Soroudi 2017

Alireza Soroudi
Power System
Optimization
Modeling in
GAMS
Power System Optimization Modeling in GAMS
Alireza Soroudi
Power System Optimization

Modeling in GAMS
123
Alireza Soroudi
School of Electrical Engineering
University College of Dublin
Belfield, Dublin, Ireland
ISBN 978-3-319-62349-8 ISBN 978-3-319-62350-4 (eBook)

DOI 10.1007/978-3-319-62350-4
Library of Congress Control Number: 2017944921
© 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
Preface
The complex structure of electric power systems as the largest human-built

machines justifies the need for an efficient and robust computation tool to conduct
in-depth analysis. This book is intended to serve as an introduction to the application
of GAMS software in solving a broad range of power system optimization problems.
Due to the innate capability of GAMS, researchers can now pursue questions that
they were unable to previously. It has been 8 years since I was introduced to GAMS.
During these years, I have explored new areas in power system studies. The main
purpose of this book is to share these experiences with other researchers/students or
industry employees who might need a powerful computation tool for their studies.
This book covers a broad range of topics in power system studies as follows:
• Chapter 1: Gives an overview and tips on how to start coding with GAMS for
beginners.
• Chapter 2: Simple but practical examples are solved.
• Chapter 3: Economic dispatch concept is covered and solved.
• Chapter 4: Deals with dynamic economic dispatch problem where the problem
is multi-period.
• Chapter 5: Covers the unit commitment problem.
• Chapter 6: Multi-period optimal DC/AC power flow is explained and solved.
• Chapter 7: The application of energy storage system in power system planning
and operation is analyzed.
• Chapter 8: The power system observability is enhanced using PMU technology.
• Chapter 9: Some different transmission network operation and planning prob-
lems are presented and solved.
• Chapter 10: Energy system integration is discussed and solved.
The intended audiences are but not limited to the following:
• Power engineers: The proposed framework of this book enables power engineers
to obtain a better insight into power system studies.
• Educators: This book can be used as a textbook in different power system
operation and planning modules.
v
vi Preface
• System operators: Power system operators usually use commercial packages

for solving their problems. This book can bring flexibility into their studies.
Sometimes, the commercial software does not include all required analysis, or
they are used as a black box.
• Researchers: Students and researchers with diverse mathematical and power
system backgrounds can use this book to get familiar with different power system
optimization issues.
This book is provided by the author “as is,” and any express or implied
warranties, including, but not limited to, the implied warranties of merchantability
and fitness for a particular purpose, are disclaimed. In no event shall the copyright
holder or author or publisher be liable for any direct, indirect, incidental, special,
exemplary, or consequential damages (including, but not limited to, procurement of
substitute goods or services; loss of use, data, or profits; or business interruption)
however caused and on any theory of liability, whether in contract, strict liability,
or tort (including negligence or otherwise) arising in any way out of the use of this
software or codes, even if advised of the possibility of such damage.
As the first book on “power system optimization modeling in GAMS,” there is
always space for improvement. Any comments and suggestions from the readers are
welcome. Please kindly share them with the author.
Dublin, Ireland Alireza Soroudi

May 2017
Acknowledgments
This book would have not been finished without the contributions of several globally
recognized experts in the power system field. A special thanks to Dr. Turaj Amraee
who introduced GAMS to me for the first time.
I would like to thank those who gave me insight about the power system studies,
especially Dr. Raphael Caire at Grenoble INP, Dr. Andrew Keane at University
College Dublin, and Dr. Jonathon O’Sullivan at EirGrid.
This book is dedicated to my lovely wife, Soudeh, who supported me in putting
together all the details. Finally, I appreciate all that my parents, Simin and Shahryar,
and my sister, Mona, have done for me.
vii
Contents
1 Introduction to Programming in GAMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 Optimization Problems in Power System . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 GAMS Installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3 GAMS Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.4 Conditional Statements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.5 Loop Definition in GAMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.5.1 LOOP Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.5.2 WHILE Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.5.3 FOR Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.5.4 REPEAT-UNTIL Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.6 Linking GAMS and Excels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.6.1 Reading from Excel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.6.2 Writing to Excel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.7 Error Debugging in GAMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.8 General Programming Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
1.9 Book Structure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2 Simple Examples in GAMS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.1 Different Types of Optimization Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.1.1 Linear Programming (LP) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.1.2 Mixed Integer Programming (MIP). . . . . . . . . . . . . . . . . . . . . . . . 38
2.1.3 Nonlinear Programming (NLP) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
2.1.4 Quadratic Constrained Programming (QCP) . . . . . . . . . . . . . . 48
2.1.5 Mixed Integer Nonlinear Programming (MINLP) . . . . . . . . 49
2.2 Random Numbers in GAMS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
2.2.1 Estimating the Number . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
2.2.2 Integration Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
2.2.3 LP Problems with Uncertain Coefficients . . . . . . . . . . . . . . . . . 56
ix
x Contents
2.3 Multi-Objective Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

2.3.1 Weighted Sum Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.3.2 Pareto Optimality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.4 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3 Power Plant Dispatching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.1 Thermal Unit Economic Dispatch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.2 Thermal Unit Environmental Dispatch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
3.3 CHP Economic Dispatch. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
3.4 Hydro Unit Economic Dispatch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
3.5 Multi-Area Mix Unit Dynamic Dispatch. . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.6 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
4 Dynamic Economic Dispatch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
4.1 Cost-Based DED . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
4.1.1 Ramp Rate Sensitivity Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
4.1.2 Multi-Objective Cost-Emission Minimization . . . . . . . . . . . . 100
4.1.3 Wind Integrated DED . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
4.2 Price-Based DED. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
4.2.1 Price-Based DED Just Energy Market. . . . . . . . . . . . . . . . . . . . . 107
4.2.2 Price-Based DED Energy and Reserve Market . . . . . . . . . . . 110
4.3 Linearized Cost-Based DED . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
4.4 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
5 Unit Commitment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
5.1 Cost-Based Unit Commitment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
5.1.1 Cost Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
5.1.2 Ramp Rate Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
5.1.3 Min Up/Down Time Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
5.1.4 Demand-Generation Balance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
5.2 Cost-Based UC with Additional Constraints. . . . . . . . . . . . . . . . . . . . . . . . 125
5.2.1 Cost-Based UC with Reserve Constraints . . . . . . . . . . . . . . . . . 126
5.2.2 Cost-Based UC Considering Generator Contingency. . . . . 128
5.2.3 Cost-Based UC with Demand Flexibility Constraints . . . . 131
5.3 Price-Based Unit Commitment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
5.4 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
5.4.1 Cost-Based UC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
5.4.2 Price-Based UC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
Contents xi
6 Multi-Period Optimal Power Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141

6.1 Single Period Optimal DC Power Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
6.1.1 Three-Bus Network DC-OPF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
6.1.2 Five-Bus Network DC-OPF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
6.1.3 IEEE Reliability Test System 24 Bus . . . . . . . . . . . . . . . . . . . . . . 150
6.2 Multi-Period Optimal Wind-DC OPF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
6.3 Multi-Period Optimal AC Power Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
7 Energy Storage Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
7.2 ESS Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
7.2.1 ESS Operation in DED . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
7.2.2 ESS Operation in Wind-DED Problem . . . . . . . . . . . . . . . . . . . . 181
7.2.3 ESS Operation in DC-OPF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
7.2.4 ESS Operation in AC-OPF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
7.3 ESS Allocation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
7.4 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
8 Power System Observability. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
8.1 Min No. PMU Placement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
8.2 Min Cost PMU placement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
8.3 Min No. PMU Placement Considering ZIB . . . . . . . . . . . . . . . . . . . . . . . . . 209
8.4 Min No. PMU Placement for Maximizing the Observability . . . . . . 213
8.5 Min No. PMU Placement Considering Contingencies . . . . . . . . . . . . . 216
8.5.1 Loss of PMU . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
8.5.2 Single Line Outage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
8.6 Multistage PMU Placement. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
8.7 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
9 Topics in Transmission Operation and Planning. . . . . . . . . . . . . . . . . . . . . . . . 227
9.1 Transmission Network Planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
9.1.1 TEP with New Lines Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
9.1.2 TEP with New Lines and Power Flow
Controller Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
9.2 Sensitivity Factors in Transmission Networks . . . . . . . . . . . . . . . . . . . . . . 238
9.2.1 Generation Shift Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
9.2.2 Line Outage Distribution Factors . . . . . . . . . . . . . . . . . . . . . . . . . . 246
9.3 Transmission Network Switching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
xii Contents
10 Energy System Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265

10.1 Water-Power Nexus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
10.2 Gas-Power Nexus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
10.3 Energy Hub Concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
10.3.1 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
10.3.2 Configuration I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
10.3.3 Configuration II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
10.3.4 Configuration III . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
List of Figures
Fig. 1.1 Some optimization problems in power system studies . . . . . . . . . . . . 2

Fig. 1.2 Optimization methods in power system studies . . . . . . . . . . . . . . . . . . . 4
Fig. 1.3 The structure of heuristic optimization techniques . . . . . . . . . . . . . . . . 5
Fig. 1.4 The GAMS interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Fig. 1.5 General GAMS code structure and elements . . . . . . . . . . . . . . . . . . . . . . 8
Fig. 1.6 The structure of GAMS listing file . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Fig. 1.7 GAMS model library . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
Fig. 1.8 GAMS help menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Fig. 2.1 Queen placement on a 4 4 board: (a) wrong placement,
(b) optimal placement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Fig. 2.2 Queen placement on a (a) 8 8 chessboard, (b) 16 16
board . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
Fig. 2.3 Optimal circle placement on a given surface . . . . . . . . . . . . . . . . . . . . . . 46
Fig. 2.4 The optimal solution of circle placement on a surface . . . . . . . . . . . . 47
Fig. 2.5 Maximizing the area of a triangle with constant
circumference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
Fig. 2.6 Optimal transportation problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Fig. 2.7 Estimating the number . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Fig. 2.8 Numerical integration using random numbers . . . . . . . . . . . . . . . . . . . . . 56
Fig. 2.9 The probability density function for each uncertain
parameter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Fig. 2.10 Uncertain coefficients in LP problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Fig. 2.11 Classification of a population to k non-dominated fronts . . . . . . . . . 59
Fig. 2.12 The Pareto optimal front for bi-objective problem . . . . . . . . . . . . . . . . 61
Fig. 3.1 Load sensitivity analysis and impact on costs and
generation schedules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
Fig. 3.2 Pareto optimal front in multi-objective environmental
dispatch strategy for load = 400 MW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
Fig. 3.3 Changes in generating level of different units in Pareto
optimal front for load = 400 MW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
xiii
xiv List of Figures
Fig. 3.4 Pareto optimal fronts for different load values . . . . . . . . . . . . . . . . . . . . 76

Fig. 3.5 Approximate heat-power operating region for a CHP unit . . . . . . . 77
Fig. 3.6 Cascaded hydro units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
Fig. 3.7 Four cascaded hydro units configuration . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Fig. 3.8 Thermal unit schedules in multi-area economic dispatch . . . . . . . . . 86
Fig. 3.9 Hydro unit schedules in multi-area economic dispatch . . . . . . . . . . . 86
Fig. 3.10 Multi-area dynamic economic dispatch problem . . . . . . . . . . . . . . . . . . 87
Fig. 3.11 Wind and demand pattern in each area . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
Fig. 3.12 Inter-area power transfer of different tie-lines . . . . . . . . . . . . . . . . . . . . . 90
Fig. 4.1 The hourly thermal unit power schedules in cost-based
DED . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
Fig. 4.2 The sensitivity analysis of ramp rates on hourly thermal
unit power schedules in DED . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
Fig. 4.3 The Pareto optimal front for emission-cost minimization
DED . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
Fig. 4.4 The best compromised solution of Pareto optimal front for
emission-cost minimization DED . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Fig. 4.5 The hourly schedules of thermal units in wind-DED
problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
Fig. 4.6 The hourly variation of total wind and thermal power
generation in wind-DED problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
Fig. 4.7 The optimal solution of price-based DED . . . . . . . . . . . . . . . . . . . . . . . . . 111
Fig. 4.8 The energy and reserve prices vs time in price-based DED . . . . . . 112
Fig. 4.9 The energy and reserve schedules vs time in price-based
DED with arbitrage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
Fig. 4.10 Linearizing the quadratic cost function . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
Fig. 5.1 Unit commitment cost calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
Fig. 5.2 Hourly demand and price values vs time . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Fig. 5.3 Power schedules in cost-based unit commitment vs time . . . . . . . . . 125
Fig. 5.4 The demand pattern change in cost-based unit commitment
with demand response flexibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
Fig. 5.5 The sensitivity analysis of demand pattern changes in
CBUC with various demand response flexibilities . . . . . . . . . . . . . . . . 135
Fig. 5.6 The variation of total operating costs vs demand flexibility . . . . . . 136
Fig. 5.7 Power schedules in price-based unit commitment vs time . . . . . . . 138
Fig. 5.8 Different uncertainty parameters and modeling methods . . . . . . . . . 139
Fig. 6.1 DC power flow for a two-bus network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
Fig. 6.2 General power flow concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
Fig. 6.3 Three bus network example (a) data and (b) power flow
solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
Fig. 6.4 The three-bus network with Pmax 12 D 50 MW . . . . . . . . . . . . . . . . . . . . . . 147
Fig. 6.5 The five-bus network data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
Fig. 6.6 The five-bus network flows and directions . . . . . . . . . . . . . . . . . . . . . . . . 150
List of Figures xv
Fig. 6.7 IEEE RTS 24-bus network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151

Fig. 6.8 Load shedding modeling using virtual generating unit . . . . . . . . . . . 158
Fig. 6.9 Load flow model considering wind power and load
shedding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
Fig. 6.10 Wind power connection points to the network . . . . . . . . . . . . . . . . . . . . 164
Fig. 6.11 Wind-demand variation patterns vs time . . . . . . . . . . . . . . . . . . . . . . . . . . 164
Fig. 6.12 Wind power generations at efferent buses vs time . . . . . . . . . . . . . . . . 167
Fig. 6.13 AC power flow considering wind generation and load
shedding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
Fig. 6.14 Active power generation of thermal units in MP-AC OPF . . . . . . . 170
Fig. 6.15 Reactive power generation of thermal units in MP-AC OPF . . . . . 170
Fig. 7.1 The hourly dispatch of ESS in DED problem . . . . . . . . . . . . . . . . . . . . . 180
Fig. 7.2 The hourly schedules of thermal units in DED-ESS problem . . . . 180
Fig. 7.3 The hourly dispatch of ESS in wind-DED problem . . . . . . . . . . . . . . . 183
Fig. 7.4 The hourly schedules of thermal units in wind-DED-ESS
problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
Fig. 7.5 The IEEE 24 RTS with ESS integration . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
Fig. 7.6 The hourly dispatch of ESS in DC-OPF problem . . . . . . . . . . . . . . . . . 189
Fig. 7.7 The hourly dispatch of thermal generating units in DC-OPF
problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
Fig. 7.8 The hourly dispatch of ESS in AC-OPF problem . . . . . . . . . . . . . . . . . 193
Fig. 7.9 The hourly dispatch of thermal generating units in AC-OPF
problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
Fig. 7.10 The hourly active power losses in AC-OPF problem . . . . . . . . . . . . . 194
Fig. 7.11 The hourly min/avegare/max values of voltage magnitudes
(pu) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
Fig. 7.12 The optimal allocation of ESS units in IEEE RTS network . . . . . . 198
Fig. 7.13 The hourly dispatch of storage unit in optimal ESS
allocation problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
Fig. 7.14 The hourly dispatch of thermal generating units in ESS
allocation problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
Fig. 8.1 PMU functionality and the observability concept . . . . . . . . . . . . . . . . . 204
Fig. 8.2 Optimal PMU allocation for IEEE 14-bus without
considering the zero injection buses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
Fig. 8.5 PMU functionality and observability concept in zero
injection buses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Fig. 8.6 Optimal PMU allocation for IEEE 14-bus with considering
the zero injection buses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
Fig. 8.7 Optimal PMU allocation for IEEE 57-bus with considering
the zero injection buses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
xvi List of Figures
Fig. 8.8 Optimal PMU allocation for IEEE 118-bus with

Fig. 8.9 Total observable buses vs NPMU in IEEE 57-bus network . . . . . . . . 216
Fig. 8.10 Total observable buses vs NPMU in IEEE 118-bus network . . . . . . . 216
Fig. 8.11 Optimal PMU allocation for IEEE 14-bus resilient against
single PMU outage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
Fig. 8.14 Multistage optimal PMU allocation for different case
studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
Fig. 9.1 Six-bus Garver transmission network (base case) . . . . . . . . . . . . . . . . . 229
Fig. 9.2 Optimal TEP solution for 2 years planning period
(T D 2 8760 h) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
Fig. 9.3 Optimal TEP solution for 10 years planning period
(T D 10 8760 h) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
Fig. 9.4 The phase shifter function in power flow control . . . . . . . . . . . . . . . . . 234
Fig. 9.5 The optimal decisions regarding the location of
phase shifter and new transmission lines 2 years plan
(T D 2 8760) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
Fig. 9.6 The optimal decisions regarding the location of phase
shifter and new transmission lines 10 years plan
(T D 10 8760) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
Fig. 9.7 Three-bus network example for sensitivity factors
calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
Fig. 9.8 The DC power flow solution for three-bus network . . . . . . . . . . . . . . . 245
Fig. 9.9 The DC power flow solution for three-bus network after
increasing the demand at bus 2 for 10 MW. . . . . . . . . . . . . . . . . . . . . . . . 245
Fig. 9.10 The DC power flow solution for three-bus network after
increasing the generation at bus 3 for 10 MW. . . . . . . . . . . . . . . . . . . . . 246
Fig. 9.11 Line outage modeling using virtual injections. (a) Intact
network. (b) Post-contingency network . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
Fig. 9.12 IEEE RTS 24-bus network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250
Fig. 9.13 IEEE 118-bus network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
Fig. 9.14 The operating and congestion cost vs number of switched
lines in IEEE 118-bus network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
Fig. 9.15 The operating cost vs the connection point of new demand
(20 MW) in with/without transmission switching cases . . . . . . . . . . 258
Fig. 9.16 The LMP values (i ) in ($/MW h) at different buses in
with/without transmission switching cases . . . . . . . . . . . . . . . . . . . . . . . . 258
Fig. 10.1 Water-energy nexus concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
Fig. 10.2 Hourly water output of different plants in water-energy
nexus problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
List of Figures xvii
Fig. 10.3 Hourly power output of different plants in water-energy

nexus problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
Fig. 10.4 Gas network linkage with electricity network . . . . . . . . . . . . . . . . . . . . . 269
Fig. 10.5 Hourly variation pattern of wind generation, electric and
gas demand (pu) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
Fig. 10.6 Hourly variation pattern of gas generation from gas sources . . . . . 275
Fig. 10.7 Hourly variation pattern of electricity power generation . . . . . . . . . 276
Fig. 10.8 The energy hub concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
Fig. 10.9 The energy hub configuration I, considering transformer,
furnace, and chiller boiler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
Fig. 10.10 The hourly purchased electricity and natural gas in energy
hub configuration I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
Fig. 10.11 The hourly output of furnace unit in energy hub
configuration I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
Fig. 10.12 The energy hub configuration II, considering transformer,
furnace, chiller boiler, CHP, and electric energy storage . . . . . . . . . 283
Fig. 10.13 The hourly purchased electricity and its division between
transformer and ESS in energy hub configuration II . . . . . . . . . . . . . . 285
configuration II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286
Fig. 10.15 The hourly state of charge (MW h), charging, and
discharging of ESS (MW) in energy hub configuration II . . . . . . . . 286
Fig. 10.16 The energy hub configuration III, considering transformer,
furnace, chiller boiler, CHP, electric energy storage, and
EHP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
Fig. 10.17 The hourly purchased electricity and its division between
transformer and ESS in energy hub configuration III . . . . . . . . . . . . . 289
configuration III . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
Fig. 10.19 The hourly state of charge (MW h), charging, and
discharging of ESS (MW) in energy hub configuration III . . . . . . . 290
Fig. 10.20 The hourly output of EHP in energy hub configuration III . . . . . . . 291
List of Tables
Table 1.1 Data for illustrative example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

Table 3.1
The economic dispatch data for five thermal units example . . . . . . 66
Table 3.2
The ED solution for five units example . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Table 3.3
The ED solution for load sensitivity analysis . . . . . . . . . . . . . . . . . . . . . . 68
Table 3.4
Solution for different environmental dispatching strategies . . . . . . 71
Table 3.5
Characteristic data of different technologies. . . . . . . . . . . . . . . . . . . . . . . 78
Table 3.6
Feasible operating regions of CHP units . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Table 3.7
Solution of dispatching CHP units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Table 3.8
Techno-economic characteristics of thermal units. . . . . . . . . . . . . . . . . 81
Table 3.9
Techno-economic characteristics of hydro units. . . . . . . . . . . . . . . . . . . 82
Table 3.10
Hourly water inflows for each hydro power and hourly
electric demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
Table 3.11 Optimal hourly dispatch of hydro and thermal units . . . . . . . . . . . . . 85
Table 3.12 Characteristics of thermal units and their connected area . . . . . . . . 88
Table 3.13 Solution of multi-area economic dispatch . . . . . . . . . . . . . . . . . . . . . . . . . 91
Table 4.1 The dynamic economic dispatch data for four units example. . . . . 96
Table 4.2 The dynamic economic dispatch solution for four units
expressed in (MW) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
Table 4.3 The numerical results for ramp rate sensitivity analysis in
dynamic economic dispatch of 4 units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
Table 4.4 The numerical results for Pareto optimal solutions
(cost-based DED) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Table 5.1 Unit commitment data for ten thermal units . . . . . . . . . . . . . . . . . . . . . . . 121
Table 5.2 Power schedules (Pi;t ) of thermal units considering reserve
constraint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
Table 5.3 Reserve provision (Ri;t ) by thermal units considering
reserve constraint. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Table 5.4 Power schedules (Pi;t ) by thermal units considering
generator contingency constraint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
xix
xx List of Tables
Table 5.5 Reserve provision (Ri;t ) by thermal units considering

generator contingency constraint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
Table 5.6 Power schedules (Pi;t ) by thermal units considering
demand response constraint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
Table 6.1 The techno-economic data of thermal units in two-bus
OPF example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
Table 6.2 The three-bus optimal power flow solution (Pmax ij D 100 MW) . . . 146
Table 6.3 The three-bus optimal power flow solution (Pmax 12 D 50 MW) . . . . 148
Table 6.4 The optimal power flow solution in five-bus network . . . . . . . . . . . . . 150
Table 6.5 Branch data for IEEE RTS 24-bus network . . . . . . . . . . . . . . . . . . . . . . . . 151
Table 6.6 Generation data for IEEE RTS 24-bus network . . . . . . . . . . . . . . . . . . . 152
Table 6.7 Base case solution of IEEE RTS 24-bus network (branch
flow limits are unchanged). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
Table 6.8 Solution of IEEE RTS 24-bus network (branch flow limits
are reduced by 30%) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
Table 6.9 Solution of IEEE RTS 24-bus network (branch outage
contingencies) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
Table 6.10 Solution of IEEE RTS 24-bus network (generator outage
contingencies) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
Table 6.11 Solution of IEEE RTS 24-bus network (generator g10
outage with load shedding modeling) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
Table 6.12 Solution of IEEE RTS 24-bus network (generator g10
outage with load shedding modeling and LMP calculation) . . . . . . 162
Table 6.13 Thermal unit power generation schedules of multi-period
wind DC-OPF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
Table 7.1 The data for ESS integrated dynamic economic dispatch . . . . . . . . . 176
Table 7.2 The dynamic economic dispatch data for four thermal
units example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
Table 7.3 The ESS integrated dynamic economic dispatch . . . . . . . . . . . . . . . . . . 179
Table 8.1 Optimal allocation of PMU devices with minimum cost
without considering the ZIB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Table 8.2 ZIB buses in different networks under study . . . . . . . . . . . . . . . . . . . . . . 210
Table 8.3 Optimal allocation of PMU devices for maximizing the
system observability of IEEE 14-bus network . . . . . . . . . . . . . . . . . . . . . 214
Table 8.4 Optimal allocation of PMU devices considering single line
outage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Table 8.5 Optimal allocation of PMU devices in multistage PMU
allocation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
Table 9.1 Transmission expansion planning data for Garver six-bus
transmission network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
Table 9.2 Generator shift factors (aijm ) for three-bus network . . . . . . . . . . . . . . . . 244
List of Tables xxi
Table 9.3 Line outage distribution factors (LODFij;nm ) for three-bus

network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
Table 9.5 LODF12;` calculated for IEEE RTS 24-bus network . . . . . . . . . . . . . . 251
Table 9.6 GSHF`;m calculated for IEEE RTS 24-bus network . . . . . . . . . . . . . . . 252
Table 9.7 Generation data for 118-bus network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
Table 9.8 Branch data for 118-bus network. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
Table 10.1 Technical and economical characteristics of thermal units . . . . . . . 266
Table 10.2 Technical and economical characteristics
of co-production units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
Table 10.3 Technical and economical characteristics of water
desalination units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
Table 10.6 Technical and economical characteristics of gas nodes . . . . . . . . . . . 272
Table 10.7 Technical characteristics of gas network . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
Table 10.8 Different hourly demand and electricity price data for
energy hub configurations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
Chapter 1
Introduction to Programming in GAMS
The General Algebraic Modeling System (GAMS) is a modeling tool for mathemat-
ical programming and optimization purpose. This chapter provides the instruction
on different programming elements in GAMS. It can be used in solving different
types of optimization problems. Some basic optimization models used in power
system literature are described in this chapter.
1.1 Optimization Problems in Power System
The power system optimization problems are broadly categorized as operation and
planning problems. The operation problems are usually related to how to exploit
the existing devices/power plants. For example, optimal power flow is an operation
problem. The planning problems usually refer to those problems which investigate
whether to invest or not in some assets. For example, the decisions regarding
the transmission network expansion belong to planning category. Some of these
problems are shown in Fig. 1.1. Some power system planning problems are listed as
follows:
• Generation expansion planning (GEP) [1, 2]: In GEP, the decision maker is trying
to find out the investment decision regarding the generation technology, size, and
time of investment.
• Transmission expansion planning (TEP) [3, 4]. In TEP, the decision maker is
trying to find out the investment decision regarding the planning option and time
of investment. The planning options include but not limited to: building new
lines, reconductoring the existing lines, building, or reinforcing the substations.
• Distribution network planning [5]: this problem is trying to make smart invest-
ments in new feeders/substations to meet the technical constraints.
© Springer International Publishing AG 2017 1

A. Soroudi, Power System Optimization Modeling in GAMS,
DOI 10.1007/978-3-319-62350-4_1
2 1 Introduction to Programming in GAMS
Fig. 1.1 Some optimization problems in power system studies
• FACTS device allocation [6]: This problem is a subset of transmission planning

problem. The FACTS devices can control the power flowing through a line and
make the transmission network more flexible.
• Distributed generation (DG) allocation in distribution networks [7–9]: The
optimal location (or size) of DG unit in a distribution network.
• Capacitor allocation in distribution networks [10]: The capacitor allocation is
usually done at the distribution level. The purpose is usually improving the
voltage profile in the network.
• PMU allocation [11, 12]: The system observability is improved using optimal
placement of phasor measurement units in the network.
• Energy Storage System (ESS) allocation [13, 14]: The ESS can deliver value
to customers. This highly depends on how they are operated and located in
the system. This problem tries to maximize its benefits by finding the optimal
connection point of ESS to the grid.
• Risk and uncertainty modeling in planning studies: the decision-making process
highly relies on accuracy of input data. As a matter of fact, the input data for
any decision-making problem (especially in practical problems) are subject to
uncertainty. If these uncertainties are not treated and handled properly, then
costly consequences might occur. Based on the data availability some of the
following techniques are applicable:
– Information gap decision theory (IGDT): ESS allocation [15]
– Scenario-based uncertainty modeling: DG planning [16]
1.1 Optimization Problems in Power System 3
– Monte Carlo simulation: FACTS allocation [17]

– Robust optimization: Transmission expansion planning [18]
– Fuzzy modeling: distribution network planning [19]
Some power system operation problems are listed as follows:
• Distribution feeder reconfiguration [20]: the on/off statuses of switches are opti-
mally determined to change the network configuration. In distribution networks,
it can reduce the active losses by improving the voltage profile and reliability.
• Generation scheduling: UC [21, 22], Dynamic economic dispatch [23], Eco-
nomic dispatch [24]: the generation level of power plants are determined to
minimize the operating costs or maximizing the economic benefits.
• Maintenance scheduling [25]: The maintenance period of assets are determined
to keep the adequacy level of the remaining system sufficient while minimizing
the costs.
• Optimal power flow: DC OPF [26] and AC OPF [27] are two forms of OPF.
In an OPF problem, the decision variables are generation and voltage level of
generating units to minimize the operating costs. In DC OPF, it is assumed that
all voltage values are 1 pu.
• Active power Loss minimization: minimizing the active losses is a way of
improving the efficiency of power system [28].
• Voltage profile improvement [29]: keeping the voltage magnitudes within the
normal operating limits is done by changing different decision variables such as
reactive power management and demand response.
• Optimal transmission switching [30]: transmission network switching is a tech-
nique for changing network topology at the transmission level. It is demonstrated
that this can lead to operating cost reduction if it is optimally managed.
• Electric vehicle charging [31]: the charging and discharging of electric vehicles
are optimally determined to provide some flexibilities for distribution network
operator.
• Offering strategy [32, 33]: the generating units submit their offers to the market
operator. The purpose is to maximize the financial benefits.
• Scheduling of reserve [34, 35]: finding the optimal reserve quantity makes the
system robust against contingencies and disturbances.
• Loss payments minimization [36]: this approach tries to minimize the payments
toward the losses, not the losses by considering the market issues.
• Congestion management [37, 38]: network congestion would reduce the com-
petition level at electricity market. Reducing the congestion would improve the
market efficiency.
• Demand side management [39, 40]: the demand side management is harvesting
the flexibility from the demand side. The customers are encouraged to shift/re-
duce their demands to minimize the system requirements for providing services.
• Risk and uncertainty modeling in operation studies:
– Information gap decision theory (IGDT): Wind operation modeling in OPF
[41], unit commitment [42]
– Scenario-based uncertainty modeling [43]
Fig. 1.2 Optimization methods in power system studies
– Monte Carlo simulation: reliability and risk analysis [44]

– Robust optimization: Demand response [40], loss payments minimization [36]
– Fuzzy modeling: DG impact assessment [45]
The methods used for solving the aforementioned optimization problems are
categorized into classic and heuristic methods as shown in Fig. 1.2.
Some of the classic methods are listed as follows:
• Interior point: optimal reactive dispatch [46]
• Branch and bound: economic dispatch with disjoint prohibited zones considering
network losses [47]
• Benders decomposition: transmission network design problems [48]
• Semi-definite programming: large scale OPF [49, 50]
• Lagrange Multipliers: pricing energy and ancillary services [51]
• Karush-Kuhn-Tucker (KKT) optimality condition: formulation of the terrorist
threat problem [52]
• Newton method: Optimal power flow [53, 54]
• Sequential quadratic programming (SQP): UC [55], VAr compensation [56]
Most of the classic methods are gradient-based techniques (in nonlinear problems).
This makes them unsuitable for large scale optimization problems. Solving the
1.1 Optimization Problems in Power System 5
Fig. 1.3 The structure of heuristic optimization techniques
nonlinear problems with integer variables or non-convex constraints would be

another challenge for classic techniques.
The heuristic methods are inspired by nature. The basic concept of these
techniques is described in Fig. 1.3. In every heuristic method, an initial random
population is generated, and it is tried to improve it by using some operator that
modify the population (solution). The way each method modifies the population
distinguishes that method from the other techniques. For example, the genetic
algorithm uses the crossover and mutation operators to improve the solution regard-
ing optimizing the objective function while satisfying the constraints. The particle
swarm optimization uses the best solution found by the group and the individual
particle to find the optimal solutions. The heuristic methods are also called iterative
techniques. The number of iterations needed for optimizing the objective function
has an inverse relation with the population number. If the population number is
increased, then it can better explore the solution space; however, it takes more time
to run. There is always a tradeoff between these two quantities. There are some
challenges associated with heuristic techniques. Some of them are listed as follows:
• The parameter tuning is a challenge in these techniques which is usually problem
dependent and should be tuned by the decision maker.
• It is not easy to check if the obtained solution is globally optimal or not.
• These methods are usually computationally expensive. In other words, it takes a

long time for the decision maker to run. This makes these methods inconvenient
for real-time applications.
• Since these methods are iterative, at every iteration, a new solution might be
found (which is somehow better than the solution found in previous iterations).
Unfortunately, if the problem is solved again (even with the same tuning
parameters), it is not guaranteed to reach the similar results.
• Setting the stopping criteria is difficult. This is because the optimal solution of the
problem in unknown. When to stop? It is a challenging question to answer. The
decision maker usually ends the procedure when there is no significant change in
objective function to a maximum number of iterations so reached.
• These solutions are not generally well accepted by the industry.
Some of the heuristic methods applied to power system studies are given as
follows:
• Single objective genetic algorithm (GA): network reconfiguration [57]
• Multi-objective Non-dominated sorting genetic algorithm (NSGA): Transmis-
sion expansion planning [58]
• Particle swarm optimization (PSO): DG planning [16]
• Immune algorithm (IA): secondary voltage control [59]
• Simulated annealing (SA): maintenance scheduling [60]
• Quantum-inspired evolutionary algorithm: Real and Reactive Power Dispatch
[61]
• Seeker Optimization Algorithm: Reactive Power Dispatch [62], coordination of
directional over-current relays [63]
• Cuckoo search algorithm: Non-convex economic dispatch [64], capacitor alloca-
tion [65]
• Shuffled frog leaping algorithm: unit commitment [66]
• Imperialist competition algorithm: dynamic economic power dispatch [67]
• Tabu search: Economic dispatch [68]
• Ant colony algorithm: Reconfiguration and capacitor placement for loss reduc-
tion of distribution systems [69]
1.2 GAMS Installation
The first step is downloading the appropriate installation file from the following
address:
https://www.gams.com/download/
1.2 GAMS Installation 7
The GAMS installation package is available for the following platforms:
Platform Description
MS Windows 32 bit Windows Vista or newer on AMD- or Intel-based
(x86) architectures.
MS Windows 64 bit Windows Vista or newer on AMD- or Intel-based
(x64) architectures.
Linux 64 bit AMD- or Intel-based 64-bit (x64) Linux systems
with glibc 2.7 or higher.
MacOS X Intel-based 64-bit (x64) Macintosh system with
MacOS X 10.10 (Yosemite) or higher.
Solaris i86pc AMD- or Intel-based 64-bit (x64) Solaris system.
Built on Solaris 11.0.
Solaris SPARC 64bit Sparc-based 64-bit (sparc-64) Solaris system.
Built on Solaris 10.
IBM AIX PowerPC based 64-bit (ppc-64) AIX system. Built
on AIX 6.1
The GAMS interface is depicted in Fig. 1.4.
Fig. 1.4 The GAMS interface

1.3 GAMS Elements
Each GAMS model consists of the following main elements:

• Sets: sets are used to define the indices in the algebraic representations of models.
For example, set of generating units, set of network buses, set of slack buses, set
of time periods, etc.
• Data: The input data of each GAMS model are expressed in the form of
Parameters, Tables, or Scalars. The parameters and tables are defined over the
sets. The scalars are single value quantities.
• Variables: The variables are decision sets and are unknown before solving the
model.
• Equations: The equations describe the relations between the data and variables.
• Model and Solve Statements: The model is defined as a set of equations which
contain an objective function. The solve statement asks GAMS to solve the
model.
• Output: There are several ways to see the outputs of the solved model such as
saving them in XLS file and displaying them.
The General GAMS code structure and elements are shown in Fig. 1.5.
The GAMS elements are explained in a simple example as follows: Suppose a
factory produces tree types of products P1;2;3 . Each product should be processed on
two different machines. The available machine hours per day and the time required
Fig. 1.5 General GAMS code structure and elements

1.3 GAMS Elements 9
Table 1.1 Data for Required time for task completion (h)
illustrative example
Machine P1 P2 P3
M1 2 5 2
M2 3 4 1
Profit per kg ($/kg) Machine availability (h)
P1 10 M1 16
P2 12 M2 12
P3 13.5
for each product are considered as the input data. The profits/kg of each item is also
known. The input data of this example is described in Table 1.1.
• Sets: Two sets should be defined: machines (M) and products (P) Sets M=m1
m2=; P=p1 p3=;
• Data: Looking at Table 1.1 shows that we have a table (required time for task
completion) and two parameters (Profit per item and Machine availability). The
task completion table is defined over two sets (machines and products), the profit
per item parameter is defined over products while the machine availability is
defined over product set. It is assumed that the minimum required amount of
each product is 1 kg.
Parameter profit(P)
/p1 10
p2 12
p3 13.5/;
Parameter availability(M)
/m1 16
m2 12/;
Table task(m,p)
p1 p2 p3
m1 2 5 2
m2 3 4 1;
• Variables: We are about to decide how many kg should be produced of each

product. This variable is defined over product set. Another variable should be
defined as the total profits which should be optimized. This variable (objective
function) should not have any set index. Variables are OF, X.p/.
• Equations: One equation should describe the objective function and another one
should enforce the constraint for availability of each machine.
Equations eq1, eq2;

eq1(m).. sum(p, task(m,p)*x(p))=l=availability(M);
eq2 .. of=e=sum(p, profit(p)*x(p));
As it can be seen in this example, there are two types of equations as

follows:
– Scalar equation: eq2 is an example of a scalar equation which is not defined
over any set.
– Indexed equation: eq1 is an example of the indexed equation. We have to
define the equation on index m since the index m exists in the formulation. In
other words, the equation is valid for any element in the set M.
Three different relations can be defined in equations as follows:
– =e= Equality: this means that both sides of the equations should be equal to
each other.
– =g= Greater than or equal: this means that the left-hand side of the equation
should be greater than or equal to the right-hand side of the equation
– =l= Less than or equal: this means that the left-hand side of the equation
should be less than or equal to the right-hand side of the equation
• Model: This example has two equations. Both should be included in the model
definition.
Model example /all/;
It can also be defined as follows:
Model example /eq1,eq2/;
• Solve statement: The solve statement tells GAMS that the model is linear and the
direction of the optimization should be toward maximizing the total benefits.
Solve example us LP Max OF;
The model type used in this code is linear programming (LP). There are
various models that can be coded in GAMS coding as follows:
1.3 GAMS Elements 11
LP: linear programming

QCP: Quadratic programming (the model can only contain linear and
quadratic terms)
NLP: Nonlinear problem with continuous constraints
DNLP: Nonlinear problem with discontinuous constraints
MIP: Mixed-integer linear programming
MIQCP: Mixed-integer quadratic constraint programming
MINLP: Mixed-integer nonlinear programming
RMIP: Mixed-integer problem where the integer variables are relaxed
• Output: The outputs of the GAMS model can be displayed and also saved in an
XSL file.
Display X . l , Of . l ;
execute u n l o a d ” Example . gdx ” X . l
execute ’ gdxxrw . e x e Example . gdx v a r =X r n g = P r o d u c t ! a1 ’
execute u n l o a d ” Example . gdx ” OF . l
execute ’ gdxxrw . e x e Example . gdx v a r =OF r n g =OF ! a1 ’
The overall GAMS code for solving the illustrative example is provided in
GCode 1.1.
GCode 1.1 Illustrative GAMS example

S e t s M / m1m2 / , P / p1p3 / ;
v a r i a b l e s OF , X( p ) ;
parameter p r o f i t (P)
/ p1 10
p2 12
p3 1 3 . 5 / ;
p a r a m e t e r a v a i l a b i l i t y (M)
/ m1 16
m2 1 2 / ;
T a b l e t a s k (m, p )
p1 p2 p3
m1 2 5 2
m2 3 4 1 ;
e q u a t i o n s eq1 , eq2 ;
eq1 (m ) . . sum ( p , t a s k (m, p ) x ( p ) ) = l = a v a i l a b i l i t y (M) ;
eq2 .. o f = l =sum ( p , p r o f i t ( p ) x ( p ) ) ;
X. lo ( p )=1;
model e x a m p l e / a l l / ;
S o l v e e x a m p l e u s LP max OF ;
d i s p l a y X . l , Of . l ;
e x e c u t e u n l o a d ” Example . gdx ” X . l
e x e c u t e ’ gdxxrw . e x e Example . gdx v a r =X r n g = P r o d u c t ! a1 ’
e x e c u t e u n l o a d ” Example . gdx ” OF . l
e x e c u t e ’ gdxxrw . e x e Example . gdx v a r =OF r n g =OF ! a1 ’
Fig. 1.6 The structure of

GAMS listing file
Once the model is solved the solve summary is available by clicking on the file.lst.
The structure of GAMS listing file is shown in Fig. 1.6.
The listing file is explained as follows:
SOLVESUMMARY
MODEL example OBJECTIVE OF
TYPE LP DIRECTION MAXIMIZE
SOLVER CPLEX FROM LINE 23
**** SOLVER STATUS 1 Normal Completion
**** MODEL STATUS 1 Optimal
**** OBJECTIVE VALUE 82.7500
The summary shows that the name of the model is example. The objective
function to be optimized is named OF. The type of the model is linear programming
(LP). The direction of the optimization is maximization. The solver used for solving
the LP model is CPLEX. The solver status is 1 which means that the model is
normally solved without error. The model status is 1 optimal. It means that the
global optimal solution is found. The value of the objective function (OF) is 82.75.
Different solver status might be reported once the model is solved.
1.3 GAMS Elements 13
SOLVER STATUS CODE DESCRIPTION

1 Normal Completion
2 Iteration Interrupt
3 Resource Interrupt
4 Terminated by Solver
5 Evaluation Error Limit
6 Capability Problems
7 Licensing Problems
8 User Interrupt
9 Error Setup Failure
10 Error Solver Failure
11 Error Internal Solver Error
12 Solve Processing Skipped
13 Error System Failure
Different model status might be reported once the model is solved.
MODEL STATUS CODE DESCRIPTION

1 Optimal
2 Locally Optimal
3 Unbounded
4 Infeasible
5 Locally Infeasible
6 Intermediate Infeasible
7 Intermediate Nonoptimal
8 Integer Solution
9 Intermediate Non-Integer
10 Integer Infeasible
11 Licensing Problems - No Solution
12 Error Unknown
13 Error No Solution
14 No Solution Returned
15 Solved Unique
16 Solved
17 Solved Singular
18 Unbounded - No Solution
19 Infeasible - No Solution
The model statistic can provide some useful information regarding the developed
model. It is indicating that there are two blocks of equations (eq1,eq2) in the
developed model. The single equations are three because eq1 has two single
equations (m1; m2) and the eq2 has only one equation. There are two blocks of
variables x, OF but since X has three variables (p1; p2; p3), then the number of total
variables is 4.
MODEL STATISTICS
BLOCKS OF EQUATIONS 2 SINGLE EQUATIONS 3
BLOCKS OF VARIABLES 2 SINGLE VARIABLES 4
By clicking on the SolVar tab, some useful information regarding the variables
will be obtained as follows:
LOWER LEVEL UPPER MARGINAL

—- VAR OF -INF 82.750 +INF .
—- VAR X
p1 1.000 1.000 +INF -3.500
p2 1.000 1.000 +INF -21.750
p3 1.000 4.500 +INF .
This means that the minimum limit of variable X is 1 and the upper limits
are C1. The marginal values are also revealed. Each variable in GAMS has five
attributes as follows:
• Variable.lo: indicates the minimum limit of a variable
• Variable.up: indicates the maximum limit of a variable
• Variable.l: indicates the level of the variable. In other words, this is the actual
value of the variable.
• Variable.fx: indicates the level of the variable is fixed and is not changing. It has
the same impact of defining the low and up attributes the same as each other.
• Variable.m: indicates the marginal value of a variable. It shows how much
sensitive is the objective function to the changes of this variable. For example,
X:m.p1/ D 3:5 this means that for 1 unit of increase in X.p1/ the objective
function will reduce by 3:5.
The variables used in this code are all real variables; however, different types of
variables can be defined in GAMS as follows:
1.4 Conditional Statements 15
Variable type Description Lower limit Upper limit

Free No bounds on variable. Both bounds can be 1 C1
changed from the default values by the user
Positive The positive variable can only take positive 0 C1
values
Negative The negative variable can only take negative 1 0
values
Binary Discrete variable that can only take values of 0 1
0 or 1
Integer Discrete variable that can only take integer val- 0 100
ues between the limits
It should be noted that the default values for variable limits can be modified by
the user.
1.4 Conditional Statements
The Dollar ($) condition is broadly used in GAMS coding. Various applications are
explained through some examples.
• Suppose we need to have a conditional statement like this if A D C then B D D.
This statement is modeled as follows:
(B=D)$(A=C)
If (A ¤ C) then no assignment to B will happen.

• If A D C then B D D otherwise B D 0. This statement is modeled as follows:
B=D$(A=C)
• Suppose that we need to filter some elements in eq1(m). In other words, this
equation should be valid for every m elements except some of them.
eq1(m)$(ord(m)=1) .. sum(p, task(m,p)*x(p))=l=availability(M);
This would force the equation only for m1.

eq1(m)$(ord(m) 2) .. sum(p, task(m,p)*x(p))=l=availability(M);
This would force the equation only for m2 and m3. It should be noted that no
variable can exist in conditional statement.
• If A and B then C D D, this is coded as follows:
(C=D)$( A and B)
1.5 Loop Definition in GAMS
The looping is used for executing one or more statements repeatedly. There are
different forms of looping in GAMS as follows:
1.5.1 LOOP Statement
One of the techniques for defining the loop in GAMS is loop statement.
Loop(setname,
Statement 1 ;
Statement 2 ;
Statement 3 ;
);
The statements 1–3 are executed N times. N is equal to the cardinality of the
“setname.” These statements can assign values to the variable limits or describe the
relation between some parameters (not the variables). No variable can appear in the
loop statements.
set counter /c1*c4/;

Loop(counter,
X.lo(p)=0.1*ord(counter);
Solve example us LP max OF;
);
1.5 Loop Definition in GAMS 17
In the first counter, the minimum values of all X variables would be 0:11 and the
model is solved. The second counter sets the minimum values of X variable equal
to 0:1 2 D 0:2 and then the model is solved. The point that should be noted here
is that the set/parameter/variable/model definition should be outside the loop.
This kind of modeling would be useful in conducting sensitivity analysis. Some
parameters are varied and then the impacts on the objective function are investigated
as follows:
set counter /c1*c4/;

parameter report(counter,*);
Loop(counter,
X.lo(p)=0.1*ord(counter);
report(counter,’lowerlimit’)=0.1*ord(counter);
report(counter,’OF’)=OF.l;
);
1.5.2 WHILE Statement
One of the techniques for defining the loop in GAMS is while statement.
while (condition,
Statement 1 ;
Statement 2 ;
Statement 3 ;
);
The statements 1–3 are executed as far as the condition is true. The condition will
be changed inside the loop based on some logic otherwise the loop will continue
forever!
1.5.3 FOR Statement
One of the techniques for defining the loop in GAMS is FOR statement.
scalar Itermax /10/;

scalar iteration ;
for (iteration=1 to Itermax,
Statement 1 ;
Statement 2 ;
Statement 3 ;
);
The statements 1–3 are executed for Itermax times.
1.5.4 REPEAT-UNTIL Statement
One of the techniques for defining the loop in GAMS is Repeat-until statement.
repeat ( Statement 1 ;
Statement 2 ;
Statement 3 ;
until condition );
The statements 1–3 are executed until the condition becomes logically true (it
should be initially false).
1.6 Linking GAMS and Excels
1.6.1 Reading from Excel
GAMS is able to read the data from xls files. It is also convenient to use xls as
an interface between GAMS and other platforms. Reading the data from xls file is
straightforward as follows:
P a r a m e t e r L e v e l (m, p ) ;
n$CALL GDXXRW. EXE Example . x l s p a r = L e v e l r n g = t a s k ! E5 : H7
n$GDXIN Example . gdx
n$LOAD L e v e l
In this case, a parameter like Level (m; p) is defined. The xls file name
(Example.xls) is indicated and the range of data is specified (sheet=task, cells:
1.7 Error Debugging in GAMS 19
E5:H7). The data is read from xls file and is written on Example.gdx file. The
parameter level is loaded and can be used in the code.
1.6.2 Writing to Excel
It is usually desirable to save the output of a GAMS code in an excel file. The
procedure is quite straightforward.
For variables, the following commands should be executed. The user can indicate
which variable (X.l,OF.l) should be written in Excel file. The name of the excel
file (Example.xls) and what would be the sheet name (product, OF). The cell a1 is
chosen as the starting cell in the specified sheet for writing the data.
execute u n l o a d ” Example . gdx ” X . l

execute ’ gdxxrw . e x e Example . gdx v a r =X r n g = P r o d u c t ! a1 ’
execute u n l o a d ” Example . gdx ” OF . l
execute ’ gdxxrw . e x e Example . gdx v a r =OF r n g =OF ! a1 ’
If the user wants to write the parameter X to an excel file then the command
would be as follows:
e x e c u t e u n l o a d ” Example . gdx ” X
e x e c u t e ’ gdxxrw . e x e Example . gdx P a r =X r n g = P r o d u c t ! a1 ’
In this case, no ‘.l’ is needed since it is a parameter, not a variable.
1.7 Error Debugging in GAMS
Once the code is written the “Shift key + F9” should be pressed. The GAMS
compiler will check the developed code without executing it. The coding errors
are unavoidable in every programming language, and GAMS is not an exception.
Usually, a list of errors is generated which should be taken care of. The first thing to
keep in mind is to start from the first error and fix them one by one. In this section,
some common coding errors and best way of debugging them are explained.
• Error 140: Unknown symbol: GAMS compiler generates this error when there
are some undefined symbols in the model.
• Error 246: Objective variable is not a free variable
This error will usually happen when the objective function (which should be
minimized/maximized) is not set as a free variable. For example, it is defined as
a positive/negative variable. In case that the decision maker needs the objective
function be a positive variable, the lower limit of the variable should be specified
using :lo D 0 command.
• Error: Unbounded variable:

This means that the objective function is equal to 1 or C1. This usually
happens when the bounds of the objective function are not properly defined, or
optimization direction is not properly defined in solve statement. For example, if
the objective is to be minimized but it is maximized by mistake. The following
error will be generated by GAMS.
**** ERRORS/WARNINGS IN VARIABLE OF

1 error(s): Unbounded variable
• Error 149: Uncontrolled set entered as constant

This error usually happens when the statements are not properly written.
Consider the following line of code
eq1 .. sum(p, task(m,p)*x(p))=l=availability(M);
The equation should be valid for every m, but the equation label is not defined
over the set m. The correct format for defining this equation is as follows:
eq1(m) .. sum(p, task(m,p)*x(p))=l=availability(M);
• Error 143: A suffix is missing

This error will happen when a variable is used outside of the equation
environment. For example, the following line is trying to assign a value to
variable X.p/ but not in any equation:
x(p)=1;
This is not a valid way of doing this. If the variable X.p/ should be fixed to a
constant value it should be defined as follows:
x.fx(p)=1;
As a general rule, the variables cannot appear outside the equations. Only
the variable’s attributes can be modified outside the equation environment.
This means X:lo.p/, X:l.p/, X:up.p/ can be changed without using equation
environment.
Another situation that this error might happen is in display command.
Consider the following part of a GAMS code:
Solve example us LP max Z; Display X , Of.l;
This will generate the following error:
143 A suffix is missing

**** 1 ERROR(S) 0 WARNING(S)
This is because X is a variable and display command can only show the
attributes of the X variable. The correct form for showing the value of X variable
is as follows:
solve example us LP max Z;

Display X.l , Of.l;
• Error 245: Objective variable not referenced in model

This means that the solve statement is trying to minimize/maximize a variable
which does not exist in the defined model.
• Error 141: Symbol declared, but no values have been assigned
This error happens when the variable or parameter is defined, but it is not
implicitly assigned any value and also does not appear in any equation. In other
words, GAMS has no info about this symbol, but we are asking GAMS some
information about it. Consider the following GAMS code:
Sets M /m1*m2/, P /p1*p3/ ;

Variables OF,x(p),Z;
parameter profit(P)
/p1 10
p2 12
p3 13.5/;
Parameter availability (M)
/m1 16
(continued)
m2 12/;
Table task(m,p)
p1 p2 p3
m1 2 5 2
m2 3 4 1 ;
Equations eq1,eq2;
eq2 .. of=l=sum(p, profit(p)*x(p));
X.lo(p)=1;
display X.l , Of.l , Z.l;
This code would generate Error 141 because although variable Z is defined but
since it has no initial assignment and does not appear in the model then GAMS
knows nothing about it.
• Error 225: Floating entry ignored
This error is a very common error specially for beginners. This happens when
a table is not properly typed in GAMS. All data should be aligned under the
column label.
P1 P2 P3
M1 2 5 2
M2 3 4 1
The correct form of typing the above table is as follows:
P1 P2 P3
M1 2 5 2
M2 3 4 1
Handling the large tables in GAMS would be much easier if the tool named
“xls2gams.exe” is used. This tool is available in the same folder that GAMS
is installed. The function of this tool is copying data from xls file into GAMS
format. Using this tool to import xls data to GAMS is highly recommended.
• Error 56: Endogenous operands for * not allowed in linear models
Consider the following code:
Sets M /m1*m2/, P /p1*p3/ ;

Variables OF,x(p),Z;
parameter profit(P)
/p1 10
p2 12
p3 13.5/;
Parameter availability (M)
/m1 16
m2 12/;
Table task(m,p)
p1 p2 p3
m1 2 5 2
m2 3 4 1 ;
Equations eq1,eq2;
eq2 .. of=l=sum(p, profit(p)*x(p)*x(p));
X.lo(p)=1;
display X.l , Of.l , Z.l;
The following error will be generated after compiling this code. In order to
find out why this error is happening we should double-check the model. The
solve statement is indicating that the model is LP but in eq2 we can see profit.p/
x.p/ x.p/. This means that the model is not LP and solve statement should be
modified as follows:
Solve example us NLP max OF;

or
Solve example us QCP max OF;
• Error 37:
Consider the following line in a GAMS code:
eq2 .. of=sum(p, profit(p)*x(p));

This line is not properly defined because the correct operator for stating the
equality is not used in equation environment. The only operators that can be used
are ‘=l=’ or ‘=e=’ or ‘=g=’. the correct format is as follows:
eq2 .. of=e=sum(p, profit(p)*x(p));
• Error 409:
Consider the following lines in a GAMS code:
eq1(m) .. sum(p, task(m,p)*x(p))=l=availability(M)

eq2 .. of=sum(p, profit(p)*x(p)*x(p));
This is because a semicolon (;) is missing at the end of eq1.
1.8 General Programming Remarks
Like other programming languages, there are different ways of modeling the
problem in GAMS. The so-called “best way” is yet to be discovered. Some
programming tips are given in this section as follows:
• GAMS compiler is not case sensitive. The lowercase letters and capital letters
are not distinguished. For example, P.t/ is the same as p.T/ or P.T/ or p.t/.
• Try to add explanations for your equations and symbols you define in your code.
Use * for commenting a line or $ontext $offtext for a block of comments.
• Choose the symbol names carefully and meaningfully. For example for the power
generated by power plants, P.g/ is a better choice than X.i/. You can define g as
the set of generating units and variable P as the generated power.
• Don’t get frightened by the errors once your code is compiled. The best thing to
see is the error flag. Sometimes although the model does not generate any error
flag (since you have obeyed the GAMS syntax) but it is not correctly modeling
the given problem.
• Keep in mind that the machine is always right so GAMS cannot go wrong.
Debugging is also a part of the coding process.
• Always check the GAMS output. The variables, model statistics, and solver
statistics should be checked to see if any flag is raised. If the output of GAMS
model is not what you expect then there should be something wrong with your
model (not the GAMS). Double-check the equations and see if the developed
model is actually what it is desired to be or not.
1.8 General Programming Remarks 25
• If you need to specify some limits for your variables then try to do it using .lo
or .up attributes. Although creating a new equation for modeling this constraint
is “legal”, however, it is totally “inappropriate”. For example, you need to state
that the minimum value of variable P.g/ is 1 then it can be easily done using
P.lo(g) = 1. Sometimes the variables should be limited using some tables of
parameters. For example, the minimum values of P.g/ are stored in a parameter
Pmin(g) then P.lo(g) = Pmin(g) will do the job. As far as no variable exists then
no equation is needed. If Pmin(g) is a variable, then the following line will be
appropriate for modeling this constraint.
eq(g) .. P(g)=g=Pmin(g);
It should be noted that eq is defined over the set g since the inequality should
be valid for every element belonging to the set g.
• Although GAMS is a robust programming language less number of variables and
equations are always welcome. Try to avoid unnecessary equations and variables.
Use filtering to omit unnecessary equations. For example, we need to define an
equation which calculates the power flow of a line connecting ‘bus’ to ‘node’
in DC power flow formulation. The following GAMS code might be the first
attempt to model it.
Eq(bus,node) .. Pij(bus,node)=e=(d(bus)-d(node))/data(bus,node,’x’);
Suppose the network has 186 lines and 118 buses. This kind of coding would
create two problems:
– The parameter data(bus, node,‘x’) is zero when bus and node are not
connected to each other. GAMS might generate a division by Zero error! It’s
not always easy to find and remove the cause of error.
– If you check the model you can see that this equation adds 13,924 single
equations to the model. You might ask yourself why? We were expecting to
have only 186 equations (or double that since flow in opposite direction should
also be calculated). It is almost 74 times bigger than what we were expecting
to have. This is because GAMS is considering every combination between
“bus” and “node” (118*118).
The better way to code this line is as follows:
Eq(bus,node)$branch(bus,node,’x’)..Pij(bus,node)=e=(d(bus)-
d(node))/data(bus,node,’x’);
In this way, the equation is filtered and is only valid for those lines which have
nonzero reactance values. It also reduces the computation time and the chance of
difficulties in finding a feasible/optimal solution.
• Don’t forget to put a semicolon (;) at the end of every line (sometimes it is not
necessary but if you have just started coding in GAMS, do it).
• Different solvers might use different approaches to solve the model. Try different
solver to find out which one is more successful/faster in finding a solution for
your problem.
• The default value for undefined elements of a table or parameter is zero. For
example, consider a parameter like Data(bus,node) representing the line reac-
tance; Suppose that we define Data(‘2’,‘4’)=0.5; If this parameter is displayed,
you can see that all elements are zero except Data(‘2’,‘4’). If you are a power
engineer, you know that the reactance between bus number 2 and bus number
4 should be the same as reactance between bus number 4 and bus number 2. In
other words, it should be symmetrical. Unfortunately (or fortunately) GAMS is
not a power system engineer and does not understand even a word from power
system or electrical engineering. GAMS only understands whatever it is told by
the user (code developer).
• Go to your GAMS model library and have a look at those developed codes. There
are loads of new things to learn even for experienced GAMS code developers.
The GAMS model library can be accessed through the interface as shown in
Fig. 1.7.
• Try to provide all of the variables appropriate upper and lower limits as well as
initial starting values. The initial values can be assigned to the variables using .l
expression (do it before solving statement). For example, voltage variables can
be assigned 1 per unit values as the initial starting values. This is a double-edged
sword. This is because providing a good starting value can highly improve the
solution procedure especially in nonlinear or mixed integer nonlinear models but
a poor starting value might slowdown the solver.
• It is very likely that you get an infeasible status for your model. Especially when
the model is large and includes a large number of variables. The following steps
might be helpful in resolving the problem:
Fig. 1.7 GAMS model library

1.9 Book Structure 27
Fig. 1.8 GAMS help menu
– Check to see if you can better express your model. It is always desired to stay
as close as possible to the linear form of expressing the equations.
– Provide better initial values for your variables.
– Relax the variable limits, rerun the model. If the problem is resolved, then
it means that the variable limits should be revised. If your model contains
integer/binary variables, then you can solve the model using relaxed option.
For example, the MIP model can be solved using RMIP. This would ask the
GAMS to neglect integer nature of the variables (the variable limits remain
unchanged). For relaxing the MINLP and MIQCP, you should use RMINLP
and RMIQCP, respectively.
– Remove some equations or add some slack variables to the model to see if you
can find the trouble making equations
– Ask support from those experienced GAMS code developers (if they have
time and are willing to contribute to your project). This option is intentionally
placed at the end of suggestion list.
There is a set of good resources that can be accessed through the GAMS interface as
it is shown in Fig. 1.8. Additionally, some other useful tutorials are listed as follows:
• A GAMS tutorial [70]
• GAMS language guide [71]
• GAMS—Modeling and Solving Optimization Problems [72]
• GAMS and MATLAB interface [73–75]
• Grid-enabled optimization with GAMS [76]
• Practical financial optimization in GAMS [77]
1.9 Book Structure
The objectives pursued in this book are the following:

• Familiarizing the readers with optimization concept through multiple examples
• Introducing different optimization problems that the decision makers might face
in power system studies
• Providing robust solutions for different operation/planning problems in power

system studies
• Exploring the capabilities of GAMS in conducting sensitivity analysis
This book consists of ten chapters. A brief description of each chapter is provided
here:
• Chapter 1 provides the instructions on how to start coding in GAMS. Different
programming elements and some coding tricks are introduced and explained.
Reading this chapter would be helpful to understand what to expect from GAMS
in dealing with optimization problems.
• Chapter 2 provides some insights to the reader on how to solve simple opti-
mization problems through some illustrative examples. These examples are pure
mathematics, and no power system or electrical engineering background is
required for the readers. This chapter can also be used for those who might be
only interested in learning GAMS for interdisciplinary applications.
• Chapter 3 discusses how to model the dispatching problem of different power
plants in a single snapshot (single period). Different power plant technologies are
explained such as thermal power, wind turbine, CHP, and hydro-power plants.
• Chapter 4 provides a solution for dynamic economic dispatch problem (multi-
period problems). The time-dependent optimal dispatch decisions are modeled
and solved.
• Chapter 5 explains how unit commitment problem can be modeled and solved
in GAMS. The developed codes in this chapter are linear and categorized as
mixed integer linear programming (MIP) models. The inputs are generator’s
characteristics, electricity prices, and demands. The outputs of these codes are
on/off status of units and their operating schedules.
• Chapter 6 provides a solution for optimal power flow (OPF) problem. The
active/reactive power output of generating units as well as the network variables
(voltage magnitudes and angles) are determined in OPF to minimize total
operating costs. Different OPF models are investigated such as single and multi-
period DC/AC optimal power flow. The appropriate linear/nonlinear models are
developed and solved in this chapter.
• Chapter 7 provides a solution for modeling the operation and planning problems
of energy storage systems (ESS). The inputs are generator’s characteristics,
electricity prices, demands, and network topology. The outputs of this code are
operating schedules/locations and sizes of ESS units.
• Chapter 8 provides a solution for allocation of Phasor Measurement Units (PMU)
problem in transmission networks to maximize the power system observability.
The PMU can measure the voltage phasor at the connection bus, and also it
measures the current phasor of any branch connected to the bus hosting the PMU.
Different cases are analyzed, and the problem is tested on some standard IEEE
cases.
• Chapter 9 provides a solution for some transmission network operation and
planning studies in GAMS. The transmission investment regarding building
new lines and power flow controllers (phase shifter), sensitivity factors, and
References 29
transmission switching have been discussed in this chapter. The GAMS code
for solving each optimization problem is developed and discussed.
• Chapter 10 provides a solution for Energy System Integration (ESI) problem in
GAMS. The question which is answered in this chapter is how to harvest the
flexibilities in different energy sectors by coordinated operation of these sectors.
References
1. J.L.C. Meza, M.B. Yildirim, A.S.M. Masud, A model for the multiperiod multiobjective power
generation expansion problem. IEEE Trans. Power Syst. 22(2), 871–878 (2007)
2. S. Kannan, S. Baskar, J.D. McCalley, P. Murugan, Application of NSGA-II algorithm to
generation expansion planning. IEEE Trans. Power Syst. 24(1), 454–461 (2009)
3. R. Fang, D.J. Hill, A new strategy for transmission expansion in competitive electricity
markets. IEEE Trans. Power Syst. 18(1), 374–380 (2003)
4. R. Romero, A. Monticelli, A. Garcia, S. Haffner, Test systems and mathematical models for
transmission network expansion planning. IEE Proc. Gener. Transm. Distrib. 149(1), 27–36
(2002)
5. V. Miranda, J.V. Ranito, L.M. Proenca, Genetic algorithms in optimal multistage distribution
network planning. IEEE Trans. Power Syst. 9(4), 1927–1933 (1994)
6. S. Gerbex, R. Cherkaoui, A.J. Germond, Optimal location of multi-type facts devices in a
power system by means of genetic algorithms. IEEE Trans. Power Syst. 16(3), 537–544 (2001)
7. C. Wang, M.H. Nehrir, Analytical approaches for optimal placement of distributed generation
sources in power systems. IEEE Trans. Power Syst. 19(4), 2068–2076 (2004)
8. W. El-Khattam, K. Bhattacharya, Y. Hegazy, M.M.A. Salama, Optimal investment planning
for distributed generation in a competitive electricity market. IEEE Trans. Power Syst. 19(3),
1674–1684 (2004)
9. A. Keane, M. O’Malley, Optimal allocation of embedded generation on distribution networks.
IEEE Trans. Power Syst. 20(3), 1640–1646 (2005)
10. S. Sundhararajan, A. Pahwa, Optimal selection of capacitors for radial distribution systems
using a genetic algorithm. IEEE Trans. Power Syst. 9(3), 1499–1507 (1994)
11. B. Milosevic, M. Begovic, Nondominated sorting genetic algorithm for optimal phasor
measurement placement. IEEE Trans. Power Syst. 18(1), 69–75 (2003)
12. B. Gou, Generalized integer linear programming formulation for optimal PMU placement.
13. Y.M. Atwa, E.F. El-Saadany, Optimal allocation of ESS in distribution systems with a high
penetration of wind energy. IEEE Trans. Power Syst. 25(4), 1815–1822 (2010)
14. H. Oh, Optimal planning to include storage devices in power systems. IEEE Trans. Power Syst.
26(3), 1118–1128 (2011)
15. P. Maghouli, A. Soroudi, A. Keane, Robust computational framework for mid-term techno-
economical assessment of energy storage. IET Gener. Transm. Distrib. 10(3), 822–831 (2016)
16. A. Soroudi, M. Afrasiab, Binary PSO-based dynamic multi-objective model for distributed
generation planning under uncertainty. IET Renew. Power Gener. 6(2), 67–78 (2012)
17. G. Blanco, F. Olsina, F. Garces, C. Rehtanz, Real option valuation of facts investments based
on the least square monte carlo method. IEEE Trans. Power Syst. 26(3), 1389–1398 (2011)
18. B. Chen, J. Wang, L. Wang, Y. He, Z. Wang, Robust optimization for transmission expansion
planning: minimax cost vs. minimax regret. IEEE Trans. Power Syst. 29(6), 3069–3077 (2014)
19. V. Miranda, M.A.C.C. Matos, Distribution system planning with fuzzy models and techniques,
in 10th International Conference on Electricity Distribution, 1989 (CIRED 1989), vol. 6,
Brighton (1989), pp. 472–476
20. S. Civanlar, J.J. Grainger, H. Yin, S.S.H. Lee, Distribution feeder reconfiguration for loss
reduction. IEEE Trans. Power Delivery 3(3), 1217–1223 (1988)
21. M. Carrion, J.M. Arroyo, A computationally efficient mixed-integer linear formulation for the
thermal unit commitment problem. IEEE Trans. Power Syst. 21(3), 1371–1378 (2006)
22. J.M. Arroyo, A.J. Conejo, Optimal response of a thermal unit to an electricity spot market.
23. D.W. Ross, S. Kim, Dynamic economic dispatch of generation. IEEE Trans. Power Apparatus
Syst. PAS-99(6), 2060–2068 (1980)
24. Z.-L. Gaing, Particle swarm optimization to solving the economic dispatch considering the
generator constraints. IEEE Trans. Power Syst. 18(3), 1187–1195 (2003)
25. J. Endrenyi, S. Aboresheid, R.N. Allan, G.J. Anders, S. Asgarpoor, R. Billinton, N. Chowdhury,
E.N. Dialynas, M. Fipper, R.H. Fletcher, C. Grigg, J. McCalley, S. Meliopoulos, T.C. Mielnik,
P. Nitu, N. Rau, N.D. Reppen, L. Salvaderi, A. Schneider, C. Singh, The present status of
maintenance strategies and the impact of maintenance on reliability. IEEE Trans. Power Syst.
16(4), 638–646 (2001)
26. A.G. Bakirtzis, P.N. Biskas, A decentralized solution to the DC-OPF of interconnected power
systems. IEEE Trans. Power Syst. 18(3), 1007–1013 (2003)
27. 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 Trans. Power
Syst. 26(1), 12–19 (2011)
28. Y.M. Atwa, E.F. El-Saadany, M.M.A. Salama, R. Seethapathy, Optimal renewable resources
mix for distribution system energy loss minimization. IEEE Trans. Power Syst. 25(1), 360–
370 (2010)
29. K.R.C. Mamandur, R.D. Chenoweth, Optimal control of reactive power flow for improvements
in voltage profiles and for real power loss minimization. IEEE Trans. Power Apparatus Syst.
PAS-100(7), 3185–3194 (1981)
30. Y. Bai, H. Zhong, Q. Xia, C. Kang, A two-level approach to ac optimal transmission switching
with an accelerating technique. IEEE Trans. Power Syst. 32(2), 1616–1625 (2017)
31. N. Rotering, M. Ilic, Optimal charge control of plug-in hybrid electric vehicles in deregulated
electricity markets. IEEE Trans. Power Syst. 26(3), 1021–1029 (2011)
32. F. Wen, A.K. David, Optimal bidding strategies and modeling of imperfect information among
competitive generators. IEEE Trans. Power Syst. 16(1), 15–21 (2001)
33. A. Baillo, M. Ventosa, M. Rivier, A. Ramos, Optimal offering strategies for generation
companies operating in electricity spot markets. IEEE Trans. Power Syst. 19(2), 745–753
(2004)
34. H.B. Gooi, D.P. Mendes, K.R.W. Bell, D.S. Kirschen, Optimal scheduling of spinning reserve.
35. K.A. Papadogiannis, N.D. Hatziargyriou, Optimal allocation of primary reserve services in
energy markets. IEEE Trans. Power Syst. 19(1), 652–659 (2004)
36. A. Soroudi, P. Siano, A. Keane, Optimal DR and ESS scheduling for distribution losses
payments minimization under electricity price uncertainty. IEEE Trans. Smart Grid 7(1), 261–
272 (2016)
37. S. Dutta, S.P. Singh, Optimal rescheduling of generators for congestion management based on
particle swarm optimization. IEEE Trans. Power Syst. 23(4), 1560–1569 (2008)
38. C. Murphy, A. Soroudi, A. Keane, Information gap decision theory-based congestion and
voltage management in the presence of uncertain wind power. IEEE Trans. Sust. Energy 7(2),
841–849 (2016)
39. B. Hayes, I. Hernando-Gil, A. Collin, G. Harrison, S. Djoki, Optimal power flow for
maximizing network benefits from demand-side management. IEEE Trans. Power Syst. 29(4),
1739–1747 (2014)
40. A.J. Conejo, J.M. Morales, L. Baringo, Real-time demand response model. IEEE Trans. Smart
Grid 1(3), 236–242 (2010)
41. A. Rabiee, A. Soroudi, A. Keane, Information gap decision theory based OPF with HVDC
connected wind farms. IEEE Trans. Power Syst. 30(6), 3396–3406 (2015)
References 31
42. A. Soroudi, A. Rabiee, A. Keane, Information gap decision theory approach to deal with wind
power uncertainty in unit commitment. Electr. Power Syst. Res. 145, 137–148 (2017)
43. A.J. Conejo, M. Carrión, J.M. Morales, Decision Making Under Uncertainty in Electricity
Markets, vol. 1 (Springer, New York, 2010)
44. E. Zio, The Monte Carlo Simulation Method for System Reliability and Risk Analysis (Springer,
London, 2013)
45. A. Soroudi, Possibilistic-scenario model for DG impact assessment on distribution networks
in an uncertain environment. IEEE Trans. Power Syst. 27(3), 1283–1293 (2012)
46. S. Granville, Optimal reactive dispatch through interior point methods. IEEE Trans. Power
Syst. 9(1), 136–146 (1994)
47. T. Ding, R. Bo, F. Li, H. Sun, A bi-level branch and bound method for economic dispatch with
disjoint prohibited zones considering network losses. IEEE Trans. Power Syst. 30(6), 2841–
2855 (2015)
48. S. Binato, M.V.F. Pereira, S. Granville, A new benders decomposition approach to solve power
transmission network design problems. IEEE Trans. Power Syst. 16(2), 235–240 (2001)
49. D.K. Molzahn, J.T. Holzer, B.C. Lesieutre, C.L. DeMarco, Implementation of a large-scale
optimal power flow solver based on semidefinite programming. IEEE Trans. Power Syst. 28(4),
3987–3998 (2013)
50. R.A. Jabr, Exploiting sparsity in SDP relaxations of the OPF problem. IEEE Trans. Power Syst.
27(2), 1138–1139 (2012)
51. T. Wu, M. Rothleder, Z. Alaywan, A.D. Papalexopoulos, Pricing energy and ancillary services
in integrated market systems by an optimal power flow. IEEE Trans. Power Syst. 19(1), 339–
347 (2004)
52. J.M. Arroyo, F.D. Galiana, On the solution of the bilevel programming formulation of the
terrorist threat problem. IEEE Trans. Power Syst. 20(2), 789–797 (2005)
53. D.I. Sun, B. Ashley, B. Brewer, A. Hughes, W.F. Tinney, Optimal power flow by Newton
approach. IEEE Trans. Power Apparatus Syst. PAS-103(10), 2864–2880 (1984)
54. F. Milano, Continuous Newton’s method for power flow analysis. IEEE Trans. Power Syst.
24(1), 50–57 (2009)
55. E.C. Finardi, E.L. da Silva, Solving the hydro unit commitment problem via dual decomposi-
tion and sequential quadratic programming. IEEE Trans. Power Syst. 21(2), 835–844 (2006)
56. I.P. Abril, J.A.G. Quintero, Var compensation by sequential quadratic programming. IEEE
Trans. Power Syst. 18(1), 36–41 (2003)
57. B. Enacheanu, B. Raison, R. Caire, O. Devaux, W. Bienia, N. HadjSaid, Radial network
reconfiguration using genetic algorithm based on the matroid theory. IEEE Trans. Power Syst.
23(1), 186–195 (2008)
58. P. Maghouli, S.H. Hosseini, M.O. Buygi, M. Shahidehpour, A scenario-based multi-objective
model for multi-stage transmission expansion planning. IEEE Trans. Power Syst. 26(1), 470–
478 (2011)
59. T. Amraee, A.M Ranjbar, R. Feuillet. Immune-based selection of pilot nodes for secondary
voltage control. Eur. Trans. Electr. Power 20(7), 938–951 (2010)
60. T. Satoh, K. Nara, Maintenance scheduling by using simulated annealing method [for power
plants]. IEEE Trans. Power Syst. 6(2), 850–857 (1991)
61. J.G. Vlachogiannis, K.Y. Lee. Quantum-inspired evolutionary algorithm for real and reactive
power dispatch. IEEE Trans. Power Syst. 23(4), 1627–1636 (2008)
62. C. Dai, W. Chen, Y. Zhu, X. Zhang, Seeker optimization algorithm for optimal reactive power
dispatch. IEEE Trans. Power Syst. 24(3), 1218–1231 (2009)
63. T. Amraee, Coordination of directional overcurrent relays using seeker algorithm. IEEE Trans.
Power Delivery 27(3), 1415–1422 (2012)
64. D.N. Vo, P. Schegner, W. Ongsakul, Cuckoo search algorithm for non-convex economic
dispatch. IET Gener. Transm. Distrib. 7(6), 645–654 (2013)
65. A.A. El-fergany, A.Y. Abdelaziz, Capacitor allocations in radial distribution networks using
cuckoo search algorithm. IET Gener. Transm. Distrib. 8(2), 223–232 (2014)
66. M. Barati, M.M. Farsangi, Solving unit commitment problem by a binary shuffled frog leaping
algorithm. IET Gener. Transm. Distrib. 8(6), 1050–1060 (2014)
67. R. Roche, L. Idoumghar, B. Blunier, A. Miraoui, Imperialist competitive algorithm for dynamic
optimization of economic dispatch in power systems, in International Conference on Artificial
Evolution (Evolution Artificielle) (Springer, Berlin, 2011), pp. 217–228
68. W.-M. Lin, F.-S. Cheng, M.-T. Tsay, An improved tabu search for economic dispatch with
multiple minima. IEEE Trans. Power Syst. 17(1), 108–112 (2002)
69. C.F. Chang, Reconfiguration and capacitor placement for loss reduction of distribution systems
by ant colony search algorithm. IEEE Trans. Power Syst. 23(4), 1747–1755 (2008)
70. R.E. Rosenthal, A GAMS tutorial. Technical note (1992)
71. A. Brooke, D. Kendrick, A. Meeraus, R. Raman, R.E. Rosenthal, Gams. A Users Guide (GAMS
Development Corporation, Washington, DC, 2005)
72. A. Geletu, Gams-Modeling and Solving Optimization Problems (TU-Ilmenau, Faculty of Math-
ematics and Natural Sciences, Department of Operation Research & Stochastrics, Ilmenau,
2008)
73. M.C. Ferris, Matlab and GAMS: interfacing optimization and visualization software. Mathe-
matical Programming Technical Report, 98:19 (1998)
74. L. Wong et al., Linking Matlab and Gams: A Supplement (University of Victoria, Department
of Economics, Victoria, BC, 2009)
75. M.C. Ferris, R. Jain, S. Dirkse, Gdxmrw: Interfacing GAMS and matlab (2011). http://www.
gams.com/dd/docs/tools/gdxmrw.pdf
76. M.R. Bussieck, M.C. Ferris, A. Meeraus, Grid-enabled optimization with GAMS. INFORMS
J. Comput. 21(3), 349–362 (2009)
77. S.S. Nielson, A. Consiglio, Practical Financial Optimization: A Library of GAMS Models
(Wiley, New York, 2010)
Chapter 2
Simple Examples in GAMS
The main concept that is developed in this chapter is explaining some optimization
categories that can be modeled in GAMS. These models include linear programming
(LP), mixed integer programming (MIP), nonlinear programming (NLP), quadratic
programming (QCP), mixed integer non-linear programming (MINLP), and multi-
objective optimization problems.
Understanding the materials presented and discussed in this chapter does not
require any background in power system studies. This makes it suitable for anybody
who might be interested to start optimization modeling in GAMS.
2.1 Different Types of Optimization Models
The general form of an optimization problem is as follows:
min f .X; I/ (2.1a)

X
G.X; I/ 0 (2.1b)
H.X; I/ D 0 (2.1c)
where f is objective function, G and H are set of equality and inequality constraints,
respectively, I is the input data of the optimization problem, and X is the set of
decision variables that should not only satisfy G and H but also optimizes the
f value.

DOI 10.1007/978-3-319-62350-4_2
34 2 Simple Examples in GAMS
2.1.1 Linear Programming (LP)
The linear programming problems are those that f , G, H are all linear in (2.1).
2.1.1.1 LP Example
A simple linear programming example is as follows:
min OF Dx1 C 3x2 C 3x3 (2.2a)

X
x1 C 2x2 3 (2.2b)
x3 C x2 5 (2.2c)
x1 C x3 D 4 (2.2d)
GCode 2.1 LP Example (2.2)
v a r i a b l e s x1 , x2 , x3 , o f ;
Equations
eq1
eq2
eq3
eq4 ;
eq1 . . x1+2x2 =g = 3 ;
eq2 . . x3+x2 =g = 5 ;
eq3 . . x1+x3 =e = 4 ;
eq4 . . x1+3x2 +3x3=e=OF ;
model LP1 / a l l / ;
S o l v e LP1 US LP min o f ;
d i s p l a y x1 . l , x2 . l , x3 . l , o f . l ;
3 2
2 3
x1 0:333
6 x2 7 6 1:333 7
The optimal solution is 6 7 6 7
4 x3 5 D 4 3:667 5. Clicking on the model statistic tap
OF 15:333
shows that this model has four blocks of equations (four single equations). It has
also four variables (x1;2;3 ;OF). The solution report would be as follows:
SOLVESUMMARY
MODEL LP1 OBJECTIVE of
TYPE LP DIRECTION MINIMIZE
SOLVER CPLEX FROM LINE 12
**** SOLVER STATUS 1 Normal Completion
**** MODEL STATUS 1 Optimal
(continued)
2.1 Different Types of Optimization Models 35
**** OBJECTIVE VALUE 15.3333

RESOURCE USAGE, LIMIT 0.016 1000.000
ITERATION COUNT, LIMIT 2 2000000000
It means that the solver has successfully solved the model and the solution is
globally optimal. The solver used for solving the model is CPLEX [1]. It states that
the value of objective function is 15.333. It also gives the user some info regarding
the computational burden needed for solving the problem. RESOURCE USAGE is
indicating how much time was needed to solve the model in seconds (0.016 s) and
what was the maximum time allowed to do so (1000 s). The number of iterations
needed for finding the optimal solution is two in this case. The default value for this
limit is 2,000,000,000.
Clicking on the SolEQU tab would show the following info regarding the model:

—- EQU eq1 3 3 +INF 0.333
—- EQU eq2 5 5 +INF 2.333
—- EQU eq3 4 4 4 0.667
—- EQU eq4 0 0 0 -1
The lower limits of eq1,eq2 have some finite values (3,5) but their upper limits
are +1. This means that these two equations are of type. Equations eq3,eq4
have equal values for lower and upper limits. This means that these equations are
of equality type. The interesting part of the analysis is given in marginal column
(the last column). As it can be seen, the level values of eq1,eq2,eq3 are equal to
their lower limits. This has a certain meaning that these constraints are binding
constraints. This means that if the lower limits are changed then the objective
function value would change. The marginal values actually show the sensitivity
coefficients of objective function to these equations. Let’s check them in more
OF
detail. The marginal value of eq1 is 0.333. This means that RHS eq1
D 0:333.
The right-hand side of eq1 is 3 so if it is increased to 3.2 then RHS eq1 D
3:2 3 D 0:2. The marginal value indicates that the new objective function would
be 15:333 C 0:333 0:2 D 15:3996. If the GAMS model is solved using the new
RHS value of eq1 (3.2) then OF would be 15.4. The obtained value is close but not
exactly what we were expecting but why? This is because the marginal values are
accurate for very small change in RHS values of the equations. If the variations are
small enough then the approximation would be accurate enough.
The marginal values constitute the values of dual variables. The decision maker
can understand which constraint is binding (has nonzero marginal value) and also
shows the most influential constraint on objective function (the biggest marginal
value).
Clicking on the SolVAR tab would show the following info regarding the model:

—- VAR x1 -INF 0.333 +INF 0
—- VAR x2 -INF 1.333 +INF 0
—- VAR x3 -INF 3.667 +INF 0
—- VAR OF -INF 15.333 +INF 0
The variable attributes are given in the table above. The lower and upper limits
of all variables are 1 and C1, respectively. The marginal values are zero (this
is because no bound is defined for variables). Suppose that we define a lower limit
for x2 which is 2 x2 . Let’s see the impact on marginal values of equations and
variables: The code would be as follows:
Variables x1,x2,x3,of;
Equations eq1,eq2,eq3,eq4;
eq1 .. x1+2*x2 =g=3;
eq2 .. x3+x2 =g=5;
eq3 .. x1+x3 =e=4;
eq4 .. x1+3*x2 +3*x3=e=OF;
Model LP1 /all/;
x2.lo=2;
Solve LP1 US LP min of;
display x1.l,x2.l,x3.l,of.l;
Clicking on the SolEQU tab would show the following info regarding the model:

—- EQU eq1 3 5 +INF 0
—- EQU eq2 5 5 +INF 2
—- EQU eq3 4 4 4 1
—- EQU eq4 0 0 0 -1
This table shows that eq1 is no longer the binding equation. This means that
small change of RHS (which is 3 here) won’t change the objective function. The
marginal value of this equation is zero. The eq2 and eq3 are binding equations (they
have nonzero marginal values and also their level is equal to their lower value).
Clicking on the SolVAR tab would show the following info regarding the model:

—- VAR x1 -INF 1 +INF 0
—- VAR x2 2 2 +INF 1
—- VAR x3 -INF 3 +INF 0
—- VAR OF -INF 16 +INF 0
The marginal values of all variables are zero except x2 which is 1. This means
that setting a lower limit for x2 caused this situation. Originally, the optimal value of
x2 was 1.333 and now the lower limit (which is set to be 2) stops it from reaching its
optimal value. Any possible decrease in x2 can help reducing the overall objective
function.
GCode 2.2 Finding the boundaries of a variable example (2.3)
V a r i a b l e s x1 , x2 , x3 , o f ;
Equations
eq1 , eq2 , eq3 , eq4 ;
eq1 . . x1+2x2 = l = 3 ;
eq2 . . x3+x2 = l = 2 ;
eq3 . . x1+x2+x3 =e = 4 ;
eq4 . . x1+2x2 3x3=e=OF ;
Model LP1 / a l l / ;
x1 . l o = 0 ; x1 . up = 5 ; x2 . l o = 0 ; x2 . up = 3 ; x3 . l o = 0 ; x3 . up = 2 ;
S o l v e LP1 US LP max o f ;
S o l v e LP1 US LP min o f ;
2.1.1.2 Boundary Determination Example
Sometimes it is needed to find the maximum and minimum of the objective function
for a given model. This means that the problem should be solved two times. The
example given in (2.3) is describing such a situation.
min = max OF D x1 C 2x2 3x3 (2.3a)

X
x1 C 2x2 3 (2.3b)
x3 C x2 2 (2.3c)
x1 C x2 C x3 D 4 (2.3d)
0 x1 5 (2.3e)
0 x2 3 (2.3f)
0 x3 2 (2.3g)
Please pay special attention to (2.3e)–(2.3g). These three constraints can be easily
treated in GAMS using :lo and :up statements. In order to reduce the number of
equations in the model it should be avoided defining them as six extra equations.
The Gcode 2.2 for solving (2.3) is provided as follows:
2 3 2 3
x1 3
6 x2 7 607
The problem is solved and the solutions are obtained as 6 7 6 7
4 x3 5 D 415 and
OF 0 max
2 3 2 3
x1 2
6 x2 7 607
6 7 D 6 7 . The application of this model is in interval optimization [2],
4 x3 5 425
OF 4 min
fuzzy optimization [3], and DC power flow (which will be discussed in Chap. 6)
2.1.2 Mixed Integer Programming (MIP)
In mixed integer programming (MIP) problems, the decision maker is faced with
constraints and objective function that are linear but there exist some integer/binary
variables.
2.1.2.1 MIP Example
A MIP example is given for clarification as follows:
GCode 2.3 MIP example (2.4)
Variables x , of ;
Binary v a r i a b l e y ;
E q u a t i o n s eq1 , eq2 , eq3 ;
eq1 . . 3x+2y =g = 1 ;
eq2 . . 8x+10y = l = 1 0 ;
eq3 . . x+y=e=OF ;
Model MIP1 / a l l / ;
x . up = 0 . 3 ;
S o l v e MIP1 US MIP max o f ;
d i s p l a y y . l , x . l , of . l ;
max OF D x C y (2.4a)
x;y
3x C 2y 1 (2.4b)
8x C 10y 10 (2.4c)
y 2 f0; 1g ; 0:3 x (2.4d)
y is a binary variable and x is a real number. The GAMS code for solving (2.4) is
provided in GCode 2.3:
2 3
x
By running the GAMS code the optimal solution is found as follows: 4 y 5 D
OF
2 3
0:3
415
1:3 max
2.1.2.2 N-Queen Example
The N-queen problem is a classic MIP problem [4]. In this problem, it is tried to
maximize the number of queens that can sit on a chessboard without attacking
each other. The procedure is simple as follows: First of all, it is needed to define
a variable xij which is a binary variable (0/1) and states whether the queen should sit
(1) on block ij (row i, column j) or not (0). Additionally, if the queen is on block ij
then no other queen can sit on row i or column j or the diagonal that contains cell ij.
This is mathematically stated as follows:
X
max OF D xij (2.5a)
xij
i;j
X
xij 1 8j (2.5b)
i
X
xij 1 8i (2.5c)
j
X ˇ ˇ
ˇi rˇ
xc;r ˇ
1 8i; j 2 ˇ ˇD1 (2.5d)
c;r
j cˇ
If a queen is on a cell then no other queen can exist on the same column (2.5b) or the
same row (2.5c) or the same diagonal (2.5d). The GAMS code for solving N-queen
problem (2.5) is given in GCode 2.4:
The N-queen problem is solved two times in Gcode 2.4. Two models are defined
in this code namely MIP2a and MIP2b. The Queen placement is solved on a 4 4
board, Fig. 2.1a shows the wrong placement. This solution is obtained from MIP2a
and considers eq1;2;3 . This is because we are not considering diagonal movement of
queen. In order to overcome this shortcoming two additional constraints eq4;5 are
considered in Model MIP2b. The correct solution is depicted in Fig. 2.1b.
It should be noted that the solution for the defined problem is optimal but not
unique. This means that other configurations might be obtained that can satisfy the
defined constraints. The developed GCode 2.4 is general and can be used for solving
the problem for various sizes of the board. The optimal solution for queen placement
on a (a) 8 8 chessboard and (b) 16 16 board is shown in Fig. 2.2.
GCode 2.4 N-queen example (2.5)
Sets i /14/ , j /14/;

a l i a s ( i , row ) ;
a l i a s ( j , col ) ;
v a r i a b l e of ;
binary variable x( i , j ) ;
E q u a t i o n s eq1 , eq2 , eq3 , eq4 , eq5 ;
eq1 ( j ) . . sum ( i , x ( i , j ) ) = l = 1 ;
eq2 ( i ) . . sum ( j , x ( i , j ) ) = l = 1 ;
eq3 . . sum ( ( i , j ) , x ( i , j ) ) =e=OF ;
eq4 ( i , j ) . . sum ( ( row , c o l ) $ ( ( o r d ( row )o r d ( i ) ) = ( o r d ( c o l )o r d ( j ) ) ) ,
x ( row , c o l ) ) = l = 1 ;
eq5 ( i , j ) . . sum ( ( row , c o l ) $ ( ( o r d ( row )o r d ( i ) ) =( o r d ( c o l )o r d ( j ) ) )
, x ( row , c o l ) ) = l = 1 ;
Model MIP2a / eq1 , eq2 , eq3 / ;
Model MIP2b / a l l / ;
S o l v e MIP2a US MIP max o f ;
d i s p l a y x . l , of . l ;
S o l v e MIP2b US MIP max o f ;
d i s p l a y x . l , of . l ;
1 2 3 4 1 2 3 4
1 1
2 2
3 3
4 4
(a) (b)
Fig. 2.1 Queen placement on a 4 4 board: (a) wrong placement, (b) optimal placement
Fig. 2.2 Queen placement on a (a) 8 8 chessboard, (b) 16 16 board
2.1.2.3 Emergency Center Allocation
Consider six cities (1–6) which are located at different distances to each other. Each
city should have access to an emergency center within a short period of time. The
time required for moving from one city to another one is given in the following table
in minutes.
1 2 3 4 5 6
1 0 30 16 22 24 29
2 0 54 32 43 24
3 0 44 50 28
4 0 14 43
5 0 12
6 0
It should be noted that the values of this table are symmetrical. For example,
distance from city 1 to city 2 is 30 min. It means that the distance from city 2 to city
1 is also 30 min. The critical time for reaching to the emergency center is assumed
to be 20 min. The question is what is the minimum number of cities that should host
emergency center? Which cities should be chosen?
By observing the first row of distance matrix, it is understood that if city 1 or
city 6 have the emergency center then city 1 meets the access requirement. The
same concept applies for city 2. City 2 should definitely have an emergency center
because no other city is located within 20 min distance of this city. If the allocation
decision is defined as a binary variable xi then the following inequality should be
satisfied for city 1:
x1 C x6 1 (2.6)
The following inequality should be satisfied for city 2:
x2 1 (2.7)
The overall constraints are as follows:
x1 C x6 1 (2.8)
x2 1 (2.9)
x3 C x5 1 (2.10)
x4 C x5 1 (2.11)
x3 C x4 C x5 C x6 1 (2.12)
x1 C x5 C x6 1 (2.13)
Two different approaches will be presented here for modeling this problem.
Scalar equations:
In this approach, we initially analyzed the distance data and understood what kind
of relations should be enforced for different variables. The GCode 2.5 is describing
how to do this.
GCode 2.5 Scalar equations for emergency centre allocation
b i n a r y v a r i a b l e x1 , x2 , x3 , x4 , x5 , x6 ;
v a r i a b l e OF ;
equations
eq1 , eq2 , eq3 , eq4 , eq5 , eq6 , eq7 ;
eq1 . . x1+x6 =g = 1 ;
eq2 . . x2 =g = 1 ;
eq3 . . x3+x5 =g = 1 ;
eq4 . . x4+x5 =g = 1 ;
eq5 . . x3+x4+x5+x6 =g = 1 ;
eq6 . . x1+x5+x6 =g = 1 ;
eq7 . . x1+x2+x3+x4+x5+x6 =e=OF ;
Model e m e r g e n c y / a l l / ;
S o l v e e m e r g e n c y u s mip min o f ;
The optimal answer is OF = 3 (three cities should host emergency centers). The
candidate cities are x1 ; x2 ; x5 . The problem with this kind of modeling is that it needs
pre-processing of the raw data and also in case, the number of cities are changed then
it is need to extensively modify the code. A much more efficient way of coding this
problem is using the extended equations.
Indexed equations:
GCode 2.6 Indexed equations for emergency centre allocation
set c i t y /16/ ;
a l i a s ( c i t y , town ) ;
binary variable x( city ) ;
v a r i a b l e OF ;
t a b l e d a t a ( c i t y , town )
1 2 3 4 5 6
1 0 30 46 22 24 19
2 0 54 32 43 24
3 0 44 16 28
4 0 14 43
5 0 12
6 0;
d a t a ( c i t y , town ) $ d a t a ( town , c i t y ) = d a t a ( town , c i t y ) ;
scalar criticaltime /20/;
Equations
eq1 , eq2 ;
eq1 ( c i t y ) . . sum ( town$ ( d a t a ( c i t y , town )< c r i t i c a l t i m e ) , x ( town ) ) =g
=1;
eq2 . . OF=e=sum ( c i t y , x ( c i t y ) ) ;
Model e m e r g e n c y / a l l / ;
S o l v e e m e r g e n c y u s mip min o f ;
The developed code will provide the same answer as before but it has the following
features:
• It is not needed to manually write one equation for each constraint.
• It works for any number of cities.
• The distance data is fed to the model using a table. This will be useful for cases
that the input data might change.
• Debugging and tracing the code are much easier for the users.
• The code does not change if the critical access time is updated.
The following line of the code is to make the data matrix symmetrical.
data(city,town)$data(town,city)=data(town,city);
The following line describes the condition for accessing the emergency center
(for each city). The equation eq1 is defined over the set “city.”
eq1(city) .. sum(town$(data(city,town)<criticaltime), x(town)) =g=1;

The optimal solution is:
—- VAR x
1 . . 1.000 1.000
2 . 1.000 1.000 1.000
3 . . 1.000 1.000
4 . . 1.000 1.000
5 . 1.000 1.000 1.000
6 . 1.000 1.000 1.000
2.1.3 Nonlinear Programming (NLP)
In nonlinear programming problems, at least one of f ; G; H in (2.1) is nonlinear.
2.1.3.1 NLP Example (2.14)
max OF D x1 x4 .x1 C x2 C x3 / C x2 (2.14a)

xi
x1 x2 x3 x4 20 (2.14b)
x12 C x22 C x32 C x42 D 30 (2.14c)
1 x1 ; x2 ; x3 ; x4 3 (2.14d)
The GAMS code for solving example (2.14) is given in GCode 2.7:
GCode 2.7 Example (2.14)
v a r i a b l e of , x1 , x2 , x3 , x4 ;
equations
eq1 , eq2 , eq3 ;
eq1 . . x1x4 ( x1+x2+x3 ) + x2=e=OF ;
eq2 . . x1x2x3x4 =g = 2 0 ;
eq3 . . x1x1+x2x2+x3x3+x4x4=e = 3 0 ;
x1 . l o = 1 ;
x1 . up = 3 ;
x2 . l o = 1 ;
x2 . up = 3 ;
x3 . l o = 1 ;
x3 . up = 3 ;
x4 . l o = 1 ;
x4 . up = 3 ;
Model NLP1 / a l l / ;
S o l v e NLP1 US NLP max o f ;
The solution for example (2.14) obtained by GCode 2.7 is as follows:
Variables Lower Level Upper Marginal

of 1 73.605 C1 0
x1 1.000 3.000 3.000 21.025
x2 1.000 2.575 3.000 0
x3 1.000 2.317 3.000 0
x4 1.000 3.000 3.000 12.025
It is worth noting that providing a starting point for the variables can help the
GAMS in finding a better solution. Generally speaking, finding the optimal solution
of the model is not guaranteed in NLP problems. The initial values for variables are
set by “X.l=initial value” command before solve statement.
2.1.3.2 Circle Placement Example (2.15)
Suppose there are n circles with known radius values (Ri ). The question is: what
is the minimum surface of the table that these circles can be placed on it without
any overlapping. The concept of optimal circle placement on a given table is shown
in Fig. 2.3. The decision variables of this problem are table dimensions fw; hg and
center locations fxi ; yi g . The objective function is defined as the surface of the table
(OF D w h). The constraints are described as:
• Each circle should be completely on the table. This requires: Ri xi w Ri
and similarly Ri yi h Ri .
• For every two circles, no overlap should happen. This means that: .xi xj /2 C
.yi yj /2 .Ri C Rj /2
The circle placement problem is formulated as:
min OF D hw (2.15a)
xi ;yi ;w;h
Ri xi w Ri (2.15b)
Ri yi h Ri (2.15c)
.xi xj /2 C .yi yj /2 .Ri C Rj /2 (2.15d)
It is assumed that Ri are known in advance and they are treated as input data. It is
also assumed that there are six circles available. The GAMS code for solving circle
placement example (2.15) is as GCode 2.8:
GCode 2.8 Circle placement example (2.15)
set i /16/; alias ( i , j ) ;

P o s i t i v e v a r i a b l e s x ( i ) , y ( i ) ,w, h ; v a r i a b l e o f ;
parameter radius ( i )
/1 2
2 1.2
3 1.8
4 0.9
5 3.2
6 0.7/;
E q u a t i o n s eq1 , eq2 , eq3 , eq4 ;
eq1 . . wh=e=OF ;
eq2 ( i ) . . x ( i ) = l =wr a d i u s ( i ) ;
eq3 ( i ) . . y ( i ) = l =hr a d i u s ( i ) ;
eq4 ( i , j ) $ ( o r d ( i )<>o r d ( j ) ) . . power ( y ( j )y ( i ) , 2 ) + power ( x ( j )x ( i ) , 2 ) =
g =( r a d i u s ( i ) + r a d i u s ( j ) , 2 ) ;
x . lo ( i )= radius ( i ) ; y . lo ( i )= radius ( i ) ;
Model NLP1 / a l l / ;
S o l v e NLP1 US NLP min o f ;
parameter report ( i ,) ;
r e p o r t ( i , ’X ’ ) =x . l ( i ) ; r e p o r t ( i , ’ y ’ ) =y . l ( i ) ;
r e p o r t ( i , ’R ’ ) = r a d i u s ( i ) ;
d i s p l a y r e p o r t , of . l ,w. l , h . l ;
As it is observable in GCode 2.8, the radius of circles are given as parameters. The
constraint given in (2.15b) is actually two constraints Ri xi and xi w Ri . The
first one should be treated using :lo statement and the second one should be modeled
using equations. This is because it involves two variables xi ; w which should be valid
for every member of set i. This is why the equation eq2 is defined over set i.
The optimal solution of circle placement on a table is given in Fig. 2.4.
Fig. 2.3 Optimal circle placement on a given surface

14 w
13 PARAMETER report
12
X y R
11 . R=3.2
10 1 2.204 5.571 2.000
2 1.200 2.533 1.200
9 3 4.600 2.622 1.800
y axis
8 4 2.520 0.900 0.900

5 3.200 10.675 3.200
7
h 6 0.700 0.700 0.700
6
. R=2
5
VARIABLE of.L = 88.801
4 VARIABLE w.L = 6.400
3 VARIABLE h.L = 13.875
R=1.2 . R=1.8
2
1 . R=0.7
R=0.9
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
x axis
Fig. 2.4 The optimal solution of circle placement on a surface
2.1.3.3 Maximizing the Area of a Right Triangle with Constant

Circumference
Suppose that we have some limited meters of wire fences (C) and are requested to
enclose an area (right triangle shape). Determine the dimensions of such a triangle.
Considering Fig. 2.5, the following optimization should be solved:
HB
max OF D (2.16a)
H;B 2
p
HCBC H 2 C B2 D C (2.16b)
where C is the length of the available wire fence. H and B are the height and base of
the triangle, respectively. The GCode 2.9 provides the solution for maximizing the
area of a triangle with constant circumference.
GCode 2.9 Maximizing the area of a triangle
Positive variables h , b;
V a r i a b l e OF ;
S c a l a r C / 1 5 / ; h . up=C ; b . up=C ;
E q u a t i o n s eq1 , eq2 ;
eq1 . . h+b+ s q r t ( hh+bb ) = e=C ;
eq2 . . OF=e =0.5 hb ;
Model t r i a n g l e / a l l / ;
S o l v e t r i a n g l e u s n l p min o f ;
Fig. 2.5 Maximizing the

area of a triangle with
constant circumference
B
H
2.1.4 Quadratic Constrained Programming (QCP)
The quadratic programming is a special case of NLP problems. In QCP problems,

at least one of f ; G; H in (2.1) is nonlinear but nonlinearity is of quadratic form.
2.1.4.1 QCP Example (2.17)
Consider the following QCP optimization problem:
max OF D 3x12 10x1 C x22 3x2 (2.17a)

xi
x1 C x22 2:5 (2.17b)

2x1 C x2 D 1 (2.17c)
The GAMS code for solving example (2.17) is as GCode 2.10:
GCode 2.10 QCP Example (2.17)
v a r i a b l e of , x1 , x2 ;
equations
eq1 , eq2 , eq3 ;
eq1 . . 3x1x1 10x1 +x2x23x2=e=OF ;
eq2 . . x1+x2x2 =g = 2 . 5 ;
eq3 . . 2x1+x2=e = 1 ;
Model QCP1 / a l l / ;
S o l v e QCP1 US QCP max o f ;

of 1 9.550 C1 0
x1 1 1.093 C1 0
x2 1 1.186 C1 0
2.1.4.2 QCP Example (2.18)
Another simple QCP problem is described as follows:
min OF Dx12 10x1 C x22 3x2 (2.18a)

xi
x12 C x2 5 (2.18b)
2x1 x2 1 (2.18c)
V a r i a b l e of , x1 , x2 ;
eq1 . . x1x1 10x1 +x2x23x2=e=OF ;
eq2 . . x1x1+x2 = l = 5 ;
eq3 . . 2x1x2=g = 1 ;
Model QCP1 / a l l / ;
S o l v e QCP1 US QCP min o f ;
The solution for example (2.18) (obtained by GCode 2.11) is as follows:

of 1 18.054 C1 0
x1 1 2.054 C1 0
x2 1 0.783 C1 0
2.1.5 Mixed Integer Nonlinear Programming (MINLP)
In MINLP problems, at least one of f ; G; H in (2.1) is nonlinear and integer variables

are involved.
Fig. 2.6 Optimal transportation problem
2.1.5.1 Minimum Transportation Cost Example (2.19)
The transportation problem is a classic example which has been modified for this
example. There are some suppliers (node i) and some demands (node j) which
should be supplied. The problem is to determine how much each supplier should
produce and how to transfer them to the demand points. The transportation costs
should be minimized.
The cost of transportation is assumed to be related to the square of quantity
transported from node i to node j . The transportation costs are proportional to
the square of product transported from i to j and the length of the rout. The
object is minimizing the total transportation costs and supplying all demand in
different nodes. The optimal transportation problem is depicted in Fig. 2.6. The cost
coefficient and maximum flow of each rout (Cij ), demand (Dj ), and capacity of each
producer are known.
sets
i / s 1s 3 /
j / D1D4 / ;
t a b l e C( i , j )
d1 d2 d3 d4
s1 0.0755 0.0655 0.0498 0.0585
s2 0.0276 0.0163 0.096 0.0224
s3 0.068 0.0119 0.034 0.0751;
table data ( i ,)
’ Pmin ’ ’ Pmax ’
s 1 100 450
s 2 50 350
s 3 30 500;
p a r a m e t e r demand ( j )
/ d1 217
d2 150
d3 145
d4 2 4 4 / ;
v a r i a b l e of , x ( i , j ) , P ( i ) ;
b i n a r y v a r i a b l e U( i ) ;
equations
eq1 , eq2 , eq3 , eq4 , eq5 ;
eq1 . . OF=e=sum ( ( i , j ) , C ( i , j ) x ( i , j ) x ( i , j ) ) ;
eq2 ( i ) . . P ( i ) = l = d a t a ( i , ’ Pmax ’ )U( i ) ;
eq3 ( i ) . . P ( i ) = g= d a t a ( i , ’ Pmin ’ )U( i ) ;
eq4 ( j ) . . sum ( i , x ( i , j ) ) = g=demand ( j ) ;
eq5 ( i ) . . sum ( j , x ( i , j ) ) = e=P ( i ) ;
P . lo ( i )=0;
P . up ( i ) = d a t a ( i , ’ Pmax ’ ) ;
x . lo ( i , j )=0;
x . up ( i , j ) = 1 0 0 ;
Model m i n l p 1 / a l l / ;
S o l v e m i n l p 1 US m i n l p min o f ;
The optimal transportation problem is formulated in (2.19).

X
min OF D Cij xij2 (2.19a)
xij
ij
(
Pmin
i Pi Pmax
i ; if unit i is on;
(2.19b)
Pi D 0 if unit i is off
X
xij Dj (2.19c)
i
X
xij D Pi (2.19d)
j
0 xij xijmax (2.19e)
The road flow limit is assumed to be xijmax D 100.
Pmin
i Pmax
i j
i 100 450 0.0755 0.0655 0.0498 0.0585
50 350 0.0276 0.0163 0.096 0.0224
30 500 0.068 0.0119 0.034 0.0751
Dj 217 150 145 244
The GAMS code for solving example (2.19) is as GCode 2.12:

Pi Xij
199.245 55.443 14.255 48.601 80.946
282.494 100 57.282 25.212 100
274.261 61.557 78.463 71.187 63.054
2.1.5.2 Benefit Maximization Transportation Example (2.20)
Reconsider the optimal transportation problem formulated in (2.19). This problem

is a minimum cost problem and the goal is service provision to the consumers. Now
suppose that the objective is benefit maximization. The machine (i) will produce
equal to Pi and sell it to different consumer j. The maximum value of purchase that
a consumer may procure is Dj . The revenue that is obtained from selling product is
k $/unit. Now the benefit maximization is modeled as follows:
X X
max OF D Pi k Cij xij2 (2.20a)
xij
i ij
(
Pmin
i Pi Pmax
i ; if unit i is on ;
(2.20b)
Pi D 0 if unit i is off
X
xij Dj (2.20c)
i
X
xij D Pi (2.20d)
j
0 xij xijmax (2.20e)
The GAMS code for solving the example (2.20) is given in GCode 2.13.
Pi Xij
i 0 0 0 0 0
137.377 32.609 55.215 9.375 40.179
0 0 0 0 0
2.2 Random Numbers in GAMS 53
2.2 Random Numbers in GAMS
Random number generation in GAMS is a simple task. There are different built-in
probability density function that can be used for generating random numbers. Some
popular random number generators are listed as follows:
• Beta function
.˛ C ˇ/
.1 x/ˇ1 .x/˛1 (2.21)
.˛/ .ˇ/
The format of this function in GAMS is beta(˛,ˇ)

• Uniform function (continuous)
(
1
.ba/
; for a x b
P.x/ D (2.22)
0; otherwise
The format of this function in GAMS is uniform(a; b)

• Uniform function (discrete)
(
1
N
; for a x b
P.x/ D (2.23)
0; otherwise
sets
i / s 1s 3 /
j / D1D4 / ;
scalar k /1.8/;
t a b l e C( i , j )
d1 d2 d3 d4
s1 0.0755 0.0655 0.0498 0.0585
s2 0.0276 0.0163 0.096 0.0224
s3 0.068 0.0119 0.034 0.0751;
table data ( i ,)
’ Pmin ’ ’ Pmax ’
s 1 100 450
s 2 50 350
s 3 30 500;
p a r a m e t e r demand ( j )
/ d1 217
d2 150
d3 145
d4 2 4 4 / ;
v a r i a b l e of , x ( i , j ) , P ( i ) ;
b i n a r y v a r i a b l e U( i ) ;
equations
eq1 , eq2 , eq3 , eq4 , eq5 ;
eq1 . . OF=e=sum ( i , kP ( i )) sum ( ( i , j ) , C ( i , j ) x ( i , j ) x ( i , j ) ) ;
eq2 ( i ) . . P ( i ) = l = d a t a ( i , ’ Pmax ’ )U( i ) ;
eq3 ( i ) . . P ( i ) = g= d a t a ( i , ’ Pmin ’ )U( i ) ;
eq4 ( j ) . . sum ( i , x ( i , j ) ) = l =demand ( j ) ;
eq5 ( i ) . . sum ( j , x ( i , j ) ) = e=P ( i ) ;
P . lo ( i )=0;
P . up ( i ) = d a t a ( i , ’ Pmax ’ ) ;
x . lo ( i , j )=0;
x . up ( i , j ) = 1 0 0 ;
Model m i n l p 2 / a l l / ;
S o l v e m i n l p 2 US m i n l p max o f ;
where N is the total integer numbers in [a; b] interval. The format of this
function in GAMS is uniformint(a; b)
2.2.1 Estimating the Number
Calculating the number is investigated and coded in GAMS. Consider a circle

inscribed in a square as shown in Fig. 2.7. If a point is randomly dropped on this
Circle area
square then the probability of sitting on the circle would be Square area
. In order to
calculate the number a simple experience is done. The point is dropped on the
square area for N times. The number of events that the point is on circle area would
be n. The following equation would be valid if N is a big number. The GAMS code
for solving this problem is described in GCode 2.14.
Circle area n
D (2.24)
Square area N
R2 n
2
D (2.25)
4R N
GCode 2.14 number estimation
s c a l a r low / 0 / , High / 1 / , p i s t i m a t e ;
s e t c o u n t e r / c1c200 / ;
parameter report ( counter , ) ;
r e p o r t ( c o u n t e r , ’ x ’ ) = u n i f o r m (LOW, HIGH ) ;
r e p o r t ( c o u n t e r , ’ y ’ ) = u n i f o r m (LOW, HIGH ) ;
p i s t i m a t e =4sum ( c o u n t e r $ ( power ( r e p o r t ( c o u n t e r , ’ x ’ )
0 . 5 , 2 ) + power ( r e p o r t ( c o u n t e r , ’ y ’ ) 0 . 5 , 2 ) < = 0 . 2 5 ) , 1 ) / c a r d ( c o u n t e r ) ;
display report , pistimate ;
As it can be seen in GCode 2.14, there is no solve statement needed. This is

because no variable is defined in the model and no optimization is going to take
place.
Fig. 2.7 Estimating the number
2.2.2 Integration Calculation
The random numbers can be used for calculating the integration problems. A simple
example is given as follows:
Z 1
ZD 1 xsin.20x/dx (2.26)
0
In order to calculate the area under the graph as it is shown in Fig. 2.8, the following
steps are followed:
• Set Counter = 1; n = 1;
• Generate a pair of random numbers .x; y/ where 0 X 1 and 0 Y 2.
• Check if y f .X/
• Set Counter = Counter + 1;
This procedure is repeated for max number of counter (N). For large values of N,
the Z can be calculated using the following relation:
Z n
D (2.27)
21 N
The following command should be added to the code (before calling the uniform
function) in order to have a set of new random numbers every time the code is run:
1.8
1.6
1.4
Y=1+xCos(20x)
1.2
0.8
0.6
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1
X
Fig. 2.8 Numerical integration using random numbers
GCode 2.15 Integration calculation using random numbers
Scalar Zstimate ;
S e t c o u n t e r / c1c200000 / ;
parameter report ( counter , ) ;
r e p o r t ( c o u n t e r , ’ x ’ )= uniform ( 0 , 1 ) ;
r e p o r t ( c o u n t e r , ’ y ’ )= uniform ( 0 , 2 ) ;
Z s t i m a t e =2sum ( c o u n t e r $ ( r e p o r t ( c o u n t e r , ’ y ’ )<1+
r e p o r t ( c o u n t e r , ’ x ’ ) s i n ( r e p o r t ( c o u n t e r , ’ x ’ ) 2 0 ) ) , 1 ) / c a r d ( c o u n t e r ) ;
Sisplay report , Zstimate ;
execseed = 1+gmillisec(jnow);
Further info on how to use random numbers in GAMS can be found in [5].
2.2.3 LP Problems with Uncertain Coefficients
Consider the following LP problem:
Z D max 750x1 C 1000x2 (2.28)

x1 ;x2
x1 C x2 a (2.29)
Fig. 2.9 The probability density function for each uncertain parameter
Fig. 2.10 Uncertain coefficients in LP problems
x1 C 2x2 b (2.30)
4x1 C 3x2 c (2.31)
x1 0 (2.32)
x2 0 (2.33)
where a, b, and c are uncertain random parameters. The probability density function
for each parameter is given in Fig. 2.9. The question is how to describe the prob-
ability density of Z? If all uncertain parameters (a,b,c) are equal to their expected
values (10,15,25) then the objective function would be 7750. The following steps
are followed to find out the distribution of Z in case of uncertain (a,b,c):
• Set Counter = 1;
• Generate a sample of random parameters (a,b,c) using the specified probability
distribution functions
• Calculate the optimal Z and save the (a,b,c,x1 ,x2 ,Z).
The variation of Z along with the average value of Z are plotted in Fig. 2.10. The
graph shows how the average value of Z converges.
2.3 Multi-Objective Optimization
In Multi-Objective Decision Making (MODM) methods, the decision alternatives

are found considering the constraints of the given problem. In most practical
optimization problems, particularly those applicable in power system, there exists
more than one objective function which should be optimized simultaneously. These
objectives functions might be in conflict, interdependent or independent of each
other, so it is impossible to satisfy them all at once. The main differences between
the multi-objective optimization and traditional single optimization techniques can
be categorized into two groups:
1. Several objective functions are to be optimized at the same time.
2. There exists a set of optimal solutions which are mathematically equally good
solutions (it means any of them cannot be preferred against others and a trade-off
should be made to select one) instead of a unique optimal solution
2.3.1 Weighted Sum Approach
Some methods try to convert the multi-objective optimization problems into single
objective one. However, this approach is not always applicable especially in the
following cases:
• The objectives are not of the same type. For example voltage deviation and cost,
weight and volume, surface and time.
• The weight coefficients in weighted sum approach cannot be easily determined.
Additionally, If decision maker’s preference is changed then the problem should
be solved again.
• The single objective approach provides just one solution to the decision maker.
It cannot show the tradeoff between two objective functions.
• The objective function is of the same type but there are multiple decision makers
with different preferences. Each decision maker is trying to optimize its own
objective function.
2.3.2 Pareto Optimality
The notion of optimality has been redefined in this context and instead of aiming to
find a single solution, it is tried to produce a set of acceptable compromises or trade-
offs from which the decision maker can choose one. The set of all optimal solutions
which are non-dominated by any other solution is known as Pareto-optimal set.
Suppose a minimizing problem with two objectives in conflict the Pareto optimal
fronts are plotted in Fig. 2.11.
For every two solution in each Pareto front (like A, B) none of them is better
than the others considering all objective functions. Here, f1 is better minimized in
solution A compared to B and f2 is better minimized in solution B compared to A.
2.3 Multi-Objective Optimization 59
Fig. 2.11 Classification of a population to k non-dominated fronts
The same concept also applies for solutions in other fronts. For every solution in
Pareto front k (for example D in the second front) there exists at least one solution
in front k 1 (here A in the first front) that dominates it (is better considering
all objective functions). Since solutions A and B belong to the first front, there is
no solution better than them in respect to all objectives. Consider the bi-objective
optimization described in (2.34):
minx f1 ; f2 (2.34)
Constraints
Each solution in Pareto optimal set has two basic characteristics:

1. For every two solutions belonging to the same Pareto front (2.35) holds:
8i9j; njfn .Nxi / > fn .Nxj / (2.35)

xN j ; xN i 2 S
This means that for every solution Xi belonging to Pareto front S, at least one
solution exists as Xj which is better than Xi at least in one objective function
(named n here). In other words, there is no solution in Pareto optimal front which
is the best among all members of this set considering all objectives.
2. For every solution belonging to an upper Pareto front and the ones in the lower
fronts, (2.36) holds:
8k 2 f1 : : : NO g fk .Nx1 / fk .Nx2 / (2.36)

0
9k 2 f1 : : : NO g fk0 .Nx1 / < fk0 .Nx2 / (2.37)
xN 1 2 S; xN 2 2 S
S < S
This means for every solution belonging to an upper Pareto front, there exists
at least one solution in a lower Pareto front which is not worse in any objective
function and is better in at least one objective function. NO is the number of
objective functions.
The classic approach for finding the Pareto optimal set is preference-based method
in which a relative preference vector is used to weight the objectives and change
them into a scalar value. By converting a multi-objective optimization problem into
a single objective one, only one optimum solution can be achieved which is very
sensitive to the given weights. The GAMS structure can solve only one objective
function at once. This means it is needed to solve the multi-objective problem
several times to obtain the Pareto optimal front. For this purpose, consider the bi-
objective problem described in (2.38):
max f1 D4x1 0:5x22 (2.38a)

x1;2
max f2 D x12 C 5x2 (2.38b)

x1;2
2x1 C 3x2 10 (2.38c)

2x1 x2 0 (2.38d)
1 x1 2 (2.38e)
1 x1 3 (2.38f)
GCode 2.16 Pareto optimal front example (2.38)
s e t c o u n t e r / c1c21 / ;
scalar E;
parameter report ( counter ,) ;
v a r i a b l e of1 , of2 , x1 , x2 ;
equations
eq1 , eq2 , eq3 , eq4 ;
eq1 . . o f 1 =e =4x1 0.5 x2x2 ;
eq2 . . o f 2 =e=x1x1+5x2 ;
eq3 . . 2x1+3x2= l = 1 0 ;
eq4 . . 2x1x2=g = 0 ;
x1 . l o = 1 ;
x1 . up = 2 ;
x2 . l o = 1 ;
x2 . up = 3 ;
Model p a r e t o 1 / a l l / ;
parameter ranges () ;
S o l v e p a r e t o 1 US nLP max o f 1 ;
r a n g e s ( ’ OF1max ’ ) = o f 1 . l ;
r a n g e s ( ’ OF2min ’ ) = o f 2 . l ;
r a n g e s ( ’ OF2max ’ ) = o f 2 . l ;
r a n g e s ( ’ OF1min ’ ) = o f 1 . l ;
2.3 Multi-Objective Optimization 61
E= r a n g e s ( ’ OF1min ’ ) ;
loop ( counter ,
E= ( r a n g e s ( ’ OF2max ’ )r a n g e s ( ’ OF2min ’ ) ) ( o r d ( c o u n t e r ) 1) / ( c a r d
( c o u n t e r ) 1)+ r a n g e s ( ’ OF2min ’ ) ;
o f 2 . l o =E ;
r e p o r t ( c o u n t e r , ’OF1 ’ ) =OF1 . l ;
r e p o r t ( c o u n t e r , ’OF2 ’ ) =OF2 . l ;
r e p o r t ( c o u n t e r , ’E ’ ) =E ;
);
Display report ;
In this problem, two objectives should be maximized simultaneously. The

procedure is as follows:
1. Find the maximum of each objective function and save them.
2. Add one of the objective functions to the constraints as follows:
f2 (2.39)
The value will be varied from f2min to f2max and the f1 is maximized.
The solution for example (2.38) obtained by GCode 2.16 is shown in Fig. 2.12.
10
OFmax
2
6
OF2
2
OFmin
2
0 OFmin
1 OFmax
1
-2
2 3 4 5 6 7
OF1
Fig. 2.12 The Pareto optimal front for bi-objective problem

2.4 Applications
Some optimization models used in power system studies are given in this section:
• LP programming: Transmission network estimation [6], short-term hydro
scheduling [7], relay coordination [8], security constrained economic
dispatch [9].
• MIP: Optimal PMU placement [10], unit commitment [11], Phase shifter place-
ment in large-scale systems [12], minimum-losses radial configuration of electri-
cal distribution networks [13].
• QCP: Topology identification in distribution network [14], optimum active and
reactive generation dispatch [15].
• NLP: Voltage stability security margin calculation [16], reactive power planning
[17], OPF [18].
• MINLP: Unit commitment with AC OPF constraints [19], flexible transmission
expansion planning with uncertainties in an electricity market [20], optimal DG
allocation [21].
• Multi-objective optimization: Transmission expansion planning [22], generation
expansion planning [23], distribution network planning [24].
References
1. A. Meeraus, A. Brooke, D. Kendrick, R. Raman, GAMS/Cplex 7.0 User Notes (GAMS

Development Corporation, Washington, DC, 2000)
2. Y. Wang, Q. Xia, C. Kang, Unit commitment with volatile node injections by using interval
optimization. IEEE Trans. Power Syst. 26(3), 1705–1713 (2011)
4. X. Hu, R.C. Eberhart, Y. Shi, Swarm intelligence for permutation optimization: a case study of
n-queens problem, in Swarm Intelligence Symposium, 2003. SIS’03. Proceedings of the 2003
IEEE (IEEE, Piscataway, 2003), pp. 243–246
5. E. Kalvelagen, Some notes on random number generation with GAMS, 2005
6. L.L. Garver, Transmission network estimation using linear programming. IEEE Trans. Power
Apparatus Syst. PAS-89(7), 1688–1697 (1970)
7. G.W. Chang, M. Aganagic, J.G. Waight, J. Medina, T. Burton, S. Reeves, M. Christoforidis,
Experiences with mixed integer linear programming based approaches on short-term hydro
scheduling. IEEE Trans. Power Syst. 16(4), 743–749 (2001)
8. B. Chattopadhyay, M.S. Sachdev, T.S. Sidhu, An on-line relay coordination algorithm for
adaptive protection using linear programming technique. IEEE Trans. Power Delivery 11(1),
165–173 (1996)
9. R.A. Jabr, A.H. Coonick, B.J. Cory, A homogeneous linear programming algorithm for the
security constrained economic dispatch problem. IEEE Trans. Power Syst. 15(3), 930–936
(2000)
11. J. Ostrowski, M.F. Anjos, A. Vannelli, Tight mixed integer linear programming formulations
for the unit commitment problem. IEEE Trans. Power Syst. 27(1), 39–46 (2012)
References 63
12. F.G.M. Lima, F.D. Galiana, I. Kockar, J. Munoz, Phase shifter placement in large-scale systems
via mixed integer linear programming. IEEE Trans. Power Syst. 18(3), 1029–1034 (2003)
13. A. Borghetti, A mixed-integer linear programming approach for the computation of the
minimum-losses radial configuration of electrical distribution networks. IEEE Trans. Power
Syst. 27(3), 1264–1273 (2012)
14. Z. Tian, W. Wu, B. Zhang, A mixed integer quadratic programming model for topology
identification in distribution network. IEEE Trans. Power Syst. 31(1), 823–824 (2016)
15. H. Nicholson, M.J.H. Sterling, Optimum dispatch of active and reactive generation by
quadratic programming. IEEE Trans. Power Apparatus Syst. PAS-92(2), 644–654 (1973)
16. L.A.L. Zarate, C.A. Castro, J.L.M. Ramos, E.R. Ramos, Fast computation of voltage stability
security margins using nonlinear programming techniques. IEEE Trans. Power Syst. 21(1),
19–27 (2006)
17. L.L. Lai, J.T. Ma, Application of evolutionary programming to reactive power planning-
comparison with nonlinear programming approach. IEEE Trans. Power Syst. 12(1), 198–206
(1997)
18. J.A. Momoh, R. Adapa, M.E. El-Hawary, A review of selected optimal power flow literature
to 1993. I. nonlinear and quadratic programming approaches. IEEE Trans. Power Syst. 14(1),
96–104 (1999)
19. A. Castillo, C. Laird, C.A. Silva-Monroy, J.P. Watson, R.P. ONeill, The unit commitment
problem with ac optimal power flow constraints. IEEE Trans. Power Syst. 31(6), 4853–4866
(2016)
20. J.H. Zhao, Z.Y. Dong, P. Lindsay, K.P. Wong, Flexible transmission expansion planning with
uncertainties in an electricity market. IEEE Trans. Power Syst. 24(1), 479–488 (2009)
21. A. Kumar, W. Gao, Optimal distributed generation location using mixed integer non-linear
programming in hybrid electricity markets. IET Gener. Transm. Distrib. 4(2), 281–298 (2010)
22. P. Maghouli, S.H. Hosseini, M.O. Buygi, M. Shahidehpour, A scenario-based multi-objective
model for multi-stage transmission expansion planning. IEEE Trans. Power Syst. 26(1),
470–478 (2011)
23. S. Kannan, S. Baskar, J.D. McCalley, P. Murugan, Application of NSGA-II algorithm to
generation expansion planning. IEEE Trans. Power Syst. 24(1), 454–461 (2009)
24. V. Miranda, J.V. Ranito, L.M. Proenca. Genetic algorithms in optimal multistage distribution
network planning. IEEE Trans. Power Syst. 9(4), 1927–1933 (1994)
Chapter 3
Power Plant Dispatching
This chapter provides the instruction on how to model the economic dispatch
problem of different power plants. Different power plant technologies will be
discussed, such as thermal power, wind turbine, CHP, and hydro power plants in
GAMS.
The idea of dispatching generating units is based on some assumptions as
follows:
• The technical characteristics of the units are known.
• The on/off status of each unit is known.
• It is solved for a single snapshot of the demand.
3.1 Thermal Unit Economic Dispatch
The thermal unit technology transforms the fuel-based source of energy into
electricity. The production costs of thermal unit i are calculated as:
th th 2
Cith .Pth th th th
i / D ai .P /i C bi Pi C ci i 2 ˝th (3.1)
where ath th th
i , bi , and ci are the fuel cost coefficients of the thermal unit i. The total
fuel cost is calculated as follows:
X
TC D Cith .Pth
i / (3.2)
i2˝th
The operating limits are defined as follows:
Pith;min Pth th;max

i Pi i 2 ˝th (3.3)

DOI 10.1007/978-3-319-62350-4_3
66 3 Power Plant Dispatching
Table 3.1 The economic dispatch data for five thermal units example
ath
i bth
i
dith eth
i
i ($=MW2 ) ($=MW) cth
i ($) (kg=MW2 ) (kg=MW) fith .kg/ Pth;min
i (MW) Pth;max
i (MW)
g1 3 20 100 2 5 3 28 206
g2 4.05 18.07 98.87 3.82 4.24 6.09 90 284
g3 4.05 15.55 104.26 5.01 2.15 5.69 68.00 189.00
g4 3.99 19.21 107.21 1.10 3.99 6.20 76.00 266.00
g5 3.88 26.18 95.31 3.55 6.88 5.57 19.00 53.00
th;max=min
where Pi are the maximum/minimum power outputs of thermal unit i. The
overall thermal unit economic dispatch is formulated as follows:
X
min TC D Cith .Pth
i / (3.4a)
Pth
i i2˝th
th th 2
Cith .Pth th th th
i / D ai .Pi / C bi Pi C ci i 2 ˝th (3.4b)
Pith;min Pth
i Pith;max i 2 ˝th (3.4c)
X
Pth
i Le (3.4d)
i2˝th
The economic dispatch data for five units example is given in Table 3.1. This table
has nine columns. The first column is showing the generating unit index. The next
three columns indicate the cost coefficients for these thermal units (ath th th
i ; bi ; ci ).
The next three columns describe the emission coefficients for these thermal units
(dith ; eth th
i ; fi ). The last two columns give the minimum and maximum generating
limits of each unit if it is on.
The GAMS code for solving the economic dispatch example (3.4) is given in
GCode 3.1.
The solution obtained by Code provides information regarding the defined
variables. For example, the generating unit schedules are given in Table 3.2.
As it can be seen in Table 3.2, four attributes of variable P.gen/ are given. The
:lo and :up values are already known and are given as the input data in Table 3.1.
There are two important information regarding the variable P.gen/, namely :l and
:m. The level column shows the optimal values of variables obtained by GAMS. The
marginal value (:m) is zero (EPS or epsilon) for unit g1, g3, and g4. This is because
these generating units are not binding variables (not reached to their limits). On the
other hand, g2 has a marginal value equal to 110.005. This means that if the value
of this variable is increased by very small quantity (P.g2/) then the increase in
objective function will be approximately equal to 110:005P.g2/. This also holds
for generating unit g5. The increase of P.g5/ will approximately decrease the
objective function equal to 199:605P.g5/. This is why g2 is at its min limit and
g5 is at its max limit. The total operating costs would be OF D $1:3146 105 .
3.2 Thermal Unit Environmental Dispatch 67
GCode 3.1 Economic dispatch problem for five units, example (3.4)
s e t Gen / g1g5 / ;
s c a l a r load / 4 0 0 / ;
T a b l e d a t a ( Gen , )
a b c d e f Pmin Pmax
g1 3 20 100 2 5 3 28 206
g2 4 . 0 5 1 8 . 0 7 9 8 . 8 7 3 . 8 2 4.24 6 . 0 9 90 284
g3 4 . 0 5 1 5 . 5 5 1 0 4 . 2 6 5 . 0 1 2.15 5 . 6 9 68 189
g4 3 . 9 9 1 9 . 2 1 1 0 7 . 2 1 1 . 1 3.99 6 . 2 76 266
g5 3 . 8 8 2 6 . 1 8 9 5 . 3 1 3 . 5 5 6.88 5 . 5 7 19 53;
v a r i a b l e s P ( gen ) ,OF ;
equations
eq1 , eq2 ;
eq1 . . OF=e=sum ( gen , d a t a ( gen , ’ a ’ ) P ( gen ) P ( gen ) + d a t a ( gen , ’ b ’ ) P
( gen ) + d a t a ( gen , ’ c ’ ) ) ;
eq2 . . sum ( gen , P ( gen ) ) =g= l o a d ;
P . l o ( gen ) = d a t a ( gen , ’ Pmin ’ ) ;
P . up ( gen ) = d a t a ( gen , ’ Pmax ’ ) ;
model ECD / eq1 , eq2 / ;
s o l v e ECD u s qcp min o f ;
Table 3.2 The ED solution Pth Lower Level Upper Marginal

i
for five units example
g1 28 102.844 206 EPS
g2 90 90 284 110.005
g3 68 76.73 189 –
g4 76 77.425 266 EPS
g5 19 53 53 199.605
Sometimes we need to perform some sensitivity analysis. For example, we might

need to know how the dispatching pattern would change if the load value changes.
The load is defined as a scalar quantity in the developed GCode 3.1. The problem
is feasible for any load value in the interval Œ281; 998. The boundary values of this
interval are obtained by the sum of min and max values of generating units given in
Table 3.1.
The solution for load sensitivity analysis in ED obtained by GCode 3.2 is given
in Table 3.3.
The results of load sensitivity analysis are shown in Fig. 3.1.
3.2 Thermal Unit Environmental Dispatch
The environmental dispatch is referred to the dispatch of electric power plants

considering environmental concerns. The environmental pollution produced by
power plants (thermal) depends on the fuel type they use to produce electricity
GCode 3.2 Example of sensitivity analysis regarding load value in ED

S e t s Gen / g1g5 /
c o u n t e r / c1c11 / ;
P a r a m e t e r r e p o r t ( c o u n t e r , ) , repGen ( c o u n t e r , Gen ) ;
S c a l a r load / 4 0 0 / ;
g1 3 20 100 2 5 3 28 206
g2 4 . 0 5 1 8 . 0 7 9 8 . 8 7 3 . 8 2 4.24 6 . 0 9 90 284
g3 4 . 0 5 1 5 . 5 5 1 0 4 . 2 6 5 . 0 1 2.15 5 . 6 9 68 189
g4 3 . 9 9 1 9 . 2 1 1 0 7 . 2 1 1 . 1 3.99 6 . 2 76 266
g5 3 . 8 8 2 6 . 1 8 9 5 . 3 1 3 . 5 5 6.88 5 . 5 7 19 53;
V a r i a b l e s P ( gen ) ,OF ;
E q u a t i o n s eq1 , eq2 ;
eq1 . . OF=e=sum ( gen , d a t a ( gen , ’ a ’ ) P ( gen ) P ( gen ) + d a t a ( gen , ’ b ’ ) P
( gen ) + d a t a ( gen , ’ c ’ ) ) ;
Model ECD / eq1 , eq2 / ;
loop ( counter ,
l o a d =sum ( gen , d a t a ( gen , ’ Pmin ’ ) )
+ ( ( o r d ( c o u n t e r ) 1) / ( c a r d ( c o u n t e r ) 1) ) sum ( gen , d a t a ( gen , ’ Pmax ’ )
d a t a ( gen , ’ Pmin ’ ) ) ;
S o l v e ECD u s qcp min o f ;
repGen ( c o u n t e r , Gen ) =P . l ( gen ) ;
r e p o r t ( c o u n t e r , ’OF ’ ) = o f . l ;
r e p o r t ( counter , ’ load ’ ) =load ;
);
d i s p l a y repgen , r e p o r t ;
Table 3.3 The ED solution for load sensitivity analysis

Iteration g1 g2 g3 g4 g5 Le (MW) OF ($/h)
c1 28.00 90.00 68.00 76.00 19.00 281.00 84,037.85
c2 67.39 90.00 68.00 76.00 51.31 352.70 105,758.02
c3 112.63 90.00 83.98 84.79 53.00 424.40 147,715.94
c4 136.77 101.55 101.86 102.93 53.00 496.10 203,091.66
c5 158.94 117.97 118.28 119.60 53.00 567.80 268,132.18
c6 181.12 134.40 134.71 136.28 53.00 639.50 342,712.41
c7 203.29 150.82 151.14 152.95 53.00 711.20 426,832.36
c8 206.00 173.71 174.02 176.18 53.00 782.90 522,137.81
c9 206.00 201.85 189.00 204.75 53.00 854.60 631,691.67
c10 206.00 237.44 189.00 240.86 53.00 926.30 760,550.50
c11 206.00 284.00 189.00 266.00 53.00 998.00 911,044.09
300
g1
P(gi) (MW) g2
200 g3
g4
g5
100
0
1 2 3 4 5 6 7 8 9 10 11
Iteration
× 105
10
OF ($/h)
0
300 400 500 600 700 800 900 1000
Load (MW)
Fig. 3.1 Load sensitivity analysis and impact on costs and generation schedules
and also their generation level. The production costs of ith thermal unit are usually
defined as:
th th 2
TEth th th th
i .Pi / D di .Pi / C ei Pi C fi
th
i 2 ˝th (3.5)
th
where dith , eth th
i , and fi are the fuel emission coefficients of the thermal unit i. TEi is
the total emission produced by thermal unit i and is expressed as (kg/h).
Different strategies have been proposed to take care of environmental issues in
dispatching the power plants, and some of them are listed below:
• Ignoring the environmental pollution and just minimizing the total fuel costs
(ED). Optimization problem described in (3.4) should be solved.
• Minimizing the total environmental pollution as the main objective function
(END). In this case, optimization problem described in (3.6) should be solved.
min TE (3.6a)
Pth
i
X
th 2
TC D ath th th th
i .Pi / C bi Pi C ci (3.6b)
i2˝th
X
2
TE D dith .Pth th th
i / C ei Pi C fi
th
(3.6c)
i2˝th
Pith;min Pth th;max

i Pi i 2 ˝th (3.6d)
X
Pth
i Le (3.6e)
i2˝th
The solution of (3.6) can be found in the first column of Table 3.4.
• Adding the environmental pollution is a penalty to the cost function (penalty).
For this strategy, the environmental pollution price (Ce ) should be known. In this
case, optimization problem described in (3.7) should be solved.
min OF D TC C Ce TE (3.7a)
Pth
i
X
th 2
TC D ath th th th
i2˝th
X
2
i / C ei Pi C fi
th
(3.7c)
i2˝th
Pith;min Pth th;max

i Pi i 2 ˝th (3.7d)
X
Pth
i Le (3.7e)
i2˝th
The solution of (3.7) can be found in the second column of Table 3.4.
• Adding the environmental pollution as a constraint to the original economic
dispatch problem (Elimit). In this case, optimization problem described in (3.8)
should be solved.
min OF D TC (3.8a)
Pth
i
X
th 2
TC D ath th th th
i2˝th
X
2
i / C ei Pi C fi
th
(3.8c)
i2˝th
Pith;min Pth th;max

i Pi i 2 ˝th (3.8d)
X
Pth
i Le (3.8e)
i2˝th
TE Elimit (3.8f)
The solution of (3.8) can be found in the third column of Table 3.4.
• Solving the dispatch problem as multi-objective problem. In this case, optimiza-
tion problem described in (3.9) should be solved.
X
th 2
min TC D ath th th th
i .Pi / C bi Pi C ci (3.9a)
Pth
i i2˝th
X
2
min TE D dith .Pth th th
i / C ei Pi C fi
th
(3.9b)
Pth
i i2˝th
Pith;min Pth th;max

i Pi i 2 ˝th (3.9c)
X
Pth
i Le (3.9d)
i2˝th
The ED, END, Penalty, and Limit strategies are solved using GCode 3.3. In this
code, you can find four solve statements.
• ED strategy: The first solve statement minimizes the total costs without any
concern about environmental issues.
• END strategy: The second solve statement minimizes the total environmental
pollution without any concern about economic issues.
• Penalty strategy: The third solve statement minimizes the total fuel costs +
environmental pollution costs.
• Limited Emission strategy: The fourth solve statement minimizes the total fuel
costs while environmental pollution is limited to some value.
Each solve statement obtains the optimal generating schedules in each strategy. A
parameter (report.gen; /) is defined which is updated after each solve statement.
The solution for different environmental dispatching strategies is given in
Table 3.4.
In some occasions, it is needed to solve the problem in a multi-objective way.
The multi-objective model is described in (3.9) and coded in GCode 3.4.
The Pareto optimal front in multi-objective environmental dispatch strategy for
load = 400 MW is depicted in Fig. 3.2. The operating schedules of different units in
solutions of Pareto optimal front are plotted in Fig. 3.3 for load = 400 MW.
The Pareto optimal front found in Fig. 3.2 depends on technical characteristics of
generating units in one hand and also on load value (D 400 MW). The question is
what happens if the load increases. How can we obtain the Pareto optimal front for
different values of loads? In order to obtain each Pareto optimal front, it is needed to
Table 3.4 Solution for different environmental dispatching strategies

P(g) ED END Penalty Elimit
g1 102.84 71.62 103.05 94.21
g2 90.00 90.00 90.00 90.00
g3 76.73 68.00 75.18 68.00
g4 77.43 129.76 78.77 94.79
g5 53.00 40.62 53.00 53.00
OF ($) 135,310.00 152,170.00 140,990.00 142,190.00
TE (kg) 96,450.75 87,089.40 94,428.55 90,000.00
TC ($) 131,460.00 148,680.00 131,550.00 133,190.00
× 104
9.7
9.6
9.5
9.4
9.3
TE
9.2
9.1
8.9
8.8
8.7
1.3 1.32 1.34 1.36 1.38 1.4 1.42 1.44 1.46 1.48 1.5
TC ($) × 105
Fig. 3.2 Pareto optimal front in multi-objective environmental dispatch strategy for
load = 400 MW
130
g1
120 g2
g3
g4
110 g5
100
Generation level
90
80
70
60
50
40
1 2 3 4 5 6 7 8 9 10 11
Solution (from environmental friendly to economic)
Fig. 3.3 Changes in generating level of different units in Pareto optimal front for load = 400 MW
3.3 CHP Economic Dispatch 73
GCode 3.3 Example of different environmental dispatch strategies

S e t Gen / g1g5 / ;
P a r a m e t e r r e p o r t ( gen , ) ;
Scalars
load / 4 0 0 /
Eprice / 0 . 1 /
Elim / 9 0 0 0 0 / ;
g1 3 20 100 2 5 3 28 206
g2 4 . 0 5 1 8 . 0 7 9 8 . 8 7 3 . 8 2 4.24 6 . 0 9 90 284
g3 4 . 0 5 1 5 . 5 5 1 0 4 . 2 6 5 . 0 1 2.15 5 . 6 9 68 189
g4 3 . 9 9 1 9 . 2 1 1 0 7 . 2 1 1 . 1 3.99 6 . 2 76 266
g5 3 . 8 8 2 6 . 1 8 9 5 . 3 1 3 . 5 5 6.88 5 . 5 7 19 53;
v a r i a b l e s P ( gen ) ,OF , TE , TC ;
equations
eq1 , eq2 , eq3 , eq4 ;
eq1 . . TC=e=sum ( gen , d a t a ( gen , ’ a ’ ) P ( gen ) P ( gen ) + d a t a ( gen , ’ b ’ ) P
( gen ) + d a t a ( gen , ’ c ’ ) ) ;
eq3 . . TE=e=sum ( gen , d a t a ( gen , ’ d ’ ) P ( gen ) P ( gen ) + d a t a ( gen , ’ e ’ ) P
( gen ) + d a t a ( gen , ’ f ’ ) ) ;
eq4 . . OF=e=TC+TE E p r i c e ;
Model END / eq1 , eq2 , eq3 , eq4 / ;
S o l v e END u s qcp min TC ;
r e p o r t ( gen , ’ED ’ ) =P . l ( gen ) ;
S o l v e END u s qcp min t e ;
r e p o r t ( gen , ’END ’ ) =P . l ( gen ) ;
S o l v e END u s qcp min OF ;
r e p o r t ( gen , ’ p e n a l t y ’ ) =P . l ( gen ) ;
TE . up= Elim ;
r e p o r t ( gen , ’ l i m i t ’ ) =P . l ( gen ) ;
Display report ;
have a loop. The constraint method [1] is used to obtain the Pareto optimal front.
The GCode 3.5 finds the Pareto optimal fronts for different load values 400, 450,
500, and 550 MW. It uses nested loops to change the load value in upper level loop
and changes the epsilon value in lower level loop.
The Pareto optimal fronts for different load values are depicted in Fig. 3.4.
3.3 CHP Economic Dispatch
The combined heat and power (CHP) technology can generate heat and electric
power simultaneously. In CHP-related dispatch problems, we are usually faced with
different sources of energy to supply heat and power demand. These sources of
GCode 3.4 Example of multi-objective environmental dispatch strategy

S e t Gen / g1g5 /
c o u n t e r / c1c11 / ;
parameter r e p o r t () , rep ( counter ,) ;
Scalars
load / 4 0 0 /
Eprice / 0 . 1 /
Elim ;
g1 3 20 100 2 5 3 28 206
g2 4 . 0 5 1 8 . 0 7 9 8 . 8 7 3 . 8 2 4.24 6 . 0 9 90 284
g3 4 . 0 5 1 5 . 5 5 1 0 4 . 2 6 5 . 0 1 2.15 5 . 6 9 68 189
g4 3 . 9 9 1 9 . 2 1 1 0 7 . 2 1 1 . 1 3.99 6 . 2 76 266
g5 3 . 8 8 2 6 . 1 8 9 5 . 3 1 3 . 5 5 6.88 5 . 5 7 19 53;
V a r i a b l e s P ( gen ) ,OF , TE , TC ;
Equations
eq1 , eq2 , eq3 ;
( gen ) + d a t a ( gen , ’ c ’ ) ) ;
( gen ) + d a t a ( gen , ’ f ’ ) ) ;
model END / eq1 , eq2 , eq3 / ;
s o l v e END u s qcp min TC ;
r e p o r t ( ’maxTE ’ ) =TE . l ;
r e p o r t ( ’ minTC ’ ) =TC . l ;
r e p o r t ( ’maxTC ’ ) =TC . l ;
r e p o r t ( ’ minTE ’ ) =TE . l ;
loop ( counter ,
Elim = ( r e p o r t ( ’maxTE ’ ) r e p o r t ( ’ minTE ’ ) ) ( ( o r d ( c o u n t e r ) 1) / ( c a r d
( c o u n t e r ) 1) ) + r e p o r t ( ’ minTE ’ ) ;
TE . up= Elim ;
r e p ( c o u n t e r , ’TC ’ ) =TC . l ;
r e p ( c o u n t e r , ’TE ’ ) =TE . l ;
);
display rep ;
GCode 3.5 Sensitivity analysis for load values in multi-objective environmental dispatch strategy
sets
Gen / g1g5 /
c o u n t e r / c1c11 /
L o a d c o u n t e r / Lc1Lc4 / ;
parameter r e p o r t ( Loadcounter ,) , rep ( Loadcounter , counter ,) ,
r e p 2 ( L o a d c o u n t e r , c o u n t e r , gen ) ;
scalars
load / 4 0 0 /
Eprice / 0 . 1 /
Elim ;
g1 3 20 100 2 5 3 28 206
g2 4 . 0 5 1 8 . 0 7 9 8 . 8 7 3 . 8 2 4.24 6 . 0 9 90 284
g3 4 . 0 5 1 5 . 5 5 1 0 4 . 2 6 5 . 0 1 2.15 5 . 6 9 68 189
g4 3 . 9 9 1 9 . 2 1 1 0 7 . 2 1 1 . 1 3.99 6 . 2 76 266
g5 3 . 8 8 2 6 . 1 8 9 5 . 3 1 3 . 5 5 6.88 5 . 5 7 19 53;
V a r i a b l e s P ( gen ) ,OF , TE , TC ;
Equations
eq1 , eq2 , eq3 ;
( gen ) + d a t a ( gen , ’ c ’ ) ) ;
( gen ) + d a t a ( gen , ’ f ’ ) ) ;
Model END / eq1 , eq2 , eq3 / ;
Loop ( L o a d c o u n t e r ,
l o a d =350+ o r d ( L o a d c o u n t e r ) 50;
r e p o r t ( L o a d c o u n t e r , ’maxTE ’ ) =TE . l ;
r e p o r t ( L o a d c o u n t e r , ’ minTC ’ ) =TC . l ;
r e p o r t ( L o a d c o u n t e r , ’maxTC ’ ) =TC . l ;
r e p o r t ( L o a d c o u n t e r , ’ minTE ’ ) =TE . l ;
loop ( counter ,
Elim = ( r e p o r t ( L o a d c o u n t e r , ’maxTE ’ )
r e p o r t ( L o a d c o u n t e r , ’ minTE ’ ) ) ( ( o r d ( c o u n t e r ) 1) / ( c a r d ( c o u n t e r )
1) )
+ r e p o r t ( L o a d c o u n t e r , ’ minTE ’ ) ;
TE . up= Elim ;
r e p ( L o a d c o u n t e r , c o u n t e r , ’TC ’ ) =TC . l ;
r e p ( L o a d c o u n t e r , c o u n t e r , ’TE ’ ) =TE . l ;
r e p 2 ( L o a d c o u n t e r , c o u n t e r , gen ) =P . l ( gen ) ;
);
TE . up= i n f ;
);
D i s p l a y rep , rep2 , r e p o r t ;
× 105
1.9
Load =400 MW
1.8 Load =450 MW
Load =500 MW
1.7 Load =550 MW
1.6
1.5
1.4
TE
1.3
1.2
1.1
0.9
1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3

TC ($) × 105
Fig. 3.4 Pareto optimal fronts for different load values
energy are usually thermal plants, CHP units, and heat only units. As described in
Sect. 3.1, the fuel cost of a thermal unit depends on its electricity generation level.
Similarly, the operating costs of CHP units depend on the level of heat and electricity
they generate. Additionally, CHP units have a feasible operating region that dictates
some limits for power and heat that they can generate. An approximate heat-power
operating region for a CHP unit is shown in Fig. 3.5 [2].
The CHP economic dispatch problem can be modeled as described in (3.10):
min OF D F th C F h C F chp (3.10a)

chp chp h
Pth
i ;Pj ;qj ;qk
X
th 2
F th D ath th th th
i2˝th
X
Fh D ahk .qhk /2 C bhk qhk C chk (3.10c)
k2˝h
X chp chp chp chp chp
F chp D aj .Pj /2 C bj Pj C cj (3.10d)
j2˝chp
chp chp chp chp chp chp chp

C dj .qj /2 C ej qj C fj Pj qj
Pith;min Pth th;max

i Pi i 2 ˝th (3.10e)
Fig. 3.5 Approximate heat-power operating region for a CHP unit
Qh;min
k qhk Qh;max
k k 2 ˝h (3.10f)
chp;min chp chp chp;max chp
Pj .qj / Pj Pj .qj / j 2 ˝chp (3.10g)
chp;min chp chp chp;max chp
Qj .Pj / qj Qj .Pj / j 2 ˝chp (3.10h)
X X chp
Pth
i C Pj Le (3.10i)
i2˝th j2˝chp
X X chp
qhk C qj Lh (3.10j)
k2˝h j2˝chp
The objective function defined in (3.10a) has three terms namely thermal unit
(F th ) [calculated in (3.10b)], heat only unit (F h ) [calculated in (3.10c)], and CHP
unit (F chp ) costs [calculated in (3.10d)]. The decision variables are electric power
chp
generation of thermal units (Pth i ) and CHP units Pj and also the heat generated
chp
by CHP units qj and heat only units qhk as described in (3.10a). The operating
limits of thermal units and heat only units are modeled in (3.10e) and (3.10f),
respectively as given in Table 3.5. Unlike the other technologies mentioned before,
the operating limits of CHP units are dependent on electricity and heat generation
level as described in (3.10g) and (3.10h) as described in Table 3.6. The electricity
and heat supply-demand balance are modeled in (3.10i) and (3.10j), respectively.
The electric and heat demands are assumed to be 605 MW and 540 MW,
respectively.
Table 3.5 Characteristic data of different technologies

Technology a b c d e f Pmin Pmax hmin hmax
g1 3.00 20.00 100.00 30.00 180.00
g2 4.05 18.07 98.87 90.00 290.00
h1 4.05 10.55 104.26 60.00 200.00
h2 3.99 9.21 107.21 70.00 270.00
chp1 0.0345 14.00 2540.00 0.03 4.20 0.031
chp2 0.0435 13.00 1460.00 0.02 0.70 0.011
Table 3.6 Feasible operating CHP unit Aq Ap Bq Bp Cq Cp Dq Dp

regions of CHP units
chp1 0 247 180 215 104.8 81 0 99
chp2 0 125 135 110 75 40 0 45
GCode 3.6 CHP dispatching GAMS code for Example (3.10)

Sets
Gen / g1g2 /
h e a t / h1h2 /
CHP / chp1chp2 / ;
scalars
Le / 6 0 5 /
Lh / 5 4 0 / ;
T a b l e d a t a T h ( Gen , )
a b c d e f Pmin Pmax hmin hmax
g1 3 20 100 0 0 0 28 206 0 0
g2 4.05 18.07 98.87 0 0 0 90 284 0 0 ;
Table dataH ( heat , )
h1 4.05 10.55 104.26 0 0 0 0 0 60 200
h2 3.99 9.21 107.21 0 0 0 0 0 70 270;
T a b l e d a t a c h p ( chp , )
chp1 0 . 0 3 4 5 14 2540 0.03 4.2 0.031 0 0 0 0
chp2 0 . 0 4 3 5 13 1460 0.02 0.7 0.011 0 0 0 0;
T a b l e FR ( chp , )
Aq Ap Bq Bp Cq Cp Dq Dp
chp1 0 247 180 215 1 0 4 . 8 81 0 99
chp2 0 125 135 110 75 40 0 45;
V a r i a b l e s P ( gen ) ,OF , q ( h e a t ) , pchp ( chp ) , qchp ( chp ) , Fth , Fh , Fchp ;

E q u a t i o n s eq1 , eq2 , eq3 , eq4 , eq5 , eq6 , eq7a , eq7b , e q 7 c ;
eq1 . . OF=e= F t h +Fh+Fchp ;
eq2 . . F t h =e=sum ( gen , d a t a T h ( gen , ’ a ’ ) P ( gen ) P ( gen ) + d a t a T h ( gen , ’
b ’ ) P ( gen ) + d a t a T h ( gen , ’ c ’ ) ) ;
eq3 . . Fh=e=sum ( h e a t , d a t a H ( h e a t , ’ a ’ ) q ( h e a t ) q ( h e a t ) + d a t a H ( h e a t
, ’ b ’ ) q ( h e a t ) + d a t a H ( h e a t , ’ c ’ ) ) ;
eq4 . . Fchp=e=sum ( CHP , d a t a c h p ( CHP , ’ a ’ ) Pchp (CHP) Pchp (CHP)
+ d a t a c h p ( CHP , ’ b ’ ) Pchp (CHP) + d a t a c h p ( CHP , ’ c ’ ) ) +
3.4 Hydro Unit Economic Dispatch 79
sum ( CHP , d a t a c h p ( CHP , ’ d ’ ) qchp (CHP) qchp (CHP)

+ d a t a c h p ( CHP , ’ e ’ ) qchp (CHP) + d a t a c h p ( CHP , ’ f ’ ) qchp (CHP) pchp (CHP
));
eq5 . . sum ( gen , P ( gen ) ) +sum ( CHP , Pchp (CHP) ) =g=Le ;
eq6 . . sum ( h e a t , q ( h e a t ) ) +sum ( CHP , Qchp (CHP) ) =g=Lh ;
eq7a (CHP) . . Pchp (CHP)FR ( chp , ’Dp ’ ) =g=

( qchp (CHP)FR ( chp , ’Dq ’ ) ) (FR ( chp , ’Dp ’ )FR ( chp , ’Cp ’ ) ) / ( FR ( chp , ’
Dq ’ )FR ( chp , ’ Cq ’ ) ) ;
eq7b (CHP) . . Pchp (CHP)FR ( chp , ’Ap ’ ) = l =
( qchp (CHP)FR ( chp , ’Dq ’ ) ) (FR ( chp , ’Ap ’ )FR ( chp , ’Bp ’ ) ) / ( FR ( chp , ’
Aq ’ )FR ( chp , ’ Bq ’ ) ) ;
eq7c (CHP) . . Pchp (CHP)FR ( chp , ’Bp ’ ) =g=
( qchp (CHP)FR ( chp , ’Bq ’ ) ) (FR ( chp , ’Bp ’ )FR ( chp , ’Cp ’ ) ) / ( FR ( chp , ’
Bq ’ )FR ( chp , ’ Cq ’ ) ) ;
P . l o ( gen ) = d a t a T h ( gen , ’ Pmin ’ ) ;
P . up ( gen ) = d a t a T h ( gen , ’ Pmax ’ ) ;
q . l o ( h e a t ) = d a t a H ( h e a t , ’ hmin ’ ) ;
q . up ( h e a t ) = d a t a H ( h e a t , ’ hmax ’ ) ;
Model c h p d i s p a t c h / a l l / ;
S o l v e c h p d i s p a t c h u s n l p min OF ;
Table 3.7 Solution of Technology p Technology q

dispatching CHP units
chp1 215 chp1 180
chp2 110 chp2 135
g1 160.714 h1 111.577
g2 119.286 h2 113.423
Le 605 Lh 540
The total operating costs are F th D 1:4068 105 ($/h), F h D 1:0418 105 ($/h),
chp
F =14,111.162 ($/h) (Table 3.7).
3.4 Hydro Unit Economic Dispatch
This section addresses the optimal operation of a hydro generating power plant. This
power plant comprises several cascaded plants along a river basin. The objective
is to minimize the total operating costs while considering different technical
constraints such as relationship between the generated power, water discharged, and
the head of the associated reservoir. The concept of cascaded hydro units is shown
in Fig. 3.6. The water balance equations that should be satisfied in each hour are as
follows:
h
2LtC1 D Lth C ItC1
h
RhtC1 StC1
h
(3.11)
Fig. 3.6 Cascaded hydro units
Xh O O
i
C RhtC1 O C StC1
h
O
h h
hO
h
Lmin Lth Lmax
h
; hO 2 up fhg (3.12)
Rht Rhmax ; Lth0 D h
Lini ; Lth24 D h
Lfin (3.13)
h
where Lth is reservoir volume, ItC1 is the water inflow, Rht is the released water,
and Sth is the spilled water at the end of period t in million m3 . Rmax is the maximum
released capacity per hour in million m3 . Lini
h h
; Lfin are the volume of the water in dam
at beginning and end of the considered horizon, respectively. This constraint means
that the volume of water in a reservoir of hydro turbine h in time t C 1 will be equal
to its value in the previous period plus the water inflow to its reservoir in time t C 1
minus its own released/spilled water and in time t C 1 plus the released/spilled water
of all reservoirs in its upstream in previous hours (with considering time delays hO ).
The concept of this cascade reservoir water balance constraint is depicted in Fig. 3.6.
The hydro power production function (HPF) (or hill chart [3]) which relates
the output power of hydro plant to the water level, inflow, and spillage is of great
importance in hydro plant scheduling. In this chapter, the method proposed in [4] has
been adopted which describes the relationship between the released water and water
head of the reservoir with the output power of the hydro power plant, as follows:
Pht D ch1 .Lth /2 C ch2 .Rht /2 C ch3 Rht Lth C ch4 Lth C ch5 Rht C ch6 (3.14)
Table 3.8 Techno-economic characteristics of thermal units

Unit ath 2
i ($=MW ) bth
i ($=MW) cth
i ($) Pth;min
i (MW) Pth;max
i (MW) RUi (MW) RDi (MW)
g1 0.00043 16.6 900 100 400 60 60
g2 0.00073 15.5 800 130 400 40 40
g3 0.00059 14.8 700 70 300 60 60
g4 0.00075 15.9 470 60 300 30 30
g5 0.00079 16.6 200 80 250 50 50
Fig. 3.7 Four cascaded hydro units configuration
Equation (3.14) states that the generated power in time t is nonlinearly related to the
water head (Lth ) and released water (Rht ) in time t simultaneously. The coefficients
ch1!6 are the characteristic factors of hydro turbine h which describe this relation.
Now we can solve a mixed hydro-thermal economic dispatch problem. Suppose we
have five thermal units which their techno-economic characteristics are described
in Table 3.8. Additionally, there are four cascaded hydro power units as shown
in Fig. 3.7 which their techno-economic characteristics are described in Table 3.9.
The loss coefficient is assumed to be 0.9 (means 10% of released/spilled water
vaporizes). The operating cost of hydro power plant is assumed to be constant
h D 14$=MW.
The hourly water inflows for each hydro power and hourly electric demand are
given in Table 3.10.
The hydro-thermal economic dispatch problem is formulated in (3.15).
Table 3.9 Techno-economic characteristics of hydro units

Unit ch1 ch2 ch3 ch4 ch5 ch6 h
Lmin h
Lmax h
Lini h
Lfin Rhmin Rhmin Ph;min Ph;max
h1 0.0042 0.42 0.030 0.90 10 50 80 150 100 120 5 15 0 500 2
h2 0.0040 0.30 0.015 1.14 9.5 70 60 120 80 70 6 15 0 500 1
h3 0.0016 0.30 0.014 0.55 5.5 40 100 240 170 170 10 30 0 500 4
h4 0.0030 0.31 0.027 1.44 14 90 70 160 120 140 6 20 0 500 0
Table 3.10 Hourly water Time Ith1 Ith2 Ith3 Ith4 Le;t (MW)
inflows for each hydro power
and hourly electric demand t1 10 8 8.1 2.8 1275
t2 9 8 8.2 2.4 1326
t3 8 9 4 1.6 1190
t4 7 9 2 0 1105
t5 6 8 3 0 1139
t6 7 7 4 0 1360
t7 8 6 3 0 1615
t8 9 7 2 0 1717
t9 10 8 1 0 1853
t10 11 9 1 0 1836
t11 12 9 1 0 1870
t12 10 8 2 0 1955
t13 11 8 4 0 1887
t14 12 9 3 0 1751
t15 11 9 3 0 1717
t16 10 8 2 0 1802
t17 9 7 2 0 1785
t18 8 6 2 0 1904
t19 7 7 1 0 1819
t20 6 8 1 0 1785
t21 7 9 2 0 1547
t22 8 9 2 0 1462
t23 9 8 1 0 1445
t24 10 8 0 0 1360
!
X X
th 2
min OF D Pht
h C ath th th th
i .Pi;t / C bi Pi;t C ci (3.15a)
DV
t i
X O O
h
LtC1 D Lth C ItC1
h
RhtC1 StC1
h
C ŒRhtC1 O C StC1
h
O
(3.15b)
h h
hO
h
Lmin Lth h
Lmax ; hO 2 up fhg (3.15c)
Rhmin Rht Rhmax (3.15d)
Lth0 D Lini
h
; Lth24 D Lfin
h
(3.15e)
Pht D ch1 .Lth /2 C ch2 .Rht /2 C ch3 Rht Lth C ch4 Lth C ch5 Rht C ch6 (3.15f)
X X
Ph;t C Pth
i;t Le;t (3.15g)
h i
Pth th
i;t Pi;t1 RUi (3.15h)
Pth
i;t1 Pth
i;t RDi (3.15i)
Pith;min Pth th;max

i Pi i 2 ˝th (3.15j)
Ph;min Pht Ph;max (3.15k)
˚
DV 2 Pth h h h h
i;t ; Pt ; Rt ; It ; St (3.15l)
The GCode 3.7 is developed to solve the hydro-thermal economic dispatch problem
described in (3.15). In this code, for saving the space, table demand and table inflow
are shown partially. The optimal hourly dispatch of hydro and thermal units are
given in Table 3.11.
GCode 3.7 Hydro-thermal economic dispatch Example (3.15)

Sets t / t 1 t 2 4 / , H / h1h4 / , i / g1g5 / ; S c a l a r z e t a / 1 4 / ;
T a b l e d a t a (H, )
c1 c2 c3 c4 c5 c6
h1 0.0042 0.44 0 . 0 4 0 0 . 8 0 1 1 . 0 53
h2 0.0043 0.32 0 . 0 1 3 1 . 2 4 9 . 7 71
h3 0.0015 0.31 0 . 0 1 2 0 . 5 4 5 . 7 42
h4 0.0032 0.33 0 . 0 2 5 1 . 4 3 1 4 . 1 91;
T a b l e g e n d a t a ( i , ) g e n e r a t o r c o s t c h a r a c t e r i s t i c s and l i m i t s
a b c l o w l i m u p p l i m RU RD
g1 0 . 0 0 0 4 3 1 6 . 6 900 100 400 60 60
g2 0 . 0 0 0 7 3 1 5 . 5 800 130 400 40 40
g3 0 . 0 0 0 5 9 1 4 . 8 700 70 300 60 60
g4 0 . 0 0 0 7 5 1 5 . 9 470 60 300 30 30
g5 0 . 0 0 0 7 9 1 6 . 6 200 80 250 50 50;
A l i a s (H, Hhat ) ; s e t u p s t r e a m (H, Hhat ) ;
u p s t r e a m ( ’ h3 ’ , Hhat ) $ ( o r d ( Hhat ) <3)= y e s ; u p s t r e a m ( ’ h4 ’ , ’ h3 ’ ) = y e s ;
P a r a m e t e r d e l a y (H)
/ h1 2
h2 1
h3 4
h4 0 / ;
T a b l e i n f l o w ( t , H)
h1 h2 h3 h4
t1 10 8 8.1 2.8
t2 9 8 8.2 2.4
t23 9 8 1 0
t24 10 8 0 0;
p a r a m e t e r demand ( t )
/ t1 1275
t2 1326
t23 1445
t24 1360/;
T a b l e c h a r a c (H, )
Vmin Vmax V i n i V f i n Qmin Qmax Pmin Pmax
h1 80 150 100 120 5 15 0 500
h2 60 120 80 70 6 15 0 500
h3 100 240 170 170 10 30 0 500
h4 70 160 120 140 6 20 0 500;
V a r i a b l e s V(H, t ) ,R (H, t ) , S p i l l (H, t ) , OBJ , PH (H, t ) , c o s t T h e r m a l , p ( i ,
t);
p . up ( i , t ) = g e n d a t a ( i , ” u p p l i m ” ) ; p . l o ( i , t ) = g e n d a t a ( i , ” l o w l i m
”) ;
V . LO(H, t ) = c h a r a c (H, ’ Vmin ’ ) ; V . UP (H, t ) = c h a r a c (H, ’Vmax ’ ) ;
V . FX (H, ’ t 2 4 ’ ) = c h a r a c (H, ’ V f i n ’ ) ;
ph . l o (H, t ) = c h a r a c (H, ’ Pmin ’ ) ; ph . up (H, t ) = c h a r a c (H, ’ Pmax ’ ) ; ;
R . LO(H, t ) = c h a r a c (H, ’ Qmin ’ ) ; R . UP (H, t ) = c h a r a c (H, ’Qmax ’ ) ;
S p i l l . LO(H, t ) = 0 ;
Equations
W a t e r l e v e l , pcalc , Genconst3 , Genconst4 , c o s t T h e r m a l c a l c , balance ,
OFdef ;
c o s t T h e r m a l c a l c . . c o s t T h e r m a l =e=sum ( ( t , i ) , g e n d a t a ( i , ’ a ’ ) power
( p ( i , t ) , 2 ) + g e n d a t a ( i , ’ b ’ ) p ( i , t ) + g e n d a t a ( i , ’ c ’ ) ) ;
G e n c o n s t 3 ( i , t ) . . p ( i , t + 1 )p ( i , t ) = l = g e n d a t a ( i , ’RU ’ ) ;
G e n c o n s t 4 ( i , t ) . . p ( i , t 1)p ( i , t ) = l = g e n d a t a ( i , ’RD ’ ) ;
W a t e r l e v e l (H, t + 1 ) . . V(H, t + 1 ) =e= c h a r a c (H, ’ V i n i ’ ) $ ( o r d ( t ) = 1 ) +
V(H, t ) $ ( o r d ( t ) >1)+ i n f l o w ( t +1 ,H)R (H, t + 1 ) S p i l l (H, t + 1 ) +
0 . 9 sum ( H h a t $ u p s t r e a m (H, Hhat ) ,R ( Hhat , t d e l a y (H) ) + S p i l l ( Hhat , t
d e l a y (H) ) ) ;
p c a l c (H, t ) . . Ph (H, t ) =e= d a t a (H, ’ c1 ’ ) V(H, t ) V(H, t ) + d a t a (H, ’ c2 ’ )
R (H, t ) R (H, t )
+ d a t a (H, ’ c3 ’ ) V(H, t ) R(H, t )
+ d a t a (H, ’ c4 ’ ) V(H, t ) + d a t a (H, ’ c5 ’ ) R(H, t
) + d a t a (H, ’ c6 ’ ) ;
b a l a n c e ( t ) . . sum ( i , p ( i , t ) ) +sum (H, ph (H, t ) ) =g=demand ( t ) ;
OFdef . . OBJ=e= z e t a sum ( ( H, t ) , Ph (H, t ) ) + c o s t T h e r m a l ;
Model h y d r o / a l l / ;
S o l v e h y d r o u s n l p min OBJ ;
The thermal unit power schedules are shown in Fig. 3.8. The hydro unit power
schedules are shown in Fig. 3.9.
Different models and methodologies related to hydro power scheduling have
been addressed in the literature such as: Self-scheduling of a hydro producer in
a pool-based electricity market [3], dispatchability enhancement of variable Wind
Generation by coordination With Pumped-Storage Hydro Units [5], coordinated
wind-hydro bidding strategies in day-ahead markets [6], and self-scheduling of
hydro-thermal Genco in smart grids [7].
Table 3.11 Optimal hourly dispatch of hydro and thermal units
h1 h2 h3 h4 Thermal
Time P V R P V R P V R P V R p1 p2 p3 p4 p5
t1 128.50 150.00 15.00 112.78 120.00 15.00 60.60 234.97 13.74 285.88 160.00 20.00 100.00 207.24 240.00 60.00 80.00
t2 75.99 101.14 7.86 54.84 82.00 6.00 52.35 157.76 12.25 285.88 160.00 20.00 139.70 247.24 300.00 90.00 80.00
t3 49.59 104.14 5.00 56.64 85.00 6.00 50.55 149.76 12.00 285.88 160.00 20.00 100.00 222.90 284.44 60.00 80.00
t4 49.82 106.14 5.00 58.36 88.00 6.00 48.09 139.86 11.90 277.27 150.80 20.00 100.00 200.00 224.44 67.01 80.00
t5 49.93 107.14 5.00 59.47 90.00 6.00 35.43 100.09 11.13 268.03 141.51 20.00 100.00 240.00 209.13 97.01 80.00
t6 55.38 108.65 5.49 60.01 91.00 6.00 60.57 240.00 13.84 285.88 160.00 20.00 142.02 280.00 269.13 127.01 80.00
t7 94.17 106.49 10.16 65.31 90.15 6.85 60.60 233.25 13.71 285.88 160.00 20.00 202.02 320.00 300.00 157.01 130.00
3.4 Hydro Unit Economic Dispatch
t8 85.19 106.64 8.85 60.09 91.15 6.00 60.40 222.93 13.51 285.88 160.00 20.00 245.04 360.00 300.00 187.01 133.38
t9 87.32 107.53 9.11 62.63 92.92 6.23 59.85 211.60 13.29 285.88 160.00 20.00 285.12 400.00 300.00 217.01 155.19
t10 88.86 109.28 9.26 67.55 95.13 6.79 58.94 200.28 13.07 285.88 160.00 20.00 251.09 400.00 300.00 247.01 136.67
t11 91.06 111.80 9.48 72.11 96.76 7.37 57.73 189.40 12.86 285.88 160.00 20.00 250.08 400.00 300.00 277.01 136.12
t12 92.43 112.14 9.66 75.87 96.76 7.99 57.08 184.49 12.76 285.88 160.00 20.00 287.34 400.00 300.00 300.00 156.40
t13 98.80 112.52 10.61 82.83 95.33 9.43 56.37 179.63 12.67 285.88 160.00 20.00 229.52 400.00 300.00 300.00 133.60
t14 83.79 116.10 8.42 72.85 96.85 7.48 55.50 174.22 12.57 285.88 160.00 20.00 169.52 400.00 300.00 299.86 83.60
t15 93.21 117.50 9.60 81.81 96.78 9.07 54.66 169.44 12.47 285.88 160.00 20.00 121.44 400.00 300.00 300.00 80.00
t16 93.99 117.80 9.70 83.50 95.19 9.58 53.71 164.38 12.38 285.88 160.00 20.00 181.44 400.00 300.00 300.00 103.49
t17 90.66 117.54 9.26 82.68 92.46 9.73 52.84 160.05 12.29 285.88 160.00 20.00 182.36 400.00 300.00 300.00 90.59
t18 97.29 115.27 10.26 85.50 87.32 11.15 52.38 157.86 12.25 285.88 160.00 20.00 242.36 400.00 300.00 300.00 140.59
t19 93.39 112.49 9.79 83.90 82.78 11.53 50.84 150.95 12.12 285.88 160.00 20.00 199.31 400.00 300.00 297.20 108.49
t20 97.52 107.85 10.63 82.85 78.68 12.10 49.78 146.54 12.03 285.88 160.00 20.00 171.77 400.00 300.00 267.20 130.00
t21 50.17 109.85 5.00 72.91 78.10 9.58 49.07 143.68 11.97 285.88 160.00 20.00 111.77 360.00 300.00 237.20 80.00
t22 50.36 112.85 5.00 67.77 78.63 8.47 50.79 151.27 11.50 285.88 160.00 20.00 100.00 320.00 300.00 207.20 80.00
t23 62.56 115.73 6.13 79.29 74.75 11.88 52.22 160.81 10.73 276.84 150.35 20.00 100.00 316.90 300.00 177.20 80.00
t24 58.59 120.00 5.73 77.98 70.00 12.75 52.85 170.00 10.00 266.48 140.00 20.00 100.00 276.90 300.00 147.20 80.00
85
P1
400
P2
P3
350 P4
Power of thermal units (MW)
P5
300
250
200
150
100
50
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 3.8 Thermal unit schedules in multi-area economic dispatch
300
250 P h1
P h2
Power of hydro unit (MW)
P h3
200
P h4
150
100
50
0
0 5 10 15 20 25
Time
Fig. 3.9 Hydro unit schedules in multi-area economic dispatch

3.5 Multi-Area Mix Unit Dynamic Dispatch 87
3.5 Multi-Area Mix Unit Dynamic Dispatch
The multi-area economic dispatch refers to a class of economic dispatch problems in

which there are several interconnected areas. In each area, there are different power
plant technologies and different demand pattern. The question is how to dispatch
different power plants to supply the demand while minimizing the operating costs.
A basic multi-area economic dispatch problem is shown in Fig. 3.10. There are ten
thermal power plants which their connection area are given in Fig. 3.10. There exist
two wind power plants, one in area 1 (250 MW) and one in area 3 (350 MW).
The multi-area dynamic economic dispatch problem is formulated in (3.16). The
objective function is defined as total operating costs (3.16a). The operating limits of
thermal units are given in (3.16b). The supply-demand balance of area a at time t
is modeled in (3.16c). The amount of transferable power between area a and a is
described in (3.16d). The actual wind dispatch (pww;t ) depends on wind availability
a
N w ) as expressed in (3.16e). Le;t
(w;t ) and wind capacity (pw is the electric load in area
a and time t in (3.16c).
X
min aa
OF D ai p2i;t C bi pi;t C ci (3.16a)
pi;t ;pww;t ;Tiet
i;t
Pmin
i pi;t Pmax
i (3.16b)
X X X
pww;t C Pi;t D Tieaa
t
a
C Le;t 8i; w 2 a; 8t (3.16c)
w i a
ˇ aa ˇ
t
N aa
ˇTie ˇ Tie (3.16d)
Fig. 3.10 Multi-area dynamic economic dispatch problem

Demand of each area (MW)

800
Area 1
Area 2
600
Area 3
400
200
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
1
Wind ratio (t)
0.8
w1
w2
0.6
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 3.11 Wind and demand pattern in each area

Table 3.12 Characteristics of thermal units and their connected area
Area Unit ai ($=MW2 ) bi ($=MW) ci ($) Pmin
i (MW) Pmax
i (MW) RUi (MW) RDi (MW)
P1 0.0056 17.87 601.75 20 150 40 40
P2 0.0079 21.62 480.29 40 200 80 80
A1
P3 0.007 23.9 471.6 30 300 100 100
P4 0.0043 21.6 958.2 30 350 120 120
P5 0.0095 22.54 692.4 10 70 30 30
A2 P6 0.009 19.58 455.6 20 80 40 40
P7 0.0063 21.05 1313.6 40 450 150 150
P8 0.0048 23.23 639.4 50 130 50 50
A3 P9 0.0039 20.81 604.97 100 340 100 100
P10 0.0021 16.51 502.7 40 130 60 60
0 pww;t w;t pw
N w (3.16e)
The wind and demand patterns in each area are shown in Fig. 3.11. The techno-
economic characteristics of thermal units and their connected area are given in
Table 3.12. The GCode 3.8 solves the multi-area economic dynamic dispatch
Example (3.16). Four sets are defined for time, area, thermal units, and wind
turbines, then it is indicated which thermal unit belongs to each area. The table wind
data describes the w;t in (3.16e). The wind capacities and their connected areas are
indicated next. Tie-line limits and thermal unit characteristics are defined next. It
3.5 Multi-Area Mix Unit Dynamic Dispatch 89
should be mentioned that some data of demand and wind tables in GCode 3.8 are
omitted to save space.
GCode 3.8 Multi-area economic dynamic dispatch Example (3.16)

S e t s t / t 1 t 2 4 / , Area / A1A3 / , i / g1g10 / , w / w1w2 / ; s e t
AreaGen ( a r e a , i ) ;
AreaGen ( a r e a , i ) =no ; AreaGen ( ’A1 ’ , i ) $ ( o r d ( i ) <5)= y e s ;
AreaGen ( ’A2 ’ , i ) $ ( o r d ( i ) >4 and o r d ( i ) <8)= y e s ;
AreaGen ( ’A3 ’ , i ) $ ( o r d ( i ) >7)= y e s ; a l i a s ( a r e a , r e g i o n ) ;
T a b l e w i n d d a t a ( t , w)
w1 w2
t1 0.989432703003337 0.515610651974288
t2 0.932703003337041 0.701561065197429
t3 0.890989988876529 0.740128558310376
t4 0.889877641824249 0.676308539944904
t5 0.914905450500556 0.770431588613407
t6 0.937708565072303 0.785123966942149
t7 0.954393770856507 0.849403122130395
t8 0.956618464961068 0.895316804407714
t24 0.937152391546162 0.557392102846648 ;
p a r a m e t e r Windcap (w)
/ w1 250
w2 3 5 0 / ;
p a r a m e t e r AreaWind ( a r e a , w) ; AreaWind ( ’A1 ’ , ’w1 ’ ) = y e s ; AreaWind ( ’
A3 ’ , ’w2 ’ ) = y e s ;
t a b l e Tielim ( area , region )
a1 a2 a3
a1 0 100 400
a2 100 0 500
a3 400 500 0;
t a b l e g e n d a t a ( i , ) g e n e r a t o r c o s t c h a r a c t e r i s t i c s and l i m i t s
a b c Pmin Pmax RU RD
g1 0 . 0 0 5 6 1 7 . 8 7 6 0 1 . 7 5 20 150 40 40
g2 0 . 0 0 7 9 2 1 . 6 2 4 8 0 . 2 9 40 200 80 80
g3 0 . 0 0 7 0 2 3 . 9 4 7 1 . 6 30 300 100 100
g4 0 . 0 0 4 3 2 1 . 6 9 5 8 . 2 30 350 120 120
g5 0 . 0 0 9 5 2 2 . 5 4 6 9 2 . 4 10 70 30 30
g6 0 . 0 0 9 0 1 9 . 5 8 4 5 5 . 6 20 80 40 40
g7 0 . 0 0 6 3 2 1 . 0 5 1 3 1 3 . 6 40 450 150 150
g8 0 . 0 0 4 8 2 3 . 2 3 6 3 9 . 4 50 130 50 50
g9 0 . 0 0 3 9 2 0 . 8 1 6 0 4 . 9 7 100 340 100 100
g10 0 . 0 0 2 1 1 6 . 5 1 5 0 2 . 7 40 130 60 60;
t a b l e demand ( t , a r e a )
a1 a2 a3
t1 258 292 237
t2 291 237 289
t3 343 214 299
t4 435 267 371
t5 408 295 393
t6 477 293 406
t7 422 311 369
t 1 0 606 422 555
t 2 4 377 251 302 ;

v a r i a b l e s Tie ( area , region , t ) , OF , p( i , t ) , Pw (w, t ) ;
t i e . l o ( a r e a , r e g i o n , t )=T i e l i m ( a r e a , r e g i o n ) ;
t i e . up ( a r e a , r e g i o n , t ) =+ T i e l i m ( a r e a , r e g i o n ) ;
t i e . fx ( area , re gi on , t ) $ ( T i e l i m ( area , r e g i o n ) =0) =0;
t i e . f x ( a r e a , a r e a , t ) = 0 ;Pw . l o (w, t ) = 0 ;
p . up ( i , t ) = g e n d a t a ( i , " Pmax " ) ; p . l o ( i , t ) = g e n d a t a ( i , " Pmin " ) ;
Pw . up (w, t ) = w i n d d a t a ( t , w) Windcap (w) ;
E q u a t i o n s t i e c o n s t , b a l a n c e , RampUp , RampDn , c o s t ;
t i e c o n s t ( a r e a , r e g i o n , t ) . . T i e ( a r e a , r e g i o n , t ) =e=T i e ( r e g i o n ,
area , t ) ;
b a l a n c e ( a r e a , t ) . . sum ( i $ A r e a G e n ( a r e a , i ) , p ( i , t ) ) +sum ( w$AreaWind
( a r e a , w) ,Pw (w, t ) ) =e=demand ( t , a r e a ) +sum ( r e g i o n , T i e ( a r e a ,
region , t ) ) ;
RampUp ( i , t ) . . p ( i , t )p ( i , t 1)= l = g e n d a t a ( i , ’RU ’ ) ;
RampDn ( i , t ) . . p ( i , t 1)p ( i , t ) = l = g e n d a t a ( i , ’RD ’ ) ;
c o s t . . OF=e= sum ( ( i , t ) , g e n d a t a ( i , ’ a ’ ) p ( i , t ) p ( i , t ) + g e n d a t a ( i , ’
b ’ ) p ( i , t )
+gendata ( i , ’c ’ ) ) ;
Model e d c / a l l / ;
S o l v e e d c min OF u s QCP ;
The inter-area power transfer of different tie-lines is shown in Fig. 3.12.

Some other examples of multi-area dynamic dispatch can be found in [8]. The
optimal solution of multi-area economic dispatch is described in Table 3.13.
300
250
200
Inter-area power transfer (MW)
150
100
50
-50
-100
TieA1-2
-150 TieA1-3
TieA3-2
-200
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 3.12 Inter-area power transfer of different tie-lines

3.6 Applications 91
Table 3.13 Solution of multi-area economic dispatch

Time P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 Pw1 Pw2
t1 20.0 40.0 30.0 30.0 10.0 20.0 40.0 50.0 100.0 40.0 247.4 159.6
t2 20.0 40.0 30.0 30.0 10.0 20.0 40.0 50.0 100.0 40.0 191.5 245.5
t3 20.0 40.0 30.0 30.0 10.0 20.0 40.0 50.0 100.0 40.0 217.0 259.0
t4 60.0 40.0 30.0 42.7 10.0 60.0 72.8 50.0 148.3 100.0 222.5 236.7
t5 100.0 40.0 30.0 30.0 10.0 67.6 40.0 50.0 100.0 130.0 228.7 269.7
t6 140.0 40.0 30.0 30.0 10.0 66.1 53.5 50.0 117.2 130.0 234.4 274.8
t7 110.0 40.0 30.0 30.0 10.0 26.1 40.0 50.0 100.0 130.0 238.6 297.3
t8 150.0 40.0 30.0 30.0 10.0 66.1 43.1 50.0 100.3 130.0 239.2 313.4
t9 150.0 40.0 30.0 65.8 10.0 80.0 88.6 50.0 173.9 130.0 236.0 297.8
t10 150.0 71.2 30.0 133.2 10.8 80.0 134.5 50.0 248.1 130.0 222.5 322.7
t11 150.0 69.1 30.0 129.3 10.0 80.0 131.9 50.0 243.9 130.0 230.8 350.0
t12 150.0 67.5 30.0 126.4 10.0 80.0 129.9 50.0 240.7 130.0 236.7 332.8
t13 150.0 40.0 30.0 64.2 10.0 80.0 87.5 50.0 172.1 130.0 245.6 336.7
t14 150.0 40.0 30.0 74.2 10.0 80.0 94.3 50.0 183.1 130.0 237.5 328.9
t15 150.0 40.0 30.0 46.5 10.0 80.0 75.4 50.0 152.6 130.0 250.0 310.5
t16 150.0 40.0 30.0 30.0 10.0 80.0 58.3 50.0 124.9 130.0 207.3 323.5
t17 128.2 40.0 30.0 30.0 10.0 40.0 40.0 50.0 100.0 130.0 161.7 344.1
t18 150.0 40.0 30.0 30.0 10.0 80.0 45.3 50.0 104.0 130.0 157.0 343.7
t19 150.0 40.0 30.0 71.7 10.0 80.0 92.6 50.0 180.3 130.0 183.0 316.4
t20 150.0 56.2 30.0 105.6 10.0 80.0 115.7 50.0 217.7 130.0 203.3 284.4
t21 150.0 40.0 30.0 53.2 10.0 80.0 80.0 50.0 159.9 130.0 178.5 285.4
t22 150.0 40.0 30.0 44.4 10.0 80.0 73.9 50.0 150.2 130.0 170.9 258.6
t23 120.2 40.0 30.0 30.0 10.0 60.0 40.0 50.0 100.0 130.0 189.8 238.0
t24 80.2 40.0 30.0 30.0 10.0 20.0 40.0 50.0 100.0 100.4 234.3 195.1
3.6 Applications
Some optimization of economic dispatch models utilized in power system studies is

given in this section:
• Economic dispatch with valve point loading [9]
• Non-smooth cost functions in economic dispatch analysis [10]
• Multi fuel handling [11]
• Transmission loss impacts on ED problem [12]
• Reserve constrained economic dispatch with prohibited operating zones [13]
• Environmentally constrained economic dispatch [14]
• Security constrained economic dispatch problem [15]
• Frequency-constrained stochastic ED [16]
• Decentralized economic dispatch [17]
Nomenclature
Indices and Sets

i Index of thermal generating units.
j Index of chp units.
k Index of heat only units.
˝th Set of all thermal generating units.
˝h Set of all heat only units.
˝chp Set of all chp units.
Parameters
Pth
i Generated electric power by thermal unit i.
chp
pj Generated electric power by chp unit j.
chp
qj Generated heat power by chp unit j.
F th Costs of thermal units.
Fh Costs of heat only units.
F chp Costs of chp units.
qhk Generated heat power by heat only unit k.
h;min=max
Qk Minimum and maximum limits of heat generation of heat only
unit k.
chp;min=max
Pj Minimum and maximum limits of electricity generation of chp
unit j.
chp;min=max
Qj Minimum and maximum limits of heat generation of chp unit j.
Le Electric power demand
Lh Heat power demand
fd; e; f gth
i Emission coefficients of thermal unit i.
fa; b; cgth
i Fuel cost coefficients of thermal unit i.
fa; b; cghi Fuel cost coefficients of heat unit k.
chp
fa; b; c; d; e; f gj Fuel cost coefficients of chp unit j.
th;max=min
Pi Maximum/minimum limits of power generation of thermal
unit i.
Elimit Maximum emission limit (kg).
Parameter used for multi-objective optimization.
Ce Emission price ($/kg).
Variables
Pi Power generated by thermal unit i (MW).
OF Total operating costs ($).
TC Total fuel cost ($).
EM Total emissions (kg).
References 93
References
1. A. Rabiee, A. Soroudi, B. Mohammadi-Ivatloo, M. Parniani, Corrective voltage control scheme

considering demand response and stochastic wind power. IEEE Trans. Power Syst. 29(6),
2965–2973 (2014)
2. T. Guo, M.I. Henwood, M. van Ooijen, An algorithm for combined heat and power economic
dispatch. IEEE Trans. Power Syst. 11(4), 1778–1784 (1996)
3. A.J. Conejo, J.M. Arroyo, J. Contreras, F.A. Villamor, Self-scheduling of a hydro producer in
a pool-based electricity market. IEEE Trans. Power Syst. 17(4), 1265–1272 (2002)
4. J. Zhang, J. Wang, C. Yue, Small population-based particle swarm optimization for short-term
hydrothermal scheduling. IEEE Trans. Power Syst. 27(1), 142–152 (2012)
5. M.E. Khodayar, M. Shahidehpour, L. Wu, Enhancing the dispatchability of variable wind
generation by coordination with pumped-storage hydro units in stochastic power systems.
6. A.A. Snchez de la Nieta, J. Contreras, J.I. Muoz, Optimal coordinated wind-hydro bidding
strategies in day-ahead markets. IEEE Trans. Power Syst. 28(2), 798–809 (2013)
7. A. Soroudi, Robust optimization based self scheduling of hydro-thermal Genco in smart grids.
Energy 61, 262–271 (2013)
8. A. Soroudi, A. Rabiee, Optimal multi-area generation schedule considering renewable
resources mix: a real-time approach. IET Gener. Transm. Distrib. 7(9), 1011–1026 (2013)
9. D.C. Walters, G.B. Sheble, Genetic algorithm solution of economic dispatch with valve point
loading. IEEE Trans. Power Syst. 8(3), 1325–1332 (1993)
10. J.-B. Park, K.-S. Lee, J.-R. Shin, K.Y. Lee, A particle swarm optimization for economic
dispatch with nonsmooth cost functions. IEEE Trans. Power Syst. 20(1), 34–42 (2005)
11. C.-L. Chiang, Improved genetic algorithm for power economic dispatch of units with valve-
point effects and multiple fuels. IEEE Trans. Power Syst. 20(4), 1690–1699 (2005)
12. Z.X. Liang, J.D. Glover, A zoom feature for a dynamic programming solution to economic
dispatch including transmission losses. IEEE Trans. Power Syst. 7(2), 544–550 (1992)
13. F.N. Lee, A.M. Breipohl, Reserve constrained economic dispatch with prohibited operating
zones. IEEE Trans. Power Syst. 8(1), 246–254 (1993)
14. K.P. Wang, J. Yuryevich, Evolutionary-programming-based algorithm for environmentally-
constrained economic dispatch. IEEE Trans. Power Syst. 13(2), 301–306 (1998)
15. R.A. Jabr, A.H. Coonick, B.J. Cory, A homogeneous linear programming algorithm for the
security constrained economic dispatch problem. IEEE Trans. Power Syst. 15(3), 930–936
(2000)
16. Y.Y. Lee, R. Baldick, A frequency-constrained stochastic economic dispatch model. IEEE
Trans. Power Syst. 28(3), 2301–2312 (2013)
17. W.T. Elsayed, E.F. El-Saadany, A fully decentralized approach for solving the economic
dispatch problem. IEEE Trans. Power Syst. 30(4), 2179–2189 (2015)
Chapter 4
Dynamic Economic Dispatch
This chapter provides a solution for dynamic economic dispatch (DED) problem in
GAMS. DED refers to dispatch a set of units over a given operating horizon (usually
24 h). The on/off status of units is assumed to be known. Two paradigms of DED
are discussed and modeled, namely cost-based DED and price-based DED. Finally,
the linearized version of DED is provided.
4.1 Cost-Based DED
The objective function of cost-based DED is minimizing the total operating costs
while satisfying the hourly demands and other technical constraints. The production
costs of g-th thermal unit is defined as:
C.Pg;t / D ag P2g;t C bg Pg;t C cg (4.1)
where ag , bg , and cg are the fuel cost coefficients of the g-th unit. The total fuel cost
(TC) is calculated as follows:
X
TC D C.Pg;t / (4.2)
g;t
The total emission (EM) is calculated as follows:

X
EM D dg P2g;t C eg Pg;t C fg (4.3)
g;t
The operating limits are defined as follows:
Pmin
g Pg;t Pmax
g (4.4)

DOI 10.1007/978-3-319-62350-4_4
96 4 Dynamic Economic Dispatch
Table 4.1 The dynamic economic dispatch data for four units example
g ag bg cg dg eg fg Pmin
g Pmax
g RUg RDg
($=MW2 ) ($=MW) ($) (kg=MW2 ) (kg=MW) .kg/ (MW) (MW) (MW) (MW)
g1 0.12 14.8 89 1.2 5 3 28 200 40 40
g2 0.17 16.57 83 2.3 4.24 6.09 20 290 30 30
g3 0.15 15.55 100 1.1 2.15 5.69 30 190 30 30
g4 0.19 16.21 70 1.1 3.99 6.2 20 260 50 50
max=min
where Pg are the maximum/minimum power outputs of g-th thermal unit. The
overall thermal unit cost-based dynamic economic dispatch is formulated as follows:
X
min TC D ag P2g;t C bg Pg;t C cg (4.5a)
Pg;t
g;t
X
EM D dg P2g;t C eg Pg;t C fg (4.5b)
g;t
Pmin
g Pg;t Pmax
g (4.5c)
Pg;t Pg;t1 RUg (4.5d)
Pg;t1 Pg;t RDg (4.5e)
X
Pg;t Lt (4.5f)
g
The economic dispatch data for four units example is given in Table 4.1. This
table has 11 columns. The first column is showing the generating unit index. The
next three columns indicate the cost coefficients for these thermal units (ag ; bg ; cg ).
The next three columns indicate the emission coefficients for these thermal units
(dg ; eg ; fg ). The next two columns give the minimum and maximum generating limits
of each unit if they are on. The last two columns indicate the ramp up/down rates of
thermal units.
GCode 4.1 The cost-based dynamic economic dispatch Example (4.5)

Sets t hours / t1t24 /
i thermal units / g1g4 / ;
a b c d e f Pmin Pmax RU0 RD0
g1 0 . 1 2 1 4 . 8 0 89 1 . 2 5 3 28 200 40 40
g2 0 . 1 7 1 6 . 5 7 83 2 . 3 4.24 6 . 0 9 20 290 30 30
g3 0 . 1 5 1 5 . 5 5 100 1 . 1 2.15 5 . 6 9 30 190 30 30
g4 0 . 1 9 1 6 . 2 1 70 1 . 1 3.99 6 . 2 20 260 50 50;
P a r a m e t e r demand ( t )
/ t1 510
t2 530
t3 516
4.1 Cost-Based DED 97
t4 510
t5 515
t6 544
t7 646
t8 686
t9 741
t10 734
t11 748
t12 760
t13 754
t14 700
t15 686
t16 720
t17 714
t18 761
t19 727
t20 714
t21 618
t22 584
t23 578
t24 544/;
Variables OBJ Objective ( revenue )
costThermal Cost of thermal u n i t s
p( i , t ) Power g e n e r a t e d by t h e r m a l power
plant
EM Emission c a l c u l a t i o n ;
p . up ( i , t ) = g e n d a t a ( i , " Pmax " ) ;
p . l o ( i , t ) = g e n d a t a ( i , " Pmin " ) ;
E q u a t i o n s G e n c o n s t 3 , G e n c o n s t 4 , c o s t T h e r m a l c a l c , b a l a n c e , EMcalc ;
c o s t T h e r m a l c a l c . . c o s t T h e r m a l =e=sum ( ( t , i ) , g e n d a t a ( i , ’ a ’ ) power (
p( i , t ) ,2)
+ g e n d a t a ( i , ’ b ’ ) p ( i , t ) + g e n d a t a ( i , ’ c ’ ) ) ;
G e n c o n s t 3 ( i , t ) . . p ( i , t + 1 )p ( i , t ) = l = g e n d a t a ( i , ’RU0 ’ ) ;
G e n c o n s t 4 ( i , t ) . . p ( i , t 1)p ( i , t ) = l = g e n d a t a ( i , ’RD0 ’ ) ;
b a l a n c e ( t ) . . sum ( i , p ( i , t ) ) =g=demand ( t ) ;
EMcalc . . EM=e=sum ( ( t , i ) , g e n d a t a ( i , ’ d ’ ) power ( p ( i , t ) , 2 )
+ g e n d a t a ( i , ’ e ’ ) p ( i , t ) + g e n d a t a ( i , ’ f ’ ) ) ;
Model DEDcostbased / a l l / ;
S o l v e DEDcostbased u s QCP min c o s t T h e r m a l ;
e x e c u t e _ u n l o a d " DEDcostbased . gdx " P . l

e x e c u t e ’ gdxxrw . e x e DEDcostbased . gdx v a r =P r n g = P t h e r m a l ! a1 ’
The thermal unit power schedules obtained by GCode 4.1 are shown in Fig. 4.1.
The numerical values of dynamic economic dispatch solution for four units
example are given in Table 4.2. The total operating costs are TC D $6:4796 105 .
The total emissions are 3:5929 103 tons.
800
700
Demand (t)
600
500
400
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Generated power (MW)
200
150
100 P1 P2 P3 P4
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 4.1 The hourly thermal unit power schedules in cost-based DED
4.1.1 Ramp Rate Sensitivity Analysis
In this section, the impact of ramp rates on optimal solution will be investigated.
For this reason, the ramp rates are gradually (in 2% steps) reduced to 60% of their
original values. The GAMS code for solving the ramp sensitivity example is given
in GCode 4.2.
GCode 4.2 The ramp rate sensitivity analysis of cost-based dynamic economic dispatch
Sets t / t1t24 / , i / p1p4 / ;
Table gendata ( i ,) g e n e r a t o r c o s t c h a r a c t e r i s t i c s and l i m i t s
p1 0 . 1 2 1 4 . 8 0 89 1 . 2 5 3 28 200 40 40
p2 0 . 1 7 1 6 . 5 7 83 2 . 3 4.24 6 . 0 9 20 290 30 30
p3 0 . 1 5 1 5 . 5 5 100 1 . 1 2.15 5 . 6 9 30 190 30 30
p4 0 . 1 9 1 6 . 2 1 70 1 . 1 3.99 6 . 2 20 260 50 50;
/ t1 510
t2 530
t3 516
t4 510
t5 515
t6 544
t7 646
t8 686
t9 741
t10 734
t11 748
t12 760
t13 754
t14 700
t15 686
t16 720
t17 714
t18 761
t19 727
t20 714
t21 618
t22 584
t23 578
t24 544
/;
p( i , t ) Power g e n e r a t e d by t h e t h e r m a l
power p l a n t
E q u a t i o n s G e n c o n s t 3 , G e n c o n s t 4 , c o s t T h e r m a l c a l c , b a l a n c e , EMcalc ;
G e n c o n s t 3 ( i , t ) . . p ( i , t + 1 )p ( i , t ) = l = g e n d a t a ( i , ’RU ’ ) ;
G e n c o n s t 4 ( i , t ) . . p ( i , t 1)p ( i , t ) = l = g e n d a t a ( i , ’RD ’ ) ;
b a l a n c e ( t ) . . sum ( i , p ( i , t ) ) =g=demand ( t ) ;
EMcalc . . EM=e=sum ( ( t , i ) , g e n d a t a ( i , ’ d ’ ) power ( p ( i , t ) , 2 ) + g e n d a t a
( i , ’ e ’ ) p ( i , t ) + g e n d a t a ( i , ’ f ’ ) ) ;
Scalar Rscale / 1 / ;
s e t c o u n t e r / c1c21 / ;
parameter report1 ( counter ,) ;
loop ( counter ,
R s c a l e =1( o r d ( c o u n t e r ) 1) 0 . 0 2 ;
g e n d a t a ( i , ’RU ’ ) = g e n d a t a ( i , ’RU0 ’ ) R S c a l e ;
g e n d a t a ( i , ’RD ’ ) = g e n d a t a ( i , ’RD0 ’ ) R S c a l e ;
S o l v e DEDcostbased u s qcp min c o s t T h e r m a l ;
report1 ( counter , ’ Scale ’ )=Rscale ;
r e p o r t 1 ( c o u n t e r , ’TC ’ ) = c o s t T h e r m a l . l ;
r e p o r t 1 ( c o u n t e r , ’EM’ ) =EM. l ;
);
display report1 ;
e x e c u t e _ u n l o a d " DEDcostbased . gdx " r e p o r t 1
e x e c u t e ’ gdxxrw . e x e DEDcostbased . gdx p a r = r e p o r t 1 r n g = P t h e r m a l ! a1 ’
The sensitivity analysis of ramp rates on hourly thermal unit power schedules in
DED is shown in Fig. 4.2.
Table 4.2 The dynamic Time p1 p2 p3 p4 Load

economic dispatch solution
for four units expressed in t1 166.20 112.10 130.50 101.30 510.00
(MW) t2 172.60 116.60 135.60 105.30 530.00
t3 168.10 113.50 132.00 102.50 516.00
t4 166.20 112.10 130.50 101.30 510.00
t5 167.80 113.20 131.70 102.30 515.00
t6 177.00 119.80 139.10 108.10 544.00
t7 200.00 145.90 168.70 131.50 646.00
t8 200.00 159.10 183.70 143.30 686.00
t9 200.00 184.80 190.00 166.30 741.00
t10 200.00 181.10 190.00 162.90 734.00
t11 200.00 188.40 190.00 169.60 748.00
t12 200.00 194.80 190.00 175.20 760.00
t13 200.00 191.60 190.00 172.40 754.00
t14 200.00 163.70 188.90 147.40 700.00
t15 200.00 159.10 183.70 143.30 686.00
t16 200.00 173.70 190.00 156.30 720.00
t17 200.00 170.50 190.00 153.50 714.00
t18 200.00 195.30 190.00 175.70 761.00
t19 200.00 177.40 190.00 159.60 727.00
t20 200.00 168.60 190.00 155.40 714.00
t21 198.00 138.60 160.00 121.40 618.00
t22 189.80 128.80 149.30 116.20 584.00
t23 187.90 127.40 147.80 114.90 578.00
t24 177.00 119.80 139.10 108.10 544.00
The numerical results for ramp rate sensitivity analysis in dynamic economic
dispatch of 4 units are given in Table 4.3. As it is seen in Fig. 4.2, the total costs will
increase by decreasing the ramp rates. The final total costs (40% reduction in ramp
rates) will be $649,226.53.
4.1.2 Multi-Objective Cost-Emission Minimization
In this section, the decision maker needs to minimize the emissions and total
operating costs simultaneously.
X
minTC D ag P2g;t C bg Pg;t C cg (4.6a)
Pg;t
g;t
X
minEM D dg P2g;t C eg Pg;t C fg (4.6b)
Pg;t
g;t
Total emissions (Tons)

3600
3595
3590
0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
Ramp scale
× 105
6.495
Total costs ($)
6.49
6.485
6.48
0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1

Ramp scale
Fig. 4.2 The sensitivity analysis of ramp rates on hourly thermal unit power schedules in DED
Table 4.3 The numerical Counter Scale TC ($) EM (kg)

results for ramp rate
sensitivity analysis in c1 1:00 647;964:46 3;592;886:80
dynamic economic dispatch c2 0:98 647;966:18 3;592;796:25
of 4 units c3 0:96 647;968:28 3;592;707:78
c4 0:94 647;970:78 3;592;622:76
c5 0:92 647;973:69 3;592;545:64
c6 0:90 647;977:11 3;592;469:20
c7 0:88 647;981:17 3;592;362:06
c8 0:86 647;985:94 3;592;261:31
c9 0:84 647;991:45 3;592;183:41
c10 0:82 647;997:77 3;592;150:64
c11 0:80 648;005:10 3;592;086:32
c12 0:78 648;014:58 3;592;007:06
c13 0:76 648;026:42 3;591;950:09
c14 0:74 648;040:63 3;591;919:82
c15 0:72 648;057:21 3;591;935:58
c16 0:70 648;076:19 3;591;985:60
c17 0:68 648;098:11 3;592;096:38
c18 0:66 648;289:75 3;593;218:29
c19 0:64 648;484:76 3;594;442:68
c20 0:62 648;853:35 3;596;838:57
c21 0:60 649;226:53 3;599;445:17
Pmin
g Pg;t Pmax
g (4.6c)

X
Pg;t Lt (4.6f)
g
The procedure is as follows:

1. Find the maximum of each objective function and save them.
2. Add one of the objective functions (here EM) to the constraints as follows:
EM (4.7)
3. The value will be varied from EMmax to EMmin and then TC is minimized.
The Pareto optimal front for emission-cost minimization DED is shown in Fig. 4.3.
The numerical results for Pareto optimal solution are provided in Table 4.4. The
question is which solution of the Pareto optimal front should be chosen as the best
solution? Which one is the “most preferred” solution among those located on Pareto
optimal front of Fig. 4.3?. A fuzzy satisfying method [1] is used in this section
to find the “the best” solution. The principles of this method are as follows: for
each solution in the Pareto optimal front, Xc , a membership function is defined as
fk .Xc / . This value, which varies between 0 to 1, shows the success degree of Xc in
minimizing the objective function k. A linear membership function is used for both
objective functions, as follows:
8
<0 Otherwise
fk .Xc / D fkmax fk .Xc / (4.8)
: max min fkmin fk .Xc / fkmax
fk fk
A conservative decision maker tries to maximize minimum satisfaction among all

objectives or minimize the maximum dissatisfaction. The final solution can then be
found as:
max . min fk .Xc / / (4.9)

cD1W11 kD1W2
× 106
3.55
3.5
Emission (kg)
3.45
3.4
3.35
3.3
6.5 6.55 6.6 6.65 6.7 6.75 6.8

Total costs ($) × 105
Fig. 4.3 The Pareto optimal front for emission-cost minimization DED
GCode 4.3 The multi-objective cost-emission minimization of dynamic economic dispatch

sets t / t1t24 / , i / g1g4 / , c o u n t e r / c1c11 / ; S c a l a r l i m / i n f / ;
g1 0 . 1 2 1 4 . 8 0 89 1 . 2 5 3 28 200 40 40
g2 0 . 1 7 1 6 . 5 7 83 2 . 3 4.24 6 . 0 9 20 290 30 30
g3 0 . 1 5 1 5 . 5 5 100 1 . 1 2.15 5 . 6 9 30 190 30 30
g4 0 . 1 9 1 6 . 2 1 70 1 . 1 3.99 6 . 2 20 260 50 50;
/ t1 510
t2 530
t3 516
t4 510
t5 515
t6 544
t7 646
t8 686
t9 741
t10 734
t11 748
t12 760
t13 754
t14 700
t15 686
t16 720
t17 714
t18 761
t19 727
t20 714
t21 618
t22 584
t23 578
t24 544/;
Variables OBJ , c o s t T h e r m a l , p ( i , t ) ,EM ;
p . up ( i , t ) = g e n d a t a ( i , " Pmax " ) ; p . l o ( i , t ) = g e n d a t a ( i , " Pmin " ) ;
E q u a t i o n s G e n c o n s t 3 , G e n c o n s t 4 , c o s t T h e r m a l c a l c , b a l a n c e , EMcalc ,
EMlim ;
b a l a n c e ( t ) . . sum ( i , p ( i , t ) ) =e=demand ( t ) ;
( i , ’ e ’ ) p ( i , t ) + g e n d a t a ( i , ’ f ’ ) ) ;
EMlim . . EM = l = l i m ;
parameter r e p o r t 1 ( counter ,) , rep () , r e p o r t 2 ( counter , i , t ) ;
r e p ( ’ TCmin ’ ) = c o s t T h e r m a l . l ;
r e p ( ’EMmax ’ ) =EM. l ;
S o l v e DEDcostbased u s qcp min EM;
r e p ( ’TCmax ’ ) = c o s t T h e r m a l . l ;
r e p ( ’EMmin ’ ) =EM. l ;
loop ( counter ,
l i m = r e p ( ’EMmax ’ ) + ( r e p ( ’EMmin ’ )r e p ( ’EMmax ’ ) ) ( o r d ( c o u n t e r ) 1) /
( c a r d ( c o u n t e r ) 1) ;
r e p o r t 1 ( counter , ’ Epsilon ’ )=lim ;
r e p o r t 1 ( c o u n t e r , ’TC ’ ) = c o s t T h e r m a l . l ;
r e p o r t 1 ( c o u n t e r , ’EM’ ) =EM. l ;
r e p o r t 2 ( c o u n t e r , i , t ) =P . l ( i , t ) ;
);
e x e c u t e ’ gdxxrw . e x e DEDcostbased . gdx p a r = r e p o r t 1 r n g = r e p o r t ! a1 ’
e x e c u t e ’ gdxxrw . e x e DEDcostbased . gdx p a r = r e p o r t 2 r n g = P t h e r m a l ! a1 ’
Considering the (4.9) and the values of Table 4.4, the best solution is c9. The
costs and emissions of this solution are $658,087.973 and 3:3459 103 tones,
respectively. The best compromised solution of Pareto optimal front for emission-
cost minimization DED is shown in Fig. 4.4.
Table 4.4 The numerical results for Pareto optimal solutions (cost-based DED)
Solution TC ($) EM (kg) TC EM min.TC ; EM /
c1 3,592,886.799 647,964.460 3,592,886.799 1.000 0.000 0.000
c2 3,562,019.341 648,050.068 3,562,019.341 0.997 0.100 0.100
c3 3,531,151.883 648,328.132 3,531,151.883 0.989 0.200 0.200
c4 3,500,284.425 648,836.461 3,500,284.425 0.974 0.300 0.300
c5 3,469,416.967 649,627.557 3,469,416.967 0.951 0.400 0.400
c6 3,438,549.510 650,778.132 3,438,549.510 0.917 0.500 0.500
c7 3,407,682.052 652,408.206 3,407,682.052 0.869 0.600 0.600
c8 3,376,814.594 654,716.176 3,376,814.594 0.800 0.700 0.700
c9 3,345,947.136 658,087.973 3,345,947.136 0.701 0.800 0.701
c10 3,315,079.678 663,531.693 3,315,079.678 0.540 0.900 0.540
c11 3,284,212.220 681,788.377 3,284,212.220 0.000 1.000 0.000
220
P1
P2
200 P3
P4
180
160
140
120
100
80
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 4.4 The best compromised solution of Pareto optimal front for emission-cost minimization
DED
4.1.3 Wind Integrated DED
In this section, the role of wind power generation in DED problem is investigated.
The integration of wind power generation in DED problem is formulated as follows:
X X
min TC D ag P2g;t C bg Pg;t C cg C VWC Pwc
t (4.10a)
DV
g;t t
˚
DV D Pg;t ; Pwt ; Pwc
t (4.10b)
Pmin
g Pg;t Pmax
g (4.10c)
X
Pwt C Pg;t Lt (4.10f)
g
Pwt C Pwc w
t t (4.10g)
GCode 4.4 The cost-based wind-DED (4.10)

Sets t hours / t1t24 / ,
g thermal units / p1p4 / ;
t a b l e g e n d a t a ( g , ) g e n e r a t o r c o s t c h a r a c t e r i s t i c s and l i m i t s
p1 0 . 1 2 1 4 . 8 0 89 1 . 2 5 3 28 200 40 40
p2 0 . 1 7 1 6 . 5 7 83 2 . 3 4.24 6 . 0 9 20 290 30 30
p3 0 . 1 5 1 5 . 5 5 100 1 . 1 2.15 5 . 6 9 30 190 30 30
p4 0 . 1 9 1 6 . 2 1 70 1 . 1 3.99 6 . 2 20 260 50 50;
table data ( t ,)
lambda l o a d wind
t1 3 2 . 7 1 510 4 4 . 1
t2 3 4 . 7 2 530 4 8 . 5
t3 3 2 . 7 1 516 6 5 . 7
t4 3 2 . 7 4 510 1 4 4 . 9
t5 3 2 . 9 6 515 2 0 2 . 3
t6 3 4 . 9 3 544 3 1 7 . 3
t7 44.9 646 3 6 4 . 4
t8 52 686 3 1 7 . 3
t9 5 3 . 0 3 741 271
t 1 0 4 7 . 2 6 734 3 0 6 . 9
t 1 1 4 4 . 0 7 748 4 2 4 . 1
t 1 2 3 8 . 6 3 760 398
t 1 3 3 9 . 9 1 754 4 8 7 . 6
t 1 4 3 9 . 4 5 700 5 2 1 . 9
t 1 5 4 1 . 1 4 686 5 4 1 . 3
t 1 6 3 9 . 2 3 720 560
t 1 7 5 2 . 1 2 714 4 8 6 . 8
t 1 8 4 0 . 8 5 761 3 7 2 . 6
t19 41.2 727 3 6 7 . 4
t 2 0 4 1 . 1 5 714 3 1 4 . 3
t 2 1 4 5 . 7 6 618 3 1 6 . 6
t 2 2 4 5 . 5 9 584 3 1 1 . 4
t 2 3 4 5 . 5 6 578 4 0 5 . 4
t 2 4 3 4 . 7 2 544 4 7 0 . 4 ;
4.2 Price-Based DED 107
cost Cost of thermal u n i t s

p(g , t ) Power g e n e r a t e d by t h e r m a l
power p l a n t
Pw ( t ) , PWC( t ) w i n f power wind c u r t a i l m e t ;
Pw . up ( t ) = d a t a ( t , ’ wind ’ ) ; Pw . l o ( t ) = 0 ;
Pwc . up ( t ) = d a t a ( t , ’ wind ’ ) ; Pwc . l o ( t ) = 0 ;
p . up ( g , t ) = g e n d a t a ( g , " Pmax " ) ;
p . l o ( g , t ) = g e n d a t a ( g , " Pmin " ) ;
s c a l a r VWC / 5 0 / ;
E q u a t i o n s G e n c o n s t 3 , G e n c o n s t 4 , c o s t T h e r m a l c a l c , b a l a n c e , wind ;
c o s t T h e r m a l c a l c . . c o s t =e=sum ( t ,VWCpwc ( t ) ) +sum ( ( t , g ) , g e n d a t a ( g ,
’ a ’ ) power ( p ( g , t ) , 2 ) + g e n d a t a ( g , ’ b ’ ) p ( g , t ) +gendata (g , ’c ’ ) ) ;
G e n c o n s t 3 ( g , t ) . . p ( g , t + 1 )p ( g , t ) = l = g e n d a t a ( g , ’RU0 ’ ) ;
G e n c o n s t 4 ( g , t ) . . p ( g , t 1)p ( g , t ) = l = g e n d a t a ( g , ’RD0 ’ ) ;
balance ( t ) . . sum ( g , p ( g , t ) ) +Pw ( t ) =g= d a t a ( t , ’ l o a d ’ ) ;
wind ( t ) . . Pw ( t ) +PWC( t ) =e= d a t a ( t , ’ wind ’ ) ;
Model DEDwindcostbased / a l l / ;
S o l v e DEDwindcostbased u s qcp min c o s t ;
parameter rep ( t ,) ;
r e p ( t , ’ P t h ’ ) =sum ( g , p . l ( g , t ) ) ;
r e p ( t , ’PW’ ) =PW. l ( t ) ;
r e p ( t , ’ Pwc ’ ) =PWc . l ( t ) ;
r e p ( t , ’ Load ’ ) = d a t a ( t , ’ l o a d ’ ) ;
e x e c u t e _ u n l o a d " DEDwindcostbased . gdx " P . l
e x e c u t e ’ gdxxrw . e x e DEDwindcostbased . gdx v a r =P r n g = P t h e r m a l ! a1 ’
e x e c u t e _ u n l o a d " DEDwindcostbased . gdx " r e p
e x e c u t e ’ gdxxrw . e x e DEDwindcostbased . gdx p a r = r e P r n g = r e p ! a1 ’
The total operating costs with wind integration are TC D $2:2692 105 as obtained
in GCode 4.4. The hourly schedules of thermal units in the wind-DED problem are
shown in Fig. 4.5. The hourly variation of total wind and thermal power generation
as well as the demand in wind-DED problem is depicted in Fig. 4.6.
4.2 Price-Based DED
The objective function of price-based DED (PBDED) is maximizing the total

benefits of generating company by selling the energy or other products. In this
case, there is no hourly demands-supply balance constraint; however, other technical
constraints should still be satisfied. The net benefit of generating company is equal
to income minus operating costs.
4.2.1 Price-Based DED Just Energy Market
This case is devoted to the case of price-based DED without arbitrage. This means
that the only sold commodity is energy. The problem is formulated in (4.11) as
follows:
160
g1
g2
140
g3
g4
120
100
(MW)
80
g,t
P
60
40
20
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 4.5 The hourly schedules of thermal units in wind-DED problem
800
Lt
700 P-thermal
Pw
t
600
500
(MW)
400
300
200
100
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 4.6 The hourly variation of total wind and thermal power generation in wind-DED problem
X
max OF D et Pg;t TC (4.11a)
Pg;t
g;t
X
TC D ag P2g;t C bg Pg;t C cg (4.11b)
g;t
X
EM D dg P2g;t C eg Pg;t C fg (4.11c)
g;t
Pmin
g Pg;t Pmax
g (4.11d)
Pg;t Pg;t1 RUg (4.11e)
Pg;t1 Pg;t RDg (4.11f)
X
Pg;t Lt (4.11g)
g;t
The objective function in (4.11a) models the benefits of Genco which is income
minus the total operating costs. et is the electricity price at time t in (4.11a). In this
formulation, we just calculate the total emission in (4.11c). Another difference of
price-based DED with cost-based DED is in Eq. (4.11g). As it can be observed in
this equation, the total power should be less than demand at any time. This means
that the maximum power that Genco can sell is limited to the demand at time t.
GCode 4.5 The code for price-based dynamic economic dispatch

i thermal units / g1g4 / ;
s c a l a r lim / i n f / ;
g1 0 . 1 2 1 4 . 8 0 89 1 . 2 5 3 28 200 40 40
g2 0 . 1 7 1 6 . 5 7 83 2 . 3 4.24 6 . 0 9 20 290 30 30
g3 0 . 1 5 1 5 . 5 5 100 1 . 1 2.15 5 . 6 9 30 190 30 30
g4 0 . 1 9 1 6 . 2 1 70 1 . 1 3.99 6 . 2 20 260 50 50;
table data ( t ,)
lambda load
t1 32.71 510
t2 34.72 530
t3 32.71 516
t4 32.74 510
t5 32.96 515
t6 34.93 544
t7 44.9 646
t8 52 686
t9 53.03 741
t10 47.26 734
t11 44.07 748
t12 38.63 760
t13 39.91 754
t14 39.45 700
t15 41.14 686

t16 39.23 720
t17 52.12 714
t18 40.85 761
t19 41.2 727
t20 41.15 714
t21 45.76 618
t22 45.59 584
t23 45.56 578
t24 34.72 544;
Variables OF Objective ( revenue )
plant
equations
G e n c o n s t 3 , G e n c o n s t 4 , c o s t T h e r m a l c a l c , b a l a n c e , EMcalc , EMlim ,
benefitcalc ;
c o s t T h e r m a l c a l c . . c o s t T h e r m a l =e=sum ( ( t , i ) , g e n d a t a ( i , ’ a ’ ) power (
p( i , t ) ,2)
+ g e n d a t a ( i , ’ b ’ ) p ( i , t ) + g e n d a t a ( i , ’ c ’ ) ) ;
balance ( t ) . . sum ( i , p ( i , t ) ) = l = d a t a ( t , ’ l o a d ’ ) ;
EMcalc . . EM=e=sum ( ( t , i ) , g e n d a t a ( i , ’ d ’ ) power ( p ( i , t ) , 2 ) + g e n d a t a ( i
, ’ e ’ ) p ( i , t )
+gendata ( i , ’ f ’ ) ) ;
b e n e f i t c a l c . . OF=e=sum ( ( i , t ) ,1 d a t a ( t , ’ lambda ’ ) p ( i , t ) )
costThermal ;
Model DEDPB / a l l / ;
S o l v e DEDPB u s qcp max o f ;
e x e c u t e _ u n l o a d "DEDPB . gdx " P . l
e x e c u t e ’ gdxxrw . e x e DEDPB . gdx v a r =P r n g = r e p o r t ! a1 ’
The optimal solution of price-based DED is depicted in Fig. 4.7.
4.2.2 Price-Based DED Energy and Reserve Market
This case is devoted to the case of price-based DED without arbitrage. This means
that the only sold commodities are energy and spinning reserve. The problem is
formulated in (4.12) as follows:
X˚
max OF D et Pg;t C rt SRg;t TC (4.12a)
Pg;t
g;t
160
P1
P2
140 P3
P4
120
100
80
60
40
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 4.7 The optimal solution of price-based DED
X
TC D ag P2g;t C bg Pg;t C cg (4.12b)
g;t
X
EM D dg P2g;t C eg Pg;t C fg (4.12c)
g;t
Pmin
g Pg;t Pmax
g (4.12d)
Pg;t Pg;t1 RUg (4.12e)
Pg;t1 Pg;t RDg (4.12f)
0 SRg;t Pmax
g Pg;t (4.12g)
X
Pg;t Lt (4.12h)
g;t
In order to solve the PBDED problem formulated in (4.12), the energy prices (et )
and reserve prices (rt ) in (4.12a) are needed to be known. The problem is usually
solved in a day ahead time horizon. Various prediction techniques exist in literature
to forecast these numbers based on historical data [2]. The energy and reserve prices
vs. time in price-based DED are depicted in Fig. 4.8. It should be noted that the
numbers shown in Fig. 4.8 are subject to uncertainties. The GAMS code for solving
the price-based DED with arbitrage is provided in GCode 4.6.
55
50
Energy and reserve prices ($/MWh)
e
45 t
r
t
40
35
30
25
20
15
10
5
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 4.8 The energy and reserve prices vs time in price-based DED
GCode 4.6 The code for price-based DED with arbitrage

Sets t / t 1 t 2 4 / , i / g1g4 / ;
S c a l a r lim / i n f / ;
tabl e gendata ( i ,) generator cost c h a r a c t e r i s t i c s and l i m i t s
g1 0 . 1 2 1 4 . 8 0 89 1 . 2 5 3 28 200 40 40
g2 0 . 1 7 1 6 . 5 7 83 2 . 3 4.24 6 . 0 9 20 290 30 30
g3 0 . 1 5 1 5 . 5 5 100 1 . 1 2.15 5 . 6 9 30 190 30 30
g4 0 . 1 9 1 6 . 2 1 70 1 . 1 3.99 6 . 2 20 260 50 50;
table data ( t ,)
lambda load Lr
t1 32.71 510.00 25.55
t2 34.72 530.00 20.83
t3 32.71 516.00 19.68
t4 32.74 510.00 21.73
t5 32.96 515.00 22.43
t6 34.93 544.00 23.94
t7 44.90 646.00 42.22
t8 52.00 686.00 36.53
t9 53.03 741.00 21.41
t10 47.26 734.00 12.58
t11 44.07 748.00 13.86
t12 38.63 760.00 9.58
t13 39.91 754.00 7.18
t14 39.45 700.00 12.16
4.3 Linearized Cost-Based DED 113
t15 41.14 686.00 18.20

t16 39.23 720.00 17.83
t17 52.12 714.00 24.13
t18 40.85 761.00 18.80
t19 41.20 727.00 18.02
t20 41.15 714.00 17.03
t21 45.76 618.00 13.37
t22 45.59 584.00 16.94
t23 45.56 578.00 19.05
t24 34.72 544.00 13.58 ;

plant
EM Emission c a l c u l a t i o n
SR ( i ,t) Spinning reserve c a l c u l a t i o n ;
p . up ( i , t ) = gendata ( i , " Pmax " ) ;
p . lo ( i , t ) = gendata ( i , " Pmin " ) ;
SR . up ( i , t ) = gendata ( i , " Pmax " ) ;
SR . l o ( i , t ) = 0 ;
Equations
G e n c o n s t 3 , G e n c o n s t 4 , c o s t T h e r m a l c a l c , b a l a n c e , EMcalc , EMlim ,
benefitcalc , reserve ;
balance ( t ) . . sum ( i , p ( i , t ) ) = l = d a t a ( t , ’ l o a d ’ ) ;
( i , ’ e ’ ) p ( i , t ) + g e n d a t a ( i , ’ f ’ ) ) ;
benefitcalc .. OF=e=sum ( ( i , t ) ,1 d a t a ( t , ’ lambda ’ ) p ( i , t ) + d a t a
( t , ’ Lr ’ ) SR ( i , t ) )c o s t T h e r m a l ;
reserve ( i , t ) . . SR ( i , t ) = l = g e n d a t a ( i , " Pmax " )p ( i , t ) ;
Model DEDPB / a l l / ;
S o l v e DEDPB u s qcp max o f ;
e x e c u t e _ u n l o a d "DEDPB . gdx " P . l
e x e c u t e ’ gdxxrw . e x e DEDPB . gdx v a r =P r n g =P ! a1 ’
e x e c u t e _ u n l o a d "DEDPB . gdx " SR . l
e x e c u t e ’ gdxxrw . e x e DEDPB . gdx v a r =SR r n g =SR ! a1 ’
The energy and reserve schedules vs time in price-based DED with arbitrage are
shown in Fig. 4.9.
4.3 Linearized Cost-Based DED
In this section, we are trying to generate a linear version of the cost-based DED
problem formulated in (4.5). The assumption is that the fuel cost function is still
expressed in quadratic form aP2 C bP C c. The procedure for linearization is
illustrated in Fig. 4.10.
SR1 SR2 SR3 SR4

Spinning reserves (MW)
250
200
150
100
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Power schedules (MW)
80
P1
P2
60
P3
P4
40
20
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 4.9 The energy and reserve schedules vs time in price-based DED with arbitrage
Fig. 4.10 Linearizing the quadratic cost function

4.3 Linearized Cost-Based DED 115

As it is shown in Fig. 4.10, the interval between Pmin ; Pmax is divided into n
h i g g
k k
equally sized intervals Pg;ini ; Pg;fin . At a specific interval k, we define a variable pkg;t
which varies between zero and the interval length Pkg as indicated in (4.13b). The
actual value of power generation i at time t is calculated in (4.13f). The slope of the
line connecting the initial and final points of the interval k is calculated in (4.13h).
X
min OF D Cg;t (4.13a)
pkg;t g;t
0 pkg;t Pkg ; 8k D 1 W n (4.13b)

Pmax
g Pmin
g
Pkg D (4.13c)
n
Pkg;ini D .k 1/Pkg C Pmin
g (4.13d)
Pkg;fin D Pkg C Pkg;ini (4.13e)
X
Pg;t D Pmin
g C pkg;t (4.13f)
k
X
2
Cg;t D ag .Pmin min
g / C bg Pg C cg C skg pkg;t (4.13g)
k
k k
Cg;fin Cg;ini
skg D (4.13h)
Pkg
k
Cg;ini D ag .Pkg;ini /2 C bg Pkg;ini C cg (4.13i)
k
Cg;fin D ag .Pkg;fin /2 C bg Pkg;fin C cg (4.13j)
Pg;t Pg;t1 RUg (4.13k)
Pg;t1 Pg;t RDg (4.13l)
X
Pg;t Lt (4.13m)
g;t
The GAMS code for linear cost-based DED

with no arbitrage in example (4.13) is
provided in GCode 4.7. The Pmin g ; Pg
max
interval is divided into 100 sections. The
Pkg ; Pkg;fin ; Pkg;ini ; Cg;fin
k k
; Cg;ini and skg are treated as parameters.
GCode 4.7 The code for linear cost-based DED with no arbitrage in example (4.13)
i thermal units / g1g4 /
k /1100/;
g1 0 . 1 2 1 4 . 8 0 89 1 . 2 5 3 28 200 40 40
g2 0 . 1 7 1 6 . 5 7 83 2 . 3 4.24 6 . 0 9 20 290 30 30
g3 0 . 1 5 1 5 . 5 5 100 1 . 1 2.15 5 . 6 9 30 190 30 30
g4 0 . 1 9 1 6 . 2 1 70 1 . 1 3.99 6 . 2 20 260 50 50;
/ t1 510
t2 530
t3 516
t4 510
t5 515
t6 544
t7 646
t8 686
t9 741
t10 734
t11 748
t12 760
t13 754
t14 700
t15 686
t16 720
t17 714
t18 761
t19 727
t20 714
t21 618
t22 584
t23 578
t24 544/;
plant
Pk ( i , t , k ) ;
Parameter data (k , i ,) ;
d a t a ( k , i , ’DP ’ ) = ( g e n d a t a ( i , " Pmax " )g e n d a t a ( i , " Pmin " ) ) / c a r d ( k ) ;
d a t a ( k , i , ’ P i n i ’ ) = ( o r d ( k ) 1) d a t a ( k , i , ’DP ’ ) + g e n d a t a ( i , " Pmin " ) ;
d a t a ( k , i , ’ P f i n ’ ) = d a t a ( k , i , ’ P i n i ’ ) + d a t a ( k , i , ’DP ’ ) ;
d a t a ( k , i , ’ C i n i ’ ) = g e n d a t a ( i , " a " ) power ( d a t a ( k , i , ’ P i n i ’ ) , 2 )
+gendata ( i , " b ")data (k , i , ’ Pini ’ )+gendata ( i , " c ") ;
d a t a ( k , i , ’ C f i n ’ ) = g e n d a t a ( i , " a " ) power ( d a t a ( k , i , ’ P f i n ’ ) , 2 )
+gendata ( i , " b ")data (k , i , ’ Pfin ’ )+gendata ( i , " c ") ;
d a t a ( k , i , ’ s ’ ) = ( d a t a ( k , i , ’ C f i n ’ )d a t a ( k , i , ’ C i n i ’ ) ) / d a t a ( k , i , ’DP ’ ) ;
Pk . up ( i , t , k ) = d a t a ( k , i , ’DP ’ ) ;
Pk . l o ( i , t , k ) = 0 ;
E q u a t i o n s eq1 , eq2 , eq3 , eq4 , eq5 ;
Nomenclature 117
eq1 . . OF =e=sum ( ( t , i ) , g e n d a t a ( i , ’ a ’ ) power ( g e n d a t a ( i , "

Pmin " ) , 2 )
+ g e n d a t a ( i , ’ b ’ ) g e n d a t a ( i , " Pmin " ) + g e n d a t a ( i , ’ c ’ )
+ sum ( k , d a t a ( k , i , ’ s ’ ) pk ( i , t , k ) ) );
eq2 ( i , t ) . . p ( i , t + 1 )p ( i , t ) = l = g e n d a t a ( i , ’RU0 ’ ) ;
eq3 ( i , t ) . . p ( i , t 1)p ( i , t ) = l = g e n d a t a ( i , ’RD0 ’ ) ;
eq4 ( t ) . . sum ( i , p ( i , t ) ) =g=demand ( t ) ;
eq5 ( i , t ) . . p ( i , t ) =e= g e n d a t a ( i , " Pmin " ) + sum ( k , Pk ( i , t , k ) ) ;
Model DEDLP / a l l / ;
S o l v e DEDLP u s LP min OF ;
The total cost is $647,972.4 and compared to the operating costs obtained by
GCode 4.1 which gives $647,960.1, the error is not significant. If more precision
is needed, the number of intervals (k) should be increased. This would increase the
number of variables in optimization problem and makes it more difficult to solve.
4.4 Applications
There are still some other technical constraints that should be considered in DED
problem formulation such as valve-point effect, prohibited operation zones (POZs),
and transmission losses. References [3] propose the imperialist competitive algo-
rithm (ICA) to solve such complicated problem. Stochastic Real-Time Scheduling
of Wind-Thermal Generation Units in an Electric Utility [4].
Nomenclature
Indices
k Blocks considered for piecewise linear fuel cost function.
i Thermal generating units.
t Time intervals.
Parameters
U=S0g Duration of period that unit i has been on/off at the beginning of the
operating horizon (end of t D 0) (h).
dg ; eg ; fg Emission coefficients of unit i.
et Electric energy price at time t ($/MW h).
Lt Electric demand at time t.
ag ; bg ; cg Fuel cost coefficients of unit i.
Pkg;ini=fin Initial and final values of power in block k of linearized cost of
thermal unit i (MW).
k
Cg;ini=fin Initial and final values of cost in block k of linearized cost of thermal
unit i ($/h).
Pkg Length of block k of linearized cost of thermal unit i (MW).

max=min
Pg Maximum/minimum limits of power generation of thermal unit i.
DT=UTg Minimum down/up time of unit i (h).
T Number of time intervals.
n Number of blocks considered for piecewise linear fuel cost function.
Parameter used for multi-objective optimization.
RUg =RDg Ramp-up/down limit of generation unit i (MW/h).
rt Reserve price at time t ($/MW).
SDCg =SUCg Shut down/Start-up cost of unit i ($/h).
SD=SUg Shut-down/start-up ramp limit of unit i (MW/h).
skg Slope of cost in block k of linearized cost of thermal unit i ($/MW).
VWC Value of wind curtailment
wt Wind availability at time t
Variables
Pwt Wind generation at time t
Pwc
t Wind curtailment at time t
Cg;t Fuel cost of thermal unit i at time t ($).
pkg;t Operating schedule of thermal unit i in block k at time t (MW).
Pg;t Power generated by thermal unit i at time t (MW).
SRg;t Spinning reserve provided by thermal unit i at time t (MW).
TC Total operating costs ($).
EM Total emissions (kg).
References
1. A. Soroudi, M. Afrasiab, Binary PSO-based dynamic multi-objective model for distributed

generation planning under uncertainty. IET Renew. Power Gener. 6(2), 67–78 (2012)
2. A.J. Conejo, M.A. Plazas, R. Espinola, A.B. Molina, Day-ahead electricity price forecasting
using the wavelet transform and ARIMA models. IEEE Trans. Power Syst. 20(2), 1035–1042
(2005)
3. R. Roche, L. Idoumghar, B. Blunier, A. Miraoui, Imperialist competitive algorithm for dynamic
optimization of economic dispatch in power systems, in International Conference on Artificial
Evolution (Evolution Artificielle) (Springer, Berlin, 2011), pp. 217–228
4. A. Soroudi, A. Rabiee, A. Keane, Stochastic real-time scheduling of wind-thermal generation
units in an electric utility. IEEE Syst. J. PP(99), 1–10 (2015)
Chapter 5
Unit Commitment
This chapter provides GAMS code for solving unit commitment (UC) problem. The
developed GAMS code is linear and is categorized as a MIP model in GAMS. The
inputs are generator’s characteristics, electricity prices, and demands. The outputs
of this code are on/off status of units and their operating schedules.
Every unit commitment problem has three main cost components namely fuel
costs, start-up, and shut-down costs. The unit commitment cost calculation is
illustrated in Fig. 5.1.
The unit commitment data for ten units are inspired by Ademovic et al. [1] and
described in Table 5.1.
5.1 Cost-Based Unit Commitment
5.1.1 Cost Calculation
X
min OF D FCi;t C STCi;t C SDCi;t (5.1)
pki;t ;ui;t ;yi;t ;zi;t
i;t
The linear version of fuel cost calculation is described in (5.2).
0 pki;t Pki ui;t ; 8k D 1 W n (5.2a)

Pmax
i Pmin
i
Pki D (5.2b)
n
Pki;ini D .k 1/Pki C Pmin
i (5.2c)
Pki;fin D Pki C Pki;ini (5.2d)

DOI 10.1007/978-3-319-62350-4_5
120 5 Unit Commitment
Fig. 5.1 Unit commitment cost calculation
X
Pi;t D Pmin
i ui;t C pki;t (5.2e)
k
k
2
Ci;ini D ai Pki;ini C bi Pki;ini C ci (5.2f)
k
2
Ci;fin D ai Pki;fin C bi Pki;fin C ci (5.2g)
k k
Ci;fin Ci;ini
ski D (5.2h)
Pki
2 X
FCi;t D ai Pmin
i C bi Pmin
i C ci ui;t C ski pki;t (5.2i)
k
ui;t is the on/off status of the unit i at time t.
5.1.2 Ramp Rate Constraints
The power generation of unit i at time t should be within the operating limits as given
in (5.3a). Pi;t and PN i;t state the minimum and maximum time-dependent operating
limits. These are not necessarily equal to Pmin i and Pmax
i , respectively.
The upper operating limit (PN i;t ) is described in (5.3b) and (5.3c) according to [2].
The ramp up/down constraints are modeled as follows:
Pi;t Pi;t PN i;t (5.3a)

PN i;t Pmax
i Œui;t zi;tC1 C SDi zi;tC1 (5.3b)
PN i;t Pi;t1 C RUi ui;t1 C SUi yi;t (5.3c)
Pi;t Pmin
i;t ui;t (5.3d)
Pi;t Pi;t1 RDi ui;t SDi zi;t (5.3e)
Table 5.1 Unit commitment data for ten thermal units
ai bi RUi RDi SDi SUi Pmin
i Pmax
i
Unit ($=MW2 ) ($=MW) ci ($) Cdi ($) Csi ($) (MW h1 ) (MW h1 ) UTi (h) DTi (h) (MW h1 ) (MW h1 ) (MW) P.MW/ Ui0 (h) ui;tD0 Si0 (h)
5.1 Cost-Based Unit Commitment
g1 0.0148 12.1 82 42.6 42.6 40 40 3 2 90 110 80 200 1 0 1

g2 0.0289 12.6 49 50.6 50.6 64 64 4 2 130 140 120 320 2 0 0
g3 0.0135 13.2 100 57.1 57.1 30 30 3 2 70 80 50 150 3 0 3
g4 0.0127 13.9 105 47.1 47.9 104 104 5 3 240 250 250 520 1 1 0
g5 0.0261 13.5 72 56.6 56.9 56 56 4 2 110 130 80 280 1 1 0
g6 0.0212 15.4 29 141.5 141.5 30 30 3 2 60 80 50 150 0 0 0
g7 0.0382 14 32 113.5 113.5 24 24 3 2 50 60 30 120 0 1 0
g8 0.0393 13.5 40 42.6 42.6 22 22 3 2 45 55 30 110 0 0 0
g9 0.0396 15 25 50.6 50.6 16 16 0 0 35 45 20 80 0 0 0
g10 0.0510 14.3 15 57.1 57.1 12 12 0 0 30 40 20 60 0 0 0
121
In order to explain the upper operating limits described in (5.3) some assumptions
should be taken into account:
• It should be always less than capacity of unit i, this means that PN i;t Pmaxi .
• In case of unit shut-down in the next hour (t C 1): PN i;t SDi zi;tC1 . Since Pi;tC1 D
0 so Pi;t cannot be more than SDi .
• If the unit has been on in the previous hour (ui;t1 D 1) and is going to remain
on then Pi;t cannot be increased more than RUi . This means that PN i;t Pi;t1 C
RUi ui;t1 .
• If the unit has been off in the previous hour (ui;t1 D 0) and it is turned on at time
t (yi;t D 1) then Pi;t cannot be more than SUi . This means that PN i;t SUi yi;t .
The combination of all these cases is enforced by (5.3b) and (5.3c).
In order to explain the lower operating limits described in (5.3) some assumptions
should be taken into account:
• If the unit is on at time t then the generated power should be greater than Pmin
i;t ui;t .
• If the unit is on at time t1 and remains to be on at time t; tC1 then the generated
power at time t should be greater than Pi;t1 RDi ui;t .
• If the unit is on at time t 1 and turned off at time t then the generated power at
time t 1 should be Pi;t1 SDi zi;t .
The combination of all these cases is enforced by (5.3d) and (5.3e).
5.1.3 Min Up/Down Time Constraints
The on/off states of unit i at time t are described by ui;t . Additionally, start-up/shut-
down are given by yi;t ; zi;t in (5.4).
yi;t zi;t D ui;t ui;t1 (5.4a)

yi;t C zi;t 1 (5.4b)
yi;t ; zi;t ; ui;t 2 f0; 1g
The minimum up time (UTi ) of unit i is modeled in (5.5) as proposed in [2].

i
X
1 ui;t D 0 (5.5a)
tD1
Xi 1
kCUT
ui;t UTi yi;k ; 8k D
i C 1 : : : T UTi C 1 (5.5b)
tDk
T
X
ui;t yi;t 0; 8k D T UTi C 2 : : : T (5.5c)
tDk
˚

i D min T; UTi Ui0 ui;tD0 (5.5d)
5.1 Cost-Based Unit Commitment 123
The minimum up time (DTi ) of unit i is modeled in (5.6) as proposed in [2].
i
X
ui;t D 0 (5.6a)
tD1
Xi 1
kCDT
1 ui;t DTi zi;k ; 8k D i C 1 : : : T DTi C 1 (5.6b)
tDk
T
X
1 ui;t zi;t 0; 8k D T DTi C 2 : : : T (5.6c)
tDk
˚
i D min T; DTi Si0 Œ1 ui;tD0 (5.6d)
The realistic start-up and shut-down costs depend on how long the unit has been
off or on, receptively. Some references have provided the detailed formulation for
modeling these cost terms. Here for simplicity, the start-up (Sdi ) and shut-down
(Sdi ) costs are modeled as constant values as (5.7).
STCi;t D Csi yi;t (5.7a)

SDCi;t D Sdi zi;t (5.7b)
5.1.4 Demand-Generation Balance
In cost-based UC, the hourly total generation should be equal to the hourly demand:
X
Pi;t Lt (5.8)
i
The hourly demand and price values vs time are depicted in Fig. 5.2.
The GAMS code for solving the cost-based unit commitment is provided in
GCode 5.1.
GCode 5.1 Cost-based unit commitment example for ten unit system
Sets t t i m e / t 1t 2 4 / ,
i g e n e r a t o r s / g1g10 / ,
k c o s t s e g m e n t s / s g 1s g 2 0 / ,
c h a r / ch1ch2 / ;
Alias ( t , h) ;
a b c CostsD c o s t s t RU RD UT DT SD SU Pmin Pmax U0 U i n i S0
g1 0.014 12.1 82 42.6 42.6 40 40 3 2 90 110 80 200 1 0 1

g2 0.028 12.6 49 50.6 50.6 64 64 4 2 130 140 120 320 2 0 0
g3 0.013 13.2 100 57.1 57.1 30 30 3 2 70 80 50 150 3 0 3
g4 0.012 13.9 105 47.1 47.9 104 104 5 3 240 250 250 520 1 1 0
g5 0.026 13.5 72 56.6 56.9 56 56 4 2 110 130 80 280 1 1 0
g6 0 . 0 2 1 1 5 . 4 29 141.5 141.5 30 30 3 2 60 80 50 150 0 0 0

g7 0 . 0 3 8 1 4 . 0 32 113.5 113.5 24 24 3 2 50 60 30 120 0 1 0
g8 0 . 0 3 9 1 3 . 5 40 42.6 42.6 22 22 3 2 45 55 30 110 0 0 0
g9 0 . 0 3 9 1 5 . 0 25 50.6 50.6 16 16 0 0 35 45 20 80 0 0 0
g10 0 . 0 5 1 1 4 . 3 15 57.1 57.1 12 12 0 0 30 40 20 60 0 0 0;
Parameter data ( k , i ,) ;
d a t a ( k , i , ’ P i n i ’ ) = ( o r d ( k ) 1)d a t a ( k , i , ’DP ’ ) + g e n d a t a ( i , " Pmin " ) ;
d a t a ( k , i , ’ C i n i ’ ) = g e n d a t a ( i , " a " )power ( d a t a ( k , i , ’ P i n i ’ ) , 2 )
+ g e n d a t a ( i , " b " )d a t a ( k , i , ’ P i n i ’ ) + g e n d a t a ( i , " c " ) ;
d a t a ( k , i , ’ C f i n ’ ) = g e n d a t a ( i , " a " )power ( d a t a ( k , i , ’ P f i n ’ ) , 2 )
+ g e n d a t a ( i , " b " )d a t a ( k , i , ’ P f i n ’ ) + g e n d a t a ( i , " c " ) ;
g e n d a t a ( i , ’ M i n c o s t ’ ) = g e n d a t a ( i , ’ a ’ )power ( g e n d a t a ( i , " Pmin " ) , 2 )
+ g e n d a t a ( i , ’ b ’ )g e n d a t a ( i , " Pmin " ) + g e n d a t a ( i , ’ c ’ ) ;
Table dataLP ( t ,)
lambda load
t1 14.72 883
t2 15.62 915
t23 20.50 915
t24 15.62 834;
Parameter unit ( i , char ) ;
u n i t ( i , ’ ch1 ’ ) = 2 4 ;
u n i t ( i , ’ ch2 ’ ) = ( g e n d a t a ( i , ’UT ’ )g e n d a t a ( i , ’U0 ’ ) )g e n d a t a ( i , ’ U i n i ’ ) ;
p a r a m e t e r u n i t 2 ( i , c h a r ) ; u n i t 2 ( i , ’ ch1 ’ ) = 2 4 ;
u n i t 2 ( i , ’ ch2 ’ ) = ( g e n d a t a ( i , ’DT ’ )g e n d a t a ( i , ’ S0 ’ ) ) (1 g e n d a t a ( i , ’ U i n i ’ ) ) ;
g e n d a t a ( i , ’ L j ’ ) = smin ( c h a r , u n i t ( i , c h a r ) ) ; g e n d a t a ( i , ’ F j ’ ) = smin ( c h a r , u n i t 2 ( i , c h a r ) ) ;
v a r i a b l e c o s t T h e r m a l ; p o s i t i v e v a r i a b l e s pu ( i , t ) , p ( i , t ) , StC ( i , t ) ,SDC( i , t ) , Pk ( i , t , k ) ;
Binary v a r i a b l e u ( i , t ) , y ( i , t ) , z ( i , t ) ;
p . up ( i , t ) = g e n d a t a ( i , " Pmax " ) ; p . l o ( i , t ) = 0 ; Pk . up ( i , t , k ) = d a t a ( k , i , ’DP ’ ) ;
Pk . l o ( i , t , k ) = 0 ; p . up ( i , t ) = g e n d a t a ( i , " Pmax " ) ; pu . up ( i , h ) = g e n d a t a ( i , " Pmax " ) ;
E q u a t i o n s Uptime1 , Uptime2 , Uptime3 , Dntime1 , Dntime2 , Dntime3 , Ramp1 , Ramp2 ,
Ramp3 , Ramp4 , s t a r t c , s h t d n c , g e n c o n s t 1 , G e n c o n s t 2 , G e n c o n s t 3 , G e n c o n s t 4 , b a l a n c e ;
Uptime1 ( i ) $ ( g e n d a t a ( i , " L j " ) >0) . .
sum ( t $ ( o r d ( t ) < ( g e n d a t a ( i , " L j " ) + 1 ) ) ,1U( i , t ) ) =e = 0 ;
Uptime2 ( i ) $ ( g e n d a t a ( i , " UT " ) >1) . .
sum ( t $ ( o r d ( t ) >24 g e n d a t a ( i , " UT " ) + 1 ) ,U( i , t )y ( i , t ) ) =g = 0 ;
Uptime3 ( i , t ) $ ( o r d ( t ) > g e n d a t a ( i , " L j " ) and o r d ( t ) <24 g e n d a t a ( i , " UT " ) +2 and
n o t ( g e n d a t a ( i , " L j " ) >24 g e n d a t a ( i , " UT " ) ) ) . . sum ( h$ ( ( o r d ( h ) > o r d ( t ) 1) and
( o r d ( h ) < o r d ( t ) + g e n d a t a ( i , " UT " ) ) ) ,U( i , h ) ) =g= g e n d a t a ( i , " UT " )y ( i , t ) ;
Dntime1 ( i ) $ ( g e n d a t a ( i , " F j " ) >0) . . sum ( t $ ( o r d ( t ) < ( g e n d a t a ( i , " F j " ) + 1 ) ) ,U( i , t ) ) =e = 0 ;
Dntime2 ( i ) $ ( g e n d a t a ( i , " DT " ) >1) . .
sum ( t $ ( o r d ( t ) >24 g e n d a t a ( i , " DT " ) + 1 ) ,1U( i , t )z ( i , t ) ) =g = 0 ;
Dntime3 ( i , t ) $ ( o r d ( t ) > g e n d a t a ( i , " F j " ) and o r d ( t ) <24 g e n d a t a ( i , " DT " ) +2
and n o t ( g e n d a t a ( i , " F j " ) >24 g e n d a t a ( i , " DT " ) ) ) . . sum ( h$ ( ( o r d ( h ) > o r d ( t ) 1)
and ( o r d ( h ) < o r d ( t ) + g e n d a t a ( i , " DT " ) ) ) ,1U( i , h ) ) =g= g e n d a t a ( i , " DT " )z ( i , t ) ;
s t a r t c ( i , t ) . . StC ( i , t ) =g= g e n d a t a ( i , " c o s t s t " )y ( i , t ) ;
s h t d n c ( i , t ) . . SDC( i , t ) =g= g e n d a t a ( i , " CostsD " )z ( i , t ) ;
Genconst1 ( i , h ) . . p ( i , h ) =e=u ( i , h )g e n d a t a ( i , " Pmin " ) +sum ( k , Pk ( i , h , k ) ) ;
G e n c o n s t 2 ( i , h ) $ ( o r d ( h ) >0) . . U( i , h ) =e=U( i , h1)$ ( o r d ( h ) >1)+ g e n d a t a ( i , " U i n i " ) $ ( o r d ( h )
=1)
+y ( i , h )z ( i , h ) ;
Genconst3 ( i , t , k ) . . Pk ( i , t , k ) = l =U( i , t )d a t a ( k , i , ’DP ’ ) ;
G e n c o n s t 4 . . c o s t T h e r m a l =e=sum ( ( i , t ) , StC ( i , t ) +SDC( i , t ) ) +sum ( ( t , i ) ,
u ( i , t )g e n d a t a ( i , ’ M i n c o s t ’ ) +sum ( k , d a t a ( k , i , ’ s ’ )pk ( i , t , k ) ) ) ;
Ramp1 ( i , t ) . . p ( i , t 1)p ( i , t ) = l =U( i , t )g e n d a t a ( i , ’RD ’ ) +z ( i , t )g e n d a t a ( i , " SD " ) ;
Ramp2 ( i , t ) . . p ( i , t ) = l =pu ( i , t ) ;
Ramp3 ( i , t ) $ ( o r d ( t ) <24) . . pu ( i , t ) = l = ( u ( i , t )z ( i , t + 1 ) )g e n d a t a ( i , " Pmax " )
+z ( i , t +1)g e n d a t a ( i , " SD " ) ;
Ramp4 ( i , t ) $ ( o r d ( t ) >1) . . pu ( i , t ) = l =p ( i , t 1)+U( i , t 1)g e n d a t a ( i , ’RU ’ )
+y ( i , t )g e n d a t a ( i , " SU " ) ;
Balance ( t ) . . sum ( i , p ( i , t ) ) =e= d a t a L P ( t , ’ l o a d ’ ) ;
Model UCLP / a l l / ;
Option o p t c r =0 .0 ;
S o l v e UCLP m i n i m i z i n g c o s t T h e r m a l u s i n g mip ;
5.2 Cost-Based UC with Additional Constraints 125
Electricity prices ($/MWh)

25
23
21
19
17
15
13
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Electric demand (MW)
1400
1200
1000
800
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 5.2 Hourly demand and price values vs time
P1 P2 P3 P4 P5 P6 P7 P8 P9 P10
300
250
200
150
100
50
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 5.3 Power schedules in cost-based unit commitment vs time
The total operating costs are $485,240.189. The start-up, shut-down, and fuel
costs are $442.100, $0, and $484,798.089, respectively. The power schedules of
thermal units in cost-based unit commitment are depicted in Fig. 5.3.
5.2 Cost-Based UC with Additional Constraints
In order to improve the security and efficiency of the energy supply, some
additional constraints are needed to be considered in UC formulation. Some of these
constraints are discussed and formulated in this section.
5.2.1 Cost-Based UC with Reserve Constraints
One of the most important resources which is used by the system operator is called
spinning reserve (SR) [3]. It is used to cover the sudden increase in demand, rapid
reduction of renewable energy production, or unplanned generating unit outage. The
SR is supplied by online generating units which are synchronized to the system
and are able to ramp-up in order to meet the demand. The UC-reserve constrained
formulation is provided as follows:
X
min OF D FCi;t C STCi;t C SDCi;t (5.9a)
i;t
X
Pi;t Lt (5.9b)
i
Ri;t PN i;t Pi;t (5.9c)

X
Ri;t Lt (5.9d)
i
Subject to:
(5.2), (5.3), (5.4), (5.5), (5.6), (5.7)
where Ri;t is the reserve provided by online unit i at time t. is the percentage
of demand which specifies the reserve requirement. It is usually expressed as a
percentage of demand at time t. The GAMS code for solving the (5.9) is described in
GCode 5.2. It should be noted that the generating unit characteristics are unchanged
but the hourly demand values specified in GCode 5.1 and Fig. 5.2 are reduced by
55%. The value is specified by the system operator and is dependent on the
system specification. Here, it is assumed to be 40% for simulation purpose. The
total operating costs without the reserve constraint is $2:1039 105 . However, if
the reserve constraint is taken into account then the operating costs would increase
to $2:1090 105 . The operating schedules of thermal units considering reserve
constraint are provided in Table 5.2.
GCode 5.2 Cost-based unit commitment with reserve constraint

S e t s t / t 1t 2 4 / , i / g1g10 / , k / s g 1s g 2 0 / , c h a r / ch1ch2 / ; a l i a s ( t , h ) ; a l i a s ( i , g ) ;
Removed f o r s a v i n g s p a c e ;

t a b l e dataLP ( t ,)
lambda load
dataLP ( t , ’ load ’ ) =dataLP ( t , ’ load ’ ) 0.45; Parameter u n i t ( i , char ) ;
u n i t ( i , ’ ch1 ’ ) = 2 4 ; u n i t ( i , ’ ch2 ’ ) = ( g e n d a t a ( i , ’UT ’ )g e n d a t a ( i , ’U0 ’ ) )g e n d a t a ( i , ’ U i n i ’ )
;
P a r a m e t e r u n i t 2 ( i , c h a r ) ; u n i t 2 ( i , ’ ch1 ’ ) = 2 4 ;
Variable costThermal ; Binary v a r i a b l e u ( i , t ) , y ( i , t ) , z ( i , t ) ;
P o s i t i v e v a r i a b l e s pu ( i , t ) , p ( i , t ) , StC ( i , t ) ,SDC( i , t ) , Pk ( i , t , k ) ;
p . up ( i , t ) = g e n d a t a ( i , " Pmax " ) ; p . l o ( i , t ) = 0 ;
Pk . up ( i , t , k ) = d a t a ( k , i , ’DP ’ ) ; Pk . l o ( i , t , k ) = 0 ;
p . up ( i , t ) = g e n d a t a ( i , " Pmax " ) ; pu . up ( i , h ) = g e n d a t a ( i , " Pmax " ) ;
E q u a t i o n s Uptime1 , Uptime2 , Uptime3 , Dntime1 , Dntime2 , Dntime3 ,
Ramp1 , Ramp2 , Ramp3 , Ramp4 , s t a r t c , s h t d n c , g e n c o n s t 1 ,
Genconst2 , Genconst3 , Genconst4 , balance , r e s e r v e ;
Uptime1 ( i ) $ ( g e n d a t a ( i , " L j " ) >0) . .
sum ( t $ ( o r d ( t ) < ( g e n d a t a ( i , " L j " ) + 1 ) ) ,1U( i , t ) ) =e = 0 ;
Uptime3 ( i , t ) $ ( o r d ( t ) > g e n d a t a ( i , " L j " ) and o r d ( t ) <24 g e n d a t a ( i , " UT " ) +2 and
n o t ( g e n d a t a ( i , " L j " ) >24 g e n d a t a ( i , " UT " ) ) )
. . sum ( h$ ( ( o r d ( h ) > o r d ( t ) 1) and ( o r d ( h ) < o r d ( t ) + g e n d a t a ( i , " UT " ) ) ) ,U( i , h ) ) =g=
g e n d a t a ( i , " UT " )y ( i , t ) ;
Dntime1 ( i ) $ ( g e n d a t a ( i , " F j " ) >0) . . sum ( t $ ( o r d ( t ) < ( g e n d a t a ( i , " F j " ) + 1 ) ) ,U( i , t ) ) =e
=0;
sum ( t $ ( o r d ( t ) >24 g e n d a t a ( i , " DT " ) + 1 ) ,1U( i , t )z ( i , t ) ) =g
=0;
Dntime3 ( i , t ) $ ( o r d ( t ) > g e n d a t a ( i , " F j " ) and o r d ( t ) <24 g e n d a t a ( i , " DT " ) +2 and
n o t ( g e n d a t a ( i , " F j " ) >24 g e n d a t a ( i , " DT " ) ) )
. . sum ( h$ ( ( o r d ( h ) > o r d ( t ) 1) and ( o r d ( h ) < o r d ( t ) + g e n d a t a ( i , " DT " ) ) ) ,1U( i , h ) ) =g=
g e n d a t a ( i , " DT " )z ( i , t ) ;
g e n c o n s t 1 ( i , h ) . . p ( i , h ) =e=u ( i , h )g e n d a t a ( i , " Pmin " ) +sum ( k , Pk ( i , h , k ) ) ;
G e n c o n s t 2 ( i , h ) $ ( o r d ( h ) >0) . . U( i , h ) =e=
U( i , h1)$ ( o r d ( h ) >1)+ g e n d a t a ( i , " U i n i " ) $ ( o r d ( h ) = 1 ) +y ( i , h )z ( i , h ) ;
G e n c o n s t 3 ( i , t , k ) . . Pk ( i , t , k ) = l =U( i , t )d a t a ( k , i , ’DP ’ ) ;
Genconst4 . . c o s t T h e r m a l =e=sum ( ( i , t ) , StC ( i , t ) +SDC( i , t ) ) +sum ( ( t , i ) ,
u ( i , t )g e n d a t a ( i , ’ M i n c o s t ’ ) +sum ( k , d a t a ( k , i , ’ s ’ )pk ( i , t , k ) ) ) ;
Ramp2 ( i , t ) . . p ( i , t ) = l =pu ( i , t ) ;
+z ( i , t +1)g e n d a t a ( i , " SD " ) ;
+y ( i , t )g e n d a t a ( i , " SU " ) ;
b a l a n c e ( t ) . . sum ( i , p ( i , t ) ) =e= d a t a L P ( t , ’ l o a d ’ ) ;
r e s e r v e ( t ) . . sum ( i , pu ( i , t )p ( i , t ) ) =g =0.40 d a t a L P ( t , ’ l o a d ’ ) ;
Model UCLP / a l l / ; O p t i o n o p t c r = 0 . 0 ;
Table 5.2 Power schedules (Pi;t ) of thermal units considering reserve constraint
Time g1 g3 g4 g5 g7 g8 g10
t1 250.0 90.4 57.0
t2 250.0 107.8 54.0
t3 94.5 250.0 80.0 30.0
t4 92.1 65.0 250.0 80.0 30.0
t5 104.0 72.2 250.0 80.0 30.0 20.0
t6 110.0 79.0 250.0 80.0 30.0 30.0 20.0
t7 123.7 95.0 250.0 80.0 30.0 30.0 20.0
t8 128.0 100.6 250.0 80.0 30.0 30.0 20.0
t9 134.0 106.8 250.0 80.0 30.0 34.0 20.0
t10 134.0 106.8 250.0 80.0 30.0 34.0 20.0
t11 133.5 105.0 250.0 80.0 30.0 30.0 20.0
t12 134.6 110.0 250.0 80.0 30.0 34.0
t13 133.7 105.0 250.0 80.0 30.0 30.0
t14 122.0 90.6 250.0 80.0 30.0 30.0
t15 128.0 97.6 250.0 80.0 30.0 30.0
t16 122.0 90.6 250.0 80.0 30.0 30.0
t17 98.0 68.2 250.0 80.0 30.0 30.0
t18 110.0 77.3 250.0 30.0 30.0
t19 104.2 52.9 250.0 30.0 30.0
t20 121.6 250.0 30.0 30.0
t21 104.9 250.0 30.0 30.0
t22 88.3 250.0 30.0 30.0
t23 101.8 250.0 30.0 30.0
t24 95.3 250.0 30.0
The reserve provision (Ri;t ) by thermal units considering reserve constraint is

given in Table 5.3.
5.2.2 Cost-Based UC Considering Generator Contingency
In this section, the cost-based UC is formulated in order to consider the generator

outage contingencies. The system operator is willing to be robust against the
unexpected outages of generating units. It means that in case of any generating unit
outage the reserve resource by the remaining units should be able to supply the lost
generating unit. This constraint is modeled as follows:
X
i;t
Table 5.3 Reserve provision (Ri;t ) by thermal units considering reserve constraint
t1 270.0 189.7 63.0
t2 99.1 38.6 27.0
t3 15.5 34.6 83.8 48.0
t4 42.5 15.0 69.4 56.0 24.0
t5 28.1 22.8 71.6 56.0 24.0 20.0
t6 34.0 104.0 56.0 8.6 25.0 12.0
t7 26.4 14.0 97.2 56.0 24.0 22.0 12.0
t8 35.7 24.5 81.3 56.0 24.0 22.0 12.0
t9 34.0 23.8 94.1 56.0 24.0 18.0 12.0
t10 40.0 30.0 77.9 56.0 24.0 22.0 12.0
t11 40.6 31.8 71.1 56.0 24.0 26.0 10.0
t12 38.9 25.0 93.5 56.0 24.0 18.0
t13 40.9 35.0 69.6 56.0 24.0 26.0
t14 51.7 44.5 42.9 56.0 24.0 22.0
t15 34.0 23.0 87.3 56.0 24.0 22.0
t16 46.0 37.1 56.0 56.0 24.0 22.0
t17 42.5 104.0 30.0 24.0 22.0
t18 28.0 21.0 104.0 24.0 22.0
t19 45.8 17.1 77.9 24.0 22.0
t20 22.6 104.0 24.0 22.0
t21 56.7 63.3 24.0 22.0
t22 56.7 56.7 24.0 22.0
t23 21.7 104.0 24.0 15.0
t24 98.1 52.0
X
Pi;t Lt (5.10b)
i
Ri;t PN i;t Pi;t (5.10c)

X
Ri;t Pi0 ;t 8i0 2 ˝c (5.10d)
i¤i0
Subject to:
(5.2), (5.3), (5.4), (5.5), (5.6), (5.7)
where ˝c is the set of contingencies for generating units. The GAMS code
for solving the (5.10) is described in GCode 5.3. As previously stated, the total
operating costs without the generator contingency constraint are $2:1039 105 .
However, if the generator contingency constraint (the outage of all generating units
except g4 is considered) is taken into account then the operating costs would
increase to $2:1041 105 .
GCode 5.3 Cost-based unit commitment considering generator contingencies

Sets t t i m e / t 1t 2 4 /
i g e n e r a t o r s i n d i c e s / g1g10 / , k c o s t s e g m e n t s / s g 1s g 2 0 /
c h a r / ch1ch2 / , g ( i ) / g1g3 , g5g10 / ; A l i a s ( t , h ) ;
parameter data ( k , i ,) ;
Table dataLP ( t ,)
lambda load
dataLP ( t , ’ load ’ ) =dataLP ( t , ’ load ’ ) 0.45;
P a r a m e t e r u n i t ( i , c h a r ) ; u n i t ( i , ’ ch1 ’ ) = 2 4 ;
u n i t ( i , ’ ch2 ’ ) = ( g e n d a t a ( i , ’UT ’ )g e n d a t a ( i , ’U0 ’ ) )g e n d a t a ( i , ’ U i n i ’ ) ;
p a r a m e t e r u n i t 2 ( i , c h a r ) ; u n i t 2 ( i , ’ ch1 ’ ) = 2 4 ;
Variable costThermal ;
p o s i t i v e v a r i a b l e s pu ( i , t ) , p ( i , t ) , StC ( i , t ) ,SDC( i , t ) , Pk ( i , t , k ) ;
p . up ( i , t ) = g e n d a t a ( i , " Pmax " ) ; p . l o ( i , t ) = 0 ;
E q u a t i o n s Uptime1 , Uptime2 , Uptime3 , Dntime1 , Dntime2 , Dntime3 , Ramp1 , Ramp2 , Ramp3 , Ramp4
, s t a r t c , shtdnc , genconst1 , Genconst2 , Genconst3 , Genconst4 , balance , r e s e r v e ;
Uptime1 ( i ) $ ( g e n d a t a ( i , " L j " ) >0) . . sum ( t $ ( o r d ( t ) < ( g e n d a t a ( i , " L j " ) + 1 ) ) ,1U( i , t ) ) =e = 0 ;
Uptime2 ( i ) $ ( g e n d a t a ( i , " UT " ) >1)
. . sum ( t $ ( o r d ( t ) >24 g e n d a t a ( i , " UT " ) + 1 ) ,U( i , t )y ( i , t ) ) =g = 0 ;
Uptime3 ( i , t ) $ ( o r d ( t ) > g e n d a t a ( i , " L j " ) and o r d ( t ) <24 g e n d a t a ( i , " UT " ) +2
and n o t ( g e n d a t a ( i , " L j " ) >24 g e n d a t a ( i , " UT " ) ) ) . . sum ( h$ ( ( o r d ( h ) > o r d ( t ) 1) and
Dntime1 ( i ) $ ( g e n d a t a ( i , " F j " ) >0) . . sum ( t $ ( o r d ( t ) <( g e n d a t a ( i , " F j " ) + 1 ) ) ,U( i , t ) ) =e = 0 ;
Dntime2 ( i ) $ ( g e n d a t a ( i , " DT " ) >1)
. . sum ( t $ ( o r d ( t ) >24 g e n d a t a ( i , " DT " ) + 1 ) ,1U( i , t )z ( i , t ) ) =g = 0 ;
n o t ( g e n d a t a ( i , " F j " ) >24 g e n d a t a ( i , " DT " ) ) ) . . sum ( h$ ( ( o r d ( h ) > o r d ( t ) 1) and
( o r d ( h ) < o r d ( t ) + g e n d a t a ( i , " DT " ) ) ) ,1U( i , h ) ) =g= g e n d a t a ( i , " DT " )z ( i , t ) ;
g e n c o n s t 1 ( i , h ) . . p ( i , h ) =e=u ( i , h )g e n d a t a ( i , " Pmin " ) +sum ( k , Pk ( i , h , k ) ) ;
G e n c o n s t 2 ( i , h ) $ ( o r d ( h ) >0) . . U( i , h ) =e=U( i , h1)$ ( o r d ( h ) >1)
+ g e n d a t a ( i , " U i n i " ) $ ( o r d ( h ) = 1 ) +y ( i , h )z ( i , h ) ;
G e n c o n s t 4 . . c o s t T h e r m a l =e=sum ( ( i , t ) , StC ( i , t ) +SDC( i , t ) )
+sum ( ( t , i ) , u ( i , t )g e n d a t a ( i , ’ M i n c o s t ’ ) +sum ( k , d a t a ( k , i , ’ s ’ )pk ( i , t , k ) ) ) ;
Ramp2 ( i , t ) . . p ( i , t ) = l =pu ( i , t ) ;
+z ( i , t +1)g e n d a t a ( i , " SD " ) ;
+y ( i , t )g e n d a t a ( i , " SU " ) ;
balance ( t ) . . sum ( i , p ( i , t ) ) =e= d a t a L P ( t , ’ l o a d ’ ) ;
r e s e r v e ( t , g ) . . sum ( i $ ( o r d ( i ) <> o r d ( g ) ) , pu ( i , t )p ( i , t ) ) =g=1p ( g , t ) ;

Model UCLP / a l l / ; O p t i o n o p t c r = 0 . 0 ;
Power schedules (Pi;t ) by thermal units considering generator contingency

constraint are provided in Table 5.4.
The reserve provision (Ri;t ) by thermal units considering generator contingencies
is provided in Table 5.5.
5.2.3 Cost-Based UC with Demand Flexibility Constraints
The balance between generation and demand is traditionally managed by scheduling

the generating units. However, this paradigm is changing gradually. In other words,
the demand values can also change intentionally to increase the efficiency of UC
Table 5.4 Power schedules (Pi;t ) by thermal units considering generator contingency constraint
t1 0.0 0.0 250.0 93.4 54.0 0.0 0.0
t2 0.0 51.8 250.0 80.0 30.0 0.0 0.0
t3 0.0 74.5 250.0 80.0 30.0 0.0 20.0
t4 110.0 104.5 250.0 0.0 30.6 0.0 22.0
t5 128.0 98.2 250.0 0.0 30.0 30.0 20.0
t6 146.0 115.0 250.0 0.0 30.0 38.0 20.0
t7 152.0 130.0 250.0 0.0 34.5 40.2 22.0
t8 158.0 130.1 250.0 0.0 34.5 42.0 24.0
t9 162.8 135.0 250.0 0.0 39.0 42.0 26.0
t10 162.8 135.0 250.0 0.0 39.0 42.0 26.0
t11 158.0 135.0 250.0 0.0 39.0 42.0 24.5
t12 158.0 130.1 250.0 0.0 34.5 42.0 24.0
t13 152.0 130.0 250.0 0.0 34.5 40.2 22.0
t14 146.0 118.6 250.0 0.0 30.0 38.0 20.0
t15 146.1 125.0 250.0 0.0 34.5 38.0 22.0
t16 146.0 118.6 250.0 0.0 30.0 38.0 20.0
t17 139.0 113.3 250.0 0.0 0.0 34.0 20.0
t18 137.3 110.0 250.0 0.0 0.0 0.0 0.0
t19 122.1 95.0 250.0 0.0 0.0 0.0 0.0
t20 106.6 75.0 250.0 0.0 0.0 0.0 0.0
t21 98.0 66.9 250.0 0.0 0.0 0.0 0.0
t22 92.0 56.3 250.0 0.0 0.0 0.0 0.0
t23 96.8 65.0 250.0 0.0 0.0 0.0 0.0
t24 125.3 0.0 250.0 0.0 0.0 0.0 0.0
Table 5.5 Reserve provision (Ri;t ) by thermal units considering generator contingency constraint
t1 0.0 0.0 270.0 186.7 66.0 0.0 0.0
t2 0.0 28.3 0.0 69.4 48.0 0.0 0.0
t3 0.0 7.3 1.8 30.0 22.8 0.0 20.0
t4 0.0 0.0 76.6 0.0 23.5 0.0 10.0
t5 22.0 0.0 64.5 0.0 24.6 25.0 14.0
t6 0.0 4.0 104.0 0.0 24.0 14.0 0.0
t7 0.0 8.7 104.0 0.0 19.5 19.9 0.0
t8 0.0 0.0 103.9 0.0 24.0 20.2 10.0
t9 0.0 15.0 96.3 0.0 19.5 22.0 10.0
t10 0.0 15.0 89.8 0.0 24.0 22.0 12.0
t11 0.0 0.0 98.5 0.0 24.0 22.0 13.6
t12 0.0 0.0 95.1 0.0 28.5 22.0 12.5
t13 0.0 0.2 104.0 0.0 24.0 23.9 0.0
t14 0.0 3.9 104.0 0.0 0.0 24.2 14.0
t15 0.0 21.1 73.5 0.0 19.5 22.0 10.0
t16 0.0 0.0 90.0 0.0 20.0 22.0 14.0
t17 0.0 14.0 104.0 0.0 0.0 11.0 10.0
t18 6.0 33.3 104.0 0.0 0.0 0.0 0.0
t19 0.0 18.1 104.0 0.0 0.0 0.0 0.0
t20 0.0 31.6 75.0 0.0 0.0 0.0 0.0
t21 0.0 31.1 66.9 0.0 0.0 0.0 0.0
t22 0.0 0.0 92.0 0.0 0.0 0.0 0.0
t23 0.0 5.0 91.8 0.0 0.0 0.0 0.0
t24 0.0 95.0 30.3 0.0 0.0 0.0 0.0
problems. In this modern context, the demand values are no longer strict and can
be dispatched. The UC problem incorporating a simple demand response model is
presented as follows:
X
i;t
X
Pi;t Dt (5.11b)
i
.1 &min /Lt Dt .1 C &max /Lt (5.11c)

X X
Dt D Lt (5.11d)
t t
Subject to:
(5.2), (5.3), (5.4), (5.5), (5.6), (5.7)
Equation (5.11c) states the demand variation ranges. &min=max determine the min/-
max flexibility of demand response. Equation (5.11d) states that the total energy of
the consumer does not change over the operating horizon. The &min=max are assumed
to be 10%.
The GAMS code for solving the (5.11) is described in GCode 5.4. As previously
stated, the total operating costs without the generator contingency constraint are
$2:1039 105 . However, if the demand response flexibility is available then the
operating costs would decrease to $2:0965 105 . The power schedules (Pi;t ) by
thermal units considering demand response constraint are given in Table 5.6.
GCode 5.4 Cost-based unit commitment considering demand response

S e t s t / t 1t 2 4 / , i / g1g10 / , k / s g 1s g 2 0 / , c h a r / ch1ch2 / ; a l i a s ( t , h ) ;
Removed f o r s a v i n g s p a c e ; parameter data ( k , i ,) ;
Table dataLP ( t ,)
lambda load
dataLP ( t , ’ load ’ ) =dataLP ( t , ’ load ’ ) 0.45;
parameter unit ( i , char ) ;
u n i t ( i , ’ ch1 ’ ) = 2 4 ; u n i t ( i , ’ ch2 ’ ) =
( g e n d a t a ( i , ’UT ’ )g e n d a t a ( i , ’U0 ’ ) )g e n d a t a ( i , ’ U i n i ’ ) ;
Parameter unit2 ( i , char ) ;
u n i t 2 ( i , ’ ch1 ’ ) = 2 4 ; u n i t 2 ( i , ’ ch2 ’ ) =
( g e n d a t a ( i , ’DT ’ )g e n d a t a ( i , ’ S0 ’ ) ) (1 g e n d a t a ( i , ’ U i n i ’ ) ) ;
g e n d a t a ( i , ’ L j ’ ) = smin ( c h a r , u n i t ( i , c h a r ) ) ;
g e n d a t a ( i , ’ F j ’ ) = smin ( c h a r , u n i t 2 ( i , c h a r ) ) ; V a r i a b l e c o s t T h e r m a l ;
P o s i t i v e v a r i a b l e s pu ( i , t ) , p ( i , t ) , StC ( i , t ) ,SDC( i , t ) , Pk ( i , t , k ) ,D( t ) ;
p . l o ( i , t ) = 0 ; Pk . up ( i , t , k ) = d a t a ( k , i , ’DP ’ ) ; Pk . l o ( i , t , k ) = 0 ;
E q u a t i o n s Uptime1 , Uptime2 , Uptime3 , Dntime1 , Dntime2 , Dntime3 , Ramp1 , Ramp2 , Ramp3 ,
Ramp4 , s t a r t c , s h t d n c , g e n c o n s t 1 , G e n c o n s t 2 , G e n c o n s t 3 , G e n c o n s t 4 , b a l a n c e , DRconst ;
Uptime1 ( i ) $ ( g e n d a t a ( i , " L j " ) >0) . . sum ( t $ ( o r d ( t ) < ( g e n d a t a ( i , " L j " ) + 1 ) ) ,1U( i , t ) ) =e = 0 ;
Uptime3 ( i , t ) $ ( o r d ( t ) > g e n d a t a ( i , " L j " ) and o r d ( t ) <24 g e n d a t a ( i , " UT " ) +2
and n o t ( g e n d a t a ( i , " L j " ) >24 g e n d a t a ( i , " UT " ) ) ) . . sum ( h$ ( ( o r d ( h ) > o r d ( t ) 1) and
Dntime1 ( i ) $ ( g e n d a t a ( i , " F j " ) >0) . . sum ( t $ ( o r d ( t ) < ( g e n d a t a ( i , " F j " ) + 1 ) ) ,U( i , t ) ) =e = 0 ;
sum ( t $ ( o r d ( t ) >24 g e n d a t a ( i , " DT " ) + 1 ) ,1U( i , t )z ( i , t ) ) =g = 0 ;
n o t ( g e n d a t a ( i , " F j " ) >24 g e n d a t a ( i , " DT " ) ) ) . . sum ( h$ ( ( o r d ( h ) > o r d ( t ) 1) and
( o r d ( h ) < o r d ( t ) + g e n d a t a ( i , " DT " ) ) ) ,1U( i , h ) ) =g= g e n d a t a ( i , " DT " )z ( i , t ) ;
genconst1 ( i , h ) . . p ( i , h ) =e=u ( i , h )g e n d a t a ( i , " Pmin " ) +sum ( k , Pk ( i , h , k ) ) ;

+ g e n d a t a ( i , " U i n i " ) $ ( o r d ( h ) = 1 ) +y ( i , h )z ( i , h ) ;
Genconst3 ( i , t , k ) . . Pk ( i , t , k ) = l =U( i , t )d a t a ( k , i , ’DP ’ ) ;
+sum ( ( t , i ) , u ( i , t )g e n d a t a ( i , ’ M i n c o s t ’ ) +sum ( k , d a t a ( k , i , ’ s ’ )pk ( i , t , k ) ) ) ;
Ramp2 ( i , t ) . . p ( i , t ) = l =pu ( i , t ) ;
+z ( i , t +1)g e n d a t a ( i , " SD " ) ;
+y ( i , t )g e n d a t a ( i , " SU " ) ;
balance ( t ) . . sum ( i , p ( i , t ) ) =e= D( t ) ;
DRconst .. sum ( t , d a t a L P ( t , ’ l o a d ’ ) ) =e=sum ( t , D( t ) ) ;
O p t i o n o p t c r = 0 . 0 ; D . up ( t ) =1.1 d a t a L P ( t , ’ l o a d ’ ) ; D . l o ( t ) =0.9 d a t a L P ( t , ’ l o a d ’ ) ;
Table 5.6 Power schedules Time g1 g3 g4 g5 g7 g8

(Pi;t ) by thermal units
considering demand response t1 250.0 80.0 34.5
constraint t2 92.9 250.0 80.0 30.0
t3 132.9 250.0 80.0 34.5
t4 146.0 80.0 250.0 34.5 38.0
t5 146.0 110.0 250.0 34.5 38.0
t6 146.0 120.0 250.0 34.5 38.0
t7 146.0 120.0 250.0 34.5 38.0
t8 146.0 120.0 250.0 34.5 38.0
t9 146.0 120.8 250.0 34.5 38.0
t10 146.0 120.8 250.0 34.5 38.0
t11 146.0 120.0 250.0 34.5 38.0
t12 146.0 120.0 250.0 34.5 38.0
t13 146.0 120.0 250.0 34.5 38.0
t14 146.0 120.0 250.0 34.5 38.0
t15 146.0 120.0 250.0 34.5 38.0
t16 146.0 120.0 250.0 34.5 38.0
t17 146.0 120.0 250.0 34.5 38.0
t18 144.0 115.0 250.0 38.0
t19 128.8 105.0 250.0 30.0
t20 128.0 96.7 250.0
t21 116.4 90.0 250.0
t22 110.0 78.1 250.0
t23 116.0 86.9 250.0
t24 97.8 65.0 250.0
The demand pattern change in cost-based unit commitment with demand

response flexibility is shown in Fig. 5.4. The demand flexibility is changed from
0 to 18% then the demand pattern changes are shown in Fig. 5.5. The variation of
total operating costs vs demand flexibility is depicted in Fig. 5.6.
650 Original demand pattern

With DR
600
550
500
450
400
350
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 5.4 The demand pattern change in cost-based unit commitment with demand response
flexibility
650 0%
3%
6%
600 9%
12%
15%
18%
550
500
450
400
350
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 5.5 The sensitivity analysis of demand pattern changes in CBUC with various demand
response flexibilities
× 105
2.104
2.102
2.1
Total costs ($)
2.098
2.096
2.094
2.092
0 2 4 6 8 10 12 14 16 18 20
Demand flexibility (%)
Fig. 5.6 The variation of total operating costs vs demand flexibility
5.3 Price-Based Unit Commitment
The benefit maximization in price-based unit commitment is formulated in (5.12).

X
max OF D ft pi;t ŒFCi;t C STCi;t C SDCi;t g (5.12)
i;t
Subject to:
(5.2), (5.3), (5.4), (5.5), (5.6), (5.7)
The GAMS code for solving the (5.12) is described in GCode 5.5.
GCode 5.5 Price-based unit commitment Example for ten unit system
Sets t / t 1t 2 4 / , i / g1g10 / , k / s g 1s g 2 0 / , c h a r / ch1ch2 / ;
Alias ( t , h)
T a b l e G d a t a ( i , ) g e n e r a t o r c o s t c h a r a c t e r i s t i c s and l i m i t s
Removed f o r s a v i n g s p a c e
d a t a ( k , i , ’DP ’ ) = ( G d a t a ( i , " Pmax " )G d a t a ( i , " Pmin " ) ) / c a r d ( k ) ;
d a t a ( k , i , ’ P i n i ’ ) = ( o r d ( k ) 1)d a t a ( k , i , ’DP ’ ) + G d a t a ( i , " Pmin " ) ;
d a t a ( k , i , ’ C i n i ’ ) = G d a t a ( i , " a " )power ( d a t a ( k , i , ’ P i n i ’ ) , 2 )
5.3 Price-Based Unit Commitment 137
+ G d a t a ( i , " b " )d a t a ( k , i , ’ P i n i ’ ) + G d a t a ( i , " c " ) ;

d a t a ( k , i , ’ C f i n ’ ) = G d a t a ( i , " a " )power ( d a t a ( k , i , ’ P f i n ’ ) , 2 )
+ G d a t a ( i , " b " )d a t a ( k , i , ’ P f i n ’ ) + G d a t a ( i , " c " ) ;
G d a t a ( i , ’ M i n c o s t ’ ) = G d a t a ( i , ’ a ’ )power ( G d a t a ( i , " Pmin " ) , 2 )
+ G d a t a ( i , ’ b ’ )G d a t a ( i , " Pmin " ) + G d a t a ( i , ’ c ’ ) ;
t a b l e dataLP ( t ,)
lambda load
p a r a m e t e r u n i t ( i , c h a r ) ; u n i t ( i , ’ ch1 ’ ) = 2 4 ;
u n i t ( i , ’ ch2 ’ ) = ( G d a t a ( i , ’UT ’ )G d a t a ( i , ’U0 ’ ) )G d a t a ( i , ’ U i n i ’ ) ;
Parameter unit2 ( i , char ) ;
u n i t 2 ( i , ’ ch1 ’ ) = 2 4 ;
u n i t 2 ( i , ’ ch2 ’ ) = ( G d a t a ( i , ’DT ’ )G d a t a ( i , ’ S0 ’ ) ) (1G d a t a ( i , ’ U i n i ’ ) ) ;
G d a t a ( i , ’ L j ’ ) = smin ( c h a r , u n i t ( i , c h a r ) ) ;
G d a t a ( i , ’ F j ’ ) = smin ( c h a r , u n i t 2 ( i , c h a r ) ) ; V a r i a b l e c o s t T h e r m a l , OF ;
P o s i t i v e v a r i a b l e s pu ( i , t ) , p ( i , t ) , StC ( i , t ) ,SDC( i , t ) , Pk ( i , t , k ) ;
binary variable u( i , t ) ,y( i , t ) , z ( i , t ) ;
p . up ( i , t ) = G d a t a ( i , " Pmax " ) ; p . l o ( i , t ) = 0 ;
p . up ( i , t ) = G d a t a ( i , " Pmax " ) ; pu . up ( i , h ) = G d a t a ( i , " Pmax " ) ;
E e q u a t i o n s Upt1 , Upt2 , Upt3 , Dntime1 , Dntime2 , Dntime3 , Ramp1 , Ramp2 , Ramp3 , Ramp4 ,
s t a r t c , shtdnc , genconst1 , Genconst2 , Genconst3 , Genconst4 , balance , b e n e f i t c a l c ;
Upt1 ( i ) $ ( G d a t a ( i , " L j " ) >0) . . sum ( t $ ( o r d ( t ) < ( G d a t a ( i , " L j " ) + 1 ) ) ,1U( i , t ) ) =e = 0 ;
Upt2 ( i ) $ ( G d a t a ( i , "UT" ) >1) . . sum ( t $ ( o r d ( t ) >24G d a t a ( i , "UT" ) + 1 ) ,U( i , t )y ( i , t ) ) =g = 0 ;
Upt3 ( i , t ) $ ( o r d ( t ) > G d a t a ( i , " L j " ) and o r d ( t ) <24G d a t a ( i , "UT" ) +2 and
n o t ( G d a t a ( i , " L j " ) >24G d a t a ( i , "UT" ) ) ) . . sum ( h$ ( ( o r d ( h ) > o r d ( t ) 1) and
( o r d ( h ) < o r d ( t ) + G d a t a ( i , "UT" ) ) ) ,U( i , h ) ) =g= G d a t a ( i , "UT" )y ( i , t ) ;
Dntime1 ( i ) $ ( G d a t a ( i , " F j " ) >0) . . sum ( t $ ( o r d ( t ) < ( G d a t a ( i , " F j " ) + 1 ) ) ,U( i , t ) ) =e = 0 ;
Dntime2 ( i ) $ ( G d a t a ( i , "DT" ) >1) . .
sum ( t $ ( o r d ( t ) >24G d a t a ( i , "DT" ) + 1 ) ,1U( i , t )z ( i , t ) ) =g = 0 ;
Dntime3 ( i , t ) $ ( o r d ( t ) > G d a t a ( i , " F j " ) and o r d ( t ) <24G d a t a ( i , "DT" ) +2 and
n o t ( G d a t a ( i , " F j " ) >24G d a t a ( i , "DT" ) ) ) . . sum ( h$ ( ( o r d ( h ) > o r d ( t ) 1) and
( o r d ( h ) < o r d ( t ) + G d a t a ( i , "DT" ) ) ) ,1U( i , h ) ) =g= G d a t a ( i , "DT" )z ( i , t ) ;
s t a r t c ( i , t ) . . StC ( i , t ) =g= G d a t a ( i , " c o s t s t " )y ( i , t ) ;
s h t d n c ( i , t ) . . SDC( i , t ) =g= G d a t a ( i , " CostsD " )z ( i , t ) ;
g e n c o n s t 1 ( i , h ) . . p ( i , h ) =e=u ( i , h )G d a t a ( i , " Pmin " ) +sum ( k , Pk ( i , h , k ) ) ;
+ G d a t a ( i , " U i n i " ) $ ( o r d ( h ) = 1 ) +y ( i , h )z ( i , h ) ;
+sum ( ( t , i ) , u ( i , t )G d a t a ( i , ’ M i n c o s t ’ )
+sum ( k , d a t a ( k , i , ’ s ’ )pk ( i , t , k ) ) ) ;
Ramp1 ( i , t ) . . p ( i , t 1)p ( i , t ) = l =U( i , t )G d a t a ( i , ’RD ’ ) +z ( i , t )G d a t a ( i , "SD" ) ;
Ramp2 ( i , t ) . . p ( i , t ) = l =pu ( i , t ) ;
Ramp3 ( i , t ) $ ( o r d ( t ) <24) . . pu ( i , t ) = l = ( u ( i , t )z ( i , t + 1 ) )G d a t a ( i , " Pmax " )
+z ( i , t +1)G d a t a ( i , "SD" ) ;
Ramp4 ( i , t ) $ ( o r d ( t ) >1) . . pu ( i , t ) = l =p ( i , t 1)+U( i , t 1)G d a t a ( i , ’RU ’ )
+y ( i , t )G d a t a ( i , "SU" ) ;
Balance ( t ) . . sum ( i , p ( i , t ) ) = l = d a t a L P ( t , ’ l o a d ’ ) ;
B e n e f i t c a l c . . OF=e=sum ( ( i , t ) , d a t a L P ( t , ’ lambda ’ )p ( i , t ) ) c o s t T h e r m a l
Option o p t c r =0 .0 ;
S o l v e UCLP m a x i m i z i n g OF u s i n g mip ;
350
P1 P2 P3 P4 P5 P6 P7 P8 P9 P 10
300
250
200
150
100
50
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 5.7 Power schedules in price-based unit commitment vs time
The start-up, shut-down, and fuel costs are $499.000, $163.800, and
$372,277.510, respectively. The net benefit of Genco is $58,186.325. The power
schedules of thermal units in cost-based unit commitment are depicted in Fig. 5.7.
5.4 Applications
The unit commitment analysis has a vast range of applications, and some of them
are described as follows:
5.4.1 Cost-Based UC
In practical applications, the thermal units might have some limitations in fuels.
This means that the unit cannot generate more than some certain amount of energy.
In other words, it might not be technically possible to keep the unit on for all time
steps for the operating period. The fuel constrained UC is formulated in 4,335,178.
Another important constraint in UC is environmental emission which is modeled
in [4].
Another aspect of UC is considering the uncertainties associated with renewable
energy resources (like wind turbines). Generally speaking, there are several methods
for dealing with uncertainties in power system studies, such as
• Stochastic methods: two-stage scenario-based [5], two-point estimate [6, 12]
• Fuzzy methods [7]
• Robust optimization [8, 9]
• Information gap decision theory [10, 11]
The different uncertainty parameters and modeling methods are shown in
Fig. 5.8 [12].
References 139
Fig. 5.8 Different uncertainty parameters and modeling methods
5.4.2 Price-Based UC
Some price-based UC formulations are listed as follows:

• Robust optimization-based self-scheduling of hydro-thermal Genco in smart
grids [13]
• Risk averse optimal operation of a virtual power plant using two-stage stochastic
programming [14]
• Smart self-scheduling of Gencos with thermal and energy storage units under
price uncertainty [15].
References
1. A. Ademovic, S. Bisanovic, M. Hajro, A genetic algorithm solution to the unit commitment

problem based on real-coded chromosomes and fuzzy optimization, in Melecon 2010 - 2010
15th IEEE Mediterranean Electrotechnical Conference, April 2010, pp. 1476–1481
2. J.M. Arroyo, A.J. Conejo, Optimal response of a thermal unit to an electricity spot market.
3. M. Ortega-Vazquez, D. Kirschen, Optimizing the spinning reserve requirements using a
cost/benefit analysis, in 2008 IEEE Power and Energy Society General Meeting - Conversion
and Delivery of Electrical Energy in the 21st Century, July 2008, p. 1
4. T. Gjengedal, Emission constrained unit-commitment (ECUC). IEEE Trans. Energy Convers.

11(1), 132–138 (1996)
5. C. Uçkun, A. Botterud, J.R. Birge, An improved stochastic unit commitment formulation to
accommodate wind uncertainty. IEEE Trans. Power Syst. 31(4), 2507–2517 (2016)
6. A. Soroudi, M. Aien, M. Ehsan, A probabilistic modeling of photo voltaic modules and wind
power generation impact on distribution networks. IEEE Syst. J. 6(2), 254–259 (2012)
8. R. Jiang, J. Wang, Y. Guan, Robust unit commitment with wind power and pumped storage
hydro. IEEE Trans. Power Syst. 27(2), 800–810 (2012)
payments minimization under electricity price uncertainty. IEEE Trans. Smart Grid 7(1), 261–
272 (2016)
10. A. Soroudi, A. Rabiee, A. Keane, Information gap decision theory approach to deal with wind
power uncertainty in unit commitment. Electr. Power Syst. Res. 145, 137–148 (2017)
11. C. Murphy, A. Soroudi, A. Keane, Information gap decision theory-based congestion and
voltage management in the presence of uncertain wind power. IEEE Trans. Sust. Energy 7(2),
841–849 (2016)
12. A. Soroudi, T. Amraee, Decision making under uncertainty in energy systems: state of the art.
Renew. Sust. Energ. Rev. 28, 376–384 (2013)
13. A. Soroudi, Robust optimization based self scheduling of hydro-thermal Genco in smart grids.
Energy 61, 262–271 (2013)
14. M.A. Tajeddini, A. Rahimi-Kian, A. Soroudi, Risk averse optimal operation of a virtual power
plant using two stage stochastic programming. Energy 73, 958–967 (2014)
15. A. Soroudi, Smart self-scheduling of Gencos with thermal and energy storage units under price
uncertainty. Int. Trans. Electr. Energy Syst. 24(10), 1401–1418 (2014)
Chapter 6
Multi-Period Optimal Power Flow
This chapter provides a solution for optimal power flow OPF problem in GAMS.
Different OPF models are investigated, such as single and multi-period DC-AC
optimal power flow.
6.1 Single Period Optimal DC Power Flow
There are some necessary conditions that make the DC power flow acceptable as an
approximate solution for AC power flow such as:
x
• The ratio of rijij should be large enough that rij can be neglected.
• The voltage magnitudes are approximately 1 pu.
The DC power flow concept for a two-bus network is shown in Fig. 6.1. The basic
variables in DC power flow are voltage angles ıi . The angle of the slack bus is
assumed to be zero as the reference for the rest of network.
The technical and economic characteristics of generating units shown in
Table 6.1 are given as follows:
The demand at bus 2 is L2 D 400 MW, line reactance is X12 D 0:2 pu, and line
flow limit is Pmax
12 D 1:50 pu (per unit on 100 MVA base). The optimization problem
which should be solved is formulated in (6.1).
X
min OF D ag .Pg /2 C bg Pg C cg (6.1a)
Pg ;ıi
g1 ;g2
ı1 ı 2
Pij D (6.1b)
X12
Pg1 D P12 (6.1c)
Pg2 C P12 D L2 (6.1d)

DOI 10.1007/978-3-319-62350-4_6
142 6 Multi-Period Optimal Power Flow
Fig. 6.1 DC power flow for a two-bus network
GCode 6.1 The OPF GAMS code for two-bus network, Example (6.1)
Sets
Gen / g1g2 /
bus / 1 2 / ;
Scalars
L2 /400/
X12 / 0 . 2 /
Sbase / 1 0 0 /
P12_max / 1 . 5 / ;
a b c Pmin Pmax
G1 3 20 100 28 206
G2 4 . 0 5 1 8 . 0 7 9 8 . 8 7 90 284;
V a r i a b l e s P ( gen ) ,OF , d e l t a ( b u s ) , P12 ;
Equations
eq1 , eq2 , eq3 , eq4 ;
eq1 . . OF=e=sum ( gen , d a t a ( gen , ’ a ’ ) P ( gen ) P ( gen ) + d a t a ( gen , ’ b ’ )

P ( gen ) + d a t a ( gen , ’ c ’ ) ) ;
eq2 . . P ( ’G1 ’ ) =e=P12 ;
eq3 . . P ( ’G2 ’ ) +P12=e=L2 / S b a s e ;
eq4 . . P12=e = ( d e l t a ( ’ 1 ’ ) d e l t a ( ’ 2 ’ ) ) / X12 ;
P . l o ( gen ) = d a t a ( gen , ’ Pmin ’ ) / S b a s e ;
P . up ( gen ) = d a t a ( gen , ’ Pmax ’ ) / S b a s e ;
P12 . l o =P12_max ;
P12 . up=+P12_max ;
d e l t a . fx ( ’1 ’ ) =0;
Model OPF / a l l / ;
S o l v e OPF u s qcp min o f ;
Pmax max
12 P12 P12 (6.1e)
ı1 D 0 Slack (6.1f)
The GAMS code for solving the (6.1) is provided in GCode 6.1.
The operating costs would be $306.108, P12 D 150 MW and ı2 D 0:3 (rad).
The general power flow concept is shown in Fig. 6.2. As it can be seen in Fig. 6.2,
every bus might host some generation and demand. Each bus might be connected
6.1 Single Period Optimal DC Power Flow 143
Table 6.1 The techno-economic data of thermal units in two-bus OPF example
g ag ($=MW2 ) bg ($=MW) cg ($) Pmin
g (MW) Pmax
g (MW)
g1 0.12 14.8 89 28 200
g2 0.17 16.57 83 20 290
Fig. 6.2 General power flow concept
to other network buses by multiple branches (with different characteristics). The

power balance between generation, demand and power transfers should be satisfied
at every bus of the network under study.
The general mathematical formulation of OPF is described in (6.2). It should be
noted that the technical terms of (6.2) are all linear. The only nonlinear part is the
cost function. This makes the DC-OPF a strong tool for power system studies at the
transmission level. This means that if the cost term can be expressed in linear form
then the DC-OPF would become a linear programming problem and can be solved
using linear solvers in GAMS such as CPLEX [1].
X
OF D ag .Pg /2 C bg Pg C cg (6.2a)
g2˝G
ıi ı j
Pij D ij 2 ˝` (6.2b)
xij
X X
Pg Li D Pij W i i 2 ˝B (6.2c)
g2˝Gi j2˝ì
Pmax
ij Pij Pmax
ij ij 2 ˝` (6.2d)
Pmin
g Pg Pmax
g (6.2e)
Fig. 6.3 Three bus network example (a) data and (b) power flow solution
6.1.1 Three-Bus Network DC-OPF
The three-bus network data is shown in Fig. 6.3. This example is taken from [2]
(Example 4B, p. 110).
The GAMS code for solving this problem is given as GCode 6.2: The developed
code for solving the OPF problem in three-bus network (Fig. 6.3) is explained here.
It is general and can be used for any network with any size.
• Three sets are defined: bus (all network buses), slack(bus) which shows slack
buses with reference angle values, Gen (set of generating units)
• A scalar value named Sbase is defined for per unit calculations
• The set node is defined as the similar set to set bus
• The table GenData defines the technical and economic characteristics of gener-
ating units
• The set GBconect defines the connection point of each generating unit
• The table BusData specifies the demand values in each bus
• The set conex specifies how each bus is connected to the other network buses
• The table branch defines the branch characteristics
• Four variables are used for this formulation namely: OF (objective function),
Pij (active power flow between bus and node), Pg(generating schedule of each
generating unit), and delta (voltage angle at each bus)
• Three equations are defined: const1 (active flow calculation between each pair of
connected buses), const2 (nodal active power balance in each bus), and const3
(objective function calculation)
• The definition of the model loadflow (specifies for GAMS that which constraints
should be taken into account)
• Variables’ limits specification
• Solve statement
• Generating report from the solved model
• The marginal value of const2.m provides the LMP at each bus. This is used for
congestion cost calculation.
GCode 6.2 The DC-OPF GAMS code for three-bus network, Example (Sect. 6.1.1)
Sets
bus / 1 3 /
s l a c k ( bus ) / 3 /
Gen / g1g3 / ;
scalars
Sbase / 1 0 0 / ;
a l i a s ( bus , node ) ;
T a b l e GenData ( Gen , ) G e n e r a t i n g u n i t s c h a r a c t e r i s t i c s
b pmin pmax
g1 10 0 65
g2 11 0 100;

s e t GBconect ( bus , Gen ) c o n n e c t i v i t y i n d e x o f e a c h g e n e r a t i n g u n i t
t o each bus
/1 . g1
3 . g2 / ;
T a b l e BusData ( bus , ) Demands o f e a c h b u s i n MW
Pd
2 100;
s e t conex Bus c o n n e c t i v i t y m a t r i x
/
1 . 2
2 . 3
1 . 3/;
c o n e x ( bus , node ) $ ( c o n e x ( node , b u s ) ) = 1 ;
T a b l e b r a n c h ( bus , node , ) Network t e c h n i c a l c h a r a c t e r i s t i c s
x Limit
1 . 2 0.2 100
2 . 3 0.25 100
1 . 3 0.4 100 ;
b r a n c h ( bus , node , ’ x ’ ) $ ( b r a n c h ( bus , node , ’ x ’ ) = 0 ) = b r a n c h ( node , bus , ’ x ’
);
b r a n c h ( bus , node , ’ L i m i t ’ ) $ ( b r a n c h ( bus , node , ’ L i m i t ’ ) = 0 ) = b r a n c h ( node
, bus , ’ L i m i t ’ ) ;
b r a n c h ( bus , node , ’ b i j ’ ) $ c o n e x ( bus , node ) = 1 / b r a n c h ( bus , node , ’ x ’ ) ;
Variables
OF
P i j ( bus , node )
Pg ( Gen )
d e l t a ( bus ) ;
Equations const1 , const2 , const3 ;
c o n s t 1 ( bus , node ) $ c o n e x ( bus , node ) . . P i j ( bus , node ) =e=
b r a n c h ( bus , node , ’ b i j ’ ) ( d e l t a ( b u s )
d e l t a ( node ) ) ;
c o n s t 2 ( b u s ) . . +sum ( Gen$GBconect ( bus , Gen ) , Pg ( Gen ) )BusData ( bus , ’

pd ’ ) / S b a s e =e=
+sum ( n o d e $ c o n e x ( node , b u s ) , P i j ( bus , node )
);
const3 . . OF=g=sum ( Gen , Pg ( Gen ) GenData ( Gen , ’ b ’ ) S b a s e ) ;
Model l o a d f l o w / const1 , const2 , const3 / ;

Pg . l o ( Gen ) =GenData ( Gen , ’ Pmin ’ ) / S b a s e ;
Pg . up ( Gen ) =GenData ( Gen , ’ Pmax ’ ) / S b a s e ;
d e l t a . up ( b u s ) = p i ; d e l t a . l o ( b u s )=p i ; d e l t a . f x ( s l a c k ) = 0 ;
P i j . up ( bus , node ) $ ( ( c o n e x ( bus , node ) ) ) =1 b r a n c h ( bus , node , ’ L i m i t ’ ) /
Sbase ;
P i j . l o ( bus , node ) $ ( ( c o n e x ( bus , node ) ) ) =1 b r a n c h ( bus , node , ’ L i m i t ’ ) /
Sbase ;
S o l v e l o a d f l o w m i n i m i z i n g OF u s i n g l p ;
p a r a m e t e r r e p o r t ( bus , ) , C o n g e s t i o n c o s t ;
r e p o r t ( bus , ’ Gen (MW) ’ ) = sum ( Gen$GBconect ( bus , Gen ) , Pg . l ( Gen ) ) s b a s e
;
r e p o r t ( bus , ’ Angle ’ ) = d e l t a . l ( b u s ) ;
r e p o r t ( bus , ’ l o a d (MW) ’ ) = BusData ( bus , ’ pd ’ ) ;
r e p o r t ( bus , ’LMP( $ /MWh) ’ ) = c o n s t 2 .m( b u s ) / s b a s e ;
C o n g e s t i o n c o s t = sum ( ( bus , node ) , P i j . l ( bus , node ) ( c o n s t 2 .m( b u s ) +
c o n s t 2 .m( node ) ) ) / 2 ;
display report , Pij . l , Congestioncost ;
The total operating costs will be 1035 $/h and the LMP for all buses are equal to
11 $/h. This means that if the demand value at any bus increases for 1 MW then
the operating cost will increase by 11 $/h. Since the generating unit 1 is generating
power at its maximum limit (because it is cheaper) the additional demand should be
supplied by generating unit 2 with operating cost equal to 11 $/MW h. The detailed
three-bus optimal power flow solution with (Pmax
ij D 100 MW) is given in Table 6.2.
Question: What would happen if the flow limit of the branch connecting bus 1 to
bus 2 is reduced to 50 MW.
Answer: The answer is straightforward. The operating cost might increase but
how much? We can easily decrease the flow limit in the GCode 6.2 (table Branch).
The new operating cost would be 1056.250 $/h. The LMP values are different for
each bus in this case. The LMP values are 1 D 10 $/MW h, 2 D 11:625 $/MW h,
3 D 11 $/MW h. The power flow solution is shown in Fig. 6.4. The LMP value
in bus 1 is 1 D 10 $/MW h because if the load increases in this node it will be
supplied by generator 1 (which the operating costs are 10 $/MW h). The LMP value
in bus 3 is 3 D 11 $/MW h because any increase in load in this node should be
supplied by generator 2 (which the operating costs are 11 $/MW h). Generator 1
Table 6.2 The three-bus Bus (i) Pg (MW) ıi (rad) Li (MW) i ($/MW h)
optimal power flow solution
(Pmax D 100 MW) 1 65 0.02 0 11
ij
2 0 0.10 100 11
3 35 0.00 0 11
Fig. 6.4 The three-bus network with Pmax

12 D 50 MW
cannot send more power by line 1 2 since it is congested and the direction of
the power flow in branch 3 1 is from bus 3 to bus 1. Finally, The LMP value in
bus 2 is 3 D 11:625 $/MW h because some of this demand will be supplied by
generating unit 1 and some part will be supplied by unit 2. If the demand should
pay the LMP value for every MW h then the total payment by the demand would be
100 MW 11:625 $/MW h = 1162.5 $/h. On the other hand, the generating units are
also paid based on the LMP value of the connection point. The total payments to the
generating units would be 43:7510C56:2511 D 1056:25 $/h. As it can be seen,
there is a difference between what demand pays and what generating units receive.
This surplus money is equal to 1162:5 1056:25 D 106:25 $/h. This is also called
the congestion cost. Another technique for calculating the congestion costs (Ccg ) is
using the following formula [3]:
X
Ccg D Pji .i j / (6.3)
ij
The power flow solution of three-bus network with Pmax

12 D 50 MW is shown in
Fig. 6.4.
The detailed three-bus optimal power flow solution with (Pmax
12 D 50 MW) is
given in Table 6.3.
Table 6.3 The three-bus Bus (i) Pg (MW) ıi (rad) Li (MW) i ($/MW h)
optimal power flow solution
(Pmax 1 43.75 0.025 0 10
12 D 50 MW)
2 0 0.125 100 11.625
3 56.25 0 0 11
Fig. 6.5 The five-bus network data
6.1.2 Five-Bus Network DC-OPF
The five-bus network data is shown in Fig. 6.5. This example is taken from [4].
The GAMS code developed for solving OPF in five-bus network is provided in
GCode 6.3.
GCode 6.3 The OPF GAMS code for five-bus network, Example (Sect. 6.1.2)
S e t s b u s / 1 5 / , s l a c k ( b u s ) / 1 / , Gen / g1g5 / ;
S c a l a r s S b a s e / 1 0 0 / ; a l i a s ( bus , node ) ;
T a b l e GenData ( Gen , )
b pmin pmax
g1 14 0 40
g2 15 0 170
g3 30 0 520
g4 40 0 200
g5 20 0 600 ;
s e t GBconect ( bus , Gen ) c o n n e c t i v i t y i n d e x o f g e n e r a t i n g u n i t
/1 . g1
1 . g2
3 . g3
4 . g4
5 . g5 / ;
Pd
2 300
3 300
4 400;
/1 . 2
2 . 3
3 . 4
4 . 1
4 . 5
5 . 1/;
c o n e x ( bus , node ) $ c o n e x ( node , b u s ) = 1 ;
T a b l e b r a n c h ( bus , node , )
x Limit
1 . 2 0.0281 400
1 . 4 0.0304 400
1 . 5 0.0064 400
2 . 3 0.0108 400
3 . 4 0.0297 400
4 . 5 0.0297 240 ;
);
, bus , ’ L i m i t ’ ) ;
V a r i a b l e s OF , P i j ( bus , node ) , Pg ( Gen ) , d e l t a ( b u s ) ;
b r a n c h ( bus , node , ’ b i j ’ ) ( d e l t a ( b u s ) d e l t a ( node ) ) ;
pd ’ ) / S b a s e =e=
+sum ( n o d e $ c o n e x ( node , b u s ) , P i j ( bus ,
node ) ) ;
d e l t a . up ( b u s ) = p i ;
d e l t a . l o ( b u s )=p i ;
d e l t a . fx ( s l a c k ) =0;
P i j . up ( bus , node ) $ ( ( c o n e x ( bus , node ) ) ) = b r a n c h ( bus , node , ’ L i m i t ’ ) /
Sbase ;
P i j . l o ( bus , node ) $ ( ( c o n e x ( bus , node ) ) ) =
b r a n c h ( bus , node , ’ L i m i t ’ ) / S b a s e ;
s o l v e l o a d f l o w m i n i m i z i n g OF u s i n g l p ;
r e p o r t ( bus , ’ Gen (MW) ’ ) = sum ( Gen$GBconect ( bus , Gen ) , Pg . l ( Gen ) ) s b a s e
;
C o n g e s t i o n c o s t = sum ( ( bus , node ) ,
P i j . l ( bus , node ) ( c o n s t 2 .m( b u s ) + c o n s t 2 .m( node ) ) ) / 2 ;
Display report , Pij . l , Congestioncost ;
Fig. 6.6 The five-bus network flows and directions
Table 6.4 The optimal Bus (i) Pg (MW) ıi (rad) Li (MW) i ($/MW h)
power flow solution in
five-bus network 1 210.000 0.000 0 16.977
2 0.000 0.070 300 26.384
3 323.495 0.065 300 30.000
4 0.000 0.057 400 39.943
5 466.505 0.014 0 10.000
The minimum operating cost is 17,479.897 $/h. The network is highly congested,
and the LMP values are different in various buses. The five-bus network flows and
directions are depicted in Fig. 6.6. The detailed five-bus optimal power flow solution
is given in Table 6.4.
6.1.3 IEEE Reliability Test System 24 Bus
The IEEE RTS 24-bus network is shown in Fig. 6.7. It is a transmission network
with the voltage levels of 138 kV, 230 kV, and Sbase = 100 MVA. The branch data
for IEEE RTS 24-bus network is given in Table 6.5 [5]. The from bus, to bus,
reactance (X), resistance (r), total line charging susceptance (b), and MVA rating
(MVA) are specified in this table. The parallel lines in MATPOWER are merged,
and the resultants are given in Table 6.5. The generation data for IEEE RTS 24-bus
network is given in Table 6.6. The data of generating units in this network is inspired
by Conejo et al. [6] and Bouffard et al. [7] with some modifications. The slack bus
is bus 13 in this network.
Fig. 6.7 IEEE RTS 24-bus network
Table 6.5 Branch data for IEEE RTS 24-bus network

Rating Rating
From To r(pu) x(pu) b(pu) (MVA) From To r(pu) x(pu) b(pu) (MVA)
1 2 0.0026 0.0139 0.4611 175 11 13 0.0061 0.0476 0.0999 500
1 3 0.0546 0.2112 0.0572 175 11 14 0.0054 0.0418 0.0879 500
1 5 0.0218 0.0845 0.0229 175 12 13 0.0061 0.0476 0.0999 500
2 4 0.0328 0.1267 0.0343 175 12 23 0.0124 0.0966 0.2030 500
2 6 0.0497 0.1920 0.0520 175 13 23 0.0111 0.0865 0.1818 500
3 9 0.0308 0.1190 0.0322 175 14 16 0.0050 0.0389 0.0818 500
3 24 0.0023 0.0839 0.0000 400 15 16 0.0022 0.0173 0.0364 500
4 9 0.0268 0.1037 0.0281 175 15 21 0.0032 0.0245 0.2060 1000
5 10 0.0228 0.0883 0.0239 175 15 24 0.0067 0.0519 0.1091 500
6 10 0.0139 0.0605 2.4590 175 16 17 0.0033 0.0259 0.0545 500
7 8 0.0159 0.0614 0.0166 175 16 19 0.0030 0.0231 0.0485 500
8 9 0.0427 0.1651 0.0447 175 17 18 0.0018 0.0144 0.0303 500
8 10 0.0427 0.1651 0.0447 175 17 22 0.0135 0.1053 0.2212 500
9 11 0.0023 0.0839 0.0000 400 18 21 0.0017 0.0130 0.1090 1000
9 12 0.0023 0.0839 0.0000 400 19 20 0.0026 0.0198 0.1666 1000
10 11 0.0023 0.0839 0.0000 400 20 23 0.0014 0.0108 0.0910 1000
10 12 0.0023 0.0839 0.0000 400 21 22 0.0087 0.0678 0.1424 500
152
Table 6.6 Generation data for IEEE RTS 24-bus network

Gen Bus Pmax
i Pmin
i bi ($/MW) Csi ($) Cdi ($) RUi (MW h1 ) RDi (MW h1 ) SUi (MW h1 ) SDi (MW h1 ) UTi (h) DTi (h) ui;tD0 Ui0 (h) Si0 (h)
g1 18 400 100 5.47 0 0 47 47 105 108 1 1 1 5 0
g2 21 400 100 5.47 0 0 47 47 106 112 1 1 1 6 0
g3 1 152 30.4 13.32 1430.4 1430.4 14 14 43 45 8 4 1 2 0
g4 2 152 30.4 13.32 1430.4 1430.4 14 14 44 57 8 4 1 2 0
g5 15 155 54.25 16 0 0 21 21 65 77 8 8 0 0 2
g6 16 155 54.25 10.52 312 312 21 21 66 73 8 8 1 10 0
g7 23 310 108.5 10.52 624 624 21 21 112 125 8 8 1 10 0
g8 23 350 140 10.89 2298 2298 28 28 154 162 8 8 1 5 0
g9 7 350 75 20.7 1725 1725 49 49 77 80 8 8 0 0 2
g10 13 591 206.85 20.93 3056.7 3056.7 21 21 213 228 12 10 0 0 8
g11 15 60 12 26.11 437 437 7 7 19 31 4 2 0 0 1
g12 22 300 0 0 0 0 35 35 315 326 0 0 1 2 0
6 Multi-Period Optimal Power Flow
6.1.3.1 IEEE-RTS: Base Case
In this case, it is assumed that the network is intact and all branches and generating
units are working in normal condition. The minimum operating cost is $29,574.275
obtained by using the GCode 6.4.
GCode 6.4 The OPF GAMS code for IEEE Reliability test 24-bus network, Example (Sect. 6.1.3)
s e t s b u s / 1 2 4 / , s l a c k ( b u s ) / 1 3 / , Gen / g1g12 / ; s c a l a r S b a s e
/ 1 0 0 / ; a l i a s ( bus , node ) ;
Pmax Pmin b CostsD c o s t s t RU RD SU SD UT DT
u i n i U0 So
g1 400 100 5.47 0 0 47 47 105 108 1 1
1 5 0 ;
t o each bus
/18 . g1 / ;
T a b l e BusData ( bus , ) Demands o f e a c h b u s i n MW/ MVar
Pd Qd
1 108 2 2 ;
r x b limit
1 . 2 0.0026 0.0139 0.4611 175;
p a r a m e t e r c o n e x ( bus , node ) ;
c o n e x ( bus , node ) $ b r a n c h ( bus , node , ’ l i m i t ’ ) = 1 ;
);
, bus , ’ L i m i t ’ ) ;
V a r i a b l e s OF , P i j ( bus , node ) , Pg ( Gen ) , d e l t a ( b u s ) ;
c o n s t 1 ( bus , node ) $ ( c o n e x ( bus , node ) ) . .
P i j ( bus , node ) =e= b r a n c h ( bus , node , ’ b i j ’ ) ( d e l t a ( b u s ) d e l t a ( node ) ) ;
pd ’ ) / S b a s e =e=
+sum ( n o d e $ c o n e x ( node , b u s ) , P i j ( bus , node ) ) ;
Pg . l o ( Gen ) =GenData ( Gen , ’ Pmin ’ ) / S b a s e ; Pg . up ( Gen ) =GenData ( Gen ,
’ Pmax ’ ) / S b a s e ;
d e l t a . up ( b u s ) = p i / 2 ; d e l t a . l o ( b u s )=p i / 2 ; d e l t a . f x ( s l a c k ) = 0 ;
Sbase ;
Sbase ;
Table 6.7 Base case solution Bus Pg (MW) ıi (rad) Load (MW) i ($/MW h)
of IEEE RTS 24-bus network
(branch flow limits are 1 152 0.150 108 20.7
unchanged) 2 152 0.151 97 20.7
3 0.113 180 20.7
4 0.185 74 20.7
5 0.191 71 20.7
6 0.230 136 20.7
7 257.15 0.105 125 20.7
8 0.186 171 20.7
9 0.136 175 20.7
10 0.172 195 20.7
11 0.044 20.7
12 0.031 20.7
13 206.85 0.000 265 20.7
14 0.027 194 20.7
15 167 0.182 317 20.7
16 155 0.168 100 20.7
17 0.251 20.7
18 400 0.276 333 20.7
19 0.146 181 20.7
20 0.164 128 20.7
21 400 0.291 20.7
22 300 0.399 20.7
23 660 0.187 20.7
24 0.069 20.7
The base case solution of IEEE RTS 24-bus network without changing the branch
flow limits is described in Table 6.7. As it can be observed in Table 6.7, the LMP
values are all the same and equal to 20.7 $/MW h. This is because there is no
congestion in this network for the given loading values. The congestion cost would
be zero in this case.
6.1.3.2 IEEE-RTS: Branch Flow Limit Reduction
Now the branch flow limits are reduced by 30%. The problem is solved again, and
the angle values are found as given in Table 6.8. The minimum operating cost is
$29,747.745 obtained by using the GCode 6.4. The congestion cost is $4597.217 in
this case. The solution of IEEE RTS 24-bus network (branch flow limits are reduced
by 30%) is given in Table 6.8. Reducing the branch flow limit will not only increase
the operating cost but also makes the LMP values different across the network.
Table 6.8 Solution of IEEE Bus Pg (MW) ıi (rad) Load (MW) i ($/MW h)
RTS 24-bus network (branch
flow limits are reduced by 1 152.00 0.162 108 20.66
30%) 2 152.00 0.163 97 20.71
3 0.129 180 19.11
4 0.197 74 20.85
5 0.203 71 20.98
6 0.241 136 21.17
7 247.50 0.130 125 20.70
8 0.205 171 21.14
9 0.147 175 20.96
10 0.183 195 21.32
11 0.053 22.69
12 0.038 20.55
13 251.24 0.000 265 20.93
14 0.012 194 25.78
15 132.26 0.158 317 16.00
16 155.00 0.149 100 15.68
17 0.230 15.79
18 400.00 0.255 333 15.85
19 0.131 181 16.86
20 0.152 128 17.87
21 400.00 0.269 15.89
22 300.00 0.378 15.85
23 660.00 0.177 18.42
24 0.049 17.19
6.1.3.3 IEEE-RTS: Branch Outage
In this case, some branch contingencies are examined. In order to simulate the
branch outage, the following equation should be satisfied.
ıi ıj
Pij Mij (6.4)
xij
ıi ıj
Pij Mij (6.5)
xij
where ij is a binary parameter which shows the status of the branch connecting bus
i to bus j. In the developed GAMS code, if the branch limit is set to 0 it is considered
as an outaged branch. This is because const1 which models the flow calculation in
line ij is calculated for every branch which has conex.bus; node/ > 0. Now some
different contingencies are evaluated as follows:
• Contingency 1: branch `2019 ; `1223 are out. Congestion costs are $4905.000
and OF = $29,888.196. The congested lines are `1323 . The LMP values are not
the same in different buses.
• Contingency 2: branch `1416 ; `1619 are out. Congestion costs are $6224.250
and OF = $32,199.855. The congested lines are `243 ; `87 . The LMP values are
not the same in different buses.
• Contingency 3: branch `15 ; `42 are out. Congestion costs are $0 and
OF = $29,574.275. No line would be congested and therefore the LMP values
are the same in all buses.
The OPF solutions for these three contingency cases are provided in Table 6.9.
Table 6.9 Solution of IEEE RTS 24-bus network (branch outage contingencies)
Contingency 1 Contingency 2 Contingency 3
Pg
Bus (MW) ıi (rad) i ($/MW h) Pg (MW) ıi (rad) i ($/MW h) Pg (MW) ıi (rad) i ($/MW h)
1 152 0.218 20.70 152 0.106 20.93 152 0.050 20.70
2 152 0.219 20.70 152 0.112 20.93 152 0.054 20.70
3 0.193 20.70 0.055 20.93 0.088 20.70
4 0.252 20.70 0.156 20.93 0.215 20.70
5 0.257 20.70 -0.173 20.93 0.239 20.70
6 0.294 20.70 0.227 20.93 0.210 20.70
7 289.15 0.123 20.70 300 0.038 20.70 257.15 0.108 20.70
8 0.224 20.70 0.145 20.93 0.189 20.70
9 0.202 20.70 0.116 20.93 0.138 20.70
10 0.235 20.70 0.181 20.93 0.176 20.70
11 0.094 20.70 0.122 20.93 0.044 20.70
12 0.116 20.70 0.037 20.93 0.032 20.70
13 206.85 20.70 436 20.93 206.85 20.70
14 0.053 20.70 0.203 20.93 0.029 20.70
15 167 0.077 20.70 66.25 0.598 5.47 167 0.189 20.70
16 155 0.060 20.70 54.25 0.631 5.47 155 0.173 20.70
17 0.144 20.70 0.691 5.47 0.257 20.70
18 400 0.171 20.70 400 0.706 5.47 400 0.283 20.70
19 0.019 20.70 0.073 20.93 0.150 20.70
20 0.419 10.89 0.109 20.93 0.167 20.70
21 400 0.185 20.70 329.5 0.711 5.47 400 0.298 20.70
22 300 0.293 20.70 300 0.827 5.47 300 0.405 20.70
23 628 0.433 10.89 660 0.142 20.93 660 0.190 20.70
24 0.026 20.70 0.390 5.47 0.083 20.70
6.1.3.4 IEEE-RTS: Generator Outage
In this case, some generating units are out of service. This can be because of
unplanned outage or maintenance purpose. Now some different contingencies are
evaluated as follows:
• Contingency 1: The generating unit g9 (connected to bus 7) is out. Solving the
problem shows that no line is congested but removing the g9 will cause an
increase in total operating costs which becomes OF = $29,633.420.
problem shows that line `87 is congested but removing the g5 will cause an
increase in total operating costs which becomes OF = $30,328.570. The LMP
values are different on different buses, and the congestion costs are $40.250.
problem shows that line `87 is congested but removing the g8 will cause an
increase in total operating costs which becomes OF = $33,078.420. The LMP
values are different on different buses, and the congestion costs are $40.250.
• Contingency 4: The generating unit g10 (connected to bus 13) is out. The problem
becomes infeasible, and GAMS cannot find any solution for it. This is because
no solution is found which satisfies the technical constraints. Some load shedding
action should take place.
The OPF solutions for first three contingency cases are given in Table 6.10. In order
to analyze the contingency case 4, the load shedding is modeled using a virtual
generating unit connected to bus i as shown in Fig. 6.8. The production cost of this
specific unit is set to a high value. It is usually called the value of the loss of load.
The production level of this generating unit is equal to the load shedding that should
happen in bus i. The maximum power of this unit is equal to the load connected
to the bus i (0 LSi Li ). The DC-OPF problem considering the load shedding
concept is formulated in (6.6).
X X
OF D ag .Pg /2 C bg Pg C cg C VOLL LSi (6.6a)
g2˝G i
ıi ı j
Pij D ij 2 ˝` (6.6b)
xij
X X
Pg C LSi Li D Pij W i i 2 ˝B (6.6c)
g2˝Gi j2˝ì
Pmax
ij Pij Pmax
ij ij 2 ˝` (6.6d)
Pmin
g Pg Pmax
g (6.6e)
0 LSi Li (6.6f)
Table 6.10 Solution of IEEE RTS 24-bus network (generator outage contingencies)
Contingency 1 Contingency 2 Contingency 3
Pg Pg Pg
Bus (MW) ıi (rad) i ($/MW h) (MW) ıi (rad) i ($/MW h) (MW) ıi (rad) i ($/MW h)
1 152 0.246 20.93 152 0.171 20.93 152 0.186 20.93
2 152 0.248 20.93 152 0.172 20.93 152 0.187 20.93
3 0.195 20.93 0.156 20.93 0.165 20.93
4 0.283 20.93 0.204 20.93 0.219 20.93
5 0.291 20.93 0.208 20.93 0.224 20.93
6 0.331 20.93 0.244 20.93 0.261 20.93
7 0.576 20.93 300 0.058 20.7 300 0.075 20.7
8 0.499 20.93 0.165 20.93 0.182 20.93
9 0.235 20.93 0.153 20.93 0.169 20.93
10 0.275 20.93 0.184 20.93 0.202 20.93
11 0.097 20.93 0.065 20.93 0.076 20.93
12 0.082 20.93 0.044 20.93 0.071 20.93
13 464 0.000 20.93 319 0.000 20.93 514 0.000 20.93
14 0.024 20.93 0.020 20.93 0.034 20.93
15 167 0.129 20.93 12 0.095 20.93 167 0.098 20.93
16 155 0.118 20.93 155 0.099 20.93 155 0.081 20.93
17 0.200 20.93 0.175 20.93 0.165 20.93
18 400 0.226 20.93 400 0.198 20.93 400 0.191 20.93
19 0.103 20.93 0.092 20.93 0.038 20.93
20 0.126 20.93 0.122 20.93 0.038 20.93
21 400 0.240 20.93 400 0.210 20.93 400 0.206 20.93
22 300 0.348 20.93 300 0.320 20.93 300 0.314 20.93
23 660 0.152 20.93 660 0.152 20.93 310 0.051 20.93
24 0.006 20.93 0.001 20.93 0.003 20.93
Fig. 6.8 Load shedding modeling using virtual generating unit

The GAMS code for modeling and calculating the load shedding values is given
in GCode 6.5:
GCode 6.5 The OPF GAMS code for IEEE Reliability test 24-bus network with load shedding
modeling
Sets
bus /124/
s l a c k ( bus ) / 1 3 /
Gen / g1g12 /
scalars
Sbase / 1 0 0 /
VOLL / 1 0 0 0 0 / ;
t a b l e GenData ( Gen , ) G e n e r a t i n g u n i t s c h a r a c t e r i s t i c s
Pmax Pmin b
REMOVED FOR SAVING SPACE
;
t o each bus
/
/ ;
Pd Qd
;
t a b l e b r a n c h ( bus , node , ) Network t e c h n i c a l c h a r a c t e r i s t i c s

r x b limit
;
);
, bus , ’ L i m i t ’ ) ;
b r a n c h ( bus , node , ’ b i j ’ ) $ b r a n c h ( bus , node , ’ L i m i t ’ ) = 1 / b r a n c h ( bus ,
node , ’ x ’ ) ;
c o n e x ( bus , node ) $ ( b r a n c h ( bus , node , ’ l i m i t ’ ) and b r a n c h ( node , bus , ’
l i m i t ’ ) ) =1;
Variables
OF , P i j ( bus , node ) , Pg ( Gen ) , d e l t a ( b u s ) , l s h ( b u s ) ;
Equations
const1 , const2 , const3 ;
c o n s t 1 ( bus , node ) $ ( c o n e x ( bus , node ) ) . . P i j ( bus , node ) =e=
b r a n c h ( bus , node , ’ b i j ’ ) ( d e l t a ( b u s ) d e l t a ( node
));
c o n s t 2 ( b u s ) . . l s h ( b u s ) +sum ( Gen$GBconect ( bus , Gen ) , Pg ( Gen ) )
BusData ( bus , ’ pd ’ ) / S b a s e
=e =+sum ( n o d e $ c o n e x ( node , b u s
) , P i j ( bus , node ) ) ;
const3 . . OF=g=sum ( ( bus , Gen ) $GBconect ( bus , Gen ) , Pg ( Gen ) GenData

( Gen , ’ b ’ ) S b a s e ) +sum ( bus , VOLL l s h ( b u s ) S b a s e ) ;
model l o a d f l o w / const1 , const2 , const3 / ;
d e l t a . up ( b u s ) = p i / 2 ;
d e l t a . l o ( b u s )=p i / 2 ;
d e l t a . fx ( s l a c k ) =0;
Sbase ;
Sbase ;
l s h . up ( b u s ) = BusData ( bus , ’ pd ’ ) / S b a s e ;
l s h . l o ( bus ) = 0 ;
s o l v e l o a d f l o w m i n i m i z i n g OF u s i n g l p ;
r e p o r t ( bus , ’ Gen (MW) ’ ) = 1sum ( Gen$GBconect ( bus , Gen ) , Pg . l ( Gen ) )
sbase ;
r e p o r t ( bus , ’LSH ’ ) = l s h . l ( b u s ) s b a s e ;
C o n g e s t i o n c o s t = sum ( ( bus , node ) , P i j . l ( bus , node ) ( c o n s t 2 .m( b u s ) +
c o n s t 2 .m( node ) ) ) / 2 ;
display report , Pij . l , Congestioncost ;
e x e c u t e _ u n l o a d " o p f . gdx " r e p o r t
e x e c u t e ’ gdxxrw . e x e OPF . gdx o=OPF . x l s p a r = r e p o r t r n g = r e p o r t ! A1 ’
d i s p l a y GBconect ;
The OPF solution for contingency case 4 is given in Table 6.11. The simulation
results show that the demand in bus 18 should be reduced by 116 MW as given in
Table 6.11. It should be noted that the LMP values are not meaningful in this case.
The LMP values are calculated and shown in Table 6.12.
In order to calculate the LMP in a network with load shedding, the calculated
amount of load shedding should be curtailed from the demand in the specific bus
(here bus 18) and the problem should be resolved.
Nomenclature
Indices and Sets

g Index of thermal generating units.
i; j Index of network buses.
˝G Set of all thermal generating units.
˝Gi Set of all thermal generating units connected to bus i.
˝ì Set of all buses connected to bus i.
˝B Set of network buses.
˝` Set of network branches.
Table 6.11 Solution of IEEE Bus Pg (MW) ıi (rad) Li (MW) LSHi (MW)
RTS 24-bus network
(generator g10 outage with 1 152 0.095 108
load shedding modeling) 2 152 0.097 97
3 0.042 180
4 0.132 74
5 0.140 71
6 0.181 136
7 300 0.006 125
8 0.101 171
9 0.084 175
10 0.124 195
11 0.004
12 0.001
13 0.000 265
14 0.093 194
15 215 0.285 317
16 155 0.259 100
17 0.360
18 400 0.396 333 116
19 0.220 181
20 0.222 128
21 400 0.405
22 300 0.511
23 660 0.237
24 0.160
Parameters
Li Electric power demand in bus i
ag ; bg ; cg Fuel cost coefficients of thermal unit g.
max=min
Pg Maximum/minimum limits of power generation of thermal unit g.
Pmax
ij Maximum power flow limits of branch connecting bus i to j.
xij Reactance of branch connecting bus i to j.
rij Resistance of branch connecting bus i to j.
Variables
Pij Active power flow of branch connecting bus i to j.
Pg Active power generated by thermal unit g(MW).
i Locational marginal price in bus i ($/MW h).
LSi Load shedding in bus i (MW).
OF Total operating costs ($/h).
ıi Voltage angle in bus i (rad).
Table 6.12 Solution of IEEE Bus Pg (MW) ıi (rad) Li (MW) i ($/MW h)

RTS 24-bus network
(generator g10 outage with 1 152 0.095 108 26.110
load shedding modeling and 2 152 0.097 97 26.110
LMP calculation) 3 0.042 180 26.110
4 0.132 74 26.110
5 0.140 71 26.110
6 0.181 136 26.110
7 300 0.006 125 20.700
8 0.101 171 26.110
9 0.084 175 26.110
10 0.124 195 26.110
11 0.004 26.110
12 0.001 26.110
13 0.000 265 26.110
14 0.093 194 26.110
15 215 0.285 317 26.110
16 155 0.259 100 26.110
17 0.360 26.110
18 400 0.396 217 26.110
19 0.220 181 26.110
20 0.222 128 26.110
21 400 0.405 26.110
22 300 0.511 26.110
23 660 0.237 26.110
24 0.160 26.110
6.2 Multi-Period Optimal Wind-DC OPF
The load flow model considering wind power and load shedding is shown in Fig. 6.9.
The multi-period wind DC-OPF is formulated as follows:
X X
OF D ag .Pg;t /2 C bg Pg;t C cg C VOLL LSi;t C VWC Pwc
i;t (6.7a)
g;t i;t
X X
Pg;t C LSi;t C Pwi;t Li;t D Pij;t W i;t (6.7b)
g2˝Gi j2˝ì
ıi;t ıj;t
Pij;t D (6.7c)
xij
Pmax
ij Pij;t Pmax
ij (6.7d)
Pmin
g Pg;t Pmax
g (6.7e)
6.2 Multi-Period Optimal Wind-DC OPF 163
Fig. 6.9 Load flow model considering wind power and load shedding
Pg;t Pg;t1 RUg (6.7f)

Pg;t1 Pg;t RDg (6.7g)
0 LSi;t Li;t (6.7h)
Pwc w w
i;t D wt i Pi;t (6.7i)
0 Pwi;t wt wi (6.7j)
The objective function in (6.7a) consists of operating costs of thermal units, load
shedding costs, and wind curtailment costs. The nodal power balance is formulated
in (6.7b). This equation is also providing the LMP value in bus i at time t (i;t ). The
active flow in branch connecting bus i to bus j is calculated in (6.7c). The branch flow
limits of every branch is formulated in (6.7d). The operating limits of the thermal
generating unit are modeled in (6.7e). The ramp rates of thermal units are described
in (6.7f) and (6.7g). The load shedding of bus i is limited to the existing demand at
time t as stated in (6.7h). The wind power curtailment is formulated in (6.7i). At any
bus i hosting wind turbine, the amount of wind power generation depends on wind
power availability and wind power capacity as described in (6.7j). It should bear in
mind that the demand and wind power availability change vs time. These variations
are reflected in Li;t and wt , respectively. The wind power connection points to the
network are indicated in Fig. 6.10.
The wind-demand variation pattern vs time is shown in Fig. 6.11. The optimal
thermal unit power generation schedules of multi-period wind DC-OPF is given in
Table 6.13.
Fig. 6.10 Wind power connection points to the network
1 Wind
Demand
0.8
Wind-demand pattern
0.6
0.4
0.2
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 6.11 Wind-demand variation patterns vs time

Table 6.13 Thermal unit power generation schedules of multi-period wind DC-OPF
Thermal unit Bus t1 t2 t3 t4 t5 t6 t7 t8 t9 t10 t11 t12 t13 t14 t15 t16 t17 t18 t19 t20 t21 t22 t23 t24
g1 18.0 4.0 4.0 3.5 3.1 2.7 2.3 2.6 3.1 3.5 4.0 4.0 4.0 4.0 3.6 3.1 3.1 3.5 4.0 4.0 4.0 4.0 3.5 3.1 2.6
g2 21.0 4.0 3.8 3.3 2.8 2.4 2.1 2.6 3.1 3.5 4.0 4.0 4.0 4.0 3.8 3.5 3.1 3.5 4.0 4.0 4.0 4.0 3.5 3.1 3.0
g3 1.0 0.4 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.4 0.4 0.4 0.4 0.3 0.3 0.4 0.6 0.7 0.9 0.7 0.6 0.4 0.3 0.3
6.2 Multi-Period Optimal Wind-DC OPF
g4 2.0 0.4 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.4 0.4 0.3 0.3 0.3 0.3 0.4 0.6 0.7 0.7 0.7 0.6 0.4 0.3 0.3
g5 15.0 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.6 0.5 0.5 0.5 0.5 0.5 0.5 0.8 1.0 1.0 1.2 1.0 0.8 0.5 0.5
g6 16.0 0.8 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.8 1.0 1.2 1.4 1.2 1.0 0.9 1.1 1.3 1.6 1.6 1.4 1.2 1.0 0.8 0.5
g7 23.0 1.4 1.2 1.1 1.1 1.1 1.1 1.1 1.2 1.4 1.6 1.8 2.0 2.2 2.1 2.3 2.5 2.7 2.9 3.1 2.9 2.7 2.5 2.3 2.1
g8 23.0 1.7 1.4 1.4 1.4 1.4 1.4 1.4 1.4 1.7 2.0 2.2 2.5 2.3 2.1 2.3 2.6 2.9 3.2 3.3 3.0 2.7 2.4 2.1 1.9
g9 7.0 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8 1.2 1.7 1.2 1.1 0.9 0.8 0.8 0.8
g10 13.0 2.1 2.1 2.1 2.1 2.1 2.1 2.1 2.1 2.1 2.1 2.1 2.1 2.1 2.1 2.1 2.1 2.3 2.5 2.3 2.1 2.1 2.1 2.1 2.1
g11 15.0 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.2 0.3 0.2 0.1 0.1 0.1 0.1 0.1
g12 22.0 3.0 3.0 3.0 2.9 3.0 3.0 2.7 2.7 3.0 3.0 3.0 3.0 3.0 3.0 2.7 2.3 2.7 3.0 3.0 3.0 3.0 3.0 2.7 3.0
165
GCode 6.6 Multi-period DC-OPF for modified IEEE RTS 24-bus system
S e t s b u s / 1 2 4 / , s l a c k ( b u s ) / 1 3 / , Gen / g1g12 / , t / t 1 t 2 4 / ;
S c a l a r s Sbase / 1 0 0 / ,VOLL / 1 0 0 0 0 / ,VWC / 5 0 / ; a l i a s ( bus , node ) ;
t a b l e GD( Gen , ) G e n e r a t i n g u n i t s c h a r a c t e r i s t i c s
u i n i U0 So
Removed f o r s a v i n g s p a c e
S e t GB( bus , Gen ) c o n n e c t i v i t y i n d e x o f e a c h g e n e r a t i n g u n i t t o
each bus
/ Removed f o r s a v i n g s p a c e / ;
O m i t t e d f o r s a v i n g s p a c e ;
O m i t t e d f o r s a v i n g s p a c e ;
T a b l e WD( t , )
w d
t1 0.0786666666666667 0.684511335492475
t2 0.0866666666666667 0.644122690036197
t3 0.117333333333333 0.61306915602972
t4 0.258666666666667 0.599733282530006
t5 0.361333333333333 0.588874071251667
t10 0.548 0.787007048961707
t20 0.561333333333333 0.936368832158506
t21 0.565333333333333 0.887597637645266
t22 0.556 0.809297008954087
t23 0.724 0.74585635359116
t24 0.84 0.733473042484283;
P a r a m e t e r Wcap ( b u s )
/8 200
19 150
21 1 0 0 / ;
);
, bus , ’ L i m i t ’ ) ;
node , ’ x ’ ) ;
b r a n c h ( bus , node , ’ z ’ ) $ b r a n c h ( bus , node , ’ L i m i t ’ ) = s q r t ( power ( b r a n c h
( bus , node , ’ x ’ ) , 2 ) + power ( b r a n c h ( bus , node , ’ r ’ ) , 2 ) ) ;
b r a n c h ( node , bus , ’ z ’ ) = b r a n c h ( bus , node , ’ z ’ ) ;
l i m i t ’ ) ) =1;
Variables
OF , P i j ( bus , node , t ) , Pg ( Gen , t ) , d e l t a ( bus , t ) , l s h ( bus , t ) ,Pw ( bus , t ) , pc
( bus , t ) ;
Equations
const1 , const2 , const3 , const4 , const5 , const6 ;
c o n s t 1 ( bus , node , t ) $ ( c o n e x ( bus , node ) ) . . P i j ( bus , node , t ) =e=
b r a n c h ( bus , node , ’ b i j ’ ) ( d e l t a ( bus , t ) d e l t a ( node , t ) ) ;
c o n s t 2 ( bus , t ) . . l s h ( bus , t ) $BusData ( bus , ’ pd ’ )
+Pw ( bus , t ) $Wcap ( b u s ) +sum ( Gen$GB ( bus , Gen ) , Pg ( Gen , t ) )
6.2 Multi-Period Optimal Wind-DC OPF 167
WD( t , ’ d ’ ) BusData ( bus , ’ pd ’ ) / S b a s e =e =+sum ( n o d e $ c o n e x ( node , b u s ) ,

P i j ( bus , node , t ) ) ;
const3 . . OF=g=sum ( ( bus , Gen , t ) $GB ( bus , Gen ) , Pg ( Gen , t ) GD( Gen , ’ b
’ ) S b a s e )
+sum ( ( bus , t ) ,VOLL l s h ( bus , t ) S b a s e $ B u s D a t a
+VWCPc ( bus , t ) sbase$Wcap ( b u s ) ) ;
c o n s t 4 ( gen , t ) . . pg ( gen , t + 1 )pg ( gen , t ) = l =GD( gen , ’RU ’ ) / S b a s e ;
c o n s t 5 ( gen , t ) . . pg ( gen , t 1)pg ( gen , t ) = l =GD( gen , ’RD ’ ) / S b a s e ;
c o n s t 6 ( bus , t ) $Wcap ( b u s ) . . pc ( bus , t ) =e=WD( t , ’w ’ ) Wcap ( b u s ) / Sbase
pw ( bus , t ) ;
model l o a d f l o w / const1 , const2 , const3 , const4 , const5 , const6 / ;
Pg . l o ( Gen , t ) =GD( Gen , ’ Pmin ’ ) / S b a s e ; Pg . up ( Gen , t ) =GD( Gen , ’ Pmax ’ ) /
Sbase ;
d e l t a . up ( bus , t ) = p i / 2 ; d e l t a . l o ( bus , t )=p i / 2 ; d e l t a . f x ( s l a c k , t ) = 0 ;
P i j . up ( bus , node , t ) $ ( ( c o n e x ( bus , node ) ) ) =1 b r a n c h ( bus , node , ’ L i m i t ’
) / Sbase ;
P i j . l o ( bus , node , t ) $ ( ( c o n e x ( bus , node ) ) ) =1 b r a n c h ( bus , node , ’ L i m i t ’
) / Sbase ;
l s h . up ( bus , t ) = WD( t , ’ d ’ ) BusData ( bus , ’ pd ’ ) / S b a s e ; l s h . l o ( bus , t ) =
0;
Pw . up ( bus , t ) =WD( t , ’w ’ ) Wcap ( b u s ) / S b a s e ; Pw . l o ( bus , t ) = 0 ;
Pc . up ( bus , t ) =WD( t , ’w ’ ) Wcap ( b u s ) / S b a s e ; Pc . l o ( bus , t ) = 0 ;
The wind power generations at efferent buses vs time are shown in Fig. 6.12.
200
180 8
19
160 21
Wind power generation (MW)
140
120
100
80
60
40
20
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 6.12 Wind power generations at efferent buses vs time

The wind power curtailment is occurring in bus 19 at time t17 equal to 38.76
and also in bus 21 at time t16 equal to 15.32 MW and also at time t17 equal to
86.93 MW. The total operating cost (including the wind curtailment cost) would
be $4:3229 105 .
Nomenclature
Indices and Sets

g Index of thermal generating units.
w Index of wind turbine units.
i; j Index of network buses.
˝G Set of all thermal generating units.
˝Gi Set of all thermal generating units connected to bus i.
˝ì Set of all buses connected to bus i.
˝B Set of network buses.
˝` Set of network branches.
Parameters
wi;t Availability of wind turbine connected to bus i at time t
wi Capacity of wind turbine connected to bus i
Li;t Electric power demand in bus i at time t
max=min
Pg Maximum/minimum limits of power generation of thermal unit g.
Pmax
VOLL Value of loss of load ($/MW h)
VWC Value of loss of wind ($/MW h)
Variables
Pij;t Active power flow of branch connecting bus i to j at time t (MW).
Pg;t Active power generated by thermal unit g at time t (MW) .
Pwi;t Active power generated by wind turbine connected to bus i at time t (MW).
Pwc
i;t Curtailed power of wind turbine connected to bus i at time t (MW).
i;t Locational marginal price in bus i at time t ($/MW h).
LSi;t Load shedding in bus i at time t (MW).
ıi;t Voltage angle of bus i at time t (rad).
6.3 Multi-Period Optimal AC Power Flow 169
6.3 Multi-Period Optimal AC Power Flow
The AC power flow considering wind generation and load shedding is shown
in Fig. 6.13.
The optimal power flow equations are described in (6.8) as follows:
X g g
X
OF D ag .Pi;t /2 C bg Pi;t C cg C VOLL PLS wc
i;t C VWC Pi;t (6.8a)
i;t i;t
g
X p
Pi;t C PLS w L
i;t C Pi;t Pi;t D Pij;t W i;t (6.8b)
j2˝ì
g
X q
Qi;t C QLS w L
i;t C Qi;t Qi;t D Qij;t W i;t (6.8c)
j2˝ì
Vi;t †ıi;t Vj;t †ıj;t bVi;t

Iij;t D C † ıi;t C (6.8d)
Zij †ij 2 2

Sij;t D .Vi;t †ıi;t /Iij;t (6.8e)
2
˚ Vi;t Vi;t Vj;t
Pij;t D real Sij;t D cos.ij / cos.ıi;t ıj;t C ij / (6.8f)
Zij Zij
2 2
˚ Vi;t Vi;t Vj;t bVi;t
Qij;t D Img Sij;t D sin.ij / sin.ıi;t ıj;t C ij / (6.8g)
Zij Zij 2
Sijmax Sij;t Sijmax (6.8h)
g;min g g;max
Pi Pi;t Pi (6.8i)
g;min g g;max
Qi Qi;t Qi (6.8j)
Fig. 6.13 AC power flow considering wind generation and load shedding
g g
Pi;t Pi;t1 RUg (6.8k)
g g
Pi;t1 Pi;t RDg (6.8l)
0 PLS L
i;t Pi;t (6.8m)
0 QLS L
i;t Qi;t (6.8n)
Pwc w w
i;t D wt i Pi;t (6.8o)
0 Pwi;t wt wi (6.8p)
The GAMS code developed for solving the minimum cost OPF is given in
GCode 6.7. The optimal active/reactive power generation of thermal units in MP-
AC OPF are given in Figs. 6.14 and 6.15, respectively.
400
Active power generation (MVAr)
350
300
250 1
2
200 7
13
150 15
16
18
100
21
22
50 23
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 6.14 Active power generation of thermal units in MP-AC OPF
200
1 2 7 13 15 16 18 21 22 23
Reactive power generation (MVAr)
150
100
50
-50
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 6.15 Reactive power generation of thermal units in MP-AC OPF

6.3 Multi-Period Optimal AC Power Flow 171
GCode 6.7 Multi-period AC OPF for modified IEEE RTS 24 bus system
S e t s i network buses /124/ , s l a c k ( i ) / 1 3 / , t / t1t24 /

GB( i ) g e n e r a t i n g b u s e s / 1 , 2 , 7 , 1 5 1 6 , 1 8 , 2 1 2 3 / ;
s c a l a r s Sbase / 1 0 0 / ,VOLL / 1 0 0 0 0 / ,VWC / 5 0 / ; a l i a s ( i , j ) ;
T a b l e GenD ( i , ) G e n e r a t i n g u n i t s c h a r a c t e r i s t i c s
pmax pmin b Qmax Qmin Vg RU RD
1 152 3 0 . 4 1 3 . 3 2 192 50 1 . 0 3 5 21 21
2 152 3 0 . 4 1 3 . 3 2 192 50 1 . 0 3 5 21 21
7 350 75 2 0 . 7 300 0 1 . 0 2 5 43 43
13 591 2 0 6 . 8 5 2 0 . 9 3 591 0 1 . 0 2 31 31
15 215 6 6 . 2 5 21 215 100 1 . 0 1 4 31 31
16 155 5 4 . 2 5 1 0 . 5 2 155 50 1 . 0 1 7 31 31
18 400 100 5 . 4 7 400 50 1 . 0 5 70 70
21 400 100 5 . 4 7 400 50 1 . 0 5 70 70
22 300 0 0 300 60 1 . 0 5 53 53
23 360 2 4 8 . 5 1 0 . 5 2 310 125 1 . 0 5 31 31 ;
T a b l e BD( i , ) Demands o f e a c h b u s i n MW
Pd Qd
removed f o r s a v i n g s p a c e ;
t a b l e LN( i , j , ) Network t e c h n i c a l c h a r a c t e r i s t i c s
r x b limit
T a b l e WD( t , )
w d
p a r a m e t e r Wcap ( i )
/8 200
19 150
21 1 0 0 / ;
LN( i , j , ’ x ’ ) $ (LN( i , j , ’ x ’ ) = 0 ) =LN( j , i , ’ x ’ ) ; LN( i , j , ’ r ’ ) $ (LN( i , j , ’ r ’ )
= 0 ) =LN( j , i , ’ r ’ ) ;
LN( i , j , ’ b ’ ) $ (LN( i , j , ’ b ’ ) = 0 ) =LN( j , i , ’ b ’ ) ; LN( i , j , ’ b i j ’ ) $LN ( i , j , ’
L i m i t ’ ) = 1 /LN( i , j , ’ x ’ ) ;
LN( i , j , ’ L i m i t ’ ) $ (LN( i , j , ’ L i m i t ’ ) = 0 ) =LN( j , i , ’ L i m i t ’ ) ;
LN( i , j , ’ z ’ ) $LN ( i , j , ’ L i m i t ’ ) = s q r t ( power (LN( i , j , ’ x ’ ) , 2 ) + power (LN( i ,
j , ’ r ’ ) ,2) ) ;
LN( j , i , ’ z ’ ) $ (LN( i , j , ’ z ’ ) = 0 ) =LN( i , j , ’ z ’ ) ;
LN( i , j , ’ t h ’ ) $ (LN( i , j , ’ L i m i t ’ ) and LN( i , j , ’ x ’ ) and LN( i , j , ’ r ’ ) )
= a r c t a n (LN( i , j , ’ x ’ ) / ( LN( i , j , ’ r ’ ) ) )
;
LN( i , j , ’ t h ’ ) $ (LN( i , j , ’ L i m i t ’ ) and LN( i , j , ’ x ’ ) and LN( i , j , ’ r ’ ) = 0 )
=pi / 2 ;
LN( i , j , ’ t h ’ ) $ (LN( i , j , ’ L i m i t ’ ) and LN( i , j , ’ r ’ ) and LN( i , j , ’ x ’ ) = 0 )
=0;
LN( j , i , ’ t h ’ ) $LN ( i , j , ’ L i m i t ’ ) =LN( i , j , ’ t h ’ ) ; P a r a m e t e r cx ( i , j ) ;
cx ( i , j ) $ (LN( i , j , ’ l i m i t ’ ) and LN( j , i , ’ l i m i t ’ ) ) = 1 ; cx ( i , j ) $ ( cx ( j , i ) )
=1;
V a r i a b l e s OF , P i j ( i , j , t ) , Q i j ( i , j , t ) , Pg ( i , t ) , Qg ( i , t ) , Va ( i , t ) ,V( i , t )
,Pw ( i , t ) ;
E q u a t i o n s eq1 , eq2 , eq3 , eq4 , eq5 , eq6 , eq7 ;
eq1 ( i , j , t ) $cx ( i , j ) . . P i j ( i , j , t ) =e = (V( i , t ) V( i , t ) c o s (LN( j , i , ’ t h ’ ) )
V( i , t ) V( j , t ) c o s ( Va ( i , t )Va ( j , t ) +LN( j , i , ’ t h ’ ) )
) / LN( j , i , ’ z ’ ) ;
eq2 ( i , j , t ) $cx ( i , j ) . . Q i j ( i , j , t ) =e = (V( i , t ) V( i , t ) s i n (LN( j , i , ’ t h ’ ) )

V( i , t ) V( j , t ) s i n ( Va ( i , t )Va ( j , t ) +LN( j , i , ’ t h ’ ) )
) / LN( j , i , ’ z ’ )
LN( j , i , ’ b ’ ) V(
i , t ) V( i , t )
/2;
eq3 ( i , t ) . . Pw ( i , t ) $Wcap ( i ) +Pg ( i , t ) $GenD ( i , ’ Pmax ’ )WD( t , ’ d ’ ) BD( i
, ’ pd ’ ) / S b a s e =e=
+sum ( j $ c x ( j , i ) , P i j ( i , j , t ) ) ;
eq4 ( i , t ) . . Qg ( i , t ) $GenD ( i , ’Qmax ’ )WD( t , ’ d ’ ) BD( i , ’
qd ’ ) / S b a s e =e=
+sum ( j $ c x ( j , i ) , Q i j ( i , j , t ) ) ;
eq5 . . OF=g=sum ( ( i , t ) , Pg ( i , t ) GenD ( i , ’ b ’ ) Sbase$GenD ( i , ’
Pmax ’ ) ) ;
eq6 ( i , t ) $ ( GenD ( i , ’ Pmax ’ ) and o r d ( t ) >1) . . Pg ( i , t )Pg ( i , t 1)= l =GenD
( i , ’RU ’ ) / S b a s e ;
eq7 ( i , t ) $ ( GenD ( i , ’ Pmax ’ ) and o r d ( t ) < c a r d ( t ) ) . . Pg ( i , t )Pg ( i , t + 1 ) =
l=
GenD ( i , ’RD ’ ) / S b a s e ;
Model l o a d f l o w / eq1 , eq2 , eq3 , eq4 , eq5 , eq6 , eq7 / ;
Pg . l o ( i , t ) =GenD ( i , ’ Pmin ’ ) / S b a s e ; Pg . up ( i , t ) =GenD ( i , ’ Pmax ’ ) / S b a s e ;
Qg . l o ( i , t ) =GenD ( i , ’ Qmin ’ ) / S b a s e ; Qg . up ( i , t ) =GenD ( i , ’Qmax ’ ) / S b a s e ;
Va . up ( i , t ) = p i / 2 ; Va . l o ( i , t )=p i / 2 ; Va . l ( i , t ) = 0 ; Va . f x ( s l a c k , t ) = 0 ;
P i j . up ( i , j , t ) $ ( ( cx ( i , j ) ) ) =+1LN( i , j , ’ L i m i t ’ ) / S b a s e ;
P i j . l o ( i , j , t ) $ ( ( cx ( i , j ) ) ) =1LN( i , j , ’ L i m i t ’ ) / S b a s e ;
Q i j . up ( i , j , t ) $ ( ( cx ( i , j ) ) ) =+1LN( i , j , ’ L i m i t ’ ) / S b a s e ;
Q i j . l o ( i , j , t ) $ ( ( cx ( i , j ) ) ) =1LN( i , j , ’ L i m i t ’ ) / S b a s e ; V . l o ( i , t )
= 0 . 9 ; V . up ( i , t ) = 1 . 1 ;
Pw . up ( i , t ) =WD( t , ’ d ’ ) Wcap ( i ) / s b a s e ; Pw . l o ( i , t ) = 0 ;
S o l v e l o a d f l o w m i n i m i z i n g OF u s i n g n l p ;
Nomenclature
Parameters
wi;t Availability of wind turbine connected to bus i at time t
wi Capacity of wind turbine connected to bus i
PLi;t Active power component of demand in bus i at time t
Pwi;t Active power generation by wind turbine connected to bus i at time t
Qwi;t Reactive power generation by wind turbine connected to bus i at time t
g;max=min
Pi Maximum/minimum limits of power generation of thermal unit g
connected to bus i.
Pmax
References 173
VWC Value of loss of wind ($/MW h)

b Total line charging susceptance of branch connecting bus i to j (pu).
Variables
Sij;t Apparent power flow of branch connecting bus i to j at time t.
g
Pi;t Active power generated by thermal unit g connected to bus i at time t (MW).
Pwi;t Active power generated by wind turbine connected to bus i at time t (MW).
p
i;t Active locational marginal price (LMP) in bus i at time t ($/MW h).
PLS
i;t Active Load shedding in bus i at time t (MW).
Pwc
i;t Curtailed power of wind turbine connected to bus i at time t (MW).
Iij;t Current flow of branch connecting bus i to j at time t.
QLSi;t Reactive Load shedding in bus i at time t (MVAr).
g
Qi;t Reactive power generated by thermal unit g connected to bus i at time t
(MW).
q
i;t Reactive locational marginal price (LMP) in bus i at time t ($/MVArh).
ıi;t Voltage angle in bus i at time t (rad).
Vi;t Voltage magnitude in bus i at time t (pu).
References
1. A. Meeraus A. Brooke, D. Kendrick, R. Raman, GAMS/Cplex 7.0 User Notes. GAMS

Development Corp. (2000)
2. A.J. Wood, B.F. Wollenberg, Power Generation, Operation, and Control (Wiley, Hoboken,
2012)
4. F. Li, R. Bo Small test systems for power system economic studies, in: IEEE PES General
Meeting (2010), pp. 1–4
planning, and analysis tools for power systems research and education. IEEE Trans. Power Syst.
26(1), 12–19 (2011)
7. F. Bouffard, F.D. Galiana, A.J. Conejo, Market-clearing with stochastic security-part II: case
studies. IEEE Trans. Power Syst. 20(4), 1827–1835 (2005)
Chapter 7
Energy Storage Systems
This chapter provides a solution for operation and planning aspects of energy
storage systems (ESS) problem in GAMS. The ESS integration has been analyzed
in operation and planning horizon. The inputs are generator’s characteristics,
electricity prices, demands, and network topology. The outputs of this code are
operating schedules of ESS as well as the investment decisions.
7.1 Introduction
The energy storage system (ESS) is an attractive option to increase the flexibility
of energy system operation and planning. These units can absorb energy in
case of low electricity price or excessive generation and return it back in high
price/low generation periods. Different technologies and objective functions have
been reported in the literature.
Some of the ESS technologies are as follows:
• Superconducting magnetic energy storage system (SMES) [1]
• Compressed Air Energy Storage (CAES) [2]
• Super-capacitor Energy Storage [3]
• Pumped Hydro Storage [4]
• Battery energy storage [5]
• Flywheel Energy Storage System [6]
• Power to gas storage method [7]
Some of the objective functions in ESS studies are listed as follows:
• Compensate grid voltage fluctuations [1]
• Overcome the Destabilizing Effect of Instantaneous Constant Power Loads in
DC Microgrids [3]
• Prevention of Transient Under-Frequency Load Shedding [8]

DOI 10.1007/978-3-319-62350-4_7
176 7 Energy Storage Systems
• Reliability enhancement [9]

• Wind uncertainty management [10]
• Fault Ride Through Support of Grid-Connected VSC HVDC-Based Offshore
Wind Farms [6]
• Phase balancing [11]
• Wind curtailment reduction and congestion management [12]
• Active power loss payment minimization [13]
7.2 ESS Operation
7.2.1 ESS Operation in DED
The total operating costs without using ESS are TC D $6:4796 105 as obtained in
GCode 4.1. The DED problem integrated with ESS is formulated in (7.1).
X
min TC D ag P2g;t C bg Pg;t C cg (7.1a)
Pg;t ;SOCt ;Pdt ;Pct g;t
Pmin
g Pg;t Pmax
g (7.1b)
Pg;t Pg;t1 RUg (7.1c)
Pg;t1 Pg;t RDg (7.1d)

SOCt D SOCt1 C Pct c Pdt =d t (7.1e)
Pcmin Pct Pcmax (7.1f)
Pdmin Pdt Pdmax (7.1g)
SOCmin SOCt SOCmax (7.1h)
X
Pg;t C Pdt Lt Pct (7.1i)
g
The simulation parameters for this problem are given in Table 7.1. The economic
dispatch data for four units example is given in Table 7.2. This table has 11 columns.
Table 7.1 The data for ESS Parameter Value

integrated dynamic economic
dispatch SOC0 100 MW
SOCmax 300 MW
Pdmax 0.2 SOCmax
Pdmin 0
Pcmax 0.2 SOCmax
Pcmin 0
c 95%
d 90%
7.2 ESS Operation
Table 7.2 The dynamic economic dispatch data for four thermal units example
Unit ag ($=MW2 ) bg ($=MW) cg ($) dg (kg=MW2 ) eg (kg=MW) fg .kg/ Pmin
g (MW) Pmax
g (MW) RUg (MW) RDg (MW)
g1 0.12 14.8 89 1.2 5 3 28 200 40 40
g2 0.17 16.57 83 2.3 4.24 6.09 20 290 30 30
g3 0.15 15.55 100 1.1 2.15 5.69 30 190 30 30
g4 0.19 16.21 70 1.1 3.99 6.2 20 260 50 50
177
The first column is showing the generating unit index. The next three columns
indicate the cost coefficients for these thermal units (ag ; bg ; cg ). The next three
columns indicate the emission coefficients for these thermal units (dg ; eg ; fg ). The
next two columns give the minimum and maximum generating limits of each unit if
they ton. The last two columns indicate the ramp up/down rates of thermal units.
The GAMS code for solving the example (7.1) is given in Code 7.1.
GCode 7.1 The cost-based dynamic economic dispatch integrated with ESS (7.1)
Set t hours / t1t24 /
Set g thermal units / p1p4 / ;
Table gendata ( g ,) g e n e r a t o r c o s t c h a r a c t e r i s t i c s and l i m i t s

p1 0 . 1 2 1 4 . 8 0 89 1 . 2 5 3 28 200 40 40
p2 0 . 1 7 1 6 . 5 7 83 2 . 3 4.24 6 . 0 9 20 290 30 30
p3 0 . 1 5 1 5 . 5 5 100 1 . 1 2.15 5 . 6 9 30 190 30 30
p4 0 . 1 9 1 6 . 2 1 70 1 . 1 3.99 6 . 2 20 260 50 50;
Table data ( t ,)

lambda l o a d wind
t1 3 2 . 7 1 510 44.1
t2 3 4 . 7 2 530 48.5
t3 3 2 . 7 1 516 65.7
t4 3 2 . 7 4 510 144.9
t5 3 2 . 9 6 515 202.3
t6 3 4 . 9 3 544 317.3
t7 44.9 646 364.4
t8 52 686 317.3
t9 5 3 . 0 3 741 271
t 1 0 4 7 . 2 6 734 306.9
t 1 1 4 4 . 0 7 748 424.1
t 1 2 3 8 . 6 3 760 398
t 1 3 3 9 . 9 1 754 487.6
t 1 4 3 9 . 4 5 700 521.9
t 1 5 4 1 . 1 4 686 541.3
t 1 6 3 9 . 2 3 720 560
t 1 7 5 2 . 1 2 714 486.8
t 1 8 4 0 . 8 5 761 372.6
t19 41.2 727 367.4
t 2 0 4 1 . 1 5 714 314.3
t 2 1 4 5 . 7 6 618 316.6
t 2 2 4 5 . 5 9 584 311.4
t 2 3 4 5 . 5 6 578 405.4
t 2 4 3 4 . 7 2 544 470.4;

p(g , t ) Power g e n e r a t e d by t h e r m a l power
plant
SOC( t ) , Pd ( t ) , Pc ( t ) ;
p . up ( g , t ) = g e n d a t a ( g , ” Pmax ” ) ;
p . l o ( g , t ) = g e n d a t a ( g , ” Pmin ” ) ;
s c a l a r SOC0 / 1 0 0 / , SOCmax / 3 0 0 / , e t a c / 0 . 9 5 / , e t a d / 0 . 9 / ;
7.2 ESS Operation 179
SOC . up ( t ) =SOCmax ; SOC . l o ( t ) =0.2SOCmax ; SOC . f x ( ’ t 2 4 ’ ) =SOC0 ;

Pc . up ( t ) =0.2SOCmax ; Pc . l o ( t ) = 0 ; Pd . up ( t ) =0.2SOCmax ; Pd . l o ( t ) = 0 ;
E q u a t i o n s Genconst3 , Genconst4 , c o s t T h e r m a l c a l c , balance , constESS1 ;
c o s t T h e r m a l c a l c . . c o s t T h e r m a l =e=sum ( ( t , g ) , g e n d a t a ( g , ’ a ’ ) power
(p(g , t ) ,2)
+ g e n d a t a ( g , ’ b ’ ) p ( g , t ) + g e n d a t a ( g , ’ c ’ ) ) ;
constESS1 ( t ) . . SOC( t ) =e=SOC0$ ( o r d ( t ) = 1 ) + SOC( t 1)$ ( o r d ( t ) >1)
+Pc ( t ) e t a c Pd ( t ) /
eta d ;
b a l a n c e ( t ) . . sum ( g , p ( g , t ) ) +Pd ( t ) =g= d a t a ( t , ’ l o a d ’ ) +Pc ( t ) ;
Model DEDESScostbased / a l l / ;
S o l v e DEDESScostbased u s qcp min c o s t T h e r m a l ;
The total operating costs with using ESS are TC D $6:4553 105 as obtained in
GCode 7.1. The optimal solution for ESS integrated dynamic economic dispatch is
given in Table 7.3. The hourly dispatch of ESS in DED problem is shown in Fig. 7.1.
The hourly schedules of thermal units in DED-ESS problem are shown in Fig. 7.2.
Table 7.3 The ESS integrated dynamic economic dispatch

Time Lt (MW) Pg;t (MW) SOCt (MW h) Pdt (MW) Pct (MW)
t1 510.00 555.92 143.62 0.00 45.92
t2 530.00 555.92 168.25 0.00 25.92
t3 516.00 555.92 206.18 0.00 39.92
t4 510.00 555.92 249.80 0.00 45.92
t5 515.00 555.92 288.68 0.00 40.92
t6 544.00 555.92 300.00 0.00 11.92
t7 646.00 646.00 300.00 0.00 0.00
t8 686.00 686.00 300.00 0.00 0.00
t9 741.00 716.28 272.53 24.72 0.00
t10 734.00 716.28 252.83 17.72 0.00
t11 748.00 716.28 217.58 31.72 0.00
t12 760.00 716.28 169.00 43.72 0.00
t13 754.00 716.28 127.08 37.72 0.00
t14 700.00 700.00 127.08 0.00 0.00
t15 686.00 686.00 127.08 0.00 0.00
t16 720.00 716.28 122.95 3.72 0.00
t17 714.00 714.00 122.95 0.00 0.00
t18 761.00 716.28 73.25 44.72 0.00
t19 727.00 716.28 61.34 10.72 0.00
t20 714.00 712.80 60.00 1.20 0.00
t21 618.00 618.00 60.00 0.00 0.00
t22 584.00 584.00 60.00 0.00 0.00
t23 578.00 582.05 63.85 0.00 4.05
t24 544.00 582.05 100.00 0.00 38.05
800
Power/energy
600
Total power by thermal units
400 Demand
SOC t
200
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Charge/discharge (MW)
60 P ct
P dt
40
20
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 7.1 The hourly dispatch of ESS in DED problem
210
200
190
180
170
(MW)
160
g,t
150
P
140
g1
130 g2
g3
120
g4
110
100
0 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
Time (h)
Fig. 7.2 The hourly schedules of thermal units in DED-ESS problem

7.2.2 ESS Operation in Wind-DED Problem
The role of ESS in efficient utilization of wind energy is investigated in this section.
The ESS-wind-DED is formulated as follows:
X X
min TC D ag P2g;t C bg Pg;t C cg C VWC Pwc t (7.2a)
DV
g;t t
˚
DV D Pg;t ; SOCt ; Pdt ; Pct ; Pwt ; Pwc
t (7.2b)
Pmin
g Pg;t Pmax
g (7.2c)

SOCt D SOCt1 C Pct c Pdt =d t (7.2f)
Pcmin Pct Pcmax (7.2g)
Pdmin Pdt Pdmax (7.2h)
SOCmin SOCt SOCmax (7.2i)
X
Pwt C Pg;t C Pdt Lt Pct (7.2j)
g
Pwt C Pwc w
t t (7.2k)
GCode 7.2 The cost-based wind-DED integrated with ESS (7.2)

Set t / t1t24 / , g / p1p4 / ;
Table gendata ( g ,) g e n e r a t o r c o s t c h a r a c t e r i s t i c s and l i m i t s
p1 0 . 1 2 1 4 . 8 0 89 1 . 2 5 3 28 200 40 40
p2 0 . 1 7 1 6 . 5 7 83 2 . 3 4.24 6 . 0 9 20 290 30 30
p3 0 . 1 5 1 5 . 5 5 100 1 . 1 2.15 5 . 6 9 30 190 30 30
p4 0 . 1 9 1 6 . 2 1 70 1 . 1 3.99 6 . 2 20 260 50 50;
table data ( t ,)
lambda l o a d wind
t1 3 2 . 7 1 510 4 4 . 1
t2 3 4 . 7 2 530 4 8 . 5
t3 3 2 . 7 1 516 6 5 . 7
t4 3 2 . 7 4 510 1 4 4 . 9
t5 3 2 . 9 6 515 2 0 2 . 3
t6 3 4 . 9 3 544 3 1 7 . 3
t7 44.9 646 3 6 4 . 4
t8 52 686 3 1 7 . 3
t9 5 3 . 0 3 741 271
t 1 0 4 7 . 2 6 734 3 0 6 . 9
t 1 1 4 4 . 0 7 748 4 2 4 . 1
t 1 2 3 8 . 6 3 760 398
t 1 3 3 9 . 9 1 754 4 8 7 . 6
t 1 4 3 9 . 4 5 700 5 2 1 . 9
t 1 5 4 1 . 1 4 686 5 4 1 . 3
t 1 6 3 9 . 2 3 720 560
t 1 7 5 2 . 1 2 714 4 8 6 . 8
t 1 8 4 0 . 8 5 761 3 7 2 . 6
t19 41.2 727 3 6 7 . 4
t 2 0 4 1 . 1 5 714 3 1 4 . 3
t 2 1 4 5 . 7 6 618 3 1 6 . 6
t 2 2 4 5 . 5 9 584 3 1 1 . 4
t 2 3 4 5 . 5 6 578 4 0 5 . 4
t 2 4 3 4 . 7 2 544 4 7 0 . 4 ;

variables cost Cost of thermal u n i t s
p(g , t ) Power g e n e r a t e d by t h e r m a l power
plant
SOC( t ) , Pd ( t ) , Pc ( t ) ,Pw ( t ) ,PWC( t ) ;
p . up ( g , t ) = g e n d a t a ( g , ” Pmax ” ) ; p . l o ( g , t ) = g e n d a t a ( g , ” Pmin ” ) ;
s c a l a r SOC0 / 1 0 0 / , SOCmax / 3 0 0 / , e t a c / 0 . 9 5 / , e t a d / 0 . 9 / , VWC
/50/;
SOC . up ( t ) =SOCmax ; SOC . l o ( t ) =0.2SOCmax ;
SOC . f x ( ’ t 2 4 ’ ) =SOC0 ; Pc . up ( t ) =0.2SOCmax ;
Pc . l o ( t ) = 0 ; Pd . up ( t ) =0.2SOCmax ; Pd . l o ( t ) = 0 ;
Pw . up ( t ) = d a t a ( t , ’ wind ’ ) ; Pw . l o ( t ) = 0 ;
Pwc . up ( t ) = d a t a ( t , ’ wind ’ ) ; Pwc . l o ( t ) = 0 ;
E q u a t i o n s G e n c o n s t 3 , G e n c o n s t 4 , c o s t T h e r m a l c a l c , constESS , b a l a n c e ,
wind ;
c o s t T h e r m a l c a l c . . c o s t =e=sum ( t ,VWCpwc ( t ) )
+sum ( ( t , g ) , g e n d a t a ( g , ’ a ’ ) power ( p ( g , t ) , 2 ) + g e n d a t a ( g , ’ b ’ )
p (g , t ) +gendata (g , ’c ’ ) ) ;
c o n s t E S S ( t ) . . SOC( t ) =e=SOC0$ ( o r d ( t ) = 1 ) + SOC( t 1)$ ( o r d ( t ) >1)+Pc ( t )
e t a c Pd ( t ) / e t a d ;
b a l a n c e ( t ) . . Pw ( t ) + sum ( g , p ( g , t ) ) +Pd ( t ) =g= d a t a ( t , ’ l o a d ’ ) +Pc ( t ) ;
wind ( t ) . . Pw ( t ) +PWC( t ) =e= d a t a ( t , ’ wind ’ ) ;
Model DEDESSwind / a l l / ;
S o l v e DEDESSwind u s qcp min c o s t ;
parameter rep ( t ,) ;
r e p ( t , ’ P t h ’ ) =sum ( g , p . l ( g , t ) ) ;
r e p ( t , ’SOC ’ ) =SOC . l ( t ) ;
r e p ( t , ’ Pd ’ ) =Pd . l ( t ) ;
r e p ( t , ’ Pc ’ ) =Pc . l ( t ) ;
r e p ( t , ’ Load ’ ) = d a t a ( t , ’ l o a d ’ ) ;
e x e c u t e u n l o a d ”DEDESSwind . gdx ” P . l
e x e c u t e ’ gdxxrw . e x e DEDESSwind . gdx v a r =P r n g = P t h e r m a l ! a1 ’
e x e c u t e u n l o a d ”DEDESSwind . gdx ” r e p
e x e c u t e ’ gdxxrw . e x e DEDESSwind . gdx p a r = r e P r n g = r e p ! a1 ’
w
Demand Total power by thermal units SOC t Pt
800
600
(MW)
400
200
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
60 P dt
P ct
(MW)
40
20
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 7.3 The hourly dispatch of ESS in wind-DED problem
160
g1
140 g2
g3
g4
120
P g,t (MW)
100
80
60
40
20
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 7.4 The hourly schedules of thermal units in wind-DED-ESS problem
The total operating costs using ESS are TC D $2:2336 105 as obtained
in GCode 7.2. The hourly dispatch of ESS in wind-DED problem is depicted in
Fig. 7.3. The hourly schedules of thermal units in wind-DED-ESS problem are
shown in Fig. 7.4.
7.2.3 ESS Operation in DC-OPF
The multi-period ESS integrated DC-OPF is formulated as follows:

X X
i;t (7.3a)
g;t i;t
X X
Pg;t C LSi;t C Pwi;t Li;t Pci;t C Pdi;t D Pij;t W i;t (7.3b)
g2˝Gi j2˝ì
ıi;t ıj;t
Pij;t D (7.3c)
Xij
Pmax
ij Pij;t Pmax
ij (7.3d)
Pmin
g Pg;t Pmax
g (7.3e)
0 LSi;t Li;t (7.3h)
Pwc w w
i;t D wi;t i Pi;t (7.3i)

SOCi;t D SOCi;t1 C Pci;t c Pdi;t =d t (7.3k)
Pci;min Pci;t Pci;max (7.3l)
Pdi;min Pdi;t Pdi;max (7.3m)
SOCi;min SOCi;t SOCi;max (7.3n)
The model is implemented on IEEE RTS 24-bus network with ESS integration
which is shown in Fig. 7.5.
It is a transmission network with the voltage levels of 138 kV, 230 kV, and
Sbase = 100 MVA. The branch data for IEEE RTS 24-bus network is given in
Table 7.4 [14]. The from bus, to bus, reactance (X), resistance (r), total line charging
susceptance (b), and MVA rating (MVA) are specified in this table. The parallel
lines in MATPOWER are merged, and the resultants are given in Table 7.4. The
generation data for IEEE RTS 24-bus network is given in Table 7.5. The data of
generating units in this network is inspired by Conejo et al. [15] and Bouffard et al.
[16] with some modifications. The slack bus is bus 13 in this network.
It is assumed that the network is intact and all branches and generating units are
working in normal condition. The SOC of ESS connected to bus 19 and 21 are 200
Fig. 7.5 The IEEE 24 RTS with ESS integration
and 100 MW h, respectively. The rest of ESS data is the same as Table 7.1. The
minimum operating cost is $4:1839 105 obtained by using the GCode 7.3.
GCode 7.3 The ESS-DCOPF GAMS code for IEEE Reliability test 24-bus network
s c a l a r s Sbase / 1 0 0 / ,VOLL / 1 0 0 0 0 / ,VOLW / 5 0 / ; a l i a s ( bus , node ) ;
T a b l e GD( Gen , ) G e n e r a t i n g u n i t s c h a r a c t e r i s t i c s
u i n i U0 So
The same a s b e f o r e
S e t GB( bus , Gen ) c o n n e c t i v i t y i n d e x o f e a c h g e n e r a t i n g u n i t t o
each bus
/ The same a s b e f o r e / ;
The same a s b e f o r e ;
T a b l e WD( t , )
w d
p a r a m e t e r Wcap ( b u s )
/8 200
19 150
21 1 0 0 / ;
p a r a m e t e r SOCMax ( b u s )
/ 1 9 200
21 1 0 0 / ;
s c a l a r e t a c / 0 . 9 5 / , e t a d / 0 . 9 / , VWC / 5 0 / ;
p a r a m e t e r SOC0 ( b u s ) ;
186

From To r(pu) x(pu) b(pu) Rating (MVA) From To r(pu) x(pu) b(pu) Rating (MVA)
1 2 0.0026 0.0139 0.4611 175 11 13 0.0061 0.0476 0.0999 500
1 3 0.0546 0.2112 0.0572 175 11 14 0.0054 0.0418 0.0879 500
1 5 0.0218 0.0845 0.0229 175 12 13 0.0061 0.0476 0.0999 500
2 4 0.0328 0.1267 0.0343 175 12 23 0.0124 0.0966 0.2030 500
2 6 0.0497 0.1920 0.0520 175 13 23 0.0111 0.0865 0.1818 500
3 9 0.0308 0.1190 0.0322 175 14 16 0.0050 0.0389 0.0818 500
3 24 0.0023 0.0839 0.0000 400 15 16 0.0022 0.0173 0.0364 500
4 9 0.0268 0.1037 0.0281 175 15 21 0.0032 0.0245 0.2060 1000
5 10 0.0228 0.0883 0.0239 175 15 24 0.0067 0.0519 0.1091 500
6 10 0.0139 0.0605 2.4590 175 16 17 0.0033 0.0259 0.0545 500
7 8 0.0159 0.0614 0.0166 175 16 19 0.0030 0.0231 0.0485 500
8 9 0.0427 0.1651 0.0447 175 17 18 0.0018 0.0144 0.0303 500
8 10 0.0427 0.1651 0.0447 175 17 22 0.0135 0.1053 0.2212 500
9 11 0.0023 0.0839 0.0000 400 18 21 0.0017 0.0130 0.1090 1000
9 12 0.0023 0.0839 0.0000 400 19 20 0.0026 0.0198 0.1666 1000
10 11 0.0023 0.0839 0.0000 400 20 23 0.0014 0.0108 0.0910 1000
10 12 0.0023 0.0839 0.0000 400 21 22 0.0087 0.0678 0.1424 500
7 Energy Storage Systems
7.2 ESS Operation

Gen Bus Pmax
i Pmin
i bi ($/MW) Csi ($) Cdi ($) RUi RDi (MW h1 ) SUi (MW h1 ) SDi (MW h1 ) UTi (h) DTi (h) ui;tD0 Ui0 (h) Si0 (h)
(MW h1 )
g1 18 400 100 5.47 0 0 47 47 105 108 1 1 1 5 0
g2 21 400 100 5.47 0 0 47 47 106 112 1 1 1 6 0
g3 1 152 30.4 13.32 1430.4 1430.4 14 14 43 45 8 4 1 2 0
g4 2 152 30.4 13.32 1430.4 1430.4 14 14 44 57 8 4 1 2 0
g5 15 155 54.25 16 0 0 21 21 65 77 8 8 0 0 2
g6 16 155 54.25 10.52 312 312 21 21 66 73 8 8 1 10 0
g7 23 310 108.5 10.52 624 624 21 21 112 125 8 8 1 10 0
g8 23 350 140 10.89 2298 2298 28 28 154 162 8 8 1 5 0
g9 7 350 75 20.7 1725 1725 49 49 77 80 8 8 0 0 2
g10 13 591 206.85 20.93 3056.7 3056.7 21 21 213 228 12 10 0 0 8
g11 15 60 12 26.11 437 437 7 7 19 31 4 2 0 0 1
g12 22 300 0 0 0 0 35 35 315 326 0 0 1 2 0
187
SOC0 ( b u s ) =0.2SOCMax ( b u s ) / s b a s e ;
b r a n c h ( bus , node , ’ x ’ ) $ ( b r a n c h ( bus , node , ’ x ’ ) = 0 ) = b r a n c h ( node ,
bus , ’ x ’ ) ;
b r a n c h ( bus , node , ’ L i m i t ’ ) $ ( b r a n c h ( bus , node , ’ L i m i t ’ ) = 0 ) = b r a n c h
( node , bus , ’ L i m i t ’ ) ;
node , ’ x ’ ) ;
b r a n c h ( bus , node , ’ z ’ ) $ b r a n c h ( bus , node , ’ L i m i t ’ ) =
s q r t ( power ( b r a n c h ( bus , node , ’ x ’ ) , 2 ) + power ( b r a n c h ( bus , node ,
’ r ’ ) ,2) ) ;
l i m i t ’ ) ) =1;
V a r i a b l e s OF , P i j ( bus , node , t ) , Pg ( Gen , t ) , d e l t a ( bus , t ) , l s h ( bus , t ) ,Pw
( bus , t ) , pwc ( bus , t ) ,
SOC( bus , t ) , Pd ( bus , t ) , Pc ( bus , t ) ;
E q u a t i o n s const1 , const2 , const3 , const4 , const5 , const6 , constESS ;
c o n s t 2 ( bus , t ) . . l s h ( bus , t ) $BusData ( bus , ’ pd ’ ) +Pw ( bus , t ) $Wcap ( b u s )
+sum ( Gen$GB ( bus , Gen ) , Pg ( Gen , t ) )
+(Pc ( bus , t ) +Pd ( bus , t ) )$SOCMAX( b u s )WD( t , ’ d ’ ) BusData ( bus , ’ pd ’ ) /
S b a s e =e=
+sum ( n o d e $ c o n e x ( node , b u s ) , P i j ( bus , node , t ) ) ;
const3 . . OF=g=sum ( ( bus , Gen , t ) $GB ( bus , Gen ) , Pg ( Gen , t ) GD( Gen , ’ b
’ ) S b a s e )
+sum ( ( bus , t ) ,VOLL l s h ( bus , t ) S b a s e $ B u s D a t a ( bus , ’ pd ’ )
+VOLWPwc ( bus , t ) sbase$Wcap ( b u s ) ) ;
c o n s t 6 ( bus , t ) $Wcap ( b u s ) . . pwc ( bus , t ) =e=WD( t , ’w ’ ) Wcap ( b u s ) / Sbase
pw ( bus , t ) ;
c o n s t E S S ( bus , t ) $SOCMax ( b u s ) . . SOC( bus , t ) =e=SOC0 ( b u s ) $ ( o r d ( t ) = 1 ) +
SOC( bus , t 1)$ ( o r d ( t ) >1)+Pc ( bus , t ) e t a c Pd ( bus , t ) / e t a d ;
Model l o a d f l o w / all /;
Sbase ;
P i j . up ( bus , node , t ) $ ( ( c o n e x ( bus , node ) ) ) = b r a n c h ( bus , node , ’ L i m i t ’ ) /
Sbase ;
P i j . l o ( bus , node , t ) $ ( ( c o n e x ( bus , node ) ) )=b r a n c h ( bus , node , ’ L i m i t ’ ) /
Sbase ;
l s h . up ( bus , t ) =WD( t , ’ d ’ ) BusData ( bus , ’ pd ’ ) / S b a s e ; l s h . l o ( bus , t ) = 0 ;
Pwc . up ( bus , t ) =WD( t , ’w ’ ) Wcap ( b u s ) / S b a s e ; Pwc . l o ( bus , t ) = 0 ;
SOC . up ( bus , t ) =SOCmax ( b u s ) / s b a s e ; SOC . l o ( bus , t ) =0.2SOCmax ( b u s ) /
sbase ;
SOC . f x ( bus , ’ t 2 4 ’ ) =SOC0 ( b u s ) ;
Pc . up ( bus , t ) =0.2SOCmax ( b u s ) / s b a s e ; Pc . l o ( bus , t ) = 0 ;
Pd . up ( bus , t ) =0.2SOCmax ( b u s ) / s b a s e ; Pd . l o ( bus , t ) = 0 ;
200 Discharge (MW)

Charge (MW)
ESS at bus 19
150 SOC (MWh)
100
50
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
120
Discharge (MWh)
100
Charge (MW)
ESS at bus 21
80 SOC (MWh)
60
40
20
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 7.6 The hourly dispatch of ESS in DC-OPF problem
400
g 12
350
g 11
g 10
300
g9
250 g8
(MW)
g7
200
g,t
g6
P
g5
150
g4
g3
100
g2
50 g1
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 7.7 The hourly dispatch of thermal generating units in DC-OPF problem
The hourly dispatch of ESS in DC-OPF problem is shown in Fig. 7.6.

The hourly dispatch of thermal generating units in DC-OPF problem is depicted
in Fig. 7.7.
7.2.4 ESS Operation in AC-OPF
The multi-period ESS integrated AC-OPF is formulated as follows:

X X
i;t (7.4a)
g;t i;t
X X p
Pg;t C LSi;t C Pwi;t Pli;t Pci;t C Pdi;t D Pij;t W i;t (7.4b)
g2˝Gi j2˝ì
X X q
Qg;t Qli;t D Pij;t W i;t (7.4c)
g2˝Gi j2˝ì
Vi;t †ıi;t Vj;t †ıj;t bVi;t

Iij;t D C † ıi;t C (7.4d)
Zij †ij 2 2

Sij;t D .Vi;t †ıi;t /Iij;t (7.4e)
2
˚ Vi;t Vi;t Vj;t
Pij;t D real Sij;t D cos.ij / cos.ıi;t ıj;t C ij / (7.4f)
Zij Zij
2 2
˚ Vi;t Vi;t Vj;t bVi;t
Qij;t D Img Sij;t D sin.ij / sin.ıi;t ıj;t C ij / (7.4g)
Zij Zij 2
Sijmax Sij;t Sijmax (7.4h)
Pmin
g Pg;t Pmax
g (7.4i)
Qmin
g Qg;t Qmax
g (7.4j)
Pg;t Pg;t1 RUg (7.4k)
Pg;t1 Pg;t RDg (7.4l)
0 LSi;t pli;t (7.4m)
Pwc
i;t D wi;t wi Pwi;t (7.4n)
0 Pwi;t wt wi (7.4o)

SOCi;t D SOCi;t1 C Pci;t c Pdi;t =d t (7.4p)
Pci;min Pci;t Pci;max (7.4q)
Pdi;min Pdi;t Pdi;max (7.4r)
SOCi;min SOCi;t SOCi;max (7.4s)
The AC-OPF model is implemented on IEEE RTS 24-bus network with ESS
integration which is shown in Fig. 7.5. The wind capacities and ESS characteristics
are the same as Sect. 7.2.3. The minimum operating cost is $4:3018 105 obtained
by using the GCode 7.4.
GCode 7.4 The ESS-ACOPF GAMS code for IEEE Reliability test 24-bus network
S c a l a r s Sbase / 1 0 0 / ,VOLL / 1 0 0 0 0 / ,VOLW / 5 0 / ; a l i a s ( bus , node ) ;
s e t GB( bus , Gen ) c o n n e c t i v i t y i n d e x o f e a c h g e n e r a t i n g u n i t t o
each bus ;
r x b limit ;
T a b l e WD( t , )
P a r a m e t e r Wcap ( b u s ) Wind c a p a c i t i e s a t e a c h b u s
p a r a m e t e r SOCMax ( b u s )
/ 19 200
21 100 / ; S c a l a r e t a c / 0 . 9 5 / , e t a d / 0 . 9 / , VWC / 5 0 / ;
P a r a m e t e r SOC0 ( b u s ) ; SOC0 ( b u s ) =0.2SOCMax ( b u s ) / s b a s e ;
( bus , node , ’ x ’ ) , 2 ) + power ( b r a n c h ( bus , node , ’ r ’ ) , 2 ) ) ;
b r a n c h ( node , bus , ’ t h e t a ’ ) $ b r a n c h ( node , bus , ’ l i m i t ’ ) =
a r c t a n ( b r a n c h ( node , bus , ’ x ’ ) / b r a n c h ( node , bus , ’ r ’ ) ) ; P a r a m e t e r
c o n e x ( bus , node ) ;
l i m i t ’ ) ) =1;
V a r i a b l e s OF , Q i j ( bus , node , t ) , P i j ( bus , node , t ) , Pg ( Gen , t ) , Qg ( Gen , t ) ,
d e l t a ( bus , t )
,V( bus , t ) , l s h ( bus , t ) ,Pw ( bus , t ) , pwc ( bus , t ) ,SOC( bus , t ) , Pd ( bus , t ) , Pc
( bus , t ) ;
E q u a t i o n s const2 , const2B , const3 , const4 , const5 , const6 , const7 ,
const8 , constESS ;
c o n s t 1 ( bus , node , t ) $ c o n e x ( bus , node ) . . P i j ( bus , node , t ) =e=
c o n s t 2 ( bus , t ) . . l s h ( bus , t ) $BusData ( bus , ’ pd ’ ) +Pw ( bus , t ) $Wcap ( b u s ) +
sum ( Gen$GB ( bus , Gen ) ,
Pg ( Gen , t ) ) +(Pc ( bus , t ) +Pd ( bus , t ) )$SOCMAX( b u s )WD( t , ’ d ’ ) BusData
( bus , ’ pd ’ ) / S b a s e =e =+sum ( n o d e $ c o n e x ( node , b u s ) , P i j ( bus , node , t ) ) ;
c o n s t 2 B ( bus , t ) . . sum ( Gen$GB ( bus , Gen ) , Qg ( Gen , t ) )WD( t , ’ d ’ ) BusData
( bus , ’Qd ’ ) / S b a s e =e =+sum ( n o d e $ c o n e x ( node , b u s ) , Q i j ( bus , node , t ) ) ;
c o n s t 3 . . OF=g=sum ( ( bus , Gen , t ) $GB ( bus , Gen ) , Pg ( Gen , t ) GD( Gen , ’ b ’ )
S b a s e ) +sum ( ( bus , t ) ,
VOLL l s h ( bus , t ) S b a s e $ B u s D a t a ( bus , ’ pd ’ ) +VOLWPwc ( bus , t )
sbase$Wcap ( b u s ) ) ;
pw ( bus , t ) ;
c o n s t E S S ( bus , t ) $SOCMax ( b u s ) . . SOC( bus , t ) =e=SOC0 ( b u s ) $ ( o r d ( t ) = 1 )
+SOC( bus , t 1)$ ( o r d ( t ) >1)+Pc ( bus , t ) e t a c Pd (
bus , t ) / e t a d ;

V( bus , t ) V( bus , t ) c o s ( b r a n c h ( node , bus , ’ t h e t a ’ ) ) / b r a n c h ( node , bus , ’
z’)
V( bus , t ) V( node , t ) c o s ( d e l t a ( bus , t ) d e l t a ( node , t )
+ b r a n c h ( node , bus , ’ t h e t a ’ ) ) / b r a n c h ( node , bus , ’ z ’ ) ;
c o n s t 8 ( bus , node , t ) $ c o n e x ( bus , node ) . . Q i j ( bus , node , t ) =e=
V( bus , t ) V( bus , t ) s i n ( b r a n c h ( node , bus , ’ t h e t a ’ ) ) / b r a n c h ( node , bus , ’
z’)
V( bus , t ) V( node , t ) s i n ( d e l t a ( bus , t ) d e l t a ( node , t )
+ b r a n c h ( node , bus , ’ t h e t a ’ ) ) / b r a n c h ( node , bus , ’ z ’ )
b r a n c h ( node , bus , ’ b ’ ) V( bus , t ) V( bus , t ) / 2 ; Model l o a d f l o w /
all /;
Sbase ;
Qg . l o ( Gen , t ) =GD( Gen , ’ Qmin ’ ) / S b a s e ; Qg . up ( Gen , t ) =GD( Gen , ’Qmax ’ ) /
Sbase ;
V . up ( bus , t ) = 1 . 0 5 ; V . l o ( bus , t ) = 0 . 9 5 ; V . l ( bus , t ) = 1 ;
Sbase ;
Sbase ;
Q i j . up ( bus , node , t ) $ ( ( c o n e x ( bus , node ) ) ) = b r a n c h ( bus , node , ’ L i m i t ’ ) /
Sbase ;
Q i j . l o ( bus , node , t ) $ ( ( c o n e x ( bus , node ) ) )=b r a n c h ( bus , node , ’ L i m i t ’ ) /
Sbase ;
0;
SOC . up ( bus , t ) =SOCmax ( b u s ) / s b a s e ; SOC . l o ( bus , t ) =0SOCmax ( b u s ) /
sbase ;
SOC . f x ( bus , ’ t 2 4 ’ ) =SOC0 ( b u s ) ; Pc . up ( bus , t ) =0.2SOCmax ( b u s ) / s b a s e ;
Pc . l o ( bus , t ) = 0 ;
Pd . up ( bus , t ) =0.2SOCmax ( b u s ) / s b a s e ; Pd . l o ( bus , t ) = 0 ;
S o l v e l o a d f l o w m i n i m i z i n g OF u s i n g n l p ;
The hourly dispatch of ESS in AC-OPF problem is shown in Fig. 7.8 and the
hourly dispatch of thermal generating units in AC-OPF problem is depicted in
Fig. 7.9.
The hourly active power losses in AC-OPF problem are shown in Fig. 7.10. The
hourly min/average/max value of voltage magnitudes (pu) is depicted in Fig. 7.11.
300
Discharge (MW)
ESS at bus 19 Charge (MW)
200 SOC (MWh)
100
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
150
Discharge (MW)
Charge (MW)
ESS at bus 21
SOC (MWh)
100
50
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 7.8 The hourly dispatch of ESS in AC-OPF problem
400
350 g 12
g 11
300 g 10
g9
250 g8
P g,t (MW)
g7
200 g6
g5
150 g4
g3
100 g2
g1
50
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 7.9 The hourly dispatch of thermal generating units in AC-OPF problem
60
55
Active power losses (MWh)
50
45
40
35
30
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 7.10 The hourly active power losses in AC-OPF problem
1.06
V max
V avg
1.04 V min
1.02
Voltage (pu)
0.98
0.96
0.94
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 7.11 The hourly min/avegare/max values of voltage magnitudes (pu)

7.3 ESS Allocation 195
7.3 ESS Allocation
The multi-period ESS integrated DC-OPF is formulated as follows:

X X
i;t (7.5a)
g;t i;t
X X
Pg;t C LSi;t C Pwi;t Li;t Pci;t C Pdi;t D Pij;t W i;t (7.5b)
g2˝Gi j2˝ì
ıi;t ıj;t
Pij;t D (7.5c)
Xij
Pmax
ij Pij;t Pmax
ij (7.5d)
Pmin
g Pg;t Pmax
g (7.5e)
0 LSi;t Li;t (7.5h)
Pwc w w
i;t D wi;t i Pi;t (7.5i)

SOCi;t D SOCi;t1 C Pci;t c Pdi;t =d t (7.5k)
Pci;min Pci;t Pci;max (7.5l)
Pdi;min Pdi;t Pdi;max (7.5m)
SOCi;min NiESS SOCi;t SOCi;max NiESS (7.5n)
X
NiESS Nmax
ESS
(7.5o)
i
Equation (7.5n) enforces the state of charge exist only in those buses that NiESS
has nonzero value. The total number of ESS units is limited by (7.5o). The model
is implemented on IEEE RTS 24-bus network with the wind power generation as
shown in Fig. 7.5. The purpose is to find the optimal allocation ESS to minimize
the total operating costs. It should be noted that this formulation is only considering
24 h. For planning purposes, longer planning horizon should be taken into account,
and load growth and investment costs should also be considered. The ESS allocation
is solved using the GCode 7.5 and the total operating cost is obtained as $4:1753
105 . The optimal locations and capacities of ESS are shown in Fig. 7.12.
GCode 7.5 The ESS allocation using DC-OPF GAMS code for IEEE Reliability test 24-bus
network
S c a l a r s Nmax , S b a s e / 1 0 0 / ,VOLL / 1 0 0 0 0 / ,VWC / 5 0 / ; a l i a s ( bus ,
node ) ;
Removed f o r s a v i n g s p a c e , t h e same a s p r e v i o u s d a t a
s e t GB( bus , Gen ) c o n n e c t i v i t y i n d e x o f e a c h g e n e r a t i n g u n i t t o
each bus
/ Removed f o r s a v i n g s p a c e , t h e same a s p r e v i o u s d a t a / ;
Removed f o r s a v i n g s p a c e , t h e same a s p r e v i o u s d a t a
r x b limit
Removed f o r s a v i n g s p a c e , t h e same a s p r e v i o u s d a t a ;
T a b l e WD( t , )
Removed f o r s a v i n g s p a c e , t h e same a s p r e v i o u s d a t a ;
p a r a m e t e r Wcap ( b u s )
/8 200
19 150
21 1 0 0 / ;
p a r a m e t e r SOCMax ( b u s ) ; s c a l a r e t a c / 0 . 9 5 / , e t a d / 0 . 9 / , VWC
/50/;
p a r a m e t e r SOC0 ( b u s ) ;
);
, bus , ’ L i m i t ’ ) ;
node , ’ x ’ ) ;
( bus , node , ’ x ’ ) , 2 )
+ power ( b r a n c h ( bus , node , ’ r ’ ) , 2 ) ) ;
P a r a m e t e r c o n e x ( bus , node ) ;
l i m i t ’ ) ) =1;
V a r i a b l e s OF , P i j ( bus , node , t ) , Pg ( Gen , t ) , d e l t a ( bus , t ) , l s h ( bus , t ) ,Pw
( bus , t ) ,
pwc ( bus , t ) ,SOC( bus , t ) , Pd ( bus , t ) , Pc ( bus , t ) ;
I n t e g e r v a r i a b l e NESS ( b u s ) ;
E q u a t i o n s c o n s t 1 , c o n s t 2 , c o n s t 3 , c o n s t 4 , c o n s t 5 , c o n s t 6 , constESS ,
constESS2 , constESS3 , constESS4 , constESS5 , c o n s t E S S 6 ;
c o n s t 2 ( bus , t ) . . l s h ( bus , t ) $BusData ( bus , ’ pd ’ ) +Pw ( bus , t ) $Wcap ( b u s )
7.3 ESS Allocation 197
+sum ( Gen$GB ( bus , Gen ) , Pg ( Gen , t ) ) +(Pc ( bus , t ) +Pd ( bus , t ) )$SOCMAX( b u s
)
WD( t , ’ d ’ ) BusData ( bus , ’ pd ’ ) / S b a s e =e =+sum ( n o d e $ c o n e x ( node , b u s ) ,
P i j ( bus , node , t ) ) ;
c o n s t 3 . . OF=g=sum ( ( bus , Gen , t ) $GB ( bus , Gen ) , Pg ( Gen , t ) GD( Gen , ’ b ’ )
S b a s e ) +sum ( ( bus , t ) ,
VOLL l s h ( bus , t ) S b a s e $ B u s D a t a ( bus , ’ pd ’ ) +VWCPwc ( bus , t )
sbase$Wcap ( b u s ) ) ;
pw ( bus , t ) ;
c o n s t E S S ( bus , t ) . . SOC( bus , t ) =e = ( 0 . 2 NESS ( b u s ) SOCMax ( b u s ) / s b a s e
) $ ( ord ( t ) =1)+
SOC( bus , t 1)$ ( o r d ( t ) >1)+Pc ( bus , t ) e t a c Pd ( bus , t ) / e t a d ;
c o n s t E S S 2 ( bus , t ) . . SOC( bus , t ) = l =NESS ( b u s ) SOCmax ( b u s ) / s b a s e ;
c o n s t E S S 3 ( bus , t ) . . Pc ( bus , t ) = l =0.2NESS ( b u s ) SOCmax ( b u s ) / s b a s e ;
c o n s t E S S 4 ( bus , t ) . . Pd ( bus , t ) = l =0.2NESS ( b u s ) SOCmax ( b u s ) / s b a s e ;
constESS5 . . sum ( bus , NESS ( b u s ) ) = l =Nmax ;
c o n s t E S S 6 ( b u s ) . . SOC( bus , ’ t 2 4 ’ ) =e =0.2NESS ( b u s ) SOCmax ( b u s ) /
sbase ;
Model l o a d f l o w / all /;
Sbase ;
Sbase ;
Sbase ;
0;
SOC . l o ( bus , t ) = 0 ; Pc . l o ( bus , t ) = 0 ; Pd . l o ( bus , t ) = 0 ;
NESS . up ( b u s ) = 5 ; SOCMax ( b u s ) = 2 0 ; SOC0 ( b u s ) =0.2SOCMax ( b u s ) / s b a s e ;
Nmax = 1 5 ;
S o l v e l o a d f l o w m i n i m i z i n g OF u s i n g mip ;
The hourly dispatch of storage unit in optimal ESS allocation problem is shown in
Fig. 7.13. The hourly dispatch of thermal generating units in ESS allocation problem
is depicted in Fig. 7.14.
Fig. 7.12 The optimal allocation of ESS units in IEEE RTS network
Discharge (MW) Charge (MW) SOC (MWh)

ESS at bus 1
100
50
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
ESS at bus 5
100
50
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
ESS at bus 13
100
50
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 7.13 The hourly dispatch of storage unit in optimal ESS allocation problem
Nomenclature
Indices
k Blocks considered for piecewise linear fuel cost function.

i Thermal generating units.
t Time intervals.
7.4 Applications 199
400
350 g 12
g 11
300 g 10
g9
250 g8
P g,t (MW)
g7
g6
200
g5
g4
150
g3
g2
100
g1
50
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 7.14 The hourly dispatch of thermal generating units in ESS allocation problem
Parameters
et Electric energy price at time t ($/MW h).
Lt Electric demand at time t.
ag ; bg ; cg Fuel cost coefficients of unit i.
max=min
Pg Maximum/minimum limits of power generation of thermal unit i.
RUg =RDg Ramp-up/down limit of generation unit i (MW/h).
VWC Value of wind curtailment
wt Wind availability at time t
Variables
ESS
Ui;t Binary variable indicating the charge/discharge mode of ESS connected to
bus i at time t
Pwt Wind generation at time t
Pwc
t Wind curtailment at time t
Cg;t Fuel cost of thermal unit i at time t ($).
Pg;t Power generated by thermal unit i at time t (MW).
TC Total operating costs ($).
7.4 Applications
Various applications of ESS in power system operation and planning studies have
been investigated in the literature. Some of these aspects are listed here as follows:
• Electrical Energy Storage and Real-Time Thermal Ratings to Defer Network
Reinforcement [17]
• Energy storage and its use with intermittent renewable energy [18]
• Sizing of Energy Storage for microgrids [19]
• Primary Frequency Control using ESS [20]
• OPF in microgrids with ESS [21]
• ESS sizing for grid-connected photovoltaic systems [22]
• Coupling pumped hydro energy storage with UC [23]
• Transmission congestion relief privately owned large-scale ESS [24]
• Phase balancing using ESS in power grids under uncertainty [11]
• Price maker ESS in nodal energy markets [25]
• Value of ESS in dynamic economic dispatch game [26]
• Impact of strategic behavior and ownership of ESS on provision of flexibility
[27]
• Reliability improvement using ESS [28]
References
1. A.M. Gee, F. Robinson, W. Yuan, A superconducting magnetic energy storage-emulator/battery

supported dynamic voltage restorer. IEEE Trans. Energy Convers. 32(1), 55–64 (2017)
2. C. Krupke, J. Wang, J. Clarke, X. Luo, Modeling and experimental study of a wind turbine
system in hybrid connection with compressed air energy storage. IEEE Trans. Energy Convers.
32(1), 137–145 (2017)
3. X. Chang, Y. Li, X. Li, X. Chen, An active damping method based on a supercapacitor energy
storage system to overcome the destabilizing effect of instantaneous constant power loads in
dc microgrids. IEEE Trans. Energy Convers. 32(1), 36–47 (2017)
4. B. Steffen, Prospects for pumped-hydro storage in germany. Energy Policy 45, 420–429 (2012)
5. J.M. Lujano-Rojas, R. Dufo-Lpez, J.L. Bernal-Agustn, J.P.S. Catalo, Optimizing daily oper-
ation of battery energy storage systems under real-time pricing schemes. IEEE Trans. Smart
Grid 8(1), 316–330 (2017)
6. M.I. Daoud, A.M. Massoud, A.S. Abdel-Khalik, A. Elserougi, S. Ahmed, A flywheel energy
storage system for fault ride through support of grid-connected VSC HVDC-based offshore
wind farms. IEEE Trans. Power Syst. 31(3), 1671–1680 (2016)
7. C. Park, R. Sedundo, V. Knazkins, P. Korba, Feasibility analysis of the power-to-gas concept
in the future swiss power system, in: CIRED Workshop 2016 (2016), pp. 1–5
8. S. Pulendran, J.E. Tate, Energy storage system control for prevention of transient under-
frequency load shedding. IEEE Trans. Smart Grid 8(2), 927–936 (2017)
9. C.L.T. Borges, D.M. Falcao, Optimal distributed generation allocation for reliability, losses
and voltage improvement. Int. J. Electr. Power Energy Syst. 28(6), 413–420 (2006)
11. S. Sun, B. Liang, M. Dong, J.A. Taylor, Phase balancing using energy storage in power grids
under uncertainty. IEEE Trans. Power Syst. 31(5), 3891–3903 (2016)
12. L.S. Vargas, G. Bustos-Turu, F. Larran, Wind power curtailment and energy storage in
transmission congestion management considering power plants ramp rates. IEEE Trans. Power
Syst. 30(5), 2498–2506 (2015)
payments minimization under electricity price uncertainty. IEEE Trans. Smart Grid 7(1),
261–272 (2016)
References 201
Syst. 26(1), 12–19 (2011)
17. D.M. Greenwood, N.S. Wade, P.C. Taylor, P. Papadopoulos, N. Heyward, A probabilistic
method combining electrical energy storage and real-time thermal ratings to defer network
reinforcement. IEEE Trans. Sustain. Energy 8(1), 374–384 (2017)
18. J.P. Barton, D.G. Infield, Energy storage and its use with intermittent renewable energy. IEEE
Trans. Energy Convers. 19(2), 441–448 (2004)
19. S.X. Chen, H.B. Gooi, M.Q. Wang, Sizing of energy storage for microgrids. IEEE Trans. Smart
Grid 3(1), 142–151 (2012)
20. A. Oudalov, D. Chartouni, C. Ohler, Optimizing a battery energy storage system for primary
frequency control. IEEE Trans. Power Syst. 22(3), 1259–1266 (2007)
21. Y. Levron, J.M. Guerrero, Y. Beck, Optimal power flow in microgrids with energy storage.
22. Y. Ru, J. Kleissl, S. Martinez, Storage size determination for grid-connected photovoltaic
systems. IEEE Trans. Sustain. Energy 4(1), 68–81 (2013)
23. K. Bruninx, Y. Dvorkin, E. Delarue, H. Pandi, W. Dhaeseleer, D.S. Kirschen, Coupling pumped
hydro energy storage with unit commitment. IEEE Trans. Sustain. Energy 7(2), 786–796
(2016)
24. H. Khani, M.R. Dadash Zadeh, A.H. Hajimiragha, Transmission congestion relief using
privately owned large-scale energy storage systems in a competitive electricity market. IEEE
Trans. Power Syst. 31(2), 1449–1458 (2016)
25. H. Mohsenian-Rad, Coordinated price-maker operation of large energy storage units in nodal
energy markets. IEEE Trans. Power Syst. 31(1), 786–797 (2016)
26. W. Tang, R. Jain, Dynamic economic dispatch game: the value of storage. IEEE Trans. Smart
Grid 7(5), 2350–2358 (2016)
27. K. Hartwig, I. Kockar, Impact of strategic behavior and ownership of energy storage on
provision of flexibility. IEEE Trans. Sustain. Energy 7(2), 744–754 (2016)
28. A. Vieira Pombo, J. Murta-Pina, V. Ferno Pires, Multiobjective formulation of the integration
of storage systems within distribution networks for improving reliability. Electr. Power Syst.
Res. 148, 87–96 (2017)
Chapter 8
Power System Observability
This chapter provides a solution for increasing the power system observability by
allocation of Phasor Measurement Units (PMU) problem in GAMS. The PMU is
able to measure the voltage phasor at the connection bus and also it measures the
current phasor of any branch connected to the bus hosting the PMU. Different cases
are analyzed and the problem is tested on some standard IEEE cases.
8.1 Min No. PMU Placement
The PMU devices are able to provide the system operator a set of synchronized
phasor measurements in the system in order to make it observable. In this case,
it is needed to make every existing bus observable no matter if it has any
generation/demand or not. Additionally, the contingencies are not considered. In
order to explain the observability concept, Fig. 8.1 provides an example. If a PMU
is installed on bus i then bus i and all adjacent buses are observable. It means that the
voltage phasors of these buses are known. The system is called observable if every
bus of the network is observable.
The PMU placement concept is simply explained as follows: If a bus or at least
one of its adjacent buses is equipped with PMU then this bus is observable. One
trivial solution for PMU placement problem is installing PMU at all buses. This
solution makes all buses observable but obviously the costs would be very high and
sometimes practical aspects do not allow this to happen. In other words, it is tried
to find the minimum number of buses to have PMU devices in order to make the
whole system observable. This PMU placement problem is formulated as follows:
X
min OF D i (8.1a)
i
i2˝B

DOI 10.1007/978-3-319-62350-4_8
204 8 Power System Observability
Fig. 8.1 PMU functionality

and the observability concept
X
i C j ˛i 8i 2 ˝B (8.1b)
j2˝ij`
˛i D 1 (8.1c)
where i; j are bus index, i is a binary variable indicating if it has PMU (1) or not (0),
˝B is the set of all network buses, ˛i is a binary parameter (observability) which is
set to 1 to make the whole network observable, and ˝i is the set of all buses adjacent
(connected) to bus i.
The GAMS code for solving the optimal PMU allocation in IEEE 14-bus network
without considering the zero bus injection is provided in GCode 8.1.
GCode 8.1 PMU allocation for IEEE 14 network without considering zero injection nodes
S e t s bus /114/;
A l i a s ( bus , node ) ;
T a b l e BusData ( bus , ) b u s s c h a r a c t e r i s t i c s
Pd gen
1 0 1
2 21.7 1
3 94.2 1
4 47.8
5 7.6
6 11.2 1
8 0 1
9 29.5
10 9
11 3.5
12 6.1
13 13.5
14 14.9 ;
8.1 Min No. PMU Placement 205
/1 . 2
1 . 5
2 . 3
2 . 4
2 . 5
3 . 4
4 . 5
4 . 7
4 . 9
5 . 6
6 . 11
6 . 12
6 . 13
7 . 8
7 . 9
9 . 10
9 . 14
10 . 11
12 . 13
13 . 14/;
V a r i a b l e s OF ;
B i n a r y v a r i a b l e PMU( b u s ) ;
Equations
const1 ,
const2 ;
const1 . . OF=g=sum ( bus ,PMU( b u s ) ) ;

c o n s t 2 ( b u s ) . . PMU( b u s ) +sum ( n o d e $ c o n e x ( bus , node ) ,PMU(
node ) ) =g = 1 ;
Model p l a c e m e n t / const1 , const2 / ;
S o l v e p l a c e m e n t m i n i m i z i n g OF u s i n g mip ;
The optimal PMU allocation for IEEE 14-bus without considering the zero
injection buses is obtained using GCode 8.1 and is shown in Fig. 8.2. The simulation
results show that four PMU devices are required to make the IEEE 14-bus network
fully observable. The optimal PMU allocation for IEEE 57-bus without considering
the zero injection buses is shown in Fig. 8.3. The simulation results show that 14
PMU devices are needed to make the IEEE 57-bus network fully observable.
The optimal PMU allocation for IEEE 118-bus without considering the zero
injection buses is shown in Fig. 8.4. The simulation results show that 32 PMU
devices are required to make the IEEE 118-bus network fully observable.
Fig. 8.2 Optimal PMU allocation for IEEE 14-bus without considering the zero injection buses
8.2 Min Cost PMU placement 207
8.2 Min Cost PMU placement
The cost of PMU device is basically dependent on the number of channels used for
measurement. In [1], it is stated that if the cost of each PMU is C and the number of
used channels is n at bus i then the total PMU cost would be wi D .1 C 0:1 n/C
in the mentioned bus. n is the number of branches connected to bus i.
X
min OF D wi i (8.2a)
i
i2˝B
X
i C j ˛i 8i 2 ˝B (8.2b)
j2˝ij`
˛i D 1 (8.2c)
where i; j are bus index, i is a binary variable indicating if it has PMU (1) or not
(0), ˝B is the set of all network buses, and ˛i is a binary parameter (observability)
which is set to 1 to make the whole network observable.
The GAMS code for solving the optimal allocation of PMU devices with
minimum cost without considering the ZIB is described in GCode 8.2.
GCode 8.2 Min cost PMU allocation for IEEE 14 network without considering zero injection
nodes
S e t s bus / 1 1 4 / ;
Set conex Bus c o n n e c t i v i t y m a t r i x
/1 . 2
1 . 5
2 . 3
2 . 4
2 . 5
3 . 4
4 . 5
4 . 7
4 . 9
5 . 6
6 . 11
6 . 12
6 . 13
7 . 8
7 . 9
9 . 10
9 . 14
10 . 11
12 . 13
13 . 14/;
p a r a m e t e r s c o s t ( bus ) ;
c o s t ( b u s ) =1+0.1sum ( n o d e $ c o n e x ( bus , node ) , 1 ) ;
Variables
OF_c ;
binary variable
PMU( b u s ) ;
Equations
const0 ,
const2 ;
const0 . . OF_c=e=sum ( bus , c o s t ( b u s ) PMU( b u s ) ) ;
c o n s t 2 ( b u s ) . . PMU( b u s ) +sum ( n o d e $ c o n e x ( bus , node ) ,
PMU( node ) ) =g = 1 ;
Option o p t c r =0;
Model p l a c e m e n t 0 / const0 , const2 / ;

S o l v e p l a c e m e n t 0 m i n i m i z i n g OF_c u s i n g mip ;
D i s p l a y pmu . l ;
8.3 Min No. PMU Placement Considering ZIB 209
Table 8.1 Optimal Network PMU locations

allocation of PMU devices
with minimum cost without IEEE 14 2 , 8 , 10 , 13
considering the ZIB IEEE 57 2 , 6 , 12 , 19 , 22 , 25
27 , 32 , 36 , 39 , 41 , 45
46 , 47 , 50 , 52 , 55
IEEE 118 1 , 5 , 10 , 12 , 15 , 17
21 , 23 , 28 , 30 , 36 , 40
44 , 46 , 50 , 52 , 56 , 62
64 , 71 , 75 , 77 , 80 , 85
87 , 91 , 94 , 101 , 105 , 110
114 , 116
Fig. 8.5 PMU functionality

and observability concept in
zero injection buses
The minimum cost PMU allocation problem is solved and the solutions are given
in Table 8.1.
8.3 Min No. PMU Placement Considering ZIB
Some buses in the network do not contain demand or generation which are called
Zero Injection Buses (ZIB). These buses can be specially treated in terms of
observability as explained in Fig. 8.5. This figure shows zero injection bus j and its
adjacent buses. The observability of each bus is guaranteed with the observability
of n 1 of them [2]. In this way, the network buses ˝B will be divided into two
categories as follows:
• Zero injection buses ˝z or adjacent to zero injection buses ˝za
• The rest of the buses (normal buses) ˝n
X
min OF D i (8.3a)
i
i2˝B
Table 8.2 ZIB buses in Network Zero injection buses

different networks under
study IEEE 14 7
IEEE 57 4,7,11,21,22,24,26,34
36,37,39,40,45,46,48
IEEE 118 5,9,30,37,38,63,64,68,71,81
Fig. 8.6 Optimal PMU allocation for IEEE 14-bus with considering the zero injection buses
X
i C j ˛i 8i 2 ˝n (8.3b)
j2˝ij`
* +
X X
a C j j˝za j 1 8i 2 ˝za (8.3c)
` `
a2˝ia j2˝aj
˛i D 1 (8.3d)
(0), ˝B is the set of all network buses, and ˛i is a binary parameter (observability)
which is set to 1 to make the whole network observable.
The zero injection buses for the networks under study are specified in Table 8.2.
without considering the zero bus injection is provided in GCode 8.1 The optimal
PMU allocation for IEEE 14 and 57 bus systems with considering the zero injection
buses are given in Figs. 8.6 and 8.7, respectively. The optimal PMU allocation for
IEEE 118-bus with considering the zero injection buses is shown in Fig. 8.8. The
8.3 Min No. PMU Placement Considering ZIB 211
simulation results show that 29 PMU devices are required to make the IEEE 118-
bus network fully observable.
GCode 8.3 PMU allocation for IEEE 14 network considering zero injection nodes
S e t b u s / 1 1 4 / ; a l i a s ( bus , node ) ; a l i a s ( bus , s h i n ) ;

T a b l e BusData ( bus , ) b u s c h a r a c t e r i s t i c s
Pd gen
1 0 1
2 21.7 1
3 94.2 1
4 47.8
5 7.6
6 11.2 1
8 0 1
9 29.5
10 9
11 3.5
12 6.1
13 13.5
14 14.9 ;
/1 . 2
1 . 5
2 . 3
2 . 4
2 . 5
3 . 4
4 . 5
4 . 7
4 . 9
5 . 6
6 . 11
6 . 12
6 . 13
7 . 8
7 . 9
9 . 10
9 . 14
10 . 11
12 . 13
13 . 14/;
p a r a m e t e r s A( bus , node ) ,B( bus , node ) , Z ( b u s ) , n o r m a l ( b u s ) ,
abnormal ( bus ) ;
Z ( b u s ) $ ( BusData ( bus , ’ Pd ’ ) =0 and BusData ( bus , ’ gen ’ ) = 0 ) =
yes ;
A( bus , node ) $ ( Z ( b u s ) and c o n e x ( bus , node ) ) = y e s ;
n o r m a l ( b u s ) $ ( Z ( b u s ) =0 and sum ( n o d e $ c o n e x ( bus , node ) ,
Z ( node ) ) = 0 ) = y e s ;
B( bus , node ) $ (A( bus , node ) ) = y e s ;
B ( bus , node ) $ ( Z ( node ) and o r d ( b u s ) = o r d ( node ) ) = y e s ;
a b n o r m a l ( b u s ) $ ( n o r m a l ( b u s ) = 0 ) = y e s ; V a r i a b l e s OF ; b i n a r y
v a r i a b l e PMU( b u s ) ;
E q u a t i o n s c o n s t 1 , eq1 , eq2 ;
const1 . . OF=g=sum ( bus ,PMU( b u s ) ) ;
eq1 ( b u s ) $ n o r m a l ( b u s ) . . PMU( b u s ) +sum ( n o d e $ c o n e x ( bus ,
node ) ,PMU( node ) ) =g = 1 ;
eq2 ( b u s ) $ a b n o r m a l ( b u s ) . . sum ( s h i n $ c o n e x ( bus , s h i n ) ,
PMU( s h i n ) +sum ( n o d e $ c o n e x ( s h i n , node ) ,PMU( node ) ) ) =g=sum
( node , B ( bus , node ) ) 1;
Model p l a c e m e n t W i t h z e r o i n j e c t i o n / c o n s t 1 , eq1 , eq2 / ;
S o l v e p l a c e m e n t W i t h z e r o i n j e c t i o n m i n i m i z i n g OF u s i n g
mip ;
8.4 Min No. PMU Placement for Maximizing the Observability 213
8.4 Min No. PMU Placement for Maximizing

the Observability
Another question that can be asked is how to maximize the system observability
using a limited number of PMU devices. How should these units be allocated in a
given system? In order to answer this question, a binary variable ˛i is defined which
states bus i is observable (1) or not (0). The optimization problem to be solved is as
follows:
X
max OF D ˛i (8.4a)
i ;˛i
i2˝B
X
i NPMU (8.4b)
i2˝B
X
i C j ˛i 8i 2 ˝B (8.4c)
j2˝ij`
1 ˛i (8.4d)
(0), ˝B is the set of all network buses, ˛i is an integer variable (observability) which
its minimum value is set to 1 to make the whole network observable, and NPMU is
the number of available PMU devices.
Table 8.3 Optimal NPMU

allocation of PMU devices for
Bus 1 2 3 4
maximizing the system
observability of IEEE 14-bus 2 1 1
network 4 1
6 1 1
7 1
9 1 1
10 1
13 1
Total observable buses 6 10 13 14
for maximizing the system observability is provided in GCode 8.4. The optimal
solution is described in Table 8.3.
GCode 8.4 PMU allocation for IEEE 14 network for maximizing the system observability
S e t s b u s / 1 1 4 / ; a l i a s ( bus , node ) ; a l i a s ( bus , s h i n ) ;

Pd gen
1 0 1
2 21.7 1
3 94.2 1
4 47.8
5 7.6
6 11.2 1
8 0 1
9 29.5
10 9
11 3.5
12 6.1
13 13.5
14 14.9 ;
/1 . 2
1 . 5
2 . 3
2 . 4
2 . 5
3 . 4
4 . 5
4 . 7
4 . 9
5 . 6
8.4 Min No. PMU Placement for Maximizing the Observability 215
6 . 11
6 . 12
6 . 13
7 . 8
7 . 9
9 . 10
9 . 14
10 . 11
12 . 13
13 . 14/;
p a r a m e t e r s A( bus , node ) ,B( bus , node ) , Z ( b u s ) , n o r m a l ( b u s ) ,
abnormal ( bus ) ;
Z ( b u s ) $ ( BusData ( bus , ’ Pd ’ ) =0 and BusData ( bus , ’ gen ’ ) = 0 ) =
yes ;
A( bus , node ) $ ( Z ( b u s ) and c o n e x ( bus , node ) ) = y e s ;
n o r m a l ( b u s ) $ ( Z ( b u s ) =0 and sum ( n o d e $ c o n e x ( bus , node ) ,
Z ( node ) ) = 0 ) = y e s ;
a b n o r m a l ( b u s ) $ ( n o r m a l ( b u s ) = 0 ) = y e s ; B( bus , node ) $ (A( bus ,
node ) ) = y e s ;
B ( bus , node ) $ ( Z ( node ) and o r d ( b u s ) = o r d ( node ) ) = y e s ;
V a r i a b l e s OF , OBI ; B i n a r y v a r i a b l e s PMU( b u s ) , a l p h a ( b u s )
; S c a l a r NPMU / 1 0 / ;
eq3 . . sum ( bus , pmu ( b u s ) ) = l =NPMU; eq4 . . OBI=e=sum
( node , a l p h a ( node ) ) ;
eq5 ( b u s ) . . PMU( b u s ) +sum ( n o d e $ c o n e x ( bus , node ) ,PMU( node )
) =g= a l p h a ( b u s ) ;
O p t i o n o p t c r = 0 ; Model p l a c e m e n t 3 / eq3 , eq4 , eq5 / ;
S o l v e p l a c e m e n t 3 m a x i m i z i n g OBI u s i n g mip ;
s e t c o u n t e r / c1c4 / ; p a r a m e t e r r e p o r t ( bus , c o u n t e r ) ,
OBIrep ( c o u n t e r ) ;
l o o p ( c o u n t e r , NPMU= o r d ( c o u n t e r ) ; S o l v e p l a c e m e n t 3
m a x i m i z i n g OBI u s i n g mip ;
r e p o r t ( bus , c o u n t e r ) =pmu . l ( b u s ) ; OBIrep ( c o u n t e r ) =OBI . l ;
);
The total observable buses vs NPMU in IEEE 57-bus network is shown in Fig. 8.9.
The total observable buses vs NPMU in IEEE 118-bus network is shown in Fig. 8.10.
60
Total number of observable buses
50
40
30
20
10
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
NPMU
Fig. 8.9 Total observable buses vs NPMU in IEEE 57-bus network
120
Total number of observable buses
100
80
60
40
20
0
0 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 31 32 33
N
PMU
Fig. 8.10 Total observable buses vs NPMU in IEEE 118-bus network
8.5 Min No. PMU Placement Considering Contingencies
The system observability should be maintained in normal operating condition as

well as in the case of contingencies. The analyzed contingencies include:
• Loss of single PMU
• Loss of single line
8.5 Min No. PMU Placement Considering Contingencies 217
8.5.1 Loss of PMU
In order to make the system observability resilient against the loss of single PMU
device, each bus should be observed at least by two PMU devices. The PMU
allocation resilient against single PMU outage is formulated as follows:
X
min OF D i (8.5a)
i
i2˝B
X
i C j 2˛i 8i 2 ˝B (8.5b)
j2˝ij`
˛i D 1 (8.5c)
˛i is a binary parameter which shows the observability of each bus and is set to 1 to
make the whole network observable. The GAMS code for PMU allocation of IEEE
14 network resilient against single PMU outage is provided in GCode 8.5.
GCode 8.5 PMU allocation for IEEE 14 network resilient against single PMU outage
Sets
bus / 1 1 4 / ;
a l i a s ( bus , s h i n ) ;
/1 . 2
1 . 5
2 . 3
2 . 4
2 . 5
3 . 4
4 . 5
4 . 7
4 . 9
5 . 6
6 . 11
6 . 12
6 . 13
7 . 8
7 . 9
9 . 10
9 . 14
10 . 11
12 . 13
13 . 14/;

Variables
OF , OBI ;
binary variable
PMU( b u s ) ;
Equations
const1 ,
const2A ;
const1 . . OF=e=sum ( bus ,PMU( b u s ) ) ;
c o n s t 2 A ( b u s ) . . PMU( b u s ) +sum ( n o d e $ c o n e x ( bus , node ) ,PMU
( node ) ) =g = 2 ;
Model p l a c e m e n t 4 / const1 , const2A / ;
S o l v e p l a c e m e n t 4 m i n i m i z i n g OF u s i n g mip ;
The optimal PMU allocation for IEEE 14-bus resilient against single PMU outage
is shown in Fig. 8.11. As it can be seen in this figure, 9 PMU devices are needed
to make the system fully observable even in the case of single PMU outage. The
optimal PMU allocation for IEEE 57-bus resilient against single PMU outage is
shown in Fig. 8.12. As it can be seen in this figure, 33 PMU devices are needed to
make the system fully observable even in the case of single PMU outage.
The optimal PMU allocation for IEEE 118-bus resilient against single PMU
outage is shown in Fig. 8.13. As it can be seen in this figure, 68 PMU devices are
needed to make the system fully observable even in the case of single PMU outage.
8.5.2 Single Line Outage
In this section, PMU units are allocated in the network in order to keep the system
observable even in case of single line outage. For this purpose, each bus should
have a PMU or should be monitored with at least two other PMU devices. This is
formulated as follows:
X
min OF D i (8.6a)
i ;˛i
i2˝B
i C ˛i 1 8i 2 ˝B (8.6b)
Fig. 8.11 Optimal PMU allocation for IEEE 14-bus resilient against single PMU outage
X
j 2 ˛i 8i 2 ˝B (8.6c)
j2˝ij`
˛i ; i 2 f0; 1g (8.6d)
The GAMS code for solving the PMU allocation considering single line outage is
developed in GCode 8.6. ˛i is a binary variable which indicates if bus i is monitored
using the PMU installed in the adjacent buses. The optimal solution is given in
Table 8.4.
GCode 8.6 PMU allocation for IEEE 14 network resilient against single line outage
S e t s bus / 1 1 4 / ; A l i a s ( bus , node ) ;

/1 . 2
1 . 5
2 . 3
2 . 4
2 . 5
3 . 4
4 . 5
4 . 7
4 . 9
5 . 6
8.6 Multistage PMU Placement 221
6 . 11
6 . 12
6 . 13
7 . 8
7 . 9
9 . 10
9 . 14
10 . 11
12 . 13
13 . 14/;
V a r i a b l e s OF ; B i n a r y v a r i a b l e PMU( b u s ) , a l p h a ( b u s ) ;
s c a l a r NPMU / 1 0 / ; E q u a t i o n s c o n s t 1 , c o n s t 2 C , c o n s t 2 D ;
c o n s t 2 C ( b u s ) . . PMU( b u s ) + a l p h a ( b u s ) =g = 1 ;
c o n s t 2 D ( b u s ) . . sum ( n o d e $ c o n e x ( bus , node ) ,PMU( node ) ) =g
=2 a l p h a ( b u s ) ;
Model p l a c e m e n t 6 / const1 , const2C , const2D / ;
S o l v e p l a c e m e n t 6 m i n i m i z i n g OF u s i n g mip ;
Table 8.4 Optimal Network Optimal locations

allocation of PMU devices
considering single line outage IEEE 14 1,3,6,8,9,10,13
IEEE 57 1,3,5,8,11,12,15,18,20,22,24,27
29,30,32,33,35,36,39,41,44,46,48
49,51,53,55,56
IEEE 118 12,15,17,19,21,23,26,27,29,32
34,36,37,38,40,42,44,46,49,50
51,53,56,59,61,64,67,68,72,73
74,75,76,78,80,83,85,87,89,91
92,94,96,100,102,105,106,109
111,112,114,116,117
8.6 Multistage PMU Placement
The optimal number of PMU units for making the system fully observable is
obtained so far in previous sections. The next step is the implementation of these
plans by the system operator. The ideal situation is installing all PMU units in
the obtained network buses. However, simultaneous installation of theses devices
is not possible due to some practical difficulties such as unavailability of crews and
difficulties in simultaneous maintenance outage of lines for PMU installations. In
this section, it is assumed that the optimal number and locations of PMU units
are already obtained. The system operator can only install a limited number of
PMU devices at a given period of time. The question is that which buses should
be equipped with PMU at each phase. This problem is called Multistage PMU
placement. The decision maker should maximize the number of observable buses
of the systems at each phase. The number of observable buses will increase in each
phase and after the final phase, the whole system will be observable. In order to
solve the problem, the following steps should be taken:
1. The optimal number of PMU units are found using the techniques discussed so
far. It might be considering ZIP or contingencies or costs. Save this number Npmu
and the locations of PMU devices.
2. How many installation stages and how many PMU units ( p ) are going to be
installed in each stage?
3. The following optimization problem is solved for stage p.
X
max OBI D ˛i (8.7a)
i ;˛i
i2˝B
X
TPMU D i p (8.7b)
i2˝B
X
i C j ˛i 8i 2 ˝B (8.7c)
j2˝ij`
i ; ˛i 2 f0; 1g (8.7d)
It is trying to maximize the number of observable buses in the network subject to

number of available installable PMU devices.
The GAMS code for solving the multistage optimal PMU allocation for IEEE
14-bus network is provided in GCode 8.7.
GCode 8.7 Multi-stage PMU allocation for IEEE 14 network
Sets
bus / 1 1 4 / ;
/1 . 2
1 . 5
2 . 3
2 . 4
2 . 5
3 . 4
4 . 5
4 . 7
8.6 Multistage PMU Placement 223
4 . 9
5 . 6
6 . 11
6 . 12
6 . 13
7 . 8
7 . 9
9 . 10
9 . 14
10 . 11
12 . 13
13 . 14/;
Variables
OF , OBI ;
B i n a r y v a r i a b l e PMU( b u s ) , a l p h a ( b u s ) , b e t a ( b u s ) ;
Equations const1 , const2 , const3 , const2E , const3A ;
c o n s t 2 ( b u s ) . . PMU( b u s ) +sum ( n o d e $ c o n e x ( bus , node ) ,PMU
( node ) ) =g = 1 ;
c o n s t 2 E ( b u s ) . . PMU( b u s ) +sum ( n o d e $ c o n e x ( bus , node ) ,PMU
( node ) ) =g= b e t a ( b u s ) ;
const3 . . OBI=e=sum ( bus , b e t a ( b u s ) ) ;
const3A . . OF=e=sum ( bus , PMU( b u s ) ) ;
Model p l a c e m e n t / const1 , const2 / ;
Model p l a c e m e n t 8 / const2E , const3 , const3A / ;
S e t cp / cp1 cp3 / ; A l i a s ( cp , cpp ) ;
P a r a m e t e r p h a s e _ r e p ( cp , ) , phase_pmu ( cp , b u s ) ; P a r a m e t e r
p h a s e ( cp )
/ cp1 2
cp2 1
cp3 1 / ;
S o l v e p l a c e m e n t m i n i m i z i n g OF u s i n g mip ;
d i s p l a y pmu . l ; Pmu . f x ( b u s ) $ ( Pmu . l ( b u s ) = 0 ) = 0 ;
b e t a . up ( b u s ) = 1 ; b e t a . l o ( b u s ) = 0 ;
l o o p ( cp ,
OF . up=sum ( cpp$ ( o r d ( cpp ) <= o r d ( cp ) ) , p h a s e ( cpp ) ) ;
s o l v e p l a c e m e n t 8 m a x i m i z i n g OBI u s i n g mip ;
p h a s e _ r e p ( cp , ’OBS ’ ) =OBI . l ;
p h a s e _ r e p ( cp , ’Tpmu ’ ) =sum ( bus , Pmu . l ( b u s ) ) ;
phase_pmu ( cp , b u s ) =pmu . l ( b u s ) ;
Pmu . f x ( b u s ) $ (PMU. l ( b u s ) ) = 1 ; ) ;
Fig. 8.14 Multistage optimal PMU allocation for different case studies
Table 8.5 Optimal allocation of PMU devices in multistage PMU allocation

IEEE 14 IEEE 57 IEEE 118
Phase OBI TPMU p OBI TPMU p OBI TPMU p
I 10 2 2 28 6 6 70 11 11
II 13 3 1 47 12 6 103 22 11
III 14 4 1 57 17 5 118 32 10
The solution for multistage optimal PMU allocation for different case studies is
shown in Fig. 8.14.
The optimal allocation of PMU devices in multistage PMU allocation is shown in
Table 8.5. It is assumed that the total number of PMU (TPMU) for full observability
of the system is known in advance. Three installation phases are considered and the
possible number of installable PMU devices ( p ) is also known.
8.7 Applications
Different objective functions as well as methods have been proposed in the literature
to improve the power system observability as follows:
• MIP formulation for optimal PMU placement [3]
• Multi-stage MIP-based PMU allocation [2]
• Optimal PMU Placement considering measurement Loss and branch outage [4]
• Weighted least squares algorithm for optimal PMU placement [5]
• Redundancy and observability analysis of conventional measurement and
PMU [6]
• Semi-definite Programming for optimal PMU placement considering fixed
channel capacity [7]
References 225
• Optimal PMU placement for full network observability using tabu search
algorithm [8]
• Multi-objective optimal PMU placement using a non-dominated sorting differ-
ential evolution algorithm [9]
• Iterated Local Search for optimal PMU allocation [10]
References
1. V. Basetti, A.K. Chandel, Optimal PMU placement for power system observability using
Taguchi binary bat algorithm. Measurement 95, 8–20 (2017)
2. D. Dua, S. Dambhare, R.K. Gajbhiye, S.A. Soman, Optimal multistage scheduling of PMU
placement: an ILP approach. IEEE Trans. Power Delivery 23(4), 1812–1820 (2008)
4. C. Rakpenthai, S. Premrudeepreechacharn, S. Uatrongjit, N.R. Watson, An optimal PMU
placement method against measurement loss and branch outage. IEEE Trans. Power Delivery
22(1), 101–107 (2007)
5. N.M. Manousakis, G.N. Korres, A weighted least squares algorithm for optimal PMU
placement. IEEE Trans. Power Syst. 28(3), 3499–3500 (2013)
6. J.B.A. London, S.A.R. Piereti, R.A.S. Benedito, N.G. Bretas, Redundancy and observability
analysis of conventional and PMU measurements. IEEE Trans. Power Syst. 24(3), 1629–1630
(2009)
7. N.M. Manousakis, G.N. Korres, Optimal PMU placement for numerical observability
considering fixed channel capacity; a semidefinite programming approach. IEEE Trans. Power
Syst. 31(4), 3328–3329 (2016)
8. J. Peng, Y. Sun, H.F. Wang, Optimal PMU placement for full network observability using Tabu
search algorithm. Int. J. Electr. Power Energy Syst. 28(4), 223–231 (2006)
9. C. Peng, H. Sun, J. Guo, Multi-objective optimal PMU placement using a non-dominated
sorting differential evolution algorithm. Int. J. Electr. Power Energy Syst. 32(8), 886–892
(2010)
10. M. Hurtgen, J.-C. Maun, Optimal PMU placement using iterated local search. Int. J. Electr.
Power Energy Syst. 32(8), 857–860 (2010)
Chapter 9
Topics in Transmission Operation and Planning
This chapter provides a solution for some transmission network operation and
planning studies in GAMS. The transmission investment regarding building new
lines and power flow controllers (phase shifter), sensitivity factors, and transmission
switching have been discussed in this chapter. The GAMS code for solving each
optimization problem is developed and discussed.
9.1 Transmission Network Planning
The question to be answered in transmission expansion planning (TEP) is when and

which right of way should be selected to build a new line, perform reconductoring,
build a new substation, or install power flow controllers. As a matter of fact,
obtaining the public acceptance for building new transmission lines has become
a challenging issue for the transmission asset owners. This also makes it difficult
for transmission system operator to keep the technical and economic performance
of transmission network high. The objective function is usually defined as the total
operation and planning costs. Different models have been proposed for TEP purpose
such as:
• Probabilistic TEP considering load and wind power generation uncertainties [1]
• MIP-based multi-stage TEP considering losses, generator costs, and the N 1
security constraints [2]
• Genetic algorithm-based TEP considering demand uncertainty [3]
• Congestion reduction-based TEP [4]
• Robust optimization-based TEP considering the uncertainties of renewable
generation and load [5]
• Branch and bound algorithm for TEP [6]
• MIP-based TEP model considering different demand levels, N 1 network
security constraints as well as environmental constraints [7]

DOI 10.1007/978-3-319-62350-4_9
228 9 Topics in Transmission Operation and Planning
• Multi-objective TEP considering total social costs, maximum regret (robustness

criterion), and maximum adjustment cost (flexibility criterion) as three objective
functions [8]
• Monte-Carlo-based TEP considering random outages of generating units and
transmission lines as well as inaccuracies in the long-term load forecasting [9]
• An interior point method considering full AC power flow constraints [10]
• Multi-objective TEP considering investment cost, reliability (both adequacy and
security), and congestion costs [11]
• Chance constrained TEP consideration of load and wind uncertainties [12].
9.1.1 TEP with New Lines Option
The transmission expansion planning is formulated in (9.1). It is assumed that the

only planning option is building new lines.
OF D T OPC C INVC (9.1a)

X X
OPC D bg Pg C VOLL LSi (9.1b)
g2˝G i
0 1
X
INVC D @0ij C ˛ijk A Cij (9.1c)
k;ij
Pkij Bij .ıi ıj / .1 ˛ijk /M (9.1d)
Pkij Bij .ıi ıj / .1 ˛ijk /M (9.1e)

X X
Pg C LSi Li D Pij W i i 2 ˝B (9.1f)
g2˝Gi j2˝ì
Pmax k k max k
ij ˛ij Pij Pij ˛ij ij 2 ˝` (9.1g)
Pmin
g Pg Pmax
g (9.1h)
if 0ij D 1 then ˛ijkD1 D 1 (9.1i)
1
Bij D (9.1j)
xij
˛ijk 2 f0; 1g (9.1k)
k 2 f1; 2; 3; 4g (9.1l)
9.1 Transmission Network Planning 229
Fig. 9.1 Six-bus Garver transmission network (base case)
The objective function in (9.1a) consists of operational costs (OPC) and invest-
ment costs (INVC). The operational costs are calculated in (9.1b) while the
investment costs are calculated in (9.1c). (9.1d), and (9.1e) model, the power flow
on branch connecting bus i to bus j. In (9.1d) and (9.1e) there is a parameter M. It is
also called big M in the literature [13]. This parameter is selected as follows:
M D maxij Bij .ıi ıj /
The power balance between the generated power, load shedding, demand, and
line flows is ensured by (9.1f). The line flow limits are modeled in (9.1g) and the
impacts of line investment decision ˛ijk are formulated. The generation operating
limits are given in (9.1h). The initial status of each line is described in (9.1i).
The six-bus Garver transmission network [14] is shown in Fig. 9.1.
Transmission expansion planning data are given in Table 9.1. The existing and
potential right of ways in addition to the reactances and flow limits are provided
there. The investment costs (Cij ) are given in million $. The VOLL is assumed to
be 1000 $/MW h. There are three generating units with different operating costs as
shown in Fig. 9.1. Although the generator number 3 is the cheapest unit, however,
it is not connected to the grid. The candidate right of ways are the existing ones
(indicated by solid lines and the dashed ones as indicated in Fig. 9.1). The GAMS
code for solving the DC power flow-based TEP is provided in GCode 9.1.
Table 9.1 Transmission From (i) To (j) xij fNij Cij 0ij
expansion planning data for
Garver six-bus transmission 1 2 0.40 100 40 1
network 1 4 0.60 80 60 1
1 5 0.20 100 20 1
2 3 0.20 100 20 1
2 4 0.40 100 40 1
2 6 0.30 100 30 0
3 5 0.20 100 20 1
4 6 0.30 100 30 0
GCode 9.1 DC-OPF-based TEP
S e t s b u s / 1 6 / , s l a c k ( b u s ) / 1 / , Gen / g1g3 / , k / k1k4 / ;

S c a l a r s S b a s e / 1 0 0 / , M / 1 0 0 0 / ; a l i a s ( bus , node ) ;
b pmin pmax
g1 20 0 400
g2 30 0 400
g3 10 0 600;
s e t GBconect ( bus , Gen ) c o n n e c t i v i t y i n d e x o f e a c h g e n e r a t i n g u n i t t o e a c h b u s
/1 . g1
3 . g2
6 . g3 / ;
Pd
1 80
2 240
3 40
4 160
5 240;
X LIMIT C o s t s t a t
1 . 2 0 . 4 100 40 1
1 . 4 0 . 6 80 60 1
1 . 5 0 . 2 100 20 1
2 . 3 0 . 2 100 20 1
2 . 4 0 . 4 100 40 1
2 . 6 0 . 3 100 30 0
3 . 5 0 . 2 100 20 1
4 . 6 0 . 3 100 30 0;
S e t c o n e x ( bus , node ) Bus c o n n e c t i v i t y m a t r i x ;
c o n e x ( bus , node ) $ ( b r a n c h ( bus , node , ’ x ’ ) ) = y e s ; c o n e x ( bus , node ) $ c o n e x ( node , b u s ) = y e s ;
b r a n c h ( bus , node , ’ x ’ ) $ b r a n c h ( node , bus , ’ x ’ ) = b r a n c h ( node , bus , ’ x ’ ) ;
b r a n c h ( bus , node , ’ c o s t ’ ) $ b r a n c h ( node , bus , ’ c o s t ’ ) = b r a n c h ( node , bus , ’ c o s t ’ ) ;
b r a n c h ( bus , node , ’ s t a t ’ ) $ b r a n c h ( node , bus , ’ s t a t ’ ) = b r a n c h ( node , bus , ’ s t a t ’ ) ;
b r a n c h ( bus , node , ’ L i m i t ’ ) $ ( b r a n c h ( bus , node , ’ L i m i t ’ ) = 0 ) = b r a n c h ( node , bus , ’ L i m i t ’ ) ;
M=smax ( ( bus , node ) $ c o n e x ( bus , node ) , b r a n c h ( bus , node , ’ b i j ’ )p i 4 / 3 ) ;
V a r i a b l e s OF , P i j ( bus , node , k ) , Pg ( Gen ) , d e l t a ( b u s ) , LS ( b u s ) ;
b i n a r y v a r i a b l e a l p h a ( bus , node , k ) ; a l p h a . l ( bus , node , k ) = 1 ;
a l p h a . f x ( bus , node , k ) $ ( c o n e x ( bus , node ) and o r d ( k ) =1 and b r a n c h ( node , bus , ’ s t a t ’ ) )
=1;
E q u a t i o n s const1A , c o n s t 1 B , c o n s t 1 C , const1D , c o n s t 1 E , c o n s t 2 , c o n s t 3 ;
c o n s t 1 A ( bus , node , k ) $ c o n e x ( node , b u s ) . . P i j ( bus , node , k )
b r a n c h ( bus , node , ’ b i j ’ ) ( d e l t a ( b u s )d e l t a ( node ) ) = l = M(1 a l p h a ( bus , node , k ) ) ;
c o n s t 1 B ( bus , node , k ) $ c o n e x ( node , b u s ) . . P i j ( bus , node , k )
b r a n c h ( bus , node , ’ b i j ’ ) ( d e l t a ( b u s )d e l t a ( node ) ) =g=M(1 a l p h a ( bus , node , k ) ) ;
c o n s t 1 C ( bus , node , k ) $ c o n e x ( node , b u s ) . . P i j ( bus , node , k ) = l =
a l p h a ( bus , node , k )b r a n c h ( bus , node , ’ L i m i t ’ ) / S b a s e ;
c o n s t 1 D ( bus , node , k ) $ c o n e x ( node , b u s ) . . P i j ( bus , node , k ) =g=
c o n s t 1 E ( bus , node , k ) $ c o n e x ( node , b u s ) . . a l p h a ( bus , node , k ) =e= a l p h a ( node , bus , k ) ;
c o n s t 2 ( b u s ) . . LS ( b u s ) +sum ( Gen$GBconect ( bus , Gen ) , Pg ( Gen ) )BusData ( bus , ’ pd ’ ) / S b a s e
=e =+sum ( ( k , node ) $ c o n e x ( node , b u s ) , P i j ( bus , node , k ) ) ;
const3 . . OF=g =28760(sum ( Gen , Pg ( Gen )GenData ( Gen , ’ b ’ )S b a s e )
+1000 s b a s esum ( bus , LS ( b u s ) ) )
+1 e6sum ( ( bus , node , k ) $ c o n e x ( node , b u s ) , 0 . 5 b r a n c h ( bus , node , ’ c o s t ’ )a l p h a ( bus , node , k )
$ ( o r d ( k ) >1 o r b r a n c h ( node , bus , ’ s t a t ’ ) = 0 ) ) ;
Model l o a d f l o w / a l l / ; o p t i o n o p t c r =0;
LS . up ( b u s ) = BusData ( bus , ’ pd ’ ) / S b a s e ; LS . l o ( b u s ) = 0 ;
P i j . up ( bus , node , k ) $ ( ( c o n e x ( bus , node ) ) ) =1 b r a n c h ( bus , node , ’ L i m i t ’ ) / S b a s e ;
P i j . l o ( bus , node , k ) $ ( ( c o n e x ( bus , node ) ) )=1b r a n c h ( bus , node , ’ L i m i t ’ ) / S b a s e ;
S o l v e l o a d f l o w min OF u s MIP ;
9.1.1.1 Two Years Planning Period
The optimal TEP solution for 2 years planning period (T D 2 8760 h) is shown in
Fig. 9.2. The total operating cost is M$227.142, and the investment costs are M$110.
The total costs would be M$337.140.
The optimal solutions dictate that the branch connecting bus 1 to bus 5 should be
reinforced with one additional line. Bus 6 to bus 2 should be connected using three
lines.
9.1.1.2 Ten Years Planning Period
The optimal TEP solution for 2 years planning period (T D 10 8760 h) is

shown in Fig. 9.3. The total operating cost is M$871.410 and the investment costs
are M$162.853. The total costs would be M$1034.263. The optimal solutions
dictates that the branch2–3 and branch1–5 should be reinforced with one additional
line. Bus 6 to bus 2 should be connected using four lines. Bus 6 to bus 4
should be connected using two lines. It should be noted that the current model is
simplistic. It needs to consider more realistic constraints. Some of them are listed
as follows:
Fig. 9.2 Optimal TEP solution for 2 years planning period (T D 2 8760 h)
Fig. 9.3 Optimal TEP solution for 10 years planning period (T D 10 8760 h)
• The model is a single period. The load duration curve (LDC) or a discrete LDC
with some demand levels and their associated duration should be considered.
• The contingencies should be taken into account. This will ensure the system
operator to keep the light on even some network/generation assets fail. The
failure of transmission lines or generating units might cause overloading of the
remaining lines and operation of over-current relays and cause cascaded failures.
It might even lead to black out.
• The AC power flow constraints should be checked to make sure no voltage
constraint or line limit is violated.
• The voltage stability issue should also be checked.
• The model is trying to minimize the total operating cost plus the investment costs.
In deregulated environment, the transmission asset owner and the transmission
system operator are two independent entities. This makes the problem more
complicated since these two entities would have different objective functions.
The asset owner is trying to maximize its benefit and make money by making
investments. On the other hand, the system operator is trying to maximize the
social welfare. The multi-objective techniques [8] or complementarity models
[15] provide the suitable answer to this challenge.
• The demand grows should be considered.
• The formulated transmission planning model is a static model. This means that
the decision is made in order to make the system capable of answering for the
needs of next N years. The dynamic of investment or timing of investment and
time value of money is neglected.
• The uncertainties of the electricity market, renewable energies, demand, and
regulatory frameworks should be considered to make the model robust against
future scenarios.
• The VOLL is assumed to be the same for all demands at different buses.
However, the importance level of all demands is not the same. This can reduce
the investment requirements for TEP.
9.1.2 TEP with New Lines and Power Flow Controller Option
This section investigates the impact of power flow controller as a planning option.
The phase shifter is used as the power flow controller device. The role of the phase
shifter is depicted in Fig. 9.4. The relation between the power flow, voltage angles
difference across the branchij , and the line susceptance is given in (9.2).
ıi ıj C ij
Pij D D Bij .ıi ıj C ij / (9.2a)
xij
ijmax ij Cijmax (9.2b)
ij D ji (9.2c)
Fig. 9.4 The phase shifter function in power flow control
Although the voltage phase shift is a discrete variable in reality but it is modeled as
a continuous variable for simplicity in (9.2).
OF D T OPC C INVC (9.3a)

X X
OPC D bg Pg C VOLL LSi (9.3b)
g2˝G i
0 1
X X
INVC D @0ij C ˛ijk A Cij C Iijk ij (9.3c)
k;ij k;ij
Pkij Bij .ıi ıj C ijk / .1 ˛ijk /M (9.3d)

Pkij Bij .ıi ıj C ijk / .1 ˛ijk /M (9.3e)
ijmax Iijk ijk Cijmax Iijk (9.3f)
ij D ji (9.3g)
Iijk ˛ijk (9.3h)
X X
Pg C LSi Li D Pij W i i 2 ˝B (9.3i)
g2˝Gi j2˝ì
Pmax k k max k
ij ˛ij Pij Pij ˛ij ij 2 ˝` (9.3j)
Pmin
g Pg Pmax
g (9.3k)
if 0ij D 1 then ˛ijkD1 D 1 (9.3l)
1
Bij D (9.3m)
xij
˛ijk ; Iijk 2 f0; 1g (9.3n)
k 2 f1; 2; 3; 4g (9.3o)
Fig. 9.5 The optimal decisions regarding the location of phase shifter and new transmission lines
2 years plan (T D 2 8760)
The phase shifter impact is shown in (9.3d) and (9.3e). The investment decision
regarding the phase shifter is reflected in (9.3f). The phase shifter can be installed
on lines that initially exist or built lines as shown in (9.3h). The investment cost for
each phase shifter is assumed to be ij = M$0.6. The optimal decisions regarding the
location of phase shifter and new transmission lines 2 years plan (T D 2 8760)
are given in Fig. 9.5. The total operating cost is M$224.256 and the investment costs
are M$110.6. The total costs would be M$334.856.
GCode 9.2 DC-OPF-based TEP considering phase shifter option

S e t s b u s / 1 6 / , s l a c k ( b u s ) / 1 / , Gen / g1g3 / , k / k1k4 / ; a l i a s ( bus , node ) ;
S c a l a r s S b a s e / 1 0 0 / , M / 1 0 0 0 / , T ; T=87602; S e t c o n e x ( bus , node ) ;
b pmin pmax
g1 20 0 400
g2 30 0 400
g3 10 0 600;
S e t GBconect ( bus , Gen ) c o n n e c t i v i t y i n d e x o f e a c h g e n e r a t i n g u n i t t o e a c h b u s
/1 . g1
3 . g2
6 . g3 / ;
Pd
1 80
2 240
3 40
4 160
5 240;
X LIMIT C o s t s t a t
;
c o n e x ( bus , node ) $ ( b r a n c h ( bus , node , ’ x ’ ) ) = y e s ; c o n e x ( bus , node ) $ c o n e x ( node , b u s ) = y e s ;
b r a n c h ( bus , node , ’ c o s t ’ ) $ b r a n c h ( node , bus , ’ c o s t ’ ) = b r a n c h ( node , bus , ’ c o s t ’ ) ;
b r a n c h ( bus , node , ’ L i m i t ’ ) $ ( b r a n c h ( bus , node , ’ L i m i t ’ ) = 0 ) = b r a n c h ( node , bus , ’ L i m i t ’ ) ;
M=smax ( ( bus , node ) $ c o n e x ( bus , node ) , b r a n c h ( bus , node , ’ b i j ’ ) 3.142) ;
V a r i a b l e s OF , P i j ( bus , node , k ) , Pg ( Gen ) , d e l t a ( b u s ) , LS ( b u s ) , P S H i j ( bus , node , k ) ;
b i n a r y v a r i a b l e a l p h a ( bus , node , k ) , I ( bus , node , k ) ; a l p h a . l ( bus , node , k ) = 1 ;
a l p h a . f x ( bus , node , k ) $ ( c o n e x ( bus , node ) and o r d ( k ) =1 and b r a n c h ( node , bus , ’ s t a t ’ ) )
=1;
E q u a t i o n s const1A , c o n s t 1 B , c o n s t 1 C , const1D , c o n s t 1 E , c o n s t 2 , c o n s t 3 ,
const4 , const5 , const6 , const7 ;
c o n s t 1 A ( bus , node , k ) $ c o n e x ( node , b u s ) . . P i j ( bus , node , k )b r a n c h ( bus , node , ’ b i j ’ ) (
d e l t a ( b u s )d e l t a ( node ) + P S H i j ( bus , node , k ) ) = l =M(1 a l p h a ( bus , node , k ) )
;
c o n s t 1 B ( bus , node , k ) $ c o n e x ( node , b u s ) . . P i j ( bus , node , k )b r a n c h ( bus , node , ’ b i j ’ ) (
d e l t a ( b u s )d e l t a ( node ) + P S H i j ( bus , node , k ) ) =g=M(1 a l p h a ( bus , node , k ) )
;
c o n s t 1 C ( bus , node , k ) $ c o n e x ( node , b u s ) . . P i j ( bus , node , k ) = l =
c o n s t 1 D ( bus , node , k ) $ c o n e x ( node , b u s ) . . P i j ( bus , node , k ) =g=
c o n s t 1 E ( bus , node , k ) $ c o n e x ( node , b u s ) . . a l p h a ( bus , node , k ) =e= a l p h a ( node , bus , k ) ;
c o n s t 2 ( b u s ) . . LS ( b u s ) +sum ( Gen$GBconect ( bus , Gen ) , Pg ( Gen ) )
BusData ( bus , ’ pd ’ ) / S b a s e =e =+sum ( ( k , node ) $ c o n e x ( node , b u s ) , P i j ( bus , node , k ) ) ;
c o n s t 3 . . OF=g=T(sum ( Gen , Pg ( Gen )GenData ( Gen , ’ b ’ )S b a s e ) +100000sum ( bus , LS ( b u s ) ) )
+1 e6sum ( ( bus , node , k ) $ c o n e x ( node , b u s ) , 0 . 5 b r a n c h ( bus , node , ’ c o s t ’ )
a l p h a ( bus , node , k ) $ ( o r d ( k ) >1 o r b r a n c h ( node , bus , ’ s t a t ’ ) = 0 ) ) +6 e5 0.5sum ( ( bus , node , k )
$ c o n e x ( node , b u s ) , I ( bus , node , k ) ) ;
c o n s t 4 ( bus , node , k ) $ c o n e x ( node , b u s ) . . P S H i j ( bus , node , k ) + P S H i j ( node , bus , k ) =e = 0 ;
c o n s t 5 ( bus , node , k ) $ c o n e x ( node , b u s ) . . P S H i j ( bus , node , k ) = l = I ( bus , node , k )p i / 8 ;
c o n s t 6 ( bus , node , k ) $ c o n e x ( node , b u s ) . . P S H i j ( bus , node , k ) =g=I ( bus , node , k )p i / 8 ;
c o n s t 7 ( bus , node , k ) $ c o n e x ( node , b u s ) . . I ( bus , node , k ) = l = a l p h a ( bus , node , k ) ;
Model l o a d f l o w / a l l / ; LS . up ( b u s ) = BusData ( bus , ’ pd ’ ) / S b a s e ; LS . l o ( b u s ) = 0 ;
Pg . l o ( Gen ) =GenData ( Gen , ’ Pmin ’ ) / S b a s e ; Pg . up ( Gen ) =GenData ( Gen , ’ Pmax ’ ) / S b a s e ;
P i j . up ( bus , node , k ) $ ( ( c o n e x ( bus , node ) ) ) =1 b r a n c h ( bus , node , ’ L i m i t ’ ) / S b a s e ;
P i j . l o ( bus , node , k ) $ ( ( c o n e x ( bus , node ) ) )=1b r a n c h ( bus , node , ’ L i m i t ’ ) / S b a s e ;
P S H i j . up ( bus , node , k ) = p i / 8 ; P S H i j . l o ( bus , node , k )=p i / 8 ; o p t i o n o p t c r = 0 ;
S o l v e l o a d f l o w min OF u s MIP ;
The optimal decisions regarding the location of phase shifter and new transmis-
sion lines 10 years plan (T D 10 8760) are given in Fig. 9.6. The total operating
cost is M$805.920 and the investment costs are M$200.600. The total costs would
be M$1006.520.
Fig. 9.6 The optimal decisions regarding the location of phase shifter and new transmission lines
10 years plan (T D 10 8760)
Nomenclature
Indices and Sets

g Index of thermal generating units
i; j Index of network buses
˝G Set of all thermal generating units
˝Gi Set of all thermal generating units connected to bus i
˝ì Set of all buses connected to bus i
˝B Set of network buses
Parameters
M Big number
T Duration of planning period (h)
Li Electric power demand in bus i at time t
bg Fuel cost coefficient of thermal unit g
0ij Initial status of branch connecting bus i to j
Cij Investment cost for branch connecting bus i to j
ij Investment cost of phase shifter in line ij
max = min
Pg Maximum/minimum limits of power generation of thermal unit g
Pmax
ij Maximum power flow limits of branch connecting bus i to j
ijmax Maximum phase shift in line connecting bus i to bus j
xij Reactance of branch connecting bus i to j
Bij Susceptance of branch connecting bus i to j
Variables
Pkij Active power flow of branch k connecting bus i to j (MW)
Pg Active power generated by thermal unit g (MW)
˛ijk Binary variable to model the investment decision regarding the line k at the
right of way ij
Iijk Binary variable to model the investment decision regarding the phase
shifter in line k at the right of way ij
i Locational marginal price in bus i ($/MW h)
LSi Load shedding in bus i (MW)
ij Phase shift in line connecting bus i to bus j
OPC Total operating costs ($)
OF Total costs ($)
INVC Total investment costs ($)
ıi Voltage angle of bus i (rad)
9.2 Sensitivity Factors in Transmission Networks
In this section two important factors are analyzed and calculated namely:
• Generation Shift Factor (GSF)
• Line Outage Distribution Factor (LODF)
9.2.1 Generation Shift Factors
The transmission network planner/operator is always interested to know what

happens to the line flows if any outage happens in generation units. In other words,
what is the influence of generation/demand change in bus i on the line ` (connecting
bus n to bus m)? It is important since there is always a chance for contingen-
cies to happen in generating units. This section will provide an answer to this
question.
The DC power flow equations of the network shown in Fig. 9.7 is provided
in (9.4).
g ı1 ı 2 ı1 ı 3
P1 L1 D C (9.4a)
x12 x13
9.2 Sensitivity Factors in Transmission Networks 239
Fig. 9.7 Three-bus network example for sensitivity factors calculation
g ı2 ı 1 ı2 ı 3
P2 L2 D C (9.4b)
x21 x23
g ı3 ı 2 ı3 ı 1
P3 L3 D C (9.4c)
x32 x31
By substituting the numerical values of xij in (9.4) we will have :
g ı1 ı 2 ı1 ı3
P1 L1 D C (9.5a)
0:1 0:2
g ı2 ı 1 ı2 ı3
P2 L2 D C (9.5b)
0:1 0:25
g ı3 ı 2 ı3 ı1
P3 L3 D C (9.5c)
0:25 0:2
0 1 0 1 0 1
P1 15 10 5 ı1
g
@ A @
(9.4) can be written as : P2 D 10 14 4 D ı2 A where Pi D Pi Li .
@ A
P3 5 4 9 ı3
0 1
15 10 5
Suppose B D @ 10 14 4 A then we can have the relation between the bus
5 4 9
angles and the net injections as a linear matrix form.
P D Bı (9.6)
In normal DC power flow, the P vector is known and the decision maker’s goal is to
find the ı vector. The problem is that matrix B is singular and does not have a matrix
inverse.
The good news is that the bus angle at slack bus is known to be zero. If the slack
bus is bus 1 then we can have the following matrix form:

P2 14 4 ı2
D (9.7)
P3 4 9 ı3
Now the matrix can be inversed as follows

! !1 !

ı2 14 4 P2 0:0818 0:0364 P2
D D (9.8)
ı3 4 9 P3 0:0364 0:1273 P3
We can write down (9.8) as a general form
ıred D Xred Pred (9.9)

Xred D B1
red (9.10)
where Bred is the B matrix after eliminating the row and column of slack bus.
If the network has n buses then Bred would be a square .n 1/ .n 1/
matrix. For the rest of this chapter, whenever X appears in any equation it is
representing Xred .
Now we are about to investigate the impact of change in power injection in bus
n on the line flow between bus i and bus j. The flow on line connecting bus i to bus
j is calculated as follows:
ıi ı j
fij D (9.11)
xij
Now we assume that any change in injected power at bus m will be compensated by
the slack bus. In order to calculate the flow change in line i j we need to use the
following equation:
ıi ıj
fij D (9.12)
xij
The value of xij remains constant but the voltage angles would change if P is
happening at bus m (and P at slack bus). Referring to (9.9) we would have
ı D XP (9.13)
0 1 0 1 0 1
ı2 X22 X2;n1 X2n 0
B ı3 C B X32 X3;n1 X3n C B 0 C
B C B C B C
B :: C B :: : : :: :: C B ::C
B : C B : : : : C B :C
B C B C B C
B ı C B Xm2 Xm;n1 C B C
B m C B Xmn C B Pm C
B : C B :: : : :: C C B C
B :: C B : :
::
: : C B ::: C
B C B B C
B C B C B C
B ıi C DB Xi2 Xi;n1 Xin C B 0 C
B : C B :: : : :: CC B C
B : C B :: B :: C
B : C B : : : : C B : C
B C B C B C
B ıj C B Xj2 Xj;n1 Xjn C B 0 C
B C B C B C
B :: C B :: :: :: :: C B :: C
B : C B : : : C
: C B C
B C B B : C
@ ın1 A @ Xn1;2 Xn1;n1 Xn1;n A @ 0 A
ın n11
Xn;2 Xn;n1 Xn;n n1n1 0 n11
(9.14)
The changes of voltage angles in bus i and j are calculated as follows:
ıi D Xim Pm (9.15)

ıj D Xjm Pm (9.16)
Xim Pm Xjm Pm
fij D (9.17)
xij
The (9.17) states that the sensitivity of the flow in line ij of power change in bus m
is obtained as follows:
fij Xim Xjm
aijm D D (9.18)
Pm xij
Using the (9.18) is useful in calculating the line flows in post-contingency period.
post pre
fij D fij C aijm Pm (9.19)
If the post-contingency line flow at line ij after the failure of a generating unit at
bus m (producing Pgm MW) is to be calculated then the following equation can be
used:
post pre
fij D fij C aijm .Pgm / (9.20)
If the post-contingency line flow at line ij after the disconnection of load at bus
m is to be calculated then the following equation can be used:
post pre
fij D fij C aijm .Lm / (9.21)
Now let’s calculate the Generation Shift Factors (aijm ) for the network shown
in Fig. 9.7.
The GCode 9.3 described the GAMS code for calculating the generation shift
factors.
GCode 9.3 Generation Shift Factor Calculation
S e t s b u s / 1 3 / , s l a c k ( b u s ) / 1 / , Gen / g1g3 / , n o n s l a c k ( b u s ) / 2 3 / ;
s c a l a r s Sbase / 1 0 0 / ;
a l i a s ( bus , node , s h i n , k n o t ) ; a l i a s ( n o n s l a c k , n o n s l a c k j ) ;
X LIMIT s t a t
1 . 2 0 . 1 100 1
1 . 3 0 . 2 80 1
2 . 3 0 . 2 5 100 1;
s e t c o n e x ( bus , node ) Bus c o n n e c t i v i t y m a t r i x l ;
c o n e x ( bus , node ) $ ( b r a n c h ( bus , node , ’ x ’ ) ) = y e s ;
c o n e x ( bus , node ) $ c o n e x ( node , b u s ) = y e s ;
b r a n c h ( bus , node , ’ L i m i t ’ ) $ ( b r a n c h ( bus , node , ’ L i m i t ’ ) = 0 ) = b r a n c h ( node , bus , ’
Limit ’ ) ;
P a r a m e t e r B m a t r i x ( bus , node ) , Binv ( bus , node ) , Flow ( bus , node ) ; A l i a s ( bus ,
knot ) ;
B m a t r i x ( bus , node ) $ ( c o n e x ( node , b u s ) )=b r a n c h ( bus , node , ’ b i j ’ ) ;
B m a t r i x ( bus , b u s ) =sum ( k n o t $ c o n e x ( k n o t , b u s ) , B m a t r i x ( bus , k n o t ) ) ;
p a r a m e t e r B r e d u c e d ( n o n s l a c k , n o n s l a c k j ) ,GSHF ( bus , node , k n o t ) ;
Breduced ( nonslack , n o n s l a c k j ) =Bmatrix ( nonslack , n o n s l a c k j ) ;
parameter inva ( nonslack , nonslackj ) ’ i n v e r s e of a ’ ;
e x e c u t e _ u n l o a d ’ a . gdx ’ , n o n s l a c k , B r e d u c e d ;
e x e c u t e ’ = i n v e r t . e x e a . gdx n o n s l a c k B r e d u c e d b . gdx i n v a ’ ;
e x e c u t e _ l o a d ’ b . gdx ’ , i n v a ;
Binv ( n o n s l a c k , n o n s l a c k j ) = i n v a ( n o n s l a c k , n o n s l a c k j ) ;
GSHF ( bus , node , k n o t ) $ c o n e x ( bus , node ) =
b r a n c h ( bus , node , ’ b i j ’ ) ( Binv ( bus , k n o t )Binv ( node , k n o t ) ) ;
D i s p l a y B m a t r i x , Binv , GSHF ;
The GCode 9.3 has no solve statement or variable. This is because no optimiza-
tion is going to be done. First of all, the B matrix is calculated in GCode 9.3 as
follows:
Parameter Bmatrix(bus,node);
Alias(bus,knot);
Bmatrix(bus,node)$conex(node,bus)=-branch(bus,node,’bij’);
Bmatrix(bus,bus)=sum(knot$conex(knot,bus),-Bmatrix(bus,knot));
This would calculate the Bmatrix as follows:
—- 68 PARAMETER Bmatrix
1 2 3
1 15.000 -10.000 -5.000
2 -10.000 14.000 -4.000
3 -5.000 -4.000 9.000
Now we need to eliminate the row and column containing the slack bus. In order
to do this, another parameter called Breduced is defined but over the non-slack bus
set. This set is defined over all buses except the slack buses.
Parameter Breduced(nonslack,nonslackj);
Breduced(nonslack,nonslackj)=Bmatrix(nonslack,nonslackj);
Parameter inva(nonslack,nonslackj) ’inverse of a’;
execute_unload ’a.gdx’, nonslack, Breduced;
execute ’=invert.exe a.gdx nonslack Breduced b.gdx inva’;
execute_load ’b.gdx’, inva;
This would calculate the inverse matrix of the reduced B matrix.
—- 68 PARAMETER Binv
2 3
2 0:082 0:036
3 0:036 0:127
Now all needed data for calculating the aijm is available. The generation shift
factors are calculated as follows:
Binv(nonslack,nonslackj)=inva(nonslack,nonslackj);
GSHF(bus,node,knot)$conex(bus,node)=branch(bus,node,’bij’)*(Binv
(bus,knot)-Binv(node,knot));
Using the GCode 9.3, the generation shift factors are calculated as Table 9.2.
The generation shift factors (GSHF) have the following interesting features as
follows:
• The procedure for calculating the GSHF does not involve any optimization
• The values of GSHF can be calculated in advance and be used in real-time
applications.
• The values of GSHF do not depend on the loading condition of the network.
These coefficients only depend on the network topology. If the network topology
is changed (due to transmission outage or switching), GSHF should be recalcu-
lated.
• The technique we used for calculating the GSHF is assuming that the changes
at any bus are quickly compensated by the slack bus. In case the changes are
compensated by multiple generating units then the calculation procedure would
be slightly different [16].
• The GSHF are also useful for understanding how to reduce the line loading. For
example, suppose we need to reduce the flow at line 3 1. As we can see that
bus 3 has the largest GSHF equal to 0.636. It means that if we can reduce the
generation at bus 3, then a negative value will be added to the initial flow of line
3 1. If the initial flow from bus 3 to bus 1 is positive, then it would reduce.
Now let’s check the values obtained in Table 9.2. For this purpose, the base power
flow is solved and is shown in Fig. 9.8. The values of line flow, as well as the voltage
angles, are specified in this figure. Suppose it is desired to increase the flow on the
0
line connecting the bus 3 to bus 2. The power flow on this line is f32 D 2:7 MW.
32 32
Using the Table 9.2 states that a2 D 0:182 and a3 D 0:364. Let’s investigate the
impact of changes in power injections in different buses on transmission lines.
9.2.1.1 Demand Increase in L2 by 10 MW
The demand in bus 2 is increased by 10 MW and the new line flows are depicted in
Fig. 9.9. As it can be seen in this figure, the new line flow of line 3 2 is 4.5 MW.
This is obtained using a GAMS code. Let’s calculate the new line flow using the
Table 9.2. The new line flow is calculated as follows:
post pre
f32 D f32 C a32
2 .P2 / D 0:027 0:182 .0:1/ D 0:0452 pu (9.22)
post pre
f12 D f12 C a12
2 .P2 / D 0:177 0:818 .0:1/ D 0:0952 pu (9.23)
Table 9.2 Generator shift Line Bus (m)

factors (aijm ) for three-bus
ij 2 3
network
1–2 0.818 0.364
1–3 0.182 0.636
2–1 0.818 0.364
2–3 0.182 0.364
3–1 0.182 0.636
3–2 0.182 0.364
Fig. 9.8 The DC power flow solution for three-bus network
Fig. 9.9 The DC power flow solution for three-bus network after increasing the demand at bus 2
for 10 MW.
post pre
f31 D f31 C a31
2 .P2 / D 0:123 C 0:182 .0:1/ D 0:1048 pu (9.24)
It should be noted that the flow values as well as the change in power injection at
bus 2 are expressed in pu. P2 is representing the change in bus injection at bus
2 which is 10 MW or 0.1 pu. It can be observed that the results confirm what is
obtained by solving the DC power flow as shown in Fig. 9.9.
Fig. 9.10 The DC power flow solution for three-bus network after increasing the generation at bus
3 for 10 MW.
g
9.2.1.2 Generation Increase in P3 by 10 MW
The generation in bus 3 is increased by 10 MW and the new line flows are depicted
in Fig. 9.10 which are obtained using a GAMS code. Let’s calculate the new line
flow using the GSHF in Table 9.2. The new line flow is calculated as follows:
post pre
f32 D f32 C a32
3 .P3 / D 0:027 C 0:364 .0:1/ D 0:0634 pu (9.25)
post pre
f12 D f12 C a12
3 .P3 / D 0:177 C 0:364 .0:1/ D 0:2134 pu (9.26)
post pre
f31 D f31 C a31
3 .P3 / D 0:123 C 0:636 .0:1/ D 0:1866 pu (9.27)
9.2.2 Line Outage Distribution Factors
The impact of line outages on power flow of other lines is investigated in this section.
Consider the line connecting the bus n and m as shown in Fig. 9.11. In Fig. 9.11a,
the intact network is shown. We need to find out the impact of the line outage of the
branch connecting bus n to bus m on the rest of the network. Suppose the flow of this
0
line is initially equal to fnm . The power from the rest of the network injected to bus n
is equal to power absorption from bus m to the rest of network when no contingency
has happened. We need to find a way to make these flows equal to zero. It is done
using a very smart trick [17]. If we add two injections to the network: CPn at bus
n and another one equal to Pn at bus m the flow on the line nm would change as
shown in Fig. 9.11b. The question is what is the new flow on this line? The change
of flow on line nm can be easily calculated using the following equation:
Fig. 9.11 Line outage modeling using virtual injections. (a) Intact network. (b) Post-contingency
network
ın ım Xnn Pn C Xnm .Pn / .Xmn Pn C Xmm .Pn //
fnm D D
xnm xnm
(9.28)
This means that

Xnn C Xmm 2 Xnm
fnm D Pn (9.29)
xnm
Now the post-contingency line flow is calculated as follows:

post 0
fnm D fnm C fnm (9.30)
post
If the virtual injection to the grid Pn is carefully chosen then fnm D Pn . This
makes the flow from the rest of the network to bus n and m equal to zero (line
outage).
post 0
Pn D fnm D fnm C fnm (9.31)
Combining the (9.29) with (9.31) gives us :
0 Xnn C Xmm 2Xnm

Pn D fnm C .Pn / (9.32)
xnm

Xnn C Xmm 2Xnm 0

Pn 1 D fnm
xnm
0
fnm
Pn D
Xnn CXmm 2Xnm
1 xnm
Now the change in power flow in line ij is calculated as follows:
ıi ıj Xin Pn C Xim .Pn / .Xjn Pn C Xjm .Pn //
fij D D (9.33)
xij xij
Xin Xim Xjn C Xjm
fij D Pn (9.34)
xij
fij Xin Xim Xjn C Xjm
LODFij;nm D D (9.35)
0
fnm xij .1 Xnn CXxmm
nm
2Xnm
/
The GCode 9.4 is developed to calculate the Line outage distribution factors for the
network shown in Fig. 9.7.
GCode 9.4 Line outage distribution factor calculation
Sets
bus / 1 3 / ,
s l a c k ( bus ) / 1 / ,
Gen / g1g3 / ,
n o n s l a c k ( bus ) / 2 3 / ;
Scalars Sbase / 1 0 0 / ;
A l i a s ( bus , node , s h i n , k n o t ) ;
Alias ( nonslack , nonslackj ) ;

X LIMIT stat
1 . 2 0 . 1 100 1
1 . 3 0 . 2 80 1
2 . 3 0 . 2 5 100 1;
S e t c o n e x ( bus , node ) Bus c o n n e c t i v i t y m a t r i x l ;
c o n e x ( bus , node ) $ ( b r a n c h ( bus , node , ’ x ’ ) ) = y e s ;

c o n e x ( bus , node ) $ c o n e x ( node , b u s ) = y e s ;

b r a n c h ( bus , node , ’ L i m i t ’ ) $ ( b r a n c h ( bus , node , ’ L i m i t ’ ) = 0 ) = b r a n c h ( node , bus , ’
Limit ’ ) ;
P a r a m e t e r B m a t r i x ( bus , node ) , Binv ( bus , node ) ;

A l i a s ( bus , k n o t ) ;
B m a t r i x ( bus , node ) $ ( c o n e x ( node , b u s ) )=b r a n c h ( bus , node , ’ b i j ’ ) ;

B m a t r i x ( bus , b u s ) =sum ( k n o t $ c o n e x ( k n o t , b u s ) , B m a t r i x ( bus , k n o t ) ) ;
p a r a m e t e r B r e d u c e d ( n o n s l a c k , n o n s l a c k j ) ,GSHF ( bus , node , k n o t ) , X0 ( bus , node ) ;
Breduced ( nonslack , n o n s l a c k j ) =Bmatrix ( nonslack , n o n s l a c k j ) ;
P a r a m e t e r i n v a ( n o n s l a c k , n o n s l a c k j ) ’ i n v e r s e o f a ’ , D f a c t o r ( bus , node , k n o t ,
shin ) ,
9.3 Transmission Network Switching 249
c o n t i n g e n c y ( bus , node , k n o t , s h i n ) ;
e x e c u t e _ u n l o a d ’ a . gdx ’ , n o n s l a c k , B r e d u c e d ;
e x e c u t e ’ = i n v e r t . e x e a . gdx n o n s l a c k B r e d u c e d b . gdx i n v a ’ ;
e x e c u t e _ l o a d ’ b . gdx ’ , i n v a ;
Binv ( n o n s l a c k , n o n s l a c k j ) = i n v a ( n o n s l a c k , n o n s l a c k j ) ;
GSHF ( bus , node , k n o t ) $ c o n e x ( bus , node ) =
b r a n c h ( bus , node , ’ b i j ’ ) ( Binv ( bus , k n o t )Binv ( node , k n o t ) ) ;
d f a c t o r ( bus , node , k n o t , s h i n ) $ ( c o n e x ( bus , node ) and c o n e x ( k n o t , s h i n )

and ( o r d ( b u s ) > o r d ( node ) )
and ( o r d ( k n o t ) > o r d ( s h i n ) )
and ( o r d ( b u s ) <> o r d ( k n o t ) o r o r d ( node ) <> o r d ( s h i n ) ) ) =
b r a n c h ( bus , node , ’ b i j ’ ) ( Binv ( bus , k n o t )Binv ( bus , s h i n )Binv ( node , k n o t )
+ Binv ( node , s h i n ) ) /(1 b r a n c h ( k n o t , s h i n , ’ b i j ’ ) ( Binv ( k n o t , k n o t )
+ Binv ( s h i n , s h i n )2Binv ( k n o t , s h i n ) ) ) ;
D i s p l a y B m a t r i x , Binv , GSHF , d f a c t o r ;
Line outage distribution factors (LODFij;nm ) are described in Table 9.3.

The branch data for IEEE RTS 24-bus network is provided in Table 9.4. The
LODF and GSF coefficients are calculated for IEEE RTS 24-bus (Fig. 9.12) and are
given in Tables 9.5 and 9.6, respectively.
9.3 Transmission Network Switching
The idea of optimal transmission switching (OTS) has been broadly investigated in
the literature [18]. Opening a set of transmission lines would change the network
topology and the line flow patterns. This can be used to relieve the line congestion
in the system and reduce the operating costs. Some research works which used the
transmission switching as a flexibility tool are listed as follows:
Table 9.3 Line outage Line m

distribution factors
ij n 1 2
(LODFij;nm ) for three-bus
network 2–1 3 1 1
3–1 2 1
3–1 3 1
3–2 2 1
3–2 3 1

Rating Rating
1 2 0.0026 0.0139 0.4611 175 11 13 0.0061 0.0476 0.0999 500
1 3 0.0546 0.2112 0.0572 175 11 14 0.0054 0.0418 0.0879 500
1 5 0.0218 0.0845 0.0229 175 12 13 0.0061 0.0476 0.0999 500
2 4 0.0328 0.1267 0.0343 175 12 23 0.0124 0.0966 0.2030 500
2 6 0.0497 0.1920 0.0520 175 13 23 0.0111 0.0865 0.1818 500
3 9 0.0308 0.1190 0.0322 175 14 16 0.0050 0.0389 0.0818 500
3 24 0.0023 0.0839 0.0000 400 15 16 0.0022 0.0173 0.0364 500
4 9 0.0268 0.1037 0.0281 175 15 21 0.0032 0.0245 0.2060 1000
5 10 0.0228 0.0883 0.0239 175 15 24 0.0067 0.0519 0.1091 500
6 10 0.0139 0.0605 2.4590 175 16 17 0.0033 0.0259 0.0545 500
7 8 0.0159 0.0614 0.0166 175 16 19 0.0030 0.0231 0.0485 500
8 9 0.0427 0.1651 0.0447 175 17 18 0.0018 0.0144 0.0303 500
8 10 0.0427 0.1651 0.0447 175 17 22 0.0135 0.1053 0.2212 500
9 11 0.0023 0.0839 0.0000 400 18 21 0.0017 0.0130 0.1090 1000
9 12 0.0023 0.0839 0.0000 400 19 20 0.0026 0.0198 0.1666 1000
10 11 0.0023 0.0839 0.0000 400 20 23 0.0014 0.0108 0.0910 1000
10 12 0.0023 0.0839 0.0000 400 21 22 0.0087 0.0678 0.1424 500
Fig. 9.12 IEEE RTS 24-bus network

Table 9.5 LODF12;` calculated for IEEE RTS 24-bus network
` D nm 1 2 3 4 5 6 8 9 10 11 12 13 14 15 16 17 18 19 20 21
3 0.611
4 0.712
5 0.706
6 0.712
9 0.291 0.712 0.054
10 0.706 0.712 0.054

11 0.067 0.022
12 0.062 0.028
13 0.002 0.010
14 0.048
16 0.048 0.031
17 0.009
18 0.006
19 0.043
20 0.043
21 0.009 0.006
22 0.003 0.003
23 0.023 0.016 0.043
24 0.154 0.154
251
252
Table 9.6 GSHF`;m calculated for IEEE RTS 24-bus network

Line Bus (m)
ìj 1 2 3 4 5 6 7 8 9 10 11 12 14 15 16 17 18 19 20 21 22 23 24
21 0.44 0.51 0.10 0.25 0.23 0.11 0.01 0.01 0.04 0.01 0.00 0.00 0.01 0.03 0.02 0.02 0.02 0.02 0.01 0.03 0.02 0.01 0.05
31 0.24 -0.22 0.20 0.09 0.16 0.10 0.03 0.03 0.01 0.06 0.00 0.01 0.02 0.06 0.04 0.05 0.05 0.03 0.02 0.05 0.05 0.02 0.11
42 0.24 0.27 0.03 0.37 0.15 0.11 0.00 0.00 0.07 0.06 0.00 0.00 0.00 0.01 0.01 0.01 0.01 0.00 0.00 0.01 0.01 0.00 0.02
51 0.32 0.29 0.10 0.15 0.38 0.01 0.01 0.01 0.05 0.07 0.00 0.00 0.01 0.03 0.02 0.03 0.03 0.02 0.01 0.03 0.03 0.01 0.06
62 0.20 0.22 0.07 0.12 0.07 0.22 0.01 0.01 0.03 0.05 0.00 0.00 0.01 0.02 0.01 0.02 0.02 0.01 0.01 0.02 0.02 0.01 0.04
87 1.00
93 0.09 0.07 0.43 0.03 0.05 0.02 0.06 0.06 0.12 0.01 0.00 0.02 0.05 0.12 0.09 0.10 0.11 0.07 0.05 0.11 0.11 0.04 0.24
94 0.24 0.27 0.03 0.63 0.15 0.11 0.00 0.00 0.07 0.06 0.00 0.00 0.00 0.01 0.01 0.01 0.01 0.00 0.00 0.01 0.01 0.00 0.02
98 0.02 0.02 0.03 0.03 0.05 0.07 0.50 0.50 0.07 0.08 0.00 0.00 0.00 0.01 0.01 0.01 0.01 0.00 0.00 0.01 0.01 0.00 0.02
10 5 0.32 0.29 0.10 0.15 0.62 0.01 0.01 0.01 0.05 0.07 0.00 0.00 0.01 0.03 0.02 0.03 0.03 0.02 0.01 0.03 0.03 0.01 0.06
10 6 0.20 0.22 0.07 0.12 0.07 0.78 0.01 0.01 0.03 0.05 0.00 0.00 0.01 0.02 0.01 0.02 0.02 0.01 0.01 0.02 0.02 0.01 0.04
10 8 0.02 0.02 0.03 0.03 0.05 0.07 0.50 0.50 0.07 0.08 0.00 0.00 0.00 0.01 0.01 0.01 0.01 0.00 0.00 0.01 0.01 0.00 0.02
11 9 0.17 0.17 0.20 0.28 0.12 0.09 0.22 0.22 0.36 0.07 0.12 0.10 0.05 0.03 0.01 0.02 0.02 0.02 0.03 0.02 0.02 0.03 0.09
11 10 0.24 0.24 0.08 0.15 0.31 0.35 0.23 0.23 0.07 0.39 0.11 0.10 0.06 0.00 0.01 0.01 0.01 0.00 0.01 0.01 0.01 0.02 0.03
12 9 0.18 0.19 0.23 0.29 0.13 0.11 0.23 0.23 0.38 0.08 0.11 0.12 0.10 0.10 0.08 0.09 0.09 0.05 0.02 0.09 0.09 0.01 0.15
12 10 0.26 0.25 0.12 0.16 0.33 0.37 0.24 0.24 0.08 0.40 0.11 0.11 0.08 0.06 0.06 0.06 0.06 0.03 0.01 0.06 0.06 0.00 0.08
13 11 0.43 0.43 0.42 0.43 0.43 0.43 0.43 0.43 0.43 0.43 0.63 0.23 0.51 0.41 0.40 0.40 0.40 0.34 0.28 0.40 0.40 0.24 0.41
13 12 0.40 0.40 0.36 0.40 0.41 0.41 0.41 0.41 0.41 0.41 0.23 0.61 0.26 0.29 0.28 0.28 0.29 0.28 0.28 0.29 0.29 0.29 0.32
14 11 0.02 0.02 0.14 0.01 0.00 0.02 0.02 0.02 0.00 0.03 0.14 0.03 0.63 0.38 0.40 0.40 0.39 0.31 0.24 0.39 0.39 0.19 0.29
16 14 0.02 0.02 0.14 0.01 0.00 0.02 0.02 0.02 0.00 0.03 0.14 0.03 0.37 0.38 0.40 0.40 0.39 0.31 0.24 0.39 0.39 0.19 0.29
16 15 0.12 0.12 0.30 0.10 0.09 0.06 0.07 0.07 0.09 0.05 0.00 0.02 0.06 0.66 0.11 0.16 0.29 0.08 0.06 0.41 0.31 0.05 0.52
(continued)
9 Topics in Transmission Operation and Planning
Table 9.6 (continued)
Line Bus (m)
ìj 1 2 3 4 5 6 7 8 9 10 11 12 14 15 16 17 18 19 20 21 22 23 24
21 15 0.03 0.03 0.07 0.02 0.02 0.02 0.02 0.02 0.02 0.01 0.00 0.00 0.01 0.15 0.03 0.31 0.45 0.02 0.01 0.57 0.47 0.01 0.12
21 18 0.02 0.02 0.06 0.02 0.02 0.01 0.01 0.01 0.02 0.01 0.00 0.00 0.01 0.13 0.02 0.27 0.46 0.02 0.01 0.37 0.12 0.01 0.11
22 17 0.00 0.00 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.02 0.00 0.04 0.01 0.00 0.00 0.06 0.41 0.00 0.02
22 21 0.00 0.00 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.02 0.00 0.04 0.01 0.00 0.00 0.06 0.59 0.00 0.02
23 12 0.04 0.04 0.02 0.05 0.05 0.06 0.06 0.06 0.05 0.07 0.01 0.16 0.08 0.13 0.14 0.14 0.14 0.20 0.25 0.14 0.14 0.28 0.09
23 13 0.17 0.17 0.22 0.17 0.16 0.16 0.16 0.16 0.16 0.15 0.13 0.16 0.23 0.31 0.31 0.31 0.31 0.38 0.44 0.31 0.31 0.47 0.27
23 20 0.13 0.13 0.23 0.12 0.11 0.10 0.10 0.10 0.11 0.09 0.14 0.00 0.31 0.44 0.46 0.45 0.45 0.58 0.69 0.44 0.45 0.25 0.36
24 3 0.15 0.15 0.37 0.13 0.11 0.08 0.08 0.08 0.11 0.06 0.01 0.03 0.07 0.18 0.14 0.15 0.16 0.10 0.07 0.17 0.16 0.06 0.35
24 15 0.15 0.15 0.37 0.13 0.11 0.08 0.08 0.08 0.11 0.06 0.01 0.03 0.07 0.18 0.14 0.15 0.16 0.10 0.07 0.17 0.16 0.06 0.65
253
• DC-OPF considering N 1 contingencies [19]

• Co-optimization of unit commitment and transmission switching with N 1
reliability constraints [20]
• Robust transmission switching considering N k contingencies [21]
• Optimal transmission switching considering short-circuit current limitation con-
straints [22]
• Probabilistic security analysis of OTS [23]
• Chance-constrained OTS with guaranteed wind power utilization [24]
• Heuristics OTS based on DC-OPF and AC-OPF [25]
In this section, the OTS is solved using GAMS and then we will discuss how
this model can be improved and extended. The transmission switching problem is
formulated as a MIP model in (9.36).
X
OF D bg Pg (9.36a)
g2˝G
Pij Bij .ıi ıj / .1

ij /M (9.36b)
Pij Bij .ıi ıj / .1

ij /M (9.36c)
X X
Pg C LSi Li D Pij W i i 2 ˝B (9.36d)
g2˝Gi j2˝ì
Pmax max
ij
ij Pij Pij
ij ij 2 ˝` (9.36e)
Pmin
g Pg Pmax
g (9.36f)
X
.1
ij / Nsw ij 2 ˝` (9.36g)
ij
1
Bij D (9.36h)
xij

ij 2 f0; 1g (9.36i)
where
ij is the on/off state of the branch connecting bus i to bus j, and Nsw is the
number of allowed switching actions in the network.
The operating and congestion cost vs number of switched lines in IEEE 118-bus
network (Fig. 9.13) are depicted in Fig. 9.14. The GAMS code for solving the (9.36)
is provided in GCode 9.5.
Fig. 9.13 IEEE 118-bus network

× 104
Operating cost ($) 9
8.9
8.8
8.7
0 1 2 3 4 5 6 7 8 9 10
N
sw
× 104
4
Congestion cost ($)
0
0 1 2 3 4 5 6 7 8 9 10
N
sw
Fig. 9.14 The operating and congestion cost vs number of switched lines in IEEE 118-bus network
GCode 9.5 The OTS GAMS code for IEEE 118-bus network
S e t s b u s / 1 1 1 8 / , s l a c k ( b u s ) / 1 3 / , c o n e x ( bus , node ) ,
GenNo / Gen1Gen54 / , c o u n t e r / c0c10 / ;
Scalars Sbase / 1 0 0 / ;
A l i a s ( bus , t o t a l b u s , node ) ;
T a b l e GenDatanew ( bus , GenNo , )

b pmin pmax ;
Pd ;
T a b l e b r a n c h ( bus , t o t a l b u s , )
x Ilim ;
b r a n c h ( bus , t o t a l b u s , ’ b i j ’ ) $ b r a n c h ( t o t a l b u s , bus , ’ x ’ ) = 1 / b r a n c h ( bus ,
totalbus , ’x ’ ) ;
c o n e x ( bus , node ) $ b r a n c h ( bus , node , ’ x ’ ) = y e s ;
p a r a m e t e r b r a n c h ( bus , t o t a l b u s , ) ,M,NSW, r e p o r t ( c o u n t e r , ) ;
M=smax ( ( bus , node ) $ c o n e x ( bus , node ) , b r a n c h ( bus , node , ’ b i j ’ ) 2 p i / 3 ) ;
P o s i t i v e v a r i a b l e Pg ( GenNo ) ;
V a r i a b l e s P i j ( bus , node ) , d e l t a ( b u s ) ,ROF ;
BINARY VARIABLE SW( bus , node ) ;
E q u a t i o n s c o n s t 0 , c o n s t 1 , c o n s t 2 , c o n s t 3 , const0A ,
c o n s t 0 B , c o n s t 0 C , const0D , c o n s t 0 E , c o n s t 0 F ;
b r a n c h ( bus , node , ’ b i j ’ ) ( d e l t a ( b u s ) d e l t a ( node ) ) ;
c o n s t 1 ( b u s ) . . sum ( GenNo$GenDatanew ( bus , GenNo , ’ Pmax ’ ) , Pg ( GenNo ) )
BusData ( bus , ’ Pd ’ ) / s b a s e =e =+sum ( n o d e $ c o n e x ( node , b u s ) , P i j ( bus , node ) ) ;

const2 . . ROF=g= sum ( ( GenNo , b u s ) $GenDatanew ( bus , GenNo , ’ Pmax ’ ) ,
GenDatanew ( bus , GenNo , ’ b ’ ) Pg ( GenNo ) S b a s e ) ;
c o n s t 0 A ( bus , node ) $ c o n e x ( bus , node ) . .
P i j ( bus , node )b r a n c h ( bus , node , ’ b i j ’ ) ( d e l t a ( b u s ) d e l t a ( node ) ) = l = M(1SW
( bus , node ) ) ;
c o n s t 0 B ( bus , node ) $ c o n e x ( bus , node ) . .
P i j ( bus , node )b r a n c h ( bus , node , ’ b i j ’ ) ( d e l t a ( b u s ) d e l t a ( node ) ) =g=M(1SW
( bus , node ) ) ;
c o n s t 0 C ( bus , node ) $ c o n e x ( bus , node ) . .
P i j ( bus , node ) = l = SW( bus , node ) b r a n c h ( bus , node , ’ I l i m ’ ) ;
c o n s t 0 D ( bus , node ) $ c o n e x ( bus , node ) . .
P i j ( bus , node ) =g=SW( bus , node ) b r a n c h ( bus , node , ’ I l i m ’ ) ;
c o n s t 0 E ( bus , node ) $ c o n e x ( bus , node ) . . SW( node , b u s ) =e=SW( bus , node ) ;
c o n s t 0 F ( bus , node ) $ c o n e x ( bus , node ) . . P i j ( node , b u s ) =e= P i j ( bus , node ) ;
const3 . . 0 . 5 sum ( ( bus , node ) $ c o n e x ( bus , node ) ,1SW( bus , node ) ) = l =NSW;
model BASE / c o n s t 0 , c o n s t 1 , c o n s t 2 / ;
model S w i t c h i n g / c o n s t 1 , c o n s t 2 , const0A , c o n s t 0 B , c o n s t 0 C ,
const0D , c o n s t 0 E , c o n s t 0 F , c o n s t 3
/;
Option Optca =0; Option Optcr =0;
BusData ( bus , ’ Pd ’ ) =1.1 BusData ( bus , ’ Pd ’ ) ;

Pg . l o ( GenNo ) =sum ( bus , GenDatanew ( bus , GenNo , ’ Pmin ’ ) ) / S b a s e ;
Pg . up ( GenNo ) =sum ( bus , GenDatanew ( bus , GenNo , ’ Pmax ’ ) ) / S b a s e ;
d e l t a . up ( b u s ) = p i / 3 ; d e l t a . l o ( b u s )=p i / 3 ; d e l t a . l ( b u s ) = 0 ; d e l t a . f x ( s l a c k )
=0;
P i j . up ( bus , node ) $ c o n e x ( bus , node ) = 1 b r a n c h ( bus , node , ’ I l i m ’ ) ;
P i j . l o ( bus , node ) $ c o n e x ( bus , node ) =1 b r a n c h ( bus , node , ’ I l i m ’ ) ;
S o l v e BASE m i n i m i z i n g ROF u s i n g l p ;
SW. l ( bus , node ) = 1 ; r e p o r t ( ’ c0 ’ , ’OF ’ ) =ROF . l ;
r e p o r t ( ’ c0 ’ , ’NSW’ ) =0.5sum ( ( bus , node ) $ c o n e x ( bus , node ) ,1SW. l ( bus , node ) ) ;
r e p o r t ( ’ c0 ’ , ’ C o n g e s t i o n ’ ) =0.5sum ( ( bus , node ) $ c o n e x ( bus , node ) ,
( c o n s t 1 .m( b u s ) + c o n s t 1 .m( node ) ) P i j . l ( bus , node ) ) ;
l o o p ( c o u n t e r $ ( o r d ( c o u n t e r ) >1) ,
NSW= o r d ( c o u n t e r ) 1;
S o l v e s w i t c h i n g m i n i m i z i n g ROF u s i n g mip ;
r e p o r t ( c o u n t e r , ’OF ’ ) =ROF . l ;
r e p o r t ( c o u n t e r , ’NSW’ ) =0.5sum ( ( bus , node ) $ c o n e x ( bus , node ) ,1SW. l ( bus , node
));
r e p o r t ( c o u n t e r , ’ C o n g e s t i o n ’ ) =0.5sum ( ( bus , node ) $ c o n e x ( bus , node ) ,
( c o n s t 1 .m( b u s ) + c o n s t 1 .m( node ) ) P i j . l ( bus , node
));
);
Increasing the demand in a given area or bus (or equivalently losing the
generation) might cause congestion and increasing the total operating costs. The
OTS can be used to enhance the grid utilization and reduce the line congestions.
A simple analysis is conducted as follows:
• The demand at bus i is increased for 20 MW.
• The DC OPF is solved without considering the transmission switching option
(LP model).
• The DC OPF along with transmission switching option is solved (MIP model).
× 10 4
9.3
No switching LP model
9.2 With switching MIP model
Operating costs ($)
9.1
8.9
8.8
1 10 20 30 40 50 60 70 80 90 100 110 120
Bus
Fig. 9.15 The operating cost vs the connection point of new demand (20 MW) in with/without
transmission switching cases
60 60
Without switching
With switching
Nsw= 2
40 Nsw= 1 40
i
λi
λ
20 20
10 20 30 40 50 60 70 80 90 100 110 120 10 20 30 40 50 60 70 80 90 100 110 120

60 80
Nsw= 3 60
40
λi
i
Nsw= 4
λ
40
20 20
10 20 30 40 50 60 70 80 90 100 110 120 10 20 30 40 50 60 70 80 90 100 110 120

60 60
40 Nsw= 5 40 Nsw= 6
i
λi
λ
20 20
10 20 30 40 50 60 70 80 90 100 110 120 10 20 30 40 50 60 70 80 90 100 110 120

60 60
40 Nsw= 7 40 Nsw= 8
λi
i
λ
20 20
10 20 30 40 50 60 70 80 90 100 110 120 10 20 30 40 50 60 70 80 90 100 110 120

60 60
40 Nsw= 9 40 Nsw= 10
λi
i
λ
20 20
10 20 30 40 50 60 70 80 90 100 110 120 10 20 30 40 50 60 70 80 90 100 110 120
Bus Bus
Fig. 9.16 The LMP values (i ) in ($/MW h) at different buses in with/without transmission
switching cases
The comparison of the total operating costs between the with/without transmission
switching cases are shown in Fig. 9.15. The impact of connecting a new demand to
different buses would cause different changes in total operating costs. In all cases,
using the transmission switching flexibility can reduce the operating costs as shown
in Fig. 9.15.
Using the transmission switching might change the LMP values at different
buses. The LMP values (i ) in $/MW h at different buses in with/without trans-
mission switching cases. As it is shown in Fig. 9.16, when no switching is allowed
(Nsw D 0) then the LMP values of switching and not switching cases are the same.
By increasing the number of switchable lines, the LMP values get closer to each
other. The generation and branch data of IEEE 118 bus are given in Tables 9.7 and
9.8, respectively.
The developed GAMS code for OTS can be improved to consider the following
issues:
Table 9.7 Generation data for 118-bus network
Bus Unit (g) bg ($/MW h) Pmin
g (MW) Pmax
g (MW) Bus Unit (g) bg ($/MW h) Pmin
g (MW) Pmax
g (MW)
1 Gen1 26.2 5.0 30.0 65.0 Gen28 8.3 100.0 420.0
4 Gen2 26.2 5.0 30.0 66.0 Gen29 12.9 80.0 300.0
6 Gen3 26.2 5.0 30.0 69.0 Gen30 15.5 30.0 80.0
8 Gen4 12.9 150.0 300.0 70.0 Gen31 26.2 10.0 30.0
10 Gen5 12.9 100.0 300.0 72.0 Gen32 26.2 5.0 30.0
12 Gen6 26.2 10.0 30.0 73.0 Gen33 37.7 5.0 20.0

15 Gen7 17.8 25.0 100.0 74.0 Gen34 17.8 25.0 100.0
18 Gen8 26.2 5.0 30.0 76.0 Gen35 17.8 25.0 100.0
19 Gen9 26.2 5.0 30.0 77.0 Gen36 12.9 150.0 300.0
24 Gen10 12.9 100.0 300.0 80.0 Gen37 17.8 25.0 100.0
25 Gen11 10.8 100.0 350.0 85.0 Gen38 26.2 10.0 30.0
26 Gen12 26.2 8.0 30.0 87.0 Gen39 10.8 100.0 300.0
27 Gen13 26.2 8.0 30.0 89.0 Gen40 12.9 50.0 200.0
31 Gen14 17.8 25.0 100.0 90.0 Gen41 37.7 8.0 20.0
32 Gen15 26.2 8.0 30.0 91.0 Gen42 22.9 20.0 50.0
34 Gen16 17.8 25.0 100.0 92.0 Gen43 12.9 100.0 300.0
36 Gen17 26.2 8.0 30.0 99.0 Gen44 12.9 100.0 300.0
40 Gen18 26.2 8.0 30.0 100.0 Gen45 12.9 100.0 300.0
(continued)
259
260

Bus Unit (g) bg ($/MW h) Pmin
g (MW) Pmax
g (MW) Bus Unit (g) bg ($/MW h) Pmin
g (MW) Pmax
g (MW)
42 Gen19 17.8 25.0 100.0 103.0 Gen46 37.7 8.0 20.0
46 Gen20 12.3 50.0 250.0 104.0 Gen47 17.8 25.0 100.0
49 Gen21 12.3 50.0 250.0 105.0 Gen48 17.8 25.0 100.0
54 Gen22 17.8 25.0 100.0 107.0 Gen49 37.7 8.0 20.0
55 Gen23 17.8 25.0 100.0 110.0 Gen50 22.9 25.0 50.0
56 Gen24 13.3 50.0 200.0 111.0 Gen51 17.8 25.0 100.0
59 Gen25 13.3 50.0 200.0 112.0 Gen52 17.8 25.0 100.0
61 Gen26 17.8 25.0 100.0 113.0 Gen53 17.8 25.0 100.0
62 Gen27 8.3 100.0 420.0 116.0 Gen54 22.9 25.0 50.0
Table 9.8 Branch data for 118-bus network
Line x Limit Line x Limit Line x Limit Line x Limit Line x Limit Line x Limit
1.2 0.10 1.75 23 . 25 0.08 5.00 44 . 45 0.09 1.75 63 . 64 0.02 5.00 68 . 81 0.02 5.00 98 . 100 0.18 1.75
1.3 0.04 1.75 26 . 25 0.04 5.00 45 . 46 0.14 1.75 64 . 61 0.03 5.00 81 . 80 0.04 5.00 99 . 100 0.08 1.75
4.5 0.01 5.00 25 . 27 0.16 5.00 46 . 47 0.13 1.75 38 . 65 0.10 5.00 77 . 82 0.09 2.00 100 . 101 0.13 1.75
3.5 0.11 1.75 27 . 28 0.09 1.75 46 . 48 0.19 1.75 64 . 65 0.03 5.00 82 . 83 0.04 2.00 92 . 102 0.06 1.75
5.6 0.05 1.75 28 . 29 0.09 1.75 47 . 49 0.06 1.75 49 . 66 0.05 10.00 83 . 84 0.13 1.75 101 . 102 0.11 1.75
6.7 0.02 1.75 30 . 17 0.04 5.00 42 . 49 0.16 3.50 62 . 66 0.22 1.75 83 . 85 0.15 1.75 100 . 103 0.05 5.00
8.9 0.03 5.00 8 . 30 0.05 1.75 45 . 49 0.19 1.75 62 . 67 0.12 1.75 84 . 85 0.06 1.75 100 . 104 0.20 1.75
8.5 0.03 5.00 26 . 30 0.09 5.00 48 . 49 0.05 1.75 65 . 66 0.04 5.00 85 . 86 0.12 5.00 103 . 104 0.16 1.75
9 . 10 0.03 5.00 17 . 31 0.16 1.75 49 . 50 0.08 1.75 66 . 67 0.10 1.75 86 . 87 0.21 5.00 103 . 105 0.16 1.75
4 . 11 0.07 1.75 29 . 31 0.03 1.75 49 . 51 0.14 1.75 65 . 68 0.02 5.00 85 . 88 0.10 1.75 100 . 106 0.23 1.75
5 . 11 0.07 1.75 23 . 32 0.12 1.40 51 . 52 0.06 1.75 47 . 69 0.28 1.75 85 . 89 0.17 1.75 104 . 105 0.04 1.75
11 . 12 0.02 1.75 31 . 32 0.10 1.75 52 . 53 0.16 1.75 49 . 69 0.32 1.75 88 . 89 0.07 5.00 105 . 106 0.05 1.75
2 . 12 0.06 1.75 27 . 32 0.08 1.75 53 . 54 0.12 1.75 68 . 69 0.04 5.00 89 . 90 0.09 10.00 105 . 107 0.18 1.75
3 . 12 0.16 1.75 15 . 33 0.12 1.75 49 . 54 0.15 3.50 69 . 70 0.13 5.00 90 . 91 0.08 1.75 105 . 108 0.07 1.75
7 . 12 0.03 1.75 19 . 34 0.25 1.75 54 . 55 0.07 1.75 24 . 70 0.41 1.75 89 . 92 0.08 10.00 106 . 107 0.18 1.75
11 . 13 0.07 1.75 35 . 36 0.01 1.75 54 . 56 0.01 1.75 70 . 71 0.04 1.75 91 . 92 0.13 1.75 108 . 109 0.03 1.75
12 . 14 0.07 1.75 35 . 37 0.05 1.75 55 . 56 0.02 1.75 24 . 72 0.20 1.75 92 . 93 0.08 1.75 103 . 110 0.18 1.75
13 . 15 0.24 1.75 33 . 37 0.14 1.75 56 . 57 0.10 1.75 71 . 72 0.18 1.75 92 . 94 0.16 1.75 109 . 110 0.08 1.75
14 . 15 0.20 1.75 34 . 36 0.03 1.75 50 . 57 0.13 1.75 71 . 73 0.05 1.75 93 . 94 0.07 1.75 110 . 111 0.08 1.75
12 . 16 0.08 1.75 34 . 37 0.01 5.00 56 . 58 0.10 1.75 70 . 74 0.13 1.75 94 . 95 0.04 1.75 110 . 112 0.06 1.75
15 . 17 0.04 5.00 38 . 37 0.04 5.00 51 . 58 0.07 1.75 70 . 75 0.14 1.75 80 . 96 0.18 1.75 17 . 113 0.03 1.75
16 . 17 0.18 1.75 37 . 39 0.11 1.75 54 . 59 0.23 1.75 69 . 75 0.12 5.00 82 . 96 0.05 1.75 32 . 113 0.20 5.00
(continued)
261
262

Line x Limit Line x Limit Line x Limit Line x Limit Line x Limit Line x Limit
17 . 18 0.05 1.75 37 . 40 0.17 1.75 56 . 59 0.13 3.50 74 . 75 0.04 1.75 94 . 96 0.09 1.75 32 . 114 0.06 1.75
18 . 19 0.05 1.75 30 . 38 0.05 1.75 55 . 59 0.22 1.75 76 . 77 0.15 1.75 80 . 97 0.09 1.75 27 . 115 0.07 1.75
19 . 20 0.12 1.75 39 . 40 0.06 1.75 59 . 60 0.15 1.75 69 . 77 0.10 1.75 80 . 98 0.11 1.75 114 . 115 0.01 1.75
15 . 19 0.04 1.75 40 . 41 0.05 1.75 59 . 61 0.15 1.75 75 . 77 0.20 1.75 80 . 99 0.21 2.00 68 . 116 0.00 5.00
20 . 21 0.08 1.75 40 . 42 0.18 1.75 60 . 61 0.01 5.00 77 . 78 0.01 1.75 92 . 100 0.30 1.75 12 . 117 0.14 1.75
21 . 22 0.10 1.75 41 . 42 0.14 1.75 60 . 62 0.06 1.75 78 . 79 0.02 1.75 94 . 100 0.06 1.75 75 . 118 0.05 1.75
22 . 23 0.16 1.75 43 . 44 0.25 1.75 61 . 62 0.04 1.75 77 . 80 0.05 10.00 95 . 96 0.05 1.75 76 . 118 0.05 1.75
23 . 24 0.05 1.75 34 . 43 0.17 1.75 63 . 59 0.04 5.00 79 . 80 0.07 1.75 96 . 97 0.09 1.75 12 . 117 0.14 1.75
75 . 118 0.05 1.75
76 . 118 0.05 1.75
References 263
• The model should be multi-period. The OTS should consider the variation pattern
of demand and then determine the optimal switching actions.
• The uncertainty of demand and renewable power generation should be taken into
account.
• The current formulation does not ensure the network connectivity. It only
tries to satisfy the nodal demand-supply constraint. The resultant system (after
switching) might contain some islands.
• The computation burden of the model should be improved to make it applicable
for large scale transmission networks.
• The AC power flow constraints should be used for getting closer to reality.
• The unit commitment constraints can be added to the formulation to consider the
on/off states of the units as the decision variables.
• Changing the transmission network topology might change the short circuit level
on each bus. This should be taken into account for protection issues.
• The current model only considers the intact condition of the network. The
contingencies should also be considered.
• The OTS flexibility can be combined with demand response and power flow
controller devices.
• The bus splitting can be regarded as a switching action.
References
1. G.A. Orfanos, P.S. Georgilakis, N.D. Hatziargyriou, Transmission expansion planning of

systems with increasing wind power integration. IEEE Trans. Power Syst. 28(2), 1355–1362
(2013)
2. H. Zhang, V. Vittal, G.T. Heydt, J. Quintero, A mixed-integer linear programming approach
for multi-stage security-constrained transmission expansion planning. IEEE Trans. Power Syst.
27(2), 1125–1133 (2012)
3. I. De J. Silva, M.J. Rider, R. Romero, C.A.F. Murari, Transmission network expansion planning
considering uncertainty in demand. IEEE Trans. Power Syst. 21(4), 1565–1573 (2006)
4. J.D. Finney, H.A. Othman, W.L. Rutz, Evaluating transmission congestion constraints in
system planning. IEEE Trans. Power Syst. 12(3), 1143–1150 (1997)
5. R.A. Jabr, Robust transmission network expansion planning with uncertain renewable
generation and loads. IEEE Trans. Power Syst. 28(4), 4558–4567 (2013)
6. S. Haffner, A. Monticelli, A. Garcia, J. Mantovani, R. Romero, Branch and bound algorithm
for transmission system expansion planning using a transportation model. IEE Proc. Gener.
Transm. Distrib. 147(3), 149–156 (2000)
7. A.K. Kazerooni, J. Mutale, Transmission network planning under security and environmental
constraints. IEEE Trans. Power Syst. 25(2), 1169–1178 (2010)
8. P. Maghouli, S.H. Hosseini, M. Oloomi Buygi, M. Shahidehpour, A scenario-based multi-
objective model for multi-stage transmission expansion planning. IEEE Trans. Power Syst.
26(1), 470–478 (2011)
9. J.H. Roh, M. Shahidehpour, L. Wu, Market-based generation and transmission planning with
uncertainties. IEEE Trans. Power Syst. 24(3), 1587–1598 (2009)
10. M.J. Rider, A.V. Garcia, R. Romero, Power system transmission network expansion planning
using AC model. IET Gener. Transm. Distrib. 1(5), 731–742 (2007)
11. P. Maghouli, S.H. Hosseini, M.O. Buygi, M. Shahidehpour, A multi-objective framework for
transmission expansion planning in deregulated environments. IEEE Trans. Power Syst. 24(2),
1051–1061 (2009)
12. H. Yu, C.Y. Chung, K.P. Wong, J.H. Zhang, A chance constrained transmission network
expansion planning method with consideration of load and wind farm uncertainties. IEEE
Trans. Power Syst. 24(3), 1568–1576 (2009)
13. M.R. Bussieck, A. Pruessner, Mixed-integer nonlinear programming. SIAG/OPT Newsl.
Views News 14(1), 19–22 (2003)
14. R. Romero, A. Monticelli, A. Garcia, S. Haffner, Test systems and mathematical models for
transmission network expansion planning. IEE Proc. Gener. Transm. Distrib. 149(1), 27–36
(2002)
15. L. Baringo, A.J. Conejo, Transmission and wind power investment. IEEE Trans. Power Syst.
27(2), 885–893 (2012)
16. A.J. Wood, B.F. Wollenberg, Power Generation Operation and Control, 2nd edn. (Wiley, New
York, 1996)
17. A.J. Wood, B.F. Wollenberg, Power Generation, Operation, and Control (Wiley, New York,
2012)
18. E.B. Fisher, R.P. O’Neill, M.C. Ferris, Optimal transmission switching. IEEE Trans. Power
Syst. 23(3), 1346–1355 (2008)
19. K.W. Hedman, R.P. O’Neill, E.B. Fisher, S.S. Oren, Optimal transmission switching with
contingency analysis. IEEE Trans. Power Syst. 24(3), 1577–1586 (2009)
20. K.W. Hedman, M.C. Ferris, R.P. O’Neill, E.B. Fisher, S.S. Oren, Co-optimization of generation
unit commitment and transmission switching with n-1 reliability. IEEE Trans. Power Syst.
25(2), 1052–1063 (2010)
21. T. Ding, C. Zhao, Robust optimal transmission switching with the consideration of corrective
actions for n-k contingencies. IET Gener. Transm. Distrib. 10(13), 3288–3295 (2016)
22. Z. Yang, H. Zhong, Q. Xia, C. Kang, Optimal transmission switching with short-circuit current
limitation constraints. IEEE Trans. Power Syst. 31(2), 1278–1288 (2016)
23. P. Henneaux, D.S. Kirschen, Probabilistic security analysis of optimal transmission switching.
24. F. Qiu, J. Wang, Chance-constrained transmission switching with guaranteed wind power
utilization. IEEE Trans. Power Syst. 30(3), 1270–1278 (2015)
25. M. Soroush, J.D. Fuller, Accuracies of optimal transmission switching heuristics based on
dcopf and acopf. IEEE Trans. Power Syst. 29(2), 924–932 (2014)
Chapter 10
Energy System Integration
This chapter provides a solution for energy system integration (ESI) problem
in GAMS. The ESI analysis refers to a class of studies which investigate the
potential in different energy sectors (water, gas, and electricity) for moving toward
a more environmentally friendly and efficient energy supply. The main idea is how
to harvest the flexibilities in each energy carrier in a larger framework. In this
chapter, the coordination between water desalination systems and power system,
gas network-power network, and finally the concept of energy hub is investigated.
10.1 Water-Power Nexus
The water-electricity interdependence is an undeniable issue. Water is used for

cooling the power plants. On the other hand, the electricity is used for collecting,
treatment, and disposal of water.
The water-energy nexus concept is shown in Fig. 10.1 [1]. The optimization is
formulated as follows:
min OF DTC C CC C WC (10.1a)

DV
˚
DV D Pg;t ; Pc;t ; Wc;t ; Ww;t (10.1b)
X
CT D ag P2g;t C bg Pg;t C cg Ug;t (10.1c)
g;t
X
CC D ˛c P2c;t C ˇc Pc;t Wc;t C c Wc;t
2
C
c Pc;t C &c Wc;t C c Uc;t
c;t
(10.1d)
X
2
CW D aw Ww;t C bw Ww;t C cw Uw;t (10.1e)
w;t

DOI 10.1007/978-3-319-62350-4_10
266 10 Energy System Integration
Fig. 10.1 Water-energy nexus concept
Table 10.1 Technical and economical characteristics of thermal units

Unit ag .$=MW2 / bg ($=MW) cg ($) Pmax
g (MW) Pmin
g (MW)
g1 0.0002069 0.1483 57.11 500 0
g2 0.0003232 0.1854 57.11 400 0
g3 0.001065 0.6026 126.8 400 0
g4 0.0004222 0.2119 57.11 350 0
Pmin max
g Ug;t Pg;t Pg Ug;t (10.1f)
Pmin max
c Uc;t Pc;t Pc Uc;t (10.1g)
Wcmin Uc;t Wc;t Wcmax Uc;t (10.1h)
Wwmin Uw;t Ww;t Wwmax Uw;t (10.1i)
Pc;t
Rmin
c Rmax
c (10.1j)
Wc;t
X X
Pg;t C Pc;t D PLt (10.1k)
g c
X X
Ww;t C Wc;t D WLt (10.1l)
w c
The simulation data are taken from [1] with slight modifications (Tables 10.1, 10.2,
and 10.3).
10.1 Water-Power Nexus 267
Table 10.2 Technical and economical characteristics of co-production units

Unit Pmax
c Pmin
c Wcmax Wcmin Rmin
c Rmax
c ˛c ˇc c
c &c c
c1 800 160 200 30 4 9 0.0004433 0.003546 0.007093 1.106 4.426 737.4
c2 600 120 150 23 4 9 0.0007881 0.006305 0.01261 1.475 5.901 737.4
c3 400 80 100 15 4 9 0.001773 0.01419 0.02837 2.213 8.851 737.4
Table 10.3 Technical and Unit aw bw cw Wwmax Wwmin

economical characteristics of
water desalination units w1 0.00182 -0.708 7.374 250 0
GCode 10.1 Water-energy nexus optimization problem

Sets t / t 1t 2 4 / , i / p1p4 / , c / c1c3 / , w / w1 / ;
a b c Pmax Pmin
p1 0 . 0 0 0 2 0 6 9 0.1483 5 7 . 1 1 500 0
p2 0 . 0 0 0 3 2 3 2 0.1854 5 7 . 1 1 400 0
p3 0 . 0 0 1 0 6 5 0.6026 1 2 6 . 8 400 0
p4 0 . 0 0 0 4 2 2 2 0.2119 5 7 . 1 1 350 0;
Table Coproduct ( c ,)
Pmax Pmin Wmax Wmin r m i n rmax A11 A12 A22 b1 b2 C
c1 800 160 200 30 4 9 0 . 0 0 0 4 4 3 3 0 . 0 0 3 5 4 6 0 . 0 0 7 0 9 3 1.106 4.426 7 3 7 . 4
c2 600 120 150 23 4 9 0 . 0 0 0 7 8 8 1 0 . 0 0 6 3 0 5 0 . 0 1 2 6 1 1.475 5.901 7 3 7 . 4
c3 400 80 100 15 4 9 0 . 0 0 1 7 7 3 0 . 0 1 4 1 9 0 . 0 2 8 3 7 2.213 8.851
737.4;
T a b l e w a t e r d a t a (w, )
a b c Wmax Wmin
w1 1 . 8 2 E02 7.081 e1 7 . 3 7 4 250 0;
T a b l e PWdata ( t , )
Pd water
t1 1250 150
t2 1125 130
t3 875 100
t4 750 150
t5 950 200
t6 1440 350
t7 1500 300
t8 1750 200
t9 2000 300
t 1 0 2250 400
t 1 6 2500 550
t 1 7 2125 550
t 1 8 2375 500
t 1 9 2250 400
t 2 0 1975 350
t 2 1 1750 300
t 2 2 1625 250
t 2 3 1500 200
t 2 4 1376 1 5 0 ;
V a r i a b l e s Of , p ( i , t ) ,TC , CC , Pc ( c , t ) ,Wc( c , t ) , Water (w, t ) , W a t e r C o s t ;
B i n a r y v a r i a b l e s Up ( i , t ) , Uc ( c , t ) ,Uw(w, t ) ;
p . up ( i , t ) = g e n d a t a ( i , " Pmax " ) ; p . l o ( i , t ) = 0 ; Pc . up ( c , t ) = C o p r o d u c t ( c , ’ Pmax ’ ) ;
Pc . l o ( c , t ) = 0 ; Wc . up ( c , t ) = C o p r o d u c t ( c , ’Wmax ’ ) ; Wc . l o ( c , t ) = 0 ;
Water . up (w, t ) = w a t e r d a t a (w, ’Wmax ’ ) ; Water . l o (w, t ) = 0 ;
E q u a t i o n s c o s t T h e r m a l , b a l a n c e P , balanceW , c o s t C o p r o d c a l c , O b j e c t i v e ,
c o s t w a t e r c a l c , r a t i o 1 , r a t i o 2 , EQ1 , EQ2 , EQ3 , EQ4 , EQ5 , EQ6 , EQ7 , EQ8 ;
c o s t T h e r m a l . . TC=e=sum ( ( t , i ) , g e n d a t a ( i , ’ a ’ )power ( p ( i , t ) , 2 ) + g e n d a t a ( i , ’ b ’ )p ( i , t )
+ g e n d a t a ( i , ’ c ’ )Up ( i , t ) ) ;
b a l a n c e P ( t ) . . sum ( i , p ( i , t ) ) +sum ( c , Pc ( c , t ) ) =e=PWdata ( t , ’ Pd ’ ) ;
balanceW ( t ) . . sum (w, Water (w, t ) ) +sum ( c , Wc( c , t ) ) =e=PWdata ( t , ’ w a t e r ’ ) ;
c o s t C o p r o d c a l c . . CC=e=sum ( ( c , t ) , C o p r o d u c t ( c , ’ A11 ’ )power ( Pc ( c , t ) , 2 )
+2 C o p r o d u c t ( c , ’ A12 ’ )Pc ( c , t )Wc( c , t ) + C o p r o d u c t ( c , ’ A22 ’ )power (Wc( c , t ) , 2 )
+ C o p r o d u c t ( c , ’B1 ’ )Pc ( c , t ) + C o p r o d u c t ( c , ’ B2 ’ )Wc( c , t ) + C o p r o d u c t ( c , ’C ’ )Uc ( c , t ) ) ;
c o s t w a t e r c a l c . . W a t e r C o s t =e=sum ( ( t , w) , w a t e r d a t a (w, ’ a ’ )power ( Water (w, t ) , 2 )
+ w a t e r d a t a (w, ’ b ’ )Water (w, t ) + w a t e r d a t a (w, ’ c ’ )Uw(w, t ) ) ;
O b j e c t i v e . . OF=e=TC+CC+ W a t e r C o s t ;
r a t i o 1 ( c , t ) . . Pc ( c , t ) = l =Wc( c , t )C o p r o d u c t ( c , ’Rmax ’ ) ;
r a t i o 2 ( c , t ) . . Pc ( c , t ) =g=Wc( c , t )C o p r o d u c t ( c , ’ Rmin ’ ) ;
eq1 (w, t ) . . Water (w, t ) = l =Uw(w, t )w a t e r d a t a (w, ’Wmax ’ ) ;
eq2 (w, t ) . . Water (w, t ) =g=Uw(w, t )w a t e r d a t a (w, ’Wmin ’ ) ;
eq3 ( c , t ) . . wc ( c , t ) = l = Uc ( c , t )C o p r o d u c t ( c , ’Wmax ’ ) ;
eq4 ( c , t ) . . wc ( c , t ) =g= Uc ( c , t )C o p r o d u c t ( c , ’Wmin ’ ) ;
eq5 ( c , t ) . . Pc ( c , t ) = l = Uc ( c , t )C o p r o d u c t ( c , ’ Pmax ’ ) ;
eq6 ( c , t ) . . Pc ( c , t ) =g= Uc ( c , t )C o p r o d u c t ( c , ’ Pmin ’ ) ;
eq7 ( i , t ) . . p ( i , t ) = l =Up ( i , t )g e n d a t a ( i , " Pmax " ) ;
eq8 ( i , t ) . . p ( i , t ) =g=Up ( i , t )g e n d a t a ( i , " Pmin " ) ;
Model DEDcostbased / a l l / ; S o l v e DEDcostbased u s MInlp min OF ;
The hourly water output of different plants in water-energy nexus problem is

depicted in Fig. 10.2. The hourly power output of different plants in water-energy
nexus problem is shown in Fig. 10.3.
Fig. 10.2 Hourly water output of different plants in water-energy nexus problem
10.2 Gas-Power Nexus 269
Fig. 10.3 Hourly power output of different plants in water-energy nexus problem
Fig. 10.4 Gas network linkage with electricity network
10.2 Gas-Power Nexus
The interaction of gas network and electricity network is modeled in this section.
The electrical network is IEEE RTS 24-bus network which is shown in Fig. 10.4.
It is a transmission network with the voltage levels of 138 kV, 230 kV and
Sbase = 100 MVA. The branch data for IEEE RTS 24-bus network is given in
Table 10.4 [2]. The from bus, to bus, reactance (X), resistance (r), total line charging
susceptance (b), and MVA rating (MVA) are specified in this table. The parallel
lines in MATPOWER are merged and the resultants are given in Table 10.4. The
generation data for IEEE RTS 24-bus network is given in Table 10.5. The data of
generating units in this network is inspired by Conejo et al. [3] and Bouffard et al.
[4] with some modifications. The slack bus is bus 13 in this network. The wind
turbines and the capacities are also shown in Fig. 10.4.

Rating Rating
1 2 0.0026 0.0139 0.4611 175 11 13 0.0061 0.0476 0.0999 500
1 3 0.0546 0.2112 0.0572 175 11 14 0.0054 0.0418 0.0879 500
1 5 0.0218 0.0845 0.0229 175 12 13 0.0061 0.0476 0.0999 500
2 4 0.0328 0.1267 0.0343 175 12 23 0.0124 0.0966 0.2030 500
2 6 0.0497 0.1920 0.0520 175 13 23 0.0111 0.0865 0.1818 500
3 9 0.0308 0.1190 0.0322 175 14 16 0.0050 0.0389 0.0818 500
3 24 0.0023 0.0839 0.0000 400 15 16 0.0022 0.0173 0.0364 500
4 9 0.0268 0.1037 0.0281 175 15 21 0.0032 0.0245 0.2060 1000
5 10 0.0228 0.0883 0.0239 175 15 24 0.0067 0.0519 0.1091 500
6 10 0.0139 0.0605 2.4590 175 16 17 0.0033 0.0259 0.0545 500
7 8 0.0159 0.0614 0.0166 175 16 19 0.0030 0.0231 0.0485 500
8 9 0.0427 0.1651 0.0447 175 17 18 0.0018 0.0144 0.0303 500
8 10 0.0427 0.1651 0.0447 175 17 22 0.0135 0.1053 0.2212 500
9 11 0.0023 0.0839 0.0000 400 18 21 0.0017 0.0130 0.1090 1000
9 12 0.0023 0.0839 0.0000 400 19 20 0.0026 0.0198 0.1666 1000
10 11 0.0023 0.0839 0.0000 400 20 23 0.0014 0.0108 0.0910 1000
10 12 0.0023 0.0839 0.0000 400 21 22 0.0087 0.0678 0.1424 500
The gas network is also shown in Fig. 10.4 which its data is taken from [5]. The
technical and economical characteristics of gas nodes are given in Table 10.6. The
technical characteristics of gas network are also provided in Table 10.7 [5]. The gas
network equations are described in (10.2).
X
GC D cn Sgn;t (10.2a)
n;t
X X
fn;m;t D fm;n;t C Sgn;t
g;t Sdn Sen;t (10.2b)
m m
q
fm;n;t D Cm;n Pr2m;t Pr2n;t Passive arcs (10.2c)
q
fm;n;t Cmn Pr2m;t Pr2n;t Active arcs (10.2d)
Sgmin
n Sgn;t Sgmax
n (10.2e)
Prmin
n Prn;t Prmax
n (10.2f)
10.2 Gas-Power Nexus

Gen Bus Pmax
i Pmin
i bi ($/MW) Csi ($) Cdi ($) RUi (MW h1 ) RDi (MW h1 ) SUi (MW h1 ) SDi (MW h1 ) UTi (h) DTi (h) ui;tD0 Ui0 (h) Si0 (h)
g1 18 400 100 5.47 0 0 47 47 105 108 1 1 1 5 0
g2 21 400 100 5.47 0 0 47 47 106 112 1 1 1 6 0
g3 1 152 30.4 13.32 1430.4 1430.4 14 14 43 45 8 4 1 2 0
g4 2 152 30.4 13.32 1430.4 1430.4 14 14 44 57 8 4 1 2 0
g5 15 155 54.25 16 0 0 21 21 65 77 8 8 0 0 2
g6 16 155 54.25 10.52 312 312 21 21 66 73 8 8 1 10 0
g7 23 310 108.5 10.52 624 624 21 21 112 125 8 8 1 10 0
g8 23 350 140 10.89 2298 2298 28 28 154 162 8 8 1 5 0
g9 7 350 75 20.7 1725 1725 49 49 77 80 8 8 0 0 2
g10 13 591 206.85 20.93 3056.7 3056.7 21 21 213 228 12 10 0 0 8
g11 15 60 12 26.11 437 437 7 7 19 31 4 2 0 0 1
g12 22 300 0 0 0 0 35 35 315 326 0 0 1 2 0
271
Table 10.6 Technical and economical characteristics of gas nodes

Gas node Sgmin
n (106 Scm) Sgmax
n (106 Scm) Sdn (106 Scm) Prmin
n (bar) Prmax
n (bar) cn $/MBtu
Anderlues 0 1.20 0.00 0.00 66.20 0.00
Antwerpen 0 0.00 4.03 1.25 80.00 0.00
Arlon 0 0.00 0.22 0.00 66.20 0.00
Berneau 0 0.00 0.00 0.00 66.20 0.00
Blaregnies 0 0.00 15.62 2.08 66.20 0.00
Brugge 0 0.00 3.92 1.25 80.00 0.00
Dudzele 0 8.40 0.00 0.00 77.00 2.28
Gent 0 0.00 5.26 1.25 80.00 0.00
Liege 0 0.00 6.39 1.25 66.20 0.00
Loenhout 0 4.80 0.00 0.00 77.00 2.28
Mons 0 0.00 6.85 0.00 66.20 0.00
Namur 0 0.00 2.12 0.00 66.20 0.00
Petange 0 0.00 1.92 1.04 66.20 0.00
Peronnes 0 0.96 0.00 0.00 66.20 1.68
Sinsin 0 0.00 0.00 0.00 63.00 0.00
Voeren 0 22.01 0.00 2.08 66.20 1.68
Wanze 0 0.00 0.00 0.00 66.20 0.00
Warnand 0 0.00 0.00 0.00 66.20 0.00
Zeebrugge 0 11.59 0.00 0.00 77.00 2.28
Zomergem 0 0.00 0.00 0.00 80.00 0.00
The electrical network equations are described in (10.3).

X X
EC D ag .Pg;t /2 C bg Pg;t C cg C VOLL LSi;t C VWC Pwc
i;t (10.3a)
g;t i;t
X X
Pg;t C LSi;t C Pwi;t Li;t D Pij;t W i;t (10.3b)
g2˝Gi j2˝ì
ıi;t ıj;t
Pij;t D (10.3c)
Xij
Pmax
ij Pij;t Pmax
ij (10.3d)
Pmin
g Pg;t Pmax
g (10.3e)
0 LSi;t Li;t (10.3h)
Table 10.7 Technical Pipe From To Active 2

Cm;n
characteristics of gas network
L1 Zeebrugge Dudzele 9.07027
L2 Zeebrugge Dudzele 9.07027
L3 Dudzele Brugge 6.04685
L4 Dudzele Brugge 6.04685
L5 Brugge Zomergem 1.39543
L6 Loenhout Antwerpen 0.10025
L7 Antwerpen Gent 0.14865
L8 Gent Zomergem 0.22689
L9 Zomergem Peronnes 0.65965
L10 Voeren Berneau 1 7.25622
L11 Voeren Berneau 1 0.10803
L12 Berneau Liege 1.81405
L13 Berneau Liege 0.02701
L14 Liege Warnand 1.45124
L15 Liege Warnand 0.02161
L16 Warnand Namur 0.86384
L17 Namur Anderlues 0.90703
L18 Anderlues Peronnes 7.25622
L19 Peronnes Mons 3.62811
L20 Mons Blaregnies 1.45124
L21 Warnand Wanze 0.05144
L22 Wanze Sinsin 1 0.00642
L23 Sinsin Arlon 0.00170
L24 Arlon Petange 0.02782
Pwc w w
i;t D wt i Pi;t (10.3i)
The overall optimization problem, constraints, and the decision variables are as
follows:
minOF D EC C GC (10.4)
DV

ıi;t ; Pg;t ; Pc;t ; Wc;t ; Ww;t
DV D
Sgn;t ; fn;m;t ; Prn;t
.10.2 and .10.3/
The hourly variation pattern of wind generation, electric and gas demand is
shown in Fig. 10.5.
Fig. 10.5 Hourly variation pattern of wind generation, electric and gas demand (pu)
The integrated electricity-gas problem is solved using the GCode 10.2. The
total electricity cost EC is $3:9760 105 . The total gas extraction costs are
GC = $5:1755 105 and the total costs are $9:1515 105 . The hourly variation
pattern of wind generation, electric and gas demand is shown in Fig. 10.5.
GCode 10.2 Gas-electricity nexus optimization problem

Sets b u s / 1 2 4 / , s l a c k ( b u s ) / 1 3 / , Gen / g1g12 / , t / t 1t 2 4 /
genD ( gen ) / g1g2 , g5 , g7g11 / , genN ( gen ) / g3 , g4 , g6 , g12 / ;
s c a l a r s Sbase / 1 0 0 / ,VOLL / 1 0 0 0 0 / ,VOLW / 5 0 / ; a l i a s ( bus , node ) ;
S e t s gn NODES / A n d e r l u e s , Antwerpen , Arlon , Berneau , B l a r e g n i e s , Brugge , Dudzele ,
Gent , L i e g e , Loenhout , Mons , Namur , P e t a n g e , P e r o n n e s , S i n s i n ,
Voeren , Wanze , Warnand , Z e e b r u g g e , Zomergem /
a PIPES / L1L24 / ; A l i a s ( gn , gm ) ; s e t Pnm ( a , gn , gm ) a r c d e s c r i p t i o n ;
t a b l e N d a t a ( gn , ) Node D a t a
slo sup Sd p l o pup c
s e t G E l i n k ( gn , gen )
/ Loenhout . g12
Voeren . g6
Sinsin . g3
Petange . g4 / ;
t a b l e AData ( a , gn , gm, ) Arc D a t a
a c t C2mn ;
t a b l e GD( Gen , ) G e n e r a t i n g u n i t s c h a r a c t e r i s t i c s ;
s e t GB( bus , Gen ) c o n n e c t i v i t y i n d e x o f e a c h g e n e r a t i n g u n i t t o e a c h b u s ;
T a b l e BusData ( bus , ) Demands o f e a c h b u s i n MW;
T a b l e b r a n c h ( bus , node , ) Network t e c h n i c a l
r x b z limit ;
T a b l e DataWDL ( t , )
w d g;
P a r a m e t e r s Wcap ( b u s ) , c o n e x ( bus , node ) ,SD ( gn ) ;
b r a n c h ( bus , node , ’ b i j ’ ) $ b r a n c h ( bus , node , ’ L i m i t ’ ) = 1 / b r a n c h ( bus , node , ’ x ’ ) ;
c o n e x ( bus , node ) $ ( b r a n c h ( bus , node , ’ l i m i t ’ ) and b r a n c h ( node , bus , ’ l i m i t ’ ) ) = 1 ;
c o n e x ( bus , node ) $ ( c o n e x ( node , b u s ) ) = 1 ; V a r i a b l e s f ( a , gn , gm , t ) , s g ( gn , t ) , p r e s s u r e ( gn , t )
,
EC , P i j ( bus , node , t ) , Pg ( Gen , t ) , d e l t a ( bus , t ) , l s h ( bus , t ) ,Pw ( bus , t ) , pc ( bus , t ) , Gc , OF ;
Pnm ( a , gn , gm ) $ a d a t a ( a , gn , gm , ’ c2mn ’ ) = y e s ;
E q u a t i o n s c o n s t 1 , c o n s t 2 , c o n s t 3 , c o n s t 4 , c o n s t 5 , c o n s t 6 , CG1 , CG2 , CG3 , CG4 , O b j e c t i v e ;
b r a n c h ( bus , node , ’ b i j ’ ) ( d e l t a ( bus , t )d e l t a ( node , t ) ) ;

c o n s t 2 ( bus , t ) . . l s h ( bus , t ) $BusData ( bus , ’ pd ’ ) +Pw ( bus , t ) $Wcap ( b u s ) +sum ( Gen$GB ( bus , Gen )
,
Pg ( Gen , t ) )DataWDL ( t , ’ d ’ )BusData ( bus , ’ pd ’ ) / S b a s e =e=
+sum ( n o d e $ c o n e x ( node , b u s ) , P i j ( bus , node , t ) ) ;
c o n s t 3 . . EC=e=sum ( ( bus , GenD , t ) $GB ( bus , GenD ) , Pg ( GenD , t )GD( GenD , ’ b ’ )S b a s e )
+sum ( ( bus , t ) ,VOLLl s h ( bus , t )S b a s e $ B u s D a t a ( bus , ’ pd ’ ) +VOLWPc ( bus , t )sbase$Wcap ( b u s )
);
c o n s t 6 ( bus , t ) $Wcap ( b u s ) . . pc ( bus , t ) =e=DataWDL ( t , ’w ’ )Wcap ( b u s ) / Sbasepw ( bus , t ) ;
Pg . l o ( Gen , t ) =GD( Gen , ’ Pmin ’ ) / S b a s e ; Pg . up ( Gen , t ) =GD( Gen , ’ Pmax ’ ) / S b a s e ;
P i j . up ( bus , node , t ) $ ( ( c o n e x ( bus , node ) ) ) =1 b r a n c h ( bus , node , ’ L i m i t ’ ) / S b a s e ;
P i j . l o ( bus , node , t ) $ ( ( c o n e x ( bus , node ) ) )=1b r a n c h ( bus , node , ’ L i m i t ’ ) / S b a s e ;
l s h . up ( bus , t ) = DataWDL ( t , ’ d ’ )BusData ( bus , ’ pd ’ ) / S b a s e ; l s h . l o ( bus , t ) = 0 ;
Pw . up ( bus , t ) =DataWDL ( t , ’w ’ )Wcap ( b u s ) / S b a s e ; Pw . l o ( bus , t ) = 0 ;
Pc . up ( bus , t ) =DataWDL ( t , ’w ’ )Wcap ( b u s ) / S b a s e ; Pc . l o ( bus , t ) = 0 ; SD ( gn ) = N d a t a ( gn , ’SD ’ ) ;
CG1 ( gn , t ) . . sum ( Pnm ( a , gn , gm ) , f ( Pnm , t ) ) =e=sum ( Pnm ( a , gm , gn ) , f ( Pnm , t ) )
+ s g ( gn , t ) $ ( N d a t a ( gn , ’ Sup ’ ) >0)DataWDL ( t , ’G ’ )SD ( gn )
sum ( ( GenN ) $ G e l i n k ( gn , GenN ) , Pg ( GenN , t )GD( GenN , ’ b ’ )S b a s e / 3 5 3 1 5 ) ;
CG2 ( Pnm ( a , gn , gm ) , t ) $ ( AData ( a , gn , gm , ’C2mn ’ ) AND AData ( a , gn , gm , ’ACT ’ ) = 0 )
. . s i g n p o w e r ( f ( Pnm , t ) , 2 ) =e= AData ( Pnm , ’C2mn ’ ) ( p r e s s u r e ( gn , t )p r e s s u r e ( gm , t ) ) ;
CG3 ( Pnm ( a , gn , gm ) , t ) $ ( AData ( a , gn , gm , ’C2mn ’ ) AND AData ( a , gn , gm , ’ACT ’ ) = 1 )
. . s q r ( f ( Pnm , t ) ) =g= AData ( Pnm , ’C2mn ’ ) ( p r e s s u r e ( gn , t )p r e s s u r e ( gm , t ) ) ;
CG4 . . Gc =e= sum ( ( gn , t ) , 35315 n d a t a ( gn , ’ c ’ )s g ( gn , t ) $ N d a t a ( gn , ’ Sup ’ ) ) ;
O b j e c t i v e . . OF=e=EC+Gc ; s g . l o ( gn , t ) = 0 ; s g . up ( gn , t ) = n d a t a ( gn , ’ s u p ’ ) ;
p r e s s u r e . l o ( gn , t ) = s q r ( n d a t a ( gn , ’ p l o ’ ) ) ; p r e s s u r e . up ( gn , t ) = s q r ( n d a t a ( gn , ’ pup ’ ) ) ;
f . l o ( Pnm ( a , gn , gm ) , t ) $ ( AData ( a , gn , gm , ’C2mn ’ ) ) =
s q r t ( AData ( a , gn , gm , ’C2mn ’ ) ( p r e s s u r e . up ( gn , t )p r e s s u r e . l o ( gn , t ) ) ) ;
f . up ( Pnm ( a , gn , gm ) , t ) $ ( AData ( a , gn , gm , ’C2mn ’ ) ) =
s q r t ( AData ( a , gn , gm , ’C2mn ’ ) ( p r e s s u r e . up ( gn , t )p r e s s u r e . l o ( gn , t ) ) ) ;
f . l o ( Pnm ( a , gn , gm ) , t ) $ ( AData ( a , gn , gm , ’C2mn ’ ) AND AData ( a , gn , gm , ’ACT ’ ) = 1 ) = 0 ;
Model o v e r a l l / a l l / ; S o l v e o v e r a l l u s i n g n l p min OF ;
Fig. 10.6 Hourly variation pattern of gas generation from gas sources
Hourly variation pattern of gas generation from gas sources are shown in
Fig. 10.6.
The hourly variation pattern of electricity power generation is shown in Fig. 10.7.
400
g1
350
g2
g3
300
g4
g5
250
Pg,t (MW)
g6
200 g7
g8
150 g9
g 10
100 g 11
g 12
50
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 10.7 Hourly variation pattern of electricity power generation
The interaction of gas network and electricity sector has been investigated in
several works. The impacts of the gas network on security constrained UC is
analyzed in [6]. A unified gas and power flow analysis in natural gas and electricity
coupled networks can be found in [7]. A robust scheduling model for wind-
integrated energy systems with the considerations of both gas pipeline and power
transmission contingencies is developed in [8]. The reliability of gas networks and
their impacts on the reliability of electricity network is modeled in [9]. The impact
of large penetration of wind generation on the UK gas network is analyzed in [10].
One of the recent efficient methods of electricity storage is storing the electricity as
gas. This method is also called power to gas or P2G technique [11].
10.3 Energy Hub Concept
The concept of Energy Hub was introduced in [12]. Energy hub may be considered
as a virtual box that can convert a set of input energy carriers into a set of energy
demands. This box contains several technologies that can store, transfer, or convert
different forms of energies to each other. A general example of Energy hub is
shown in Fig. 10.8. Different aspects of energy hubs are investigated in the literature
such as economic dispatch of energy hubs [13], demand response and energy hub
[14], energy hub concept applied on car manufacturing plants [15], and wind power
uncertainty modeling in energy hubs [16, 17].
The technologies shown in Fig. 10.8 are explained as follows:
• Combined heat and power (CHP): receives the natural gas (Gt ) and converts it
into heat (Ht ) and electricity (Et ) The CHP economic dispatch problem can be
modeled in (10.5) [18]:
Ht D chp
ge Gt (10.5a)
10.3 Energy Hub Concept 277
Fig. 10.8 The energy hub concept
chp
Et D gh Gt (10.5b)
• Electric heat pump (EHP): It is fed by electricity and generates heat demand
(Ht ) or cool demand (Ct ) based on the operating mode. The operation of EHP is
mathematically formulated as follows:
Ct C Ht D Et COP (10.6a)
Htmin Ith Ht Htmax Ith (10.6b)
Ctmin Itc Ct Ctmax Itc (10.6c)
Itc C Ith 1 (10.6d)
Itc ; Ith 2 f0; 1g
The EHP can be in heat or cool generation mode. COP is the coefficient of
performance for EHP.
• Chiller boiler (CB): It receives heat and transforms it into cool demand The
chiller boiler operation is mathematically formulated as follows:
Ct D hc Ht (10.7)
The hc is the efficiency of heat to cooling conversion for chiller boiler.
• Electricity storage system (ESS): It can store (electricity) and then discharge
electricity The ESS operation is mathematically formulated as follows:
SOCt D SOCt1 C .Etch c Etdch =d /t (10.8a)

ch
Emin Etch Emax
dch
(10.8b)
dch
Emin Etdch Emax
dch
(10.8c)
SOCmin SOCt SOCmax (10.8d)
Itdch C Itch 1 (10.8e)
Itch ; Itdch 2 f0; 1g

ch=dch
SOCt is the state of charge in ESS. Et is for demonstrating the charged and
discharged electricity in ESS. The binary variables Itch ; Itdch show the charge or
discharge mode of ESS at time t.
• Transformer (Tr): It receives electricity and the output is also electricity (with
different voltage level)
Etout D ee Etin (10.9)
• Furnace (F): receives the natural gas and generates the heat demand
Ht D gh Gt (10.10)
Three different energy hub configurations will be analyzed to investigate the level
of achievable operational flexibility.
10.3.1 Data
The energy hubs which are analyzed in this chapter would have three types of energy
demands namely electric, heat, and cooling demand. Different hourly demand and
electricity price data for three energy hub configurations are given in Table 10.8.
• The charging and discharging efficiencies (ch=dch ) of ESS are assumed to be 0.9.
The ESS capacity is SOCmax D 600 MW h and SOCmin D 120 MW h. The initial
stored energy in ESS is 120 MW h. The minimum charging and discharging
ch=dch
limits are Emin D 0 and the maximum charging and discharging limits are
ch=dch
Emax D 120 MW.
• The transformer efficiency is ee D 0:98.
• The CHP efficiencies for gas to electricity is ge D 0:35 and for gas to heat is
gh D 0:45. The CHP capacity is 250 MW.
Table 10.8 Different hourly Time Dht (MW) Det (MW) Dct (MW) et $/MW h
demand and electricity price
data for energy hub t1 21.41 52.10 11.51 36.67
configurations t2 23.21 66.70 13.68 40.41
t3 26.09 72.20 16.01 38.48
t4 26.72 78.37 21.42 38.00
t5 25.59 120.20 21.97 40.24
t6 26.45 83.48 30.80 38.55
t7 39.54 110.40 38.94 52.26
t8 47.28 124.29 46.78 67.34
t9 52.12 143.61 50.97 70.47
t10 49.13 149.28 48.86 66.20
t11 69.26 154.19 34.77 73.30
t12 61.97 147.30 32.68 60.82
t13 68.04 200.71 27.77 63.15
t14 68.56 174.37 32.02 70.77
t15 56.40 176.54 33.22 63.09
t16 41.32 136.11 34.13 52.53
t17 37.43 108.71 40.78 57.00
t18 25.44 96.90 43.56 49.15
t19 25.66 89.08 51.48 47.47
t20 21.94 82.49 43.15 49.46
t21 22.44 76.93 36.49 53.07
t22 24.63 66.85 27.68 51.60
t23 22.72 47.17 19.14 50.53
t24 22.59 64.67 11.04 36.38
ehp ehp
• For EHP, CMax D HMax D 500 MW; The COP is assumed to be 2.5.
f
• The furnace efficiency gh is 0.9. The furnace capacity is 600 MW.
• The chiller boiler has a capacity equal to 500 MW and the efficiency is hc D 0:95
• The natural gas price is assumed to be constant for different hours and it is equal
g
to t D 12$/MW h
10.3.2 Configuration I
This configuration contains transformer, furnace, and chiller boiler as shown in

Fig. 10.9.
X g
min OF D et Et C t Gt (10.11a)
t
ee Et D Det (10.11b)

Fig. 10.9 The energy hub configuration I, considering transformer, furnace, and chiller boiler
f
gh Gt D H1;t C H2;t (10.11c)
hc H2;t D Dct (10.11d)
H1;t D Dht (10.11e)
The GAMS code for solving the hub configuration I described in (10.11) is given in
GCode 10.3.
The developed model in GCode 10.3 is linear and can be solved using any lp
solver. The problem is solved and the total operating costs are $1:1327 105 . The
hourly purchased electricity and natural gas in energy hub configuration I are shown
in Fig. 10.10.
The output of furnace system will be divided into two streams. The first
one will supply the chiller and the second one will directly supply the heat
demand. The hourly output of furnace unit in energy hub configuration-I is shown
in Fig. 10.11.
GCode 10.3 The optimal operation of energy hub configuration I
Set t hours / t1t24 /

Table data ( t ,)
Dh De Dc lambda
t1 21.4 52.1 11.5 36.7
t2 23.2 66.7 13.7 40.4
t3 2 6 . 1 7 2 . 2 16 38.5
t4 2 6 . 7 7 8 . 4 2 1 . 4 38
t5 2 5 . 6 1 2 0 . 2 22 40.2
t6 26.4 83.5 30.8 38.6
t7 39.5 110.4 38.9 52.3
t8 47.3 124.3 46.8 67.3
t9 5 2 . 1 1 4 3 . 6 51 70.5
t10 49.1 149.3 48.9 66.2
t11 69.3 154.2 34.8 73.3
t 1 2 62 147.3 32.7 60.8
t 1 3 68 200.7 27.8 63.2
t 1 4 6 8 . 6 1 7 4 . 4 32 70.8
t15 56.4 176.5 33.2 63.1
t16 41.3 136.1 34.1 52.5
t 1 7 3 7 . 4 1 0 8 . 7 4 0 . 8 57
t18 25.4 96.9 43.6 49.2
t19 25.7 89.1 51.5 47.5
t20 21.9 82.5 43.1 49.5
t21 22.4 76.9 36.5 53.1
t22 24.6 66.8 27.7 51.6
t23 22.7 47.2 19.1 50.5
t 2 4 2 2 . 6 6 4 . 7 11 36.4

Variable cost ;
Positive variables E ( t ) ,G( t ) , H1 ( t ) , H2 ( t ) ;
Scalar
CBmax / 5 0 0 / , e t a _ e e / 0 . 9 8 / , e t a _ g h f / 0 . 9 / , e t a _ h c / 0 . 9 5 / ;
H2 . up ( t ) =CBmax ;
Equations
eq1 , eq2 , eq3 , eq4 , eq5 ;
eq1 .. c o s t =e=sum ( t , d a t a ( t , ’ lambda ’ )E ( t )+12G( t ) ) ;
eq2 ( t ) .. e t a _ e e E ( t ) = e= d a t a ( t , ’E ’ ) ;
eq3 ( t ) .. H1 ( t ) = e= d a t a ( t , ’ h ’ ) ;
eq4 ( t ) .. e t a _ g h f G( t ) = e=H1 ( t ) +H2 ( t ) ;
eq5 ( t ) .. e t a _ h c H2 ( t ) = e= d a t a ( t , ’ c ’ ) ;
Model Hub / a l l / ;
S o l v e hub u s l p min c o s t ;
10.3.3 Configuration II
This configuration contains transformer, furnace, chiller boiler, CHP, and ESS as
shown in Fig. 10.12.
300
Power (MW) Et
200
100
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
150
Gt
Power (MW)
100
50
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 10.10 The hourly purchased electricity and natural gas in energy hub configuration I
80
H1,t
H2,t
70
60
Power (MW)
50
40
30
20
10
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 10.11 The hourly output of furnace unit in energy hub configuration I
Fig. 10.12 The energy hub configuration II, considering transformer, furnace, chiller boiler, CHP,
and electric energy storage
X g
t
ee E2;t C Etdch C ge G1;t D Det (10.12b)

Et D E1;t C E2;t (10.12c)
E1;t D Etch (10.12d)
SOCt D SOCt1 C .Etch c Etdch =d /t (10.12e)

ch
Emin Etch Emax
dch
(10.12f)
dch
Emin Etdch Emax
dch
(10.12g)
Itdch C Itch 1 (10.12i)
Gt D G1;t C G2;t (10.12j)

f
gh G1;t C H1;t D Dht (10.12k)
gh G2;t D H1;t C H2;t (10.12l)
hc H2;t D Dct (10.12m)
The GAMS code for solving the hub configuration II described in (10.12) is given
in GCode 10.4.
GCode 10.4 The optimal operation of energy hub configuration II

set t hours / t 1t 2 4 /
t a b l e data ( t ,)
Dh De Dc lambda
t1 21.4 52.1 11.5 36.7
t2 23.2 66.7 13.7 40.4
t3 2 6 . 1 7 2 . 2 16 38.5
t4 2 6 . 7 7 8 . 4 2 1 . 4 38
t5 2 5 . 6 1 2 0 . 2 22 40.2
t6 26.4 83.5 30.8 38.6
t7 39.5 110.4 38.9 52.3
t8 47.3 124.3 46.8 67.3
t9 5 2 . 1 1 4 3 . 6 51 70.5
t10 49.1 149.3 48.9 66.2
t11 69.3 154.2 34.8 73.3
t 1 2 62 147.3 32.7 60.8
t 1 3 68 200.7 27.8 63.2
t 1 4 6 8 . 6 1 7 4 . 4 32 70.8
t15 56.4 176.5 33.2 63.1
t16 41.3 136.1 34.1 52.5
t 1 7 3 7 . 4 1 0 8 . 7 4 0 . 8 57
t18 25.4 96.9 43.6 49.2
t19 25.7 89.1 51.5 47.5
t20 21.9 82.5 43.1 49.5
t21 22.4 76.9 36.5 53.1
t22 24.6 66.8 27.7 51.6
t23 22.7 47.2 19.1 50.5
t 2 4 2 2 . 6 6 4 . 7 11 36.4
Variable cost ;
Positive variables
E ( t ) , E1 ( t ) , E2 ( t ) ,G( t ) , G1 ( t ) , G2 ( t ) , H1 ( t ) , H2 ( t )
SOC( t ) , Ec ( t ) , Ed ( t ) ;
Binary v a r i a b l e s Idch ( t ) , Ich ( t ) ;
s c a l a r SOC0 / 2 0 / , SOCmax / 6 0 0 / , e t a _ c / 0 . 9 / , e t a _ d / 0 . 9 / , e t a _ e e / 0 . 9 8 / , e t a _ g e / 0 . 4 5 / ,
e t a _ g h / 0 . 3 5 / , e t a _ h c / 0 . 9 5 / , Chpmax / 2 5 0 / , CBmax / 5 0 0 / , Fmax / 6 0 0 / , e t a _ g h f / 0 . 9 / ;
SOC0= 0.2SOCmax ;
SOC . up ( t ) =SOCmax ; SOC . l o ( t ) =0.2SOCmax ; SOC . f x ( ’ t 2 4 ’ ) =SOC0 ;
Ec . up ( t ) =0.2SOCmax ; Ec . l o ( t ) = 0 ;
Ed . up ( t ) =0.2SOCmax ; Ed . l o ( t ) = 0 ;
G1 . up ( t ) =Chpmax ;
G2 . up ( t ) =Fmax ;
H2 . up ( t ) =CBmax ;
E q u a t i o n s eq1 , eq2 , eq3 , eq4 , eq5 , eq6 , eq7 , eq8 , eq9 , eq10 , eq11 , eq12 ;
eq1 .. c o s t =e=sum ( t , d a t a ( t , ’ lambda ’ )E ( t ) +12G( t ) ) ;
eq2 ( t ) .. e t a _ e eE2 ( t ) +Ed ( t ) + e t a _ g eG1 ( t ) =e= d a t a ( t , ’E ’ ) ;
eq3 ( t ) .. E ( t ) =e=E1 ( t ) +E2 ( t ) ;
eq4 ( t ) .. E1 ( t ) =e=Ec ( t ) ;
eq5 ( t ) .. SOC( t ) =e=SOC0$ ( o r d ( t ) = 1 ) + SOC( t 1)$ ( o r d ( t ) >1)+Ec ( t )e t a _ cEd ( t ) / e t a _ d
;
eq6 ( t ) .. Ed ( t ) = l =0.2SOCmaxI d c h ( t ) ;
eq7 ( t ) .. Ec ( t ) = l =0.2SOCmaxI c h ( t ) ;
eq8 ( t ) .. Idch ( t ) + Ich ( t ) = l =1;
eq9 ( t ) .. G( t ) =e=G1 ( t ) +G2 ( t ) ;

eq10 ( t ) .. e t a _ g hG1 ( t ) +H1 ( t ) =e=1 d a t a ( t , ’ h ’ ) ;
eq11 ( t ) .. e t a _ g h fG2 ( t ) =e=H1 ( t ) +H2 ( t ) ;
eq12 ( t ) .. e t a _ h cH2 ( t ) =e= d a t a ( t , ’ c ’ ) ;
model Hub2 / a l l / ;
S o l v e hub2 u s mip min c o s t ;
solver. The problem is solved, and the total operating costs are $0:85504 105 .
The hourly purchased electricity and its division between transformer and ESS in
energy hub configuration-II are shown in Fig. 10.13. The output of furnace system
will be divided into two streams. The first one will supply the chiller and the second
one will supply the heat demand. The hourly output of furnace unit in energy hub
configuration II is shown in Fig. 10.14. The hourly state of charge, charging, and
discharging of ESS in energy hub configuration II is shown in Fig. 10.15.
10.3.4 Configuration III
This configuration contains transformer, furnace, chiller boiler, CHP, and ESS as
shown in Fig. 10.16.
200
Et
Power (MW)
150
E 1,t
E 2,t
100
50
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
300
Gt
Power (MW)
200 G1,t
G2,t
100
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 10.13 The hourly purchased electricity and its division between transformer and ESS in
energy hub configuration II
60
H1,t
H2,t
50
40
Power (MW)
30
20
10
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 10.14 The hourly output of furnace unit in energy hub configuration II
600 SOC
t
500 Edch
t
Power (MW)
400 Ech
t
300
200
100
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 10.15 The hourly state of charge (MW h), charging, and discharging of ESS (MW) in energy
hub configuration II
X g
t
ee E2;t C Etdch C ge G1;t D Det (10.13b)

Et D E1;t C E2;t C E3;t (10.13c)
E1;t D Etch (10.13d)

Fig. 10.16 The energy hub configuration III, considering transformer, furnace, chiller boiler, CHP,
electric energy storage, and EHP
SOCt D SOCt1 C .Etch c Etdch =d /t (10.13e)

ch ch
Emin It Etch Emax
ch ch
It (10.13f)
dch dch
Emin It Etdch Emax
dch dch
It (10.13g)
Itdch C Itch 1 (10.13i)
Gt D G1;t C G2;t (10.13j)
gh G1;t C H1;t C HtEHP D Dht (10.13k)

f
gh G2;t D H1;t C H2;t (10.13l)
hc H2;t C CtEHP D Dct (10.13m)
CtEHP C HtEHP D E3;t COP (10.13n)
Htmin Ith HtEHP Htmax Ith (10.13o)
Ctmin Itc CtEHP Ctmax Itc (10.13p)

Itc C Ith 1 (10.13q)
Itc ; Ith 2 f0; 1g
solver.
GCode 10.5 The optimal operation of energy hub configuration III

Set t hours / t 1t 2 4 /
Table data ( t ,)

Dh De Dc lambda
t1 21.4 52.1 11.5 36.7
t2 23.2 66.7 13.7 40.4
t3 2 6 . 1 7 2 . 2 16 38.5
t4 26.7 78.4 21.4 38
t5 2 5 . 6 1 2 0 . 2 22 40.2
t6 26.4 83.5 30.8 38.6
t7 39.5 110.4 38.9 52.3
t8 47.3 124.3 46.8 67.3
t9 5 2 . 1 1 4 3 . 6 51 70.5
t10 49.1 149.3 48.9 66.2
t11 69.3 154.2 34.8 73.3
t 1 2 62 147.3 32.7 60.8
t 1 3 68 200.7 27.8 63.2
t 1 4 6 8 . 6 1 7 4 . 4 32 70.8
t15 56.4 176.5 33.2 63.1
t16 41.3 136.1 34.1 52.5
t17 37.4 108.7 40.8 57
t18 25.4 96.9 43.6 49.2
t19 25.7 89.1 51.5 47.5
t20 21.9 82.5 43.1 49.5
t21 22.4 76.9 36.5 53.1
t22 24.6 66.8 27.7 51.6
t23 22.7 47.2 19.1 50.5
t 2 4 2 2 . 6 6 4 . 7 11 36.4 ;
d a t a ( t , ’ lambda ’ ) =0.6 d a t a ( t , ’ lambda ’ ) ;

variable cost ;
p o s i t i v e v a r i a b l e s E ( t ) , E1 ( t ) , E2 ( t ) , E3 ( t ) ,G( t ) , G1 ( t ) , G2 ( t ) , H1 ( t ) , H2 ( t )
SOC( t ) , Ec ( t ) , Ed ( t ) , H_ehp ( t ) , C_ehp ( t ) ;
Binary v a r i a b l e s Idch ( t ) , Ich ( t ) , Ic ( t ) , Ih ( t ) ;
s c a l a r SOC0 / 2 0 / , SOCmax / 6 0 0 / , e t a _ c / 0 . 9 / , e t a _ d / 0 . 9 / , e t a _ e e / 0 . 9 8 /
, eta_ge / 0 . 4 5 / , eta_gh / 0 . 3 5 /
e t a _ h c / 0 . 9 5 / , COP / 2 . 5 / , H_ehpMax / 2 0 0 / , C_ehpMax / 2 0 0 / , Chpmax / 3 0 0 / ,
CBmax / 3 0 0 / , Fmax / 3 0 0 / , e t a _ g h f / 0 . 9 / ;
SOC0= 0.2SOCmax ; SOC . up ( t ) =SOCmax ; SOC . l o ( t ) =0.2SOCmax ; SOC . f x ( ’ t 2 4 ’ ) =SOC0 ;
Ec . up ( t ) =0.2SOCmax ; Ec . l o ( t ) = 0 ; Ed . up ( t ) =0.2SOCmax ; Ed . l o ( t ) = 0 ;
C_ehp . up ( t ) =C_ehpMax ; H_ehp . up ( t ) =H_ehpMax ;
G1 . up ( t ) =Chpmax ; G2 . up ( t ) =Fmax ; H2 . up ( t ) =CBmax ; E . up ( t ) = 1 0 0 0 ;
Equations
eq1 , eq2 , eq3 , eq4 , eq5 , eq6 , eq7 , eq8 , eq9 , eq10 , eq11 , eq12 , eq13 , eq14 , eq15 , eq16 ;
eq1 .. c o s t =e=sum ( t , d a t a ( t , ’ lambda ’ )E ( t ) +12G( t ) ) ;

eq2 ( t ) .. e t a _ e eE2 ( t ) +Ed ( t ) + e t a _ g eG1 ( t ) =e= d a t a ( t , ’E ’ ) +E3 ( t ) ;
eq3 ( t ) .. E ( t ) =e=E1 ( t ) +E2 ( t ) ;
eq4 ( t ) .. E1 ( t ) =e=Ec ( t ) ;
eq5 ( t ) .. SOC( t ) =e=SOC0$ ( o r d ( t ) = 1 ) + SOC( t 1)$ ( o r d ( t ) >1)+Ec ( t )e t a _ cEd ( t ) / e t a _ d
;
eq6 ( t ) .. Ed ( t ) = l =0.2SOCmaxI d c h ( t ) ;
eq7 ( t ) .. Ec ( t ) = l =0.2SOCmaxI c h ( t ) ;
eq8 ( t ) .. Idch ( t ) + Ich ( t ) = l =1;
eq9 ( t ) .. G( t ) =e=G1 ( t ) +G2 ( t ) ;
eq10 ( t ) .. e t a _ g hG1 ( t ) +H1 ( t ) +H_ehp ( t ) =e=1 d a t a ( t , ’ h ’ ) ;
eq11 ( t ) .. e t a _ g h fG2 ( t ) =e=H1 ( t ) +H2 ( t ) ;
eq12 ( t ) .. e t a _ h cH2 ( t ) +C_ehp ( t ) =e= d a t a ( t , ’ c ’ ) ;
eq13 ( t ) .. C_ehp ( t ) +H_ehp ( t ) =e=E3 ( t )cop ;
eq14 ( t ) .. H_ehp ( t ) = l =H_ehpMax I h ( t ) ;
eq15 ( t ) .. C_ehp ( t ) = l =C_ehpMax I c ( t ) ;
eq16 ( t ) .. Ic ( t ) +Ih ( t ) = l =1;
Model Hub / a l l / ;
S o l v e hub u s mip min c o s t ;
200
Et
Power (MW)
150
E 1,t
E 2,t
100
50
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
300
Gt
Power (MW)
200 G1,t
G2,t
100
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 10.17 The hourly purchased electricity and its division between transformer and ESS in
energy hub configuration III
The problem is solved and the total operating costs are $0:84430 105 . The
hourly purchased electricity and its division between transformer and ESS in energy
hub configuration-III are shown in Figs. 10.17, 10.18. The output of furnace system
will be divided into two streams. The first one will supply the chiller and the second
one will supply the heat demand. The hourly output of furnace unit in energy hub
configuration II is shown in Fig. 10.14. The hourly state of charge, charging, and
discharging of ESS in energy hub configuration III is shown in Fig. 10.19. The
hourly output of EHP in energy hub configuration III is shown in Fig. 10.20.
60
H1,t
H2,t
50
40
Power (MW)
30
20
10
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 10.18 The hourly output of furnace unit in energy hub configuration III
600 SOCt
500 Edch
t
Power (MW)
400 Ech
t
300
200
100
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 10.19 The hourly state of charge (MW h), charging, and discharging of ESS (MW) in energy
hub configuration III
References 291
55
50 HEHP
t
CEHP
t
45
40
35
Power (MW)
30
25
20
15
10
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (h)
Fig. 10.20 The hourly output of EHP in energy hub configuration III
References
1. A. Santhosh, A.M. Farid, K. Youcef-Toumi, Real-time economic dispatch for the supply side
of the energy-water nexus. Appl. Energy 122, 42–52 (2014)
Syst. 26(1), 12–19 (2011)
Markets, vol. 1 (Springer, Berlin, 2010)
5. D. De Wolf, Y. Smeers, The gas transmission problem solved by an extension of the simplex
algorithm. Manag. Sci. 46(11), 1454–1465 (2000)
6. T. Li, M. Eremia, M. Shahidehpour, Interdependency of natural gas network and power system
security. IEEE Trans. Power Syst. 23(4), 1817–1824 (2008)
7. A. Martinez-Mares, C.R. Fuerte-Esquivel, A unified gas and power flow analysis in natural gas
and electricity coupled networks. IEEE Trans. Power Syst. 27(4), 2156–2166 (2012)
8. L. Bai, F. Li, T. Jiang, H. Jia, Robust scheduling for wind integrated energy systems considering
gas pipeline and power transmission n-1 contingencies. IEEE Trans. Power Syst. 32(2), 1582–
1584 (2017)
9. J. Munoz, N. Jimenez-Redondo, J. Perez-Ruiz, J. Barquin, Natural gas network modeling for
power systems reliability studies. in 2003 IEEE Bologna Power Tech Conference Proceedings,
vol. 4, June 2003, p. 8
10. M. Qadrdan, M. Chaudry, J. Wu, N. Jenkins, J. Ekanayake, Impact of a large penetration of
wind generation on the GB gas network. Energy Policy 38(10), 5684–5695 (2010)
11. H.S. de Boer, L. Grond, H. Moll, R. Benders, The application of power-to-gas, pumped hydro
storage and compressed air energy storage in an electricity system at different wind power
penetration levels. Energy 72, 360–370 (2014)
12. M. Geidl, G. Koeppel, P. Favre-Perrod, B. Klöckl, G. Andersson, K. Fröhlich, The energy hub–
a powerful concept for future energy systems. in Third Annual Carnegie Mellon Conference
on the Electricity Industry, Pittsburgh (2007), pp. 13–14
13. S.D. Beigvand, H. Abdi, M. La Scala, A general model for energy hub economic dispatch.
Appl. Energy 190, 1090–1111 (2017)
14. M. Batić, N. Tomašević, G. Beccuti, T. Demiray, S. Vraneš, Combined energy hub optimisation
and demand side management for buildings. Energy Build. 127, 229–241 (2016)
15. K. Kampouropoulos, F. Andrade, Energy hub optimization applied on car manufacturing
plants. in ANDESCON, 2016 IEEE (IEEE, Piscataway, 2016), pp. 1–4
16. A. Soroudi, B. Mohammadi-Ivatloo, A. Rabiee, Energy hub management with intermittent
wind power. in Large Scale Renewable Power Generation (Springer, Singapore, 2014),
pp. 413–438
17. A. Dolatabadi, B. Mohammadi-Ivatloo, M. Abapour, S. Tohidi, Optimal stochas-
tic design of wind integrated energy hub. IEEE Trans. Ind. Inf. 99, 1–1 (2017).
doi:10.1109/TII.2017.2664101
18. M. Geidl, Integrated modeling and optimization of multi-carrier energy systems. PhD thesis,
TU Graz, 2007
Index
A Dynamic economic dispatch (DED), 3, 28, 30,

Allocation, 2, 3, 6, 28, 41–43, 62, 195–200, 87, 95–118, 176–179, 200
204 –211, 213, 214, 217–220, 222, 224,
225
E
Economic dispatch, 3, 4, 6, 28, 65–67, 70,
B 73–91, 95–118, 176–200, 276
Battery, 175 Electric heat pump (EHP), 277, 278, 289, 291
Branch flow, 154–155, 163 Emission, 66, 69, 71, 92, 95–97, 100–105, 109,
117, 118, 138, 178
Energy hub, 276–291
C Energy storage system (ESS), 2, 28, 175–200,
Cascaded reservoir, 79, 80 278, 281, 285, 286, 289, 290
Charging, 3, 150, 173, 184, 269, 278, 285, 286, Environmental pollutions, 67, 69–71
289, 290 Excel, 18–19
Chiller boiler, 277–281, 283, 285, 287
Combined heat and power (CHP), 28, 65,
73–79, 276, 278, 281, 283, 285, 287
Conditional statement, 15, 16 F
Congestion costs, 145, 147, 154, 156, 157, Furnace, 278–283, 285–287, 289, 290
228, 254 Fuzzy, 3, 4, 38, 102, 138
Contingency, 128, 129, 131–133, 156–158,
160, 246
Cost based DED, 95–107, 109, 113–117 G
Cost based unit commitment, 119–126, 130, Gas-power nexus, 269–276
133–135, 138 Generation shift factor (GSF), 238–246, 249
Generator outage, 128, 157–162
D
Debugging, 19–24, 43 H
DED. See dynamic economic dispatch (DED) Heat, 73, 76, 77, 92, 276–278, 280, 285, 289
Demand flexibility, 131–136 Heat demand, 77, 277, 278, 280, 285, 289
Discharging, 3, 278, 285, 286, 289, 290 Heuristic optimization, 5
Display, 8, 11, 21–23, 26, 36 Hydro Unit, 79–86

DOI 10.1007/978-3-319-62350-4
294 Index
I Optimal DC power flow, 3, 141–162

Information gap decision theory (IGDT), 2, 3, Optimal power flow (OPF), 1, 3, 4, 28, 62,
138 141–173, 200, 257
Installation, 6–7, 221, 222, 224 Optimal transmission switching (OTS), 3, 249,
Interface, 7, 18, 26, 27 254, 256–258, 263
Investment cost, 196, 228, 229, 231, 233,
235–238
P
Pareto front, 58–60
L Pareto optimal, 58–61, 71, 72, 73, 76, 102–105
Library, 26 Phase shifter, 28, 62, 233–237
Linearization, 113 Phasor measurement unit (PMU), 2, 28, 62,
Linear programming (LP), 10–12, 16, 17, 203–225
21–23, 28, 33–38, 56–57, 62, 143, 257 Placement, 2, 6, 39, 40, 45–47, 62, 203–225
Line outage distribution factor (LODF), 238, Power flow controller, 28, 227, 233–238, 263
246–249 Price based DED (PBDED), 107–114
LMP. See Locational marginal price (LMP) Price based unit commitment, 136–138
Load shedding, 157–163, 168, 169, 173, 175,
229, 238
Locational marginal price (LMP), 145–147,
Q
150, 154, 156, 157, 160–163, 168, 173,
Quadratic constrained programming (QCP),
238, 258
11, 23, 48–49, 62
Loop, 16–18, 71, 73
Loss of PMU, 217–218
R
M Ramp rate, 98–101, 120–122, 163
Marginal, 14, 35–37, 43, 45, 48, 49, 66, 67, Random numbers, 53–57
145, 161, 168, 173, 238 Reserve, 3, 91, 110–114, 126–129, 131, 132
Mixed integer non-linear programming Resistance, 150, 161, 168, 172, 184, 269
(MINLP), 11, 27, 49–52, 62 Robust optimization, 3, 4, 138, 139, 227
Mixed integer programming (MIP), 11, 27, 28,
38–44, 62, 119, 224, 227, 254, 257
Multi-area economic dispatch, 86, 87, 90, 91 S
Multistage PMU placement, 221–224 Sensitivity analysis, 17, 28, 67–69, 75, 98–101,
135
Single line outage, 218–221
N Solve statement, 8, 10, 20, 21, 23, 45, 55, 71,
Natural gas, 276, 278–280, 282 145, 242
Non-linear programming (NLP), 11, 23, Spinning reserve, 110, 114, 118, 126
44–52, 62 State of charge, 195, 278, 285, 286, 289, 290
Stochastic, 91, 117, 138, 139
Susceptance, 150, 173, 184, 233, 238, 269
O
Observability, 2, 28, 203–225
Operating costs, 3, 28, 76, 79, 81, 87, 95, 97, T
100, 107, 109, 117, 125, 126, 129, 133, Thermal unit, 65–73, 76, 77, 81, 83–88, 95–99,
136, 142, 146, 150, 153, 154, 157, 161, 101, 107, 108, 121, 125, 126, 128, 129,
163, 168, 176, 179, 183, 185, 191, 195, 131–134, 138, 139, 143, 161, 163, 165,
196, 199, 229, 231, 233, 235, 236, 249, 170, 172, 173, 177–180, 183, 199, 237,
257, 258, 280, 285, 289 238, 266
OPF. See Optimal power flow (OPF) Transmission expansion planning (TEP), 1, 3,
Optimal AC power flow, 28, 169–173 6, 62, 227–238
Index 295
U Weighted sum, 58
Uncertainty, 2, 3, 138, 139, 176, 200, 227, 263, Wind curtailment, 118, 163, 168, 176, 199
276 Wind turbine, 28, 65, 88, 138, 163, 168, 172,
173, 269
W
Water, 79–82, 265–268 Z
Water-power Nexus, 265–269 Zero injection bus, 205–207, 209–211, 213

Soroudi 2017

Uploaded by

Copyright:

Available Formats

Soroudi 2017

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Soroudi 2017

Uploaded by

Copyright:

Available Formats

What topics are covered in the book?

What topics are covered in the book?

What optimization problems are discussed?

What optimization problems are discussed?

Alireza Soroudi

Power System Optimization

ISBN 978-3-319-62349-8 ISBN 978-3-319-62350-4 (eBook)

Library of Congress Control Number: 2017944921

© Springer International Publishing AG 2017

Printed on acid-free paper

This Springer imprint is published by Springer Nature

The complex structure of electric power systems as the largest human-built

• System operators: Power system operators usually use commercial packages

Dublin, Ireland Alireza Soroudi

1 Introduction to Programming in GAMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2.3 Multi-Objective Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

6 Multi-Period Optimal Power Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141

10 Energy System Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265

Fig. 1.1 Some optimization problems in power system studies . . . . . . . . . . . . 2

Fig. 3.4 Pareto optimal fronts for different load values . . . . . . . . . . . . . . . . . . . . 76

Fig. 6.7 IEEE RTS 24-bus network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151

Fig. 8.8 Optimal PMU allocation for IEEE 118-bus with

Fig. 10.3 Hourly power output of different plants in water-energy

Table 1.1 Data for illustrative example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

Table 5.5 Reserve provision (Ri;t ) by thermal units considering

Table 9.3 Line outage distribution factors (LODFij;nm ) for three-bus

1.1 Optimization Problems in Power System

© Springer International Publishing AG 2017 1

Fig. 1.1 Some optimization problems in power system studies

• FACTS device allocation [6]: This problem is a subset of transmission planning

– Monte Carlo simulation: FACTS allocation [17]

Fig. 1.2 Optimization methods in power system studies

– Monte Carlo simulation: reliability and risk analysis [44]

Fig. 1.3 The structure of heuristic optimization techniques

nonlinear problems with integer variables or non-convex constraints would be

• These methods are usually computationally expensive. In other words, it takes a

1.2 GAMS Installation

The GAMS installation package is available for the following platforms:

The GAMS interface is depicted in Fig. 1.4.

Fig. 1.4 The GAMS interface

1.3 GAMS Elements

Each GAMS model consists of the following main elements:

Fig. 1.5 General GAMS code structure and elements

• Variables: We are about to decide how many kg should be produced of each

Equations eq1, eq2;

As it can be seen in this example, there are two types of equations as

Model example /all/;

It can also be defined as follows:

Model example /eq1,eq2/;

Solve example us LP Max OF;

LP: linear programming

GCode 1.1 Illustrative GAMS example

Fig. 1.6 The structure of

SOLVER STATUS CODE DESCRIPTION

Different model status might be reported once the model is solved.

MODEL STATUS CODE DESCRIPTION

LOWER LEVEL UPPER MARGINAL

Variable type Description Lower limit Upper limit

1.4 Conditional Statements

eq1(m)$(ord(m) 2) .. sum(p, task(m,p)*x(p))=l=availability(M);

Sets M /m1m2/, P /p1p3/ ;

Sets M /m1m2/, P /p1p3/ ;