SMART GRID
IEEE Press
445 Hoes Lane
Piscataway, NJ 08854
IEEE Press Editorial Board
Lajos Hanzo, Editor in Chief
R. Abhari
J. Anderson
G. W. Arnold
F. Canavero
M. El-Hawary
B-M. Haemmerli
M. Lanzerotti
D. Jacobson
O. P. Malik
S. Nahavandi
T. Samad
G. Zobrist
Kenneth Moore, Director of IEEE Book and Information Services (BIS)
A complete list of titles in the IEEE Press Series on Power Engineering appears
at the end of this book.
SMART GRID
Fundamentals of
Design and Analysis
James Momoh
IEEE PRESS
A JOHN WILEY & SONS, INC., PUBLICATION
Copyright © 2012 by the Institute of Electrical and Electronics Engineers.
Published by John Wiley & Sons, Inc., Hoboken, New Jersey. All rights reserved.
Published simultaneously in Canada.
No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or
by any means, electronic, mechanical, photocopying, recording, scanning, or otherwise, except as
permitted under Section 107 or 108 of the 1976 United States Copyright Act, without either the prior
written permission of the Publisher, or authorization through payment of the appropriate per-copy fee to
the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, (978) 750-8400, fax
(978) 750-4470, or on the web at www.copyright.com. Requests to the Publisher for permission should be
addressed to the Permissions Department, John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030,
(201) 748-6011, fax (201) 748-6008, or online at http://www.wiley.com/go/permissions.
Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in
preparing this book, they make no representations or warranties with respect to the accuracy or
completeness of the contents of this book and specifically disclaim any implied warranties of
merchantability or fitness for a particular purpose. No warranty may be created or extended by sales
representatives or written sales materials. The advice and strategies contained herein may not be suitable
for your situation. You should consult with a professional where appropriate. Neither the publisher nor
author shall be liable for any loss of profit or any other commercial damages, including but not limited to
special, incidental, consequential, or other damages.
For general information on our other products and services or for technical support, please contact our
Customer Care Department within the United States at (800) 762-2974, outside the United States at (317)
572-3993 or fax (317) 572-4002.
Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may
not be available in electronic formats. For more information about Wiley products, visit our web site at
www.wiley.com.
Library of Congress Cataloging-in-Publication Data:
Momoh, James A., 1950Smart grid : fundamentals of design and analysis / James Momoh.
p. cm.
ISBN 978-0-470-88939-8 (hardback)
1. Electric power distribution–Automation. I. Title.
TK3226.M588 2012
333.793'2–dc23
2011024774
Printed in the United States of America.
10
9
8
7
6
5
4
3
2
1
CONTENTS
Preface
1
SMART GRID ARCHITECTURAL DESIGNS
1.1
Introduction
1.2
Today’s Grid versus the Smart Grid
1.3
Energy Independence and Security Act of 2007: Rationale
for the Smart Grid
1.4
Computational Intelligence
1.5
Power System Enhancement
1.6
Communication and Standards
1.7
Environment and Economics
1.8
Outline of the Book
1.9
General View of the Smart Grid Market Drivers
1.10 Stakeholder Roles and Function
1.10.1 Utilities
1.10.2 Government Laboratory Demonstration Activities
1.10.3 Power Systems Engineering Research Center (PSERC)
1.10.4 Research Institutes
1.10.5 Technology Companies, Vendors, and Manufacturers
1.11 Working Definition of the Smart Grid Based on Performance
Measures
1.12 Representative Architecture
1.13 Functions of Smart Grid Components
1.13.1 Smart Devices Interface Component
1.13.2 Storage Component
1.13.3 Transmission Subsystem Component
1.13.4 Monitoring and Control Technology Component
1.13.5 Intelligent Grid Distribution Subsystem Component
1.13.6 Demand Side Management Component
xiii
1
1
2
2
4
5
5
5
5
6
6
9
9
10
10
10
11
12
12
13
13
14
14
14
14
v
vi
CONTENTS
1.14 Summary
References
Suggested Readings
2
3
SMART GRID COMMUNICATIONS AND MEASUREMENT
TECHNOLOGY
2.1
Communication and Measurement
2.2
Monitoring, PMU, Smart Meters, and Measurements
Technologies
2.2.1
Wide Area Monitoring Systems (WAMS)
2.2.2
Phasor Measurement Units (PMU)
2.2.3
Smart Meters
2.2.4
Smart Appliances
2.2.5
Advanced Metering Infrastructure (AMI)
2.3
GIS and Google Mapping Tools
2.4
Multiagent Systems (MAS) Technology
2.4.1
Multiagent Systems for Smart Grid Implementation
2.4.2
Multiagent Specifications
2.4.3
Multiagent Technique
2.5
Microgrid and Smart Grid Comparison
2.6
Summary
References
PERFORMANCE ANALYSIS TOOLS FOR SMART GRID DESIGN
3.1
Introduction to Load Flow Studies
3.2
Challenges to Load Flow in Smart Grid and Weaknesses of the
Present Load Flow Methods
3.3
Load Flow State of the Art: Classical, Extended Formulations,
and Algorithms
3.3.1
Gauss–Seidal Method
3.3.2
Newton–Raphson Method
3.3.3
Fast Decouple Method
3.3.4
Distribution Load Flow Methods
3.4
Congestion Management Effect
3.5
Load Flow for Smart Grid Design
3.5.1
Cases for the Development of Stochastic Dynamic
Optimal Power Flow (DSOPF)
3.6
DSOPF Application to the Smart Grid
3.7
Static Security Assessment (SSA) and Contingencies
15
15
15
16
16
19
20
20
21
22
22
23
24
25
25
26
27
27
27
29
29
30
31
31
32
33
33
37
38
41
41
43
CONTENTS
4
vii
3.8
Contingencies and Their Classification
3.8.1
Steady-State Contingency Analysis
3.8.2
Performance Indices
3.8.3
Sensitivity-Based Approaches
3.9
Contingency Studies for the Smart Grid
3.10 Summary
References
Suggested Readings
44
46
47
48
48
49
50
50
STABILITY ANALYSIS TOOLS FOR SMART GRID
4.1
Introduction to Stability
4.2
Strengths and Weaknesses of Existing Voltage Stability Analysis
Tools
4.3
Voltage Stability Assessment
4.3.1
Voltage Stability and Voltage Collapse
4.3.2
Classification of Voltage Stability
4.3.3
Static Stability (Type I Instability)
4.3.4
Dynamic Stability (Type II Instability)
4.3.5
Analysis Techniques for Dynamic Voltage Stability
Studies
4.4
Voltage Stability Assessment Techniques
4.5
Voltage Stability Indexing
4.6
Analysis Techniques for Steady-State Voltage Stability Studies
4.6.1
Direct Methods for Detecting Voltage Collapse Points
4.6.2
Indirect Methods (Continuation Methods)
4.7
Application and Implementation Plan of Voltage Stability
4.8
Optimizing Stability Constraint through Preventive Control of
Voltage Stability
4.9
Angle Stability Assessment
4.9.1
Transient Stability
4.9.2
Stability Application to a Practical Power System
4.9.3
Boundary of the Region of Stability
4.9.4
Algorithm to Find the Controlling UEP
4.9.5
Process Changes in Design of DSA for the Smart Grid
4.10 State Estimation
4.10.1 Mathematical Formulations for Weighted Least
Square Estimation
4.10.2 Detection and Identification of Bad Data
4.10.3 Pre-Estimation Analysis
51
51
51
56
57
58
59
59
60
62
65
68
69
69
70
71
73
75
76
77
80
80
81
84
86
86
viii
CONTENTS
4.10.4 Postestimation Analysis
4.10.5 Robust State Estimation
4.10.6 SE for the Smart Grid Environment
4.10.7 Real-Time Network Modeling
4.10.8 Approach of the Smart Grid to State Estimation
4.10.9 Dynamic State Estimation
4.10.10 Summary
References
Suggested Readings
5
6
88
90
94
95
95
97
98
98
98
COMPUTATIONAL TOOLS FOR SMART GRID DESIGN
5.1
Introduction to Computational Tools
5.2
Decision Support Tools (DS)
5.2.1
Analytical Hierarchical Programming (AHP)
5.3
Optimization Techniques
5.4
Classical Optimization Method
5.4.1
Linear Programming
5.4.2
Nonlinear Programming
5.4.3
Integer Programming
5.4.4
Dynamic Programming
5.4.5
Stochastic Programming and Chance Constrained
Programming (CCP)
5.5
Heuristic Optimization
5.5.1
Artificial Neural Networks (ANN)
5.5.2
Expert Systems (ES)
5.6
Evolutionary Computational Techniques
5.6.1
Genetic Algorithm (GA)
5.6.2
Particle Swarm Optimization (PSO)
5.6.3
Ant Colony Optimization
5.7
Adaptive Dynamic Programming Techniques
5.8
Pareto Methods
5.9
Hybridizing Optimization Techniques and Applications to the
Smart Grid
5.10 Computational Challenges
5.11 Summary
References
100
100
101
102
103
103
103
105
106
107
PATHWAY FOR DESIGNING SMART GRID
6.1
Introduction to Smart Grid Pathway Design
6.2
Barriers and Solutions to Smart Grid Development
122
122
122
107
108
109
111
112
112
113
113
115
117
118
118
119
120
CONTENTS
Solution Pathways for Designing Smart Grid Using Advanced
Optimization and Control Techniques for Selection Functions
6.4
General Level Automation
6.4.1
Reliability
6.4.2
Stability
6.4.3
Economic Dispatch
6.4.4
Unit Commitment
6.4.5
Security Analysis
6.5
Bulk Power Systems Automation of the Smart Grid
at Transmission Level
6.5.1
Fault and Stability Diagnosis
6.5.2
Reactive Power Control
6.6
Distribution System Automation Requirement of the Power Grid
6.6.1
Voltage/VAr Control
6.6.2
Power Quality
6.6.3
Network Reconfiguration
6.6.4
Demand-Side Management
6.6.5
Distribution Generation Control
6.7
End User/Appliance Level of the Smart Grid
6.8
Applications for Adaptive Control and Optimization
6.9
Summary
References
Suggested Reading
ix
6.3
7
RENEWABLE ENERGY AND STORAGE
7.1
Renewable Energy Resources
7.2
Sustainable Energy Options for the Smart Grid
7.2.1
Solar Energy
7.2.2
Solar Power Technology
7.2.3
Modeling PV Systems
7.2.4
Wind Turbine Systems
7.2.5
Biomass-Bioenergy
7.2.6
Small and Micro Hydropower
7.2.7
Fuel Cell
7.2.8
Geothermal Heat Pumps
7.3
Penetration and Variability Issues Associated with Sustainable
Energy Technology
7.4
Demand Response Issues
7.5
Electric Vehicles and Plug-in Hybrids
125
125
125
127
127
128
130
130
131
132
132
132
135
136
136
137
137
137
138
138
139
140
140
141
141
142
142
144
145
147
147
148
148
150
151
x
CONTENTS
8
9
7.6
PHEV Technology
7.6.1
Impact of PHEV on the Grid
7.7
Environmental Implications
7.7.1
Climate Change
7.7.2
Implications of Climate Change
7.8
Storage Technologies
7.9
Tax Credits
7.10 Summary
References
Suggested Reading
151
151
152
153
153
154
158
159
159
159
INTEROPERABILITY, STANDARDS, AND CYBER SECURITY
8.1
Introduction
8.2
Interoperability
8.2.1
State-of-the-Art-Interoperability
8.2.2
Benefits and Challenges of Interoperability
8.2.3
Model for Interoperability in the Smart Grid
Environment
8.2.4
Smart Grid Network Interoperability
8.2.5
Interoperability and Control of the Power Grid
8.3
Standards
8.3.1
Approach to Smart Grid Interoperability Standards
8.4
Smart Grid Cyber Security
8.4.1
Cyber Security State of the Art
8.4.2
Cyber Security Risks
8.4.3
Cyber Security Concerns Associated with AMI
8.4.4
Mitigation Approach to Cyber Security Risks
8.5
Cyber Security and Possible Operation for Improving
Methodology for Other Users
8.6
Summary
References
Suggested Readings
160
160
161
161
161
RESEARCH, EDUCATION, AND TRAINING FOR THE SMART
GRID
9.1
Introduction
9.2
Research Areas for Smart Grid Development
9.3
Research Activities in the Smart Grid
162
162
163
163
163
166
166
169
171
171
173
174
174
174
176
176
176
178
CONTENTS
9.4
9.5
Multidisciplinary Research Activities
Smart Grid Education
9.5.1
Module 1: Introduction
9.5.2
Module 2: Architecture
9.5.3
Module 3: Functions
9.5.4
Module 4: Tools and Techniques
9.5.5
Module 5: Pathways to Design
9.5.6
Module 6: Renewable Energy Technologies
9.5.7
Module 7: Communication Technologies
9.5.8
Module 8: Standards, Interoperability, and Cyber
Security
9.5.9
Module 9: Case Studies and Testbeds
9.6
Training and Professional Development
9.7
Summary
References
10
11
CASE
10.1
10.2
10.3
10.4
10.5
10.6
xi
178
179
180
180
181
181
181
181
182
182
182
182
183
183
STUDIES AND TESTBEDS FOR THE SMART GRID
Introduction
Demonstration Projects
Advanced Metering
Microgrid with Renewable Energy
Power System Unit Commitment (UC) Problem
ADP for Optimal Network Reconfiguration in Distribution
Automation
10.7 Case Study of RER Integration
10.7.1 Description of Smart Grid Activity
10.7.2 Approach for Smart Grid Application
10.8 Testbeds and Benchmark Systems
10.9 Challenges of Smart Transmission
10.10 Benefits of Smart Transmission
10.11 Summary
References
184
184
184
185
185
186
EPILOGUE
200
Index
191
196
196
196
197
198
198
198
199
203
PREFACE
The term “smart grid” defines a self-healing network equipped with dynamic optimization techniques that use real-time measurements to minimize network losses, maintain
voltage levels, increase reliability, and improve asset management. The operational data
collected by the smart grid and its sub-systems will allow system operators to rapidly
identify the best strategy to secure against attacks, vulnerability, and so on, caused by
various contingencies. However, the smart grid first depends upon identifying and
researching key performance measures, designing and testing appropriate tools, and
developing the proper education curriculum to equip current and future personnel with
the knowledge and skills for deployment of this highly advanced system.
The objective of this book is to equip readers with a working knowledge of fundamentals, design tools, and current research, and the critical issues in the development
and deployment of the smart grid. The material presented in its eleven chapters is an
outgrowth of numerous lectures, conferences, research efforts, and academic and industry debate on how to modernize the grid both in the United States and worldwide. For
example, Chapter 3 discusses the optimization tools suited to managing the randomness, adaptive nature, and predictive concerns of an electric grid. The general purpose
Optimal Power Flow, which takes advantage of a learning algorithm and is capable of
solving the optimization scheme needed for the generation, transmission, distribution,
demand response, reconfiguration, and the automation functions based on real-time
measurements, is explained in detail.
I am grateful to several people for their help during the course of writing this book. I
acknowledge Keisha D’Arnaud, a dedicated recent graduate student at the Center for
Energy Systems and Control, for her perseverance and support in the several iterations
needed to design the text for a general audience. I thank David Owens, Senior Executive
Vice President of the Edison Electric Institute, and Dr. Paul Werbos, Program Director of
the Electrical, Communication and Cyber Systems (ECCS), National Science Foundation (NSF), for encouraging and supporting my interest in unifying my knowledge of
systems through computational intelligence to address complex power system problems
where traditional techniques have failed. Their support was especially valuable during
my stint at NSF as a Program Director in ECCS from 2001 to 2004. I am also grateful for
the Small Grant Expository Research award granted by the NSF to develop the first
xiii
xiv
PREFACE
generation of Dynamic Stochastic Optimal Power flow, a general purpose tool for use in
smart grid design and development.
I thank my family for their encouragement and support. I am grateful to my students and colleagues at the Center for Energy Systems and Control, who, as audience
and enthusiasts, let me test and refine my ideas in the smart grid, and also for honorary
invited presentations to top utility executive management in addressing the emergence
of the smart grid across the country. All these exposures rekindled my interest in the
design and development of the grid for the future.
James Momoh
1
SMART GRID
ARCHITECTURAL DESIGNS
1.1
INTRODUCTION
Today’s electric grid was designed to operate as a vertical structure consisting of generation, transmission, and distribution and supported with controls and devices to maintain
reliability, stability, and efficiency. However, system operators are now facing new challenges including the penetration of RER in the legacy system, rapid technological change,
and different types of market players and end users. The next iteration, the smart grid,
will be equipped with communication support schemes and real-time measurement techniques to enhance resiliency and forecasting as well as to protect against internal and
external threats. The design framework of the smart grid is based upon unbundling and
restructuring the power sector and optimizing its assets. The new grid will be capable of:
•
•
•
•
Handling uncertainties in schedules and power transfers across regions
Accommodating renewables
Optimizing the transfer capability of the transmission and distribution networks
and meeting the demand for increased quality and reliable supply
Managing and resolving unpredictable events and uncertainties in operations and
planning more aggressively.
Smart Grid: Fundamentals of Design and Analysis, First Edition. James Momoh.
© 2012 Institute of Electrical and Electronics Engineers. Published 2012 by John Wiley & Sons, Inc.
1
2
SMART GRID ARCHITECTURAL DESIGNS
TABLE 1.1. Comparison of Today’s Grid vs. Smart Grid [4]
Preferred Characteristics
Today’s Grid
Smart Grid
Active Consumer
Participation
Consumers are uninformed and
do not participate
Accommodation of all
generation and storage
options
Dominated by central
generation—many obstacles
exist for distributed energy
resources interconnection
Limited, poorly integrated
wholesale markets; limited
opportunities for consumers
Informed, involved consumers—
demand response and
distributed energy resources
Many distributed energy
resources with plug-and-play
convenience focus on
renewables
Mature, well-integrated
wholesale markets; growth of
new electricity markets for
consumers
Power quality a priority with a
variety of quality/price
options—rapid resolution of
issues
Greatly expanded data
acquisition of grid parameters;
focus on prevention,
minimizing impact to
consumers
Automatically detects and
responds to problems; focus on
prevention, minimizing impact
to consumers
Resilient to cyber attack and
natural disasters; rapid
restoration capabilities
New products, services,
and markets
Provision of power
quality for the digital
economy
Focus on outages—slow
response to power quality
issues
Optimization of assets
and operates efficiently
Little integration of operational
data with asset
management—business
process silos
Anticipating responses
to system disturbances
(self-healing)
Responds to prevent further
damage; focus on protecting
assets following a fault
Resiliency against cyber
attack and natural
disasters
Vulnerable to malicious acts of
terror and natural disasters;
slow response
1.2
TODAY’S GRID VERSUS THE SMART GRID
As mentioned, several factors contribute to the inability of today’s grid to efficiently
meet the demand for reliable power supply. Table 1.1 compares the characteristics of
today’s grid with the preferred characteristics of the smart grid.
1.3 ENERGY INDEPENDENCE AND SECURITY ACT OF 2007:
RATIONALE FOR THE SMART GRID
The Energy Independence and Security Act of 2007 (EISA) signed into law by President
George W. Bush vividly depicts a smart grid that can predict, adapt, and reconfigure
itself efficiently and reliably. The objective of the modernization of the U.S. grid as
outlined in the Act is to maintain a reliable and secure electricity [2] infrastructure that
ENERGY INDEPENDENCE AND SECURITY ACT OF 2007
3
Figure 1.1. Rationale for the smart grid.
will meet future demand growth. Figure 1.1 illustrates the features needed to facilitate
the development of an energy-efficient, reliable system.
The Act established a Smart Grid Task Force, whose mission is “to insure awareness, coordination and integration of the diverse activities of the DoE Office and elsewhere in the Federal Government related to smart-grid technologies and practices” [1].
The task force’s activities include research and development; development of widely
accepted standards and protocols; the relationship of smart grid technologies and
4
SMART GRID ARCHITECTURAL DESIGNS
practices to electric utility regulation; the relationship of smart grid technologies and
practices to infrastructure development, system reliability, and security; and the relationship of smart grid technologies and practices to other facets of electricity supply,
demand, transmission, distribution, and policy. In response to the legislation, the U.S.
research and education community is actively engaged in:
1.
2.
3.
4.
5.
Smart grid research and development program
Development of widely accepted smart grid standards and protection
Development of infrastructure to enable smart grid deployment
Certainty of system reliability and security
Policy and motivation to encourage smart grid technology support for generation, transmission and distribution
As Figure 1.2 shows, there are five key aspects of smart grid development and
deployment.
1.4
COMPUTATIONAL INTELLIGENCE
Computational intelligence is the term used to describe the advanced analytical tools
needed to optimize the bulk power network. The toolbox will include heuristic, evolution programming, decision support tools, and adaptive optimization techniques.
Power System
Enhancement
Communication
and Standards
Testbed
Smart Grid
Environment and
Economics
Computational
Intelligence
Figure 1.2. Five key aspects of smart grid development.
OUTLINE OF THE BOOK
1.5
5
POWER SYSTEM ENHANCEMENT
Policy-makers assume that greatly expanded use of renewable energy [4,5] resources
in the United States will help to offset the impacts of carbon emissions from thermal
and fossil energy, meet demand uncertainty, and to some extent, increase reliability of
delivery.
1.6
COMMUNICATION AND STANDARDS
Since planning horizons can be short as an hour ahead, the smart grid’s advanced
automations will generate vast amounts of operational data in a rapid decision-making
environment. New algorithms will help it become adaptive and capable of predicting
with foresight. In turn, new rules will be needed for managing, operating, and marketing
networks.
1.7
ENVIRONMENT AND ECONOMICS
Based on these desired features, an assessment of the differences in the characteristics
of the present power grid and the proposed smart grid is needed to highlight characteristics of the grid and the challenges. When fully developed the smart grid system will
allow customer involvement, enhance generation and transmission with tools to allow
minimization of system vulnerability, resiliency, reliability, adequacy and power quality.
The training tools and capacity development to manage and operate the grids and hence
crate new job opportunities is part of the desired goals of the smart grid evolution which
will be tested using test-bed. To achieve the rapid deployment of the grids test bed and
research centers need to work across disciplines to build the first generation of smart
grid.
By focusing on security controls rather than individual vulnerabilities and threats,
utility companies and smart-grid technology vendors can remediate the root causes that
lead to vulnerabilities. However, security controls are more difficult and sometimes
impossible to add to an existing system, and ideally should be integrated from the
beginning to minimize implementation issues. The operating effectiveness of the implemented security controls-base will be assessed routinely to protect the smart grid against
evolving threats.
1.8
OUTLINE OF THE BOOK
This book is organized into 10 chapters. Following this chapter ’s introduction, Chapter
2 presents the smart grid concept, fundamentals, working definitions, and system architecture. Chapter 3 describes the tools using load flow concepts, optimal power flows,
and contingencies and Chapter 4 describes those using voltage stability, angle stability,
and state estimation. Chapter 5 evaluates the computational intelligence approach as a
6
SMART GRID ARCHITECTURAL DESIGNS
feature of the smart grid. Chapter 6 explains the pathways design of the smart grid
using general purpose dynamic stochastic optimization. Chapter 7 reviews renewable
supply and the related issues of variability and probability distribution functions, followed by a discussion of storage technologies, capabilities, and configurations. Demand
side managemen (DSM) and demand response, climate change, and tax credits are
highlighted for the purpose of evaluating the economic and environmental benefit of
renewable energy sources. Chapter 8 discusses the importance of developing national
standards, followed by a discussion of interoperability such that the new technologies
can easily be adapted to the legacy system without violating operational constraints.
The chapter also discusses cyber security to protect both RER and communication
infrastructure. Chapter 9 explains the significant research and employment training for
attaining full performance and economic benefits of the new technology. Chapter 10
discusses case studies on smart grid development and testbeds to aid deployment. The
chapter outlines the grand challenges facing researchers and policy-makers before the
smart grid can be fully deployed, and calls for investment and multidisciplinary collaboration. Figure 1.3 is a schematic of the chapters.
1.9
GENERAL VIEW OF THE SMART GRID MARKET DRIVERS
To improve efficiency and reliability, several market drivers and new opportunities
suggest that the smart grid must:
1. Satisfy the need for increased integration of digital systems for increased efficiency of the power system. In the restructured environment, the deregulated
electric utility industry allows a renovation of the market to be based on system
constraints and the seasonal and daily fluctuations in demand. Competitive
markets increase the shipment of power between regions, which further strains
today’s aging grid and requires updated, real-time controls.
2. Handle grid congestion, increase customer participation, and reduce uncertainty
for investment. This requires the enhancement of the grid’s capability to handle
demand reliably.
3. Seamlessly integrate renewable energy systems (RES) and distributed generation. The drastic increase in the integration of cost-competitive distributed
generation technologies affects the power system.
In addition to system operators and policy-makers, stakeholders are contributing to the
development of the smart grid. Their specific contributions and conceptual understanding of the aspects to be undertaken are discussed below.
1.10
STAKEHOLDER ROLES AND FUNCTION
As in the legacy system, critical attention must be paid to the identification of the
stakeholders and how they function in the grid’s development. Stakeholders range from
7
STAKEHOLDER ROLES AND FUNCTION
Chapter 1
Chapter 2
Introduction to the Smart Grid
Working Definition Smart Grid Architecture
Smart Grid Functions GIS and Google mapping tools
Multi-agent System MAS Technology
Chapter 3
Performance Measures for Tool Development
Smart Grid Tools and Techniques
Load flow Concepts and New Approach to Smart Grid
Optimal Power Flow
Contingencies
Chapter 4
Voltage, Angle Stability and Estimation Application to Smart Grid
Chapter 5
Computational Techniques
Communication and Measurement
Monitoring PMU, Smart Meters, Measurements Technologies
Barriers and Solutions to Smart Grid Development
Pathways for Design using Advanced Optimization and Control
Generation level Automation
Bulk Power Systems Automation
Distribution System Automation
End-User/Appliance Level
Chapter 6
Development and Applications of DSOPF
Evaluation of Techniques for Smart Grid Design
Market and Pricing
Chapter 7
Renewable Energy Technologies
Storage Technologies
Demand Response
Electric Vehicles and Plug
Environmental Implications and Climate Change
Tax Credit and Incentives
Standards
Chapter 8
Interoperability
Cyber Security
Chapter 9
Chapter 10
Research Areas and Needs for Smart Grid Development
Education Needs for the Smart Grid Environment
Training and Professional Development
Case Studies and Test Beds for Smart Grid
Figure 1.3. Schematic of chapters.
utility and energy producers to consumers, policy-makers, technology providers, and
researchers. An important part of the realization of the smart grid is the complete buy-in
or involvement of all stakeholders.
Policy-makers are the federal and state regulators responsible for ensuring the
cohesiveness of policies for modernization efforts and mediating the needs of all parties.
The primary benefit of smart grid development to these stakeholders concerns the mitigation of energy prices, reduced dependence on foreign oil, increased efficiency, and
reliability of power supply. Figure 1.4 shows the categories of stakeholders.
Other participants in the development of the smart grid include government agencies, manufacturers, and research institutes. The federal Department of Energy’s (DOE)
8
SMART GRID ARCHITECTURAL DESIGNS
SMART
GRID:
Stakeholders
UTILITIES: Installation
and implementation of
power grid technologies
POLICY-MAKERS: Establishment of
standards for operation, monitoring,
interoperability etc.
TECHNOLOGY PROVIDERS: Development of
smart grid technologies for the grid
enhancement
RESEARCHERS: Development of tools and technologies for
the smart grid
CONSUMERS: Consumer input and participation, consumer buy-in etc.
Figure 1.4. Stakeholders and their functions.
National Renewable Energy Laboratory (NREL) and state agencies such as the California Energy Commission and the New York State Energy Research and Development
Authority are among the pioneers. In the monograph, “The Smart Grid: An Introduction,” the DOE discusses the nature, challenges, opportunities, and necessity for smart
grid implementation. It defines the smart grid as technology which “makes this transformation of the electric industry possible by bringing the philosophies, concepts and
technologies that enabled the internet to the utility and the electric grid and enables the
grid modernization” [1]. The characteristics of the smart grid are two-way digital communication, plug-and-play capabilities, advanced metering infrastructure for integrating
customers, facilities for increased customer involvement, interoperability based on
standards, and low-cost communication and electronics.
Additional features identified include integration and advancement of grid visualization technology to provide wide-area grid awareness, integrating real-time sensor
data, weather information, and grid modeling with geographical information [1].
However, the DOE’s definitions in our opinion do not provide measures for
addressing uncertainty, predictivity, and foresight. Another federal entity, the Federal
Energy Regulatory Commission (FERC), has mandated the development of:
1. Cyber Security: require NIST define standard and protocol consistent with the
overarching cyber security and reliability requirements of the Energy Independence and Security Act (EISA) and the FERC Reliability Standards.
STAKEHOLDER ROLES AND FUNCTION
9
2. Intersystem Communications: Identify standards for common information
models for communication among all elements of the bulk power system regional
market operators, utilities, demand response aggregators, and customers
3. Wide-Area Situational Awareness: Ensure that operators of the nation’s bulk
power system have the equipment that gives them a complete view of their
systems so they can monitor and operate their systems.
4. Coordination of the bulk power systems with new and emerging technologies: Identify standards development that help to accommodate the introduction
and expansion of renewable resources, demand response, and electricity storage
to address several bulk power system challenges. Also identify standards
development that help to accommodate another emerging technology, electric
transportation.
1.10.1
Utilities
South California Edison (SCE) and other utility companies undertook to reinvent electrical metering. Vendors are migrated to an open standards–based advanced metering
infrastructure. These contributions have led to the continual improvement of associated
features such as customer service, energy conservation, and economic efficiency.
PEPCO Holdings has been working on an Advanced Metering Infrastructure
(AMI). The technology is an integral component of the smart grid [5]. The features
proposed include investment in and implementation of innovative, customer-focused
technologies and initiatives for efficient energy management, increased pricing options
and demand response, reduction of total energy cost and consumption, and reduction
of the environmental impacts of electric power consumption.
1.10.2
Government Laboratory Demonstration Activities
Much of the fundamental thinking behind the smart grid concept arose from the DOE’s
Pacific Northwest National Laboratory (PNNL) more than 20 years ago. In the middle
1980s researchers at PNNL were already designing first-generation data collection
systems that were installed in more than 1000 buildings to monitor near real time
electricity consumption for every appliance. PNNL developed a broad suite of analytical tools and technologies that resulted in better sensors, improved diagnostics, and
enhanced equipment design and operation, from phasor measurement and control at the
transmission level to grid-friendly appliances [2]. In January 2006, four years after its
first presentation, PNNL unveiled the GridWise Initiative whose objective was the
testing of new electric grid technologies [3]. This demonstration project involved 300
homeowners in Washington and Oregon.
The GridWise Alliance manages the GridWise Program in the DOE’s Office of
Electricity and Energy Assurance. Members include Areva, GE, IBM, Schneider Electric; American Electric Power, Bonneville Power Administration, ConEd, the PJM
Interconnection; Battelle, RDS, SAIC, Nexgen, and RockPort Capital Partners [2]. The
GridWise Architecture Council [4], a primary advocate for the smart grid, promotes the
10
SMART GRID ARCHITECTURAL DESIGNS
benefits of improving interoperability between the automation systems needed to enable
smart grid applications.
1.10.3
Power Systems Engineering Research Center (PSERC)
The Power Systems Engineering Research Center (PSERC) [6] consists of 13 universities and industrial collaborators involved in research aimed at solving grid problems
using state-of-the-art technologies. The direction of PSERC is the development of new
strategies, technologies, analytical capabilities, and computational tools for operating
and planning practices that will support an adaptive, reliable, and stable power grid.
1.10.4
Research Institutes
The Electric Power Research Institute (EPRI) and university consortium groups have
developed software architecture for smart grid development. These tools focus on the
development of the grid’s technical framework through the integration of electricity
systems, communications, and computer controls. The Intelligrid software from EPRI,
an open-standard, requirements-based approach for integrating data networks and
equipment, enables interoperability between products and systems. It provides methodology, tools, and recommendations for standards and technologies for utility use in
planning, specifying, and procuring IT-based systems.
1.10.5
Technology Companies, Vendors, and Manufacturers
IBM is a major player in the provision of information technology (IT) equipment for
the smart grid on a global level. In 2008, IBM was chosen to spearhead IT support and
services for smart-grid energy-efficiency programs by American Electric Power, Michigan Gas and Electric, and Consumers Energy. IBM serves as the systems integrator for
its GridSmart program that displays energy usage and participate in energy-efficiency
program. Its Intelligent Power Grid is characterized by increased grid observability with
modern data integration and analytics to support advanced grid operation and control,
power delivery chain integration, and high-level utility strategic planning functions [7].
Some key characteristics of the Intelligent Power Grid are:
•
•
•
•
Grid equipment and assets contain or are monitored by intelligent IP-enabled
devices (digital processors).
Digital communication networks permit the intelligent devices to communicate
securely with the utility enterprise and possibly with each other.
Data from the intelligent devices and many other sources are consolidated to
support the transformation of raw data into useful information through advanced
analytics.
Business intelligence and optimization tools provide advanced decision support
at both the automatic and human supervisory level.
WORKING DEFINITION OF THE SMART GRID
11
The data base and architecture consist of five major components: data sources, data
transport, data integration, analytics, and optimization. In addition there are means for
data distribution which includes publish-and-subscribe middleware, portals, and Webbased services [8].
CISCO has also contributed with its IP architecture. CISCO describes the smart
grid as a data communication network integrated with the electrical grid that collects
and analyzes data captured in near-real time about power transmission, distribution,
and consumption. Predictive information and recommendations to stakeholders are
developed based on the data for power management. Integration of the generation,
transmission, distribution, and end user components is a critical feature.
There is no one acceptable or universal definition for the smart grid; rather it is
function-selected. Below we give a working definition to encompass the key issues of
stakeholders and developers.
1.11 WORKING DEFINITION OF THE SMART GRID BASED ON
PERFORMANCE MEASURES
A working definition should include the following attributes:
•
•
•
•
•
•
•
Assess grid health in real time
Predict behavior, anticipate
Adapt to new environments like distributed resources and renewable energy
resources
Handle stochastic demand and respond to smart appliances
Provide self-correction, reconfiguration, and restoration
Handle randomness of loads and market participants in real time
Create more complex interactive behavior with intelligent devices, communication protocols, and standard and smart algorithms to improve smart communication and transportation systems.
In this environment, smart control strategies will handle congestion, instability, or reliability problems. The smart grid will be cyber-secure, resilient, and able to manage
shock to ensure durability and reliability. Additional features include facilities for the
integration of renewable and distribution resources, and obtaining information to and
from renewable resources and plug-in hybrid vehicles. New interface technologies will
make data flow patterns and information available to investors and entrepreneurs interested in creating goods and services.
Thus, the working definition becomes:
The smart grid is an advanced digital two-way power flow power system capable of
self-healing, and adaptive, resilient, and sustainable, with foresight for prediction under
different uncertainties. It is equipped for interoperability with present and future standards
of components, devices, and systems that are cyber-secured against malicious attack.
12
SMART GRID ARCHITECTURAL DESIGNS
Transmission
Automation
System Coordination,
Situation Assessment
Distribution
Automation
System
Operations
Renewables
Integration
Energy
Efficiency
Demand Participation
Signals & Options
Smart Appliances,
PHEVs, & Storage
Distributed
Generation & Storage
Figure 1.5. DOE representative architecture of the smart grid design (architecture 1).
It is enabled to perform with robust and affordable real-time measurements and
enhanced communication technology for data/information transmission. It allows smart
appliances and facilitates the deployment of advanced storage technologies including
plug-in electric and hybrid vehicles and control options, and supports DSM and demand
response schemes.
1.12
REPRESENTATIVE ARCHITECTURE
Several types of architecture have been proposed by the various bodies involved in smart
grid development. We present two: one from the DOE and one illustrated by Figure 1.5,
which shows how the DOE’s proposed smart grid divides into nine areas: transmission
automation, system coordination situation assessment, system operations, distribution
automation, renewable integration, energy efficiency, distributed generation and storage,
demand participation signals and options, and smart appliances, PHEVs, and storage.
Figure 1.6 shows how the second architectural framework is partitioned into subsystems with layers of intelligence and technology and new tools and innovations. It
involves bulk power generation, transmission, distribution, and end user level of the
electric power system. The function of each component is explained in the next section.
1.13
FUNCTIONS OF SMART GRID COMPONENTS
For the generation level of the power system, smart enhancements will extend from the
technologies used to improve the stability and reliability of the generation to intelligent
controls and the generation mix consisting of renewable resources.
FUNCTIONS OF SMART GRID COMPONENTS
13
Figure 1.6. The intelligent grid (architecture 2).
1.13.1
Smart Devices Interface Component
Smart devices for monitoring and control form part of the generation components’ real
time information processes. These resources need to be seamlessly integrated in the
operation of both centrally distributed and district energy systems.
1.13.2
Storage Component
Due to the variability of renewable energy and the disjoint between peak availability
and peak consumption, it is important to find ways to store the generated energy for
later use. Options for energy storage technologies include pumped hydro, advance batteries, flow batteries, compressed air, super-conducting magnetic energy storage, supercapacitors, and flywheels. Associated market mechanisms for handling renewable
energy resources, distributed generation, environmental impact and pollution are other
components necessary at the generation level.
Associated market mechanism for handling renewable energy resources, distributed generation, environmental impact and pollution has to be introduced in the design
of smart grid component at the generation level.
14
1.13.3
SMART GRID ARCHITECTURAL DESIGNS
Transmission Subsystem Component
The transmission system that interconnects all major substation and load centers is the
backbone of an integrated power system. Efficiency and reliability at an affordable cost
continue to be the ultimate aims of transmission planners and operators. Transmission
lines must tolerate dynamic changes in load and contingency without service disruptions. To ensure performance, reliability and quality of supply standards are preferred
following contingency. Strategies to achieve smart grid performance at the transmission
level include the design of analytical tools and advanced technology with intelligence
for performance analysis such as dynamic optimal power flow, robust state estimation,
real-time stability assessment, and reliability and market simulation tools. Real-time
monitoring based on PMU, state estimators sensors, and communication technologies
are the transmission subsystem’s intelligent enabling tools for developing smart transmission functionality.
1.13.4
Monitoring and Control Technology Component
Intelligent transmission systems/assets include a smart intelligent network, selfmonitoring and self-healing, and the adaptability and predictability of generation and
demand robust enough to handle congestion, instability, and reliability issues. This new
resilient grid has to withstand shock (durability and reliability), and be reliable to
provide real-time changes in its use.
1.13.5
Intelligent Grid Distribution Subsystem Component
The distribution system is the final stage in the transmission of power to end users.
Primary feeders at this voltage level supply small industrial customers and secondary
distribution feeders supply residential and commercial customers. At the distribution
level, intelligent support schemes will have monitoring capabilities for automation
using smart meters, communication links between consumers and utility control, energy
management components, and AMI. The automation function will be equipped with
self-learning capability, including modules for fault detection, voltage optimization and
load transfer, automatic billing, restoration and feeder reconfiguration, and real-time
pricing.
1.13.6
Demand Side Management Component
Demand side management options and energy efficiency options developed for effective means of modifying the consumer demand to cut operating expenses from expensive generators and defer capacity addition.
DSM options provide reduced emissions in fuel production, lower costs, and contribute to reliability of generation. These options have an overall impact on the utility
load curve. A standard protocol for customer delivery with two-way information
highway technologies as the enabler is needed. Plug-and-play, smart energy buildings
and smart homes, demand-side meters, clean air requirements, and customer interfaces
for better energy efficiency will be in place.
SUGGESTED READINGS
1.14
15
SUMMARY
This chapter has discussed the progress made by different stakeholders in the design
and development of the smart grid. A working definition of the smart grid was given.
Two design architectures and the specific aspects of prospective smart grid function
were provided. The next chapters discuss the tools and techniques needed for smart
grid analysis and development.
REFERENCES
[1] “The Smart Grid: An Introduction and Smart Grid System Report.” Litos Strategic Communication, U.S. Department of Energy, 2009.
[2] L.D. Kinter-Meyer, M.C. Chassin, D.P. Kannberg, et al. “GridWiseTM: The Benefits of a
Transformed Energy System.” Pacific Northwest National Laboratory, PNNL-14396, 2003.
[3] “Overview of the Smart Grid: Policies, Initiatives and Needs.” ISO New England, 2009.
[4] “The Modern Grid Initiative.” GridWise Architecture Council, Pacific Northwest National
Laboratory, 2008.
[5] “Our Blueprint for the Future.” PEPCO Holdings, 2009.
[6] “PSERC Overview, 2008.” PSERC, 2008.
[7] J. Taft. “The Intelligent Power Grid.” IBM Global Services, 2006.
[8] “A National Vision for Electricity’s Second 100 Years.” Office of Electric Transmission and
Distribution, U.S. Department of Energy, 2003.
SUGGESTED READINGS
American Recovery and Reinvestment Act of 2009. Public Law No. 111-5, 2009.
P. Van Doren and J. Taylor. “Rethinking Electricity Restructuring.” Policy Analysis 2004, 530, 1–8.
EPRI Intelligrid. Electric Power Research Institute, 2001–2010.
“Smart Grid System Report.” U.S. Department of Energy, 2009.
The Energy Independence and Security Act of 2007. S. 1419, 90d Congress, 2007.
“The Smart Grid: An Introduction and Smart Grid System Report.” U.S. Department of Energy,
2009.
2
SMART GRID
COMMUNICATIONS AND
MEASUREMENT TECHNOLOGY
2.1
COMMUNICATION AND MEASUREMENT
Because much of the existing transmission and distribution systems in the United States
still uses older digital communication and control technology, advanced communication
systems for distribution automation, such as Remote Terminal Unit (RTU) [3] and
SCADA, are under development as well as innovative tools and software that will
communicate with appliances in the home [1]. Ultimately, high-speed, fully integrated,
two-way communication technologies will allow the smart grid to be a dynamic, interactive mega-infrastructure for real-time information and power exchange.
The technology exists for the measure, monitor, and control in real time in the
Smart Grid, and this technology plays an essential role in the functioning of the Smart
Grid. Issues of standards, cyber security, and interoperability which are dealt with more
extensively in Chapter 8 impact most definitely on communication. There is need for
the formalization of the standards and protocols which will be enforced for the secured
transmission of critical and highly sensitive information within the communications
scheme.
Obviously, existing measuring, monitoring, and control technology will have a role
in smart grid capability. Establishing appropriate standards, cyber security, and interop-
Smart Grid: Fundamentals of Design and Analysis, First Edition. James Momoh.
© 2012 Institute of Electrical and Electronics Engineers. Published 2012 by John Wiley & Sons, Inc.
16
COMMUNICATION AND MEASUREMENT
17
erability (discussed in Chapter 8) requires careful study, for example, formalizing the
standards and protocols for the secure transmission of critical and highly sensitive
information within the proposed communication scheme. Moreover, open architecture’s
plug-and-play environment will provide secure network smart sensors and control
devices, control centers, protection systems, and users. Possible wired and wireless
communications technologies can include:
1. Multiprotocol Label Switching (MPLS): high-performance telecommunications
networks for data transmission between network nodes
2. Worldwide Interoperability for Microwave Access (WiMax): wireless telecommunication technology for point to multipoint data transmission utilizing Internet technology
3. Broadband over Power Lines (BPL): power line communication with Internet
access
4. Wi-Fi: commonly used wireless local area network
Additional technologies include optical fiber, mesh, and multipoint spread spectrum.
The five characteristics of smart grid communications technology are:
1.
2.
3.
4.
5.
High bandwidth
IP-enabled digital communication (IPv6 support is preferable)
Encryption
Cyber security
Support and quality of service and Voice over Internet Protocol (VoIP)
Reliable intercommunication of hardware and software will require configuring
several types of network topologies. Below is a summary of the most likely candidates.
Local Area Network [5,6] consists of two or more components and high-capacity
disk storage (file servers), which allow each computer in a network to access a common
set of rules. LAN has operating system software which interprets input, instructs
network devices, and allows users to communicate with each other. Each hardware
device (computer, printer, and so on) on a LAN is a node. The LAN can operate or
integrate up to several hundred computers. LAN combines high speed with a geographical spread of 1–10 km. LAN may also access other LANs or tap into Wide Area Networks. LAN with similar architectures are bridges which act as transfer points, while
LAN with different architectures are gateways which convert data as it passes between
systems.
LAN is a shared access technology, meaning that all of the attached devices share
a common medium of communication such as coaxial, twisted pair, or fiber optics cable.
A physical connection device, the Network Interface Card (NIC), connects to the
network. The network software manages communication between stations on the
system.
18
SMART GRID COMMUNICATIONS AND MEASUREMENT TECHNOLOGY
The special attributes and advantages of LAN include:
•
•
•
•
•
Resource sharing: allows intelligent devices such as storage devices, programs,
and data files to share resources, that is, LAN users can use the same printer on
the network; the installed database and the software can be shared by multiple
users
Area covered: LAN is normally restricted to a small geographical area, for
example, office building, utility, campus
Cost and availability: application software and interface devices are affordable
and off-the-shelf
High channel speed: ability to transfer data at rates between 1 and 10 million
bits per second
Flexibility: grow/expand with low probability of error; easy to maintain and
operate
LAN has three categories of data transmission:
1. Unicast transmission: a single data packet is sent from a source node to a destination (address) on the network
2. Multicast transmission: a single data packet is copied and sent to a specific
subset of nodes on the network; the source node addresses the packet by using
the multicast addresses
3. Broadcast transmission: a single data packet is copied and sent to all nodes on the
network; the source node addresses the packet by using the broadcast address
LAN topologies define how network devices are organized. The four most common
architectural structures are:
1. Bus topology: linear LAN architecture in which transmission from network
station propagates the length of the medium and is received by all other stations
connected to it
2. Ring bus topology: a series of devices connected to one another by unidirectional transmission links to form a single closed loop
3. Star topology: the end points on a network are connected to a common central
hub or switch by dedicated links
4. Tree topology: identical to the bus topology except that branches with multiple
nodes are also possible
The devices and software used in LAN utilize a standard protocol such as Ethernet/
IEEE 802.3, Token Ring/IEEE 802.5 or 880.2 (available through IEEE Press).
Home Access Network [2,3] is a LAN confined to an individual home. It enables
remote control of automated digital devices and appliances throughout the house. Smart
meters, smart appliances and Web-based monitoring can be integrated into this level.
MONITORING, PMU, SMART METERS, AND MEASUREMENTS TECHNOLOGIES
19
Neighborhood Area Network (NAN) is a wireless community currently used for
wireless local distribution applications. Ideally, it will cover an area larger than a LAN.
Some architectural structures will focus on the integration and interoperability of
the various domains within the smart grid. Domains consist of groups of buildings,
systems, individuals, or devices which have similar communications characteristics:
•
•
•
•
•
•
•
Bulk generation: includes market services interface, plant control system, and
generators; this domain interacts with the market operations and transmission
domains through wide area networks, substation LANs, and the Internet
Transmission: includes substation devices and controllers, data collectors, and
electric storage; this domain interacts with bulk generation and operations
through WANs and substation LANs; integrated with the distribution domain
Distribution: this domain interacts with operations and customers through Field
Area Networks
Customer: includes customer equipment, metering, Energy Management
Systems (EMS), electric storage, appliances, PHEVs, and so on
Service Providers: includes utility and third party providers which handle billing
customer services, and so on; this domain interacts with operations and customers
primarily through the Internet
Operations: includes EMS, Web Access Management System (WAMS), and
SCADA; this domain can be sub-divided into ISO/RTO, transmission, and
distribution
Market: includes /ISOs/RTOs, aggregators, and other market participants
2.2 MONITORING, PMU, SMART METERS, AND MEASUREMENTS
TECHNOLOGIES
The smart grid environment requires the upgrade of tools for sensing, metering, and
measurements at all levels of the grid. These components will provide the data necessary for monitoring the grid and the power market. Sensing provides outage detection
and response, evaluates the health of equipment and the integrity of the grid, eliminates
meter estimations, provides energy theft protection, enables consumer choice, DSM,
and various grid monitoring functions.
With regard to metering and measurement, new digital technologies using two-way
communications, a variety of inputs (pricing signals, time-of-day tariff, regional transmission organization (RTO) [7] curtailments for congestion relief), a variety of outputs
(real time consumption data, power quality, electric parameters), the ability to connect
and disconnect, and interfaces with generators, grid operators, and customer portals to
enhance power measurement. This is facilitated by the increased utilization of digital
electronics for metering and measurements, advancement of the electric meter at the
customer level, and installation of wide area monitoring system for advanced utility
monitoring and protection [11,12]. Details of these measurements are discussed in later
sections.
20
SMART GRID COMMUNICATIONS AND MEASUREMENT TECHNOLOGY
New digital technologies will employ two-way communication, a variety of inputs
(pricing signals, time-of-day tariffs, RTO) [7] curtailments for congestion relief), a
variety of outputs (real-time consumption data, power quality, electric parameters), the
ability to connect and disconnect, and interfaces with generators, grid operators, and
customer portals to enhance power measurement. This introduces the increased utilization of digital electronics for metering and measurements, advancement of the electric
meter at the customer level, and installation of wide area monitoring systems (WAMs)
for advanced utility monitoring and protection [11,12]. Details of these measurements
are discussed in later sections.
2.2.1
Wide Area Monitoring Systems (WAMS)
WAMS are designed by the utilities for optimal capacity of the transmission grid and
to prevent the spread of disturbances. By providing real-time information on stability
and operating safety margins, WAMS give early warnings of system disturbances for
the prevention and mitigation of system-wide blackouts. WAMS utilize sensors distributed throughout the network in conjunction with GPS satellites for precise time stamping of measurements in the transmission system. The integrated sensors will interface
with the communication network. Phasor Measurements are a current technology that
is a component of most smart grid designs.
2.2.2
Phasor Measurement Units (PMU)
Phasor Measurement Units or Synchrophasors give operators a time-stamped snapshot
of the power system. The PMUs consist of bus voltage phasors and branch current
phasors, in addition to information such as locations and other network parameters
[9,10]. Phasor measurements are taken with high precision from different points of the
power system at the same instant, allowing an operator to visualize the exact angular
difference between different locations. PMUs are equipped with GPS receivers which
allow synchronization of readings taken at distant points [12]. Microprocessor-based
instrumentation such as protection relays and Disturbance Fault Recorders (DFRs)
incorporate the PMU module with other existing functionalities as an extended feature.
The IEEE standard on Synchrophasors specifies the protocol for communicating the
PMU data to the Phasor Data Concentrator. Figure 2.1 illustrates the PMU measurement
system from Reference 8.
PMUs ensure voltage and current with high accuracy at a rate of 2.88 kHz. They
can calculate real power, reactive power, frequency, and phase angle 12 times per 60
hertz cycle. The actual sampling rate used to achieve this output is 1.4 MHz [3]. Recent
trends now require fast controls and online implementations for mitigating voltage
collapse in the shortest, least-cost time [8]. Over the years, researchers and engineers
have found PMUs are suitable for monitoring and control of voltage stability PMUs)
[6–9].
Offering wide-area situational awareness, phasor measurement work to ease congestion, bottlenecks and mitigate—or even prevent—blackouts. When integrated with
Smart Grid communications technologies, the measurements taken will provide dynamic
MONITORING, PMU, SMART METERS, AND MEASUREMENTS TECHNOLOGIES
21
Control
Center
Figure 2.1. Conceptual diagram of a synchronized phasor measuring system [9].
visibility into the power system. Adoption of the Smart Grid with real time measurement will enhance every facet of the electric delivery system including generation,
transmission, distribution, and consumption. It will increase the possibilities of distributed generation, bringing generation closer to those it serves.
Additional utility monitoring systems include dynamic line rating technology,
conductor sensors, insulation contamination leakage current, backscatter radios technology, electronic instrument transformers, and monitors for circuit breaker, cables,
batteries, temperature, and current frequency and so on [10, 11]. Research into the
application and integration of these measurements into the smart grid is particularly
important and continuing work.
2.2.3
Smart Meters
Smart meters have two functions: providing data on energy usage to customers (endusers) to help control cost and consumption; sending data to the utility for load factor
control, peak-load requirements, and the development of pricing strategies based
on consumption informationand so on Automated data reading is an additional
22
SMART GRID COMMUNICATIONS AND MEASUREMENT TECHNOLOGY
component of both smart meters and two-way communication between customers
and utilities. The development of smart meters is planned for electricity, water, and gas
consumption [9].
Smart meters equip utility customers with knowledge about how much they pay
per kilowatt hour and how and when they use energy. This will result in better pricing
information and more accurate bills in addition to ensuring faster outage detection and
restoration by the utility. Additional features will allow for demand-response rates, tax
credits, tariff options, and participation in voluntary rewards programs for reduced
consumption. Still other features will include remote connect/disconnect of users, appliance control and monitoring, smart thermostat, enhanced grid monitoring, switching,
and prepaid metering.
With governmental assistance, large-scale deployment of smart meters has begun
throughout the United States. Academic participation in the research and development
of metering tools and techniques for network analysis enhancement and the use of smart
meter outputs for voltage stability and security assessment and enhancement have been
proposed.
2.2.4
Smart Appliances
Smart appliances cycle up and down in response to signals sent by the utility. The
applicances enable customers to participate in voluntary demand response programs
which award credits for limiting power use in peak demand periods or when the grid
is under stress. An override function allows customers to control their appliances using
the Internet.
Air conditioners, space heaters, water heaters, refrigerators, washers, and dryers
represent about 20% of total electric demand during most of the day and throughout
the year [10]. Grid-friendly appliances use a simple computer chip that can sense disturbances in the grid’s power frequency and can turn an appliance off for a few minutes
to allow the grid to stabilize during a crisis.
2.2.5
Advanced Metering Infrastructure (AMI)
AMI is the convergence of the grid, the communication infrastructure, and the supporting information infrastructure. The network-centric AMI coupled with the lack of a
composite set of cross industry AMI security requirements and implementation guidance, is the primary motivation for its development. The problem domains to be
addressed within AMI implementations are relatively new to the utility industry;
however, precedence exists for implementing large-scale, network-centric solutions
with high information assurance requirements. The defense, cable, and telecom industries offer many examples of requirements, standards, and best practices that are directly
applicable to AMI implementations.
The functions of AMI can be subdivided into three major categories:
•
Market applications: serve to reduce/eliminate labor, transportation, and infrastructure costs associated with meter reading and maintenance, increase accuracy
GIS AND GOOGLE MAPPING TOOLS
•
•
23
of billing, and allow for time-based rates while reducing bad debts; facilitates
informed customer participation for energy management
Customer applications: serves to increase customer awareness about load
reduction, reduces bad debt, and improves cash flow, and enhances customer
convenience and satisfaction; provides demand response and load management
to improve system reliability and performance
Distribution operations: curtails customer load for grid management, optimizes
network based on data collected, allows for the location of outages and restoration of service, improves customer satisfaction, reduces energy losses, improves
performance in event of outage with reduced outage duration and optimization
of the distribution system and distributed generation management, provides
emergency demand response
An extension of AMI will be smart meters which handle customers’ gas and water usage
data. The issues associated with ensuring network and data security are discussed in
Chapter 8.
2.3
GIS AND GOOGLE MAPPING TOOLS
GIS is useful for managing traditional electric transmission and distribution and telecom
networks. It can also help to manage information about utility assets for data collection
and maintenance.
Google’s free downloadable Google Earth software offers geographical contextual
information in an updated user-friendly platform that facilitates inquiry-based study
and analysis. Users can create and share many types of dynamically-updating data over
the Internet. Keyhole Markup Language (KML) allows them to overlay basic data types
such as images, point data, lines, and polygons [7]. Through satellite imagery, maps
are available from space to street-level. The integration of GIS with Google Earth or
other mapping tools will aid in understanding the relationship of the grid network to
its surroundings, for example, determining the optimal location of rights of way, placement of sensors and poles, and so on. GIS technology will provide partial context to
operators and planners, for example, real-time sensors that collect the data needed to
reconfigure networks for reducing outages and equipment failures.
The trends in the development of the electric power system and the expectation of
future demand suggest the following needs:
1.
2.
3.
4.
5.
6.
Reducing outage time
Preventing power theft which causes significant unaccounted losses
Effective system for collection and billing system
Expanding services for customers
Effective asset management
Improving reliability such as SAIDI (System Average Interruption Duration
Index) and SAIFI (System Average Interruption Frequency Index) for distribution networks
24
SMART GRID COMMUNICATIONS AND MEASUREMENT TECHNOLOGY
7. Improving analysis of customer complaint logs
8. Enhancing load flow power quality analysis and fault study for current and
anticipated problems
9. Scheduling of actions such as load shedding and vegetation control
2.4
MULTIAGENT SYSTEMS (MAS) TECHNOLOGY
MAS are a computational system in which several agents cooperate to achieve a desired
task. The performance of MAS can be decided by the interactions among various
agents. Agents cooperate to achieve more than if they act individually.
Increasingly, MAS are the preferential choice for developing distributed systems.
The development of monitoring and measurement schemes within the smart grid environment can be enhanced through the use of MAS architecture (Fig. 2.2). As an
example, MAS have been utilized as a detection and diagnosis device and in system
monitoring. Such architectures utilize a collection of agents such as Arbitrator Agents
(AA), System Monitoring Agents (SMA), Fault Detection Agents (FDA), Diagnosis
Agents (DA), a Judgment Index Agent (JIA), and a Scheduling Agent (SA). Information
passes between the agents about the appropriate actions to be taken. When implemented, the process repeats itself to constantly monitor the system so that management
of system conditions can be implemented instantaneously.
Arbitrator
Enforced MAS Behavior and
Assessment Control
MAS Structure for Shared
Ontology
Agent Function
Organization of Agent
Observer
Fault Tolerant
Programming
Diagnose
Computing Platform /
language
Decide
Cost Effectiveness
Computation
Act
Agent Diagnosis and
Reporting
Report
Adaptability and
Flexibility
Communication Layer – Language and Protocols
Network Layer (System Under Control)
Figure 2.2. Simplified multiagent architecture.
MULTIAGENT SYSTEMS (MAS) TECHNOLOGY
2.4.1
25
Multiagent Systems for Smart Grid Implementation
As mentioned in Chapter 1, the DOE’s Modern Grid Initiative [1] explains that a smart
grid integrates advanced sensing technologies, control methods, and integrated communications into the present electricity grid at both transmission and distribution levels.
The smart grid is expected to have the following key characteristics:
1.
2.
3.
4.
5.
6.
7.
self-healing
consumer friendly
attack resistant
provide power quality for 21st-century needs
accommodate all generation and storage options
enable markets
optimize assets and operate efficiently
Central to the operation of any power system is its control architecture consisting of
hardware and software protocols for exchanging system status and control signals. In
conventional electric power systems, this is accomplished by SCADA [4, 5]. Current
trends to control and monitor system operations are moving toward the use of MAS.
A multiagent system is a combination of several agents working in collaboration pursuing assigned tasks to achieve the overall goal of the system. The multiagent system has
become an increasingly powerful tool in developing complex systems that take advantages of agent properties: autonomy, sociality, reactivity and pro-activity [6].
The multiagent system is autonomous in that they operate without human interventions. The multiagent system is sociable in that they interact with other agents via some
kind of agent communication language. The agents also perceive and react to their
environment. Lastly, the multiagent system is proactive in that they are able to exhibit
goal oriented behavior by taking initiatives.
2.4.2
Multiagent Specifications
In this section, the specifications of a control agent, a distributed energy resource (DER)
agent, a user agent, and a database agent in the Intelligent Distributed Autonomous
Power System (IDAPS) MAS are defined.
1. Control agent: responsibilities include monitoring system voltage and frequency
to detect contingency situations or grid failures, and sending signals to the main
circuit breaker to isolate the IDAPS microgrid from the utility when an upstream
outage is detected; receiving electricity price ($/kWh) signal from the main grid,
which may be obtained from AMI, and publishing them to the IDAPS entities
2. DER agent: responsibilities include storing associated DER information, monitoring and controlling DER power levels and connect/disconnect status; DER
information to be stored may include DER identification number, type (solar
26
SMART GRID COMMUNICATIONS AND MEASUREMENT TECHNOLOGY
cells, microturbines, fuel cells), power rating (kW), local fuel availability, cost
function or price at which users agree to sell, DER availability, that is, planned
maintenance schedule
3. User agent: acts as a customer gateway that makes features of an IDAPS
microgrid accessible to users; responsibilities include providing users with realtime information on entities residing in the IDAPS system; monitors electricity
consumption by each critical and noncritical load; allows users to control the
status of loads based on user ’s predefined priority
4. Database agent: serves as a data access point for other agents as well as users;
responsibilities include storing system information, recording messages and
data shared among agents.
2.4.3
Multiagent Technique
An agent of a MAS may be defined as an entity with attributes considered useful in a
particular domain. In this framework, an agent is an information processor that performs
autonomous actions based on information. Common agent attributes include:
•
•
•
•
•
Autonomy: goal-directedness, proactive and self-starting behavior
Collaborative behavior: the ability to work with other agents to achieve a common
goal
Knowledge-level communication ability: the ability to communicate with other
agents with language resembling human speech acts rather than typical symbollevel program-to-program protocols
Reactivity: the ability to selectively sense and act
Temporal continuity: persistence of identity and state over long periods
MAS can be characterized by:
•
•
•
•
Each agent has incomplete capabilities to solve a problem
No global system control
Decentralized data
Asynchronous computation
For instance, the system outage of a ship could be caused by an internal system
error or any external contingency from battle. To pursue the best ship performance, it
is very important to restore the electric power supply as much as possible. When a fault
occurs on the ship power system, the protection systems will isolate the fault from the
remaining power grid. Then the system should restore the electric power to a target
configuration after the outage.
An example of MAS architecture in action is a power failure on board a ship that
is caused by an internal system error, an external contingency from battle, and so on.
REFERENCES
27
Clearly, the goal is rapid restoration of the onboard power supply; hence, when a fault
occurs, the protection system will isolate the fault, allowing the system to restore power
to a target configuration after the outage.
2.5
MICROGRID AND SMART GRID COMPARISON
Research has been conducted to understand the differences between a microgrid and a
smart grid. Basically, a microgrid is a local island grid that can operate as a stand-alone
or as a grid-connected system. It is powered by gas turbines or renewable energy and
includes special purpose inverters and a link for plug-and-play to the legacy grid.
Special purpose filters overcome harmonics problems while improving power quality
and efficiency. Several demonstration projects and a testbed are operating in university
and government facilities. In summary, think of the microgrid as a local power provider
with limited advanced control tools and the smart grid is a wide area provider with
sophisticated automated decision support capabilities.
2.6
SUMMARY
This chapter has focused on various communication aspects of the smart grid. The
measurement techniques described included PMUs and smart meters. GIS was introduced as a planning tool to facilitate locating important components. The relationship
of MAS to the smart grid developmental process was also described.
REFERENCES
[1] J.A. Momoh. Electric Power System Application of Optimization. Marcel Dekker, New
York, 2001.
[2] J.L. Marinho and B. Stott. “Linear Programming for Power System Network Security
Applications.” IEEE Transactions on Power Apparatus and Systems 1979, PAS-98,
837–848.
[3] R.C. Eberhart and J. Kennedy. “A New Optimizer Using Particle Swarm Theory.” In Proceedings on the Sixth International Symposium on Micromachine and Human Science 1995,
39–31.
[4] G. Riley and J. Giarratano. Expert Systems: Principles and Programming. PWS Publisher,
Boston, 2003.
[5] A. Englebrecht. Computational Intelligence: An Introduction, John Wiley & Sons, 2007.
[6] M. Dorigo and T. Stuzle. Ant Colony Optimization. Massachusetts Institute of Technology,
Cambridge, 2004.
[7] A.G. Barto, W.B. Powell, D.C. Wunsch, and J. Si. Handbook of Learning and Approximate
Dynamic Programming. IEEE Press Series on Computational Intelligence, 2004.
[8] J.A. Momoh. Electrical Power System Applications of Optimization, CRC Press, 2008.
28
SMART GRID COMMUNICATIONS AND MEASUREMENT TECHNOLOGY
[9] W.H. Zhange and T. Gao. “A Min-Max Method with Adaptive Weightings for Uniformly
Spaced Pareto Optimum Points.” Computers and Structures 2006, 84, 1760–1769.
[10] P.K. Skula and K. Deb. “On Finding Multiple Pareto-Optimal Solutions Using Classical
and Evolutional Generating Methods.” European Journal of Operational Research 2007,
181, 1630–1652.
[11] M. Dorigo and T. Stutzle. “The Ant Colony Optimization Metaheuristic: Algorithms, Applications and Advances.” In F. Glover and G. Kochenberger, eds.: Handboook of Metaheuristics. Norwell, MA, Kluwer, 2002.
[12] A.G. Phadhke. “Synchronized Phasor Measurements in Power Systems.” IEEE Comput.
Appl. Power 1993, 6, 10–15.
3
PERFORMANCE ANALYSIS
TOOLS FOR SMART
GRID DESIGN
3.1
INTRODUCTION TO LOAD FLOW STUDIES
Load flow studies are critical to system planning and system operation. For example,
data on peak load conditions assists planners in determining the size of components
(conductors, transformers, reactors, and shunt capacitors), siting new generation and
transmission, and planning interties with neighboring systems to meet predicted demand
consistent with the North American Electric Reliability Corporation’s (NERC) reliability requirements. Load flow studies identify line loads and bus voltages out of range,
inappropriately large bus phase angles (and the potential for stability problems), component loads (principally transformers), proximity to Q-limits at generation buses, and
other parameters having the potential to create operating difficulties. Intermediate load
and off-peak (minimum) load studies are also useful, since off-peak loads can result in
high voltage conditions that are not identified during peak loads. Load flow studies
assist system operators in calculating power levels at each generating unit for economic
dispatch, analyzing outages and other forced operating conditions (contingency analysis
[7]), and coordinating power pools. In most instances, load flow studies are used to
assess system performance and operations under a given condition.
Smart Grid: Fundamentals of Design and Analysis, First Edition. James Momoh.
© 2012 Institute of Electrical and Electronics Engineers. Published 2012 by John Wiley & Sons, Inc.
29
30
PERFORMANCE ANALYSIS TOOLS FOR SMART GRID DESIGN
3.2 CHALLENGES TO LOAD FLOW IN SMART GRID AND
WEAKNESSES OF THE PRESENT LOAD FLOW METHODS
Current legacy methods have weaknesses that need to be addressed prior to their use
in analyzing smart grid performance and operations. Four fundamental questions should
be answered:
1.
2.
3.
4.
What are the special features of the smart grid compared to the legacy system?
What computations are needed in the case of smart grid?
What specific directions are needed for developing a new power flow?
What new features of the load flow make it suitable for smart grid performance
and evaluation?
Table 3.1 compares the old and desired load flow techniques.
Other features to be considered in the development of the new load flow include:
1. Condition adaptiveness [9–11] of transmission and distribution to accommodate
load flows comprising renewable generation
2. Self-adaptiveness to ensure proper coordination
3. High impedance topology matching for distribution network with randomness
and uncertainty requiring intelligence analytical tools
4. Since reverse power flow technique is possible, the use of FACTs devices to
power electronics building blocks is essential.
Existing load flow performance tools capable of determining voltage, angle, flows,
MW/MVAr, and scheduling dispatch are mostly offline although a few can give realtime results. To enhance load flow capabilities, the smart grid load flow process consists
of the following steps:
1. Data acquisition for radial or mesh network
2. Existence of connection of data to assure network feasibility
TABLE 3.1. Load Flow Techniques Comparison
Old Load Flow Technique
Central generation and control
Load flow by Kirchhoff ’s laws
Power generation according to demand
Manual switching and trouble response
Simulation and response tracking
Desired Load Flow Technique
Central and distributed generation control and
distributed intelligence
Load flow by electronics
Controllable generation, fluctuating/random
sources and demand in dynamic [1,6] equilibrium
Automatic response and predictive avoidance
Monitoring overload against bottlenecks
LOAD FLOW STATE OF THE ART
31
3. Formulation of Y-bus for representing the interconnection of the system under
study and determination of initial conditions
4. Solution of mismatch real and reactive powers and checking of mismatch by
adjusting the initial conditions typically called the hot state
5. With the snapshot power demand (real power and reactive power demand),
determining a feasible static voltage and angle to minimize mismatch
The process is defined as the iterative power simulation which is described mathematically in three different formulation frameworks. The process iterates until convergence
is reached so that the program terminates with solutions of voltage, angle, flows, losses,
and minimum mismatch power injected.
3.3 LOAD FLOW STATE OF THE ART: CLASSICAL, EXTENDED
FORMULATIONS, AND ALGORITHMS
The traditional load flow techniques used for distribution load flow are characterized by:
1.
2.
3.
4.
Distribution systems are radial or weakly meshed network structures
High X/R ratios in the line impedances
Single phase loads handled by the distribution load flow program
Distributed Generation (DG), other renewable generation, and/or cogeneration
power supplies installed in relative proximity to some load centers
5. Distribution systems with many short line segments, most of which have low
impedance values
For the purpose of load flow study we model the network of buses connected by lines
or switches connected to a voltage-specific source bus. Each bus may have a corresponding load composite form (consisting of inductor, shunt capacitor, or combination).
The load and/or generator are connected to the buses.
The classical methods of studying load flow include:
1. Gauss–Seidal
2. Newton–Raphson
3. Fast Decouple
We summarize each of them here for easy reference.
3.3.1
Gauss–Seidal Method
This method uses Kirchhoff ’s current law nodal equations given as Iinjection current at
the node. Suppose Iinjection = current at the node of a given connected load, then
32
PERFORMANCE ANALYSIS TOOLS FOR SMART GRID DESIGN
n
I inj ( j ) =
∑I
ji
i =1
where Iinj(j) is the injection current at bus j and Iji = current flow from jth bus to ith
bus. Rewriting, we obtain Iinj(j) = Ybus Vbus where Ybus admittance matrix is given
as Vbus vector of bus voltages.
If we sum the total power at a bus, the generation and load is denoted as complex
power. The nonlinear load flow equation is:
Sinj −k = Pg + jQg − ( PL D + jQLD )
= Vk
*
YkjV j
j =1
n
∑
This equation is solved by an iterative method for Vj if P and Q are specified.
Additionally, from
k
)
Vi( k +1) = g(Vbus
=
1 PLsch − jQLsch
−
Yii
VL*( k )
n
∑Y V
ij
(k )
j
j =1
where Yij are the elements of bus admittance matrix, and Pi sch and Qisch are scheduled P
and Q at each bus.
After a node voltage is updated within iteration, the new value is made available
for the remaining equations within that iteration and also for the subsequent iteration.
Given that the initial starting values for voltages are close to the unknown, the iterative
process converges linearly.
Notably, the classical method performance is worse in a radial distribution system
because of the lack of branch connections between a large set of surrounding buses. It
should be noted that the injection voltage correction propagates out to the surrounding
buses on the layer of neighboring buses for each iteration.
3.3.2
Newton–Raphson Method
The Newton–Raphson Method assumes an initial starting voltage use in computing
k ∗
k
mismatch power ΔS where ∆S = Sijsch
−i − (Vi ) ( ∑ YijV j ). The expression ΔS is called the
mismatch power. In order to determine convergence criteria given by ΔS ≤ ε, where
ε is a specific tolerance, accuracy index, and a sensitivity matrix is derived from the
inverse Jacobian matrix of the injected power equations:
Pi = Vi ∑ Yij V j cos (θi − θ j −ψ ij )
Qi = Vi ∑ Yij V j sin (θi − θ j −ψ ij )
where θi is the angle between Vi and Vj, and ψij is admittance angle.
LOAD FLOW STATE OF THE ART
33
The complex power ΔS can be expressed in polar or rectangular form |ΔV| = (Δe + Δf)
or ΔV = |ΔV|∠θv or ΔS = ΔP + ΔQ respectively.
This method is excellent for large systems but does not take advantage of the radial
structure of distribution and hence is computationally inefficient. The method fails
when the Jacobian matrix is singular or the system becomes ill-conditioned as in the
case of a low distribution X/R ratio.
3.3.3
Fast Decouple Method
The fast decouple method simplifies the Jacobian matrix by using small angle approximations to eliminate relatively small elements of the Jacobian. The method is one of
the effective techniques used in power system analysis. However, it exhibits poor convergence with a high R/X ratio system. The interaction of V and θ magnitudes with
active and reactive power flows cause poor convergence as well.
A variation solves current injection instead of model power injection power
equations.
3.3.4
Distribution Load Flow Methods
Due to the limitation of the fast decouple method in solving an ill-conditioned system
with a high X/R ratio, the distribution load flow [12] techniques above require alternative methods. We summarize the commonly used methods.
1. Forward/backward sweep methods solves branch current or load flow by using
the forward sweeping method
2. Compute the nodal voltages using backward sweep approach
3. Newton method uses power mismatches at the end of feeders and laterals to
iteratively solve the nodal voltage
4. Gauss method on the bus impedance matrix equation solves iteratively for the
branch currents.
Method 1: Forward/Backward Sweep. This method models the distribution system
as a tree network, with the slack bus denoted as the root of the tree and the branch
networks as the layers which are far away from the root nodal. Weakly meshed networks
are converted to a radial network by breaking the loops and injection currents
computation.
The backward sweep primarily sums either the line currents or load flows from the
extreme feeder (leaf) to the root. The steps of the algorithm are:
1. Select the slack bus and assume initial voltage and angle at the root, node, and
other buses
2. Compute nodal current injection at the Kth iteration
34
PERFORMANCE ANALYSIS TOOLS FOR SMART GRID DESIGN
S sch
I i( k ) = (ik −1)
Vi
*
3. Start from the root with known slack bus voltages and move toward the feeder
and lateral ends
4. Compute the voltage at node j
V jk −1 = Vi k − Z ij I ij( k )
where Zij is the branch impedance between bus i and j and Vj is the latest voltage value
of bus j
5. Compute the power mismatch from and check the termination criteria using
∆Si( k ) = Sisch − Vi( k ) ( I i( k ) ) ≤ ε
*
6. If step above is not reached repeat the previous steps until convergence is
achieved.
Note that in Step 2 from each known load power S, the lateral voltages are computed
or assumed. This involves Vi k −1 as the k − l past iteration of bus voltage and I i( k ) is the
Kth current iteration of injected current. We do this by starting from the last branch
from the lateral feeder and moving back through the tree node. This is done using the
expression
S sch
I i( k ) = i( k )
Vi
as before for all interconnected branches.
Method 2: Load Flow Based on Sensitivity Matrix for Mismatch Calculation. The
distribution load flow is an improved forward/backward method utilizing a sensitivity
matrix scheme to compensate the mismatch between slack bus power injection and the
load flow at the feeder and lateral ends. This results in the Newton–Raphson method
for distribution load flow.
V0
Vk-1
Vk
Vn
Pk+jQk
rk+jxk
Yk1
Yk2
Figure 3.1. Single feeder representation.
Pn+jQn
LOAD FLOW STATE OF THE ART
35
Consider a single feeder, as shown in Figure 3.1. The steps for this method are:
1. Assume the slack bus as the root node
2. Assume P0, Q0 power injection at the slack bus node equal to the sum of all of
the loads in the system
3. Load flows in each branch are equal to the sum of downstream connected loads.
At kth iteration, start from root node with known voltage at slack bus
4. Obtain the latest Vk, Pijk , Qijk (voltage and flows)
5. Compute power loss = f (V k , Pij* , Qij* )
6. From the loss compute receiving and power Pji, Qji, and Vj
7. The loads and shunt power are taken from the received power and the remaining
power is sent to the next feeder at lateral branches
8. At network solutions ΔPL, ΔQL = 0, when mismatch power is approximately
0; if load flow mismatch is less than the tolerance, ε, then load flow has
converged
9. Update the slack bus power from the sensitivity matrix
Method 3: Bus Impedance Network. This method uses the bus impedance matrix
and equivalent current injection to solve the network equation in a distribution system.
It employs a simple superposition [3,4] to find the bus voltage through the system. The
voltage in each bus is computed after specifying the slack bus voltage and then computing the incremental change ΔV due to current injection flowing into the network.
The steps are:
1. Assume no load system
2. Initialize the load bus voltage throughout the system using the value of the slack
bus voltage
3. Modify nodal voltages due to current flow which are function of loads
connected
4. The injection current is modified in Kth iteration as level changes
5. Use I i( k ) = ( Sisch Vi k −1 )* for the first equivalent current injection until getting I kj at the
last iteration I 0( k )
(k )
(k )
= Z bus I inj
6. Compute the vector of voltage denoted as ΔV using ∆Vbus
, where Zbus
is a η × η bus impedance matrix
7. Determine the bus voltage updates throughout the network as Vi k −1 = V0 − ∆Vi( k −1)
where V0 is slack bus voltage at root node
8. Check mismatch power at each load bus using specified and calculated values
to obtain ∆S = S spec − ∑Vi calc I ijcalc and stop if the value of ΔS ≤ ε.
9. Otherwise, go to step 3.
Note that the load flow techniques for transmission or distribution system are not sufficient for smart grid load flow. These methods can easily be implemented using (i)
sparsity techniques, (ii) implicit bus matrix, or (iii) computational techniques. In
36
PERFORMANCE ANALYSIS TOOLS FOR SMART GRID DESIGN
summary these load flow techniques are well documented in the literature. Software
package are available for demonstrating their tradeoffs and compatibility.
ATC is the maximum amount of additional MW transfer possible between two
parts of a power system. Additional means that existing transfers are considered part
of the base case and are not included in the ATC number. Typically, these two parts are
control areas or any group of power injections. Maximum refers to cases of either no
overloads occurring in the system as the transfer is increased or no overloads occurring
in the system during contingencies as the transfer is increased in online real time.
By definition, ATC is computed using the formula
ATC = TTC − ∑(CBM, TRM, and “existing TC”)
where the components are Total Transfer Capability (TTC), Capacity Benefit Margin
(CBM), Transmission Reliability Margin (TRM), and existing Transmission
Commitments.
ATC is particularly important because it signals the point where power system
reliability meets electricity market efficiency. It can have a huge impact on market
outcomes and system reliability, so the results of ATC tend to be of great interest to all
players.
The load flow is solved using the iterate solution on
∆d
−1 ∆P
∆V = [ J ] ∆Q
to obtain the incremental changes in the LHS vector until the terminating conditions
of the power mismatch (or maximum number of iterations) is reached. After convergence, derived quantities such as losses and power factors can be computed using
network equations.
A smart decoupled load flow equation could be developed based on the following
assumptions.
Let
δk − δm ≈ 0 such that cos(δk − δm) ≈ 1 and sin(δk − δm) ≈ 0
Vk ≈ 1
rkm << xkm, which implies gkm ≈ 0.
This allows a reduction of the Jacobian such that
∆d
∆V
can be solved without using an iterative method to solve simultaneous equations. The
approximation takes advantage of the lower sensitivities of real power with respect to
(w.r.t.) voltage magnitude and reactive power w.r.t. voltage arguments.
The DC load flow equations are simply the real part of the decoupled load flow
equations. This is achieved via the following additional assumptions: only angles and
real power in real time are solved for by iterating
37
CONGESTION MANAGEMENT EFFECT
−1
∆d = [ B′] ∆P where
N
∂P
∂P
∂P
= B′ with k =
bkm and k = −bkm .
∂δ k m =1
∂δ m
∂´
∑
m≠k
Recall the load flow that is solved using the iterate on
∆d
−1 ∆P
∆V = [ J ] ∆Q
until the terminating conditions of the power mismatch (or maximum number of iterations) is reached. The solution, via any of the three approaches summarized above, can
be used to compute the transfer of power from one node or set of nodes to another
within a power system network. Thus, incorporating online real-time data measurement
allows us to formulate the load flow for a smart grid network.
Role and Calculation of Power Transfer Distribution Factors (PTDFs). Power
Transfer Distribution Factors (PTDFs) measure the sensitivity of line MW flows to a
MW transfer. The line flows, from network theory, are simply functions of the voltages
and angles at its terminal buses—sending and receiving. Using the chain rule, the
PTDFs are represented as a function of these voltage and angle sensitivities.
The Single Linear Step ATC is extremely fast. Linearization is quite accurate in
modeling the impact of contingencies and transfers. However, it only uses derivatives
around the present operating point and does not model the control changes resulting
from ramping up to the transfer limit. Also, special devices such as phase shifters which
are generally used to redirect load flow across critical transmission interface can be
included.
The possibility of generators participating in the transfer hitting limits is not
modeled. Iterated Linear Step ATC considers these control changes.
3.4
CONGESTION MANAGEMENT EFFECT
Congestion is defined as a situation where load flows lead to a violation of operational
limits, that is, maximal thermal loading, voltage stability, or the n − 1 security rule. Net
Transmission Capacity (NTC) is the additional transaction potential between zones A
and B given a base scenario of injections.
In a meshed network, intrazonal injection variations will influence cross-border
load flows and are an important factor in the determination of transmission capacities.
These variations are a superposition of load flows on a given scenario of injections.
Accounting for the injection variations allows for a better estimation of the transfer
capacities, indifferent to the base case variations and creating thus more accurate longterm capacity index.
Due to the large number of linked equations that describe an actual electricity
network, numerical methods are used to determine load flows and node voltages. The
base load flow calculation method uses Newton–Raphson iterations, also called AC
load flow. The computational efforts can be relieved by using decoupled load flow
38
PERFORMANCE ANALYSIS TOOLS FOR SMART GRID DESIGN
equations. This assumes line impedances to be particularly inductive, which is an
acceptable assumption in the case of high voltage transmission lines. Going one step
further, a DC load flow can be used. In this method ohmic losses are ignored by stating
that lines are purely inductive. The injection of reactive power is assumed to be sufficient to keep the voltage profile in the network at a constant and nominal level. The
iterative algorithm of AC load flow equations is reduced to a system of linear equations.
These DC load flow equations will be used to simulate the effect of power injections
on load flows in the transmission network.
ISF1,1 ⋯ ISF1,k I nj1 flow1
⋮
⋱
⋮ ⋮ = ⋮
ISFn,1 ⋯ ISFn,k I njk flown
The injection in node i is generated power minus the load at node i. Injection Shift
Factor represents the proportion of the power injected in node k which flows in transmission line n. The load flows to a certain reference node which is not of importance
for the global system solution, but is used in the calculation of the ISFs and is necessary
for a proper understanding of these factors. The difference between two ISFs gives a
PTDF. A PTDF indicates the proportion of a transaction between two nodes that flows
across a certain line.
3.5
LOAD FLOW FOR SMART GRID DESIGN
Load flow tools that incorporate the stochastic and random study of the smart grid could
be modeled with the following implementation algorithm. Conditioning the load flow
topology will require a new methodology and algorithm that will include feeders and
the evolution of a time-dependent load flow. This method has been proven in terms of
characteristics and usage in power system planning and operation. Hence, the interoperability of RER with smart grid specifications could account for adequate use of current
methodology to perform analysis in both usual and alert states.
The implementation algorithm proposed will extend the following capability:
1. Model input of RER [2,5] and load will be changed to account for variability;
the input will have to include some power distribution flow so as to advance
the congested value of new estimate of Pf, Qg, and Pd, Qd. These attributes also
have a unique load appropriate effectiveness in the performance study.
2. Sparsity may be affected because the loads of RER may be widely distributed,
that is, load and size of RER has to be considered.
3. Computational challenges in new load flow with RER for smart grid that include
the stochastic model may affect the independent computation.
The load flow is also used in distributed networks (see Fig. 3.2).
ATC is the maximum amount of additional MW transfer possible between two
parts of a power system. Additional means that existing transfers are considered part
39
LOAD FLOW FOR SMART GRID DESIGN
Start Subroutine
For time, k
Read static network input and line data-switch
connectivity, layer info, line from to Z, Pload,
Qload, B, Vroot, Sroot
Is there
probabilistic load
data available?
Update input file with
probabilistic load data
Update input file with
deterministic source data
Are there available
PV or wind
resources?
Obtain probabilistic
parameters for PV/Wind
Assign flat voltages to all nodes v = 1 < 0
Update input file with
probabilistic source data
Set tolerance, ε iteration counter i = 0
Compute current injection at each node
Compute line current “backward”, current
Compute node voltage “forward”, voltage
Compute power mismatch
i=i+1
Check termination
criteria ∆S ≤ ε
Output: V, Θ, Sloss,
Ploss
Solve recursive
equations for injected
power P and Q for
network
Figure 3.2. Proposed load flow methodology.
Continue
40
PERFORMANCE ANALYSIS TOOLS FOR SMART GRID DESIGN
of the “base case” and are not included in the ATC number. Typically these two parts
are control areas or it can be any group of power injections. The term “maximum”
refers to cases of either no overloads occurring in the system as the transfer is increased
or no overloads occurring in the system during contingencies as the transfer is increased
in online real-time.
By definition, ATC is computed using the formula
ATC = TTC − ∑ (CBM, TRM, and “existing TC”)
where the components are Total Transfer Capability (TTC), Capacity Benefit
Margin (CBM), Transmission Reliability Margin (TRM), and “existing Transmission
Commitments”.
ATC is particularly important because it signals the point where power system
reliability meets electricity market efficiency. It can have a huge impact on market
outcomes and system reliability; so the results of ATC are of great interest to all
involved – power marketers, operators, planning engineers, and policy makers.
The load flow is solved using the iterate solution on
∆d
−1 ∆P
∆V = [ J ] ∆Q
to obtain the incremental changes in the LHS vector until the terminating conditions
of the power mismatch (or maximum number of iterations) is reached. After convergence, derived quantities such as losses and power factors can be computed using
network equations.
A smart decoupled load flow equation could be developed based on the following
assumptions.
Let
δ k − δ m ≈ 0 such that cos (δ k − δ m ) ≈ 1 and sin (δ k − δ m ) ≈ 0
Vk ≈ 1
rkm << xkm, which implies gkm ≈ 0.
This allows a reduction of the Jacobian such that
∆d
∆V
can be solved without using an iterative method to solve simultaneous equations. The
approximation takes advantage of the lower sensitivities of real power with respect to
(w.r.t.) voltage magnitude and reactive power w.r.t voltage arguments.
The “DC Load flow” equations are simply the real part of the decoupled load flow
equations. This is achieved via the following additional assumptions:
Only angles and real power in real time are solved for by iterating
41
DSOPF APPLICATION TO THE SMART GRID
−1
∆d = [ B′] ∆P where
N
∂P
∂P
∂P
= B′ with k =
bkm and k = −bkm .
∂δ k m =1
∂δ m
∂d
∑
m≠k
Recall the load flow that was solved using the iterate on
∆d
−1 ∆P
∆V = [ J ] ∆Q
until the terminating conditions of the power mismatch (or maximum number of iterations) is reached. The solution, via any of the three approaches summarized above, can
be used to compute the transfer of power from one node or set of nodes to another
within a power system network. Thus, by incorporating online real-time data measurement, we could formulate the load flow for a smart grid network.
3.5.1 Cases for the Development of Stochastic Dynamic Optimal
Power Flow (DSOPF)
DSOPF is being developed by the author. The DSOPF computational algorithm [8] has
the following built-in performance measures that are also defined for other general
purpose tools:
•
•
•
•
3.6
Controllability and interoperability: This is important for enabling different
devices, systems, and subsystems to provide greater observability when different
devices interact as agents for cooperation and benefits.
Reliability: quality measure of electricity delivered to achieve adequacy and
performance using intelligence tools, support devices, and software; ability to
achieve power quality and improve voltage profile is one of the attributes of the
smart grid
Adaptability and sustainability: ability of the grid to adapt to changes; meeting
energy needs in a way that can sustain life and civilization
Anticipatory behavior and affirmation of security: ability of the grid to anticipate
different scenarios and prepare to handle the dynamic changes while guaranteeing system security.
DSOPF APPLICATION TO THE SMART GRID
Adaptive Dynamic Programming (ADP) is a computational intelligence technique that
incorporates time Framework for Implementation of DSOPF [1,2].
There is a need for a generalized framework for solving the many classes of power
system problems where programmers, domain experts, and so on, can submit their
challenge problem. The collective knowledge will published and posted on the Web for
dissemination. Figure 3.3 shows the general framework for the application of ADP to
develop a new class of OPF problems called DSOPF; it is divided into three modules.
42
PERFORMANCE ANALYSIS TOOLS FOR SMART GRID DESIGN
Figure 3.3. Framework of ADP applications to power systems [3].
Module 1: Read power system parameters and obtain distribution function for state
estimation of measurement errors inherent in data; ascertain and improve accuracy of
data. Infer relationships between the past data and future of unknown period using time
series and dynamical systems; in all cases determine the time-dependent model approximation behavior of the generation data. Define the model with the uncertainties, including defining the problem objective and constraint functions for each problem.
Module 2: Determine the feasibility region of operation of the power system and
the emergency state with corresponding violations under different contingencies. Enumerate and schedule different control options over time for the different contingency
scenarios. Coordinate the controls and perform post optimizations of additional changes.
Evaluate results and perform sensitivity analysis studies.
Module 3: Address the postoptimization process through cost benefit analysis to
evaluate the various controls (cost effectiveness and efficiency). In the power system
parlance, a big network, which will perform this evaluation, is essential and indispens-
43
STATIC SECURITY ASSESSMENT (SSA) AND CONTINGENCIES
able. The critic network from ADP techniques will help realize the dual goals of cost
effectiveness and efficiency of the solution via the optimization process.
3.7
STATIC SECURITY ASSESSMENT (SSA) AND CONTINGENCIES
System security refers to the ability of the power system to withstand probable disturbance with minimal disruption of service. In an operational environment, security
assessment involves predicting the vulnerability of the system to possible disruptive
events in real time.
Actual operating conditions change constantly because of maintenance requirements, forced outages, and load patterns. The important options available to improve
upon an insecure condition include starting an available unit, rescheduling generation,
or asking for assistance from a neighboring system. The concept of DyLiacco’s securitystate diagram, shown in Figure 3.4, shows the principal operating states:
Transition due to
control action
Transition due to
a disturbanc
SECURE
NORMAL
STATE
Preventive
Control
Corrective
Control
INSECURE
NORMAL
STATE
Restorative
Control
RESTORATIVE
STATE
Cascade Events
EMERGENCY
STATE
Emergency Control
Figure 3.4. DyLiacco’s security-state diagram.
44
PERFORMANCE ANALYSIS TOOLS FOR SMART GRID DESIGN
1. Secure or normal state: all system loads are satisfied at the specified voltage
levels
2. Emergency state: some operating limits are violated, for example, overloaded
lines
3. Restorative state: some loads are not met, that is, partial or total blackout, but
the operating portion of the system is in a normal state.
Assume that a transmission line sustains a fault, which causes an outage. This will
result in a redistribution of power flows and changes in all voltages in the system. If
this redistribution results in a normal system condition then the predisturbance state
was both normal and secure relative to this event. If an emergency condition occurred,
then the predisturbance state was normal, but insecure. Similarly, a system in the emergency state can be forced to return to normal following some corrective control measures.
However, depending on the severity of the emergency, loads may be shed to alleviate
a more catastrophic situation leading to a partially normal system (i.e., restorative state).
System restoration involves activities to restore service to all interrupted loads.
Steady-state security involves situations where the transients following a disturbance have decayed, but where some limit violations could not be tolerated for long.
The loss of a transmission link, for example, after the transients have died out, may
result in an overloaded line, or an over-voltage condition. The system may tolerate such
limit violations for a short period. During such a period corrective action should be
taken. If corrective action is not possible, then the predisturbance state is seriously
insecure and some preventive measures should be carried out. Analysis tools required
address steady-state operation, that is, load flow and related analysis methods.
In Figure 3.5 a block diagram illustrates the functions associated with on-line
security analysis. The output of the state estimator can be used directly to determine
the security state (normal or emergency). For an emergency state, the next step is to
specify the required corrective action and apply it before it is too late. For a normal
state, it is not usually known if a postulated disturbance will or will not cause an emergency. As a result, contingency analysis is carried out using three data sources: the
precontingency state (state estimator output); a model of the external system; and a
specified list of contingencies. The results of contingency analysis are reevaluated
further to examine the level of system security. This is referred to as security analysis.
The outcome of security analysis will yield information on what to do subsequently. If
the system is deemed secure, then nothing is done till the next cycle of analysis (30
minutes or an hour later). If the system is deemed insecure, then preventive measures
are evaluated but not necessarily carried out. At this stage the system operator will
exercise some judgment over the desirability of preventive action since, invariably, it
may lead to less favorable operational economics.
3.8
CONTINGENCIES AND THEIR CLASSIFICATION
Steady-state contingency analysis predicts power flows and bus voltage conditions
following events such as line outages, transformer outages, and generator outages.
45
CONTINGENCIES AND THEIR CLASSIFICATION
EXTERNAL
SYSTEMS
Interties
Power System
Under Control
Disturbances
SCADA
(On-Line Data)
Network
and
Other Data
Specify
Security State
State
Estimation
Emergency
Specify
Security State
Specify
Security
Specify Preventive
Measures. Stay Alert, or,
Implement Measures
Specify
Security State
Normal
Contingency
Analysis
No
Contingency
Analysis
Stop
Yes
Is
System
Secure?
Security
Analysis
Figure 3.5. System security and associated functions.
There are many reasons for transmission line and transformer outages. The simplest
reason is planned maintenance. Another involves switching operations to control power
flows in the network and/or to overcome voltage problems. In either case, the outages
are caused by operators performing daily dispatching and maintenance functions.
Another class of outages is called forced outages, for example, a line that has
experienced a permanent fault is automatically de-energized by circuit breakers, or an
46
PERFORMANCE ANALYSIS TOOLS FOR SMART GRID DESIGN
overloaded line or transformer is de-energized to protect it from damage. In either
case, the operator needs to know the effects of the outage on power flows and voltage
conditions throughout the system in order to take preventive measures before the
outages occur. In the case of planned maintenance, the operator will require a forecasted load flow case at the time of planned outage. In the case of switching operations
for flow and voltage control, the operator will require a load flow solution for the
present preoutage condition.
Forced outage cases are more complicated since they occur with very low probabilities that are time- and weather-dependent. Usually, single-line outages are more
probable than double or multiple outages, but double and multiple outages occasionally
occur during severe weather. As a result of the large amount of computations involved,
only single-line contingencies are considered. Generator outages occur for reasons
similar to those of line and transformer outages. Power plants are taken off-line for
operational and maintenance requirements. There are also forced generator outages
caused by equipment failures, line faults near the generator, and so on.
3.8.1
Steady-State Contingency Analysis
Steady-state contingency analysis predicts power flows and bus voltage conditions following transmission line outages, transformer outages, and generator outages. Models
of Contingency Evaluation
In transmission line or transformer outage contingencies it is assumed that input/
demand load flow variables will not change due to the outage. Therefore, specified real
and reactive loads, together with real generation and generator bus voltages, will be
constant before and after the outage. Clearly, this line-outage model is only approximate. The loss of a major transmission line will cause changes in voltage and power
flow conditions. The first consequence is that power system losses will change. In the
above model, these losses will be accounted for by a changed slack-bus generation
level. Changes in voltage conditions at load busses normally mean changes in the load
itself, especially when it is represented as an impedance load. Only with an adequate
load model for each bus can the accuracy of contingency analysis be improved. For
most purposes, the errors introduced by the above modeling assumption are fewer than
those resulting from inaccuracies in the input data.
The case of a generator outage is more complicated for a number of reasons.
Immediately following the outage (first few seconds), the remainder of the generating
system will be unable to respond to the resulting generation/load unbalance by increasing its generation level. System frequency will drop, with the net effect being an overall
and fairly uniform reduction in system load because loads are frequency-dependent and
tending to decrease or increase with corresponding decreases and increases in frequency. Because of this drop in frequency, as well as the serious violation of the
requirements on scheduled net power flow interchanges with neighboring systems,
generation levels of various generators will be automatically controlled to restore both
normal system frequency and net interchanges with neighboring systems through automatic generation control (AGC) within a few minutes. The use of economic dispatch
allows the desired optimal generation levels to be accomplished, although they may
47
CONTINGENCIES AND THEIR CLASSIFICATION
not be economically and/or environmentally optimal. This final steady-state condition
is reached in several minutes after the disturbance.
In practice, many smart grid contingencies will require the same number of fast
load flow techniques for performing contingency analysis. In all of these techniques,
the starting point is that of the preoutage solution. Some of these techniques are discussed below.
3.8.2
Performance Indices
A security-type performance index ranks the severity of various contingencies. The
following index is an illustration:
J=
1
2
∑ (V − V
k −ref
k
)2 Wk
k
where Vk is the magnitude of the voltage at bus, Vk−ref is the nominal (usually rated)
voltage at bus k, and Wk is the weighting number associated with the relative importance, or the allowable range of voltage variation, at bus k. For a precontingency condition, J will take the value of J0. For the ith contingency J will take the value of J1. The
contingencies are then ranked according to the J1 values. Because of the quadratic
nature of J1, more severe contingencies will result in higher J1 values.
A similar performance index can be formulated for line power flow limit violationsfor example,:
J′ =
1
2
∑W T
k
k
Tk
k max
2
where Tk is the real power flow on line k, Tk max is the corresponding maximum power
flow limit, and Wk is a weighting factor.
External System Equivalents. In the online control and operational context,
the system being controlled is usually interconnected to other systems. Normally, a
contingency in one system will have the highest repercussions within that system. There
are always cases, however, where a contingency in one system is strongly felt in
another, for example, the loss of a major generating unit may cause power flow limit
violations elsewhere. The difficulty in predicting the impact of a contingency arises
from the fact that the external network is not monitored as carefully as the internal
network. Through state estimation, all internal system voltage magnitudes and angles,
power flows, generations, load, and network topology are known online. As for the
external system, online information is normally restricted to items such as intertie
power flows, status of major lines and generators, and possibly individual unit outputs.
When searching for an exact power flow solution to a postulated contingency, the
state of the entire network (internal and external) should be known to establish the
precontingency base case. Since the state of the external network is not fully known,
some approximations are required. In this section, two types of approximations are
48
PERFORMANCE ANALYSIS TOOLS FOR SMART GRID DESIGN
considered. The first is based on sensitivity approaches, and the second on network
reduction approaches. These will be illustrated, following some important definitions
of terms.
The starting point is the so-called base case load flow solution. In a planning study
this can be a complete network solution for a given set of loads, generation, and network
configuration. In the case of online operations, it can be a solution of the internal plus
the boundary system obtained from on-line state estimation using live measurement data.
In performing studies on the internal system, it is possible to deviate from the base
case to account for line outage or generator contingencies. Normally, the existing classification of boundary nodes are retained, but there are instances when a load boundary
bus may be reclassified as a generation bus. For example, if the boundary bus is close
to a strong external system generation center, its voltage level will be maintained at a
constant value, or nearly so, because of the external influence. In fact, some operators
have reclassified all boundary load busses as generation busses, claiming higher accuracy of solutions.
The steps in the utilization of network equivalents are:
1.
2.
3.
4.
5.
3.8.3
Perform network reduction using an appropriate technique
Given the base case (preoutage) solution, compute boundary bus injections
Reclassify some boundary busses as generation busses if necessary
Postulate a contingency list
For each contingency in the list, solve the load flow problem using a fast contingency evaluation technique that is initiated by the base case solution
Sensitivity-Based Approaches
As shown above, the external equivalent network depends on real and reactive external
bus injections and voltage magnitudes. Following a line outage or generator contingency, the external voltages, and possibly some injections will be different. This means
that a postoutage equivalent is required. Obviously, it defeats the entire purpose of
network reduction. A more crucial problem is that external system injections and voltages are actually unknown in an online environment..
There are two possible solution paths. In the first path, during offline studies, it is
important to obtain equivalent networks that are insensitive to external conditions and
to internal outages. If this objective is achieved, the next step uses online measurements
to calibrate the parametric values of equivalent lines and injections.
3.9
CONTINGENCY STUDIES FOR THE SMART GRID
Contingency studies are proposed planning/operation tool for assessing the impact of
unit or line outage in an integrated smart grid environment. This could be a single or
multiple line outages called N − 1, N − 2, N − 3, . . . contingency. There are two types
of contingency:
SUMMARY
49
1. AC automatic contingency screening/filtering
2. AC automatic contingency control
Use of the contingency set includes:
1. These contingencies sets are used in SSA as well as in DSA. The SSA consists
of the following elements.
a. Load flow base case
b. Schedule contingency
c. Develop and model a PI for each contingency
2. Provide the studies for violation check against a given threshold, PI ≤ Threshold
for violated contingency case in ascending order
3. Develop security measure to aid improvement in recommendation and display
SSA is well known for classical power system. The weak points are: includes
1.
2.
3.
4.
Selection of weighting measure
Exponent factors
The probability of selecting a contingency
Lack of human intelligence and relative information on knowledge-base for
decision-making by nonexperts. Work by the author has been planned using ES,
ANN, and fuzzy sets. These schemes do not include time against sources, variability in selecting parameters for contingency studies.
The author proposes an ACS capable of including:
1.
2.
3.
4.
RER with variability in output power as contingency
Loss of measured unit needed for performance study
Probabilistic measure of contingency to be added to the input
The resulting PI index needs to be replaced with time stamp measurements using
PI- OSIsoft to predict, analyze, and view recommendations by classified behavior and impact of contingency under study in real time, hence creating a new
opportunity in research.
There are many opportunities for designing new computational algorithms to support
the deployment of a smart grid equipped with advanced tools for system security, reliability, sustainability, and affordability.
3.10
SUMMARY
This chapter has discussed several needed tools for analysis of smart grid design, operation, and performance. Such tools include load flow, optimal power flow, and static
security assessments and contingencies. Reviews of classical methods were presented
along with details about each solution tool’s incorporation into the smart grid.
50
PERFORMANCE ANALYSIS TOOLS FOR SMART GRID DESIGN
REFERENCES
[1] M.H. Mickle and T.W. Sze. Optimization in Systems Engineering. Scranton, 1972.
[2] M.E. El-Hawary. Electrical Power Systems Design and Analysis. IEEE Press Series on
Power Engineering, 2008.
[3] J.A. Momoh. Electric Power System Application of Optimization. Marcel Dekker, New
York, 2001.
[4] J.L. Marinho and B. Stott. “Linear Programming for Power System Network Security Applications.” IEEE Transactions on Power Apparatus and Systems 1979, PAS-98837–848.
[5] R.C. Eberhart and J. Kennedy. “A New Optimizer Using Particle Swarm Theory.” In Proceedings on the Sixth International Symposium on Micromachine and Human Science, 1995,
39–31.
[6] A.G. Barto, W.B. Powell, D.C. Wunsch, and J. Si. Handbook of Learning and Approximate
Dynamic Programming. IEEE Press Series on Computational Intelligence, 2004.
[7] J. Momoh. Electrical Power System Applications of Optimization. CRC Press, 2008.
[8] L. Zhao and A. Abur. “Multiarea State Estimation Using Synchronized Phasor Measurements.” IEEE Transactions on Power Systems 2005, 20, 611–617.
[9] “Appendix B2: A Systems View of the Modern Grid-Sensing and Measurement.” National
Energy Technology Laboratory, 2007.
[10] T. Bottorff. “PG&E Smart Meter: Smart Meter Program,” NARUC Summer Meeting 2007.
[11] D. Zhengchun, N. Zhenyong, and F. Wanliang. “Block QR decomposition based Power
System State Estimation Algorithm.” ScienceDirect 2005.
[12] M.S. Srinivas. “Distribution Load Flows: A Brief Review.” IEEE Power Engineering Winter
Meeting 2000. Vol. 2, pp. 942–945. August 2002.
SUGGESTED READINGS
M. Dorigo and T. Stutzle. “The Ant Colony Optimization Metaheuristic: Algorithms, Applications
and Advances.” In F. Glover and G. Kochenberger, eds.: Handboook of Metaheuristics.
Norwell, MA, Kluwer, 2002.
M. Dorigo and T. Stuzle. Ant Colony Optimization. Massachusetts Institute of Technology, Cambridge, 2004.
A. Englebrecht. Computational Intelligence: An Introduction. John Wiley & Sons, 2007.
B. Milosevic and M. Begovic. “Voltage-Stability Protection and Control Using a Wide-Area Network
of Phasor Measurements.” IEEE Transactions on Power Systems 2003, 18, 121–127.
A.G. Phadhke. “Synchronized Phasor Measurments in Power Systems.” IEEE Computer Applications in Power 1993, 6, 10–15.
G. Riley and J. Giarratano. Expert Systems: Principles and Programming. PWS Publisher,
Boston, 2003.
P.K. Skula and K. Deb. “On Finding Multiple Pareto-Optimal Solutions Using Classical and
Evolutional Generating Methods.” European Journal of Operational Research 2007, 181,
1630–1652.
C.W. Taylor. “The Future in On-Line Security Assessment and Wide-Area Stability Control.”
IEEE Power Engineering Society 2000, 1, 78–83.
W.H. Zhange and T. Gao. “A Min-Max Method with Adaptive Weightings for Uniformly Spaced
Pareto Optimum Points.” Computers and Structures 2006, 84, 1760–1769.
4
STABILITY ANALYSIS TOOLS
FOR SMART GRID
4.1
INTRODUCTION TO STABILITY
As electric power networks worldwide expand to accommodate more generation, RER,
and control devices, the physical and technical consequences resulting from weaknesses
in generation, transmission, and distribution will become more costly to society and
the environment. To date, analysis tools that provide operational and planning guidance
have been limited to studying systems with static and dynamic models of generators,
busses, controls (exciters and governors), and FACTs devices. Little work has been
undertaken to upgrade these tools to include predictability and adaptiveness, and to
provide solutions for managing the power system of the future. Thus, this chapter first
reviews the extant research in stability assessment, and follows with a proposed framework for the design of the new tools needed for smart grid analysis and design. The
concept and techniques used in voltage stability assessment are described next.
4.2 STRENGTHS AND WEAKNESSES OF EXISTING VOLTAGE
STABILITY ANALYSIS TOOLS
The characteristics listed here are inherent in the analytic tools for the smart grid, yet
are not incorporated in the analytic tools for the existing power system network:
Smart Grid: Fundamentals of Design and Analysis, First Edition. James Momoh.
© 2012 Institute of Electrical and Electronics Engineers. Published 2012 by John Wiley & Sons, Inc.
51
52
STABILITY ANALYSIS TOOLS FOR SMART GRID
•
•
•
•
•
•
Robustness: Persistence of a system’s characteristic behavior under perturbations
or conditions of uncertainty.
Scalability: Ability of a system, network, or process to accommodate growing
amounts of work; ability to be enlarged to accommodate growth.
Stochasticity: Time development (be it deterministic or essentially probabilistic)
that is analyzable in terms of probability.
Predictivity: Rigorous, often quantitative, forecast of what will occur under
specific conditions.
Adaptability: System can adapt its behavior according to changes in its environment or in parts of the system itself.
Online real-time data acquisition: Instantaneous acquisition of data.
Other suitable tools and techniques include:
WAM techniques: measurements of voltage, angle, frequency, control series, and
available resources for load state conditions; data is usually assumed or computed for
static model with advent of GPS importance; new advances in the design of PMUs,
smart meters, state estimation (SE), and FIDR monitor and control data for assessing
stability, mitigating volatility, and achieving high-order efficiency and reliability.
Phasor Measurement Unit: PMUs are high-speed, time-synchronized digital
recorders that measure voltage, current, and frequency on the transmission system, and
calculate voltage and current magnitudes, phase angles, and real and reactive power
flows. PMU data can be applied to the following:
1.
2.
3.
4.
Asset management
Voltage stability
Angle stability assessment
Designing optimum controls
Smart Meters: Two-way electronic communication meter or other device measuring electricity, natural gas, or water consumption.
Similar meters, usually referred to as interval or time-of-use meters, have existed
for years, but smart meters usually involve a real-time or near real-time sensors, power
outage notification, and power quality monitoring. These additional features are more
than simple automated meter reading (AMR). They are similar in many respects to
Advanced Metering Infrastructure (AMI) meters. Smart meters are also believed to be
a less costly alternative to traditional interval or time-of-use meters and are intended
to be used on a wide scale with all customer classes, including residential customers.
Interval and time-of-use meters are more of a legacy technology that historically have
been installed to measure commercial and industrial customers, but typically provide
no AMR functionality. Smart meters may be part of a smart grid, but alone do not
constitute a smart grid.
Design of Architecture for Smart Grid with RER: The legacy system is built around
central generation power sources and static load models. We propose a new expandable
model schematic of a smart grid with an RER single line diagram.
STRENGTHS AND WEAKNESSES OF EXISTING VOLTAGE STABILITY ANALYSIS TOOLS
53
Defining stability methods using these conditions requires research efforts that
encompass:
1. Voltage stability: Use of probability distribution function to model the resources
(wind, solar, and so on) that will provide the average power output included in
the network model; load models are measured using stochastic, time series
methods; line voltages are provided with some probabilistic measure to assure
their capture. The steps are:
2. Determine the operating point for stability assessment by performing a probabilistic load flow which starts from the base case of load flow for prime loading
condition
3. Define or rank contingency based on model of choice for which selection
methods include:
a) Real-time usage of ant colony
b) ANN
c) Mixture to increase the selection of condition
d) Perform the screening evaluation using one of the new index methods
e) Determine distance from voltage collapse
These points include the collaboration of model and sensitivity analysis methods to
determine the equilibrium point and to incorporate control measures to mitigate stability
methods for the different scenarios of events studied.
To enhance the algorithm for smart grid, we further include a real time voltage
index which is based on phasor measurement data as a means of state estimation. This
involves a modification of the step above, that is, computation of the new state of the
system with respect to RER and use of the data from base case load flow for different
time scales to compute new voltage stability based on RER data.
The results of stability condition here require a stochastic central approach compared with the classical method of mitigation, unlike offline study with the improved
system for enhancing dispatching in study mode. It can be run at fixed intervals as it
can provide the most recent estimate of system conditions.
It can also automatically trigger computational engine of fast steady state. The
output or input data use different scenarios which are available for developing visualization tools. The revised stability tool for smart grid will form the basis of new software
for next generation of SCADA or EMS systems.
The stability result margin concept is a simple way to determine the current or
actual system condition from the critical state in which a source perturbation can cause
instability in the smart grid environment. This will be important so that communication
infrastructure can account for necessary corrective and preventive actions. The indicators generally used are:
Stability power flow = 100% ∗ (Pmax − Pbase ) / Pbase
Stability rescue voltage = 100% ∗ (Vmax − Vbase ) / Vbase
54
STABILITY ANALYSIS TOOLS FOR SMART GRID
TABLE 4.1. Old and New Grid Methodology
Methodology
Model load
Resources
FACTS devices and controls
Risk management
Protection platform
Old Grid
New Grid
Static
Deterministic
Specified
Deterministic
Defined
Dynamic
Stochastic
Adaptive
Random
Adaptive
Load base Voltage Stability Communication Point (VSCP): Based on required successive load changes and delivery of voltage desired to determine the MW/MVAr that
could lead to voltage collapse
These techniques could be used to solve the classical voltage stability of old networks. Because of the new features, as discussed in Table 4.1, we need to define the
tools needed to achieve this ultimate simulation platform and challenges.
The various indications of instability to be determined are meant to rank sensitivity
of stability. Simply, they measure maximum transfer capability across the interfacing
region in an interconnected transmission line. This is particularly useful for the smart
grid where distributed resources, RER, and controls are widely spread.
Transient stability studies are needed in this case and optimization of RER and
other resources to ensure that stability is derived. We propose the algorithm in Figure
4.1 and enumerate the steps to be taken in realizing this:
1. Analyze and define the system configuration with all networks in service
2. Simulate faults such as single-line-to-ground, three phase, or server contingency
study
3. Evaluate the impact of contingency for branch, unit outage with voltage stability
criteria
4. Perform stability analysis for voltage and angle.
The nonlinear system model used for transient stability analysis describes the system
by a set of differential equations and a set of algebraic equations. Generally, the differential equations are machine equations, control system equations, and so on, and the
algebraic equations are system voltage equations involving the network admittance
matrix. The time simulation method and direct method are often used for transient
stability analysis. The first method determines transient stability by solving the system
differential equation step by step, while the second method determines the system
transient stability without explicitly solving the system differential equations. This
approach is appealing and has received considerable attention. Energy-based methods
are a special case of the more general Lyapunov’s second or direct method, the energy
function being the possible Lyapunov function.
The challenges of implementation are:
1. They do not include time stamp of load change
2. They do not account for real-time information on system topology and devices
contributing to system changes
STRENGTHS AND WEAKNESSES OF EXISTING VOLTAGE STABILITY ANALYSIS TOOLS
55
Start
Input of Data
Options available for obtaining real time
data based on PMU and State Estimation
Sensitivity studies
Is VSM
Violated?
No
Yes
Display
System State
and VSM
Determination of type and location of violation
Stop
Utilize OPF subject to voltage security
constraint
Re-compute the improvement to the voltage stability
index
Carry out necessary control action/corrective measures to the input file and rerun the load flow
Figure 4.1. Flowchart of voltage stability and PMU that enhances the current grid.
56
STABILITY ANALYSIS TOOLS FOR SMART GRID
Start
Configuration of System
Faults Simulation
Is the system
power smart?
Contingencies Studies
System Stability Analysis
Figure 4.2. Voltage stability study algorithm.
3. They do not inform the decision-maker of real-time status (location of system
parameters)
We propose that future and ongoing algorithms should include:
1. SE or PMU data filtering so that appropriate data on system parameters
2. A practical index for voltage stability based on input data or estimated system
parameters
3. Assessment (preventive/corrective) scheme to guide the decision-maker
4. OPF formulation to overcome the collapse or instability
Note that this scheme is useful for both bulk power systems and distribution networks
which require stability.
4.3
VOLTAGE STABILITY ASSESSMENT
Voltage stability has many definitions. It is a fast phenomenon for engineers involved
with the operation of induction motors, air conditioning loads, or HVDC links, while
it is a slow phenomenon (involving, for example, mechanical tap changing) for others.
Voltage instability or collapse is a dynamic process. The term stability as used
VOLTAGE STABILITY ASSESSMENT
57
here implies that a dynamic system is being discussed. A power system is a dynamic
system. In contrast to rotor angle (synchronous) stability, the dynamics involve mainly
the loads and the means for voltage control. Voltage stability is alternatively called load
stability.
The loss of lines or generators can sometimes cause voltage quality degradation,
a phenomenon that is attributed to lack of sufficient reactive reserve when the power
system experiences a heavy load or a severe contingency. Thus, voltage instability is
characterized in such a way that voltage magnitude of the power system decreases
gradually and then rapidly as the system nears the collapsing point. Voltage stability is
classified as either static voltage stability or dynamic voltage stability. The latter is
further classified into small signal stability and large disturbance stability problems.
In dynamic voltage stability analysis, exact models of transformers, SVCs, induction motors, and other types of loads are usually included in problem formulations in
addition to models of generators, exciters, and other controllers. Small signal voltage
stability problems are formulated as a combination of differential and algebraic equations that are linearized about an equilibrium point. eigenvalue analysis methods are
used to analyze system dynamic behavior. Small signal analysis can provide useful
information on modes of voltage instability and is instructive in locating VAR compensations and designing controllers. Large disturbance voltage stability is mainly
approached using numerical simulation techniques. Since system dynamics are described
by nonlinear differential and algebraic equations that cannot be linearized in nature, the
mechanism of voltage collapse may be explained as a saddle node bifurcation. Voltage
collapse is analyzed based on a center manifold voltage collapse model.
4.3.1
Voltage Stability and Voltage Collapse
Voltage stability has often been viewed as a steady-state viability problem suitable for
static analysis techniques. The ability to transfer reactive power from production
sources to consumption sinks during steady operating conditions is a major aspect of
voltage stability. A power system in a given operating state and subject to a given
disturbance undergoes voltage collapse if postdisturbance equilibrium voltages are
prespecified acceptable limits in a significant part of the system. Voltage collapse may
be total (blackout) or partial.
Assume that a power system undergoes a sudden increase of reactive power
demand following a system contingency; the additional demand is met by the reactive
power reserves carried by the generators and compensators. Because of a combination
of events and system conditions it is possible that the additional reactive power demand
may lead to voltage collapse in part or all of the system. A typical sequence of events
leading to a voltage collapse is:
•
•
The power system is experiencing abnormal operating conditions with large
generating units near the load centers being out of service. Some EHV lines are
heavily loaded and reactive power resources are low.
A heavily loaded line is lost which causes additional loading on the remaining
adjacent lines. This increases the reactive power losses in the lines (Q absorbed
58
STABILITY ANALYSIS TOOLS FOR SMART GRID
•
by a line increases rapidly for loads above surge impedance loading), causing a
heavy reactive power demand on the system.
Immediately following the loss of the line, a considerable voltage reduction
occurs at adjacent load centers due to the extra reactive power demand. This
causes a load reduction, and the resulting reduction in power flow through the
lines should have a stabilizing effect. However, the generator AVRs quickly
resolve terminal voltages by increasing excitation. The resulting additional reactive power flow through the inductances associated with generator transformers
and lines increases the voltage drop across each of these elements.
At this stage, generators are likely to be within the limits of P-Q output capabilities,
that is, within the armature and field current heating limits. The speed governors regulate the frequency by reducing MW output.
•
•
The EHV level voltage reduction at load centers is reflected in the distribution
system. The under load tap changers (ULTCs) of substation transformers restore
distribution voltages and loads to prefault levels in about 2 to 4 minutes. With
each tap change operation, the resulting increment in load on EHV lines increases
the line XI 2 and RI 2 losses, which in turn cause greater drops in the EHV lines.
If an EHV line is loaded considerably above the SIL, each MVA increase in line
flow would cause several MVArs of line losses.
With each tap-changing operation, the reactive output of generators throughout
the system increases. One by one, generators gradually exceed their reactive
power capability limits (imposed by maximum allowable continuous field
current). When the first generator reaches its field current limit, its terminal
voltage drops. At the reduced terminal voltage for a fixed MW output, the armature current increases, and might further limit reactive output to keep the armature current within allowable limits. Its share of reactive loading transfers to other
generators and causes more overloading. With fewer generators on automatic
excitation control, the system is prone to voltage instability that is likely compounded by the reduced effectiveness of shunt compensators at low voltages.
Eventually, there is voltage collapse or cascading, possibly leading to loss of
generation synchronism.
Voltage security is the ability of a system to operate in a stable mode and to remain
stable following credible contingencies or load increases. It often means that the existence of considerable margin from an operating point to the voltage instability point
involves credible contingencies.
4.3.2
Classification of Voltage Stability
The three categories of voltage problems are:
1. Primary phenomena related to system structure: Reflect the autonomous response
of the system to reactive supply/demand imbalances.
59
VOLTAGE STABILITY ASSESSMENT
2. Secondary phenomena related to control actions: Reflect the counterproductive
nature of some manual or automatic control actions.
3. Tertiary phenomena resulting from interaction of the above.
This classification implies that the problems involve both static and dynamic aspects
of system components. Voltage collapse dynamics span a range in time from a fraction
of a second to tens of minutes. Time frame charts are used to describe dynamic phenomena which show time responses from equipment that may affect voltage stability.
The time frames could be considered to become very fast transient, transient, and long
term. The main characteristics of the three time frames are:
1. Very fast transient voltage collapse: Involves network RLC components having
very fast response; the time range is from microseconds to milliseconds.
2. Transient voltage collapse: Involves a large disturbance and loads having a rapid
response; motor dynamics following a fault are often the main concern; the time
frame is one to several seconds.
3. Long-term voltage collapse: Involving a load increase or a power transfer
increase; shows load restoration by tap-changer and generator current limiting;
manual actions by system operators may be important; the time frame is usually
0.5 to 30 minutes.
Since voltage stability is affected by various system components in a wide time range,
in order to tackle this problem, one must consider proper modeling and analysis
methods. Currently, voltage stability approaches mainly include static and dynamic,
that is, transient voltage collapse and long-term voltage collapse.
4.3.3
Static Stability (Type I Instability)
Suppose that ∆xɺ S = ∆xɺ F = 0. Then we have a static situation with all equations being algebraic. Let all the voltage deviations in ∆V̂g and ∆V̂l be denoted by ∆V̂, then the rest of
the algebraic variables can be eliminated (assuming constant power load) to express
∆V̂ = J s−1 H∆pL . If det(Js) → 0 as load is increased it is called a Type I static instability,
that is, the system cannot handle increased load.
4.3.4
Dynamic Stability (Type II Instability)
Eliminating the algebraic variables and assuming Δu ≡ 0,
∆xɺ S ASS
∆xɺ = A
F FS
ASF ∆xS
+ [ H1 ] ∆pL
AFF ∆xF
Two types of dynamic instability (Type II) can be distinguished as slow and fast. In
both cases assume ΔpL = 0.
60
STABILITY ANALYSIS TOOLS FOR SMART GRID
Slow Instability. Theoretically it should be possible to eliminate ΔxF using singular perturbation theory and obtain the linearized slow system as ∆xɺ S = AS ∆xS . The
time scale of the phenomenon is so large that the linearized results may not reflect the
true picture. For such long time phenomena, a nonlinear simulation is recommended.
Fast Instability. First, we rearrange the variables ∆I g , ∆Vˆg , ∆Vˆl as [ΔIg, θ1,
ΔV1, . . . , ΔVm|Δθ2, Δθ3, . . . , Δθn, ΔVm+1, . . . , ΔVn] = [Δz, Δv]. Next, we assume xS as
constant and load parameters as constant which implies ΔpL = 0, to get
∆xɺ F A1 A2
0 = B B
2
1
0 C1 C2
4.3.5
A3 ∆xF 0 E
B3 ∆z + ∆S1 + 0 ∆u
C3 ∆v ∆S2 0
Analysis Techniques for Dynamic Voltage Stability Studies
It is only recently that the effects of system and load dynamics have begun to be investigated in the context of voltage collapse. The dynamics considered are:
1. Machine and excitation system dynamics including power system stabilizer
(PSS)
2. Load dynamics
3. Dynamics of SVC controls and FACTS devices
4. Tap-changer dynamics
5. Dynamics due to load frequency control, AGC, and so on.
While 1, 2, and 3 involve fast dynamics, 4 and 5 represent the slow dynamics. A nice
classification process of dynamic voltage stability vis-à-vis static stability is shown in
Figure 4.3, where oad implies demand; u represents the set points of LFC, AGC and
voltage/var controls at substations; xS represents the slow variables such as the state
variables belonging to tap-changing transformers, AGC loop and center of angle variables in the case of a multi-area representation; and xF represents the fast variables
including PSS and governor, induction motor load dynamics, SVC dynamics, and so
on, belonging to the generating unit. The overall mathematical model takes the form:
xɺ S = gs (Vˆ , u )
xɺ F = gF (xS , xF ,Vˆ , y, I g ,u, pL )
0 = hF (xS , xF ,Vˆ , I g ,u, pL )
y = hN ( xS , xF ,Vˆ ,u )
Ignoring the slower AGC dynamics and the faster network transients (60 Hz), we can
categorize the variables such that
61
VOLTAGE STABILITY ASSESSMENT
Power System
xS=fS(xS, xF, u, Load)
xF=fF(xF, xS, u, Load)
Subsystem S:
xS=fS(x , x , u, Load)
0=fF(xS, xF, u, Load)
Stable?
N
Subsystem F:
xF=fF(xS, xF, u, Load)
(Voltage Collapse) N
Type II
Instability
Stable?
Y
Y
Both Subsystems S and F are Stable
Small
(Normal State)
Stable
Change
in xS?
Large (Voltage Drop)
Type I
Instability
Figure 4.3. Classification of voltage instabilities.
xS = [ ni ]
i = 1, … , p
where
ni = transformer tap ratio
δi
rotor angle of i th machine
ω
i
rotor velocity of i th machine
Eqi′
xF = Edi′
rotor electrical variables i = 1, … , m
E ′fdi
VRi
excitation system variables
R fi
θi
VDi
V̂i = Rectangular variables of ith bus voltage or
Vi
VQi
PGi + PLi
y=
Injected real and reactive power i = 1, 2, . . . , n
QGi + QLi
I di
I gi = Machine terminal currents in machine reference frame i = 1, . . . , m
I gi
62
STABILITY ANALYSIS TOOLS FOR SMART GRID
desired real power of i th generator
pMi
i = 1, 2, … , m
u = Vref ,i
desired voltage at i th generator bus
i = 1, 2, … , p
Voi desired voltage at the bus controlled by tap − changer i
pL = Vector of load parameters
The state variables of the static VAr system (SVC) control and induction motor
will appear in xF if included in the overall model. The example below gives the equations for an m machine n bus system having p tap-changing transformers. Only the
synchronous machine and tap-changer dynamics are included.
4.4
VOLTAGE STABILITY ASSESSMENT TECHNIQUES
Now, we discuss the selected voltage stability indices, the VIPI method, the method
based on singular value decomposition, condition number of the Jacobian matrix, and
the method based on the energy margin. Figure 4.4 depicts the techniques.
VIPI Method. VIPI was developed by Tamura et al. [12] based on the concept
of multiple load flow solutions. According to Reference 12, a pair of load flow solutions
x1 and x2 are represented by two vectors a and b as follows:
x1 = a + b
x2 = a − b
and are equivalent to:
a=
x1 + x2
2
b=
x1 − x2
2
Voltage Stability Assessment
Techniques
VIPI
Singular value
decomposition
Bifurcation
Continuation
Power Flow
Figure 4.4. Voltage stability assessment techniques.
Energy
Function
63
VOLTAGE STABILITY ASSESSMENT TECHNIQUES
Y(b)
Y(a)
b
×1
Ys
-b
a
×2
θ
Figure 4.5. Concept of VIPI in the node specification space.
where x1 is the normal (high) power flow solution, x2 its corresponding low voltage
power flow solution, a is a singular vector in the space of node voltages, and b is a
margin vector in the same space.
Define two other vectors, Ys, and Y(a), called singular vectors in the space of node
specifications. The relationship between these vectors is shown in Figure 4.5.
VIPI is defined by the following equation:
Y T ×Y (a)
VIPI = cos−1 s
Ys × Y ( a )
where vector Ys consists of bus injections computed with respect to x1, but the injection
values corresponding to the reactive powers of PV busses are replaced by the squared
values of voltage magnitudes. Y(a) consists of bus injections with respect to vector a:
||χ|| is the l2 norm of vector x.
The computation of VIPI is easy once the relevant low voltage power flow solutions are obtained. Exhaustive and simplified methods of locating low voltage power
flow solutions of a power system are presented in References 1 and 5, respectively.
Generally speaking, finding all the relevant low voltage solutions is a computational
burden for practical-sized systems.
Minimum Singular Value. When an operating state approaches the collapse
point, the Jacobian matrix of the power flow equations (J) approaches singularityf. The
minimum singular value of the Jacobian matrix expresses the closeness of Jacobian
singularity. Singular value decomposition method is used to solve the minimum singular value for static voltage stability analysis.
According to the theory of singular value decomposition, the Jacobian matrix of
the power flow can be decomposed as:
J =U
∑V
T
where J ∈ R2n×2n is the Jacobian matrix of the power flow and Σ = Diag (σ1, σ2, . . . ,
σn) with σmax = σ1 ≥ σ2 ≥ . . . ≥ σn = σmin ≥ 0.
64
STABILITY ANALYSIS TOOLS FOR SMART GRID
If matrix J has rank r (r ≤ 2n), its singular values are the square roots of the r
positive eigenvalues of ATA (or AAT), U and V are orthonormal matrices of order 2n,
and their columns contain the eigenvectors of AAT and ATA respectively. Thus:
AVi = σ i ui
AT ui = σ iVi
Next, define
E j = u jV jT
This equation can be written as:
J = σ 1E1 + σ 2 E2 + ⋯ + σ n En
If we allow
J ′ = σ 1E1 + σ 2 E2 + ⋯ + σ n−1En−1
Concerning the l2 norm of the J matrix, J′ is a matrix of rank n−1 nearest to the J matrix
of rank n, that is, the smallest singular value of a matrix is a measure of the distance
between matrices J and J′. The minimum singular value for the power flow equations
expresses the proximity of the Jacobian matrix to singularity and can be used as an
index for static voltage stability.
Condition Number of the Jacobian Matrix. The condition number is used
in numerical analysis to analyze the propagation of errors in matrix A or vector b in
solving variable vector x for the linear equation Ax = b. If matrix A is ill-conditioned,
even very small variations in vector b (or A) may result in significant changes in solution vector x.
For the linearized load flow equations, the condition number of the Jacobian matrix
can be used to measure its conditioning and whether any small variations in vector b
(or A) may result in significant changes in solution vector x. For the linearized load
flow equations, the condition number of the Jacobian matrix can be used to measure
its conditioning and whether any small variations in loads may lead to large changes
in bus voltages. If the condition number is greater than a specified threshold, the current
operating state is close to the collapse point.
A precise measure of the sensitivity of a linear system solution with respect to
matrix A or vector b can be defined as:
( max stretch of A T A )
Cond 2 ( J ) =
( min stretch of A T A )
1
2
1
2
65
VOLTAGE STABILITY INDEXING
If matrix A is symmetric with eigenvalues σ1, σ2, . . . , σn, then Cond2 (A) is
expressed as:
Cond 2 ( A ) =
max σ i
min σ i
For power flow Jacobian matrix J, the value of Cond2 (J) can give an indication of the
condition of J with respect to inversion. A small value of Cond2 (J) (1∼10) refers to a
well-conditioned Jacobian matrix (relatively large voltage stability margin); a large
value of Cond2 (J) (>100) means that the operating state is very close to the point of
Jacobian singularity and has a low voltage stability margin. The extreme condition is
that J is singular and Cond2 (J) is infinite. Hence, the condition number Cond2 (J) can
be used to measure the proximity of the operating states to voltage collapse.
4.5
VOLTAGE STABILITY INDEXING
Many indices characterizing the proximity of an operating state to the collapse point
have been developed. The degeneracy of the load flow Jacobian matrix is used as an
index of power system steady-state stability. Under certain conditions, a change in the
sign of the determinant of the Jacobian matrix during continuous variations of parameters means that a real eigenvalue of the linearized swing equations crosses the imaginary axis into the right half of the complex plane and stability is lost. Some authors
have considered that a change in the sign of the Jacobian matrix probably does not
indicate the loss of steady-state stability when there are an even number of eigenvalues
whose real parts cross the imaginary axis. Voltage stability is related to multiple load
flow solutions.
A proximity indicator for VCPI was defined for a bus, an area, or the complete
system as a vector of ratios of the incremental generated reactive power at a generator
to a given reactive load demand increase. A different indicator (L index) is calculated
from normal load flow results with reasonable computations. The minimum singular
value of the Jacobian matrix was proposed as a voltage security index, since the magnitude of the minimum singular value coincides with the degree of Jacobian ill-conditioning
and the proximity to collapse point. Based on a similar concept, the condition number
of the Jacobian was applied as an alternative voltage instability indicator.
Bifurcation theory was used to analyze static stability and voltage collapse.
Static bifurcation of power flow equations was associated with either divergent-type
instability or loss of causality. The necessary and sufficient conditions for steady-state
stability are based on the concept of feasibility regions of power flow maps and feasibility margins but with high computational efforts. A security measure was derived to
indicate system vulnerability to voltage collapse using an energy function for system
models that include voltage variation and reactive loads. The conclusion is that the
key to applications of the energy method is to find the appropriate Type-1 low voltage
solutions.
66
STABILITY ANALYSIS TOOLS FOR SMART GRID
In addition to the above methods for direct computation of a stability index, some
indirect approaches, based on either the continuation methodc or optimization methods,
have been developed to compute the exact point of collapse. In applying the continuation methods, different assumptions about load changing patterns are needed.
To sum up, the methods for static voltage instability analysis are based on multiple
load flow solutions (VIPI, energy method), load flow results (L index, VCPI), or eigenvalues of the Jacobian matrix (minimum singular value and condition number). While
studies on dynamic voltage collapse explain control strategy design (offline applications), static voltage stability analysis can provide operators with guideline information
on the proximity of the operating state to the collapse point (online applications). In
this case, an index, which can give advance warning about the proximity to collapse
point, is useful.
The differential change in voltage at each bus for a given differential change in
system load is available from the tangent vector. Therefore, one way to identify the
weakest bus with respect to voltage stability limits is to find the bus with the largest
dVi/dPtotal, where dPtotal is the differential change in active load for the entire system
and is given by
dPtotal =
∑ dP
Li
n
= S∆ BASE
∑K
n
Li
cos (ψ i ) dλ = Cdλ
The weakest bus j is therefore given by
dV j
dV
dV2
dVn
= max 1 ,
,…,
Cdλ
Cdλ
Cdλ Cdλ
When the weakest bus j reaches its steady-state, the voltage stability limit dλ approaches
zero and the ratio |dVj/Cdλ| will become infinite or, equivalently, the ratio |Cdλ/dVj|
tends to zero. The latter ratio which is easier to handle numerically and makes a good
voltage stability index for the entire system. Instead of dPtotal, dQtotal can also be used.
The International Council on Large Electric Systems (CIGRE) task force categorizes
indices as given operating state-based and large deviation-based. Large deviation indices
account for discontinuities such as generator current limiting. Based on the system’s
ability to withstand load or power transfer increases, an MW or MVAr distance or
margin from the operating point to the maximum power transfer point is determined.
Given operating state indices are based on a solved power flow case or an actual
real system operating point. Observation of reactive power reserve is a simple and
valuable index that can be calibrated to system security.
Determining the index for DC systems is somewhat different. The index may be
calculated as ∆V = ∑in=1Vi − Vref where Vi is the voltage at bus i, Vref is the reference bus
voltage, and choosing a threshold is the security.
The modal or eigenvalue analysis method is akin to sensitivity analysis but the
modal separation provides additional insight. The system partitioned matrix equations
of the Newton-Raphson method can be rewritten as
67
VOLTAGE STABILITY INDEXING
J PV ∆θ
JQθ ∆V
∆P J Pθ
∆Q = J
Qθ
Where the partitioned Jacobian matrix reflects a solved power flow condition and
includes enhanced device modeling, by letting ΔP = 0, we can write
∆Q = JQV − JQθ J P−θ1J PV ∆V = J R ∆V
where JR is a reduced Jacobian matrix of the system. JR directly relates the bus voltage
magnitude and bus reactive power injection.
Let λi be the ith eigenvalue of JR and ξi and ηi be the corresponding right-column
and left-row eigenvectors respectively. The ith modal reactive power variation is
∆Qmi = K iξi
where
K i2
∑ξ
2
ji
=1
j
with ξji the jth element of ξi. The corresponding ith modal voltage variation is
∆Vmi =
1
∆Qmi
λi
The magnitude of each eigenvalue λi determines the weakness of the corresponding
modal voltage. The smaller the magnitude of λi, the weaker the corresponding modal
voltage. If λi = 0, the ith modal voltage will collapse because any change in that modal
power will cause infinite modal voltage variation.
If all eigenvalues are positive, the system is considered voltage-stable. This is a
different dynamic system where eigenvalues with negative real parts are stable. The
relationship between system voltage stability and eigenvalues of the JR matrix is best
understood by relating the eigenvalues with the Q-V sensitivity of each bus. JR can be
taken as a symmetric matrix and therefore the eigenvalues of JR are close to being
purely real. If all the eigenvalues are positive, JR is positive definite and the V-Q sensitivities are also positive, indicating that the system is voltage-stable.
The system is considered voltage-unstable if at least one of the eigenvalues is
positive. A zero eigenvalue of JR means that the system is on the verge of voltage
instability. Furthermore, small eigenvalues of JR determine the proximity of the system
to be voltage-unstable.
The participation factor of bus k to mode i is defined as
PKi = ξ kiηik
For all of the small eigenvalues, bus participation factors determine the areas close to
voltage instability. In addition to the bus participations, modal analysis also calculates
68
STABILITY ANALYSIS TOOLS FOR SMART GRID
branch and generator participations. Branch participations indicate which branches are
important in the stability of a given mode. This provides insight into possible remedial
actions as well as contingencies which may result in loss of voltage stability. Generator
participations show which machines must retain reactive reserves to ensure stability of
a given mode.
For a system with several thousand busses it is impractical and unnecessary to
calculate all of the eigenvalues. Calculating only the minimum eigenvalue of JR is not
sufficient because there is usually more than one weak mode associated with different
parts of the system, and the mode associated with the minimum eigenvalue may not be
the most troublesome mode in the stressed system. The m smallest eigenvalues of JR
are the m least stable modes of the system. If the biggest of the m eigenvalues, say
mode m, is a strong enough mode, the modes that are not computed can be neglected
because they are known to be stronger than mode m. An implicit inverse lopsided
simultaneous iteration technique is used to compute the m smallest eigenvalues of JR
and the associated right and left eigenvectors.
Similar to sensitivity analysis, modal analysis is only valid for the linearized model.
Modal analysis can, for example, be applied at points along P-V curves or at points in
time in a dynamic simulation.
4.6 ANALYSIS TECHNIQUES FOR STEADY-STATE VOLTAGE
STABILITY STUDIES
In their early stages, voltage collapse studies mainly concerned steady-state voltage
behavior. Voltage collapse is often described as a problem that results when a transfer
limit is exceeded. The transfer limit of an electrical power network is the maximal real
or reactive power delivered by the system from generation sources to the load areas.
Specifically, the transfer limit is the maximal amount of power that corresponds to at
least one power flow solution. From the well-known PV or QV curves, one can observe
that the voltage gradually decreases as the power transfer amount is increased. Beyond
the maximum power transfer limit, the power flow solution does not exist, which
implies that the system has lost its steady-state equilibrium point. From an analytical
view, the criteria for detecting the point of voltage collapse is the point where the
Jacobian matrix of power flow equations becomes singular.
The steady-state operation of the power system network is represented by power
flow equations given by
F (θ ,V , λ ) = 0
where θ represents the vector of bus voltage angles and V represents the vector of bus
voltage magnitudes. λ is a parameter of interest we wish to vary. In general, the dimension of F will be 2n1 + n2, where n1 and n2 are the number of PQ and PV busses,
respectively.
It should be emphasized that the singularity of the power flow Jacobian ∂F/∂x is
a necessary but not sufficient condition to indicate voltage instability. The method
ANALYSIS TECHNIQUES FOR STEADY-STATE VOLTAGE STABILITY STUDIES
69
proposed to observe the voltage instability phenomenon is closely related to multiple
power flow solutions caused by the nonlinearity of power flow solutions. The drawback
of the method is that it relies on the Newton-Raphson method of power flow analysis,
which is unreliable in the vicinity of the voltage stability limit.
4.6.1
Direct Methods for Detecting Voltage Collapse Points
This approach tries to find the maximum allowable variation of λ, that is, an operating
point (x*, λ*) of F(x, λ) = 0 such that the Jacobian at this point is singular. It solves
the following system of equations:
F ( x, λ )
G ( y ) = FxT ( x, λ ) h = 0
hk − 1
This procedure basically augments the original set of power flow equations F(x, λ) = 0
by Fx(x, λ)h = 0 where λ is an n-vector with hk = 1. The disadvantages of this approach
are:
•
•
The dimension of the nonlinear set of equations to be solved is twice that for the
conventional power flow
The approach requires a good estimate of the vector λ
The advantage is that convergence of the direct method is very fast if the initial operating point is close to the turning point. The enlarged system is solved in such a way that
it requires the solution of four n × n (n is the dimension of the Jacobian Fx(x, λ)) linear
systems, each with the same matrix.
4.6.2
Indirect Methods (Continuation Methods)
Assuming that the first solution (x0, λ0) of F(x, λ) = 0 is available, the continuation
problem is to calculate further solutions, (x1, λ1), (x2, λ2), until a target point is reached,
say at λ = λ*. The ith continuation step starts from an approximation of (xi, λi) and
attempts to calculate a next solution. However there is an intermediate step. With predictor, corrector type continuation, the step i → i + 1 is split into two parts. The first
part tries to predict a solution, and the second part tries to make this predicted part
converge to the required solution:
Predictor
Corrector
( xi , λi ) → ( xi +1, λi +1 )
( xi +1, λi +1 ) → ( xi +1, λi +1 )
Continuation methods differ in choice of predictor, type of the parameterization strategy, type of corrector method, and step length control.
Parameterization. The branch consisting of solutions forming a curve in the
(x, λ) space has to be parameterized. A parameterization is a mathematical way of
70
STABILITY ANALYSIS TOOLS FOR SMART GRID
identifying each solution on a branch. There are many different kinds of parameterization. For instance, looking at a PV curve shows that the voltage is continually decreasing
as the load nears maximum. Thus, the voltage magnitude at some particular bus could
be changed by small amounts and the solution is found for each given value of the
voltage. Here the load parameter would be free to take on any value it needed to satisfy
the equations. This is called local parameterization. In local parameterization the original set of equations is augmented by one equation that specifies the value of one of the
state variables. In equation form this is expressed as:
F ( y)
y − η = 0,
k
θ
y = V
λ
where η is an appropriate value for the kth element of y. Once a suitable index k and
the value of η are chosen, a slightly modified N-R power flow method (altered only in
that one additional equation and one additional state variable are involved) can be used
to solve the set of equations. This provides the corrector needed to modify the predicted
solution discussed in the previous section.
A simple example to explain the static analysis procedure follows.
4.7 APPLICATION AND IMPLEMENTATION PLAN OF VOLTAGE
STABILITY
Consider the power flow equation defined in the equation below. The vector function
F consists of n scalar equations defining a curve in the n + 1 dimensional (x, λ) space.
Continuation means tracing of this curve. For a convenient graphical representation of
the previous solution (x, λ) we need a one-dimensional measure of x. The frequently
used measures are:
(i) x = ∑in=1 xi2 (square of the Euclidean norm)
(ii) x = max xi (maximum norm)
i =1,n
(iii) x = xk For some index k, 1 ≤ k ≤ n
Power systems generally use the measure of (iii). As Figure 4.6 illustrates, we have a
type of a critical solution for λ = λ*, where for λ > λ* there are no solutions. For λ < λ*
we have two solutions (high voltage solution, low voltage solution). When λ approaches
λ*(λ < λ*), both solutions merge. At this point, the Jacobian of the power flow solution
is singular. In the mathematical literature these points are called turning points, or fold
points. An algebraic feature of the turning point given by Fx is
•
•
Fx(x*, λ*) is singular for rank <n
Fx(x*, λ*)/Fλ(x*, λ*) has a full rank n and satisfies some nondegeneracy
conditions
71
OPTIMIZING STABILITY CONSTRAINT THROUGH PREVENTIVE CONTROL
|x|
•
(x−i+1,li+1)
•
(xi,li)
(xi+1,li+1)
l
Figure 4.6. The fold type curve including predictor-corrector step.
Several techniques proposed to calculate these points are based on two approaches:
direct and indirect.
4.8 OPTIMIZING STABILITY CONSTRAINT THROUGH PREVENTIVE
CONTROL OF VOLTAGE STABILITY
The controllable reactive power sources include generators, shunt reactors, shunt capacitors, and on load tap changers of transformers (OLTC). Generators can generate or
absorb reactive power depending on the excitation. When overexcited they supply the
reactive power, and when under-excited they absorb reactive power. The automatic
voltage regulators of generators can continually adjust the excitation. Reactors shunt
capacitors and OLTCs are traditionally switched on/off through circuit breakers on
command from the operator. Since the early 1980s, advances in flexible AC transmission systems (FACTS) controllers in power systems have led to their application to
improve voltage profiles of power networks. Figure 4.7 depicts the control scheme.
The most frequently used devices are Reactive Power Controllers (RPC) and Static
Var Compensators (SVC). The RPC connects or disconnects capacitor stages automatically by detecting the phase divergence between the fundamentals of current and
voltage. The measured divergence is compared with several segmental set phase divergence regions, and the capacitor contactors will be switched on or off accordingly. SVC,
more advanced electronics, provides continuous capacitive and inductive reactive
72
STABILITY ANALYSIS TOOLS FOR SMART GRID
Input system data
Select contingencies
Solve base case load flow
Calculate stability index
Give limits to
critical states
Check stability
index
NO
All satisfied ?
Modal and
sensitivity analysis
YES
Check stability index for
other contingencies
NO
Stability index satisfied for all
contingencies ?
YES
Stop
Figure 4.7. Voltage instability preventive control scheme for the power grid.
supply to the power system. The SVC typically consists of a thyristor controlled reactor
(TCR), a thyristor switched capacitor (TSC) and AC filters (ACF). From the view
system operation, the SVC is equivalent to a controllable reactor and a fixed capacitor.
Its output can vary depending on the level of generation and absorption of reactive
power needed to maintain its terminal voltage at a certain level. Both the RPC and SVS
require significant investments. Since they work only in a local area, the reactive power
sources of a network must be coordinated to achieve network voltage stability. A new
optimal power flow that includes dynamics of DSOPF is needed.
In an operations planning environment, it is desirable to determine whether a set
of dynamic disturbance scenarios is stable or unstable and to assign a measure to its
degree of stability, in order to compare the various configurations and determine the
ANGLE STABILITY ASSESSMENT
73
transfer level that exhausts the available stability margin. Although transient and
voltage stability—due to distinct mechanisms and separate time frames—have sometimes been perceived as decoupled phenomena, in reality they ought to be treated as
aspects of coupled short-term power system dynamics [21]. Various researchers have
used energy function methods to quickly assess whether a transient disturbance will
result in loss of stability. A focus on transient stability assessment methods that yield
results are fast but operations planners synthesize the output of simulations resulting
in a multitude of time-domain curves into a single, easy-to-interpret number, namely,
the energy margin (EM).
For the transmission and distribution system to become more affordable, reliable
and sustainable the grid needs to become smarter. During the past few years a considerable number of activities have been carried out in to achieve a smart power grid. Smart
grid is envisioned to use all present technologies of voltage stability in transforming
the grid intelligently and creating better situational awareness and operation friendliness. At the same time blackouts in the past and reliability of the grid are major issues
for system engineers. Features of the DSOPF that aid the development of new OPF
with dynamic OPF as constraint, solving performance index, new optimizer DSOPF
that incorporate the performance in real time under different control situation are
modeled as follows:
1. Obtain load and generation control profile and real time data contribution.
2. Develop and optimize the minimum imbalance between load and generation
considering the load generation randomness.
3. Design an optimizer that formulates the OPF tool as a constant where generation
availability is a constraint in random space.
4. Design computational intelligence or a combination of classical optimization to
solve the scheme problem.
5. Utilize the priority list scheme from dynamic programming concept to rank
order load generation balance index.
4.9
ANGLE STABILITY ASSESSMENT
Transient stability studies of a typical 500-bus, 100-machine system require an hour of
run time, even for a single contingency. Therefore, direct methods of stability assessment, such as those based on Lyapunov or energy functions, are attractive alternatives.
It should be noted that a transient stability study is often more than an investigation of
whether the synchronous generators, following the occurrence of disturbance, will
remain in synchronism, for it can be a general-purpose transient analysis to investigate
the quality of the dynamic system behavior. The transient period of primary interest is
the electromechanical transient, usually lasting up to a few seconds in duration. If
growing oscillations or the behavior of special controls are a concern, the study may
involve a longer transient period. The problem solution uses one of the family of solutions depicted in Figure 4.8.
74
STABILITY ANALYSIS TOOLS FOR SMART GRID
Figure 4.8. Overview of methods for angle stability.
In actual system operation, the parameters and loading conditions are quite different
from the assumptions made in the planning stage. As a result, to ensure power system
security against possible abnormal conditions due to contingencies, the system operator
needs to simulate contingencies in advance, assess the results, and take preventive
control action if required. The equation for solving this can be stated and modeled as:
Md 2δ / dt 2 = Pm − Pe
which is written in state space as
dδ / dt = f (δ , u)
d 2δ / dt 2 = acceleration
where δ = angle
u = Control (power system generator taps, and so on)
Pm, Pe = mechanical and electrical power
To solve the time domain stability problem, numerical methods are utilized based on
predictive and corrective methods and to provide suggestions for further interpretation,
including:
75
ANGLE STABILITY ASSESSMENT
1. Numerical method: RK4, Trapezoidal method, modified Euler method, Theta
method. The Euler approach converts differential equations from digital to analogue;
numerical methods solve numerical method technique by analyzing decoupled first- and
second-order state differential equation with the predictor corrector technique; the
trapezoidal method is faster for still networks than other methods. All of these technique
have similarities: the initial condition of the network, step-size selection, and integration application.
Although these methods are used today, they lack characteristics such as the robustness, scalability, stochasticity, predictivity, and adaptability of the future grid, which is
completely absent in the current power system grid network.
2. Artificial Intelligence Method. This technique uses heuristic and computational
intelligence to estimate the strability margin and compuatational indices. The method
includes the expert system for classifying contingencies, identifying and estimating the
instability margin, and so on.
The neural network is also used as a classifer for different stability margins based
on network features and parameters contributing to the estimation of instabilities. Other
AI methods include particle swarm optimization, genetic algorithm. These methods are
used to model the inherent and stochastic elements in the network. Several of these
techniques have been tested in the literature and they form the basis for a new methodology for angle stability margin or assessment in real time.
An important techique for stability assessment includes the dynamic security
assessment used over several years. These method is useful as it can take in real time
data measurement, unlike other methods.
3. Evaluation of dynamic security assessment and its potential for real-time stability in a smart grid environment. DSA includes the study of power system oscillation in
terms of generating power plants and network configuration under different contingencies. To classify the impact of the contingency, an assessment of the energy margin of
unit impact based on loss of line or unit consists of measures of the potential and kinetic
energy associated with the machines in the system with respect to a reference machine,
or COA. The sum of the kinetic and potential energy is compiled and compressed to a
given threshold in order to determine the network impacts, and rank and classify them.
4.9.1
Transient Stability
Consider an autonomous system described by the ordinary differential equation
xɺ = F ( x )
where x = x(t), and F(x) are n-vectors. F(x) is generally a nonlinear function of x. Stability in the sense of Lyapunov is referred to as an equilibrium state. The equilibrium
state is defined as the state xe at which x(t) remains unchanged for all t. That is,
xɺe = F ( xe ) = 0
76
STABILITY ANALYSIS TOOLS FOR SMART GRID
The solution for xe is a fixed state since F(x) is not an explicit function of t. For convenience, any nonzero xe is to be translated to the origin (x = 0.) That is, replace x by
x + xe to have
d
( x + xe ) = F ( x + xe ) = f ( x )
dt
which gives
xɺ = F ( x )
Note that the current x differs from the old one by xe. As can be seen later from
the definitions, this translation does not affect the stability of the system. Thus, the
origin is always an equilibrium state. Note that t may be any independent variable
including the time.
Let a = d2δ/dt2 denote the acceleration of the system with a < 0 above the Pm line
and a > 0 below it after fault clearance. There are two angles: δs = sin−1Pm/A(Xc) and
δu = π − δs that correspond to zero acceleration. δ increases from δ0 with ω = 0 due to
decreasing acceleration, until reaching δm < δu, at which point the speed, ω, is zero.
Since the acceleration is negative at δm, δ begins to decrease until the machine speed
is zero at an angle less than δs and then comes back because a > 0. As such, the power
angle swings back and forth around δs. This is the case for E − ∇ < 0 at δu because δu
is unreachable (ω is imaginary).
However, one fault clearance may result in large A1 which makes δ cross δu with
ω ≥ 0. Then, δ increases further without return due to a > 0. This is the case for
E − ∇ ≥ 0. Thus, the transient is stable if the power angle swings around δs and unstable
otherwise. This definition makes it possible to assess the stability with other techniques
such as transient energy function (TEF) as follows:
(a) The transient is stable if E − ∇ < 0 at δu, and large magnitude yields better
stability
(b) The transient is unstable if E − ∇ ≥ 0 at δu
4.9.2
Stability Application to a Practical Power System
Since the application of the direct method to actual power systems is quite difficult, it
becomes necessary to make a number of simplifying assumptions. To date, the analysis
has been mostly limited to power system representation with generators represented by
classical models and loads modeled as constant impedances. Recently several attempts
to extend the method now include more detailed load models.
In a multi-machine power system, the energy function E describing the total system
transient energy for the post-disturbance system is given by:
77
ANGLE STABILITY ASSESSMENT
∇=−
1
2
n −1
n
∑
n
J iωi2 −
i =1
n
∑ P′ (θ − θ ) −
mi
s
i
i =1
∑ ∑ C (cosθ
ij
i =1 j =i +1
i
ij
− cosθijs ) −
∫
θi +θ j
θis +θ sj
Dij coosθij d (θi + θ j )
where
θis = angle of bus i at the post-disturbance SEP
Ji = 2Hiω0 = per unit moment of inertia of the ith generator
The transient energy function consists of the following four terms:
1.
1
2
∑J iω i2: change in rotor kinetic energy of all generators in the COI reference
frame
2. ∑Pmi′ (θ i − θ is ): change in rotor potential energy of all generators relative to COI
3. ∑ ∑Cij (cos θ ij − cos θ ijs ): change in stored magnetic energy of all branches
4. ∑ ∑ ∫Dij cos θ ij d (θ i + θ j ): change in dissipated energy of all branches
The first term is called the kinetic energy (Eke) and is a function of generator speeds
only. The sum of terms 2, 3, and 4 is called the potential energy (Epe) and is a function
of generation angles only.
The assessment procedure has three steps:
Step 1: Calculate the critical energy Vcr
Step 2: Calculate the total system energy at the instant of fault-clearing Ecl
Step 3: Calculate the stability index: Vcr − Vcl; the system is stable if the stability
index is positive
A time domain simulation is run up to the instant of fault clearing to obtain the
angles and speeds of all generators. These are used to calculate the total system energy
(Vcl) at fault clearing. The flowchart of the TEF for transient stability analysis is shown
in Figure 4.9.
4.9.3
Boundary of the Region of Stability
The calculation of the boundary of the region of stability, Vcr, is the most difficult step
in applying the TEF method. Three different approaches are briefly described here.
Algorithm for the closest unstable equilibrium point (UEP) approach.
Early papers on the application of the TEF method for transient stability analysis used
the following approach to determine the smallest Ecr.
Step 1: Determine all of the UEPs by solving the post-disturbance system steadystate equations with different initial values of bus angles
78
STABILITY ANALYSIS TOOLS FOR SMART GRID
Start
Input system data
Power flow calculation
Contigency specification
Form during fault Y matrix
and reduced Y matrix
Form post-fault Y matrix
and reduced Y matrix
Calculate post fault SEP
Calculate critical energy Vcr
Calculate system total energy
Vcl at clearing time using an
Exit Technique (Sec. 5.6-b to c)
Yes
Vcr > Vcl
?
No
unstable
stable
Stop
Figure 4.9. Flowchart of TEF for transient stability analysis.
Step 2: Calculate system potential energy at each of the UEPs obtained in step 1;
the critical energy Vcr is given by the system at the UEP, which results in the
minimum potential energy
Because this approach computes the critical energy by implicitly assuming the worst
fault location, the results are very conservative.
Algorithm for the controlling UEP approach. The degree of conservatism
introduced by the closest UEP approach is such that the results are usually of little
practical value. The controlling UEP approach removes much of this conservatism by
computing the critical energy depending on the fault location. This approach is based
79
ANGLE STABILITY ASSESSMENT
on the observation that the system trajectories for all critically stable cases get close to
the UEPs that are closely related to the boundary of system separation. The UEPs are
called the controlling or relevant UEPs.
The essence of the controlling UEP [6,7] method as shown by Chiang uses the
constant energy surface through the controlling UEP to approximate the relevant part
of the stability boundary (stable manifold of the controlling UEP) to which the fault-on
trajectory is heading.
For any fault-on trajectory xf (t) starting from a point p ∈ A(xs) with V ( p ) < V ( xˆ ),
if the exit point of the fault-on trajectory lies in the stable manifold of x̂, the fault-on
trajectory must pass through the connected constant energy surface ∂Vc ( xˆ ) before it
passes through the stable manifold of xˆ (W S ( xˆ ) ) (thus exiting the stability boundary
∂A ( xˆ S )). Therefore, the connected constant energy surface ∂Vc ( xˆ ) can be used to
approximate the part of the stability boundary ∂A ( xˆ S ) for the fault-on trajectory xf (t).
The four steps in the computation process are:
1. Determine the controlling UEP, xco for the fault-on trajectory xf (t)
2. Setting the critical energy Vc as the value of the energy function V(·)at the controlling UEP, that is, Vc = V(xco)
3. Calculate the value of the energy function E(·) at the time of fault clearance
(say, tcl) using the fault-on trajectory Vf = V(xf (tcl))
4. If Vf < Vc, then the post-fault system is stable, otherwise, it is unstable
The key is how to find the controlling UEP for a fault-on trajectory. Much of the
recent work on the controlling UEP method is based on heuristics and simulations. A
theory-based algorithm to find the controlling UEP for the classical power system
model with transfer conductance Gij is discussed in the next section.
The energy function is of the form:
n −1
V (δ , ω ) =
n
∑ ∑ 2 M
1
i =1 j = i +1
−
T
Mi M j (ωin − ω jn
2
1
( Pi M j − Pj Mi )(δ in − δ jn − δ inS + δ jnS )
MT
− ViV j Bij {cos(δ in − δ jn ) − cos(δ inS − δ jSn )
− ViV j Gij
= Vp (δ ) +
δ in + δ jn − (δ inS + δ jnS )
{sin(δ in − δ jn ) − sin(δ inS − δ jnS )}
δ in − δ jn − (δ inS − δ jnS )
1
2 MT
n −1
n
∑ ∑ M M (ω
i
j
in
− ω jn )2
i =1 j = 1
where MT = ∑in=1 Mi, xS = (δ S,0) is the stable equilibrium point (SEP) under consideration. Note that V(δ,ω) represents the energy margin, while Vi and Vj are nodal voltages
within the network.
80
STABILITY ANALYSIS TOOLS FOR SMART GRID
4.9.4
Algorithm to Find the Controlling UEP
The reduced system is
1
M
δɺin =
( Pi − Pei ) − i ( Pn − Pen )
Mi
Mn
= fi (δ )
i = 1, 2, ⋯ , n − 1
The algorithm for finding the controlling UEP consists of five steps:
Step 1: From the fault-on trajectory (δ(t),ω(t)), detect the point δ* at which the
projected trajectory δ(t) reaches the first local maximum of Vp(.), and then
compute the point δ ¯ that is one step ahead of δ* along δ(t), and the point δ +
that is one step after δ*
Step 2: Use the point δ* as initial condition and integrate the post-fault reduced
system Equation (5–31) to find the first local minimum of ∑in=1 fi (δ ) , say at δ 0*
Step 3: Use δ ¯ and δ + as initial conditions and repeat Step 2 to find the corresponding points, say δ ō and δ +o, respectively
Step 4: Compare the values of |f(δ)| at δ o−, δ o*, and δ o+; use the one with the smallest
value as the initial guess to solve (5–31), fi(δ) = 0, say the solution is δco
Step 5: The controlling UEP for the fault-on trajectory is (δco, 0)
4.9.5
Process Changes in Design of DSA for the Smart Grid
We now give the possible options for designing DSA for the smart grid.
Option 1. Real Time DSA with PMU is currently being used by some researchers, but we propose a modification that includes system topology in time stamp for a
new topology of the power system with or without RER in three steps:
Step 1 Develop a computer index for assessing stability based on different scheduled parameter changes
Step 2 Develop a new ranking procedure in time interval and utilize some form
of probability function
Step 3 Identify source of instability and margin of stability for next corrective or
preventive action
Option 2. The second option includes variability in the definition of system
components and devices as well as the model of load in stochastic form. Doing so
places a construct on the network solution as a probabilistic load flow. The model of
the generators are RER mixed generation (some deterministic and some probabilistic).
The aggregate of each requires some correlative probability distribution function. This
dynamic equation is now treated as a probabilistic method with results given in terms
STATE ESTIMATION
81
of mean and variance. We then apply reversed function computation to address the state
of margin calculation computation. The interpretation of the results also accounts for
new variability. Figure 4.10 shows the form. The three steps are:
1. Design nominal integration with variation time to complete values of P as function of δ(t)
2. Analyze the algorithm or differential equation coupled
3. Interpret results and display
Option 3. Research is in progress to include the variability of RER and stochastic
load with the appropriate probability distribution function to establish a probabilistic
equation and couple it with the differential equation. As before, at each time step,
corresponding values of power from control generation and RER using computational intelligence or artificial neural network provide prediction and estimation of
instability.
4.10
STATE ESTIMATION
Energy management systems (EMS) run in real time to compute and maintain security
of operation at minimum cost. The power system measurements provide information
to the SE program for processing and analysis. The functions performed include topology processing which gathers data about breakers and disconnect switches. State estimation (SE) of voltage and angles are obtained for all busses using the weighted least
square (WLS) method. Detection of inaccurate data is obtained from network parameters, tap changing transformers, shunt capacitors, or breakers. The general SE model
discussed below is based on steady-state functions. Many of the algorithms in use
contain problems including:
1. Convergence problem: traditional SE may not converge if the power system
state changes faster than the SCADA data
2. Possibility of missing critical data from communication channels
3. System parameter data base with many errors
4. Report changes in system parameters or devices such as breaker, taps, or disconnected switches
This is true because when topology changes are undefined, the estimation is likely
to fail because the SE develops on network topology, bus admittance matrix which may
fail to lead to invertible Y-bus matrix. The explicit modeling of switches facilitates bad
data analysis when topology errors are involved (incorrect status of switching devices).
In this case, SE is performed on a model in which parts of the network can be represented physically, allowing for the inclusion of measurements made on zero impedance
branches and switching devices. The conventional states of bus voltages and angles are
augmented with new state variables. Observability analysis is extended to voltages at
bus sections and flows in switching devices; if their values can be computed from the
82
STABILITY ANALYSIS TOOLS FOR SMART GRID
ANN
List of Contingencies
Determine stability margin
Do TSP (transient stability performance)
Do Control Solution
Re-assert stability
Computation of new margin of stability
Is stability margin ≤1
Propose new stability for the grid
Power System
Real time measurement / DataProcessing
Predictive TSA / DSA
Control Action
Unstable power system
Stability margin
Design Corrective Action
Figure 4.10. DSA for the smart grid.
83
STATE ESTIMATION
available measurements, they are considered to be observable. For a zero impedance
branch or a closed switch, the following constraints or pseudo- measurements, are
included in SE [14]:
Vk − Vm = 0 and θ km = θ k − θ m = 0
In this case, Pkm and Qkm are used as additional state variables independent of the
complex nodal voltages Vk e jθk and Vk e jθ m, since Ohm’s law (in complex form) cannot
be used to compute the branch current as a function of these voltages.
For open switches (zkm = ∞), the same additional state variables are included in SE.
No pseudomeasurements are added in the case of switches with unknown status.
Situations occur in which the incorrect status of a switching device can affect SE convergence. In these cases it may be preferable to treat such status as unknown and
proceed with SE, which hopefully will include the estimation of the correct status.
The ideas above can be extended to branches with unknown impedances [11].
Branch impedance zkm is unknown, whereas for simplicity all branches incident to k
and m are assumed to have known impedances. As with zero impedance branches and
with closed/open breakers, Ohm’s law cannot be used to relate the state variables Vk e jθk
and Vk e jθ m associated with the terminal nodes k and m with the branch complex power
flows Pkm + jQkm, and Pmk + jQmk. These power flows can be used as additional states,
although they are not independent, since they are linked by the constraint Ikm + Imk = 0
which can be expressed by the two following pseudomeasurements:
PkmVm + ( Pmk cosθ km − Qmk sin θ km ) Vk = 0
QkmVm + ( Pmk sin θ km + Qmk cosθ km ) Vk = 0
A power injection measurement at node k can be expressed as the summation of the
flow state variables Pkm, Qkm and the flows in all other branches incident to k. Since
only the flows in regular branches are functions of the nodal state variables, the
unknown impedance will not form part of the measurement model. A similar analysis
holds for power injection measurement at nodal point and power flow measurements
made in the unknown impedance branch. Once the network state is estimated, the value
of the unknown parameter can be computed from the estimates.
Complex network elements, such as a transmission line π equivalent model, require
consideration of other constraints (pseudomeasurements) in addition to the inclusion
of flow state variables. Consider, for example, the equivalent π model, where the series
branch impedance is to be estimated. In this case, power flows Pkm, Qkm, Pmk and Qmk
are considered to be additional states. Then, the terminal power flows Pkm, Qkm, Pmk and
Qmk are expressed in terms of the new state variables rather than as a function of the
terminal bus voltages; as a consequence, the series branch impedance will not appear
in the measurement model, and it can be written as
′
Pkm = Pkk + Pkm
′
and Qkm = Qkk + Qkm
′
Pmk = Pmm + Pmk
′
and Qmk = Qmm + Qmk
84
STABILITY ANALYSIS TOOLS FOR SMART GRID
In these equations, note that the power flows Pkk, Qkk, Pmm and Qmm are written in terms
of the shunt parameters as usual. The bus injection measurements at busses k and m
are expressed in terms of the terminal power flows as described above. These added
states are not entirely independent, so it is necessary to include the following relationship in the model:
′ Vm + ( Pmk
′ cosθ km − Qmk
′ sin θ km ) Vk = 0
Pkm
′ Vm + ( Pmk
′ sin θ km + Qmk
′ cosθ km ) Vk = 0
Qkm
th
The constraint linking the state variables in the case of a balanced π model is ykkth = ymm
.
Expressing the shunt admittances in terms of the corresponding active and reactive
power flows yields two pseudomeasurements:
Pkk Vm2 − PmmVk2 = 0
Qkk Vm2 − QmmVk2 = 0
The term static state estimation refers to the process of computing solutions of the
basic load flow problem using online data telemetered periodically to the energy control
center. Data exchanges with neighboring systems to develop an external network
equivalent model are facilitated if each system has an online state estimator. An external
equivalent representation is required to perform online state contingency analysis.
Without an external equivalent model, online SE will be limited to monitoring voltage
levels, phase angles, line flows, and network topology.
In load flow, the input/load variables describe the steady-state behavior of the
system. In actual online systems it is not usually possible to measure these inputs and
loads directly, because loads are net injection quantities, each of which is the sum of
several power flow quantities. A single measurement of power flow on a transmission
line requires less instrumentation and may be useful in contributing to the load flow
solution. In principle, one can measure any meaningful set of system quantities and use
those measurements as inputs to a system of equations whose solution is the values of
state variables (bus voltage magnitudes and angles). We can compute net bus injections
from the estimated state variables.
4.10.1 Mathematical Formulations for Weighted Least
Square Estimation
Because of redundant measurements, the solution x̂ of x is obtained by minimizing the
WLS performance index J given by:
J = [ z − h ( x, p )] R −1 [ z − h ( x, p )]
T
with respect to x. The vector p is assumed to be known exactly. Hence, we can drop p
from the above expressions. The concise SE problem statement becomes:
Given z = h(x) + v such that E(v) = 0, E(vvT) = R compute the best estimate x̂ of x
which minimizes:
85
STATE ESTIMATION
J = [ z − h ( x )] R −1 [ z − h ( x )]
T
with respect to x. At the minimum of J we should expect that:
∂J
=0
∂x x̂
where x̂ is the state vector at the minimum of J and is referred to as the best estimate
of x. Given the above definition of J, we assert that the zero gradient condition just
stated will yield the n-dimensional vector equation:
0 = HT ( xˆ ) R −1 ( z − h ( xˆ ) )
The necessary conditions for solution are a set of nonlinear algebraic equations requiring an iterative solution method, that is, following the Newton-Raphson method, we
linearize the system equations around a nominal value of the state vector x.
Statistical Properties of State Estimator Outputs. The expected value of
the optimal estimate vector x̂ is given by:
E ( x̂ ) = x
Moreover, the covariance matrix of x̂ is given by:
−1
cov ( x̂ ) HT R −1H
We are also interested in the expected values of the index (J/m) corresponding to the
fit of estimates of measured quantities to the measurements themselves. This can be
given by:
E (J / m) =
m−n
n
For m − n (no redundancy), E(J/m) = 0 and the estimates fit the measurements perfectly. For m → ∞ (infinite redundancy), E(J/m) → 1 and the estimates approach the
true value. In addition, the index Jt/m corresponds to the fit of the estimates to the true
values of the noisy measurements. This can be given by:
E ( J t / m) = n / m
For m = n, we get E(Jt/m) = 1, and for m → ∞, E(Jt/m) = 0, m → 0, d.n.e.
In general the index J is chi-square distributed with m − n degrees of freedom.
86
STABILITY ANALYSIS TOOLS FOR SMART GRID
4.10.2
Detection and Identification of Bad Data
Redundant measurement information allows us to identify and locate bad data consisting of gross measurement errors and/or large modeling errors, for example, wrong
network topology, or large parameter errors. Without redundancy the estimates will fit
the data perfectly, but it does not provide ways to locate bad data. With redundancy,
the WLS algorithm will try to minimize the performance index J (or Jp as the case may
be). In the absence of bad data and parameter errors, the expected value of J/m is
(m − n)/m. Thus we have a ready means to check if the data lies within its postulated
error bounds. If J >> m − n , then we can be sure that something is wrong. This is the
m
m
detection step. In this case, we have to look for the source of trouble by means of bad
data identification. The four sources of problems are:
1.
2.
3.
4.
Gross measurement errors
Small modeling errors
Small parameter errors
Inaccurate knowledge of measurement variances
In a real situation (and especially in initial implementation phases) all of these
problems will occur simultaneously. The solution key consists of two primary
considerations:
•
•
Exploitation of the structure of power flow equations and associated sensitivities
Creative hypothesis testing
Extra care is required to make the bad data identification step workable. This is achieved
in the two steps, pre- and postestimation analysis, described below.
4.10.3
Pre-Estimation Analysis
Before a given snapshot measurement, vector z of the system is processed in the WLS
algorithm. Its components can undergo a series of so-called consistency tests with the
following objectives:
•
•
•
•
Detection of obviously bad measurements
Detection of obviously bad network topology
Classification of data as (a) valid, (b) suspect, and (c) raw
Tuning the measurement variance values
We discuss some of these topics as follows.
Detection of Obviously Bad Measurements. In this preliminary stage, measurements whose values are outside reasonable limits are automatically discarded. For
example, line flow limits can be set at twice the theoretical capacity of the line. Power
87
STATE ESTIMATION
factors, voltage levels, and so on can be safely limited. In most cases, almost all of the
bad data will be in this category and can be quickly discarded.
Detection of Obviously Bad Network Topology. Normally, a special
network configuration program constructs the system network on the basis of breaker
status information. Open lines are not represented in the model forwarded to the state
estimator, that is, all lines that are in the model used by the estimator should be closed
(energized). However, cases may occur where a breaker is closed but has an open
disconnect switch. If the disconnect switch status is not reported, the line is mistakenly
assumed to be energized. One way to check for this anomaly is to see if the power flow
on the line is zero at both ends. If this is the case then the line is most probably open.
Other cases of bad topology may be detected from incoming data and are usually
peculiar to the system being analyzed.
Classification of Raw Data. Because of the structure of load flow equations a
considerable number of hypothesis consistency tests can be conducted to verify the
validity of most of the data and to tune the values of the measurement variables. Typical
examples of these tests are:
1. Line flow measured at both ends: For real flows, the magnitudes of flows from
both ends differ only by the amount of line losses evaluated from:
TLoss = Tij + T ji
= gij (Vi 2 + V j2 − 2ViV j cos (δ i − δ j ) )
≈ Vi 2 gij ( Tij / bij )
2
As a result, we conclude that Tij + Tji − TLoss = eij where eij is an error whose
variance is approximately the sum of the variances of the two line flow measurements under consideration. If the ratio
eij2
≤ 9.0,
σ T2ij + σ T2ji
then the two measurements under consideration are consistent with one another.
This assures that the combined errors of both measurements are within the limit.
This statement is true with a probability of ≈97%.
2. Real and reactive line flows measured at both ends, together with a voltage
measurement at one end: This information allows computation of the real and
reactive power flows at the end and comparison with the measured values of
the same quantities; easily validates or invalidates the consistency of the measurements under consideration.
3. Bus injection and line flows measured at same and/or opposite bus ends: Since
an injection measurement is equal to the sum of the corresponding line flow
measurements, it is a quick consistency check.
88
STABILITY ANALYSIS TOOLS FOR SMART GRID
4. Local estimators: Dividing the overall network into a set of small observable
networks allows rapid computation of SEs for these small networks to check
for potential locations of bad data.
5. Pseudo-measurements: By definition, the net power injection at the many
unconnected transmission busses is zero and thus is not measured; these zero
injections are exact and can be used in the state estimator as pseudo-measurements
with very small variances; Alternatively, reformulate an optimization problem
to minimize the WLS error subject to equality constraints imposed by the
pseudo-measurements; in either case, the pseudo-measurements are automatically valid and useful measurements that are very useful.
We now discuss the process of pre-estimation analysis.
Denote the set of measurements used in a particular consistency test of the problems above by Si. Let ai2 and ei2 be the respective overall variance and squared error
associated with the test. Let a2 be a consistency threshold such that if ei2 > a 2, then the
set of measurements in Si is inconsistent. Otherwise, it is consistent. If the measurements are consistent they are declared valid. If the tests indicate inconsistency, only
those measurements in Si that have not been previously validated are declared suspect.
If, by this process Si contains strictly one suspected measurement, then that measurement is bad and is deleted from the measurement vector.
Figure 4.11 shows the overall block diagram for pre-estimation analysis. At the
end of the analysis every measurement will either be raw, valid, suspect, or bad. Only
bad measurements are discarded. The variances of valid measurements are slightly
modified to reflect information derived from the consistency analysis. Raw measurements are those for which a consistency test cannot be performed, that is, they do not
belong to any set Si. This is usually the case for nonredundant portions of the measurement system. Finally, suspected measurements will contain the desired set of yetundetermined bad data. The final decision on the suspected measurements is performed
in the postestimation analysis process.
4.10.4
Postestimation Analysis
Postestimation analysis looks at the results of state and parameter estimation and
attempts to establish hypotheses for the most probable causes of poor performance, if
any. This is based on the analysis of the normalized measurement residuals defined as:
ri′ =
zi − h ( xˆ , pˆ )
, i = 1, … , m
ρi
where ρi2 = var ( zi − hi ( xˆ , pˆ ) ).
Obviously ρi2 is the i-th diagonal term of the covariance matrix S defined earlier.
On average, ri′ = 1.0. Statistically, ri′ can vary from zero to three with a high probability. This is true only when all data lie within the specified statistical accuracies. If
89
STATE ESTIMATION
INITIALIZE STATUS OF ALL MEASUREMENTS
Pseudo-measurements: VALID: Others: RAW
i=0
i = i +1
YES
IS THE SET SI CONSISTENT?
DECLARE STATUS OF ALL ELEMENTS OF SI AS
VALID. COMPUTE ADJUSTED VARIANCES
DECLARE STATUS OF ALL NONVALIDATED ELEMENTS OF SI AS
SUSPECT
i = imax
YES
RESET VARIANCES OF VALID
MEASUREMENTS
STOP
Figure 4.11. Flow chart for pre-estimation analysis.
90
STABILITY ANALYSIS TOOLS FOR SMART GRID
there are bad data in the measurements and/or parameters, some of the normalized
residual terms will have a large magnitude. In many cases, the measurement with the
largest normalized residual is a bad measurement, but this has not been mathematically
proven. However, we can show that if measurement zi is bad and if it is a redundant
measurement, its normalized residual will be large. In practice, the computation is
time-consuming. Therefore, without much loss in information, the residuals defined
as: ri ′ = [ zi − h ( xˆ , pˆ )] σ i , i = 1, … , m are analyzed instead of the normalized residuals
defined above. Usually ri2 ≤ 1.0 is on the average. However, a bad measurement which
is redundant will have a large residual.
Without parameter errors and pre-estimation analysis, all measurements are classified as raw with a small probability that some of them are bad. By analyzing the
residuals one can start the hypothesis testing process by discarding the measurement
with the largest residual and performing the estimation process again. If this fails
to yield acceptable estimates, the discarded measurements are put back and the measurement with the largest residual in the new estimate is removed. This process is
repeated until a satisfactory answer is obtained. Obviously, successful identification
of the bad data will occur only if a single bad measurement is present. For two or
more bad measurements this process may fail. In this case, the above process is
repeated by discarding two or more measurements with large residuals at a time and
so on. However, in this case, computational times will start to become unacceptably
excessive.
Hypothesis Testing for the Identification of a Single Bad Data Point.
With noisy parameter values, the preceding bad data detection process will be even less
attractive because the noisy parameters themselves cause large measurement residuals.
Thus, we rely heavily on pre-estimation and bad data analysis. In this case, we assume
that bad measurement data resides with high probability among the suspected measurements (see Fig. 4.12). Normally, suspected measurements in a set Si are highly correlated to one another with minimal or no sensitivity to parameter errors. Thus, we can
implement the algorithm whose flowchart is shown in Figure 4.13. In this algorithm,
following state and parameter estimation stages, the largest unacceptable residual
among suspected measurements in every consistency set Si points to a bad measurement. Small and acceptable residuals of previously suspected measurements cause the
corresponding measurements to be declared valid. This is iterated several times until
the system performance reaches acceptable limits.
4.10.5
Robust State Estimation
We consider an approach to SE that automatically compensates for the effects of bad
data. The basic concept involves the smearing property of WLS estimation. Recall that
in WLS if a measurement is bad the tendency is to spread the effect of the bad measurement residuals over the rest of the system. To develop this concept we write:
91
STATE ESTIMATION
z
State Estimation
ˆ r
z,
Compute
Remove Measurement with
State Estimation
YES
Is Performance
Adequate?
STOP
NO
Put Back Removed Measurements
Figure 4.12. Flow chart for postestimation analysis.
J = ( z − h ( xˆ ) ) R −1 ( z − h ( xˆ ) )
T
m
=
∑
i =1
zi − hi ( xˆ )
σi
2
m
=
∑ (r )
2
i
i =1
m
=
∑J
i
i =1
Assume that if ri for some j is bad, then its real error variance is much larger than the
assumed variance. In trying to minimize J, the WLS algorithm attempts to make the
terms Ji as equal as possible to each other. In robust estimation, the functions Ji are
quadratic in the neighborhood of the origin and flat as the residuals ri become large.
An example is the following Ji:
92
STABILITY ANALYSIS TOOLS FOR SMART GRID
Enter Raw Measurements
Pre-Estimation Analysis
State and Parameter
Estimation
Is Performancea
Adequate?
YES
STOP
NO
Discard Suspected Measurements with Largest
Unacceptable Residual in Every Set Si
NO
Declare as Valid Those Suspected Measurements
with Small Residuals
Figure 4.13. Overall flowchart for online, state, and parameter estimation, and bad data
analysis.
93
STATE ESTIMATION
ri ≤ a
ri2 ,
Ji = 3 1
2
4 a 2 ri 2 − 3a
ri ≥ a
where a is a variable parameter. In this formulation, as ri becomes large, Ji increases
as |ri|1/2. Thus, large residuals will not strongly influence the estimation process. The
above form of Ji requires that x̂ minimize J such that:
m
J=
∑J
i
i =1
The optimality conditions are:
g ( xˆ ) = 0 = −
1
2
m
∂J i
∑ ∂x
i =1
x = xˆ
where
1
− 1 ∂J i = σ i
2 ∂x 1
σ i
∂hi
ri ( x ) ,
∂x
ri ≤ a
3
a 2 ∂hi
ri ( x ) ,
ri ∂x
ri > a
From this we can conclude that if |ri| > 0, the effective standard deviation for the ith
measurement is:
3
r 2
σ i′ = σ i i ,
a
that is, bad measurements will effectively have large deviations of error. The information matrix associated with the above problem is given by:
F (x) =
m
∑ F ( x ),
i
i
where
1 ∂hi ∂hi T
,
2
σ i ∂x ∂x
Fi ( x ) =
3
T
1 a 2 ∂hi ∂hi
σ 2 r ∂x ∂x
i i
The resulting iterative algorithm is given by:
xˆ k +1 = xˆ k + F −1 ( xˆ k ) g ( xˆ k ) .
ri ≤ a
ri > a
94
STABILITY ANALYSIS TOOLS FOR SMART GRID
In the context of pre- and postestimation analysis, robust estimation can be made quite
effective. Basically all validated measurements will have a quadratic Ji function. Only
suspected and raw measurements will have Ji functions of the type above. By extending
this concept to parameter estimation, we can also identify large parameter errors. Figure
4.14 depicts a simplified flow chart for performing robust state and parameter estimation with pre-estimation bad data analysis.
4.10.6
SE for the Smart Grid Environment
These attributes are desirable in the development of future SE for smart grid computational tools:
Enter Raw Data
Pre-Estimation Analysis
Set JI, Functions for Raw/Suspected Measurements and
Suspected Parameters as Non-Quadratic
Perform Robust State
And Parameter Estimation
STOP
Figure 4.14. Flowchart for robust state estimation with pre-estimation analysis.
STATE ESTIMATION
95
1. PMU-based SE includes the following steps suggested by Abur and Zhao [17]:
a) Obtain PMU measurements at precisely controlled instants to estimate slow
errors
b) Measure for magnitude and angle of bus voltages (do not have to estimate
angle)
c) Measure the magnitude and angle of current
d) Detect the state changes to reduce errors in building the bus admittance
matrix
e) Directly compute active and reactive power flows at the substation, or simply
use voltage and current phasors to calculate active, reactive, and apparent
power
In the new environment, the pattern of power flows in the network is less predictable than in the vertically integrated systems, in view of the possibilities associated
with open access and the operation of the transmission network under energy market
rules. Although reliability remains a central issue, the need for real-time network
models us more important due to the new energy market-related functions which have
to be added to new and existing EMS. These models are based on the results yielded
by SE and are used in network applications such as optimal power flow, available
transfer capability, voltage, and transient stability. The new role of SE and other
advanced analytical functions in competitive energy markets is discussed by Shirmohammadi et al. [2]. Based on these network models, operators will be able to justify
their technical and economic decisions, such as congestion management and the procurement for adequate ancillary services, and to identify potential operational problems
related to voltage and transient stability [1].
4.10.7
Real-Time Network Modeling
Real-time models are built from a combination of snapshots of real-time measurements
and static network data. The real-time measurements consist of analog measurements
and the status of switching devices, whereas static network data corresponds to the
system’s parameters and basic substation configurations. Hence, the real-time model is
a mathematical representation of the network’s current conditions extracted at intervals
from SE results. Ideally, SE should run at the scanning rate, for example, every two
seconds, but due to computational limitations, most practical estimators run every few
minutes or when a major change occurs.
4.10.8
Approach of the Smart Grid to State Estimation
SE is an important tool for detecting and diagnosing errors in measurement such as
network error and/or device malfunctions. The technique is used in estimating voltage
and power flow errors as a result of system parameters errors. The problem is formulated as one of the minimum J(x) subject to different constraint conditions and solved
96
STABILITY ANALYSIS TOOLS FOR SMART GRID
by using a DC or AC power model for determining optimum performance. The formulation is reproduced as J = (x −x*) R(x − x*) subject to varying constraints.
To solve this, a typical optimization technique could be employed with the following methodology and algorithm:
1.
2.
3.
4.
Convergence of the algorithm
Clearly defined problem location
Manage/modify noise server
Time and quality of measurements
Components suggested by SE researchers for the smart grid include:
1. Real PMU-based measurements to account for integrated RER/CG and protected load (stochastic) stored
2. A distributed SE around the different (partitioned) network configuration on
appropriate assessment of data performed for subsequent assessment.
Figure 4.15 shows how PMU can be used to enhance the capability of the state
estimator with the PMU schemes by adding phasor measurements with a linear step to
traditional measurements and with a nonlinear iterative procedure. This approach is
concerned with identifying major changes in the PMU in the SE, a final conclusion is
that the data provide direct assessment of the state, which can significantly simplify the
traditional SE and involves no change in existing SE. The changes could be in the
techniques of solution. For example, the problem can be formulated as two subproblems, one composed only of SCADA measurements and another by PMU and SCADA.
The first model is solved in the traditional way. The second can be solved as a local
joint SE, which divides the system into locally observable islands. During the estimation process, based on the method of WLS, the measurements of PMU and SCADA
Traditional State Estimator
Traditional
Measurements
State Estimator with PMU
Synchrophasor
Measurements +
Traditional
Measurements
Linear State
Estimator (Static)
Revised State
Estimator (Dynamic)
Estimated State
Vector
Estimated State
Vector
Figure 4.15. Scheme for state estimation with PMU.
STATE ESTIMATION
97
can be mixed to correct the local state variables from the PMU. When the states of the
PMU are reliable and accurate, the local estimator could be studied for omission. A
system with enough PMU would give a direct measure of itself, or if a small number
of PMUs exist, achieve further approximation in the estimate.
The hybrid of SE and PMU could generate a new data set for security assessment.
Following this stochastic nature we can generate a predictive strategy for data analysis
while suggesting real-time approximate control measures.
Conceptual visualization of an interconnected network can be used for vertically
integrated utilities, pools, and ISOs. Ideally, the control area for which a specific control
center is responsible is observable, although this is not always the case, since parts of
it can be permanently or temporarily unobservable, for example, lower-voltage subnetworks. On the other hand, parts of the network outside the control area, which are
normally unobservable, can be made observable by direct metering or data exchange.
Observable islands are handled with full SE including bad data [18, 19] analysis.
SE can be extended to the rest of the system of interest through the addition of pseudomeasurements based on load prediction and generation scheduling. In executing SE for
this augmented system, care must be taken to avoid corrupting the states estimated from
telemetry data.
Hence, the state estimator is used to build the model for the observable part of the
network and optionally to attach a model of the unobservable part. With adequate
redundancy level, SE can eliminate the effect of bad data and allow the temporary loss
of measurements without significantly affecting the quality of the estimated values. SE
is mainly used to filter redundant data, to eliminate incorrect measurements, and to
produce reliable state estimates, although to a certain extent it allows the determination
of the power flows in parts of the network that are not directly metered. Not only will
traditional applications, such as contingency analysis, optimal power flow, and dispatcher training simulation, rely on the quality of the real-time network model obtained
via SE, but the new functions needed by the emerging energy markets will do so even
more [2].
4.10.9
Dynamic State Estimation
The accuracy of the state estimator is very important because it feeds EMS functions,
that is, voltage/angle stability, economic dispatch, security analysis, and so on. Today’s
power system is clearly more dynamic since both load and source vary with the introduction of RER as distributed resources. A technique to address this new dynamic
system, dynamic state estimation (DSE), is currently being researched [17].
DSE incorporates advanced mathematical modeling, in which time stamping at
instant k is noted, the state for k + 1 is predicted, and the errors are calculated to attain a
single advanced time stamp. Information on one advance time stamp will help system
operators make better decisions. The DSE module uses Holt’s double exponential
smoothing technique for predicting the state vector one time stamp ahead and the
extended Kalman filter technique for filtering. To date, the usual SCADA data is utilized
in lieu of PMU data and therefore the accuracy of the system is not very high.
98
4.10.10
STABILITY ANALYSIS TOOLS FOR SMART GRID
Summary
Chapter 4 has discussed several performance tools for incorporation into smart grid
design. The tools included voltage and angle stability and state estimation. Reviews of
these methods were presented along with the research challenges.
REFERENCES
[1] M.H. Mickle and T.W. Sze. Optimization in Systems Engineering. Scranton, 1972.
[2] J.A. Momoh. Electric Power System Application of Optimization. Marcel Dekker, New
York, 2001.
[3] G. Riley and J. Giarratano. Expert Systems: Principles and Programming. PWS Publisher,
Boston, MA, 2003.
[6] A. Englebrecht. Computational Intelligence: An Introduction. John Wiley & Sons, 2007.
[7] M. Dorigo and T. Stuzle. Ant Colony Optimization. Massachusetts Institute of Technology,
Cambridge, MA, 2004.
[6] P.K. Skula and K. Deb. “On Finding Multiple Pareto-optimal Solutions Using Classical and
Evolutional Generating Methods.” European Journal of Operational Research 2007, 181,
1630–1652.
[7] C.W. Taylor. “The Future in On-Line Security Assessment and Wide-Area Stability Control.”
IEEE Power Engineering Society 2002, 1, 78–83.
[8] “Appendix B2: A Systems View of the Modern Grid-Sensing and Measurement.” National
Energy Technology Laboratory 2007.
[9] T. Bottorff. “PG&E Smart Meter: Smart Meter Program.” NARUC Summer Meeting, 2007.
[10] D. Zhengchun, N. Zhenyong, and F. Wanliang. “Block QR Decomposition Based Power
System State Estimation Algorithm.” ScienceDirect 2005.
[11] A. Jain and N.R. Shivakumar. “Power System Tracking and Dynamic State Estimation.”
[12] Y. Tamura, K. Sakamoto, and Y. Tayama. “Voltage Instability Proximity Index (VIPI) Based
on Multiple Load Flow Solutions in Ill-Conditioned Power System,” Sixth IEEE Conference
on Decision and Control, vol. 3, pp. 2114–2119, August 2002.
SUGGESTED READINGS
A.G. Barto, W.B. Powell, D.C. Wunsch, and J. Si. Handbook of Learning and Approximate
Dynamic Programming. IEEE Press Series on Computational Intelligence, 2004.
M. Dorigo and T. Stutzle. “The Ant Colony Optimization Metaheuristic: Algorithms, Applications
and Advances.” In F. Glover and G. Kochenberger, eds.: Handboook of Metaheuristics.
Norwell, MA, Kluwer, 2002.
R.C. Eberhart and J. Kennedy. “A New Optimizer Using Particle Swarm Theory.” In Proceedings
of the Sixth International Symposium on Micromachine and Human Science, 1995; 39–43.
M. Gibescu, C-C Liu, H. Hashimoto, and H. Taoka. “Energy-Based Stability Margin Computation Incorporating Effects of ULTCs.” In IEEE Transactions on Power Systems 2005, 20.
J.L. Marinho and B. Stott. “Linear Programming for Power System Network Security Applications.” IEEE Transactions on Power Apparatus and Systems 1979, PAS-98, 837–848.
SUGGESTED READINGS
99
B. Milosevic and M. Begovic. “Voltage-Stability Protection and Control Using a Wide-Area
Network of Phasor Measurements.” IEEE Transactions on Power Systems 2003, 18,
121–127.
J. Momoh. Electrical Power System Applications of Optimization. CRC Press, 2008.
A.G. Phadhke. “Synchronized Phasor Measurments in PowerSystems.” IEEE Computer Applications in Power 1993, 6, 10–15.
W.H. Zhange and T. Gao. “A Min-max Method with Adaptive Weightings for Uniformly Spaced
Pareto Optimum Points.” Computers and Structures, 2006, 84, 1760–1769.
L. Zhao and A. Abur. “Multiarea State Estimation Using Synchronized Phasor Measurements.”
IEEE Transactions on Power Systems 2005, 20, 2.
5
COMPUTATIONAL TOOLS
FOR SMART GRID DESIGN
5.1
INTRODUCTION TO COMPUTATIONAL TOOLS
Previous work undertaken in the operation and research, systems engineering, and
computer sciences communities to design various optimization and computational intelligence techniques has already been incorporated into large-scale grids; for example,
the work in References 1–37 using artificial intelligence, heuristic and evolutional
optimization to analyze optimal power flow, power flow, SE, stability, and unit commitment. This chapter underscores the importance of computational tools by discussing
the answers to the following three questions:
1. Are the tools sufficient for modeling and accounting for adequate models of the
system as it incorporates variability and randomness of RER?
2. Can the tools manage stochasticity and randomness in the system?
3. Can the tools address predictivity and the anticipatory nature of the problems
encountered?
The classical optimization tools currently used cannot handle the adaptability and
stochasticity of smart grid functions. Thus, the computational tools and techniques
Smart Grid: Fundamentals of Design and Analysis, First Edition. James Momoh.
© 2012 Institute of Electrical and Electronics Engineers. Published 2012 by John Wiley & Sons, Inc.
100
DECISION SUPPORT TOOLS (DS)
101
required are defined as a platform for assessment, coordination, control, operation, and
planning of the smart grid under different uncertainties. We define the competitive
schemes which are able to handle:
1. Inadequate models of the real world.
2. Complexity and large size of the problems which prohibit computation using
computational intelligence.
3. Solution method employed by the operator which is incapable of being expressed
in an algorithm or mathematical form, usually involves many rules of thumb,
and is limited.
4. Decision-making by operator is based on fuzzy linguistics description.
We next discuss the concepts, algorithms, and some of the drawbacks of the proposed computational tools.
5.2
DECISION SUPPORT TOOLS (DS)
Decision support tools (DS) combining game theory, decision support systems, and
analytical hierarchical processes (AHP) are used for computation of multiobjectives
and risk assessment in smart grid planning and operations. Decision analysis (DA) is
a powerful tool that makes a total uncertainty problem appear as a perfectly rational
decision that is based on numerical values for comparing and yielding fast results. It
looks at the paradigm in which an individual or group decision-maker contemplates a
choice of action in an uncertain environment.
Importantly, the process relies on information about the alternatives. The quality
of information varies from hard data to subjective interpretations, from certainty about
decision outcomes (deterministic information) to uncertain outcomes represented by
probabilities and fuzzy numbers. This diversity in type and quality of information about
a decision problem calls for methods and techniques that can assist in information
processing [7].
DA includes many procedures, methods, and tools for identifying, clearly representing, and formally assessing the important aspects of a decision situation. It computes the recommended course of action by applying the maximum expected utility
action axiom to a well-formed representation of the decision, translating the formal
representation of a decision and its corresponding recommendation into insights for the
decision-maker(s). Multi-criteria decision analysis (MCDA), a form of DA, supports
decision-maker(s) when a problem involves numerous conflicting evaluations. MCDA
highlights these conflicts and derives the path to a compromise in a transparent process.
Analytical hierarchical processing (AHP) is another form of MCDA.
DA support requires two stages: Stage 1: Evaluate the expected monetary value
(EMV) from the profit and loss data and their associated probabilities. Draw
the first decision flow tree which should yield a best decision based on the
highest EMV and/or the lowest expected loss.
102
COMPUTATIONAL TOOLS FOR SMART GRID DESIGN
Stage 2: Consider the possibilities of sampling and accurate information and
reevaluate the new EMV. Draw the new decision flow diagram and include the
tree in stage 1. This should yield a best decision based on the highest EMV and/
or the lowest expected loss.
DA must be implemented with care. If the available data are inadequate to support
the analysis, it is difficult to evaluate the effectiveness, which will lead to oversimplification of the problem. The outcomes of DA are not amenable to traditional statistical
analysis. Strictly, by the tenets of DA, the preferred strategy or treatment is the one that
yields the greatest utility (or maximizes the occurrence of favorable outcomes) no
matter how narrow the margin of improvement. DA is particularly useful in handling
multiobjective functions, or attaining several goals of the smart grid where risk and
possibility are included.
5.2.1
Analytical Hierarchical Programming (AHP)
AHP is a decision-making approach that presents the alternatives and criteria, evaluates
the trade-offs, and performs a synthesis to arrive at a final decision. AHP is especially
appropriate for cases involving both qualitative and quantitative analyses. Figure 5.1
illustrates the AHP method.
AHP has found its widest application in multicriteria decision-making, in planning and resource allocation, and in conflict resolution [7]. In its general form, AHP
is a nonlinear framework for carrying out both deductive and inductive thinking
without use of the syllogism by considering several factors simultaneously, allowing
for dependence and feedback, and making numerical tradeoffs to arrive at a synthesis
Cognitive
differences
Action
Possibly poorly
defined problem
Action
Action
Uncertain
Multi-dimension
outcomes
Value of
Outcomes
Action
Unknown causes
& event
Random
Variations
Figure 5.1. Analytical Hierarchical Programming (AHP).
Different
constituency
103
CLASSICAL OPTIMIZATION METHOD
or conclusion. The composite priorities of each alternative at the bottom level of a
hierarchy may be represented as a multilinear form:
∑
i1
i2
ip
x1 x2 … x p
i1,…,i p
Consider a single term of this sum; for simplicity, denote it by x1, x2 . . . xp. The
product integral is given by:
n
n
x1 x2 ⋯ x p = e log x1 x2 ⋯ x p =
∏e
log xi
∑ log xi
= e i=1
i =1
5.3
→e
∫
log x (α ) dα
OPTIMIZATION TECHNIQUES
Optimization techniques consist of static and dynamic techniques for optimization, such
as linear programming, mixed integer, dynamic programming, and so on, for development of smart grid optimization and planning activities (see Fig. 5.2). Linear programming, nonlinear mixed-integer programming (MIP), dynamic programming (DP) and
Lagrangian relaxation methods are used for power system and operation issues, but
they are limited for use in the smart grid due to the static network of the programs they
can solve. They work better when computed in conjunction with DS tools and computational tool techniques. Below is a summary of the highlights of their formulation,
implementation, and problem process.
5.4
CLASSICAL OPTIMIZATION METHOD
5.4.1
Linear Programming
Linear programming uses a mathematical model to describe the problem with linear
objectives and linear constraints. The general structure of problems solved by this
method is:
Maximize cTx
s.t. Ax ≤ b
and xi ≥ 0 ∀i ∈ {1, n}
Assume that the model is statically linear. The process to achieve the global
optimum uses Simplex-like techniques, variants of the interior point method, or integer
programming. These methods are applicable to problems involving linear objective
functions and linear constraints. The solution procedure includes:
1. Initialization Step: Introduce slack variables (if needed) and determine initial
point as a corner point solution of the equality constraints.
2. At each iteration, move from the current basic feasible solution to a better,
adjacent basic feasible solution.
104
Optimization Techniques
Decision System
Decision
Analysis (DA)
Analytical
Hierarchical
Programming
(AHP)
Analytical
Network Process
(ANP)
Cost Benefit
Analysis (CBA)
Sensitivity Analysis
Risk Assessment
Game Theory
Adaptive Dynamic
Programming(ADP)
Static
Linear
Programming (LP)
Non linear
Programming
(NLP)
Integer
Programming
(IP)
Barrier Method
Intelligent Systems
Neural Dynamic
Programming
Heuristic
Programming
Expert Systems
Heuristic
Dynamic
Programming
(HDP)
Particle Swarm
Optimization
Fuzzy Logic
System
Ant Colony
Optimization
Artificial Neural
Network (ANN)
Dual Heuristic
Programming
(DHP)
Quadratic
Programming
(QP)
Dynamic
Programming
Evolutionary
Programming
Action
Dependent
Heuristic
Programming
(ADHDP)
Action
Dependent Dual
Heuristic
Programming
(ADDHP)
Figure 5.2. Optimization techniques.
Tabu Search
Annealing
Method
Pareto Multiples
Methods
CLASSICAL OPTIMIZATION METHOD
105
3. Determine the entering basic variable: Select the nonbasic variable that, when
increased, would increase the objective at the fastest rate. Determine the leaving
basic variable: Select the basic variable that reaches zero first as the entering
basic variable is increased.
4. Determine the new basic feasible solution.
5. Optimality Test and Termination Criteria: Check if the objective can be increased
by increasing any non-basic variable by rewriting the objective function in terms
of the nonbasic variables only and then checking the sign of the coefficient of
each non-basic variable. If all coefficients are nonpositive, this solution is
optimal; stop. Otherwise, go to the iterative.
Appling the linear programming method to the smart grid requires improving the
it to accommodate the grid’s stochasticity, predictivity, adaptivity, and randomness. The
traditional linear method is confined to static problems and thus is insufficient for smart
grid implementation.
5.4.2
Nonlinear Programming
Nonlinear programming (NLP) typically employs Lagrangian or Newtonian techniques
for constrained and unconstrained optimization problems. The approach assumes that
all objective functions are modeled as smooth and continuous functions. However, the
mathematical response of the power system diverts from this assumption in many
instances.
The general structure of problems solved by this method is:
Maximize f(y)
Subject to
Ci ≤ fi(y) ≤ Di ∀i ∈ {1, n}
The procedure used in solving the NLP class of problems stems from identifying
the feasibility of satisfaction. Convergence is due to sufficiency condition as given in
work defined by extended KT conditions. Reference 36 provides a detailed formulation
and methodology to solve the NLP class of problems. The sensitivity method, barrier
method, and quadratic programming all feature in this class. Briefly stated, for all NLP:
1. Determine the initial feasible set based on investigation of extrema of the functions with or without constraints.
2. Check the optimality conditions.
3. Determine candidate solution for local or global optimum.
4. Perform further optimization and evaluate the optimal value to the objective
function that satisfies the constraints.
This process may involve the application of Kuhn-Tucker (KT) and Karush-KuhnTucker (KKT) first and second order necessary and sufficient conditions [36, 37]. It
106
COMPUTATIONAL TOOLS FOR SMART GRID DESIGN
can be applied to functions as well as functional. NLP problems can be classified based
on the nature of the objective functions and constraints.
The two challenges of NLP are its large computational burden and its limitation
to static variables in the objectives and constraints. Moreover, NLP including QP often
suffers if the approximations to the actual models for these devices are not sufficiently
accurate.
Apart from being a very common form for many important problems, QP is also
useful because many of the problems in a power system are often solved as a series of
QP problems or Sequential Quadratic Programming (SQP).
Again, as in LP, the nonlinear programming classes are not suitable or designed to
handle variability and predictivity. Therefore, research is needed to include this smart
grid feature if they are to be used as computational tools. More work is needed to
improve this technique to include the special requirement for the smart grid optimization process which requires adaptive, predictive, and stochastic algorithms.
5.4.3
Integer Programming
This is a special case of LP where all or some of the decision variables are restricted
to discrete integer values, for example, where the discrete values are restricted to zero
and one only, that is, yes or no decisions, or binary decision variables. The general
structure of the MIP problem is:
n
Maximize P( x ) =
∑c x
j i
j =1
≤ bi
i =1
i =1
And xj ≥ 0 ∀ j ∈{1, n}, and xj is an integer ∀i ∈ {1, I}.
Subject to the constraints:
m
n
∑ ∑a x
ij
j
The branch-and-bound procedure features:
1. Initialization: Set P* = −∞, where P* is the optimal value of P.
2. Branching: This step involves developing subproblems by fixing the binary
variables at 0 or 1, or choosing the first element in the natural ordering of the
variables as the branching variable.
3. Bounding: For each of the subproblems, a bound can be obtained to determine
the goodness of its best feasible solution. For each new subproblem, obtain its
bound by applying the Simplex method to its LP relaxation and use the value
of the P for the resulting optimal solution.
4. Fathoming: If a subproblem has a feasible solution, it should be stored as the
first incumbent (the best feasible solution found so far) for the whole problem
along with its value of P. This value is denoted P*, which is the current incumbent for P.
5. Optimality Test: The iterative procedure stops when no subproblems remain. At
this stage, the current incumbent for P is the optimal solution.
CLASSICAL OPTIMIZATION METHOD
107
Pure integer or MIP problems pose a great computational challenge. While highly
efficient LP techniques can enumerate the basic LP problem at each possible combination of the discrete variables (nodes), the problem lies in the astronomically large
number of combinations to be enumerated. If there are N discrete variables, the total
number of combinations becomes 2N! The branch-and-bound technique for binary
integer or reformulated MIP problems overcomes this challenge by dividing the overall
problem into smaller and smaller sub-problems and enumerating them in a logical
sequence. Note that the direct use of these techniques for solving the smart grid optimization problem will be limited, because they are generally static and are not designed
for handling real-time and dynamic optimization problems.
5.4.4
Dynamic Programming
This approach was developed to solve sequential, or multistage, decision problems.
Basically, it solves a multivariable problem by solving a series of single variable problems. This is achieved by tandem projection onto the space of each of the variables. In
other words, it projects first onto a subset of the variables, then onto a subset of these,
and so on. It is a candidate optimization technique for handling time variability and
noise in the objective and constrained optimization problem.
Two common techniques derived from Bellman principles are
1 Backward and forward recursion or table look up
2 Calculus based on composition
Both techniques have the drawback of the curse of dimensionality and thus are not
suitable for large, complex power systems.
5.4.5 Stochastic Programming and Chance Constrained
Programming (CCP)
Stochastic programming solves LP problems where the uncertainty assumption is so
badly violated that the same parameters must be treated explicitly as random variables.
The two ways to handle LP with variability are:
1. Stochastic programming (SP)
2. Chance-constrained programming (CCP)
SP requires all constraints to hold with probability whereas CCP permits a small
probability of validating any functional constraint.
Formulation of Stochastic Programming. Consider max Z = ∑ c j x j between
the limits 1 and n. Because we can replace Z by some deterministic function, we pose
it as E ( Z ) = ∑ cjxj between the limits 1 and n. Similarly, the functional constraints
∑ aij x j ≤ b j for i = 1, 2, 3, 4 … m. The solution requires feasibility requirements of the
constraints for all possible combinations of parameters in aij, bj which must be assigned
to xj.
108
COMPUTATIONAL TOOLS FOR SMART GRID DESIGN
The problem is solved in two categories:
1. One-stage problem: Here, aij, bj are random variables and mutually independent
where both aij, bj must satisfy: ∑ max aij x j ≤ min b j
2. Multistage problem: Here, xj have 2 or more points in the time stage. xj are a
firststage variable, others are secondstage variables, and so on. This method is
based on first obtaining the first time and the subsequent time depending on
what happens in the preceding stages. Therefore, SP for multistage problems is
more adjustable for later stages decisions based on unfolding encounters.
When some of the parameters of the model are random variables (RVs), the SP
formulation requires that all the functional constraints must hold for all possible combination of values for RV parameters. However, the CCP formulation requires only that
each constraint be held for most of the combinational period. This formulation replaces
the original LP constituents of ∑ aij xj ≤ b j by P{∑ aij x j ≤ b j } ≥ α1 where α1 are specified
between zero and one. They are normally chosen to be reasonably close to one. Therefore, a non-negative solution (x1, x2, x3, . . . , xn) is considered feasible if and only if
P{∑ aij x j ≤ b j } ≥ α1 for n = 1, 2, 3 . . . n. Each contingency probability 1 − α1 represents the allowable risk such that the RV will take on value stated as aijxj ≤ bj for j = 1,
2, 3, . . . . n. Thus, the best objective is to select the best non-negative solution that
will likely satisfy each of the original constraints when RV (aij, bi, ci) take on the values.
In general, the procedure for computing CCP, LP includes all aij parameters which
must be correlative so that bi, ci are RV probability distribution of bi and cj is slightly
independent of bj (where j = 1, 2, 3, 4 . . . . n; for i = 1, 2, 3, . . . m).
SP is positioned to solve network, transportation, and power grid problems with
uncertainties, randomness, and noise. To date, work on SP is able to handle small-sized
systems, but is still unable to account for randomness and predictivity.
5.5
HEURISTIC OPTIMIZATION
Future work that accounts for the predictive and stochastic nature of the smart grid involves:
1. Modeling components to account for predictivity and stochasticity
2. Selecting new optimization methods such as adaptive dynamic programming
(ADP)
1. The security of power systems is affected by various contingencies which may
move the system from normal and alert to emergency states. Their impacts are
studied with the optimization tools under development. The limitations to be
resolved include the fact that a model cannot account for uncertainty in load,
central location, and sources.
2. Selectivity of contingency sets appropriate for study (in real time for offline
study).
3. Accounting for stability of the solution and model price and economic marginal
benefits of new RER.
HEURISTIC OPTIMIZATION
109
A new optimization strategy is proposed:
1. Model a new objective function to account for customer and welfare power
uncertainty.
2. Model RER to account for stochasticity and variability.
3. Update the model and simulate.
4. Define new system components’ performance in time.
5. Following this, solve the probabilistic load flow for base case studies and
include new contingency set to define violations.
6. Select appropriate optimization technique which may include heuristic program
technique and hybrid or ADP, methods that define the impact of predictivity and
stochasticity as optimum.
5.5.1
Artificial Neural Networks (ANN)
ANN are based on the natural genetics of the brain. Common techniques include back
and forward propagation techniques. ANN have the ability to classify and recognize
patterns in a large quantity of data through training and tuning of the algorithm. The
key element of this paradigm is the novel structure of the information processing
system, which comprises a large number of highly interconnected processing elements
(neurons) working in unison to solve specific problems. ANN learn by example. They
are configured for a specific application, such as pattern recognition or data classification, through a learning process.
ANN, with their remarkable ability to derive meaning from complicated or imprecise data, can be used to extract patterns and detect trends that are too complex to be
noticed by either humans or other computer techniques. A trained neural network can
be thought of as the expert in the category of information it has been given to analyze.
The expert provides projections, given new situations of interest and answers the “what
if ” questions.
Other advantages of ANN include:
1. Adaptive learning: Ability to learn how to do tasks based on the data given for
training or initial experience.
2. Self-organization: Creates its own organization or representation of the information received during learning period.
3. Real-Time Operation: Computations may be carried out in parallel; special
hardware devices are being designed and manufactured to exploit this
capability.
4. Fault Tolerance via Redundant Information Coding: Partial destruction of a
network leads to the corresponding degradation of performance.
Back-propagation nets are probably the most well-known and widely applied of
the neural networks in use today. The net is a perceptron with multiple layers, a
110
COMPUTATIONAL TOOLS FOR SMART GRID DESIGN
different threshold function in the artificial neuron, and a more robust and capable
learning rule. A unit in the output layer determines its activity in two steps:
Step 1. It computes the total weighted input xj, using the formula: X j =
∑yW
i
ij
where yi is the activity level of the jth unit in the previous layer and Wij is the weight
of the connection between the ith and the jth unit.
Step 2: The unit calculates the activity yj using some function of the total weighted
input. Typically we use the sigmoid function:
1
1 + e− x j
yj =
Once the activities of all output units have been determined, the network computes the
error E, which is defined by the expression:
E=
1
2
∑ (y − d )
2
i
i
i
where yj is the activity level of the jth unit in the top layer and dj is the desired output
of the jth unit.
The back-propagation algorithm consists of four steps:
Step 1: Compute how fast the error changes as the activity of an output unit is
changed. This error derivative (EA) is the difference between the actual and the
desired activity.
EA j =
∂E
= yj − dj
∂y j
Step 2: Compute how fast the error changes as the total input received by an output
unit is changed. This quantity (EI) is the answer from Step 1 multiplied by the
rate at which the output of a unit changes as its total input is changed.
EI j =
∂E ∂E dy j
=
×
= EA j y j (1 − y j )
∂x j ∂y j dx j
Step 3: Compute how fast the error changes as a weight on the connection into an
output unit is changed. This quantity (EW) is the answer from Step 2 multiplied
by the activity level of the unit from which the connection emanates.
EWij =
∂E
∂E ∂x j
= EI j yi
=
×
∂Wij ∂x j ∂Wij
111
HEURISTIC OPTIMIZATION
Step 4: Compute how fast the error changes as the activity of a unit in the previous
layer is changed. To compute the overall effect on the error, add all of these
separate effects on output units.
EAi =
∂E
=
∂yi
∂E
∑ ∂x
j
j
×
∂x j
=
∂yi
∑ EI W
j
ij
j
Power system applications include: adaptive control, fault detection and classification, network reconfiguration, voltage stability assessment, and transient stability
assessment. These functions require real-time data for monitoring and control. The
deployment of ANN as a computational platform will facilitate the assessment of smart
grid performance for each of the above applications.
5.5.2
Expert Systems (ES)
ES are used as a method of optimization that relies on heuristic or rule-driven decisionmaking. They are sometimes used for fault diagnosis with prescription for corrective
actions. While the expert system/computer application performs a task that would
otherwise be performed by a human, the method is only as reliable as the designed
engineering rule-base. Figure 5.3 depicts the components.
Expert systems have several advantages over human experts, including: increased
availability and reliability, lower cost and response time, increased confidence in
decision-making ability via provision of clear reason for a given answer. Its uses for
the power system include optimal load shedding, resource allocation such as VAr,
discrete control (series capacitors, ULTCs, and so on), and economic dispatch. The
operator-assisted functions in the management of the smart grid will benefit from the
use of this knowledge-based system or its hybrid with ANN, or other variations of
computational/heuristic optimization techniques. Some of the problems can be reformulated using expert systems, real time data, and hybrid ANN.
KNOWLEDGE BASE
FACTS
USER
EXPERTISE
INFERENCE ENGINE
Expert System
Figure 5.3. Fundamental components of an expert system.
112
COMPUTATIONAL TOOLS FOR SMART GRID DESIGN
START
Fuzzyfication
Decision
Making
Defuzzyfication
END
Figure 5.4. Simplified block diagram of fuzzy logic approach.
Fuzzy logic is a superset of conventional logic that has been extended to handle
the concept of partial truth, which are the truth values between “completely true” and
“completely false.” Special membership functions transform uncertainty in data to a
crisp form for analysis. Fuzzy-logic control is often used to determine whether the
process variables are within acceptable tolerances. A fuzzy set is a generalization of
ordinary sets that allows assigning a degree of membership for each element to range
from (0, 1) interval. Figure 5.4 depicts the simplified block diagram of the fuzzy logic
approach.
A fuzzy system combines and applies sets with fuzzy rules to problems with overall
complex nonlinear behavior. In smart grid design, many real-time decisions require the
attribute of fuzziness. The design of automatic generation control (AGC), SSA, and SE
will therefore be used for proposed smart grid functions and control.
5.6
EVOLUTIONARY COMPUTATIONAL TECHNIQUES
Based on natural genetics, EC solve combinatorial optimization problems. The techniques in this category, including particle-swarm, ant-colony, genetic algorithms (GA),
and artificial intelligence, learn or adapt to new situations, generalize, abstract, discover,
and associate. The evolutionary algorithms use a population of individuals. An individual is referred to as a chromosome which defines the characteristics of individuals
in the population. The characteristic of each individual is termed a gene. Individuals
with the best survival capabilities have the best chance to reproduce [7]. Offspring are
generated by combining parts of the parents, a process referred to as crossover. Each
individual in the population can also undergo mutation which alters some of the alleles
of the chromosome.
5.6.1
Genetic Algorithm (GA)
GA mimics biological evolution such that the elements in the algorithm are synonymous
with genetic system terminology. Figure 5.5 from References 7 and 8 shows a typical
GA in which offspring are produced from selected parents, modified through crossover
or mutation, and evaluated to find the fittest offspring. They are placed in the population
to become parents while the unfit offspring are discarded. The process can be repeated
until a suitable offspring or solution is created.
GA has been applied to power system expansion and structural planning, operation
planning, and generation, transmission, and distribution operation and analysis for VAr
planning requiring real-time operation with uncertainty and randomness.
113
EVOLUTIONARY COMPUTATIONAL TECHNIQUES
reproduction
children
modification
modified
children
parents
evaluation
population
evaluated children
deleted
members
discard
Figure 5.5. Typical GA cycle.
5.6.2
Particle Swarm Optimization (PSO)
(PSO) is a population-based stochastic optimization technique developed by Eberhart and
Kennedy in 1995. It uses a simple mechanism that mimics the social behavior of bird
flocking and fish schooling to guide the particles’ search for globally optimal solutions.
Individuals in a particle swarm follow a very simple behavior: to emulate the
success of neighboring individuals and their own successes. The collective behavior
that emerges is that of discovering optimal regions of a high dimensional search space
[5]. The population of the study is referred to as the population and each individual is
termed a particle. The equation shows that for each time, t, the position of the particle
xi is altered by adding a velocity, vi(t):
xi (t + 1) = xi (t ) + vi (t + 1)
The velocity adjusts the particle based on both the particle and its neighbors within
the swarm, which is calculated as:
vij (t + 1) = vij (t ) + c1r1 j (t ) [ yij (t ) − xij (t )] + c2 r2 j (t ) [ yˆ j (t ) − xij (t )]
where vij(t) is the velocity of particle i in dimension j = 1, . . . , n, xij(t) is the position
of particle i in dimension j at time step t, c1 and c2 are the positive acceleration constants
used to scale the contribution of the cognitive and social components, respectively
(discussed in Section 16.4), and r1j(t), r2j(t) are the random values in the range (0, 1)
sampled from a uniform distribution which introduce a stochastic element to the
algorithm.
Two fundamental variants of PSO are Local Best (lbest) and Global Best (gbest).
The lbest provides a local best, utilizing a ring network topology and a subset of the
swarm as the neighborhood of the particle. The gbest presents the global best where
the neighborhood for the particle taken to be the swarm.
5.6.3
Ant Colony Optimization
ACO is a class which is applied to combinatorial optimization problems. The essential
trait of ACO algorithms is the combination of a priori information about the structure
114
COMPUTATIONAL TOOLS FOR SMART GRID DESIGN
of a promising solution with a posteriori information about the structure of previously
obtained good solutions [6, 7]. ACO uses computational concurrent and asynchronous
agents termed a colony of ants that move through states of the problem corresponding
to the partial solutions. The measurement generally involves a stochastic local decision
policy based on the two parameters, trails and attractiveness. The pheromone information will direct the search of the future. During the construction phase or upon completing the solution, the ant evaluates the solution and modifies the trail value of the
components used in its solution.
An ACO algorithm includes trail evaporation and daemon actions. The trail evaporation action decreases all trail values over time, in order to avoid unlimited accumulation of trails over some component. Conversely, daemon actions implement centralized
actions which cannot be performed by single ants, such as the invocation of a local
optimization procedure, or the update of global information used to decide whether to
bias the search process from a nonlocal perspective [7]. The process is:
begin
Intialize τij(0)
Let t = 0;
Set location, nk = origin node;
while stopping criteria is not true
for k = 1, . . . , nk
// Construct a path, xk(t);
xk(t) = 0;
for destination node is reached;
remove loops from xk(t);
calculate f(xk(t));
end
end
for each link(i, j) of the graph
//pheromone evaporation;
τij(t) = (1 − ρ)τij(t)
end
for k = 1, . . . , nk
for link(i, j) of xk(t)
∆τ k =
1
( (
k
f x t lim
δ x→0
τ ij (t + 1) = τ ij (t ) +
))
nk
∑ ∆τ
k =1
end
k
ij
(t )
115
ADAPTIVE DYNAMIC PROGRAMMING TECHNIQUES
end
t = t + 1;
end
return xk(t) for smallest f(xk(t)).
5.7
ADAPTIVE DYNAMIC PROGRAMMING TECHNIQUES
ADP incorporates time dependency of the deterministic or stochastic data required for
the future. ADP is also termed reinforcement learning, adaptive critics, neural-dynamic
programming, and approximate dynamic programming [6, 7] (see Fig. 5.6). ADP consider the optimization over time by using learning approximation to handle problems
that severally challenge conventional methods due to their very large scale and lack of
sufficient prior knowledge. ADP overcomes the curse of dimensionality in DP. ADP is
designed to maximize the expected value of the sum of future utility over all future
time periods:
∞
Maximize
∑ (1 + r )
−k
U (t + k )
k =0
Some common nomenclature in ADP is:
u(t): Action vectors
U(t): Utility which the system is to maximize
X(t): Sensor inputs
r: Discount rate or interest rate (needed only in infinite-time-horizon problems)
J(t)
Critical Network
X(t)
Dynamic Model of
System
U (t)
Action Network
Derivatives
calculated by
back
propagation
Figure 5.6. Structure of adaptive dynamic programming (ADP) system.
116
COMPUTATIONAL TOOLS FOR SMART GRID DESIGN
<>: Denote the expectation value
J: Secondary or strategic utility function
R(t): Complete state description of the plant to be controlled at time t
A: Action network
F_Wij: Derivatives of error with respect to all weights Wij
Adaptive critic designs (ACDs) use neural networks based designs for optimization
over time using combine concepts of reinforcement learning and approximate dynamic
programming.
ACDs use two neural networks, the critic and action networks, to solve the
Hamilton-Jacobi-Bellman equation of optimal control. The critic network approximates
the cost to-go function J of Bellman’s equation of dynamic programming and is referred
to as the heuristic dynamic programming (HDP) approach in ACDs.
The action network provides optimal control to minimize or maximize the cost-togo function J.
ACD use combined concepts of reinforcement learning and approximate dynamic
programming. Two neural networks, the critic and action networks, are used to solve
the Hamilton-Jacobi-Bellman equation of optimal control. The critic network approximates the cost-to-go function J of Bellman’s equation of dynamic programming,
termed the HDP approach, while the action network provides optimal control to minimize or maximize the cost-to-go function J.
Several critic designs have been proposed based on dynamic programming:
•
Heuristic dynamic programming (HDP) adapts a critic network whose output is
an approximation of J(R(t))H:
J ( R(t )) = Max(U ( R(t ), u(t )) +
u(t )
•
< J ( R(t + 1)) >
1+ r
Dual Heuristic programming (DHP) adapts a critic network whose outputs represent the derivatives of J(R(t)):
J ( R(t )) = (U ( R(t ), u(t )) + < J ( R(t + 1)) > − U 0
∂J ( R(t ))
∂
∂J ( R(t + 1))
>
=
U ( R(t ), u( R(t )) + <
∂Ri (t )
∂Ri (t )
∂Ri (t )
∂J ( R(t + 1)) ∂R j (t + 1)
∂J ( R(t ), u(t ))
∂U ( R, u) ∂u j R(t )
+
=
+
⋅
⋅
∂
∂Ri (t )
u
R
t
(
)
∂Ri (t )
∂
∂Ri (t + 1)
j
i
j
j
λi ( R(t )) =
∑
+
∑
j ,k
•
∑
∂J ( R(t + 1) ∂R j (t + 1) ∂uk (t )
⋅
⋅
∂uk (t ) ∂R j (t )
∂R j (t + 1)
Globalized DHP (GDHP) adapts a critic network whose output is an approximation of J(R(t)) to minimize errors in the implied derivatives of J. GDHP tries to
combine the best of HDP and DHP.
117
PARETO METHODS
HDP intends to break down through very slow learning as the problem size grows,
however, DHP is more difficult to implement. The three methods all yield actionindependent critics, and there are also ways to adapt a critic network that inputs R(t)
and u(t) (ADHDP, ADDHP).
Action dependent heuristic dynamic programming (ADHDP or Q-learning) adapts a
critic network, J'(R(t),u(t),W ), which attempts to approximate J' as defined in equations:
J ′( R(t ), u(t )) = U ( R(t ), u(t )) +
J ′( R(t ), u(t )) = U ( R(t ), u(t )) + Max
u ( t +1)
5.8
< J ( R(t + 1)) >
1+ r
< J ′( R(t + 1), u(t + 1)) >
1+ r
PARETO METHODS
Pareto analysis is used for solving the multiobjective optimization problem:
min F ( x )
x ∈N
where F(x) = (f1, f2, . . . , fm)T
s.t
g(x) = 0
h(x) ≤ 0
a≤x≤b
The goal is to seek Pareto-optimal solutions through various approaches. Due to
the conflicting nature of the objective functions based on the concepts of convergence
and diversity, solutions to the above problem are both multiple and Pareto [3] optimal.
By definition, a feasible solution X* is Pareto optimal if there is no other improved
feasible point X such that fk(X ) fk(X*) with strict inequality for at least one condition.
Two methods for solving such problems are:
•
•
One-at-a-time strategy: A multiobjective optimizer may be applied repeatedly
with the goal each time of finding one single Pareto-optimal solution. These are
classical generating multiobjective optimization methods which can be static or
deterministic and which use such an iterative scalarization scheme of standard
procedures, such as weighted-sum, ε-constraint method, trade-off method, or the
min-max method [5, 6]. The drawbacks include inefficiency as well as the difficulty in maintaining diversity in the objective space. The solution to every
subproblem involves contending with infeasible region and local optimums
(which may feature in every step of the solution). Overcoming such difficulties
requires an optimization algorithm to learn how to solve the problem independently each time for a different initial point [2].
Simultaneous strategy: This approach utilizes EA due to the population or
archive-based approach to facilitate a parallel search. The efficiency of this
method is greatly improved due to the reduced need for multiple applications of
118
COMPUTATIONAL TOOLS FOR SMART GRID DESIGN
TABLE 5.1. Hybridized Computational Intelligence Tools for Smart Grid Analysis and
Design.
Fuzzy
Fuzzy
Genetic Algorithm (GA)
Ant colony optimization (ACO)
Artificial Neural Networks (ANN)
Evolution programming (EP)
Particle Swamp Optimization (PSO)
Adaptive Dynamic Programming (ADP)
X
X
X
X
X
X
GA
AOC
ANN
EP
PSO
ADP
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
the optimization. Critical information (including the information needed to
solve a subproblem) is shared among the population through exchange (or
recombination).
Pareto can be used to augment static optimization techniques.
5.9 HYBRIDIZING OPTIMIZATION TECHNIQUES AND
APPLICATIONS TO THE SMART GRID
To facilitate the computational methods we incorporate hybrid of intelligent systems
such as immunized-neuro systems, immunized-swarm systems, neuro-fuzzy systems,
neuro-swarm systems, fuzzy-PSO systems, fuzzy-GA systems, and neuro-genetic
systems. These hybrids (Table 5.1) are necessary to harness the advantages of the
various paradigms of computational intelligence [7] for addressing the smart grid functions and capability to achieve self-learning capability.
These advanced techniques have a host of characteristics which make them viable
for application to the smart grid environment, such as handling the nonlinear modeling
of time varying dynamics of the power system, and handling dynamics and stochasticity. These techniques will be utilized for FACTS controllers, learning for systems,
training sets for voltage profile, unit commitment, discrete adjustment of controls,
power system planning, and parallel computing.
5.10
COMPUTATIONAL CHALLENGES
The computational challenges associated with utilizing advanced tools involve their
optimal selection based upon the specific application and location. Data availability is
another challenge, enforcing the need to develop sensor and communication technology
to facilitate the acquisition of real-time or just-in-time data.
Some of the key questions to be answered are:
•
Software/hardware: Is the tool to be developed software only or is some hardware required for integration and implementation?
SUMMARY
•
•
•
•
•
119
Integration: How will software/hardware be integrated into the system? Will it
replace an existing tool or be integrated into an existing package/tool?
Location of installation: Where will a tool be located and used; for example,
LAN or WAN? Will it be necessary at all generating points or load points, at
substation levels, or at the customer level?
Robustness: How will the tool be implemented for ease of application and
robustness?
Sensitivity: What degree of sensitivity is required for satisfactory functioning?
Standards: Have existing standards been identified or developed for utilization
of the tool in similar environments?
In practice, the tools will be used by engineers and operators who lack knowledge
and training. Thus, the advanced tools must be readily interpretable, user friendly, and
self-teaching.
An expanded power system, deregulation (liberalization), and environmental concerns produce new challenges for operation, control, and automation. Power system
models used for intelligent operation and control are highly dependent on the task
purpose. Smart solutions to economic, technical (secure, stable, and good power
quality), and environmental goals, must incorporate forecasting of load, price, and
ancillary services; penetration of new and renewable energy sources; bidding strategies
of participants; system planning and control; operating decision-making under missing
information; increased distributed generations and demand response; tuning controller
parameters in different operating conditions, and so on.
Risk management and financial management in the electric sector attempt to find
the ideal trade-off between maximizing the expected returns and minimizing the risks
associated with investment. Computational intelligence (CI) is a new tool for solving
the complex problems posed by smart grid technology. For example, heuristic optimization techniques have been combined with elements of nature-based methods or those
with a foundation in stochastic and simulation approaches. Research occurring in the
field of intelligent system technologies is using digital signal processing techniques,
expert systems, artificial intelligence, and machine learning. Developing solutions with
these and other such tools offers two major advantages: development time is much
shorter, and the systems are relatively insensitive to nois and/or missing data/information
known as uncertainty. Due to environmental, rights-of-way, and cost problems, there
is increased interest in better utilization of the capacity available in both bundled and
unbundled power systems.
5.11
SUMMARY
Chapter 5 has summarized the classical optimization techniques and computational
methods currently being applied in legacy and smart grid design, planning, and operations. The Pareto methods used in grid design were described. A discussion of computational challenges highlighted several problems faced by researchers and engineers in
determining the best mitigation and solution approaches.
120
COMPUTATIONAL TOOLS FOR SMART GRID DESIGN
REFERENCES
[1] H.W. Dommel and W.F. Tinney. “Optimal Power Flow Solutions,” IEEE Transactions On
Power Apparatus and Systems, 1968, PAS.87, 1866–1876.
[2] W.L. Winston. Operations Research: Applications and Algorithms, Boston: Duxbury, 1987.
[3] J.A. Momoh. Electric Power System Application of Optimization, New York: Marcel
Dekker, 2001.
[4] Y.L. Abdel-Magid, M.A. Abido, and A.H. Mantaway. “Robust Tuning of Power Systems Stabilizers in Multimachine Power Systems,” IEEE Transactions on Power Systems 2000, 15, 735–740.
[5] M. A. Abido. “Robust Design of Multimachine Power System Stabilizers Using Simulated
Annealing.” IEEE Transactions on Energy Conversions 2000, 15, 297–304.
[6] R.A.F. Saleh and H.R Bolton. “Genetic Algorithm-aided Design of a Fuzzy Logic Stabilizer for
a Superconducting Generator,” IEEE Transactions on Power Systems 2000, 15 1329–1335.
[7] P. Zhang and A.H. Coonick. “Coordinated Synthesis of PSS Parameters in Multi-machine
Power Systems Using The Method of Inequalities Applied to Genetic Algorithms,” IEEE
Transactions on Power Systems 2000, 15, 811–816.
[8] Y.L. Abdel-Magid, M.A. Abido, S. Al-Baiyat, and A.H. Mantawy. “Simultaneous Stabilization of Multimachine Power Systems Via Genetic Algorithms,” IEEE Transactions on
Power Systems 1999, 14, 1428–1439.
[9] A.L.B. do Bomfim, G.N Taranto, and D.M Falcao. “Simultaneous Turning of Power System
Damping Controllers Using Genetic Algorithms,” IEEE Transactions on Power Systems
2000, 15, 163–169.
[10] J. Wen, S. Cheng, and O.P. Malik. “A Synchronous Generator Fuzzy Excitation Controller
Optimally Designed With a Genetic Algorithm,” IEEE Transactions on Power Systems
1998, 13, 884–889.
[11] M.A. Abido and Y.L. Abdel-Magid. “Hybridizing Rule-based Power System Stabilizers
With Genetic Algorithms,” IEEE Transactions on Power Systems 1999, 14, 600–607.
[12] R. Asgharian and S.A. Tavakoli. “A Systematic Approach to Performance Weights Selection
in Design of Robust H/sup/spl infin/PSS Using Genetic Algorithms,” IEEE Transactions on
Energy Conversion 1996, 11, 111–117.
[13] K.Y. Lee and J.G. Vlachogiannis. “Ant Colony Optimization For Active/reactive Operational Planning,” World Congress, vol. 16, 2005.
[14] M.R. Kalli, I. Musirin, and M.M. Othman. “Ant Colony Based Optimization Technique for
Voltage Stability Control,” Proceedings of the 6th WSEAS International Conference on
Power System, 2006.
[15] M.R. Kalli, I. Musirin, and M.M. Othman. “Optimal Transformer Tap Changer Setting For
Voltage Stability Improvement,” International Journal of Power, Energy and Artificial intelligence 2009, 1, 89–95.
[16] M.F. Mustafar, I. Musirin, M.R. Kalli, and M.K. Jebatan. “Ant Colony Optimization (ACO)
Based Technique For Voltage Control and Loss Minimization Using Transformer Tap
Setting,” Proceedings of Research and Development 2007, 1–6, December 2007.
[17] H. Yoshida, K. Kawata, Y. Fukuyama, S. Takayama, and Y. Nakanishi. “A Particle Swarm
Optimization For Reactive Power and Voltage Control Considering Voltage Security Assessment,” IEEE Transactions on Power Systems 2000, 15.
[18] K.S. Pandya and S.K. Joshi. “A Survey of Optimal Power Flow Methods,” Journal of
Theoretical and Applied Information Technology, 450–458.
REFERENCES
121
[19] P.A. Jeyanthy and D. Devaraj. “Optimal Reactive Power Dispatch For Voltage Stability
Enhancement Using Real Coded Genetic Algorithm,” International Journal of Computer
and Electrical Engineering 2010, 2, 734–740.
[20] A.P. Alves da Silva and P.J. Abrao. “Applications of Evolutionary Computation in Electric
Power Systems,” IEEE, 2002.
[21] J.G. Vlachogiannis and J. Ostergaard. “Reactive Power and Voltage Control Based On General
Quantum Genetic Algorithm,” Expert Systems with Applications 2009, 36, 6118–6126.
[22] H. Ying-Yi and L. Yi-Feng. “Reactive Power Control in Distribution System Using Genetic
Algorithms,” Proceeding of the 14th International Conference on Intelligent System Application to Power System, 2007, 44–50.
[23] R. Shivakumar, R. Lakshmipathi, and Y. Suresh. “Implementation of Bio Inspired Genetic
Optimizer in Enhancing Power System Stability,” IACSIT International Journal of Engineering and Technology 2010, 2, pp. 263–268.
[24] B. Mahdad, T. Bouktir, and K. Srairi. “Genetic Algorithm and Fuzzy Rules Applied to
Enhance the Optimal Power Flow with Considerations of FACTS,” International Journal
of Computational Intelligence Research 2008, 4, 229–238.
[25] A. Salami. “Control of Capacitors in Distribution Networks Using Neural Network Based
On Radial Basis Function,” Proceeding of the 5th IASME/WSEAS International Conference
on Energy and Environment, 2010.
[26] M.A.H. El-Sayed. “Artificial Neural Network Reactive Power Optimization,” Neurocomputing, 1998.
[27] L. Chen-Ching and T. Kevin. “An Expert System Assisting Decision-making of Reactive
Power/Voltage Control,” IEEE Transactions on Power Systems 2007, 1, 195–201.
[28] A.K. Tajfar, M. Iravani, and R.K. Zare. “Cordinated Fuzzy Logic Voltage/Var Controller in
Distribution Networks,” IEEE International Conference on Industrial Technology, 2008, 1–5.
[29] R. Shivakumar, V. Miranda, and P. Calisto. “A Fuzzy Inference System to Voltage/Var
Control in Distribution Management System,” Proceedings of 14th PSCC, Seville, 2002,
vol. 1, 1–6.
[30] A.M.A. Haidar, A. Mohamed, and A. Hussain. “Vulnerability Control of Large Scale Interconnected Power System Using Neuro-fuzzy Load Shedding Approach,” Expert Systems
with Applications 2010, 10, issue 4.
[31] S.A. Karzalis, A.G. Barkitzis, and V. Petridis. “A Genetic Algorithm Solution of the Unit
Commitment Problem,” IEEE Transactions 1996, PWRS 11, 83–90.
[32] G.B. Sheble and T.T. Maifeld. “Unit Commitment by Genetic Algorithms and Expert
System,” EPSR 1994, 30, 115–121.
[33] D. Dasgupta and D.R. McGregor. “Thermal Unit Commitment Using Genetic Algorithms,”
IEE Proceedings 1994, Part C 141 5, 459–465.
[34] R. Meziane, Y. Massim, A. Zeblah, A. Ghoraf, and R. Rahli. “Reliability Optimization Using
Ant Colony Algorithm Under Performance and Cost Constraints,” Electric Power Systems
Research, 2005.
[35] M. Sasson. “Non-linear Programming Solutions For Load Flow, Minimum Loss, and Economic
Dispatching Problems,” IEEE Transactions on Power Apparatus and Systems 1969, Pas-88.
[36] K.Y. Lee, Y.M. Park, and J.L. Ortiz. “A United Approach to Optimal Real and Reactive
Power Dispatch,” IEEE Transactions on Power Systems 1985, PAS-104, 1147–1153.
[37] H.W. Dommel and W.F. Tinney. “Optimal Power Flow Solutions,” IEEE Transactions on
Power Apparatus and Systems, 1968, PAS.87, 1866–1876.
6
PATHWAY FOR DESIGNING
SMART GRID
6.1
INTRODUCTION TO SMART GRID PATHWAY DESIGN
The design of the smart grid involves the coupling of tools, technologies, and techniques
[14]. An introduction to the tools and technologies which are required for the computational tool to capture the attribute of smart grid is given in this chapter. We will develop
an algorithm which will support the computational tools for smart grid development.
Due to the increased availability of real time or near-real time data and signals, it is
necessary to reinvestigate the most suitable methods or techniques that will allow for
handling dynamics and stochasticity of the grid [15]. Barriers to the smart grid development are developed based on the current existing grid power system structure.
6.2
BARRIERS AND SOLUTIONS TO SMART GRID DEVELOPMENT
Figure 6.1 illustrates the multilayered approach to designing the smart grid. The layers,
or levels, [1, 2] are:
1. System Planning and Maintenance
2. Energy Auction
Smart Grid: Fundamentals of Design and Analysis, First Edition. James Momoh.
© 2012 Institute of Electrical and Electronics Engineers. Published 2012 by John Wiley & Sons, Inc.
122
• Too many decision makers/Local opposition to new plants and lines
• Uncertainty in planning data (e.g., regional growth, costs)
• Lack of predictive real-time system controls (limits capacity of present grid)
• Focus on supply-side reliability solutions
System Planning &
Maintenance Level
• Create and fund intelligent grid center of leaming in Washington DC
• Mandate consumer access to time-dependent pricing information
• Offer tax incentives for deployment of predictive real-time system controls
• Offer tax incentives for demand side load management (DCM) technology
Energy Auction Level
System Operations Level
Device Level
• Public resistance to deregulation (California experiment)
• No time-dependent pricing information available to consumers
• Lack of consumer participation (e.g., using smart DSM technology)
• Lack of environmental credits/imposition of taxes
• Lack of predictive control signals to operate devices
• Lack of location-and time-dependent pricing information
• Lack of energy storage devices (e.g., themo storage)
• Fund development of predictive control technology
• Offer tax incentives for deployment of technology
• Offer tax incentives for DSM technology
Barriers
Solutions
• Mandate use of transparent location-and time-dependent pricing
• Mandate consumer access to pricing information to enable DSM Technology
• Offer environmental credits/impose taxes
• Communications between system operators
• Lack of predictive real-time management tools for contingency planning
• Lack of predictive optimal power flow control software and hardware
• Lack of demand-side load management technology
• Fund development of predictive grid management and control technology
• Fund intelligentgrid demonstration project
• Offer tax incentives for deployment of smart control software and hardware
• Offer tax incentives for demand-side load management technology
123
Figure 6.1. Barriers and solutions to the smart grid.
124
PATHWAY FOR DESIGNING SMART GRID
3. Systems Operations
4. Device
System Planning and Maintenance Level: Issues include too many decision-makers,
local opposition to new plants and lines, planning uncertainties [3, 4], lack of predictive
real-time system controls, and not enough focus on supply-side reliability solutions.
To overcome these barriers, the smart grid design needs consumer access to timedependent pricing information, tax incentives for deployment of predictive real-time
system controls, and tax incentives for DSM technologies. The solution method will:
•
•
•
•
•
Create and fund an intelligent grid center of learning
Mandate consumer access to time-dependent pricing information
Offer tax incentives for deployment of predictive real-time system controls
Offer tax incentives for DSM
These solutions should minimize maintenance at the level of implementation.
Energy Auction Level: Issues include public resistance to deregulation, inadequate time-dependent pricing information unavailable to consumers, lack of consumer
participation (for example, using smart DSM technology), and lack of environmental
credits/imposition of taxes.
The solution will
•
•
•
Mandate use of transparent location-and-time-dependent pricing
Mandate consumer access to pricing information to enable DSM technology
Offer environmental credits/impose taxes
Energy auction level (EAL) purports to redefine pricing structure for the energy market.
This will go a long way toward making the accessibility and usage of the grid affordable by the general populace.
System Operation Level: Issues include communication pitfalls among system
operators to handle emergencies, lack of predictive real-time management tools for
efficient handling, and lack of predictive optimal power flow control software and
hardware and DMS technology.
The solution will:
•
•
•
•
Fund development of predictive grid management and control technology
Fund intelligent demonstration projects
Offer tax incentives for deployment of smart control software and hardware
Offer tax incentives for DSM technology
Device Level: Issues include lack of predictive control signals to operate devices,
lack of locational pricing and time-dependent pricing information, and lack of energy
storage devices.
GENERAL LEVEL AUTOMATION
125
The solution will:
•
•
•
Fund development of predictive control technology
Offer tax incentives for deployment of technology
Offer tax incentives for DSM technology
6.3 SOLUTION PATHWAYS FOR DESIGNING SMART GRID USING
ADVANCED OPTIMIZATION AND CONTROL TECHNIQUES
FOR SELECTION FUNCTIONS
Electric power grids are highly complex dynamical systems vulnerable to a number of
disturbances in day-to-day operations. Random disturbances from weather and accidents rarely produce wide area [6, 10] catastrophic failures. However, the availability
of detailed real-time operational data opens the door for maliciously designed disturbances. Figure 6.2 illustrates the four advanced optimization and control techniques
required to meet the criteria for smart grid performance discussed in Chapter 3.
6.4
GENERAL LEVEL AUTOMATION
Automation at this level involves the use of addressed computation technologies and a
new algorithm for dispatch and unit commitment to ensure:
•
•
•
•
•
•
Reliability
Stability
Optimal dispatch
Unit commitment under different uncertainties and constraints
Security analysis
Distributed generation control
Forecasting techniques must be incorporated into real-time operating practices as well
as day-to-day operational planning, and consistent and accurate assessment of variable
generation availability to serve peak demand is needed in longer-term system planning.
High-quality and real-time data must be integrated into existing practices and software.
The electricity industry is being encouraged to pursue research and development in
these areas [1].
6.4.1
Reliability
Power system reliability is defined as the ability to deliver electricity to all points of
power utilization within acceptable standards. The traditional reliability analysis
methods are deterministic, for example, n-1 criterion where the system is deemed reliable if it can operate under a single unplanned outage [2], whereas in the smart grid
Generation
Level
Automation
Bulk Power
Systems
Automation
Reliability
Unit Commitment
Stability
Security Anaylsis
Economic Dispatch
Distributed Generation
Control
Fault and Stability
Diagnosis
Reactive Power
Control
Network
Reconfiguration
Features of Computational tool
Selection Criteria / Performance
Measures for Selection of tool
Controllability
Adaptability
System Restoration
Distributed Generation
Control
Anticipatory
Behaviour
Voltage/Var Control
Distribution
System
Automation
Stochasticity
Fault Diagnosis
Demand Side
Management
Power Quality
Distributed Generation
Control
Network
Reconfiguration
Trouble Call
System Restoration
Load Management
Power Quality
End-User /
Appliance
Level
126
Criterion/Assessment Tools for SMART GRID functions
Reliability
Sustainability
Dynamics
Load Management
Metering
Robustness
Market and Price
Signals
Menu of analytical, computational intelligence and decision support and hybrid Implementation Tool
Figure 6.2. Smart grid using advanced optimization and control techniques.
GENERAL LEVEL AUTOMATION
127
environment, reliability analysis may not be used for extreme events. Thus, a new reliability and sustainability metric should be capable of handling:
•
•
•
•
Nonlinearity of the system parameters, especially in the presence of the new mix
of generation resources (wind, storage, hydro, and so on) and high power conversion devices
Uncertainty in load demand and generation availability that are functions of time
and contingencies
System dynamics that reflect the market, availability of natural resources, network
reconfiguration, and load switching
Stochasticity of the system parameters due to man-made or natural events over
different time-scales of operations
Indices such as loss of load probability (LOLP) and expected unserved energy (EUE)
are used to determine the energy/load that can be supplied over time for different reliability services. The cost component balances the cost of reliability improvements with
the cost of service disruptions. The ultimate goal is to achieve the maximum reliability
under a variety of probable contingencies. Autonomous control actions will serve to
meet this goal.
6.4.2
Stability
Power system stability is the ability of the bulk power electric power system to withstand sudden disturbances such as electric short circuits or unanticipated loss of system
components [3]. At the generation level, transient stability is the primary consideration;
as the system expands, the significance of transient stability increases.
In general, voltage and angle stability assessments are needed to ensure dynamic
reconfiguration in response to system faults and disturbances. The methods discussed
earlier in chapter 4 are conventional analytical methods including risk assessment and
Lyapunov stability assessment. Since the margin of stability cannot be obtained using
numerical methods, an energy-based method has been utilized for angle stability. Traditional methods for solving stability problems, including continuation power flow
method, singular value decomposition, Jacobian condition number, and energy margin
methods are generally used, but all are based on offline studies.
WAMS and PMU measurements prepared for the smart gird environment will
allow for the real-time evaluation of the system under different loading and unknown
contingencies. Advancements in stability studies aim to address the impacts of contingency with the help of developing new indices for stability assessment.
6.4.3
Economic Dispatch
Economic dispatch is a computational process where the total required generation
including renewable energy resources is distributed among the generation units in
operation by minimizing the selected cost criterion [4] subject to load and operational
constraints. For any specified load condition, the output of each plant (and each
128
PATHWAY FOR DESIGNING SMART GRID
generating unit within the plant) which will minimize the overall cost of fuel needed
to serve the system load is calculated [3]. Traditionally, the economic dispatch problem
is formulated as an optimization with cost as the quadratic objective function:
F ( Pg ) =
Ng
∑ (α + β P
i
i gi
+ γ i Pgi2 )
i =1
Ng
∑
i =1
ND
Pgi −
∑P
Di
+ Ploss = 0
i =1
s.t Vmin ≤ V ≤ Vmax
Pg min ≤ Pg ≤ Pg max
The constraints include equalities and inequalities which represent the power equations
as well as generator, bus, voltage, and line flow limits. This can be solved using analytical mathematical programming such as NLP, QP, LP, Newton method, interior point
methods (IPM), and decision support methods such as AHP.
Computational methods used to solve this form of problem require simplifications
of the objective function to be piecewise linear and monotonically increasing, for solution, but they suffer from the power system’s problem of nonconvexity behavior which
then leads to local minimization or local optimality. Alternative solution methods such
as (EP) [4], GA [5], tabu search [6], neural networks [7], particle swarm optimization
[8, 9], and ADP are proposed for improving the performance of the economic dispatch
algorithm.
6.4.4
Unit Commitment
UC, an operation scheduling function, is sometimes termed “predispatch.” In the overall
hierarchy of generation resources management [10], the UC function coordinates economic dispatch and maintenance and production scheduling over time. UC scheduling
covers the scope of hourly power system operation decisions with a one-day to oneweek horizon.
The UC decisions are coupled or iteratively solved in conjunction with coordinating the use of hydro units including pumped storage with capabilities and to ensure
system reliability. The function may also include labor constraints due to normal crew
policy that a full operating crew will be available without committing to overtime costs.
A prime consideration is the adoption of environmental controls, such as fuel switching.
The application of computer-based UC programs by electric utilities has been slow due
to the following reasons:
i.
ii.
They are not readily transferred between systems. The problem is so large and
complex that only the most important features can be included; these vary
among systems, thus requiring tailor-made applications.
Political problems, constraints, and peculiarities of systems that are not easily
amenable to mathematical solutions and may be difficult to model.
129
GENERAL LEVEL AUTOMATION
iii. Operating situation changes quickly; there is so much objective and subjective
information about the system that the input requirements of sophisticated,
computerized schedulers are discouraging.
iv. As in other computer applications areas, developing fully workable systems
has been difficult; building an operator ’s confidence is also difficult.
The UC schedule is obtained by considering:
•
•
•
•
Unit operating constraints and costs
Generation and reserve constraints
Plant startup constraints
Network constraints
Formulation of Unit Commitment. To account for constraints, randomness,
and stochastic variables, several assumptions are needed for the formulation of UC.
Several classifications for reserve include units on spinning reserve and units on cold
reserve under the conditions of banked boiler or cold start part of the formulation. The
first constraint to be met is that net generation must be greater than or equal to the sum
of total system demand and required system reserve:
N
∑ P (t ) ≥ (Net Demand + Reserve)
gi
i =1
In case the units should maintain a given amount of reserve, the upper bounds must be
modified accordingly. Therefore:
Pgimax = Pgicapacity − Pgireserve
N
N
Demand + Losses ≤
∑
Pgi −
i =1
αt
∑P
reserve
gi
i =1
Ccold = CO (1 − e ) + CL
where
Ccold: Cost to start an off-line boiler
α: Unit’s thermal time constant;
t: Time in seconds
CL: Labor cost to up the units
CO: Cost to start up a cold boiler
Cbanked = C B t + CL
where
CB: Cost to start up a banked boiler
t: Time in seconds
130
PATHWAY FOR DESIGNING SMART GRID
Additional constraints include uncertainty and some of the security index or vulnerability index and congestion margin used in the formulation of the UC system for a smart
grid.
Lagrangian relaxation is among the classical techniques used for the UC problem
where the constraints are based on stochastic variables and predictivity is considered.
The stochastic optimization method is useful for solving this problem as its randomness
is a good solution approach, but its predictivity is not as good compared to ADP [13,
14]. ADP can handle discrete systems which use computational techniques and IPM as
features of the more recent techniques which have been able to handle stochastic and
dynamics indices.
6.4.5
Security Analysis
Power system security analysis detects whether the power system is in a secure state
or an alert state. Secure state implies that the load is satisfied and no limit violations
will occur under the current operating conditions and in the presence of unforeseen
contingencies. Alert state implies that particular limits are violated and/or the load
demand cannot be met, and corrective actions are necessary to bring the system back
to the secure state [10].
Dynamic analysis evaluates the time-dependent transition from the precontingent
to the postcontingent state. Dynamic security has been analyzed either by deriving
dynamic security functions only, or along with the development of some preventive
action techniques [11–13]. Traditionally, n-1 contingency analysis has been used for
the evaluation of security defined as the system’s ability to supply load under
contingency.
The combination of computational intelligence and NP provides an improved
security assessment which accounts for randomness or stochasticity in the data. The
index for measuring security is based on real-time measurements. An example of the
new indices to estimate system performance is vulnerability which could be based on
loss of load index, voltage index, or power flow index.
6.5 BULK POWER SYSTEMS AUTOMATION OF THE SMART GRID
AT TRANSMISSION LEVEL
The automation of different functions of the bulk grid level of the transmission system
is important for achieving resilience and sustainability of the system. The following
functions are evaluated, and the appropriate intelligent technology is proposed:
•
•
•
Real-time angle, voltage stability, and collapse detection and prevention via
intelligent-based data
Reactive power control based on intelligent coordination controls
Fault analysis and reconfiguration schemes based on intelligent switching
operations
131
BULK POWER SYSTEMS AUTOMATION OF THE SMART GRID
Fault and stability
diagnosis, network
switching and cnotrol
Reactive power
control
Intelligent control
schemes and device
management
Managing control
coordination
Intelligent grid
automation
Min. Loss, Load
balancing, opt switching
Intelligent controls and
demand reduction
Network switching/
restoration
Distributed
generation for
emergency use
Figure 6.3. Intelligent grid automation.
•
•
Power generation and load balance via intelligent switching operation and minimizing demand interruption
DG and DSM via DR strategy for peak shaving, including increased proliferation
and control of RER
The integration of these functions is shown in Figure 6.3.
Some of the functions and roles are presented next.
6.5.1
Fault and Stability Diagnosis
Voltage and angle stability assessments ensure dynamic reconfiguration in response to
system faults and disturbances. Since the margin of stability cannot be obtained using
numerical methods, an energy-based method has been proposed for angle stability. To
date, real-time evaluation of the system under different loading and unknown contingencies has not been implemented. Analytical techniques used for fault analysis include
the circuit theory-based method and the traveling wave-based method. Methods for
132
PATHWAY FOR DESIGNING SMART GRID
detection of voltage instability are generally based solely on local measurements of
voltage, for example under voltage load shedding (UVLS). WAM and PMUs also play
a major role in stability monitoring and analysis.
6.5.2
Reactive Power Control
The main factor causing instability is the inability of the power system to meet the
demand for reactive power. The heart of the problem is usually the voltage drop that
occurs when active power and reactive power flow through the inductive reactance
associated with the transmission network [14]. Reactive power control uses optimization techniques to minimize loss and schedule reactive resources to remove the voltage
degradation problem. The use of real-time power margin as a constraint in optimization
can be solved in real time with intelligent optimization techniques based on DSOPF.
Other functions such as network switching and restoration use knowledge based
systems completed with GA and fuzzy logic to select the outage units and lines and
appropriate control measures. for scheduling outages. ADP and other variants of EP,
such as GA and ACO, are adapted to address RER penetration into the smart grid and
to manage variability.
6.6 DISTRIBUTION SYSTEM AUTOMATION REQUIREMENT
OF THE POWER GRID
Distribution automation and control involves the delivery of energy to customers.
Figure 6.4 shows the distribution automation schemes for distribution systems.
Smart Distribution solutions are designed to minimize energy losses, mitigate
power disruptions, and optimally utilize distributed smart grid components, including
alternative energy sources, power storage, and PHEV charging infrastructure.
Greater access to larger pools of available generation and demand may also be
important to the reliable integration of large-scale variable generation. As the level of
variable generation increases in a balancing area, the resulting variability may not be
manageable with its existing conventional generation resources. Base load generation
[15] may need to be frequently cycled in response to these conditions, which introduces
reliability concerns as well as economic consequences. These efforts may also help to
address minimum load requirements of conventional generation and contribute to the
effective use of off-peak, energy-limited resources.
6.6.1
Voltage/VAr Control
Voltage control within a specified range of limits and capacitor switching is an effective
means of minimizing loss, improving voltage profiles, and deferring construction in the
end within the reliability and power quality constraints.
Voltage/VAr Control considers a multiphase unbalanced distribution system operation, dispersed generation, and control equipment in the large system. In distribution
automation, functions using voltage/VAr control options must maintain proper com-
133
DISTRIBUTION SYSTEM AUTOMATION REQUIREMENT OF THE POWER GRID
Stop Stop
Power
Quality
Fault
Diagnosis
Voltage Sag and
Harmonics
Sitting and Breaker
Coordination
DSM
Analysis
Optimize units and
energy usage
Control Secured
System?
DAs for
Distribution
Systems
Min. Loss/Voltage &
optimal switching
Analysis and Crew
Dispatch
Trouble
Call
Min. Loss, Load
balancing, switching
Network
Reconfiguration
System
Restoration
Figure 6.4. Distribution automation for distribution systems.
munication throughout decision-making to install capacitors, thereby realizing the cost
benefit.
Formulation of Voltage/VAr Problem. This includes integrating voltage/VAr
with the load management problem to improve efficiency. The objectives are:
a. Customer Outage Cost
Min Outage Cost =
∑ (X
ik
⋅ PCki ⋅ CCk )
i
where
Xik: Level of Curtailable Load Selection of type k at bus i (p.u. MW)
PCki: Max Curtailable MW of type k at the ith bus (p.u. MW)
CCk: Curtailment Cost of Customer Type k ($/p.u. MW)
134
PATHWAY FOR DESIGNING SMART GRID
b. Loss Minimization
The objective of the loss minimization function is given by:
Min I 2 r =
∑r
Pij2 + Qij2
ij
ij
Vi
2
c. Load Balancing
Max
Si
Si max
where:
N: Total number of buses
Pij, Qij: Transfer Power (branch i-j at p.u.)
Vi: Voltage of bus i (p.u.)
d. Multiple objective functions
Z = Min [ a + b − c ]
System Operating Constraints include:
a. Branch Flow Equations
Yi +1 = fi +1 (Yi )
where Yi = [P,Q,|V|2,Xk,Qs,δi]
Branch flow considers the recursive relationships between the successive nodes in
the radial distribution system. The demand in PD, QD has an interruptible component
which is the load management control options X. In addition, the reactive power equation has the capacity switching option Qs.
b. Voltage Limits/Current Limits
The voltage and current limits are given as:
Vi min ≤ Vi ≤ Vi max
I ijmin ≤ I ij ≤ I ijmax
c. Capacitor Control Limits
Qsimin < Qsi ≤ Qsimax
DISTRIBUTION SYSTEM AUTOMATION REQUIREMENT OF THE POWER GRID
135
d. Curtailable Load Control Limits
Pi × X ki ≤ Pcki
Qi × X ki ≤ Qcki
6.6.2
Power Quality
Power quality is the ability of a system to operate without causing disturbance or
damage to loads and components. Power quality is a major concern because of the
sensitivity of digital and modern control equipment to distortion/PQ deterioration. A
basic requirement for maintaining power quality is balancing supply and demand.
Typical indices utilized as power quality measures include:
•
•
•
•
Voltage transient impulse
Voltage sag
Total harmonic distortion
Flicker factor
A complete listing of quality measures employed in the global electric power industry
is given in Table 6.1.
Several accepted power quality standards which apply to electrical distribution
systems are itemized in “Voltage Disturbances Standard EN 50160: Voltage Characteristics in Public Distribution Systems.” A real-time power quality study features several
aspects [13]:
TABLE 6.1. Typical Power Quality Indices
Phenomena Indices
Calculations
Parameters
Voltage Sag or Swell ΔV = |Vi −Vnom|
Flicker factor
FF =
∆V
V
Total Harmonic
Distortion (THD)
∞
2
∑ (I ( ) )
i
THD =
i=2
I (1)
∞
2
∑ (V ( ) )
i
THD =
i=2
V (1)
Networks/Systems
where index is applied
Networks which feature:
Vi—voltage at bus i
Vnom—nominal voltage Remote System Faults
Networks which feature:
V—voltage
Intermittent loads
Motor starting
Arc Furnaces
Networks which feature:
I(i)—ith harmonic
Non-linear loads
current
System resonance
V(i)—ith harmonic
voltage
136
PATHWAY FOR DESIGNING SMART GRID
1. Sensitive detection of power disturbances
2. Real-time measurement of parameters of signal components in power
disturbances
3. Quantification of power disturbances and their negative impacts on power
systems
4. Identification of types and causes of power disturbances
5. Location of power disturbances
Power quality monitoring and analysis in the smart grid will utilize real-time data.
6.6.3
Network Reconfiguration
Cost-effective reconfiguration assessment strategies are needed to address the following
challenges:
•
•
•
•
Nonlinearity of power system parameters
Uncertainty in load demand and generation availability
System dynamics
Stochastic of system parameters
Intelligent control methods including adaptive critics designs, fuzzy logic, and ANN
methods may provide adaptive stochastic solutions to the general problem formulation.
6.6.4
Demand-Side Management
DSM is an effective means of modifying customer demand to reduce operating expenses
from expensive generators and defer capacity addition in the long run. Environmental
conservation is improved with DMS options, which also sustain industrialization at
minimum cost. In addition DSM options provide reduced emission reduction in fuel
production and reduced costs, and contribute to system reliability. DSM options are
categorized into:
•
•
•
•
Peak shifting
Valley filling
Peak clipping
Storage conservation
These options have an overall impact on the utility load curve. For distribution
automation functions, DSM is classified into three categories:
1. Direct Control of Load [12]: Uses a communication system such as a power
line carrier/radio to transmit direct control of load, small generators, and storage
from utility to customers.
APPLICATIONS FOR ADAPTIVE CONTROL AND OPTIMIZATION
137
2. Local Load Control Option: Enables customers to self-adjust loads to limit peak
demand, for example, demand activated breakers, load interlocks, timers, thermostats, occupancy sensors, cogeneration heating, cooling storage, and so on.
3. Distribution Load Control: Utility controls customer loads by sending real-time
prices.
6.6.5
Distribution Generation Control
More comprehensive planning approaches from the distribution system to the bulk
power system are needed, including probabilistic approaches at the bulk system level.
This is particularly important with the increased prevalence of distributed variable
generation, for example, local wind plants and rooftop solar, on distribution systems.
In aggregate, distributed variable generation needs to be treated, where appropriate,
similarly to transmission-connected variable generation. The issues of note include
forecasting, restoration, voltage ride through, safety, reactive power, observability, and
controllability [11, 14]. Standard, nonconfidential, and nonproprietary power flow and
stability models are needed to support improved planning efforts and appropriately
account for new variable resources. Variable generation manufacturers are encouraged
to support the development of these models.
6.7
END USER/APPLIANCE LEVEL OF THE SMART GRID
At the end-user level, significant changes in metering and monitoring are being introduced. DSM and DR are the two fundamental automation functions which will be
enhanced by developments at the end-user and appliance levels. DSM includes customers as utility planning options. The total system cost including DSM cost is minimized
to obtain an optimal mix of the supply side of the generation and demand side load
reduction. Analysis of DSM uses techniques such as daily load curve or mathematical
programming methods. DSM has been carried out using the context of unit commitment
studies, optimal power flow studies, load reduction forecasting methods, engineering
features of the end-user equipment interruptible load management program, survey
methods (data collection), and DP to optimize energy procurement and load management by utilities.
Standards for integrating advanced mobile power recovery units, energy storage,
and on-demand or peak-shaving technologies will invariably necessitate the development of standards for coordination, communication, and interoperability of multiple
appliances competing for energy at the optimum grid price available. These and other
related issues are research topics that require urgent attention.
6.8
APPLICATIONS FOR ADAPTIVE CONTROL AND OPTIMIZATION
There are several potential applications for adaptive controls and optimization in the
development of the smart grid’s design framework. It is important that adaptation and
138
PATHWAY FOR DESIGNING SMART GRID
control strategy scheduling consider the stochastic and dynamically interdependent
attributes of power systems. Conventional techniques to handle these problems often
suffer from the curse of dimensionality and somewhat heuristic methods which tend to
give nonoptimal solutions. Therefore, system stochasticity and dynamic interdependence lead to unstable, nonfeasible, and nonoptimal solutions.
In DSOPF, the efficient optimization technique typically based on two-stage action
and critic networks are used to achieve:
•
•
•
•
•
6.9
Multiobjective, time-dependent optimization for complex systems
Optimal scheduling subject to technical constraints of the plant or system
Adaptation to perturbation of power system dynamics over time
Adaptation to varying random noise, uncertainties, and/or missing or corrupted
measurements
Adaptation to changes in system structure while distinguishing between observable and unobservable measurements
SUMMARY
This chapter has discussed the development of predictive grid management and control
technology for the deployment of smart control software and hardware to enhance smart
grid performance. This will lead to benefits for customers and stockholders [14, 15].
The lack of predictive control signals to operate devices and the lack of energy storage
devices were described. Research on the integration of renewable and storage and new
software and control technology was suggested to deploy, operate, and maintain the
new grid’s infrastructure.
REFERENCES
[1] “Accommodating High Levels of Variable Generation,” North American Electric Reliability
Corporation, 2009.
[2] J.A. Momoh and M.E. El-Hawary. Electric Systems, Dynamics and Stability with Artificial
Intelligence Applications. New York: Marcel Dekker, 2000.
[3] A.J. Wood and B.F. Wollenberg. Power Generation, Operation and Control. New York:
John Wiley & Sons, 1996.
[4] H. Yang, P. Yang, and C. Huang. “Evolutionary Programming Based Economic Dispatch
with Non-Smooth Fuel Cost Functions,” IEEE Transactions on Power Systems 1996, 11,
112–118.
[5] D.C. Walters and G.B. Sheble. “Genetic Algorithm Solution of Economic Dispatch With
The Valve-point Loading,” IEEE Transactions on Power Systems 1993, 8, 1325–1332.
[6] W.M. Lin, F.S. Cheng, and M.T. Tsay. “An Improved Tabu Search For Economic Dispatch
With Multiple Minima,” IEEE Transactions on Power Systems 2002, 17, 108–112.
[7] K.Y. Lee, A. Sode-Yome, and J.H. Park. “Adaptive Hopfield Neural Network For Economic
Load Dispatch,” IEEE Transactions on Power Systems 1998, 13, 519–526.
SUGGESTED READING
139
[8] R.C. Eberhart and Y. Shi. “Comparing Inertia Weights and Constriction Factors in Particle
Swarm Optimization,” Proceedings of the 2000 Congress on Evolutionary Computation, 1,
84–88.
[9] R.C. Eberhart and Y. Shi. “Particle Swarm Optimization: Developments, Applications, and
Resources,” Proceedings of the 2001 Congress on Evolutionary Computation, 1, 81–86.
[10] OSI. Smart Grid Initiatives White Paper: Revision 1.1.
[11] A.K. Al-Othman, F.S. Al-Fares, and K. M. EL-Nagger. “Power System Security Constrained
Economic Dispatch Using Real Coded Quantum Inspired Evolution Algorithm,” World
Academy of Science, Engineering and Technology 2007, 29, 7–14.
[12] R.B. Alder. “Security Consideration Economic Dispatch With Participation Factors,” IEEE
Transactions on Power Apparatus and Systems 1977, PAS-96.
[13] M. El-Sharkawy and D. Niebur. “Artificial Neural Networks With Application to Power
Systems,” IEEE Power Engineering Society, A Tutorial Course, 1996.
[14] F.F. Wu, J.H. Chow, and J.A. Momoh, Eds. Applied Mathematics for Restructured Electric
Power Systems: Optimization, Control, and Computational Intelligence. Springer Science
and Business Media, Inc., 2005: pp. 11–24.
[15] T. Lin and A. Domijan. “On Power Quality Indices and Real Time Measurements,” IEEE
Transactions on Power Delivery 2005, 20, 2552–2562.
SUGGESTED READING
P. Kundur. Power System Stability and Control. Washington, D.C.: McGraw-Hill Inc., 1994.
7
RENEWABLE ENERGY
AND STORAGE
7.1
RENEWABLE ENERGY RESOURCES
The design and development of the smart grid requires modeling renewable energy
sources and technologies such as wind, PV, solar, biomass, and fuel cells, analyzing
their levels of penetration, and conducting impact assessments of the legacy system for
the purpose of modernization. The roadmap envisions widespread deployment of Distributed Energy Resources (DERs) in the near future. Renewable technologies have
been positioned to reduce both America’s dependence on foreign oil and the environmental impacts of energy production [1]. Renewable energy technologies and their
integration introduce several issues including enhancement of efficiency and reliability,
and the development of state-of-the-art tracking to manage variability.
Architecture designs which include optimal interconnections, optimal sizing and
siting DERs for optimum reliability, security, and economic benefits are also critical
aspects. Additionally, computational development of the smart grid to permit estimation
and forecasting models for fast real-time accurate predictions of these variable power
sources need to be addressed. The challenges and the technologies including conversion, storage, and PHEV, are discussed in this chapter.
Smart Grid: Fundamentals of Design and Analysis, First Edition. James Momoh.
© 2012 Institute of Electrical and Electronics Engineers. Published 2012 by John Wiley & Sons, Inc.
140
SUSTAINABLE ENERGY OPTIONS FOR THE SMART GRID
7.2
141
SUSTAINABLE ENERGY OPTIONS FOR THE SMART GRID
Sustainable energy is derived from natural sources that replenish themselves over
of time. Sometimes called green power, because they are considered environmentally
friendly and socially acceptable, they include sun, wind, hydro, biomass, and
geo-thermal.
The future electric grid [3, 4] will feature rapid integration of alternative forms for
energy generation as a national priority. This will require new optimization for energy
resources that are distributed with interconnection standards and operational constraints.
The legacy grid was designed to handle this change. The stability, reliability, and cost
implications of renewable energy resources in the development of technical and economic software are vital.
Renewable energy options are meant to provide the smart grid with:
i. Remote utilization and storage of RER resources output
ii. Enhancement of functionality of grid-connected renewable energy systems
(RES)
a. Facilitating give-and-take of energy from the system
b. Redistribution/reallocation of unused power from grid-connected RES
c. Facilitating storage of grid-generated and RER-generated energy by backup storage technologies at customer end
d. Tracking interactions for billing and study
iii. Enhancement of functionality of electric vehicles and plug-in hybrids
iv. Utilization of vehicle battery packs as energy storage devices
The common RER uses in smart grid networks are presented below.
7.2.1
Solar Energy
Solar energy harnessed by the use of photovoltaic (PV) cells was first discovered in
1839 by French physicist Edmund Becquerel. The technology can be a single panel, a
string of PV panels, or a multitude of parallel strings of PV panels. Solar PV has no
emissions, is reliable, and requires minimum maintenance.
The PV system generally considers:
i. Insolation: The availability of solar energy conversion to electricity. Insolation
levels are affected by the operating temperature of PV cells intensity of light
(location-dependent), and the position of the solar panels (maximize the power
tracking while maximizing perpendicular incident light rays).
ii. Emission: PV emission levels are environmental friendly.
To improve efficiency, the materials used for manufacturing PV cells include amorphous silicon, polycrystalline silicon, cadmium telluride, microcrystalline silicon, and
copper indium selenide.
142
7.2.2
RENEWABLE ENERGY AND STORAGE
Solar Power Technology
Solar power technology enhances PV output by concentrating a large area of sunlight
into a small beam using lenses, mirrors, and tracking systems. Parabolic troughs and
solar power towers are examples of such technologies [12].
Cost Implication: Manufacturers continue to reduce the cost of installation as new
technology is developed for manufacturing materials. Much work is being conducted
in the manufacturing of PV and the development of superior materials. Basic materials
include monocrystalline and polycrystalline panels, cast polycrystalline silicon, and
string and ribbon silicon, as well as amorphous silicon or thin film panels which are
created by the application of amorphous silicon, copper indium diselenide (CIS),
and cadmium telluride (CdTe) as a thin semiconductor film. This application allows for
the manufacturing of panels that are less time-consuming to manufacture with lower
manufacturing costs and that are applicable to varied applications, albeit with reduced
efficiency.
7.2.3
Modeling PV Systems
There are several PV simulation programs which allow for series analysis for time such
as PV-DesignPro [5]. The input typically consists of one year ’s worth of hourly globaldirect irradiance, and temperature and wind speed data, for example, TMY data, array
geometry, and other PV and balance of system (BOS) parameters. Many models exist
for the calculation of the power output of a PV cell or bank. Due to the varying efficiencies and numerous technologies presently available, power output is affected by
environmental conditions and module specifications. The I-V characteristic model of a
single cell is commonly used for PV technologies. The model is given by:
q (V + IRs ) V + IRs
I = I sc − I os e nkT − 1 −
Rsh
− Eg
I os = AT γ e nkT
q Ed
I d = I sc − I os e nkT − 1
where for an array of Ns × Nsh solar cells:
Id =
I d _ mod
N sh
Ed _ mod = Ed × N s
N
Rs _ mod = Rs s
N sh
SUSTAINABLE ENERGY OPTIONS FOR THE SMART GRID
N
Rsh _ mod = Rsh s
N sh
where
I: current flowing into load of a solar cell (A)
Isc: short circuit current (A)
Ios: saturation current (A)
s: insolation (kW/m2)
q: electron charge (1.6 x 10-19 (C))
k: Boltzmann constant (1.38 x 10-23 (JK-1))
T: p-n junction temperature (K), t (oC)
N: junction constant
A: temperature constant
γ: temperature dependency exponent
Eg: energy gap (eV)
V: voltage across solar cell (V)
Voc: open circuit voltage of a solar cell (V)
Rs, Rs_mod: series parasitic resistance for cell and entire module (Ω)
Rsh, Rsh_mod: shunt parasitic resistance for cell and entire module (Ω)
Ed, Ed_mod: across voltage of an ideal solar cell and entire module (V)
Id, Id_mod: current of an ideal solar cell and entire module (A)
Ns: number of series cell junctions of a PV module
Nsh: number of parallel cell junctions of a PV module
Vout: across voltage of a PV module (V)
Iout: current of a PV module (A)
R: connected load (Ω)
An alternative equation for the modeling of the output of the PV panels is [6]:
Pmp =
G
Pmp,ref [1 + γ (T − Tref )]
Gref
where
G is the incident irradiance
Pmp is the maximum power output
Pmp,ref is the maximum power output under standard testing conditions
T is the temperature
Tref is the temperature for standard testing conditions reference (25°C)
Gref = 1000 Wm−2
γ is the maximum power correction for temperature
143
144
RENEWABLE ENERGY AND STORAGE
VPV < 850 V
Full Bridge
Inverter
PV
Array
Utility
Grid
Filter
Vgrid
Figure 7.1. PV inverter system for DC-AC conversion.
The PV output is variable due to in relation to the solar insolation and surface temperature. The data for predicting the solar input is several years of measurements of
irradiance on the proposed locator. These statistical measures may be estimated from
meteorological data available from the site, from a nearby site having similar irradiance,
or from an official solar atlas or database [7]. Solar insolation has been modeled probabilistically for variability studies of PV systems utilizing various distributions including
Gaussian (normal) and Beta probability density functions.
Conversion and Power Electronic Technology. Several inverter systems
convert or transform the DC into AC for grid-connected PV systems (see Fig. 7.1).
Penetration of PV into the smart grid requires studies in variability and conversion
technology. The mathematical models and probability density used to model PV behavior include Beta and Rayleigh density functions. The tracking method based on fuzzy/
GA technologies is able to obtain enough power points for effective delivery. Siting
and sizing of PV uses classical and computational intelligence methods that involve
decisions based on real-time data.
7.2.4
Wind Turbine Systems
Wind is one of the fastest-growing sources of renewable energy throughout the world
(see Fig. 7.2). Turbines produce electricity at affordable cost without additional investments in infrastructure such as transmission lines. A wind turbine consists of a rotor,
generator, blades, and a driver or coupling device. Compared with PV, wind is the most
economically competitive renewable.
Table 7.1 shows various wind turbine configurations. Both wind speed and the
height of pole-mounted units contribute to the power output.
Although turbines produce no CO2 or pollutants [9], wind has three drawbacks:
output cannot be controlled, wind farms are most suited for peaking applications, and
power is produced only when there is sufficient wind.
145
SUSTAINABLE ENERGY OPTIONS FOR THE SMART GRID
120
Megawatts
80
60
40
20
U
0
1996 1998 2000 2002 2004 2006 2008
Added in 2008
Existing in 2002
ni
te
d
S
G tate
er s
m
an
Sp y
a
C in
hi
n
In a
di
a
U
I
ni
te Fr taly
d a
Ki nc
n e
D gdo
en m
m
Po ark
r tu
ga
l
Glgawatts
100
26,000
24,000
22,000
20,000
18,000
16,000
14,000
12,000
10,000
8,000
6,000
4,000
2,000
0
Figure 7.2. (a) Wind power, existing world capacity, 1996–2008; (b) wind power capacity, top
ten countries, 2008.
Modeling Wind Turbines: Machines which consume reactive power and produce
real power. The quantification of the capacity/real power output is given by:
Pm =
1
ρ ⋅ π R2 ⋅ V 3 ⋅ C p
2
where
ρ is the air density (kg/m3)
R is the turbine radius (m)
Cp is the turbine power coefficient power conversion efficiency of a wind turbine
V is the wind speed (m/s)
The electrical power output is given by:
Pe = no Pm
where
no = ηmηg
ηm, and ηg are the efficiency of the turbine and the generator, respectively
Wind energy options are typically random and require a modeling technique based
on probability such as Gaussian, Rayleigh, and Beta pdfs. Another challenge is congestion management. Due to time-varying behavior and randomness, EP such as GA, ANN,
and fuzzy logic and its hybrids are used to allocate wind energy to meet in-time local
demand. The process includes obtaining real-time measurement, and using the system
state, location, and network limitation to define new algorithms.
7.2.5
Biomass-Bioenergy
Bioenergy is the energy derived from organic matter such as corn, wheat, soybeans,
wood, and residues that can produce chemicals and materials. Biopower is obtained
146
TABLE 7.1. Wind Turbine Modeling
Type
Type A: Constant Speed Wind Turbine
Type B : Variable Speed WT
with variable rotor resistance
By-pass
Configuration
SCIG
wind
G
Gear
box
Capacitor
Bank
AC
WRIG
transformer
AC
Soft-starter
wind
grid
Type A
Gear
box
Variable
Resistance
G
Type C :Variable Speed
WT with partial-scale
frequency converter
DFIG
By-pass
AC
transformer
AC
grid
Soft-starter
Capacitor
Type B
Bank
wind
SCIG
WRSG / PMSG
transformer
G
Gear
box
Type D :Variable Speed
WT with full-scale
frequency converter ()
wind
AC
AC
Power Converter
grid
Type C
Generator type
Squirrel Cage Induction Generator (SCIG)
Wound rotor induction generator
(WRIG)
Doubly-fed induction
generator (DFIG)
Reactive power
Draws reactive power from grid; must be
accompanied by a capacitor back.
Converter performs
the reactive power
compensation and
a smooth grid
connection
Key characteristics
Wind fluctuations are converted into
mechanical fluctuations and further
into electrical power fluctuations
Does not support any speed control
requires a stiff grid, mechanical
construction must be able to support high
mechanical stress caused by wind gusts.
Limited active power
A capacitor bank performs the
reactive power compensation
and smooth grid connection
occurs by means of a soft
starter.
Total rotor resistance is
controllable by variable
additional rotor resistance,
which is changed by
an optically controlled
converter mounted on the rotor
shaft
G
Gear Box/
Gearless
transformer
AC
AC
grid
Full Scale
Power
Converter
Type D
Electrically excited
(wound rotor
synchronous
generator WRSG)
or permanent magnet
excited type
(permanent magnet
synchronous generator
PMSG).
Converter performs
the reactive power
compensation and
a smooth grid
connection
Full variable speed
controlled wind
turbine, with the
generator
connected to the grid
through a full-scale
frequency converter
SUSTAINABLE ENERGY OPTIONS FOR THE SMART GRID
147
from a process called gasification, converting gas to gas turbines to generate electricity.
Biomass can be converted directly into fluid fuels such as ethanol, alcohol or biodiesel
derived from corn ethanol.
Biomass plants are commercially available in the United States for up to 11GW of
installed capacity. Biomass power ranges from 0.5 GW to 3.0 GW using landfill gas
and forest products, respectively. Biomass has traditionally been used for domestic
cooking and heating in developing countries. It can produce power only when sufficient
bioproducts are available and the conversion process is undertaken. Biomass produces
CO2 and other emissions. The desirable scheduling and allocation strategy for biomass
in real time requires the capability to include variability of the modeling using new
system theory concepts.
7.2.6
Small and Micro Hydropower
Hydropower is by far the largest renewable source of power/energy [8]. Small hydropower systems vary from 100 kW to 30 MW while micro hydropower plants are smaller
than 100 kW. Small hydropower generators work in variable speed because of water
flow. Induction motors provide a generator for a turbine system. The hydraulic turbine
converts the water energy to mechanical rotational energy.
Small and micro hydropower systems are RER optimizations to enhance the smart
grid. The issues of reliability and modeling are addressed as in PV and wind energy.
CI technology is used for performance study and commitment.
7.2.7
Fuel Cell
Fuel cells can also be used to enhance power delivery in the smart grid. They are simply
fuels from hydrogen, natural gas, methanol, and gasoline. The efficiency for fuel to
electricity can be high as 65% since it does not depend on Carnot limits. Fuel cells are
environmentally friendly by efficient use of fuel. Different forms include phosphoric
acid (PAFC), proton exchange membrane (PEM), solid polymer molten carbonate
(MCFC), solid oxide (SOFC), alkaline (a direct methanol), regenerative fuel cells, and
botanic ceramic fuel cells (PCFC). Fuel cells produce virtually no emissions. Their cost
is significantly high compared with conventional technologies.
The topology of a fuel cell is a stack which consists of the part of the fuel cell that
holds the electrodes and electrolytic material. Hydrogen is extracted from gasoline
propane, with natural gas refineries to operate fuel cells commercially.
Emission is very low and so these cells have minimum environmental impacts.
The high efficiency leads to low fuel costs and minimum maintenance due to lack of
moving parts. They have virtually no pollutant emission as CO2 is rather low. They are
a good fit for green power and premium power due to their quality. In addition, they
provide a moderate high thermal quantity output and hence are ideal for CHP. However,
they perform poorly as peaking power due to extremely high capital costs.
The efficiency of fuel cells ranges from 40–80%. Two commonly used fuel cell
types are phosphoric acid fuel cells (PAFC), which operate at relatively high temperatures and use an external water cooling system to cool the stack, and proton exchange
148
RENEWABLE ENERGY AND STORAGE
membrane fuel cells, which operate at a lower temperature than most other fuel cells
and contain no chemicals such as liquid acids or molten bases that would cause concerns
about materials of construction.
7.2.8
Geothermal Heat Pumps
This form of power is based on accessing the underground steam or hot water from
wells drilled several miles into the earth. Conversion occurs by pumping hot water to
drive conventional steam turbines which drive the electrical generator that produces
the power. The water is then recycled back into earth to recharge the reservoir for a
continuous energy cycle. There are several types of geothermal power plants including
dry steam, flash stream, and binary cycle. Dry steam plants draw water from the reservoirs of steam, while both flash steam and binary cycle plants draw their energy from
the recycled hot water reservoir.
In the United States, geothermal power is produced in California, Utah, Nevada,
and Hawaii. Various applications of geothermal power include heat pumps, agriculture,
fishing, farming, and food processing. Geothermal projects require significant upfront
capital investment for exploration, drilling wells, and equipment. Exploration risk and
environmental impacts are considered in geothermal power plant projects.
7.3 PENETRATION AND VARIABILITY ISSUES ASSOCIATED WITH
SUSTAINABLE ENERGY TECHNOLOGY
The degree of penetration by sustainable energy into today’s grid varies by location,
depending on the availability of the conditions necessary for viable utility. This can be
quantified by:
∑P
Penetration Level =
∑P
DG ,i
∀i
demand , j
∀j
The selection and implementation of the available sustainable energy technologies
are subject to the issues of variability associated with the sources, for example, solar
and wind. For solar energy, the variation of the solar radiation is a function of the time,
location, height above the earth’s surface, local weather conditions, and degrees of
shading. The variability of the PV source is a function of the solar insolation which
has been modeled using a Beta distribution function. Utilizing historical data for location and time allows calculation of parameters of the Beta model by solving equations
given as
(1 − µ ) µ
α = µ
− 1
σ
(1 − µ ) µ
β = (1 − µ )
− 1
σ
149
PENETRATION AND VARIABILITY ISSUES
The corresponding probability distribution function is formulated as f(s):
f ( si ) =
β −1
siα −1 (1 − si )
Γ (α + β )
Γ (α ) Γ (β )
There are three apparent sources of variability in wind technology: wind speed random
variability, turning off turbines to prevent damage when wind speed values are greater
than the tolerable limit, and no output when wind speed is less than the cut-in speed.
The Weibull model, which has been used for studying wind speeds since 1978, models
wind speed trends in numerous instances. The probability density function is:
k
k V k −1
V
exp , V ≥ 0
λ
f (V ) = λ λ
0, V < 0
with mean and variance calculated in terms of the shape and scale parameters
1
µ = λΓ 1 +
k
2
Var = λ 2 Γ 1 + − µ 2
k
Utilizing historical data for location and time, the parameters of the Weibull model
can be determined by simultaneously solving the mean and variance equations
given as
σ
k =
V
λ=
−1.086
V
1
Γ 1 +
k
Another common model is the Rayleigh distribution function, a special case of the
Weibull model, which offers the added simplicity of only one parameter. This assumption can be made in several instances where the wind scale parameter in the location
under study is approximately [2].
The concept of smoothing mitigates the impact of variability by utilizing several
turbines over an area. From this penetration level an optimal scheme based on the
computational intelligence technique is proposed for evaluating system and stability
assessment in real time.
150
RENEWABLE ENERGY AND STORAGE
DEMAND RESPONSE
Smart Grid Communication
Sensing and Measurement
Control
Local Area Networks
(LANs)
Sensing
Appliance Controllers
Web-enabled Architecture
Measurement
On-site Generation
Control
Software
Dynamic Demand
Figure 7.3. Demand Response technology tree.
7.4
DEMAND RESPONSE ISSUES
Energy management involves controlling electrical and mechanical systems to reduce
power needs and the associated costs. DR helps to reduce customer demand on the grid
that is dependent on that demand. Abundant data, including price signals and grid
conditions, allow DR technologies to play a significant role. Monitoring operating
parameters such as voltage, angle, and frequency of the system are utilized through
real-time sensors in addition to controllers, metering signals, and two-way digital communication, in responding to changes in the grid and electricity prices. Automatic DR
in times of disruption is a key feature of the smart grid, as are smart meters, smart
appliances, and distributed RER. To achieve optimal control of demand and fulfill
economic and environmental goals, utilities can show customers how to adjust their
consumption to off-peak time demand to assist in efficient supply.
Figure 7.3 shows the DR applications that can be categorized into four components:
A. Energy Efficiency
B. Price-based DR
a. Time-of-use (TOU)
b. Day-ahead hourly pricing
C. Incentive-based DR
a. Capacity/ancilliary services
b. Demand bidding buy-back
c. Direct load control
D. Time scale commitments and dispatch
a. Years of system planning
b. Months of operational planning
PHEV TECHNOLOGY
151
c. Day-ahead economic scheduling
d. Day-of economic dispatch
e. Minutes dispatch
7.5
ELECTRIC VEHICLES AND PLUG-IN HYBRIDS
The integration of electric vehicles and hybrids is another component of the smart grid
system. Vehicle-to-grid power (V2G) uses electric-drive vehicles (battery, fuel cell, or
hybrid) to provide power for specific electric markets. V2G can provide storage for
renewable energy generation and stabilize large-scale wind generation via regulation.
Plug-in hybrids can dramatically cut local air pollution. Hybridization of electric vehicles and connections to the grid overcome limitations of their use including cost, battery
size/weight, and short range of application. PHEVs provide the means to replace the
use of petroleum-based energy sources with a mix of energy resources (encountered in
typical electric power systems) and to reduce overall emissions [9].
PHEVs offer advantages and disadvantages compared to other proposed solutions.
The deployment of PHEVs potentially has a substantial positive impact on the electric
power system from the point of view of increasing electric energy consumption, offsetting petroleum fuels with alternative sources of energy.
7.6
PHEV TECHNOLOGY
Through V2G power, a parked vehicle can provide power to the grid as a batteryelectric, fuel-cell, or even a plug-in hybrid vehicle. Stored or unused energy that utilities
reserve at night or during off-peak times can be used to support the vehicles in cases
of extreme emergencies, for example, a significant decrease in oil supplies or a sudden
rise in oil prices. It should be noted that electric drive systems are considered 70%
efficient, for example, a first-generation plug-in hybrid can travel approximately 3–4
miles per kWh, or about 75 cents per gallon of gas.
Each PHEV vehicle will be equipped with a connection to the grid for electrical
energy flow, a control or logical connection necessary for communication with the grid
operator, and onboard controls and metering. Figure 7.4 schematically illustrates the
connections between vehicles and the grid from Reference 11. Electricity flows throughout the grid from generators to end-users while unused energy flow backs from the
EDVs as shown by the lines with two arrows in Figure 7.4 Note that during times of
low demand, battery EDVs can charge and discharge as necessary for power supply or
demand. Fuel cells can create power from liquid and gaseous fuels and plug-in hybrids
can function in either capacity [11].
7.6.1
Impact of PHEV on the Grid
PHEVs are expected to take 10% of the U.S. market share of new vehicles sold in 2015,
and 50% of the market share by 2025 [8]. Utilities have become concerned with the
152
RENEWABLE ENERGY AND STORAGE
ISO
Figure 7.4. Schematic of proposed power line and wireless control connections between
electric vehicles and the grid.
number of PHEVs coming on the market because there may be insufficient supply for
the increased demand resulting from additional load for battery charging. By 2040, the
addition of PHEV battery charging in the United States will increase existing load by
18% [8]. Unfortunately, this increase in load will eventually cause voltage collapse in
amounts up to 96% of the nominal voltage in some areas, requiring the integration of
transformers, capacitors, and other power distribution devices for mitigation.
It will be critical to study the trends of daily PHEV power usage and the average
power consumption over one day to determine the impacts on the grid, market, environment, and economy. Peak driving levels will coincide with existing early and end-ofday peaking hours when users are waking up and heading to work and when they are
home, having supper, and so on. During peak hours the increased need for energy may
require users to charge their PHEVs in high-peak hours. Regulation of charge times
will help maintain stability throughout the network.
7.7
ENVIRONMENTAL IMPLICATIONS
Environmental impact mitigation is a major driver of smart grid development. The
integration/facilitation of the use of renewable energy resources for generation and
the move towards the use of PHEVs are two critical aspects of the environmental
implications of the smart grid. In addition the increased energy efficiency, demand
response and load management will result in a stable development of clean power
market.
ENVIRONMENTAL IMPLICATIONS
7.7.1
153
Climate Change
Climate change is the term commonly linked to the issue of global warming and cooling
resulting from the increased emissions of greenhouse gases (GHG). The term climate
change refers to any distinct change in measures of climate lasting for a long period
of time, that is, major changes in temperature, rainfall, snow, or wind patterns lasting
for decades or longer. Climate change may result from:
•
•
•
Natural factors, such as changes in the Sun’s energy or slow changes in the
Earth’s orbit around the Sun.
Natural processes within the climate system, for example, changes in ocean
circulation
Human activities that change the atmosphere’s composition (for example, burning
fossil fuels) and the land surface (for example, cutting down forests, planting
trees, and expanding cities and suburbs)
Global warming is an average increase in temperatures near the Earth’s surface and
in the lowest layer of the atmosphere. Increases in temperatures in the Earth’s atmosphere can contribute to changes in global climate patterns. Global warming can be
considered part of climate change along with changes in precipitation, sea level, and
so on.
Global change is a broad term that refers to changes in the global environment,
including climate change, ozone depletion, and land use change.
7.7.2
Implications of Climate Change
The key implications of climate change include:
•
•
•
•
Energy: Increased temperatures will cause an increase in energy bills as consumers use more air conditioning.
Health: Extremes of temperature such as excessive and long-term exposure to
heat will contribute to disease.
Agriculture and Wildlife: Irregular weather variability implies lack of proper
water supply, and increased temperature may result in worsening crop production
and, ultimately, rising food costs.
Water Resources: Temperature and weather irregularity increase the possibility
of flooding and droughts and impact the quality and availability of global fresh
water supply. As the water supply is affected, farmers will need to irrigate crops.
The DOE’s strategic plan for a U.S. Climate Change technology program involves:
•
•
•
The facilitation of DG, particularly RER technology
The development of energy storage technology to conserve generated energy
The advancement of DR to decrease demand
154
RENEWABLE ENERGY AND STORAGE
Consumer education and participation should increase sensitivity to the issues.
Discussion of the potential contribution of RER on climate change will consider the
projected increase, from the current 13% to 30%, as well as the following improvements
to the grid:
i.
ii.
iii.
iv.
v.
7.8
Improvement from 5% to 15% in DR systems
Improvement from 1% to 10% in consumer generation
Improvement from 47% to 90% of asset utilization
Improvement in transmission asset utilization from 50% to 80%
Improvement in distribution asset utilization from 30% to 80%
STORAGE TECHNOLOGIES
Energy storage is important for utility load leveling, electrical vehicles, solar energy
systems, uninterrupted power supply, and energy systems in remote locations. Energy
storage has always been closely associated with solar installations, including solar
heating and PV. Storage options are particularly essential when variable sources are
used in islanding and standalone power systems. Figure 7.5 presents a sample topology
for a microgrid system.
Photovoltaic (PV) Option Wind Power Option
Diesel Power Option
Switching
Logic
Charge
Controller
*Charge
Controller
Circuit Breaker
AC Bus
DC Bus
AC
DC
Storage Options
Power
Inverter
Switch
Battery Bank
Fuse
Supercapacitor
Load I
Load II
Figure 7.5. Microgrid topology with storage technologies.
Load III
STORAGE TECHNOLOGIES
155
There are two principal reasons why energy storage will grow in importance with
the increased development of RER:
1. Many important RER are intermittent, generating when the weather dictates,
rather than when energy demand dictates.
2. Many transportation systems require energy to be carried with the vehicle.
Storage options can be evaluated based on the characteristic of the application, for
example, whether the application requires portable or fixed storage methods, the
duration when storage will be operational, and the maximum power needed for the
application. The selection of the proper storage technology is based on the following
parameters:
•
•
•
•
•
•
•
•
•
•
•
Unit Size: Scale of technology. Storage technologies have an associated
range for application, for example, large units support grid-connected RER
technologies.
Storage Capacity: Total store of available energy after charging.
Available Capacity: Average value of power output based on the state of charge/
depth of discharge.
Self-discharge Time: Time required for a fully charged, non-interconnected
storage device to reach a certain depth of discharge (DOD), this is contingent on
the operational condition of the system.
Efficiency: Ratio of energy output from the device to the energy input issue of
conversion technology and design of RER and storage and conversion needed.
Durability or Life-cycle: Number of consecutive charge-discharge cycles a
storage installation can undergo while maintaining the installations and other
specifications within limited ranges. Life-cycle specifications are made against
a chosen DOD depending on the applications of the storage device.
Autonomy: Ratio between energy capacity and maximum discharge power;
indicates the maximum amount of time the system can continuously release
energy.
Mass and Volume Densities: Amount of energy accumulated per unit mass or
volume of the storage unit.
Cost: Cost of installation, operation, and maintenance of storage technology;
cost should be analyzed throughout system lifespan.
Feasibility: Degree of adaptability to the storage applications.
Reliability: Guarantee of service.
Additional information characteristics include monitoring and control equipment,
operational constraints, environmental impacts, ease of maintenance, simplicity of
design, operational flexibility, and response time for energy release. Table 7.2 compares
the options.
156
RENEWABLE ENERGY AND STORAGE
TABLE 7.2. Comparison of Storage Technology Options
Storage
Technology
Flow
Batteries
Characteristics/Particulars
Similar to lead-acid batteries, but the
electrolyte is stored in a external
container and it circulates through the
battery cell stack
Power Out
Ion Exchange
Membrane
Electrode
Advantages
Disadvantages
•
Unlimited electrical
storage capacity, the
only limitation is the
size of the
electrolyte storage
reservoir
•
Limited number
of cycles of
usage, after three
(3) to five (5)
years the system
has to be
changed
•
Use less space than
lead acid batteries
•
Too expensive
for large scale
applications
•
Virtually unlimited
cycle life
Low impedance
Rapid charging
Simple charge
methods
•
Linear discharge
voltage prevents
use of the full
energy spectrum
Low energy
density
Cells have low
voltages
High selfdischarge
Electrolyte
Flow
Electrode
Electrolyte Tanks
Advanced
Batteries
Advanced batteries include lithium-ion,
polymerion, nickel metal hybrid and
sodium sulfur type
–
Discharge
e–
+
+
OH–
+
+ battery
Li
Li
Metallic
Li
OH–
O2
O2
Air
Anode
Air Electrode (Porous Carbon)
Catalyst
Oragnic Electrolyte
Solid State Aqueous Electrolyte
Electrolyte
Super
capacitors
Electronic device with the capacity to
provide high power and energy which
have the characteristics of capacitors and
electrochemical batteries except there is
no chemical reaction.
•
•
•
•
•
The Technology
•
Current Collecting Plate
Active Electrode
Sepatator
Active Electrode
Current Collecting Plate
157
STORAGE TECHNOLOGIES
TABLE 7.2. (Continued)
Storage
Technology
Super
Conducting
Magnetic
Energy
Storage
Characteristics/Particulars
Energy stored in the magnetic field created
by the flow of direct current in a coil of
superconducting material that has been
cryogenically cooled.
Adiabatic vacuum
vessel
Radiation heat
shield
Center Inner coil case
Support member
cylinder
between coils
Outer coil case
Coolant port
•
•
•
•
•
•
Visitors Center
Switchyard
Resorvoir
Intake
Pumped-Storage Plant
Elevator
•
Energy content is
small and
short-lived
Cryogenics, cold
temperature
technology, can
be challenging
Readily available
and widely used in
high power
applications
Lower cost of
power, frequency
regulation on the
grid, and reserve
capability
Spends years in
regulatory and
environmental
review
• Can only be
implemented in
areas with hills
Conserves some
natural gas by using
low-cost, heated
compressed air to
power turbines and
create off-peak
electricity
•
Main Access Tunnel
Surge Chamber
Transformer Vault
Compressed
Air
•
Adiabatic
support leg
The process of water being pumped from
a lower reservoir uphill then allowing it
to flow downhill to through turbines to
produce electricity
Discharge
Power is available
almost
instantaneously
High power output
for a brief period
No loss of power
No moving parts
Disadvantages
Current lead
port
Superconducting
coil
Common base
Pumped
Hydro
Advantages
Powerplant Chamber
Breakers
Compressed Air Energy Storage (CAES)
utilities use electricity generated during
off-peak hours (i.e., storage hours) to
compress air and store it in airtight
underground caverns. When the air is
released from storage, it expands through
a combustion turbine to create electricity.
•
•
Low efficiency
due to the extra
reheating energy
needed to turn on
the turbines
For every
kilowatt-hour
of energy going
in, only.5
kilowatt-hour of
energy can be
taken out
(Continued)
158
RENEWABLE ENERGY AND STORAGE
TABLE 7.2. (Continued)
Storage
Technology
Flywheels
Characteristics/Particulars
Advantages
A cylinder that spins at a very high speed,
storing kinetic energy.
•
•
Utility
Rectifier
Motor/
Generator
Bus
Converter
Load
Bi-directional
Inverter
•
Flywheel
Vacuum
Housing
Magnetic
Bearings
•
•
•
Hub
Courtesy of Beacon Power
7.9
Composite
Rim
6 kWh unit
Charge and
discharge rapidly
Affected little by
temperature
fluctuations
Take up relatively
little space
Long life span
Tolerant of abuse
Lower maintenance
requirements than
batteries
Disadvantages
Power loss faster
than for batteries
TAX CREDITS
Tax credits are examples of incentives to promote customer buy-in and participation in
the utilization of RER and practice of energy efficiency. The DOE offers a number of
tax credits to consumers [9]. These include:
•
•
•
•
•
Home Energy Efficiency Improvement Tax Credit: Consumers who purchase and
install specific products, such as energy-efficient windows, insulation, doors,
roofs, and heating and cooling equipment in existing homes can receive a tax
credit for 30% of the cost, up to $1,500.
Residential Renewable Energy Tax Credit: Consumers who install solar energy
systems (including solar water heating and solar electric systems), small wind
systems, geothermal heat pumps, and residential fuel cell and microturbine
systems can receive a 30% tax credit.
Automobile Tax Credit for Hybrid Gas-Electric and Alternative Fuel Vehicles:
Individuals or businesses who buy or lease a new hybrid gas-electric car or truck
are eligible. Hybrid vehicles that use less gasoline than the average vehicle of
similar weight and that meet an emissions standard qualify for the credit.
Alternative-fuel vehicles, diesel vehicles with advanced lean-burn technologies,
and fuel-cell vehicles are also eligible for tax credits.
Automobile Tax Credit for Plug-In Electric Vehicles: The minimum amount of
the credit is $2,500 and the credit tops out at $7,500, depending on the battery
capacity.
Automobile Tax Credit for Plug-In Hybrid Conversion Kits: The credit is equal
to 10% of the cost of converting a vehicle to a qualified plug-in electric drive
motor vehicle. The maximum amount of the credit is $4,000.
SUGGESTED READING
•
7.10
159
Automobile Tax Credit for Low Speed & 2- or 3-Wheeled Vehicles: The Federal
Recovery Act creates a special tax credit for certain low-speed electric vehicles
and 2- or 3-wheeled vehicles. The amount of the credit is 10% of the cost of the
vehicle, up to a maximum credit of $2,500.
SUMMARY
Chapter 7 has described progress in the utilization of cleaner, more environmentally
responsible technologies for the electric system. Managing variability and interoperability were considered for the smart grid. Besides the increased utilization of RER, the
chapter explained the need for conversion and storage technologies which must be
further studied to determine the parameters necessary for optimal selection and implementation. New technologies covered included PHEV that advance transportation technology while providing additional storage for the smart grid.
REFERENCES
[1] J. Twidell and A.D. Weir. Renewable Energy Resources. USA: Taylor & Francis, 2006.
[2] R. Wiser and G. Barbose. “Renewables Portfolio Standards in the United States: A Status
Report with Data Through 2007,” Lawrence Berkeley National Laboratory, 2008.
[3] Thermal Energy System Specialists (TESS). The Transient Energy System Simulation Tool
(TRNSYS).
[4] National Renewable Energy Laboratory 2010, Renewable Resource Data Center: PVWatts.
[5] J.I. Rosell and M. Ibanez. “Modeling Power Output in Photovoltaic Modules for Outdoor
Operating Conditions,” Science Direct: Energy Conversion and Management 2005, 472,
424–430.
[6] M. Kaltschmitt, W. Streicher, and A. Wiese, Eds., Renewable Energy Technology, Economics
and Environment. New York: Springer, 2007.
[7] G. Boyle, “Renewable Electricity and the Grid: The Challenge of Variability,” Earthscan,
USA, 2007.
[8] C. Roe, J. Meisel, A.P. Meliopoulos, F. Evangelos, and T. Overbye. “Power System Level
Impacts of PHEVs,” 42nd Hawaii International Conference on System Sciences, 2009,
1–10.
[9] W. Kempton and J. Tomic. “Vehicle to Grid Power Fundamentals: Calculating Capacity and
Net Revenue,” Journal of Power Sources 2005, 1–12.
SUGGESTED READING
National Renewable Energy Laboratory 2010, HOMER: The Optimization Model for Distributed
Power.
8
INTEROPERABILITY,
STANDARDS, AND
CYBER SECURITY
8.1
INTRODUCTION
Deployment of the smart grid’s components and interoperability requires a substantial
overhaul of today’s standards and protocols. In addition, improving the physical and
cyber security of the network, which is notoriously vulnerable, is a top priority for the
new architectural framework. Today’s power distribution and monitoring are still in the
initial stages of becoming a smart grid, with some substation network intelligence connected by microwave, power line, and/or fiber optic point to points. Although these
core network backbones are very basic, they were never meant to securely connect
two-way digital communication devices for every home, building, and appliance
throughout a utility’s service territory. In fact, adding millions of these connections to
a distribution system is no easy task, and power companies are in the precarious position of having to prepare for the future.
If smart grids can realize their full potential, consumers, utilities, nations, and even
the environment will benefit. Unfortunately, as with nearly any new technology, much
of the focus has been on getting smart grid pilots and meters up and running, often with
little consideration for cyber security. Worse, some experts appear to believe that the
new grid’s IT networks and industrial control systems can be secured by adapting exist-
Smart Grid: Fundamentals of Design and Analysis, First Edition. James Momoh.
© 2012 Institute of Electrical and Electronics Engineers. Published 2012 by John Wiley & Sons, Inc.
160
INTEROPERABILITY
161
ing countermeasures. Therefore, the following discussion outlines the challenges for
planners and designers and the role of policy-makers in attaining reliability and secure
operations.
8.2 INTEROPERABILITY
Interoperability is “the ability of two or more systems or components to exchange
information and to use the information that has been exchanged.” The issues include
the interoperation of system components supposedly conforming to a particular standard as well as the interoperation of components across standards. Merely having
complete compliance to applicable smart grid standards is not enough to ensure interoperability. For example, the International Electrotechnical Commission (IEC) identifies
the various objects and attributes as mandatory, optional, or conditional. Communications, management, security, and application execution messages must all be well
understood by the interoperating equipment. A careful approach will include:
•
•
•
8.2.1
Reviewing the activities of governing bodies: The outcome will determine the
activities to be undertaken by smart grid users
Reviewing components before deployment: Ensure compatibility with functional
requirements
Developing internal project standards: Address continuing issues and the governing body efforts.
State-of-the-Art-Interoperability
Interoperability is a key assumption in the development of smart metering to facilitate
the competitive energy retail market [1]. There are two elements to interoperability:
technical and commercial. Technical interoperability is largely about defining the functionality for gas and electricity metering interfaces providing smart metering (format
and data content) service requirements. The definition of technical interoperability will
depend on the market model the use of smart metering. WAN and LAN communications interfaces will be explored. Consumer engagement essential to delivering consumer benefits will be identified in a cost benefit analysis. Use of case studies will have
to be forward looking to ensure that opportunities such as DSM and smart grids are
not precluded by any solution.
8.2.2
Benefits and Challenges of Interoperability
Interoperability allows a network to seamlessly and autonomously integrate all components of electric power supply, particularly monitoring and measurement equipment,
distribution and substation equipment, and management and communication equipment. The minimization of human intervention in this process is an important benefit
of this functionality.
162
INTEROPERABILITY, STANDARDS, AND CYBER SECURITY
The challenges include the need for technical enhancement of the network, adoption and adaptation of existing technologies, and development and implementation of
comprehensive standards. Procedures to address vandalism, hacking, and malicious
attacks will require the development of security protocols for authentication and validation before access is granted.
8.2.3
Model for Interoperability in the Smart Grid Environment
Several key features in the interoperable efficiency of a network have to be considered
in relation to each other and also independently [2]. The following illustrates a conceptual model of the smart grid developed by the GridWise Architecture Council (GWAC).
An eight-layer stack, termed the GWAC stack, provides a context for determining smart
grid interoperability requirements and defining exchanges of information [1]. The layers
represent the chronological processes that enable various interactions and transactions
within the smart grid. Each layer depends upon the layer beneath it and so each layer
must function properly for the entire stack to be effective [2]. As more complex functions are required by the network, more layers will be required to achieve interoperability. According to GWAC, each category/driver subdivided by layers has a special
purpose, as follows:
•
•
•
8.2.4
Technical: Emphasizes the syntax or format of the information, focusing on how
the information is represented on the communication medium
Informational: Emphasizes the semantic aspects of interoperation, paying attention to what information is exchanged and its meaning
Operational: Emphasizes the pragmatic (business and policy) aspects of interoperation, especially those pertaining to the management of electricity. [3]
Smart Grid Network Interoperability
Machines require specific data and instructions to complete tasks. The challenge is to
design language and protocols to ensure effective communication between machines
that are governed by the same protocol and even more so for those using different
protocols. The main issue is to facilitate efficient, quick, and speedy transfer of data
among and across devices. Interoperability is not limited to a physical aspect of the
network, since design engineers must also consider that as two devices try to exchange
data, the messages must now “speak the language” of network navigation and be properly “addressed” to reach the destination device.
This issue creates the need for networking standards so that machines that need to
be connected can communicate without interruption or disruption. Unfortunately, this
is not a simple task to undertake. For example, some machines may use a particular
language protocol that requires another machine attempting to communicate to complete specific system requirements, which may be outside its scope, before transmission
can be completed. In this case, the data transfer/communication may be prohibited or
incomplete, making the network inefficient.
STANDARDS
8.2.5
163
Interoperability and Control of the Power Grid
As discussed in the previous section, network control is a major issue for utilities. To
upgrade to a smart system, the network should be outfitted with equipment that can
detect problems, report back to the utility, receive control or restorative commands, and
execute them. Full control will require that all machines communicate, interpret, and
perform tasks that most machines today cannot do.
The use of SCADA and EMS have become ineffective, and leaves control centers
with the need to communicate with other control centers as well as regulatory agencies,
energy markets, independent power producers, large customers and suppliers, to keep
up with the evolving market environment [6]. Control centers must be able to connect
to machines that have smart technology to facilitate effective performance with little
or no interruptions. Ultimately, the user/customer should have some degree of autonomy over consumption with a faster, more effective response to supply disruptions.
8.3
STANDARDS
Standards are the specifications that establish the fitness of a product for a particular
use or that define the function and performance of a device or system [1]. Many standards bodies, including the National Institute of Standards and Technology (NIST),
International Electrotechnical Commission (IEC), Institute of Electrical and Electronic
Engineers (IEEE), Internet Engineering Task Force (IETF), American National Standards Institute (ANSI), North American Reliability Corporation (NERC), and the World
Wide Web Consortium (W3C) are addressing interoperability issues for a broad range
of industries, including the power industry. Table 8.1 provides a summary of the standards under development by key standards bodies.
The urgent need for the development of standards has led NIST to develop a plan
to accelerate the identification and establishment of standards “while establishing a
robust framework for the longer term evolution of the standards and establishment of
testing and certification procedures.” Based on the first phase of this work, NIST published the NIST Framework and Roadmap for Smart Grid Interoperability Standards
Release 1.0 in September 2009 [8]. In this publication, nearly 80 existing standards are
identified.
8.3.1
Approach to Smart Grid Interoperability Standards
The roadmap for interoperability by NIST includes the following applications:
•
•
•
•
•
•
Demand Response and Consumer Energy Efficiency
Wide Area Situational Awareness
Electric Storage
Electric Transportation
Advanced Metering Management
Distribution Grid Management
164
TABLE 8.1. Summary of Relevant Standards for Smart Grid Developed by Key Standards Bodies [4]
Standard Body
The International
Electrotechnical
Commission (IEC)
Institute of Electrical
and Electronic
Engineers (IEEE)
Internet Engineering
Task Force (IETF)
Description of Roles
Leading global organization which
publishes standards for electrical
electronic and related technologies
for the electric power industry.
Applicable standards have been
developed in the area of
communication for the power
industry.
Standards in all areas of electrical,
electronic and related technologies.
Standards developed in the area of
communications and interoperability.
Responsible for Internet standards,
dissemination of request for
comment (RFC) documents for
finalization of standards
Key Standards applicable to the Smart Grid Environment
IEC 61850
IEC 61968
IEC 61850
IEC 61968
IEC TC 13 and 57
IEEE 802.3
IEEE 802.11
IEEE 802.15.1
IEEE 802.15.4
IEEE 802.16
RFC 791
RFC 793
RFC 1945
RFC 2571
RFC 3820
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Substation automation, distributed generation (photovoltaics, wind
power, fuel cells, etc.), SCADA communications, and distribution
automation. Work is commencing on Plug—in Hybrid Electric
Vehicles (PHEV).
Distribution management and AMI back office interfaces
Substation automation, distributed generation (photovoltaics, wind
power, fuel cells, etc.), SCADA communications, and distribution
automation. Work is commencing on Plug—in Hybrid Electric
Vehicles (PHEV)
distribution management and AMI back office interfaces
Metering and communications for metering, specifically for AMI.
Ethernet
WiFi
Bluetooth
Zigbee
WiMax
Internet Protocol (IP)
Transport Control Protocol (TCP)
HyperText Transfer Protocol (HTTP)
Simple Network Management Protocol (SNMP)
Internet X.509 Public Key Infrastructure (PKI) for security
Standard Body
American National
Standards Institute
(ANSI)
National Institute of
Standards and
Technology (NIST)
North American
Electric Reliability
Corporation
(NERC)
World Wide Web
Consortium (W3C)
Description of Roles
Key Standards applicable to the Smart Grid Environment
Developed relevant standards for
interoperability of AMI systems
ANSI C12.19
ANSI C12.22
Publications which provide guidelines
toward secured interoperability.
NIST SP-800.53
NIST SP-800.82
Security standards for the bulk power
system which may be extended to
the distribution and AMI systems.
NERC CIP
002-009
Interoperable technologies
(specifications, guidelines, software,
and tools) for the world wide web
HTML
XML
SOAP
•
•
•
•
•
•
•
•
Metering “tables” internal to the meter
Communications for metering tables)
Recommended Security Controls for Federal Information Systems.
Guide to Industrial Control Systems (ICS) Security.
Bulk Power Standards with regards to Critical Cyber Asset
Identification, Security Management Controls, Personnel and
Training, Electronic Security Perimeter(s), Physical Security of
Critical Cyber Assets, Systems Security Management, Incident
Reporting and Response Planning, and Recovery Plans for Critical
Cyber Assets
Web page design
Structuring documents and other object models
Web services for application-to-application communications for
transmitting data
165
166
INTEROPERABILITY, STANDARDS, AND CYBER SECURITY
•
•
Cyber Security
Network Communications
These standards are applicable to transmission and/or distribution systems and
pertain to electrical power system components. Table 8.2 presents such standards.
Additionally, there are several standards identified for different levels of communication, demand response, and measurement devices. As an example, in the area of
power quality, the IEEE and other governing bodies have their own sets of standards.
8.4
SMART GRID CYBER SECURITY
Cyber security is a concept that has become increasingly prevalent with the development of the smart grid technology with the increased use of digital information and
controls technology to improve reliability, security, efficiency of the electric grid and
the deployment of smart technologies (real-time, automated, interactive technologies
that optimize the physical operation of appliances and consumer devices) for metering,
communications concerning grid operations and status, and distribution automation.
The interaction of the power, communication, and information networks are critical to
facilitating resiliency and sustainability of the infrastructures which further enhance the
provision of adequate power and support economic and social growth of the nation.
Technologies and protocols are developed for the maintenance of system, network, data,
and SCADA security while conducting vulnerability assessment, incident recognition,
recording, reporting, and recovery. Protection of network data as well as web-based or
stored data is conducted.
8.4.1
Cyber Security State of the Art
Cyber security is a critical priority of smart grid development. However, the cyber
security requirements for the smart grid are in a considerable state of flux. Cyber security includes measures to ensure the confidentiality, integrity, and availability of the
electronic information communication systems necessary for the management and
protection of the smart grid’s energy, information technology, and telecommunications
infrastructure.
This infrastructure includes information and communications systems and services
and the information contained in these systems and services. Information and communications systems and services are comprised of the hardware and software that
process, store, and communicate information, or any combination of all of these elements. Processing includes paper, magnetic, electronic, and all other media types.
Cyber security is defined as security from threats conveyed by computer or computer terminals and the protection of other physical assets from modification or damage
from accidental or malicious misuse of computer-based control facilities [7]. Smart grid
security protocols contain elements of deterrence, prevention, detection, response, and
mitigation; a mature smart grid will be capable of thwarting multiple, coordinated
TABLE 8.2. Standards for the Various Electric Grid Levels
Level
Standard
Transmission
/Distribution
Level
IEEE Standards for
Synchrophasors for Power
Systems (IEEE C37.118-2005)
Transmission
/Distribution
Level
IEEE Standard for Interconnecting
Distributed Resources with the
Electric Power System (IEEE
1547-2003)
Common Information Model
(CIM) for Power Systems (IEC
61968/61970)
Transmission
/Distribution
Level
Distribution
Communication networks and
systems in substations (IEC
61850, Ed. 1 - 2009)
IEC 61850
Distribution
/End User
Advanced Metering Infrastructure
(AMI) System Security
Requirements—AMI-SEC
(June 2009)
Distribution
/End User
American National Standard For
Utility Industry End Device Data
Tables (ANSI C12.19-2008)
Description
This standard defines synchronized phasor measurements used in power
system applications. It provides a method to quantify the measurement,
tests to be sure the measurement conforms to the definition, and error
limits for the test. It also defines a data communication protocol including
message formats for communicating this data in a real-time system.
Itemizes criteria and requirements for the interconnection of distributed
generation resources into the power grid.
167
Describes the components of a power system and power system software
data exchange such as asset tracking, work scheduling and customer
billing at an electrical level and the relationships between each
component.
Standard for the design of electrical substations which addresses issues of
interoperability, integration, intuitive device and data modeling and
naming, fast and convenient communication. It includes abstract
definitions of services, data and common data class, independent of
underlying protocols.
Provides the utility industry and vendors with a set of security
requirements for Advanced Metering Infrastructure (AMI) to be used in
the procurement process, and represent a superset of requirements
gathered from current cross-industry accepted security standards and best
practice guidance documents.
Defines a table structure for utility application data to be passed between
an end device and a computer. Does not define device design criteria nor
specify the language or protocol used to transport that data. The purpose
of the tables is to define structures for transporting data to and from end
devices.
Application
Phasor measurement
units communication
Physical and electrical
interconnections
between utilized and
distributed generation.
Application level
energy management
system interfaces.
Telecontrol
/Telemetering
Substation automation
Advanced metering
infrastructure and SG
end to end security
Revenue metering
information model
168
INTEROPERABILITY, STANDARDS, AND CYBER SECURITY
attacks over a span of time. Enhanced security will reduce the impact of abnormal
events on grid stability and integrity, ensuring the safety of society and the economy.
There is a cyber security coordination task force directed at the design of grid
security at the architectural level. Tasks include identifying used case studies with cyber
security considerations, performing risk assessment such as vulnerabilities, threats and
impacts as well as developing security architecture [2].
The underlying concept is that security should be built-in, not added-on. This
strategy includes:
•
•
Review of the system functionality and data flows with particular attention to
their similarities and differences with identified dsmart grid use cases (as documented in the NIST Roadmap).
Identification of relevant threats and the consequences/impacts if the confidentiality, integrity, availability, or accountability of the system data flows are
compromised.
Security requires many different solutions and is not relegated to encryption and
password protection. Facets of the cyber security include:
•
•
•
•
•
Security assessment and hardening of the existing systems
Vulnerability assessment
Disaster recovery
Intrusion detection incident response
Event logging, aggregation, and correlation
Another critical understanding of the issue of security for the purpose of development is the realization of the inevitability of the occurrence of breaches. This leads to
the development of contingency and recovery plans.
Table 8.3 itemizes the metamorphosis of the threats faced by the power system in
the legacy power system grid and the new grid.
Critical objectives for the development of cyber security for the smart grid environment are ensuring the confidentiality, integrity, and availability of device and system
data and communications channels, and securing logging, monitoring, alarming, and
notification. Data protection will require confidentiality of communicated and stored
data for the power system facilitated by authentication methods and the use of cryptography which includes encrypted authentication, to make it difficult for hackers. A
combination of detective, corrective, and preventive controls will address cyber security
risks. As the risks are determined, the best approach will balance efficiency, cost, and
effectiveness.
All known cyber threats are monitored with a form of detective control. These
include technologies such as host-based intrusion detection systems (HIDS) to monitor
unauthorized changes to servers and deployed systems, network-based intrusion detection systems (NIDS) to detect network-based attacks, and platform-specific controls
like virus detection and malware detection. These detective controls provide vital data
SMART GRID CYBER SECURITY
169
TABLE 8.3. Threats Facing the Electric Power System
Traditional Threats faced
by Legacy System
Impact
Location of
origination of threat
Target
Direct damage to
physical utility
Local
Indirect damage to physical assets
through damage to software systems
Local or remote
Individuals
Individuals, competitors, and
organizations
Multiple point simultaneously
Attack may be undetected or lie
dormant and then be triggered later
Continued damage associated with
attack
Attacker may have continued impact
preventing restoration
Occurrence
Single site
Immediate damage
causing obvious damage
Single episode
Restoration
Restoration after attack
Point of Attack
Duration of Damage
Threats faced by the New System
to design and testing staff about the real attacks and threats that the system faces.
Detective controls often provide forensic-quality evidence that can be used to reconstruct attacks and potentially identify and/or prosecute attackers. Additionally, data
from detective controls also assist in the selection of corrective and preventive
controls.
Corrective controls seek to restore normal operations in the event of a successful
cyber attack. Such controls are both manual, for example, a standardized procedure for
switching to a backup system, and automatic, for example, failover designs that automatically disable compromised systems and replace them with known good systems.
Corrections often seek to isolate and preserve successful attacks so that forensic analysis can proceed and permanent corrections, for example, in design, construction, or
deployment, can be established.
For example, the secure development approach seeks to address risk throughout
the life-cycle and throughout the engineering process, but preventive controls are a
logical, efficient, and effective complement to secure implementation. Some attacks,
for example, denial-of-service, can only be mitigated with a combination of secure
engineering and preventive controls. Other attacks can be partially mitigated with a
protective control, for example, network rules or role-based access controls, until a
more comprehensive change to the system can be developed, tested, and deployed. As
required, GridPoint deploys preventive controls to satisfy regulations, adhere to best
practices, or to mitigate cyber security risks, including network-level filtering and rules,
operating system-level mechanisms, and hardware-level mechanisms.
8.4.2
Cyber Security Risks
Cyber security risks appear in each phase of the project life-cycle and include risks to
managerial, operational, and technical processes. These risks may impact equipment
170
INTEROPERABILITY, STANDARDS, AND CYBER SECURITY
and systems, network management and integration, communications, control and operations, and system availability. The primary components that may be vulnerable to
security risks include IT applications, communications network, and endpoints (for
example, meters, in-home displays, and thermostats). There are substantial risks to the
integrity of data and control commands due to the exchange of information through
publicly accessible equipment, for example, smart meters using over-the-air communications technologies, for example, wireless or radio frequency, which may be intercepted and altered if not appropriately secured.
A number of system constraints need to be taken into account when satisfying
security requirements. The requirements described do not prescribe which solution, for
example, the use of narrow- or wide-band communications technologies, is most appropriate in a given setting. Such a decision is typically based on making prudent trade-offs
among a collection of competing concerns. The following trade-off must include the
considering cyber security risk:
•
•
•
•
•
•
•
•
•
•
•
•
Other business or non-functional requirements
Performance (for example, response time)
Usability (for example, complexity of interactions for users)
Upgradability (for example, ease of component replacement)
Adaptability (for example, ease of reconfiguration for use in other applications)
Effectiveness (for example, information relevant and pertinent to the business
process as well as being delivered in a timely, correct, consistent, and usable
manner)
Efficiency (for example, the provision of information through the most productive and economical use of resources)
Confidentiality (for example, protection of sensitive information from unauthorized disclosure)
Integrity (for example, accuracy, completeness, and validity of information in
accordance with business values and expectations)
Availability (for example, information being available when required by the
business process)
Compliance (for example, complying with the laws, regulations, and contractual
arrangements)
Reliability (for example, the provision of appropriate information for management
to operate the entity and exercise its fiduciary and governance responsibilities)
It is important to consider system constraints when developing applying security
requirements, which include:
•
•
•
•
Constraints
Computational (for example, available computing power in remote devices)
Networking (for example, bandwidth, throughput, or latency)
Storage (for example, required capacity for firmware or audit logs)
SMART GRID CYBER SECURITY
•
•
•
•
•
•
•
•
•
•
•
•
•
•
8.4.3
171
Power (for example, available power in remote devices)
Personnel (for example, impact on time spent on average maintenance)
Financial (for example, cost of bulk devices)
Temporal (for example, rate case limitations)
Technology
Availability
Maturity
Integration/Interoperability (for example, legacy grid)
Life-cycle
Interconnectedness of infrastructure
Applications (for example, automated user systems and manual procedures that
process the information)
Information (for example, data, input, processed and output by the information
systems in whatever form is used by the business)
Infrastructure (for example, technology and facilities, that is, hardware, operating
systems, database management systems, networking, multimedia, and the environment that houses and supports them, that enable processing the applications)
People (e.g., the personnel required to plan, organize, acquire, implement, deliver,
support, monitor, and evaluate the information systems and services. They may
be internal, outsourced or contracted as required) will consider: time, financial,
technical, operational, cultural, ethical, environmental, legal, ease of use, regulatory requirements, scope/sphere of influence
Cyber Security Concerns Associated with AMI
AMI is the convergence of the power grid, the communications infrastructure, and the
supporting information infrastructure [5]. This system of systems is constituted by a
collection of software, hardware, operators, and information and has applications to
billing, customer service and support, and electrical distribution. These applications
each have associated cyber security concerns as summarized in Table 8.4.
The development of the security domain for AMI systems is addressed in Reference
5 and a security domain model was developed to bound the complexity of specifying
the security required to implement a robust, secure AMI solution and to guide utilities
in applying the security requirements to their AMI implementation. The services shown
in Table 8.5 are descriptions of each of the six security domains. Each utility’s AMI
implementation will vary based on the specific technologies selected, the policies of the
utility, and the deployment environment.
8.4.4
Mitigation Approach to Cyber Security Risks
This process of mitigation of errors or sources of insecurity includes the following:
•
•
Identifying and classifying the information that needs to be protected
Defining detailed security requirements
TABLE 8.4. Cyber Security Concerns Associated with AMI Systems
Application
Market
Applications:
Billing
Cyber-Security Concerns
•
Confidentiality of:
Privacy of customer data, signals and location data
Integrity of:
Meter data
Signals for message and location and tamper indication
Availability of:
Meter data (for remote read), connect/disconnect service
Confidentiality of:
Access control for customer equipment via controls, price signals and
messages
Privacy of customer data and payments
Integrity of:
Control messaging and message information containing prepayment
data, usage data, rate information
Meter data for remote reading
Signals for message and location and tamper indication
Availability of:
Meter data (for remote read), connect/disconnect service, usage data,
rate information
Customer payment data and usage balances customer devices
Confidentiality of:
Access control of customer equipment including remote service switch
and HAN devices
Integrity of:
Control messaging and message information
System Data
Availability of:
Customer devices
System data
䊊
•
䊊
䊊
•
䊊
Customer
Applications
•
䊊
䊊
•
䊊
䊊
䊊
•
䊊
䊊
Distribution
System
Application
•
䊊
•
䊊
䊊
•
䊊
䊊
TABLE 8.5. AMI Security Domain Descriptions
Security Domain
Utility Edge Services
Premise Edge
Services
Communications
Services
Management
Services
Automated Services
Business Services
172
Description
All field services applications including monitoring, measurement
and control controlled by the utility
All field services applications including monitoring, measurement
and control controlled by the customer (the customer has the control
to delegate to third party)
Applications that relay, route, and field aggregation, field
communication aggregation, field communication management
information
Attended support services for automated and communication services
(includes device management)
Unattended collection, transmission of data and performs the
necessary translation, transformation, response, and data staging
Core business applications (includes asset management)
CYBER SECURITY AND POSSIBLE OPERATION
•
•
•
•
•
•
•
•
•
•
•
•
173
Reviewing the proposed security architecture that is designed to meet the
requirements
Procuring a system that is designed to meet the specified security requirements
and includes the capability to be upgraded to meet evolving security standards
Testing the security controls during the test and installation phase
Obtaining an independent assessment of the security posture before deployment
Developing a remediation plan to mitigate the risks for identified vulnerabilities
Installing a system with built-in management, operational, and security controls
Monitoring and periodically assessing the effectiveness of security controls
Migrating to appropriate security upgrades as security standards and products
mature
Monitoring of communication channels
Monitoring spike in usage (meter reading) to detect possible failures or tampering
with the devices
Making sure devices synchronize with the network within a given time frame to
detect tampering, potential problems, and device failures.
Penetration testing will be performed using the latest hacking techniques, to
attempt to break into the systems, identifying possible vulnerabilities, and
remotely validating the authenticity of the software running in the meters.
8.5 CYBER SECURITY AND POSSIBLE OPERATION FOR IMPROVING
METHODOLOGY FOR OTHER USERS
Every communication path that supports monitoring and control of the smart grid is a
two-way communication path. Each path is a potential attack path for a knowledgeable
attacker. There are many potential entry points physically unprotected. Wireless networks can be easily monitored by attackers and may be susceptible to man-in-themiddle (MitM) attacks. Security mechanisms in place are intended to prevent
unauthorized use of these communication paths, but there are weaknesses in these
mechanisms. The history of security in complex networks implies that more vulnerability is yet to be discovered. Thus, the key points include:
1. Using spot checks on systems to go beyond the current paper chase approach
to validating CIP compliance
2. Acknowledging that attackers and malware will find ways around/through
current outer- wall–based network defenses, instituting a less-perimeter, defenseoriented approach to security controls with guidance on use of DMZs between
internal networks.
174
8.6
INTEROPERABILITY, STANDARDS, AND CYBER SECURITY
SUMMARY
This chapter has delved into the fundamental tools and techniques essential to the
design of the smart grid. The tools and techniques were classified into: (1) computational techniques and (2) communication, measurement, and monitoring technology.
Based on the performance measures, that is, controllability, interoperability, reliability,
adaptability, sustainability, efficiency, stochasticity, and predictivity, the chapter identified the most suitable applications of the tools. Ongoing work in the critical area of
standards development by NIST and IEEE was explained, including consideration
of the available standards to be adopted and/or augmented for application. The issue
of interoperability was presented as it pertains to present grid technologies and the
introduction of newer technologies. Acknowledging the grid’s increasing dependence
on communication and information systems is necessary to any discussion of the challenges of developing and deploying adequate cyber security protections.
REFERENCES
[1] J.A. Momoh. Electric Power System Application of Optimization, New York: Marcel
Dekker, 2001.
[2] J.L. Marinho and B. Stott. “Linear Programming for Power System Network Security
Applications,” IEEE Transactions on Power Apparatus and Systems 1979, vol. PAS-98,
pp. 837–848.
[3] R.C. Eberhart and J. Kennedy. “A New Optimizer Using Particle Swarm Theory,” Proceedings of the Sixth International Symposium on Micromachine and Human Science, 39–31,
1995.
[4] G. Riley and J. Giarratano. Expert Systems: Principles and Programming, Boston: PWS
Publisher, 2003.
[5] A. Englebrecht. Computational Intelligence: An Introduction. John Wiley & Sons, Ltd., 2007.
[6] M. Dorigo and T. Stuzle. Ant Colony Optimization. Cambridge, MA: Massachusetts Institute
of Technology, 2004.
[7] “Appendix B2: A Systems View of the Modern Grid-Sensing and Measurement,” National
Energy Technology Laboratory, 2007.
[8] Report to NIST on the Smart Grid Interoperability Standards Roadmap—Post Comment
Period Version. http://www.nist.gov/smartgrid/upload/Report_to_NIST_August10_2.pdf.
SUGGESTED READINGS
Z. Alaywan and J. Allen. “California Electric Restructuring: A Broad Description of the Development of the California ISO,” IEEE Transactions on Power Systems, 1998, 13, 1445–1452.
A.G. Barto, W.B. Powell, D.C. Wunsch, and J. Si. Handbook of Learning and Approximate
Dynamic Programming. IEEE Press Series on Computational Intelligence, 2004.
T. Bottorff. “PG&E Smart Meter: Smart Meter Program,” NARUC Summer Meeting, 2007.
B. Brown. “AMI System Security Requirements,” UCAIUG: AMI-SEC-ASAP, 2008.
SUGGESTED READINGS
175
M. Dorigo and T. Stutzle. “The Ant Colony Optimization Metaheuristic: Algorithms, Applications
and Advances.” In F. Glover and G. Kochenberger, eds.: Handboook of Metaheuristics.
Norwell, MA, Kluwer, 2002.
B. Milosevic and M. Begovic. “Voltage-Stability Protection and Control Using a Wide-Area
Network of Phasor Measurements,” IEEE Transactions on Power Systems 2003, 18,
121–127.
J. Momoh, Electrical Power System Applications of Optimization, Boca Raton, FL: CRC Press,
2008.
“NIST Framework and Roadmap for Smart Grid Interoperability Standards,” Office of the
National Coordinator for Smart Interoperability, Release 1.0, 2009.
A.G. Phadhke. “Synchronized Phasor Measurements in Power Systems,” IEEE Computer Applications in Power 1993, 6, 10–15.
D. Shirmohammadi, B. Wollenberg, A. Vojdani, P. Sandrin, M. Pereira, F. Rahimi, T. Schneider,
and B. Stott. “Transmission Dispatch and Congestion Management in the Emerging Energy
Market Structures,” IEEE Transactions on Power Systems 1998, 13, 1466–1474.
H. Singh and F.L. Alvarado. “Weighted Least Absolute Value State Estimation Using Interior
Point Methods”, IEEE Transactions on Power Systems 1994, 9.
P.K. Skula and K. Deb. “On Finding Multiple Pareto-optimal Solutions Using Classical and
Evolutional Generating Methods,” European Journal of Operational Research 2007, 181,
1630–1652.
“Smart Grid Communication Architecture,” Pike Research, 2011.
“Smart Grid Initiatives White Paper,” OSI Revision 1.1, 2009.
T.M. Smith and V. Lakshmanan. “Utilizing Google Earth as a GIS Platform for Weather Applications,” 22nd International Conference on Interactive Information Processing Systems for
Meteorology, Oceanography, and Hydrology, 2006.
C.W. Taylor. “The Future in On-Line Security Assessment and Wide-Area Stability Control,”
IEEE Power Engineering Society Winter, 2000, 1, 78–83.
W.H. Zhange and T. Gao. “A Min-Max Method with Adaptive Weightings for Uniformly Spaced
Pareto Optimum Points,” C4omputers and Structures 2006, 84, 1760–1769.
L. Zhao and A. Abur. “Multiarea State Estimation Using Synchronized Phasor Measurements,”
IEEE Transactions on Power Systems 2005, 20, 2005.
D. Zhengchun, N. Zhenyong, and F. Wanliang. “Block QR Decomposition Based Power System
State Estimation Algorithm,” ScienceDirect 2005.
9
RESEARCH, EDUCATION,
AND TRAINING
FOR THE SMART GRID
9.1
INTRODUCTION
The smart grid will require engineers and professionals with greater expertise and training than the skilled workforce of today. In addition to the technological aspects of
development [1], engineers will need to study manufacturing, data management, asset
optimization, and policy and protocol development. The smart grid will also depend
upon expanding current research efforts in the areas of cyber security, controls, communication, computational intelligence techniques, and decision support tools [2].
Operation and management of electricity generation, transmission, and distribution
are changing due to technological and power marketing developments. Technological
changes are being driven by the introduction of emerging technologies such as power
electronics, DG, RER, microgrids, digital protection coordination, supervisory control,
and energy management. The market-driven power business environment has created a
need for nonelectrical engineering topics such as operations research and economics.
9.2
RESEARCH AREAS FOR SMART GRID DEVELOPMENT
Work is under way in developing research support for the intelligent grid in FACTS
technology coordination and placement for real-time application, PMUs for real-time
Smart Grid: Fundamentals of Design and Analysis, First Edition. James Momoh.
© 2012 Institute of Electrical and Electronics Engineers. Published 2012 by John Wiley & Sons, Inc.
176
RESEARCH AREAS FOR SMART GRID DEVELOPMENT
177
TABLE 9.1. Smart Grid Deployment
Problem
Classical method
Stability
Lyapunov’s
Method;
Transient energy
function;
Bifurcation
method;
Eigenvalue
evaluation;
Interior point
method;
Trust region
method;
minimum cost
while meeting
the constraints;
OPF
Economic
Dispatch/
Unit
Commitment
Reliability
State
enumeration;
Effects analysis;
Reliability indices
Improvement
Currently for static model
Recommendation for
smart grid deployment
Quasi-steady state
modeling;
Index for fast and accurate
evaluation;
Real time computation;
using leaning algorithm
and ADP for control
coordination
SQP, rSQP
ADP and heuristic, hybrid
methods to account for
prediction and
stochasticity.
Knowledge based system;
Computational Intelligent
method; to account for
uncertainties and
randomness
Computational Intelligence
and its hybrid method
together with dynamics
in the data
Variants of nonlinear
interior point methods;
Probabilistic input data
for reliability study
voltage stability, and reliability monitoring and control. Other research includes
advanced distribution automation capable of handling DSM functions with real-time
pricing options and mechanisms.
The research activities can be classified by the tools required (as shown in Table 9.1):
•
•
•
•
Simulation and analysis tools: Simulate energy markets and energy systems
and validate the vision. These tools will combine operations and economics in a
single model to analyze and monitor the system as changes are implemented to
determine the impacts and ensure fairness.
Development of smart technologies from the government and industry: Jump-start the transformation
Testbeds and demonstration projects: Provide experiments of increasing scale
to prove the worth of these technologies or reveal their faults; build momentum
for change, reduce the perception of risk, and build acceptance of the concept of
a transformed energy grid [3].
New regulatory, institutional, and market frameworks: Support a climate of
innovation as technologies develop and evolve; fulfill the need for examination,
provided to the network through education and development of research and
training.
178
RESEARCH, EDUCATION, AND TRAINING FOR THE SMART GRID
9.3
RESEARCH ACTIVITIES IN THE SMART GRID
The critical objectives of technical research include:
1. Develop advanced techniques for measuring peak load reductions and energyefficiency savings from smart metering, DR, DG, and electricity storage systems
2. Investigate means for DR, DG, and storage to provide ancillary services
3. Conduct research to advance the use of wide-area measurement and control
networks, including data mining, visualization, advanced computing, and secure
and dependable communications in a highly distributed environment
4. Test new reliability technologies including those concerning communications
network capabilities in a control room environment [4, 5] against a representative set of local and wide area outage scenarios
5. Identify communications network capacity needed to implement advanced
technologies
6. Investigate the feasibility of a transition to time of use (TOU) and real-time
electricity pricing
7. Develop algorithms for use in electric transmission system software applications
8. Promote the use of underutilized electricity generation capacity in any substitution of electricity for liquid fuels in the transportation system of the United
States
9. Develop interconnection protocols to enable electric utilities to access electricity
stored in vehicles to help meet peak loads
9.4
MULTIDISCIPLINARY RESEARCH ACTIVITIES
In addition to technical research, interdisciplinary research areas, such as economics,
finance, policy, and environmental science, will incorporate aspects of development
and implementation. Systems engineering courses such as intelligent systems and adaptive controls, pricing for new and emerging power markets, financial engineering,
socioeconomics, and studies on climate change and environmental implications are all
relevant to this discipline. The following lists some of the activities which demonstrate
the need for interdisciplinary research and education:
•
•
•
•
•
Development of advanced techniques for measuring peak load reductions and
energy-efficiency savings
Investigate means for demand response, distributed generation, and storage to
provide ancillary services
Conduct research to identify and advance the use of wide-area measurement and
control networks
Research on network asset management and architectures
Test new reliability technologies, including those concerning communications
network capabilities, in a grid control room environment
SMART GRID EDUCATION
•
•
9.5
179
Investigate the feasibility of a transition to TOU and real-time electricity pricing
Development of algorithms for use in electric transmission system software
applications
SMART GRID EDUCATION
Creating a professional development curriculum is a vital component for the future
power industry. New areas of study, delivery tools, and professional development career
paths need to be introduced periodically. The development of educational schemes will
require skills and technologies outside of power engineering. Development of infrastructure to facilitate the delivery of courses such as the development of a Web-based
information data bank, interactive and distant power laboratory experiments, and
encouraging international cooperation for knowledge and expertise exchange are
necessary.
Smart grid fundamentals will include definitions, architecture, metrics of performance requirements, discussion of development of analytical and decision support
tools, as well as RER. Grid design will be based on cross boundaries of knowledge in
communication theory, optimization, control, social and environmental constraints, and
dynamic optimization techniques.
Conventional Electrical Engineering courses are directed toward the development
of engineers for jobs relating to the operation of the electric power distribution and
telecommunications infrastructure, designing communication systems, and developing
electronics or control systems. The power system curriculum allows for an introduction
to theory or practical work in many of these areas with a major focus on the following
fundamental modules of a current modular power engineering program [1]:
•
•
•
•
Utilization and management of power and energy
Application of new emerging technologies for power quality
High voltage engineering; power electronics control
Multidisciplinary power engineering related topics
Desired elements of the curriculum have been identified to facilitate the development
of the smart grid include:
(i)
(ii)
(iii)
(iv)
(v)
(vi)
(vii)
direct digital control
roles of system operators
power systems dynamics and stability
electric power quality and concomitant signal analysis
transmission and distribution hardware and the migration to middleware
new concepts in power system protection
environmental and policy issues
180
RESEARCH, EDUCATION, AND TRAINING FOR THE SMART GRID
(viii) reliability and risk assessment
(ix) economic analysis, energy markets, and planning
Course development can be established at the senior undergraduate level as well
as for graduate students. Outside the university environment, modules can be geared
toward the technical staff of power systems and policy-makers. Integration of technical
components aimed at providing exposure to the smart grid design is needed. The new
curriculum requires the following key features:
•
•
•
•
•
•
Planning and operation under uncertainties
Use of real-time measurements, techniques and tools such as PMUs and SE for
analysis of stability, reliability, and efficiency.
Renewable energy with vulnerability and penetration strategies and associated
storage technologies
Performance measures and issues which include sustainability, power quality,
interoperability, and cyber security
Development of new adaptive and stochastic optimization techniques that will
facilitate resource allocations and scheduling such as unit commitment, restoration, and reconfiguration
Marketing and pricing ancillary services and business cases for smart grid
deployment
A sample syllabus for a smart grid fundamentals course will be divided into several
modules [2].
9.5.1
•
•
•
Module 1: Introduction
What is the smart grid?
Working definitions and associated concepts
Smart grid functions
This module will introduce the smart grid along with a history of the development of
the smart grid including the contributions of key players and policy developments. A
working definition and characteristics will also be included.
9.5.2
•
Module 2: Architecture
Components and Architecture of Smart Grid Design
This module will review the proposed architecture. The fundamental components of
smart grid designs will include: transmission automation, system coordination, situation
assessment, system operations, distribution automation, renewable integration, energy
SMART GRID EDUCATION
181
efficiency, DG and storage, demand participation signals and options, smart appliances,
PHEVs, and storage.
9.5.3
•
•
Module 3: Functions
Review the functions of the smart grid
Evaluation of components: generation, transmission, distribution, and end-user
This module will partition the smart grid with layers of intelligence and functions by
technology, new tools, and innovations in energy supply and consumption for efficiency
and survivability.
9.5.4
•
Module 4: Tools and Techniques
Computational Techniques
Analytic methods (decision support tools, static and dynamic optimization
techniques)
Computational Intelligence Techniques
Introduction to Technologies
Communication technology
Sensing, metering, and measurement technologies
RER
䊊
•
•
䊊
䊊
䊊
This module will introduce classical tools and techniques, new global optimization
techniques, and cyber technology for anticipatory, predictability, and adaptability to
sustainability and resiliency of the infrastructure and its interdependencies. Evaluation
of the tools will be conducted considering the performance measures for their development, particularly controllability, interoperability, reliability, adaptability, sustainability,
anticipatory behavior, and security.
9.5.5
•
•
•
Module 5: Pathways to Design
Selection criteria for tools and techniques
Advanced optimization and control techniques
Automation at generation, transmission, distribution, and end-user levels
This module will apply advanced optimization techniques for the automation of various
grid level functions.
9.5.6
•
•
Module 6: Renewable Energy Technologies
Introduction
Storage Technologies
182
RESEARCH, EDUCATION, AND TRAINING FOR THE SMART GRID
•
•
•
EVs and plug-in hybrids
Environmental impacts and climate change
Economics
This module will introduce renewable energy technologies and study the sources,
storage and electronic technologies, as well as EVs and plug-in hybrids. Characteristics,
advantages and disadvantages of these technologies will be discussed and the associated
environmental and economic issues emphasized.
9.5.7
•
•
•
Module 7: Communication Technologies
Introduction Network topologies
WAMS
AMI
This module will discuss communication technology for the development of open
architecture that securely integrates smart sensors and control devices, control centers,
protection systems, and end-users.
9.5.8
Module 8: Standards, Interoperability, and Cyber Security
This module will involve studies on the development and implementation of standards,
interoperability, and cyber security for the smart grid environment.
9.5.9
Module 9: Case Studies and Testbeds
This module will present case studies of generation, transmission, and distribution.
Implementation of various technologies at industry locations and the establishment
and utilization of academic and research testbeds will be included. Instruction will
extend beyond the lecture format, with project based learning (PBL), case studies, and
experiments.
9.6
TRAINING AND PROFESSIONAL DEVELOPMENT
For several years, the power industry has faced the realization that the trained workforce
is aging much like the system itself. Smart grid deployment is an opportunity to address
these two issues at once. Programs in place to upgrade the skills of existing workers
and to train future workers include the new Digital and Distributed Power Systems
Training Program. This expertise entails that both operators and engineers have high
coordination skills including multidisciplinary knowledge [5].Training and re-education
must equip current and potential employees at all levels of grid development with
knowledge of advanced cyber security tools and technologies for critical controls
systems, which the Department of Homeland Security has identified as top priorities.
REFERENCES
183
Industry experts and regulatory bodies are unanimous in stating that resilience and
cyber security are critical areas of concern for the future grid.
9.7
SUMMARY
This chapter has highlighted research and education in a proactive approach for achieving a robust, scalable power system network development. Nine educational modules
to promote smart grid awareness and enhancement were described. The importance of
upgrading workforce skills and training the future workforce was emphasized.
REFERENCES
[1] G. Heydt et al. “Professional Resources to Implement the ‘Smart Grid,’ ” NAPS Power Symposium, 2009.
[2] J. Momoh. “Fundamentals of Analysis and Computation for the Smart Grid,” IEEE PES
General Meeting, 2010.
[3] J. Rosell and M. Ibanez. “Modeling Power Output in Photovoltaic Modules for Outdoor
Operating Conditions,” Science Direct: Energy Conversion and Management 2005, 47,
2424–2430.
[4] M. Kaltschmitt, W. Streicher, and A. Wiese, eds. Renewable Energy Technology, Economics
and Environment. Springer, 2007.
[5] G. Boyle. Renewable Electricity and the Grid: The Challenge of Variability, Earthscan, 2007.
10
CASE STUDIES AND TESTBEDS
FOR THE SMART GRID
10.1
INTRODUCTION
Much work has been undertaken in the areas of policy and standards, framework development, and the assessment of new, intelligent tools to achieve the objectives of the
smart grid. With the financial support of utilities and government agencies, the work is
primarily directed to the preparation of the smart grid environment, particularly in terms
of retrofitting and increasing the efficiency of the grid. Installation of WAMs and smart
metering upgrades are two additional areas to which utilities are committing time and
resources. The development of tools, packages, demonstration environments, and testbeds is important for the research, education, and technological growth that will
support successful deployment.
10.2
DEMONSTRATION PROJECTS
Demonstration projects related to the development of WAMS and voltage monitoring
were completed by the DOE feasibility and design of WAMS [1] as early as 1996.
DOE stimulus funding has been provided for numerous utilities to engage in technical
Smart Grid: Fundamentals of Design and Analysis, First Edition. James Momoh.
© 2012 Institute of Electrical and Electronics Engineers. Published 2012 by John Wiley & Sons, Inc.
184
MICROGRID WITH RENEWABLE ENERGY
185
manufacturers upgrade, and research support to academic institutions. Projects using
smart meter and distribution technology and consumer/end-user technology integration
and deployment are organized and used by some utilities. [3]
Smart grid stimulus funds awarded by DOE [2] for 2009 totaled $620 million. Two
categories of demonstration projects were funded: (i) fully integrated regional smart
grid demonstrations for communication technologies, sensing and control devices,
smart meters and in-home monitoring systems, energy storage options, and renewable
energy integration; and (ii) utility-scale energy storage projects.
Commonwealth Edison, Dominion Virginia Power, and Duke Energy were among
the utilities receiving $10–200 million primarily for smart meters throughout the United
States. Southern California Edison and several other companies received grants for
synchrophasor installation, wind energy storage, DR, distribution automation, and
regional smart grid demonstration projects. The GridComm project by San Diego Gas
& Electric in conjunction with Cisco, IBM, and Arcadian Networks developed a wireless smart grid network.
10.3
ADVANCED METERING
Utilities such as Pepco are investing in the installation and integration of smart meters.
The Pacific Northwest Smart Grid Demonstration Project aims to validate new smart
grid technologies and business models and provide two-way communication between
distributed generation, storage, demand assets, and the existing grid infrastructure while
quantifying smart grid costs and benefits and advanced standards for interoperability
and cyber security approaches.
10.4
MICROGRID WITH RENEWABLE ENERGY
The feasibility of a microgrid as a standalone system will be improved if it can generate
enough power for the load it is designed for. Microgrid development is a customer
system-oriented project meant to enable smart grid technologies and functions for
buildings and facilities. The development of an integrated system encompasses the
implementation of smart grid equipment, devices, and software, as well as installation
and the improvement of advanced smart meter technologies into a community to impact
the distribution system. In most implementations, microgrid development encompasses
the retrofitting infrastructure and DER resource installation within the identified system.
The intellectual research which will support this collaborative will involve an economic
and benefit analysis, a platform for the analysis of network performance which constitutes the system level of the intelligent microgrid in addition to a platform for decision
support.
Figure 10.1 depicts a sample testbed topology design for such a study, with two
PV and wind technology. In the microgrid topology [2], storage technologies are necessary, particularly for islanded or stand-alone operation of the microgrid.
186
CASE STUDIES AND TESTBEDS FOR THE SMART GRID
PV Option
Subtransmission
line
Bus 1
G1
PV Modules
Switching
Logic
Power
Inverter
GridCo.
Synchronizer
Switch
Bus 3
AC
Control
Charge
Controller
Bus 2
DC
AC
Wind
Option
DC
Switching
Logic
Charge
Controller
Storage
Technologies
Bus 4
Load
Battery
Bank
Load I
Switching
Logic
Switching
Logic
Load II Load III
CustLoad
Super
Capacitor
Figure 10.1. Sample microgrid testbed environment.
10.5
POWER SYSTEM UNIT COMMITMENT (UC) PROBLEM
The objective function of the UC problem can be formulated as the sum of the costs
of all units over time, and presented mathematically as [4, 5]:
T
F=
N
∑ ∑ [u (t )F (t )) + S (t )]
i
i
i
t =1 i =1
The constraint models for the UC optimization problem are as follows:
•
System energy balance
N
∑ [u (t )P (t ) + u (t − 1)P (t − 1)] = P (t )
0.5
i
gi
i
gi
i =1
•
Energy and power exchange
Ei (t ) = 0.5[ Pgi (t ) + Pgi (t − 1)]
D
187
POWER SYSTEM UNIT COMMITMENT (UC) PROBLEM
•
Spinning reserve requirements
N
∑ u (t ) P (t ) ≥ P (t ) + P (t )
i
gi
D
R
i =1
•
Unit generation limits
Pgimin ≤ Pgi (t ) ≤ Pgimax
With t ∈ {1,T} and i ∈ {1,N} in all cases where:
F : Total operation cost on the power system
Ei(t) : Energy output of the ith unit at hour t
Fi(Ei(t)) : Fuel cost of the ith unit at hour t
ui(t) : Ratio of generation output and capability
N : Total number of units in the system
T : Total time under which UC is performed
Pgi(t) : Power output of the ith unit at hour t
Pgimax : Maximum power output of the ith unit
Pgimin : Minimum power output of the ith unit
Si(t) : Start-up cost of the ith unit at hour t
In the reserve constraints, the classifications for reserve include units on spinning
reserve and units on cold reserve under the conditions of banked boiler or cold start.
Lagrangian relaxation is commonly used to solve UC problems. It is much more
beneficial for utilities with a large number of units since the degree of suboptimality
goes to zero as the number of units increases. It is easily modified to model characteristics of specific utilities and It is relatively simple to add unit constraints. The drawback
of Lagrangian relaxation is its inherent suboptimality.
T
L (λ , μ , ν ) =
N
∑ ∑ [C (P (t )) + S ( x (t ))] + λ (t ) (P (t ) + P (t ) − ∑ P ) + μ(t )(P
i
gi
i
i
d
R
gi
max
gi
− Pgi )
t =1 i =1
where λ(t), µ(t) are the multipliers associated with the requirement for time t.
Solution Approach using ADP Variant for the UC Problem: ADP is able to optimize
the system over time under conditions of noise and uncertainty. If optimal operation
samples are used to train the networks of the ADP, the neural network can learn to
commit or adapt the generators, follow the operators’ patterns, and change the operation
according to the load changing. Figure 10.2 is a schematic diagram for implementations
of heuristic dynamic programming (HDP). The input of the action network is the states
of the generators and the action is how to adjust their output. Output J represents the
cost-to-go function and the task is to minimize the J function.
188
CASE STUDIES AND TESTBEDS FOR THE SMART GRID
Action
Network
u(t + 1)
Critic
Network
J(t + 1)
X(t +1)
Σ
System
U(t)
X(t)
Action
Network
Critic
Network
u(t)
J(t)
Figure 10.2. Scheme of implementation of HDP.
In this diagram, the input is the state variable of the network, and it is the cost of
the generation vector presented as X = [C(Pgi)]. The output is the control variables of
units, and it is the adjustment of unit generation presented as: u = [ΔPg]. The utility
function is local cost, so it is a cost function about unit generation within any time
interval presented as U = f(P,t). After transposing the power system variables using
these guidelines, the scheme of implementation of HDP include the following
computations.
The error of the critic network is:
eC (t ) = γ J (t ) − J (t + 1) − U (t )
and the updating weight using:
wC (t + 1) = wC (t ) + ΔwC (t )
and
⎡ ∂e ( t ) ⎤
ΔwC (t ) = ηeC ⎢ − C ⎥
⎣ ∂wC (t ) ⎦
where
∂hCj′
∂EC
∂E ∂e ∂h
= C ⋅ C ⋅ Ck ⋅ (1)
(1)
∂wCij ∂eC ∂yCk ∂hCk
′ ∂wCij
⎤ (2)
⎡1
xi
= γ eC ⋅ ⎢ (1 − hCj2 ) ⎥ ⋅ wCj
⎦
⎣2
∂EC
∂E ∂e
∂y
= C ⋅ C ⋅ Ck
= γ eC yCk
(2)
(2)
∂wCjk ∂eC ∂yCk ∂wCjk
189
POWER SYSTEM UNIT COMMITMENT (UC) PROBLEM
I : Number of elements in R vector
J : Number of hidden layer nodes
K : Number of output layer nodes
M : Number of elements in u (action) vector
hC′ : Hidden layer input nodes
hC : Hidden layer output nodes
yC′ : Output layer input nodes
yC : Output layer output nodes
wC(1) : Weights between input and hidden layers
wC( 2 ) : Weights between hidden and output layers
x : Input layer nodes
The error of the action network is computed as:
e A (t ) = J (t ) − U (t )
and the updating weight is
w A (t + 1) = w A (t ) + Δw A (t )
and
⎡ ∂e ( t ) ⎤
Δw A (t ) = ηeA ⎢ − A ⎥
⎣ ∂w A ( t ) ⎦
where
∂E A
∂E A ∂eA ∂J k ∂y Ak ∂y Ak
′
⋅
=
⋅
⋅
⋅
2)
2)
∂eA ∂J k ∂y Ak ∂y Ak
∂w(Ajk
′ ∂w(Ajk
⎡
⎡1
2 ⎤
)⎥ ⋅ ⎢
= γ eA hAj ⋅ ⎢ (1 − hAj
⎦ ⎢⎣
⎣2
J
∑w
(2)
Cj
j =1
⎤
1
(1)
(1 − hCj2 )wCij
⎥
2
⎥⎦
∂E A
∂E A ∂eA ∂J k ∂y Ak ∂y Ak
′
⋅
=
⋅
⋅
⋅
1)
1)
∂eA ∂J k ∂y Ak ∂y Ak
∂w(Aij
′ ∂w(Aij
⎡1
⎤ ⎡
2)
2 ⎤ ⎡1
xi ⋅ ⎢ (1 − hAj
) ⎥ ⋅ ⎢ (1 − y 2Ak ) ⎥ ⋅ ⎢
= γ eA w(Ajk
⎣2
⎦ ⎣2
⎦ ⎢⎣
J
∑w
(2)
Cj
j =1
The corresponding calculation steps are as follows:
⎤
1
(1)
(1 − hCj2 )wCij
⎥
2
⎥⎦
190
CASE STUDIES AND TESTBEDS FOR THE SMART GRID
Step 1: Use the sample data to pretrain the action network. The error is the difference between the output and the real value.
Step 2: Use the sample data to train the critic network with the pre-trained and
unchanged action network. Use Equations (7) ∼ (12) to update the weights. Then
begin to apply the mature ADP network in the real work.
Step 3: Input the current state data X(t) to the action network.
Step 4: Get the output u(t) of the action network. Input u(t) to the system function
to get the state of next time X(t + 1).
Step 5: Use the state of next time X(t + 1) to get the action of next time u(t + 1).
Step 6: Input the action and state of different time u(t), X(t) and u(t + 1), X(t + 1)
to different critic network, respectively, and J functions for different time J(t),
J(t + 1) are obtained.
Step 7: Back propagate and update the weights of the critic and action networks.
Then time t = t + 1.
Results: Figure 10.3 from Reference 6 shows the structure of the neural network in
HDP for a 3-generator system.
Figure 10.4 shows the load curve of a 3-generator, 6-node system. The closeness
of the line graphs indicates that the ADP method generates correct results.
After training, the HDP can give the generation plan, which is very close to the
optimal plan. The HDP [4] method can deal with the dynamic process of UC, and easily
find a global optimal solution, which is difficult for classical optimization methods.
Figure 10.5 shows the generation schedule of a 3-generator system.
In Figure 10.5, X1, X2, and X3 present the output of the three generators, respectively, and [X1], [X2], and [X3] present their expected (or optimal) output.
Action Network
Critic Network
R(t)
J(t)
R(t)
Weight wa
Control action
u(t)
Weight wc
System
function
Figure 10.3. Structure of the neural network in HDP.
191
ADP FOR OPTIMAL NETWORK RECONFIGURATION
Power(pu)
Load Curve
4
3.5
3
2.5
2
1.5
1
0.5
0
1
3
5
7
9
11
13
15
17
19
Time(hour)
Figure 10.4. Load curve of 3-generator, 6-node system.
2.5
2
output power
X1(t)
1.5
[X1(t)]
X2(t)
1
[X2(t)]
X3(t)
0.5
[X3(t)]
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
time
Figure 10.5. Generation schedule for the UC problem solved using ADP.
10.6 ADP FOR OPTIMAL NETWORK RECONFIGURATION IN
DISTRIBUTION AUTOMATION
Distribution networks are generally configured radially for effective and noncomplicated protection schemes. Under normal operating conditions, distribution feeders
may be reconfigured to satisfy the objectives of minimum distribution line losses,
optimum voltage profile, and overloading relief. The minimum distribution line loss is
formulated as:
Minimize
∑zi
subject to:
[ A]i = I
b b
192
CASE STUDIES AND TESTBEDS FOR THE SMART GRID
where:
Zb : Impedance of the branch
Ib : complex current flow in branch b
i : m-vector of complex branch currents
A : n × m network incidence matrix, whose entries are
= +1 if branch b starts from node p
= −1 if branch b starts from node b
= zero if branch b is not connected to node p
m : Total number of branches
n : Total number of network nodes
I : n-vector of complex nodal injection currents
The illustrative example problem is solved using the integer interior point method
presented in Reference 7. Figure 10.8 shows the ADP method for a 5-bus system.
It involves the development of an ADP framework which involves (a) action
network, (b) critic network, and (c) the plant model as shown in Figure 10.6 for network
distribution reconfiguration.
The algorithm to solve this problem using ADP is presented in Figure 10.8.
To solve the optimal distribution reconfiguration problem using the ADP algorithm,
we need to model the four parts of system structure shown in Figure 10.8: action
vectors, state vectors, immediate rewards, and the plant. The system is tested with a
5-bus and a 32-bus system. We discuss the different parts of the ADP implementation
structure as follows:
Source Node
S1
1
S2
S3
2
S6
S4
3
NO
S5
4
NO S7
5
Closed Line
Open Line
Figure 10.6. Small power system for network distribution reconfiguration problem.
193
ADP FOR OPTIMAL NETWORK RECONFIGURATION
Critic
Network
A(t)
R(t)
Action
Network
γ
N
J(t)
R(t)
J(t – 1)
J(t)/A(t)
N
Plant
(Dist flow)
U(t)
R(t)
Figure 10.7. ADP structure network reconfiguration.
Rewards (Utility function): Optimal reconfiguration involves selecting the best
set of branches to be opened, one from each loop, such that the resulting radial
distribution system has the desired performance. Amongst the performance
criteria considered for optimal network reconfiguration, the one selected is the
minimization of real power losses. Application of the ADP to optimal reconfiguration of radial distribution systems is linked to the choice of an immediate
reward U, such that the iterative value of J is minimized, while the minimization
of total power losses is satisfied over the whole planning period. Thus, we
compute the immediate reward as:
U = −Total Losses
Action vectors: If each control variable Ai is discretized in dui levels (for example,
branches to be opened one at each loop of radial distribution systems), the total
number of action vectors affecting the load flow is:
m
A=
∏d
ui
i =1
where m expresses the total number of control variables (for example, the total
number of branches to be switched out).
The control variables comprise the sets of branches to be opened, one from each
loop. From the network above, we can easily deduce from the simple system the entire
set of action vectors that can maintain the radial structure of the network. The combinations are:
194
CASE STUDIES AND TESTBEDS FOR THE SMART GRID
Start
Initialize:
• the number of input, hidden, and output nodes of
Action and Critic Neural network
• the original state vector R
• the original weights of Critic and Action Neural
Network by random value
Calculate action vector A via
Action Neural Network based on R
Calculate J via Critic Neural Network based on R
and new action Vector A
Jpre = J
Calculate state vector R from Distribution Power Flow
based on action vector u
Calculate new action vector u via
Action Neural Network based on R
Calculate new J via Critic Neural Network based on R
and new action Vector A
Calculate reward U from Distribution Power Flow
Calculate Critic Neural Network error
errorc = aJ – Jpre + U
Update weights in Critic Neural
Network
Calculate Action Neural Network error
errorA = J – U0
Update weights in Action Neural Network
Calculate new action vector u via
Action Neural Network based on R
No
Convergence
criteria met?
Summarize output results
Stop
Figure 10.8. Flowchart for ADP-based optimal reconfiguration strategy using ADP.
195
ADP FOR OPTIMAL NETWORK RECONFIGURATION
6
× 10–3
Total Power Losses
5
4
3
2
1
0
1
2
3
4
5
6
7
8
9
Action Vector
10 11 12
Figure 10.9. Action vector performance during system training.
A1: {open switches 2, 3; close all other switches}
A2: {open switches 6, 3; close all other switches}
A3: {open switches 2, 5; close all other switches}
A4: {open switches 6, 5; close all other switches}
A5: {open switches 2, 4; close all other switches}
A6: {open switches 3, 4; close all other switches}
A7: {open switches 6, 4; close all other switches}
A8: {open switches 5, 4; close all other switches}
A9: {open switches 2, 7; close all other switches}
A10: {open switches 3, 7; close all other switches}
A11: {open switches 6, 7; close all other switches}
A12: {open switches 5, 7; close all other switches}
Some Simulation Results: The purpose of the algorithm is to find the optimal switches
status combination; for the 5-bus case, the optimal solution is Action Vector 15.
Figure 10.9 shows the minimization of the losses as action vectors for the optimal
switching sequence. After the initialization, the action network generates the first action
vector by random number. The action vector then inputs into the critic vector with state
variables. The output of critic network J and immediate cost U obtains the new error
for the action and critic networks. The weights in the two networks can be updated
based on back propagation rules. After sufficient iterations, the system will output the
result, in our case, the optimal action vector, which is the best switching status combination with the minimum losses.
196
CASE STUDIES AND TESTBEDS FOR THE SMART GRID
We recommend extending this study to a large-scale aerospace power system while
addressing the multiobjective challenges of restoration, reconfiguration, and remedial
control.
10.7
CASE STUDY OF RER INTEGRATION
1. AEP, which has its own existing EHV 765 kilovolt (kV) network, consisting of
19,000 miles of transmission lines, would establish a combined additional
capacity of perhaps 200–400 GW of bulk transmission, thus spurring significantly increased levels of wind energy in the overall energy portfolio. AEP
estimated the cost in the range of $60 billion in 2007.
2. Cape Wind took one step closer to becoming the first offshore wind producer
in the United States when Massachusetts regulators approved a power purchase
agreement with National Grid for fifty percent of the output. The state’s Supreme
Judicial Court is expected to rule on the PPA by December 31, 2011.
3. Southern California Edison is moving toward smaller PV plants as the price of
solar panels continues to drop. It recently signed contracts for 239.5 MW to be
provided by 20 small solar farms.
4. A U.K.-based electricity company is developing an electric sports car that gets
its power from wind turbines, to break the British land speed record of 139 mph
in 2012.
5. With 45 wave and tidal prototypes slated for ocean testing this year and next,
IHS Emerging Energy Research says the global ocean power market looks
poised for takeoff.
6. The U.S. Navy is planning to roll out about 10 ships, planes and submarines
powered by a blend of biofuels and nuclear power.
10.7.1
Description of Smart Grid Activity
Two important ways that smart grid technology creates value in the transmission
network are
•
•
Direct interconnection of large-scale resources
Using remote sensing (and eventually, remote devices for the injection of dynamic
reactive power) to increase network capacity by improving flow management
10.7.2
Approach for Smart Grid Application
Advanced metering infrastructure (AMI). AMI is a catch-all term for smart
meters. The smart grid is often associated with electrical meters, or smart meters. This
association helps to simplify a very complex process, but it sells the vision of the smart
grid short.
TESTBEDS AND BENCHMARK SYSTEMS
197
The smart grid is no more about the meter than the economy is only about the cash
register. Both are merely collection points: interfaces between buyer and seller where
data can be gathered and analyses. As in many business and technology scenarios, you
can’t improve what you can’t measure. AMI systems capture data, typically at the meter,
to provide information to utilities and transparency to consumers. It’s also usually the
case that AMI systems piggyback on a variety of wireless systems. One simple illustration: the days of the electric company representative swinging by to read your meter
could be phased out with AMI.
Demand response (DR). To date, consumers have used energy whenever they
want to, and utilities have built the power plants and delivery infrastructure to support
it, no matter what the cost or environmental impacts. To achieve economic and environmental goals, consumers need to become equal participants in the process, tuning
their energy consumption to when clean resources are available and avoiding peak
energy consumption times as much as they can. If some electricity-consuming devices
can be deferred to nonpeak time, everyone wins. While you may need to turn on your
lights when you arrive home, there’s no reason that you can’t run your dishwasher at
3 AM when rates and demand are lower.
That’s the reasoning behind demand response programs and they have been very
successful. Research bears out again and again that when consumers are asked to “do
the right thing” when it comes to energy usage, they will do so.
Smart thermostats, for example, can prompt consumers to lower their air conditioner by a degree or two. To date, [5] some U.S. citizens are in some kind of demand
response program. In-home displays and similar devices can lower energy use by up
to 6%. Demand response programs yield immediate reductions on the average.
Critical peak pricing (CPP). An off-shoot of demand response, critical peak
pricing simply means that utilities have the technology infrastructure in place to charge
consumers more for energy during peak periods. It allows customers to decide whether
or not to pay more on the specific critical days, rather than paying an average cost. It
helps balance cost and risk between the consumer and the utility, as well as providing
a further incentive for consumers to reduce energy consumption.
Time-of-Use Pricing (TOU). Time-of-use pricing is similar to critical peak
pricing, except extrapolated across every hour for every day. Time-of-use pricing allows
utility rates and charges to be assessed based on when the electricity was used. Not
only the time of day, but also the season, as well as accounting for local weather patterns that might prompt a rate adjustment (at which point it becomes a demand response
application).
10.8
TESTBEDS AND BENCHMARK SYSTEMS
The development of computational tools for power system applications requires extensive testing and validation for efficiency, speed, accuracy, reliability, and robustness.
198
CASE STUDIES AND TESTBEDS FOR THE SMART GRID
We will require data and/or users to test the final product based on the uniqueness of
the test system being studied in normal, alert, emergency, and restoration conditions,
with approved standards for interoperability, cyber security, and managing
uncertainty.
We hope to continue development of a MATLAB-based [4, 5] environment for
generalizing the dynamic stochastic OPF, the variants of ADP, decision analysis tools,
and other tools for solving different testbeds and benchmarks in civilian and military
power networks. The results will be discussed a future book entitled, “Adaptive Essential Optimization Power Techniques (AOPT),” which will provide a development platform for smart grids and other complex system networks.
10.9
CHALLENGES OF SMART TRANSMISSION
Two key issues in developing a super transmission grid involve citing decisions and
providing reasonable policies for equitable cost allocation. In both cases, a congressional response is needed that gives FERC authority to take bold and decisive action
(see DOE announcement and http://www.nappartners.com/news/doe-fercannouncenew-collaborative-effort-ontransmission-projects).
10.10
BENEFITS OF SMART TRANSMISSION
Smart transmission investment provides many benefits to power customers and electricity markets. Although they vary according to the type and location of smart technologies, installation of new digital technologies materials and implementation of the
functionalities identified in this chapter are aimed at achieving:
•
•
•
•
•
•
•
Increased reliability
Increased electricity throughput at lower delivered cost
More efficient fuel use for generation, yielding lower air emissions
Greater use of RER and clean generation resources, with lower operational integration costs.
More effective use of energy storage, reduce the costs of the peak electricity
provision
Facilitate third party participation in the power system
Foster wholesale and retail markets, improving information available to customers and market participants about grid connections, electricity prices, and usage
10.11
SUMMARY
This chapter has discussed how the existing transmission system can be made
even smarter by investing in new technology. Most of these smart grid technology ele-
REFERENCES
199
ments are well-tested, mature, and cost-effective, and their use will make the North
American bulk power system more reliable, secure, efficient, economic, diverse, and
environmentally sustainable. But while communications, computer analytical tools [8],
sensors, and controls are critical elements, such technologies [9] cannot themselves
deliver electricity from a power plant to the end-user. The chapter pointed out the need
for a strong platform of wires, cables, and substations, investment in existing transmission infrastructure, and additional investment in new wires and transformers. The
combination of conventional transmission technologies with advanced smart grid elements that will optimize and enhance the value of transmission investments was
explained.
REFERENCES
[1] J.A. Momoh, M.E. El-Hawary, and R. Adapa. “A Review of Selected Optimal Power Flow
Literature to 1993, Part I: Nonlinear and Quadratic Programming Approaches,” IEEE Transactions on Power Systems 1999, 14, 96–104.
[2] J. Si, A.G. Barto, W.B. Powell, and D. Wunsch. Handbook of Learning and Approximate
Dynamic Programming, Hoboken, NJ: Wiley, 2004.
[3] G.K. Venayagamoorthy, R. G. Harley, and D.C. Wunsch. “Dual Heuristic Programming
Excitation Neuro-Control for Generators in a Multi-machine Power System,” IEEE Transaction on Industry Applications 2003, 39, 382–394.
[4] J.A. Momoh. Electric Power System Application of Optimization, New York: Marcel Dekker,
2001.
[5] N.P. Phady. “Unit Commitment—A Bibliographical Survey,” IEEE Transactions on Power
Systems 2004, 19, 1196–1205.
[6] J.A. Momoh and Y. Zhang. “Unit Commitment Using Adaptive Dynamic Programming,”
Proceedings of the IEEE Intelligent Systems Application to Power Systems (ISAP) Conference, 2005.
[7] J.A. Momoh and A.C. Caven. “Distribution System Reconfiguration Scheme Using Integer
Interior Point Programming Technique,” Transmission and Distribution Conference and
Exposition, 2003, PES1, 7–12.
[8] Smart Grid Interoperability Standards Project, National Institute of Standards and Technology. January 2010.
[9] L. Karisny. “Will Security Start or Stop the Smart Grid?” Digital Communities, November
18, 2010.
11
EPILOGUE
The smart grid initiative, also known as “intelligent grid” or “smart transmission,” is
presently being pursued throughout the United States by various entities. It is a completely new idea with initial concepts and work undertaken by utilities, academics, and
other stakeholders. The development of supporting technologies is being provided
mainly through stimulus funding programs administered by the DOE.
Stakeholders understand that the technological development of new components,
such as smart meters, geolocational monitoring equipment, new DR including sustainable energy resources, and energy storage mediums must be complemented by developing the requisite communication infrastructure, retrofitting/updating current grid
components, and establishing standards for interoperability and integration. Testbed and
experimentation environments are presently being undertaken. Stakeholders in the
smart grid have also asserted the importance of training the existing workforce and
educating future personnel.
As with any cutting-edge technology, many challenges must be met. The list
includes:
1. The integration of multiagent system frameworks and adaptive system thinking
to improve system operability for congestion management.
Smart Grid: Fundamentals of Design and Analysis, First Edition. James Momoh.
© 2012 Institute of Electrical and Electronics Engineers. Published 2012 by John Wiley & Sons, Inc.
200
EPILOGUE
201
2. The UC problem with stochasticity for network uncertainty challenges using
a DSOPF technology. This technique will ultimately provide the robust cooptimization need for planning and operation or resources.
3. Microgrid development with conversion and protection schemes that will facilitate efficiency. This may involve the use of micro electro-mechanical systems
(MEMS) and smart protection schemes.
4. Interdisciplinary approaches to development of large-scale smart buildings and
components involving electrical engineering, mechanical engineering, material
sciences and information systems.
5. Future aerospace automation technology advancements that will be integrated
for improvement of quality of technology and design.
6. Advancement of materials and components such as nanotechnology to be utilized for storage technologies.
7. Pricing studies based on dynamics and stochastic models with both technical
and human behavior and serious study to justify the marginal cost benefit associated with investment.
8. Standardization of tools, software, and hardware for design and deployment to
ensure smart grid acceptability by the public and stakeholders.
9. Training/retraining and education of present and future workforces through the
facilitation of cross-discipline academic research supported by utilities.
Overall examples are needed to represent the designs and prototypes of performance metrics that accurately define the new smart grid technology with the attributes
discussed in this book as well as those referenced herein through:
1. Increased use of digital information and controls technology to improve reliability, security, and efficiency of the electric grid.
2. Dynamic optimization of grid operations and resources with full cyber
security.
3. Deployment and integration of distributed resources and generation, including
RER.
4. Development and incorporation of DR, demand-side resources, and energy
efficiency resources.
5. Deployment of smart technologies (real-time, automated, interactive technologies that optimize the physical operation of appliances and consumer devices)
for metering, communications concerning grid operations and status, and
distribution automation.
6. Integration of smart appliances and consumer devices.
7. Deployment and integration of advanced electricity storage and peak-shaving
technologies, including PHEVs and thermal-storage air conditioning.
8. Provision to consumers of timely information and control options.
202
EPILOGUE
9. Development of standards for communication and interoperability of appliances and equipment connected to the grid, including the infrastructure serving
the grid.
10. Identification and lowering of unreasonable or unnecessary barriers to adoption of smart grid technologies, practices, and services.
Once the above is achieved we will be able to boast of a truly intelligent grid with
system applications in space, naval ship systems, and smart technology for public
housing and energy consuming facilities.
INDEX
Note: Page numbers in italics figures; tables are noted with t.
AA. See Arbitrator Agents
AC automatic contingency control, 49
AC automatic contingency screening/filtering,
49
ACDs. See Adaptive critic designs
AC filters, 72
AC load flow, 37
ACO. See Ant colony optimization
Action dependent heuristic dynamic
programming, 117
Action networks, 116
Action vectors
adaptive dynamic programming and,
115
ADP implementation structure for, 193
performance of, during system training,
195
Adaptability, smart grid and, 52
Adaptive controls, applications for, in smart
grid design, 137–138
Adaptive critic designs, 116
Adaptive dynamic programming, 41, 118t
framework of applications to power
systems, 42
generation schedule for UC problem
solved with, 191
implementation structure, 192–193,
195–196
action vectors, 193, 195
rewards (Utility function), 193
some simulation results, 195
for optimal network reconfiguration in
distribution automation, 191–196
structure network reconfiguration, 193
system, structure of, 115
techniques, 115–117
common nomenclature, 115
variant for UC problem, 187
ADHDP. See Action dependent heuristic
dynamic programming
ADP. See Adaptive dynamic programming
ADP-based optimal reconfiguration strategy,
using ADP, flowchart for, 194
Advanced batteries, characteristics,
advantages, and disadvantages, 156t
Advanced metering, 185
Advanced Metering Infrastructure, 9
customer applications, 23
cyber security concerns associated with,
171, 172t
distribution operations, 23
market applications, 22–23
meters, 52
RER integration case study, 196
security domain descriptions, 171, 172t
System Security Requirements, 167t
Smart Grid: Fundamentals of Design and Analysis, First Edition. James Momoh.
© 2012 Institute of Electrical and Electronics Engineers. Published 2012 by John Wiley & Sons, Inc.
203
204
AEP, 196
AGC. See Automatic generation control
Agriculture, climate change and, 153
AHP. See Analytical hierarchical processing
Alkaline fuel cells, 147
American Electric Power, 9, 10
American National Standard for Utility
Industry End Device Data Tables,
167t
American National Standards Institute, 163,
165t
AMI. See Advanced Metering Infrastructure
Amorphous silicon, in manufacture of PV
cells, 141
AMR. See Automated meter reading
Analytical hierarchical processing, 101, 102,
102–103
Angle stability, overview of methods for, 74
Angle stability assessment, 73–81
algorithms
for closest unstable equilibrium point
(UEP) approach, 77–78
for controlling UEP approach, 78–79
for finding controlling UEP, 80
boundary of region of stability, 77–79
stability application to a practical power
system, 76–77
transient stability, 75–76
ANNs. See Artificial neural networks
ANSI. See American National Standards
Institute
Ant colony optimization, 113–115, 118t
Appliance control/monitoring, smart meters
and, 22
Appliance level, of smart grid, 137
Appliances, smart, 22
Arbitrator Agents, 24
Arcadian Networks, 185
Areva, 9
Artificial intelligence, 112, 119
Artificial Intelligence Method, 75
Artificial neural networks, 109–111, 118t
ATC
computed, formula for, 36
defined, 38
Automated meter reading, 52
Automatic generation control, 46, 112
Automatic Tax Credit for Low Speed & 2- or
3-Wheeled Vehicles, 159
INDEX
Automobile Tax Credit for Hybrid GasElectric and Alternative Fuel Vehicles,
158
Automobile Tax Credit for Plug-In Electric
Vehicles, 158
Automobile Tax Credit for Plug-In Hybrid
Conversion Kits, 158
Back propagation, artificial neural networks
and, 109
Backscatter radios technology, 21
Bad data, detection and identification of, 86
Bandwidth, high, smart grid communications
technology, 17
Base case load flow solution, 48
Battelle, 9
Batteries, advanced, characteristics,
advantages, and disadvantages, 156t
Becquerel, Edmund, 141
Bellman principles, techniques derived from,
107
Beta density function
conversion and power electronic
technology and, 144
solar insolation modeling and, 144
Bifurcation theory, 65
Billing, smart meters and, 22
Biodiesels, 147
Biomass-bioenergy, 145, 147
Biomass plants, commercial availability of, in
U.S., 147
Biopower, derivation of, 145, 147
Blackouts, 20, 73
Bonneville Power Administration, 9
Botanic ceramic fuel cells, 147
Bottlenecks, phasor measurement and, 20
BPL. See Broadband over Power Lines
Branch participations, 68
Broadband over Power Lines, 17
Broadcast transmission, LAN and, 18
Bulk power systems
automation of smart grid at transmission
level, 130–131
coordinating with new and emerging
technologies, 9
Bush, George W., 2
Bus impedance network method, 35–36
Bus topology, 18
Bus voltages, 81
INDEX
Cadmium telluride (CdTe)
in manufacture of PV cells, 141
as thin semiconductor film, solar power,
142
California Energy Commission, 8
Capacity Benefit Margin, in ATC
computation, 36
Cape Wind, 196
Case studies/testbeds for smart grid, 184–199
ADP for optimal network reconfiguration
in distribution automation, 191–196
advanced metering, 185
benefits of smart transmission, 198
challenges of smart transmission, 198
demonstration projects, 184–185
microgrid with renewable energy, 185
power system unit commitment problem,
186–191
RER integration, 196–197
testbeds and benchmark systems, 197–198
Cast polycrystalline silicon panels, 142
CBM. See Capacity Benefit Margin
Chance-constrained programming (CCP),
107
Chemical and materials residues, bioenergy
derived from, 145
CI. See Computational intelligence
CIS. See Copper indium diselenide
Cisco, 11, 185
Classical optimization method, 103
Climate change
causes of, 153
implications of, 153
Coaxial cable, 17
Commercial interoperability, 161
Common Information Model (CIM) for
Power Systems, 167t
Commonwealth Edison, 185
Compressed air, characteristics, advantages,
and disadvantages, 157t
Computational intelligence, 4, 4, 119
Computational tools
adaptive dynamic programming techniques,
115–117
computational challenges, 118–119
decision support tools, 101–102
evolutionary computational techniques,
112–115
heuristic optimization, 108–112
205
hybridizing optimization techniques and
applications to smart grid, 118
introduction to, 100–101
optimization techniques, 103
Pareto methods, 117–118
for smart grid design, 100–119
Conductor sensors, 21
ConEd, 9
Congestion
defined, 37
phasor measurement and, 20
Congestion management effect, 37–38
Consumers Energy, 10
Contingencies
classification of, 44–46
Static Security Assessment and, 43–44
Contingency studies, for smart grid, 48–49
Continuation power flow method, 127
Control agent, multiagent specifications of,
25
Control architecture, power systems and, 25
Copper indium diselenide, 142
Copper indium selenide, in manufacture of
PV cells, 141
Corn, bioenergy derived from, 145
Corn ethanol, 147
Corrective controls, cyber threats and, 169
Critical peak pricing, RER integration case
study, 197
Critic networks, 116
Curriculum, smart grid education, 179–182
Cyber security
AMI and, 171, 172t
built-in vs. added-on, 168
contingency and recovery plans, 168
defined, 166
facets and objectives of, 168
possible operation for improving
methodology for other users, 173
risks, 169–171
considering, trade-off, 170
mitigation approach to, 171, 173
system constraints, 170–171
smart grid, 166–173
smart grid and mandating development of,
8
smart grid communications and, 17
standards for various electric grid levels,
167t
206
Cyber security (cont’d)
state of the art, 166, 168–169
tools and technologies, training in, 182
Cyber threats, monitoring, 168–169
DA. See Decision analysis; Diagnosis Agents
Database agent, multiagent specifications of,
26
Day-ahead economic scheduling, 151
Day-ahead hourly pricing, 150
Day-of economic dispatch, 151
DC-AC conversion, PV Inverter system for,
144
Decision analysis, stages in, 101–102
Decision support tools, uses for, 101
Deforestation, climate change and, 153
Demand bidding buy-back, 150
Demand response
renewable energy, categorizing, 150–151
RER integration case study, 197
technology tree, 150
Demand-response rates, smart meters and,
22
Demand-side management, 6, 14, 136–137
categories of options for, 136
classification of, 136–137
Demonstration projects, development of
WAMS and, 184–185
Department of Energy, 7, 8, 200
Modern Grid Initiative, 25
Office of Electricity and Energy Assurance,
9
representative architecture of smart grid
design, 12
Department of Homeland Security, 182
DERs. See Distributed Energy Resources
Detective controls, cyber threat detection and,
168–169
Development of Stochastic Dynamic Optimal
Power Flow
application to smart grid, 41–43
module 1, 42
module 2, 42
module 3, 42–43
cases for, 41
efficient optimization technique and
objectives with, 138
features of, that aid development of new
OPF, 73
INDEX
Device level, smart grid development, 123,
124–125
Device malfunctions, state estimation and, 95
DFRs. See Disturbance Fault Recorders
DHP. See Dual heuristic programming
Diagnosis Agents, 24
Digital and Distributed Power Systems
Training Program, 182
Digital signal processing techniques, 119
Direct Control of Load, 136
Direct load control, 150
Discount rate, adaptive dynamic
programming and, 115
Distributed energy resource agent, multiagent
specifications of, 25–26
Distributed Energy Resources, 140
Distribution automation, ADP for optimal
network reconfiguration in, 191–196
Distribution generation control, 137
Distribution Load Control, 137
Distribution load flow methods, 33–37
bus impedance network: method 3, 35–36
forward/backward sweep: method 1, 33–34
load flow based on sensitivity matrix for
mismatch calculation: method 2,
34–35
Distribution systems, distribution automation
for, 133
Disturbance Fault Recorders, 20
DOE. See Department of Energy
Dominion Virginia Power, 185
DP. See Dynamic programming
DSA
for smart grid, 82
designing, 80–81
DSE. See Dynamic state estimation
DSM. See Demand-side management
DSOPF. See Development of Stochastic
Dynamic Optimal Power Flow
Dual heuristic programming, 116
Duke Energy, 185
DyLiacco’s security-state diagram, 43
Dynamic analysis, 130
Dynamic line rating technology, 21
Dynamic programming, 103, 107
Dynamic security, 130
Dynamic stability (Type II instability), 59–60
fast instability, 60
slow instability, 60
207
INDEX
Dynamic state estimation, 97
Dynamic voltage stability studies, analysis
techniques for, 57, 60–62
EAL. See Energy Auction level
Economic dispatch, 127–128
Education
smart grid, 179–182
architecture: module 2, 180–181
case studies and testbeds: module 9, 182
communication technologies: module 7,
182
functions: module 3, 181
introduction: module 1, 180
pathways to design: module 5, 181
renewable energy technologies: module
6, 181–182
standards, interoperability, and cyber
security: module 8, 182
tools and techniques: module 4, 181
EDVs, battery, power supply or demand and,
151
Efficiency
improving, smart grid and, 6
transmission, 14
Eigenvalue analysis methods, system
dynamics and, 57
880.2, 18
EISA. See Energy Independence and Security
Act
Electricity, smart meters for, 22
Electric parameters, 20
Electric Power Research Institute, 10
Electric power system
future demand and trends in, 23–24
threats facing, 169t
Electric vehicles, 151. See also Plug-in
hybrid electric vehicles
proposed power line and wireless control
connections between grid and, 152
Electronic instrument transformers, 21
EM. See Energy margin
EMS. See Energy management systems
Encryption, smart grid communications and,
17
End-user level, of smart grid, 137
Energy, climate change and, 153
Energy and power exchange model, UC
optimization problem and, 186
Energy Auction level, smart grid
development, 122, 123, 124
Energy Independence and Security Act of
2007, 8
rationale for smart grid, 2–4, 3
Energy management systems, 19, 81, 163
Energy margin, 73
Energy margin method, 127
Energy method, 66
Energy storage, 154–158
EP. See Evolution programming
EPRI. See Electric Power Research
Institute
ES. See Expert systems
Ethanol, 147
Ethernet/IEEE 802.3, 18
EUE. See Expected unserved energy
Euler method, modified, 75
Evolutionary computational techniques,
112–115
ant colony optimization, 113–115
genetic algorithm, 112, 113
particle swarm optimization, 113
Evolution programming, 118t
Expected unserved energy, 127
Expert systems, 111–112, 119
advantages over human experts, 111
fundamental components of, 111
FACTS. See Flexible AC transmission
systems
Fast decouple method, 33
Fast instability, 60
Fathoming, in branch-and-bound procedure,
106
Fault and stability diagnosis, 131–132
Fault Detection Agents, 24
Fault-on trajectories, finding controlling UEP
for, 79
FDA. See Fault Detection Agents
Federal Energy Regulatory Commission
(FERC), 8
Fiber optics cable, 17
Flexible AC transmission systems, 71
Flicker factor, calculations, parameters, and
network/systems where index is
applied, 135t
Flow batteries, characteristics, advantages,
and disadvantages, 156t
208
Flywheels, characteristics, advantages, and
disadvantages, 158t
Fold type curve, including predictor-corrector
step, 71
Food costs, climate change and rise in, 153
Forced outages, 46
Forward/backward sweep method, 33–34
Forward propagation, artificial neural
networks and, 109
Fossil fuel burning, climate change and, 153
Fuel cells, 140, 147–148
Fuel to electricity efficiency, 147
Fuzzy-GA systems, 118
Fuzzy logic, 112, 112
Fuzzy-PSO systems, 118
Fuzzy set, 112
Fuzzy system, 112
GA. See Genetic algorithm
Gas consumption, smart meters for, 22
Gasification, 147
Gaussian functions, solar insolation modeling
and, 144
Gauss-Seidal method, of studying load flow,
31–32
GDHP. See Globalized HDP
GE, 9
Generator outages, 46
Genetic algorithm, 112, 118t, 128
applications for, 112
typical cycle, 113
Geolocational monitoring equipment, 200
Geothermal heat pumps, 148
GHGs. See Greenhouse gases
GIS, Google mapping tools and, 23–24
Global Best (gbest) variant, of PSO, 113
Global change, defined, 153
Globalized HDP, 116
Global warming, defined, 153
Google Earth software, 23
Google mapping tools, GIS and, 23–24
GPS satellites, wide area monitoring systems
and, 20
Greenhouse gases, climate change and, 153
GridComm project, 185
Grid-friendly appliances, 22
Grid methodology, old and new, 54t
Grid monitoring, enhanced, smart meters and,
22
INDEX
GridPoint, 169
GridWise Architecture Council, 9, 162
GridWise Program, 9
GWAC. See GridWise Architecture Council
Hamilton-Jacobi-Bellman equation, of
optimal control, 116
Hardware, computational tools for smart grid
and, 118
HDP. See Heuristic dynamic programming
Health, climate change and, 153
Heuristic dynamic programming, 187
in ACDs, 116
implementation of, schematic for, 188
neural network in, structure of, 190
Heuristic optimization, 108–112
artificial neural networks, 109–111
expert systems, 111–112
HIDS. See Host-based intrusion detection
systems
Home Access Network, 18
Home Energy Efficiency Improvement Tax
Credit, 158
Host-based intrusion detection systems,
168
Hybridization optimization techniques,
applications to smart grid, 118
Hybridized computational intelligence tools,
for smart grid analysis and design,
118t
Hybrid vehicles, tax credits and, 158
Hydropower, small and micro, 147
IBM, 9, 10, 185
IDAPS microgrid, 26
IEC. See International Electrotechnical
Commission
IEEE. See Institute of Electrical and
Electronic Engineers
IETF. See Internet Engineering Task Force
IHS Emerging Energy Research, 196
Immunized-neuro systems, 118
Immunized-swarm systems, 118
Incentive-based demand response, 150
Indirect methods (continuation methods),
parameterization, 69–70
Injection Shift Factor, 38
Institute of Electrical and Electronic
Engineers, 163, 164t, 166, 174
209
INDEX
Standard for Interconnecting Distributed
Resources with the Electric Power
System, 167t
Standards for Synchophasors for Power
Systems, 167t
Insulation contamination leakage current, 21
Integer programming, 106–107
branch-and-bound procedure, 106
Integration, computational tools for smart
grid and, 119
Intelligent grid, 200, 202
automation, 131
distribution subsystem component, 14
Intelligent Power Grid
components of, 11
key characteristics of, 10
Intelligent system technologies, 119
Intelligrid software, 10
Interest rate, adaptive dynamic programming
and, 115
Interior point methods, 128
International Electrotechnical Commission,
163, 164t
Internet Engineering Task Force, 163, 164t
Interoperability, 161–163
benefits and challenges with, 161–162
control of power grid and, 163
defined, 161
model for, in smart grid environment, 162
smart grid network, 162
state-of-the-art, 161
Intersystem communication, smart grid and
mandating development of, 9
IP-enabled digital communication, 17
IPM. See Interior point methods
ISF. See Injection Shift Factor
ISO/RTO, 19
Iterated Linear Step ATC, 37
I-V characteristic model of single cell, 142–143
Jacobian condition number method, 127
Jacobian matrix
condition number of, 64–65
eigenvalues of, 66
fast decouple method and, 33
minimum singular value of, 63–64
Newton-Raphson method and, 32
voltage stability indexing and, 65
Judgment Index Agent (JIA), 24
Karush-Kuhn-Tucker, 105
Keyhole Markup Language, 23
Kinetic energy, 77
Kirchhoff’s current law nodal equations,
Gauss-Seidal method and, 31–32
KKT. See Karush-Kuhn-Tucker
KML. See Keyhole Markup Language
Kuhn-Tucker (KT), 105
Lagrangian relaxation method, 103
solving UC problems and, 187
for unit commitment problem, 130
LAN. See Local area network
Land use change, global change and,
153
L index, 65, 66
Linearization, 37
Linear programming, 103
Lithium-ion batteries, 156t
Load base VSCP, 54
Load curve, of 3-generator, 6-node system,
191
Load flow
challenges to, in smart grid and
weaknesses of present methods,
30–31
proposed methodology, 39
for smart grid design, 38, 40–41
Load flow state of the art
distribution load flow methods, 33–37
Fast Decouple method, 33
Gauss-Seidal method, 31–32
Newton-Raphson method, 31, 32–33
Load flow studies, function of, 29
Load flow techniques
comparison of, 30t
traditional, characteristics of, 31
Load stability, 57. See also Voltage
stability
Local area network
attributes and advantages with, 18
data transmission categories, 18
description of, 17
topologies, 18
Local Best (lbest) variant, of PSO, 113
Local Load Control Option, 137
LOLP. See Loss of load probability
Long-term voltage collapse, 59
Loss of load probability, 127
210
Lyapunov function, 54
Lyapunov stability assessment, 127
Machine learning, 119
Man-in-the-middle (MitM) attacks, wireless
networks and, 173
MCDA. See Multi-criteria decision analysis
Metering, 19
Michigan Gas and Electric, 10
Microcrystalline silicon, in manufacture of
PV cells, 141
Microgrid
with renewable energy, 185
sample testbed environment, 186
smart grid vs., 27
topology, storage technologies, 154
Micro hydropower, 147
Minimum distribution line loss, formulation
for, 191–192
Minimum singular value, of Jacobian matrix,
63–64
Minutes dispatch, 151
MIP. See Mixed-integer programming
MIP problem, general structure of, 106
Mismatch calculation
load flow based on sensitivity matrix for,
34–35
single feeder representation, 34, 35
Mismatch power, computing, 32
Mixed-integer programming, 103
Modal (or Eigenvalue) analysis method, 66
Modern Grid Initiative (DOE), 25
Modified Euler method, 75
Monitoring and control technology
component, 14
Monocrystalline solar panels, 142
MPLS. See Multiprotocol Label Switching
Multiagent architecture, simplified view of,
24
Multiagent systems
characteristics of, 26
performance of, 24
Multiagent systems (MAS) technology,
24–27
multiagent specifications, 25–26
multiagent systems for smart grid
implementation, 25
multiagent technique, 26
Multicast transmission, LAN and, 18
INDEX
Multi-criteria decision analysis, 101
Multiobjective optimization problem, solving,
Pareto analysis for, 117
Multiprotocol Label Switching, 17
National Institute of Standards and
Technology, 163, 165t, 174
National Renewable Energy Laboratory, 8
n-1 contingency analysis, 130
Neighborhood Area Network, characteristics
of, 19
NERC. See North American Electric
Reliability Corporation
Net Transmission Capacity, 37
Network-based intrusion detection systems,
168
Network distribution reconfiguration
problem, small power system for,
192
Network equivalents, utilization of, steps in,
48
Network errors, state estimation and, 95
Network Interface Card, 17
Network reconfiguration, 136
Network switching and restoration, 132
Neural networks, 128
Neuro-fuzzy systems, 118
Neuro-genetic systems, 118
Neuro-swarm systems, 118
Newton method, 128
Newton-Raphson iterations, basic load flow
calculation method and, 37
Newton-Raphson method, 32–33, 34
for power flow analysis, 69
for studying load flow, 31, 32–33
system partitioned matrix equations of,
rewriting, 66–67
New York State Energy Research and
Development Authority, 8
Nexgen, 9
NIC. See Network Interface Card
Nickel metal hybrid batteries, 156t
NIDS. See Network-based intrusion detection
systems
NIST. See National Institute of Standards and
Technology
Nonlinear programming (NLP), 105–106
challenges related to, 106
structure of problems solved by, 105
INDEX
North American Electric Reliability
Corporation, 29, 163, 165t
NREL. See National Renewable Energy
Laboratory
NTC. See Net Transmission Capacity
Ocean circulation, climate change and, 153
Ocean wave energy, 196
OLTCs. See On load tap changers of
transformers
One-at-a-time strategy, 117
On load tap changers of transformers, 71
Optimality test, 106
Optimality Test and Termination Criteria, 105
Optimization, applications for, in smart grid
design, 137–138
Optimization techniques, 103
Outage detection, sensing, 19
Ozone depletion, global change and, 153
Pacific Northwest National Laboratory, 9
Pacific Northwest Smart Grid Demonstration
Project, 185
PAFCs. See Phosphoric acid fuel cells
Parabolic troughs, 142
Parameterization, 69–70
Pareto methods, 117–118
Particle swamp optimization, 118t
Particle swarm optimization, 113, 128
Peak clipping, demand-side management and,
136
Peak shifting, demand-side management and,
136
Pepco, 185
PEPCO Holdings, 9
Performance indices, external system
equivalents, 47–48
Phase shifters, 37
Phasor Measurement Units, 20–21, 52
enhancing current grid, 55
scheme for state estimation with, 96
stability monitoring and analysis, 132
PHEVs. See Plug-in hybrid electric vehicles
Phosphoric acid fuel cells, 147
Photovoltaic cells, harnessing solar energy
with, 141
Photovoltaic panels, modeling output of,
alternative equation for, 143
Photovoltaic systems, modeling, 142–144
211
PJM Interconnection, 9
Plug-in hybrid electric vehicles, 140, 151,
201
advantages and disadvantages with, 151
description of technology, 151
impact of, on grid, 151–152, 152
standards related to, 164t
PMU. See Phasor Measurement Units
PNNL. See Pacific Northwest National
Laboratory
Policy-makers, 7
Polycrystalline silicon, in manufacture of PV
cells, 141
Polycrystalline solar panels, 142
Polymerion batteries, 156t
Power flow, Jacobian matrix of, decomposed,
63–64
Power grid, interoperability and control of,
163
Power quality, 20
defined, 135
indices, typical, 135t
measures, indices utilized as, 135
Power system enhancement, smart grid and,
4, 5
Power Systems Engineering Research Center,
10
Power system stabilizer, 60
Power system unit commitment problem,
186–191
Power Transfer Distribution Factors, role and
calculation of, 37
Predictivity, smart grid and, 52
Prepaid metering, 22
Price-based demand response, 150
Pricing signals, 19, 20
Proton exchange membrane (PEM) fuel cells,
147
PSERC. See Power Systems Engineering
Research Center
PSO. See Particle swarm organization
PSS. See Power system stabilizer
PTDFs. See Power Transfer Distribution
Factors
Pumped hydro, characteristics, advantages,
and disadvantages, 157t
PV-DesignPro, 142
PV inverter system, for DC-AC conversion,
144
212
Q-learning, 117
Raw data, classification of, 87–88
Rayleigh density function, conversion and
power electronic technology and, 144
Rayleigh distribution function, wind
technology and, 149
RDS, 9
Reactive power control, 132
Reactive Power Controllers, 71
Real-time consumption data, 20
Real-time network modeling, state estimation
and, 95
Reforestation, climate change and, 153
Regenerative fuel cells, 147
Regional transmission organization, 19, 20
Regulatory frameworks, smart grid
development and, 177
Reliability
improving, smart grid and, 6
power system, defined, 125
transmission, 14
Remote Terminal Unit, 16
Renewable energy resources, 140
demand response issues, 150–151
design of architecture for smart grid with,
52
energy storage and increased development
of, 155
microgrid with, 185
tax credits and, 158–159
Renewable energy systems, smart grid and
integration of, 6
RER. See Renewable energy resources
RER integration case study, 196–197
approach for smart grid application,
196–197
advanced metering infrastructure, 196
critical peak pricing, 197
demand response, 197
time-of-use pricing, 197
description of smart grid activity, 196
RES. See Renewable energy systems
Research activities
multidisciplinary, 178–179
in smart grid, 178
Research areas, for smart grid development,
176–177
Residential Renewable Energy Tax Credit, 158
INDEX
Ribbon silicon, in solar panels, 142
Ring bus topology, 18
Risk assessment, 127
RK4, 75
Robustness, smart grid and, 52, 119
Robust state estimation, 90–91, 93–94
with pre-estimation analysis, flow chart for,
94
RockPort Capital Partners, 9
RPCs. See Reactive Power Controllers
RTO. See Regional transmission organization
RTU. See Remote Terminal Unit
SA. See Scheduling Agent
SAIC, 9
SAIDI, 23
SAIFI, 23
San Diego Gas & Electric, 185
SCADA, 19, 25, 96, 97, 163, 166
Scalability, smart grid and, 52
SCE. See South California Edison
Scheduling Agent, 24
Schneider Electric, 9
SE. See State estimation
Security analysis, power, 130
Security-type performance index, 47
Sensing, 19
Sensitivity, computational tools for smart grid
and, 119
Sensitivity-based approaches, to outages, 48
Sensor inputs, adaptive dynamic
programming and, 115
Sequential Quadratic Programming, 106
Simulation and analysis tools, smart grid
development and, 177
Simultaneous strategy, 117–118
Single Linear Step ATC, 37
Single-line outages, 46
Singular value decomposition method, 127
Slow instability, 60
SMA. See System Monitoring Agents
Small hydropower, 147
Smart appliances, 22
Smart distribution solutions, 132
Smart grid. See also Case studies/testbeds for
smart grid; Cyber security; Stability
analysis tools for smart grid
advanced optimization and control
techniques used with, 126
INDEX
capabilities of, 1
characteristics of, 8, 17
communication and standards for, 4, 5
contingency studies for, 48–49
demand side management component, 14
deployment of, 177t
DSA for, 82
education about, 179–192
end user/appliance level of, 137
Energy Independence and Security Act of
2007: rationale for, 2–4, 3
environmental implications for
development of, 152–154
climate change, 153–154
environment and testbed, 4, 5
hybridization optimization techniques
applied to, 118
intelligent grid distribution subsystem
component, 14
key characteristics of, 25
microgrid vs., 27
monitoring and control technology
component, 14
network interoperability and, 162
penetration of PV into, requirements for,
144
power system enhancement and, 4, 5
process changes in design of DSA for,
80–81
representative architecture, 12, 12
research activities in, 178
smart devices interface component, 13
stakeholder roles and function, 6–9
state estimation and approach with, 95–97
storage component, 13
today’s grid vs., 2t
training and professional development
related to, 182–183
transmission subsystem component, 14
utilities and, 9
working definition of, 11–12
Smart grid components, functions of, 12
Smart grid design
DOE representative architecture of, 12
load flow for, 38, 40–41
second architectural framework, 13
Smart grid development
key aspects of, 4
research areas for, 176–177
213
Smart grid environment, state estimation for,
94–95
Smart grid initiative
challenges related to, 200–201
design and prototypes of performance
metrics, 201–202
Smart grid market drivers, general view of, 6
Smart grid pathway design, 122–138
applications for adaptive control and
optimization, 137–138
barriers and solutions to smart grid
development, 122, 124–125
bulk power systems automation of smart
grid at transmission level, 130–132
distribution system automation requirement
of power grid, 132–137
demand-side management, 136–137
distribution generation control, 137
network reconfiguration, 136
power quality, 135–136
voltage/VAr control, 132–135
end user/appliance level of smart grid, 137
general level automation, 125, 127–130
economic dispatch, 127–128
reliability, 125, 127
security analysis, 130
stability, 127
unit commitment, 128–130
solution pathways for designing smart grid
using advanced optimization and
control techniques, 125
Smart Grid Task Force, 3
Smart meters, 52, 200
functions of, 21–22
Smart transmission, 200
benefits of, 198
challenges of, 198
Smoothing, wind technology and, 149
Sodium sulfur type batteries, 156t
Software, computational tools for smart grid
and, 118
Solar energy, 140, 141
Solar insolation, probabilistic modeling of, for
variability studies of PV systems, 144
Solar power technology, 142
Solar power towers, 142
Solid oxide (SOFC) fuel cells, 147
Solid polymer molten carbonate (MCFC) fuel
cells, 147
214
Southern California Edison, 9, 185, 196
Soybeans, bioenergy derived from, 145
SP. See Stochastic programming
Spinning reserve requirements, UC
optimization problem and, 187
SQP. See Sequential Quadratic Programming
Stability, power system, 127
Stability analysis tools for smart grid, 51–98
analysis techniques for steady-state voltage
stability studies, 68–70
angle stability assessment, 73–81
application/implementation plan of voltage
stability, 70–71
dynamic state estimation, 97
optimizing stability constraint through
preventive control of voltage stability,
71–73
state estimation, 81–97
strengths/weaknesses of existing voltage
stability analysis tools, 51–56
voltage stability assessment, 56–62
techniques for, 62–65
voltage stability indexing, 65–68
Stakeholders, smart grid and roles/function
of, 6–9, 8
Standards, 163–166
cyber security, for various electric grid
levels, 167t
defined, 163
smart grid interoperability, approach to,
163, 166
summary of, for smart grid development
by key standards bodies, 164t–165t
Star topology, 18
State estimation, 52, 81, 83–97
detection and identification of bad data, 86
dynamic, 97
mathematical formulations for weighted
least square estimation, 84–85
overall flow chart for online, state, and
parameter estimation, and bad data
analysis, 92
PMU-based, steps for, 95
postestimation analysis, 88
flow chart for, 91
pre-estimation analysis, 86–88
classification of raw data, 87–88
detection of obviously bad
measurements, 86–87
INDEX
detection of obviously bad network
topology, 87
flow chart for, 89
flow chart for robust state estimation
with, 94
hypothesis testing for identification of
single bad data point, 90
real-time network modeling, 95
robust, 90–91, 93–94
flow chart for, with pre-estimation
analysis, 94
scheme for, with PMU, 96
smart grid and approach toward,
95–97
for smart grid environment, 94–95
static, 84
State estimator, outputs, statistical properties
of, 85
Static Security Assessment, contingencies
and, 43–44
Static stability (Type I instability), 59
Static state estimation, 84
Static Var Compensators, 71, 72
Stead-state security, 44
Steady-state contingency analysis, 44–45,
46–47
Steady-state voltage stability studies
analysis techniques for, 68–70
direct methods for detecting voltage
collapse points, 69
indirect methods (continuation methods),
69–70
Stochasticity, smart grid and, 52
Stochastic programming, formulation for,
107–108
Storage conservation, demand-side
management and, 136
Storage technologies, 154–158
microgrid topology with, 154t
options, comparison of, 156–158t
parameters for, 155
String silicon, in solar panels, 142
Super capacitors, characteristics, advantages,
and disadvantages, 156t
Super conducting magnetic energy storage,
characteristics, advantages, and
disadvantages, 157t
Sustainable energy options for smart grid,
141–148
INDEX
biomass-bioenergy, 145, 147
fuel cell, 147–148
geothermal heat pumps, 148
goals for, 141
modeling PV systems, 142–144
small and micro hydropower, 147
solar energy, 141
solar power technology, 142
wind turbine systems, 144–145, 146t
Sustainable energy technology, penetration
and variability issues associated with,
148–150
SVCs. See Static Var Compensators
Switching, smart meters and, 22
Synchophasors, 20, 21
System Average Interruption Duration Index.
See SAIDI
System Average Interruption Frequency
Index. See SAIFI
System energy balance model, UC
optimization problem and, 186
System Monitoring Agents, 24
System Operations level, smart grid
development, 123, 124
System Planning and Maintenance level,
smart grid development, 122, 123, 124
System security
associated functions of, 45
defined, 43
Tabu search, 128
Tax credits, to consumers, RER and energy
efficiency, 158–159
TCR. See Thyristor controlled reactor
Technical interoperability, 161
TEF. See Transient energy function
TEF method, calculation of boundary of
region of stability and, 77
Testbeds. See also Case studies/testbeds for
smart grid
smart grid development and, 177
Thermal-storage air conditioning, 201
Thermostat control, smart meters and, 22
Theta method, 75
Thin film panels, 142
3-generator, 6-node system, load curve of, 191
Thyristor controlled reactor, 72
Tidal power, 196
Time-of-day tariffs, 19, 20
215
Time-of-use, 150
Time-of-use pricing, RER integration case
study, 197
Token Ring/IEEE802.5, 18
Total harmonic distortion, calculations,
parameters, and network/systems
where index is applied, 135t
Total Transfer Capability, in ATC
computation, 36
TOU. See Time-of-use
Training and professional development,
182–183
Transfer limit, of electrical power network,
68
Transient energy function, 76
terms for, 77
Transient stability, 75–76
analysis of
flowchart of TEF for, 78
nonlinear system model for, 54
Transient voltage collapse, 59
Transmission Reliability Margin, in ATC
computation, 36
Transmission subsystem component, 14
Trapezoidal method, 75
Tree topology, 18
TRM. See Transmission Reliability Margin
TTC. See Total Transfer Capability
Twisted pair cable, 17
Two-way communications, 20
smart meters and, 22
UC. See Unit commitment
UK-based Electricity, 196
Under voltage load shedding, 132
Unicast transmission, LAN and, 18
Unit commitment, 128–130
formulation of, 129
slow application of computer-based
programs by electric utilities, 128–129
Unit commitment (UC) optimization problem
constraint models for, 186–187
objective function of, 186
Unit generation limits, UC optimization
problem and, 187
Unstable equilibrium point (UEP) approach
closest, algorithm for, 77
controlling, algorithm for, 78–79, 80
Urbanization, climate change and, 153
216
U.S. Climate Change technology program,
DOE strategic plan for, 153
User agent, multiagent specifications of, 26
Utilities, smart grid and, 9
UVLS. See Under voltage load shedding
Valley filling, demand-side management and,
136
VCPI, 65, 66
Vehicle-to-grid power (V2G), 151
Very fast transient voltage collapse, 59
VIPI, 62, 62, 66
computing, 63
defining, equation for, 63
in node specification space, 63
Voice over Internet Protocol (VoIP), 17
Volatage/VAr problem
capacitor control limits, 134
curtailable load control limits, 135
customer outage cost, 133
formulation of, 133–135
load balancing, 134
loss minimization, 134
voltage limits/current limits, 134
Voltage assessment techniques, 62, 62–65
condition number of the Jacobian matrix,
64–65
minimum singular value, 63–64
VIPI method, 62, 62–63
Voltage collapse
description of, 68
events leading to, 57–58
Voltage collapse dynamics, time frames and,
59
Voltage collapse points, detecting, direct
methods for, 69
Voltage instabilities, classification of, 61
Voltage instability, 57
preventive control scheme for power grid,
72
Voltage sag or swell, calculations,
parameters, and networks/systems
where index is applied, 135t
Voltage security, defined, 58
Voltage stability
application and implementation plan of,
70–71
assessment, 56–62
classification of, 58–59
INDEX
defining, 56–57
flowchart of, 55
indexing, 65–68
optimizing stability constraint through
preventive control of, 71–73
study algorithm, 56
voltage collapse and, 57–58
Voltage stability analysis tools, strengths/
weaknesses of, 51–56
Voltage/VAr control, 132–135
WAMS. See Web Access Management System
Water consumption, smart meters for, 22
Water resources, climate change and, 153
Web Access Management System, 19
demonstration projects for development of,
184–185
stability monitoring and analysis and, 132
techniques, 52
WEibull model, 149
Weighted least square estimation,
mathematical formulations for, 84–85
Weighted least square (WLS) method, 81
Wheat, bioenergy derived from, 145
Wide area monitoring systems, 20
Wide-area situational awareness, smart grid
and mandating development of, 9
Wi-Fi, 17
Wildlife, climate change and, 153
Wind energy, 140
congestion management and, 145
Wind power, existing world capacity,
1996–2008, 145
Wind technology, variability in, sources of, 149
Wind turbines
components of, 144
drawbacks with, 144
modeling, 145, 146t
systems of, 144–145
Wired communications technologies, 17
Wireless communications technologies, 17
Wireless networks, man-in-the-middle
(MitM) attacks and, 173
Wood, bioenergy derived from, 145
Worldwide Interoperability for Microwave
Access (WiMax), 17
World Wide Web Consortium (W3C), 163, 165t
Y-bus matrix, 81
IEEE Press Series on
Power Engineering
1. Principles of Electric Machines with Power Electronic Applications, Second
Edition
M. E. El-Hawary
2. Pulse Width Modulation for Power Converters: Principles and Practice
D. Grahame Holmes and Thomas Lipo
3. Analysis of Electric Machinery and Drive Systems, Second Edition
Paul C. Krause, Oleg Wasynczuk, and Scott D. Sudhoff
4. Risk Assessment of Power Systems: Models, Methods, and Applications
Wenyuan Li
5. Optimization Principles: Practical Applications to the Operations of Markets of
the Electric Power Industry
Narayan S. Rau
6. Electric Economics: Regulation and Deregulation
Geoffrey Rothwell and Tomas Gomez
7. Electric Power Systems: Analysis and Control
Fabio Saccomanno
8. Electrical Insulation for Rotating Machines: Design, Evaluation, Aging, Testing,
and Repair
Greg Stone, Edward A. Boulter, Ian Culbert, and Hussein Dhirani
9. Signal Processing of Power Quality Disturbances
Math H. J. Bollen and Irene Y. H. Gu
10. Instantaneous Power Theory and Applications to Power Conditioning
Hirofumi Akagi, Edson H. Watanabe, and Mauricio Aredes
11. Maintaining Mission Critical Systems in a 24/7 Environment, Second Edition
Peter M. Curtis
12. Elements of Tidal-Electric Engineering
Robert H. Clark
13. Handbook of Large Turbo-Generator Operation Maintenance, Second Edition
Geoff Klempner and Isidor Kerszenbaum
14. Introduction to Electrical Power Systems
Mohamed E. El-Hawary
15. Modeling and Control of Fuel Cells: Disturbed Generation Applications
M. Hashem Nehrir and Caisheng Wang
16. Power Distribution System Reliability: Practical Methods and Applications
Ali A. Chowdhury and Don O. Koval
17. Introduction to FACTS Controllers: Theory, Modeling, and Applications
Kalyan K. Sen and Mey Ling Sen
18. Economic Market Design and Planning for Electric Power Systems
James Momoh and Lamine Mili
19. Operation and Control of Electric Energy Processing Systems
James Momoh and Lamine Mili
20. Restructured Electric Power Systems: Analysis of Electricity Markets with Equilibrium Models
Xiao-Ping Zhang
21. An Introduction to Wavelet Modulated Inverters
S.A. Saleh and M. Azizur Rahman
22. Probabilistic Transmission System Planning
Wenyuan Li
23. Control of Electric Machine Drive Systems
Seung-Ki Sul
24. High Voltage and Electrical Insulation Engineering
Ravindra Arora and Wolfgang Mosch
25. Practical Lighting Design with LEDs
Ron Lenk and Carol Lenk
26. Electricity Power Generation: The Changing Dimensions
Digambar M. Tagare
27. Electric Distribution Systems
Abdelhay A. Sallam and Om P. Malik
28. Maintaining Mission Critical Systems in a 24/7 Environment, Second Edition
Peter M. Curtis
29. Power Conversion and Control of Wind Energy Systems
Bin Wu, Yongqiang Lang, Navid Zargan, and Samir Kouro
30. Integration of Distributed Generation in the Power System
Math Bollen and Fainan Hassan
31. High Voltage Protection for Telecommunications
Steven W. Blume
32. Doubly Fed Induction Machine: Modeling and Control for Wind Energy
Generation
Gonzalo Abad, Jesús Lopéz, Miguel Rodríguez, Luis Marroyo, and Grzegorz Iwanski
33. Smart Grid: Fundamentals of Design and Analysis
James Momoh
Sharafat Hussain