivery, vol. 9, no. 2, pp. 712–719, April 1994.
Index
A Active power, 144, 147, 151, 152, 155, 158, 159, 175, 178, 179, 185 Alternating Current Transmission System (FACTS), 317, 319, 320, 323, 324, 329, 337, 342 Apparent power, 144, 147–149, 151, 152, 155, 159, 163, 164, 167, 168 Artificial Bee Colony (ABC), 255 Aurora, 618 Automatic compensation, 154, 156 Automatic Repeat reQuest (ARQ), 550 Automatic Secondary Voltage Regulation, 228, 242 B Back-to-Back (B2B), 288, 310, 312 Balanced loads, 155 Bandwidth, 552, 553, 556 Banks capacitors, 240, 241 Battery Energy Storage System (BESS), 296 Best compromise solution, 478, 483, 487, 492, 494, 496, 498, 499, 501, 503, 505 C Cascaded H-Bridge (CHB), 302–304 Cicode programs, 592 CitectSCADA, 591–593 Code Division Multiple Access (CDMA), 554, 555, 558 Combined Heat and Power (CHP), 549 Comma separated value (CSV), 461 Communications, 547, 549, 551, 552 Compensator, 137–139, 141, 145, 146, 151, 159, 169, 185 Constraints, 337 Control, 227, 233 Cost function, 329
Coverage, 548, 553, 555 Current Source Converter (CSC), 289 Customised-Of-The-Shelf (COTS), 551 Cyber, 565, 567, 579 Cyber attacks, 574, 578 Cyber security, 565, 579 Cyclodissipativity, 137, 138, 144, 185 D Decision variable, 476, 489, 490, 510 Delta configuration, 159 Deterministic ORPD (DMO-ORPD), 476, 478, 491–496, 498, 500–503, 505, 507, 508, 510 Differential Evolution (DE), 255 Digital signal processor, 175 DIgSILENT Programming Language (DPL), 417 Diode clamped, 295, 302 Dissipative systems, 137, 146 Distributed generation (DG), 191, 203, 204, 276, 287 Distribution Network Operator (DNO), 253 Double fed induction generator (DFIG), 224 Dynamical system, 145 Dynamic Power Flow Controller (DPFC), 287, 288, 309, 312 Dynamic Voltage Restorer (DVR), 288 E Earth’s magnetosphere, 618 Electric Power Systems, 547, 549 Electric Vehicle (EV), 549 Electromagnetic Interference (EMI), 276, 289 Energy loss, 329 Energy storage system, 296, 311 Energy transmission, 137–139, 144, 182 Evolutionary programming (EP), 346 Expected value of power loss, 484
© Springer International Publishing AG 2017 N. Mahdavi Tabatabaei et al. (eds.), Reactive Power Control in AC Power Systems, Power Systems, DOI 10.1007/978-3-319-51118-4
631
632 Expected value of voltage deviation, 484 F Fibre Optic Communication, 547, 552 Fibre to the Home (FTTH), 552 Fixed capacitor, 518 Fixed Capacitor Thyristor-Controlled Reactor (FCTCR), 124, 125 Flexible AC Transmission Systems (FACTS), 252, 275, 312 Flying capacitor MLC (FC-MLI), 302 Forward Error Correction (FEC), 550 Fuzzy Adaptive Particle Swarm Optimization (FAPSO), 255 Fuzzy decision making, 320, 333 G Gate Turn-Off Thyristor (GTO), 287, 295, 298, 302 Generalized Unified Power Flow Controller (GUPFC), 319, 320, 325–328, 337, 339, 340, 342 General Packet Radio Service (GPRS), 553 Genetic algorithm (GA), 192, 346, 365, 378 Geodynamo, 614–616 Geomagnetic activity, 612, 618 Geomagnetic storm, 611, 613, 619 Geomagnetism, 613, 614, 618 Global System for Mobile Communications (GSM), 551, 553, 555 Global System for Mobile Communications Railway (GSM-R), 551 Goodput, 556 H High Voltage Direct Current (HVDC), 287–289, 310–312 HMI, 563, 564, 572, 573, 576, 577, 581, 582, 591, 607 Homogenous operators, 141, 142 Honey Bee Mating Optimization (HBMO), 265 Hybrid Flow Controller (HFC), 319–321, 323, 329, 330, 337, 339, 340, 342 Hybrid Stochastic Search (HSS), 255 I IEEE 802.11 Mesh Networking, 547, 557 IEEE 802.15.4 Wireless Sensor Network, 547, 558 IEEE Standard, 547, 558 Improved Genetic Algorithm (IGA), 265 Improved PSO (IPSO), 255 Incidence matrix, 622, 625
Index Induction machine (IM), 518, 519, 521, 526, 542 Insulated Gate Bipolar Transistor (IGBT), 287, 288, 295 Insulated Gate Commutated Thyristor (IGCT), 287, 288, 295 Interline Power Flow Controller (IPFC), 127, 128, 310, 312, 319 Internal product, 142 Internetwork Protocol version 6 (IPv6), 550 Inter-phase power controller (IPC), 296, 319 ISA100.11a, 558 K Kirchhoff current law (KCL), 50, 57, 58, 66, 72, 97, 98, 100, 103, 105, 106, 108, 109, 111 Kirchhoff voltage law (KVL), 57, 58, 66, 67, 97, 100, 103, 111 L LAN, 572, 574, 587, 608 Latency, 553, 556, 557 Life expectancy, 556 Linear Programming (LP), 255 Long Term Evolution (LTE), 553, 555 Loss, 193, 195, 196, 198, 199, 207, 209, 219, 221–223 Low-pass filter, 154, 173, 174 M Machine-To-Machine (M2M), 550, 554, 555 Magnetic storm, 611, 612, 618–620, 626, 627 Media Access Control (MAC), 558 Metering, 547, 548, 550–552 Minimal value, 148 Mixed integer non-linear program (MINLP), 476 Modular Multilevel Converter (MMC), 310–312 Monte Carlo simulation (MCS), 478 MOPSO-NTVE, 331, 333, 334, 337, 338, 340, 341 MOPSO-TVAC, 338, 339, 341 MOPSO-TVIW, 338, 339, 341 MS SQL database, 587 MTU, 608 Multi-converter FACTS (M-FACTS), 319, 320, 323, 324, 329 Multilevel Converter, 276, 298, 302 Multi-objective, 318, 320, 321, 331, 342 Multi-Objective Optimal Reactive Power Dispatch (MO-ORPD), 475, 477, 478, 491, 494, 507
Index Multi-objective optimization, 320, 321, 331 N Network, 612, 613, 619–623, 625–627 Neutral Point Clamped (NPC), 131 Nodal Method, 50, 97 Nonlinear, 137, 138, 144–146, 151–156, 168, 171, 177 Non-Linear Programming (NLP), 255 Nonsinusoidal, 137, 138, 140, 141, 172 Number of Function Evaluation (NFE), 394, 457, 463, 465, 466, 469, 470 O On-Load Tap Changer (OLTC), 252, 253, 255 Operating point, 517, 518, 528–531, 533–535, 543 Optimal power flow (OPF), 214, 217, 252, 264, 346, 475, 476 Optimal reactive power dispatch, 475, 476, 510 Optimal reactive power flow (ORPF), 252, 362 P Pareto front, 221, 222 Particle Swarm Optimization (PSO), 255, 346, 366–371, 375, 378, 399, 427, 428, 435, 437, 438, 439, 442, 443, 449, 463, 465–468, 474 Pattern search, 347, 372–378, 382–385, 387, 389, 391, 397 Phase-shifting transformer (PST), 310, 319, 321 Planetary magnetism, 615 Point of Common Coupling (PCC), 262, 276, 281 Polyphase, 139, 141 Power factor, 17, 31, 35–38, 43, 44, 276–278 Power Flow Controller, 287–290, 308–310, 312 Power Line Communications (PLC), 547, 552 Power loss, 475–477, 483, 484, 492, 494, 498, 507, 510 Power quality, 227, 228, 237 Power system, 3–6, 21, 26, 31, 45, 227–229, 233, 234, 237, 240, 246, 247 Power transfer theorem, 50, 97 Power transformer sockets, 241 Power triangle, 172, 177 Probability distribution function (PDF), 478–480, 482 Pulse Width Modulation (PWM), 287
633 Q Quadratic Programming (QP), 255, 346 R Reactive power, 4, 12, 17, 34–38, 42–44, 137–139, 141, 145, 146, 148, 154, 155, 159, 163, 167, 169, 171–174, 185, 191, 193–200, 205, 209, 211–214, 216–218, 222–224, 227–237, 240–243, 245, 246 Reactive power compensation, 36, 38, 44 Reactive power consumption, 34, 35, 38, 44 Reactive Power Dispatch (RPD), 255–258, 261, 271 Reactive Power Planning (RPP), 263, 265, 269, 318 Real Genetic Algorithms (RGA), 255 Reliability, 191, 192, 196, 203, 204 Renewable energy source, 276 Resilience, 556 Resistive-inductive load, 518 Rikitake system, 616 Root mean square, 51 S SCADA, 561–579, 581–591, 595, 597, 601–604, 606–608 Scalability, 549, 556 Security, 549, 551, 555, 556 Self-excited induction generator, 518, 526, 528 Sequential Quadratic Programming (SQP), 346 Service Level Agreement (SLA), 550, 551, 555 Sinusoidal signal, 50–53, 97, 111 Smart grid, 554–557 Smart Integrated Generation, 549 Smart Market, 549 Smart Network Management, 549 Solar activity, 611, 617, 619 Solar wind, 613, 618, 619 Standards for power systems, 3, 26 Static Synchronous Compensator (STATCOM), 276, 296–298, 319 Static Synchronous Series Compensator (SSSC), 127, 128 Static VAr Compensation (SVC), 261, 265 Static VAR Compensator, 276, 287, 289 Static Voltage Restorer (SVR), 288 Static Voltage Stability (SVS), 264 STC, 329, 330, 334, 340 Stochastic MO-ORPD (SMO-ORPD), 475, 476, 478, 486, 490, 491, 497, 501, 504, 507
634 Sub-Synchronous Resonance (SSR), 123, 288, 294, 303–305, 312 Superconducting Magnetic Energy Storage (SMES), 296 Surge Impedance Loading (SIL), 118, 132 Switched, 179 Synchronous condenser, 228, 235, 236, 241 T Tabu Search (TS), 255 Tellegen’s theorem, 145 Three-phase networks, 154, 185 Throughput, 553 Thyristor-Controlled Braking Resistor (TCBR), 296 Thyristor-Controlled Reactor (TCR), 124, 125, 287, 290, 293, 294, 309, 517, 536 Thyristor-Controlled Series Capacitor (TCSC), 126 Thyristor-Controlled Series Compensator (TCSC), 287, 294, 295 Thyristor-Controlled Series Reactor (TCSR), 126 Thyristor-Controlled Voltage Limiter (TCVL), 296 Thyristor-Switched Capacitor (TSC), 124, 125, 287, 290–292, 309 Thyristor-Switched Capacitor-Thyristor-Controlled Reactor (TSC-TCR), 124, 125 Thyristor-Switched Reactor (TSR), 124, 125 Thyristor-Switched Series Capacitor (TSSC), 125, 319 Thyristor Switched Series Reactor (TSSR), 319 Time delay, 172, 173, 177 Total harmonic distortion, 28, 41, 80, 277 Traditional Genetic Algorithm (TGA), 265 U Unbalanced, 140, 154, 158, 161–163, 165, 168 Unified Power Flow Controller (UPFC), 127, 128, 287, 288, 296, 306–308, 310, 312, 319, 320
Index V Vijeo Citect Surveillance Softwares, 561, 565, 566, 591 Volt Ampere Reactive (VAR), 275, 276, 287, 289, 296 Voltage collapse, 318 Voltage Deviation (VD), 195, 221–223, 329, 362, 385, 386, 388, 390, 476, 478, 483, 484, 492, 494, 507, 510 Voltage profile, 318 Voltage Source Converter (VSC), 276, 287–289, 295–301, 304, 306, 308, 310–312 Voltage stability, 193, 221–223, 227, 228, 233, 246, 247, 318, 475–478, 483–485, 507, 510 Voltage Stability Index (VSI), 362, 384, 386, 387, 389 Voltage Stability Margin (VSM), 255 W WiFi, 552, 558 WiMAX, 556, 558 Wind farm, 259, 261, 265, 266, 271, 476, 482, 483, 488, 489, 493, 496, 503, 506, 508 Wind power plant, 192, 201, 202, 211, 221, 223, 241, 243 Wind turbine, 191, 193–199, 201, 202, 204, 205, 210, 211, 213–215, 222–224, 259, 265 WirelessHART, 558 Wireless network, 553, 558 Wireless technologies, 547, 553, 556 X X Digital Subscriber Line/Loop (xDSL), 547, 552 Z Zero-Voltage Switching (ZVS), 290 ZigBee, 558