Power distribution planning_reference_book__second_edition__power_engineering__23_

Power Distribution
Planning Reference Book
Second Edition, Revised and Expanded

H. Lee Willis
ABB, Inc.
Raleigh, North Carolina, U.S.A.

MARCEL

MARCEL DEKKER, INC. NEW YORK • BASEL
lo
DEKKER

Although great care has been taken to provide accurate and current information, neither the author(s)
nor the publisher, nor anyone else associated with this publication, shall be liable for any loss,
damage, or liability directly or indirectly caused or alleged to be caused by this book. The material
contained herein is not intended to provide specific advice or recommendations for any specific
situation.

Trademark notice: Product or corporate names may be trademarks or registered trademarks and are
used only for identification and explanation without intent to infringe.

Library of Congress Cataloging-in-Publication Data
A catalog record for this book is available from the Library of Congress.

ISBN: 0-8247-4875-1

This book is printed on acid-free paper.

Headquarters
Marcel Dekker, Inc.
270 Madison Avenue, New York, NY 10016, U.S.A.
tel: 212-696-9000; fax: 212-685-4540

Distribution and Customer Service
Marcel Dekker, Inc.
Cimarron Road, Monticello, New York 12701, U.S.A.
tel: 800-228-1160; fax: 845-796-1772

Eastern Hemisphere Distribution
Marcel Dekker AG
Hutgasse 4, Postfach 812, CH-4001 Basel, Switzerland
tel: 41-61-260-6300; fax: 41-61-260-6333

World Wide Web
http://www.dekker.com

The publisher offers discounts on this book when ordered in bulk quantities. For more information,
write to Special Sales/Professional Marketing at the headquarters address above.

Copyright © 2004 by Marcel Dekker, Inc. All Rights Reserved.

Neither this book nor any part may be reproduced or transmitted in any form or by any means,
electronic or mechanical, including photocopying, microfilming, and recording, or by any information
storage and retrieval system, without permission in writing from the publisher.

Current printing (last digit):

10 9 8 7 6 5 4 3 2 1

PRINTED IN THE UNITED STATES OF AMERICA

POWER ENGINEERING

Series Editor

H. Lee Willis
ABB Inc.
Raleigh, North Carolina

1. Power Distribution Planning Reference Book, H. Lee Willis
2. Transmission Network Protection: Theory and Practice, Y. G.
Paithankar
3. Electrical Insulation in Power Systems, N. H. Malik, A. A. Al-
Arainy, and M. I. Qureshi
4. Electrical Power Equipment Maintenance and Testing, Paul Gill
5. Protective Relaying: Principles and Applications, Second Edition, J.
Lewis Blackburn
6. Understanding Electric Utilities and De-Regulation, Lorrin Philipson
and H. Lee Willis
7. Electrical Power Cable Engineering, William A. Thue
8. Electric Systems, Dynamics, and Stability with Artificial
Intelligence Applications, James A. Momoh and Mohamed E. El-
Haw ary
9. Insulation Coordination for Power Systems, Andrew R. Hileman
10. Distributed Power Generation: Planning and Evaluation, H. Lee
Willis and Walter G. Scott
11. Electric Power System Applications of Optimization, James A.
Momoh
12. Aging Power Delivery Infrastructures, H. Lee Willis, Gregory V.
Welch, and Randall R. Schrieber
13. Restructured Electrical Power Systems: Operation, Trading, and
Volatility, Mohammad Shahidehpour and Muwaffaq Alomoush
14. Electric Power Distribution Reliability, Richard E. Brown
15. Computer-Aided Power System Analysis, Ramasamy Natarajan
16. Power System Analysis: Short-Circuit Load Flow and Harmonics,
J. C. Das
17. Power Transformers: Principles and Applications, John J. Winders,
Jr.
18. Spatial Electric Load Forecasting: Second Edition, Revised and Ex-

panded, H. Lee Willis
19. Dielectrics in Electric Fields, Gorur G. Raju
20. Protection Devices and Systems for High-Voltage Applications,
Vladimir Gurevich
21. Electrical Power Cable Engineering: Second Edition, Revised and
Expanded, William A. Thue
22. Vehicular Electric Power Systems: Land, Sea, Air, and Space Ve-
hicles, Ali Emadi, Mehrdad Ehsani, and John M. Miller
23. Power Distribution Planning Reference Book: Second Edition, Re-
vised and Expanded, H. Lee Willis
24. Power System State Estimation: Theory and Implementation, Ali
Abur and Antonio Gomez Expos/to

ADDITIONAL VOLUMES IN PREPARATION

Series Introduction

The original edition of Power Distribution Planning Reference Book was the first in
Marcel's Dekker's Power Engineering series. It is a sign of the growing maturity of this
series, and the continued development and evolution of power engineering and the utility
industry, that many of the books in it are now going into their second editions. Power
engineering is certainly the oldest and was for many decades the most traditional of the
various areas within electrical engineering. Without doubt, electric power and utilities are
also the oldest technology-based sector of modern industry. Yet no other facet of our
technology and culture is undergoing such a lengthy and comprehensive revolution.
It would be a gross simplification to attribute all the changes the power industry is
seeing to de-regulation, which is in fact more an effect of much larger forces than a root
cause of the industry's continuing transformation. As proof, one only has to look at power
distribution. Distribution is the level of the power system least impacted by de-regulation,
yet it has changed dramatically in the past decade, as Power Distribution Planning
Reference Book, Second Edition bears witness. Fully 70% of the book is new compared to
the first edition. Overall, the second edition contains more than twice the content of the
first, changes fairly reflecting the growth and change of power distribution planning in the
21st century.
As both the editor of the Power Engineering series and the author of this book, I am
proud to include Power Distribution Planning Reference Book, Second Edition, in this
important group of books. Following the theme we have set from the beginning in Marcel
Dekker's Power Engineering series, this book provides modem power technology in a
context of proven, practical application; useful as a reference book as well as for self-study
and advanced classroom use. Marcel Dekker's Power Engineering series includes books
covering the entire field of power engineering, in all of its specialties and sub-genres, all
aimed at providing practicing power engineers with the knowledge and techniques they
need to meet the electric industry's challenges in the 21st century.

H. Lee Willis

in

Preface
This second edition of the Power Distribution Planning Reference Book is, like the first,
both a reference book and a tutorial guide for the planning of electric utility power delivery
systems. But it is much more than just a revision of the former edition. During the decade
since the first edition was published a number of forces have come together to greatly
complicate and expand the role of the power distribution planner. This new book, which
contains more than twice the content of the original, has been designed to fit the wider
scope and greater complexity of modern power distribution planning.
Foremost among the changes affecting planners has been an industry-wide move toward
very explicit management of reliability. Modern electric utilities cannot meet their customer
and regulatory expectations simply by designing to good standards and criteria as utilities
did in the past. A 21st-century power company must aim at and achieve specific targets with
respect to reliability of service.
But while reliability-targeted performance may be the greatest new technical challenge
affecting many modern power delivery planners, changes in utility business and financial
orientation have had a much larger impact on most. Performance-based and frozen-rate
schedules, disaggregation of the traditional vertical structure, and subtle changes in the
attitude of the investment community toward utilities in general have created tremendous
pressure to reduce spending and improve financial performance. While this means planners
are challenged more than ever to reduce cost, there is a subtle difference between
"spending" and "cost." Traditional methods of evaluating merit and ranking alternatives
against one another do not always fit well in this new environment.
The third change has been a rather sudden and still-growing recognition that "aging
infrastructures" have become a significant problem. A few utilities have systems whose
average age exceeds the nominal design lifetime of major electrical equipment. Almost all
have significant areas of their system where this is the case. Aging equipment has a higher
failure rate, requires more maintenance, and has a shorter expected lifetime than new
equipment, all factors that potentially reduce reliability and increase cost.
Beyond the very real challenge that dealing with increasing amounts of worn out
equipment creates, aging infrastructures have brought about a subtle change in technical

vi Preface

approach and cultural attitudes at utilities. Traditionally, the fact that all equipment would
eventually "die" was understood by all concerned, but it had no role in utility engineering
planning. Unavailability of equipment for brief periods, perhaps unexpectedly, was a
recognized problem and addressed by contingency planning methods. But the concept of
"lifetime" was completely absent from traditional T&D planning, which effectively
assumed that equipment already in the system as well as that being added would be there
"forever."
Today, planners, managers, and operators at many utilities realize that much of their
system will fail within the span of their career, a good deal of it in the very foreseeable
future. Lifetime, as affected by specification, utilization and required reliability, and care
given equipment, is suddenly something to be explicitly studied and managed. These
considerations, and their obvious interaction with targeted reliability and tight budget
considerations, greatly complicate T&D planning.
One reaction to the combination of increasing focus on business coupled with this need
to manage "lifetime" has been the advent of "Asset Management" as a business paradigm
for electric utilities. While that term has many variations in meaning within the industry, it
invariably means a closer integration of business and engineering, and of capital and O&M
planning, aimed at a "lifetime optimum" business-case for the acquisition, use, and
maintenance of equipment and facilities. The author does not mean to imply that the use of
asset management methods in the power industry has been driven by the aging
infrastructure issue. In fact it has been driven mostly by a search for an improved business
paradigm suitable for the modern regulatory and financial environment. But asset
management methods provided a very sound basis for considering all of the issues that
surround equipment aging and "lifetime management," and thus fit not only the business,
but many of the new technical issues in the industry.
Despite all these changes, distribution planning still involves developing and justifying a
schedule of future additions and changes that will assure the utility's goals for electric
delivery are met. Tactically, distribution planners must accomplish three tasks. First, they
must identify the goals for their system. Exactly what constitutes "satisfactory
performance?" How is that measured? What does "lowest cost" and "least budget" mean?
Unambiguous, quantitative targets must be established for all planning goals.
Second, planners must understand how differences in distribution system design and
equipment will affect the achievement of these goals. Distribution systems are complicated
combinatorial entities, whose performance and economy depend on the coordinated
interaction of tens of thousands of individual equipment and circuit elements. Worldwide,
there are several fundamentally different "design philosophies" for laying out a distribution
system and engineering it to work well — what might be called differing design paradigms.
All work well. But while each paradigm has its plusses and minuses, there is usually one
best design to achieve the planner's specific desired targets and end results.
Third, planners must find that best design, every time, from among the thousands or
even tens of thousands of feasible designs that might work for their system. Their planning
methodology must be comprehensive and complete, assuring that nothing is overlooked and
that every opportunity for savings or improvement is fully exploited.
This second edition has been written to address this distribution planning process in a
way that meets all of the new challenges discussed above. Following his own advice - that
reliability and business prioritization must be built into the system and the planning process,
and not "tacked on at the end" - the author has not updated the first edition by simply
adding a few new chapters on reliability-based planning, aging equipment, and business-
based prioritization methods.

Preface VII

Instead, the bulk of the original book has been entirely re-written in company with the
addition of many new topics and chapters new material. As a result, most of this second
edition is new. There are two completely new chapters on reliability (Chapters 21 and 23),
but more important, reliability engineering and planning concepts are distributed throughout
the book, beginning in Chapter 1 and continuing through to the end. Similarly, there are
new chapters on business-based "bang for the buck" prioritization methods (Chapter 6), on
aging infrastructures and their impacts (Chapters 7 and 8) and on business-based and asset
management planning methods (Chapter 28). But more importantly, those concepts are all
woven throughout all of the book, in both old and new chapters. The net result of these
additions and enhancements, along with other new chapters on distributed resources
(Chapter 10) and objectivity and accuracy in planning (Chapter 29), is that the Power
Distribution Planning Reference Book has more than doubled in content, from just under
290,000 words in the first edition to more than 600,000 here, an increase that the author
believes is rather representative of the increased challenge that modem planners face.
Roughly speaking, the job is about twice as difficult as it was in the past.
The much greater length of this second edition has brought some changes in
organization to make the book more useful both as a tutorial guide and as a reference for
practicing planners. First, to facilitate tutorial usage each chapter has been written as much
as possible as a stand-alone, serial treatise on its topic. (Chapters 13-15 are a notable
exception, being in reality a single 130+ page discussion of feeder planning.) Second, to
facilitate usage as a reference, numerous cross-references by topic and interaction have
been given among chapters, and each chapter concludes with a one-page table summarizing
its key concepts - useful for quick reference. A concluding chapter has been added that
summarizes key concepts and guidelines, and gives references to chapters and sections
where detail can be found on each point. Finally, the author has endeavored to make the
index particularly comprehensive and useful.
This book is organized into four parts. The first ten chapters constitute basic "resource"
material, each chapter being more or less a stand-alone tutorial on a specific area of modern
T&D planning or systems (e.g., Chapter 6 on prioritization and ranking methods). The
second part of the book, Chapters 11-19, is a bottom-up look at T&D systems in detail,
including their electrical performance, reliability, and cost. The third part of the book,
Chapters 20-25, covers T&D planning tools and technologies. Chapters 26-30 conclude
with a look at the planning process: how it is organized, how it meshes with other utility
functions, and how planners work within it.
Chapter 1 provides a brief introduction to distribution systems, their mission, the rules
that govern their behavior, and their performance and economics. This one-chapter
summary is provided for those unfamiliar with or new to power delivery systems.
Experienced planners can skip this chapter, although it is recommended, as several key
concepts, particularly the systems approach and Two-Q, are introduced here.
Ultimately, the T&D system exists solely to deliver power to energy consumers.
Chapter 2 looks at consumer demand for electric power. It explains Two-Q, a concept of
looking at both customer demand and system capability as composed of two dimensions:
the quantity (peak kW load, system capability) of power needed, and the quality of power
need (value of connectivity, reliability of service). Successful electric utilities keep
spending low while managing to provide satisfactory levels of both.
Chapter 3 looks at consumer demand as it looks to the power system, as electric load.
This chapter covers basic electric load concepts such as types of electric load (resistive,
impedance), appliance duty cycles and their interaction with one another and weather, load
curve shapes and factors, coincidence of load and diversity of peaks, and measurement of
demand and load curve shape.

viii Preface

Chapter 4 is a basic introduction to electric service reliability. It provides basic
definitions and concepts and builds a foundation for the considerable amount of reliability-
based planning and engineering material later in the book. The chapter describes the use of
reliability indices including the purposes and pitfalls of comparison and benchmarking.
Chapter 5 reviews traditional engineering economics from the perspective of the
distribution planner, mostly time-value-of-money analysis and its pragmatic application as a
decision-making tool. Although many utilities use newer asset management approaches to
prioritize their spending, traditional engineering economics are still used pervasively in the
analysis and cost evaluation that leads up to the decision-making system, regardless of what
"management paradigm" the utility is using.
Chapter 6 is perhaps the first that experienced planners will find a break from the past. It
looks at project and spending prioritization methods in some detail. While it deals
comprehensively with the traditional least-cost approach, the chapter is definitely focused
on modern "bang for the buck" methods as used in reliability-based planning, in budget-
constrained situations, and for "asset management." Both traditional and new analytical
methods as well as processes and procedures for their use are discussed and compared.
Chapters 7 and 8 provide tutorial and reference material on equipment aging, failure
rates, and in particular how aging infrastructures impact the reliability of a power system.
Failure rate does increase with age. But as these chapters show, very old equipment is
seldom the problem. It is usually equipment in "late middle age" that proves most
problematic.
Chapter 9 covers load reach, a measure of the distance that a circuit or system can move
power while providing suitable service voltage, and one of the author's favorite distribution
planning tools. Many power engineering techniques optimize equipment selection and
design on a per unit basis, which for circuits means a per foot, per kilometer, or similar
basis. Such methods represent a current-related optimization approach and are quite
important. By contrast, load reach can be thought of as a voltage or voltage drop
optimization of a system. It permits a planner to look at the overall needs of a feeder system
- power must be moved from substations out to the mid-point between substations. Circuit
designs and specifications that have sufficient load reach to meet this distance need
efficiently, but no more, generally prove most economical and reliable. The best performing
circuits are optimized on both a voltage and current basis. The chapter also addresses volt-
VAR performance and planning, something closely related to load-reach concepts and
attaining optimum voltage performance.
Chapter 10 is a tutorial on distributed resources (DR), which includes distributed
generation (DG), distributed energy storage (DS), and demand-side management (DSM).
Although not part of classical T&D planning, DR is increasingly a factor in many utility
decisions and is a required alternative that planners must consider in some regulatory
jurisdictions. But the author has included this chapter for another reason, too. DR methods
also include customer-side methods such as UPS and backup generation (a form of DG),
that principally affect service reliability. From a Two-Q perspective their capability pertains
mostly to the quality, not quantity, dimension. They are therefore of great interest to
modern planners looking for maximum bang for the buck in terms of reliability.
The second part of the book, Chapters 11 through 19, constitutes a detailed look at the
design, performance, and economics of a power distribution system, based on two
overriding principles. The first is the systems approach: sub-transmission, substation,
feeder, and service levels are integrated layers of the system. These disparate levels must
work well together and their overall performance and cost should be optimized. That comes
as much from sound coordination of each level with the others as it does from optimization
of any one level.

Preface ix

The second concept is cost. While the aforementioned changes in financial priorities
have caused many utilities to optimize spending (cash flow), not long-term cost as in the
past, detailed and explicit knowledge of what various alternatives will truly cost, now and in
the future, is vital for sound management of both the system and the utility's finances no
matter what paradigm is driving the budget prioritization. Therefore, Chapters 11-19 focus
a great deal of attention on determining all costs.
Chapters 11 and 12 look at conductor and transformer economic sizing in some detail,
beginning with basic equipment sizing economics in Chapter 11. The approaches discussed
there will be familiar to most planners. However, Chapter 12's extension of these concepts
to conductor or equipment type set design, which involves finding an optimal group of line
types or equipment sizes to use in building a distribution system, will be new to many.
Primary feeder circuit layout and performance is examined in Chapters 13 through 15.
Although there are exceptions in order to put important points in context, generally Chapter
13 deals with the equipment selection, layout, and performance of individual circuits, while
Chapter 15 looks at multi-feeder planning - for groups of feeders serving a widespread
area. Chapter 14 looks at reliability and reliability planning at the feeder level, and serves as
a conceptual and methodological bridge between Chapters 13 and 15, for while distribution
reliability begins with a sound design for each individual feeder, it is quite dependent on
inter-feeder contingency support and switching, which is the multi-feeder planning venue.
These three chapters also present and compare the various philosophies on system layout,
including "American" or "European" layout; large trunk versus multi-branch; loop, radial,
or network systems; single or dual-voltage feeder systems; and other variations.
Chapter 16 looks at distribution substations and the sub-transmission lines that route
power to them. Although they are not considered part of "distribution" in many utilities,
coordination of sub-transmission, substation, and distribution feeders is a critical part of the
system approach's optimization of overall performance and cost. Substations can be
composed of various types of equipment, laid out in many different ways, as described here.
Cost, capacity, and reliability vary depending on equipment and design, as do flexibility for
future design to accommodate uncertainty in future needs.
Chapter 17 takes a "systems approach" perspective to the performance and economics
of the combined sub-transmission/substation/feeder system. Optimal performance comes
from a properly coordinated selection of voltages, equipment types, and layout at all three
levels. This balance of design among the three levels, and its interaction with load density,
geographic constraints, and other design elements, is explored and evaluated in a series of
sample system design variations of distribution system performance, reliability, and
economy.
Chapter 18 focuses on locational planning and capacity optimization of substations,
looking at the issues involved, the types of costs and constraints planners must deal with,
and the ways that decisions and different layout rules change the resulting system
performance and economy. No other aspect of power delivery system layout is more
important than siting and sizing of substations. Substation sites are both the delivery points
for the sub-transmission system and the source points for the feeder system. Thus even
though the substation level typically costs less than either of the two other levels, its siting
and sizing largely dictate the design of those two other levels and often has a very
significant financial impact if done poorly.
Chapter 19 looks at the service (secondary) level. Although composed of relatively
small "commodity" elements, cumulatively this closest-to-the-customer level is surprisingly
complex, and represents a sizable investment, one that benefits from careful equipment
specification and design standards.

Preface

The third portion of the book, Chapters 20-25, discusses the tools and skills that modern
distribution planners need to do their work. In the 21st century, nearly all of these tools
make heavy use of computers, not just for analysis, but for data mining and decision
support. Regardless, this portion of the book begins with what many planners may not
consider to be tools: the engineering and operating guidelines (standards) and the criteria
that set limits on and specify the various engineered aspects of a T&D system. Chapters 20
and 21 focus respectively on voltage, loading, equipment and design guidelines, and
reliability, maintainability and service quality guidelines and criteria. Their overall
perspective is that criteria and guidelines are tools whose primary purposes should be to
assist planners in performing their job quickly. The more traditional (and obsolete) view
was that guidelines (standards) and criteria existed primarily to assure adequate service
quality and equipment lifetime. Modem utilities explicitly engineer aspects like voltage,
flicker, and reliability of service, so that role has been somewhat superceded. Regardless, an
important point is that good guidelines and criteria can greatly speed the planning process, a
critical point for modern utilities that must "do more with less."
Chapter 22 presents electrical performance analysis methods - engineering techniques
used to assess the voltage, current, power factor and loading performance of distribution
systems. The chapter focuses on application, not algorithms, covering the basic concepts,
models, methods and approaches used to represent feeders and evaluate their performance
and cost in short-range planning and feeder analysis.
Chapter 23 focuses on reliability analysis methods - engineering techniques used to
assess the expected reliability performance of distribution systems, including frequency and
duration of interruptions seen by customers, and severity and frequency of voltage sags on
the system. Focus is on application and use of modern, predictive reliability engineering
methods, not on algorithms and theory. Most important from the standpoint of modem
distribution planners, dependable and proven reliability-engineering methods do exist and
are effective. To an electric utility trying to meet regulatory targets and customer
expectations in the most effective business manner possible, these methods are vitally
important.
Chapter 24 looks at decision-making and optimization tools and methods for
distribution planning. Although modern planners depend on computerized methods for
nearly all planning, there is a substantial difference between the computerized performance
analysis methods of Chapters 22 and 23, and the automated planning methods presented
here. Automated planning tools can streamline the work flow while also leading to better
distribution plans. Most are based on application of optimization methods, which, while
mathematically complicated, are simple in concept. The chapter presents a brief, practical,
tutorial on optimization, on the various methods available, and on how to pick a method
that will work for a given problem. As in Chapters 22 and 23, the focus is on practical
application, not on algorithms. Automated methods for both feeder-system planning and
substation-siting and sizing are discussed, along with typical pitfalls and recommended
ways to avoid them.
Chapter 25 looks at spatial load forecasting tools and methods. All T&D plans begin
with some concept of what the future demand patterns will be: the forecast is the initiator of
the planning process - anticipation of demand levels that cannot be served satisfactorily is
the driver behind capital expansion planning. Characteristics of load growth that can be
used in forecasting are presented, along with rules of thumb for growth analysis. Various
forecasting procedures are delineated, along with a comparison of their applicability,
accuracy, and data and resource requirements.
The final portion of the book, Chapters 26-30, focuses on planning and the planning
process itself: how planners and their utilities can use the tools covered in Chapters 20-25,

Preface XI

and organize themselves and their work flow, and interact with other aspects and groups
within a utility, so that ultimately they both produce and execute good plans for the system
and accommodate the other needs that the utility has with respect to the distribution system
(mostly safety and business needs).
Planning processes and approaches have undergone considerable change since the first
edition of this book was published, and are in a state of continual change as this is written.
This portion of the book is nearly all new. It presents an integrated look at the three largest
changes affecting distribution planners, and how they can deal with their combined impact.
These changes are the increasing emphasis on reliability and achieving reliability targets, an
increasingly business-basis in all spending and resource decisions, and far less emphasis on
long-range system planning. Reliability is important because in the 21st century it has
become and will remain an explicitly measured and tracked aspect of utility performance,
something the utility (and its planners) must engineer and manage well. The increasing
business basis is due to a host of factors discussed in various parts of this book, but its net
impact on planners is to change the context within which their recommendations are made:
modern utilities make all spending decisions on a business-case basis.
The shift to more short-range and less long-range focus is not, as often thought, due to a
narrow "profit now" business focus by utilities. Instead, it reflects a shift in the importance
of both engineering and business considerations. The higher utilization ratios and the
greater emphasis on reliability faced by modern utilities mean that planning mistakes are
more costly and may take longer to fix. "Getting it right, now" is vital to a modern T&D
utility: very good, rather than just adequate, short-range planning is more important than
ever. Beyond this, modern business planning and management methods have proven very
successful at reducing long-term risk and retaining long-term flexibility: optimizing or even
organizing long-range T&D plans is simply not as critical to success as it once was.
Chapter 26 begins by looking at the distribution planning process itself. Despite all the
changes, the core of the planning process is much as it always was, even if the metrics used
and context within which it works have changed dramatically. Planning still involves
setting goals, identifying and evaluating options, and selecting the best option. Each of
these planning steps is examined in detail, along with its common pitfalls. Short- and long-
range planning are defined and their different purposes and procedures studied. Finally, the
functional steps involved in T&D planning, and the tools they use, are presented in detail.
Chapter 27 looks at how the forecasting tools covered in Chapter 25 are used, and how
planners organize themselves and their forecast-translated-to-capability-need methodology
so that they can efficiently accommodate their company's requirements. In a modern high-
utilization ratio, just-in-time utility, correctly anticipating future needs could be said to be
the critical part of planning. Fully half of all recent large-scale outages and blackouts have
been due, at least in part, to deficiencies that started with poor recognition or anticipation of
future system capability needs.
Chapter 28 looks at balancing reliability against spending. Many power engineers
buying this book will consider this its key chapter. It presents the basics of reliability-based
distribution planning including processes and procedures aimed at achieving specific
reliability targets. But at most electric delivery utilities there is more involved than just a
reliability-based approach and an explicit focus on attaining reliability targets. To be
successful, reliability-based planning must be interfaced into the "business" framework of
an electric utility. This includes a three-way melding of the traditional T&D planning tools
and methods (Chapters 17, 20, 23, 24 and 26), reliability-based analysis engineering
(Chapters 14, 21 and 23) and "bang for the buck" prioritization methodologies and budget-
constrained planning methods covered in Chapter 6. When successfully integrated into a
utility's investment, system, and operational planning functions, this is often called asset

xii Preface

management. Since these concepts are new, and represent a considerable step change to
most utilities, the chapter also includes a simple but effective "cookbook" on CERI (Cost-
Effective Reliability Improvement) projects used to bootstrap a utility's move toward
explicit cost- or spending-based management of reliability.
Chapter 29 is a considerable departure from previous chapters and sure to be somewhat
controversial and discomforting to some planners. It discusses objectivity and accuracy in
distribution planning, frankly and explicitly addressing the fact that some planning studies
are deliberately biased in both their analysis and their reporting, so that they do not present
a truly balanced representation and comparison of alternatives based on their merits. As
shown, there is a legitimate, and in fact critical, need for such reports in the power industry,
and there is nothing unethical about a planning study that declares its intent to "make a
case" for a particular option with an advantageous set of assumptions and analysis. But the
sad truth is that some planning studies contain hidden bias, because of unintended mistakes,
or deliberate efforts to misrepresent a truthful evaluation. Part of this chapter is a "tutorial
on cheating" - giving rules and examples of how to disguise a very biased analysis so that it
gives the appearance of objectivity and balance. This is then used to show how planners can
review a report for both accuracy and objectivity, and how to detect both unintended
mistakes as well as bias that has been carefully hidden.
Chapter 30 concludes with a summary and integrating perspective on the book's key
points as well as a set of guidelines and recommendations for modern distribution planners.
Along with Chapter 1, it provides a good "executive summary" to both T&D systems and
planning for a modern utility.
This book, along with a companion volume (Spatial Electric Load Forecasting, Marcel
Dekker, 2002), took more than a decade to complete. I wish to thank many good friends
and colleagues, including especially Jim Bouford, Richard Brown, Mike Engel, James
Northcote-Green, Hahn Tram, Gary Rackliffe, Randy Schrieber and Greg Welch, for their
encouragement and willing support. I also want to thank Rita Lazazzaro and Lila Harris at
Marcel Dekker, Inc., for their involvement and efforts to make this book a quality effort.
Most of all, I wish to thank my wife, Lorrin Philipson, for the many hours of review and
editorial work she unselfishly gave in support of this book, and for her constant, loving
support, without which this book would never have been completed.

H. Lee Willis

Contents
Series Introduction Hi

Preface iv

Power Delivery Systems 1
1.1 Introduction 1
1.2 T&D System's Mission 2
1.3 Reliability of Power Delivery 3
1.4 The "Natural Laws of T&D" 6
1.5 Levels of the T&D System 8
1.6 Utility Distribution Equipment 16
1.7 T&D Costs 21
1.8 Types of Distribution System Design 29
1.9 The Systems Approach and Two-Q Planning 35
1.10 Summary of Key Points 41
References and Bibliography 45

Consumer Demand and Electric Load 47
2.1 The Two Qs: Quantity and Quality of Power 47
2.2 Quantity of Power Demand: Electric Load 49
2.3 Electric Consumer Demand for Quality of Power 59
2.4 The Market Comb and Consumer Values 75
2.5 Two-Q Analysis: Quantity and Quality Versus Cost 78
2.6 Conclusion and Summary 82
References 84

XIII

xiv Contents

Electric Load, Coincidence, and Behavior 85
3.1 Introduction 85
3.2 Peak Load, Diversity, and Load Curve Behavior 85
3.3 Measuring and Modeling Load Curves 94
3.4 Summary 102
References 102

4 Power System Reliability 103
4.1 Introduction 103
4.2 Outages Cause Interruptions 107
4.3 Reliability Indices 111
4.4 Comparison of Reliability Indices Among Utilities 117
4.5 Benchmarking Reliability 120
References and Further Reading 133

5 Economics and Evaluation of Cost 135
5.2 Costs 136
5.3 Time Value of Money 141
5.4 Variability of Costs 158
5.5 Conclusion 163
References 165

6 Evaluation, Prioritization, and Approval 167
6.1 Decisions and Commitments 167
6.2 Evaluation, Comparison, Prioritization, and Approval 167
6.3 Traditional Regulated Utility Least-Cost Planning 178
6.4 The Benefit/Cost Ratio Paradigm 185
6.5 Incremental Benefit/Cost Evaluation 195
6.6 Profit-Based Planning Paradigms 219
6.7 Summary, Comments, and Conclusion 221

7 Equipment Ratings, Loadings, Lifetime, and Failure 231
7.2 Capacity Ratings and Lifetime 232
7.3 Aging, Deterioration, and Damage 246
7.4 Measures to Improve Equipment Reliability and Life 259
For Further Reading 266

Contents xv

8 Equipment Failures and System Performance 267

8.2 Equipment Failure Rate Increases with Age 267
8.3 A Look at Failure and Age in a Utility System 274
References 282

9 Load Reach and Volt-VAR Engineering 283
9.2 Voltage Behavior on a Distribution System 285
9.3 Load Reach and Distribution Capability 291
9.4 Load Reach, the Systems Approach, and Current and 298
Voltage Performance Optimization
9.5 Managing Voltage Drop on Distribution Systems 301
9.6 Volt-VAR Control and Correction 310
References 330

10 Distributed Resources 331
10.1 Managing Two-Q Demand on the Consumer Side 331
10.2 Energy and Demand Management Methods 332
10.3 Conservation Voltage Reduction 356
10.4 Distributed Generation 363
10.5 Electric Energy Storage Systems 373
10.6 Distributed Resource Cost Evaluation 378
10.7 Summary 387
Bibliography 387

11 Basic Line Segment and Transformer Sizing Economics 389
11.2 Distribution Lines 389
11.3 Transformers 399
11.4 Basic Equipment Selection Economics 407
11.5 Conclusion 418

12 Choosing the Right Set of Line and Equipment Sizes 419
12.2 Using Economic Loading and Voltage Drop Well 423
12.3 Economy and Performance of a Conductor Set 428
12.4 Conductor Set Design: Fundamental Aspects 436
12.5 Recommended Method for Conductor Set Design 443
12.6 Standard Transformer Sets 446
12.7 Conclusion 448

xvi Contents

13 Distribution Feeder Layout 449
13.2 The Feeder System 449
13.3 Radial and Loop Feeder Layout 465
13.4 Dual-Voltage Feeders 470
References 476

14 Feeder Layout, Switching, and Reliability 477
14.2 Designing Reliability into the Primary Feeder (MV) Level 486
14.3 Feeder System Strength 494
14.4 Contingency-Based Versus Reliability-Based Planning 497
14.5 Contingency Support and Switching Design 505
14.6 Protection and Sectionalization of the Feeder System 523

15 Multi-Feeder Layout 553
15.2 How Many Feeders in a Substation Service Area? 554
15.3 Planning the Feeder System 558
15.4 Planning for Load Growth 564
15.5 Formulae for Estimating Feeder System Cost 570
References 577

16 Distribution Substations 579
16.2 High-Side Substation Equipment and Layout 581
16.3 Transformer Portion of a Substation 591
16.4 Low-Side Portion of a Substation 598
16.5 The Substation Site 602
16.6 Substation Costs, Capacity, and Reliability 604
16.7 Substation Standardization 606
16.8 Substation Planning and the Concept of "Transformer Units" 610

17 Distribution System Layout 615
17.2 The T&D System in Its Entirety 615
17.3 Design Interrelationships 625
17.4 Example of a System Dominated by Voltage Drop, Not Capacity 651

Contents xvii

18 Substation Siting and System Expansion Planning 661
18.2 Substation Location, Capacity, and Service Area 661
18.3 Substation Siting and Sizing Economics 666
18.4 Substation-Level Planning: The Art 682
18.5 Guidelines to Achieve Low Cost in Substation Siting and Sizing 685
18.6 Substation-Level Planning: The Science 689
18.7 Planning with Modular Substations 698
18.8 Summary: The Most Important Point About Substation-Level Planning 703

19 Service Level Layout and Planning 705
19.2 The Service Level 705
19.3 Types of Service Level Layout 706
19.4 Load Dynamics, Coincidence, and Their Interaction with the Service Level 711
19.5 Service-Level Planning and Layout 716
19.6 High Reliability Service-Level Systems 725
19.7 Conclusion 733
References 733

20 Planning Goals and Criteria 735
20.2 Voltage and Customer Service Criteria and Guidelines 737
20.3 Other Distribution Design and Operating Guidelines 749
20.4 Load Ratings and Loading Guidelines 751
20.5 Equipment and Design Criteria 752

21 Reliability-Related Criteria and Their Use 757
21.2 Reliability Metrics, Targets, and Criteria 761
21.3 Practical Issues of Reliability-Based Criteria 772
21.4 Approaches and Criteria for Targeted Reliability Planning 775

22 Distribution Circuit Electrical Analysis 785
22.2 Models, Algorithms, and Computer Programs 787
22.3 Circuit Models 790
22.4 Models of Electric Load 798
22.5 Types of Electrical Behavior System Models 803
22.6 Coincidence and Load Flow Interaction 810

xviii Contents

23 Distribution System Reliability Analysis Methods 819
23.2 Contingency-Based Planning Methods 823
23.3 Engineering Reliability Directly 844
23.4 Analytical Distribution System Reliability Assessment 848
23.5 Important Aspects of Reliability Assessment 851
23.6 Reliability Simulation Studies and Financial Risk Assessment 857
23.7 Conclusion and Key Points 863

24 Automated Planning Tools and Methods 869
24.2 Fast Ways to Find Good Alternatives 870
24.3 Automated Feeder Planning Methods 881
24.4 Substation-Level and Strategic Planning Tools 892
24.5 Application of Planning Tools 900

25 T&D Load Forecasting Methods 909
25.1 Spatial Load Forecasting 909
25.2 Load Growth Behavior 911
25.3 Important Elements of a Spatial Forecast 916
25.4 Trending Methods 923
25.5 Simulation Methods for Spatial Load Forecasting 939
25.6 Hybrid Trending-Simulation Methods 954
25.7 Conclusion and Summary of Key Points 961

26 Planning and the T&D Planning Process 967
26.2 Goals, Priorities, and Direction 968
26.3 Tactical Planning: Finding the Best Alternative 978
26.4 Short- Versus Long-Range Planning 987
26.5 Uncertainty and Multi-Scenario Planning 992
26.6 The Power Delivery Planning Process 995
26.7 Summary and Key Points 1008

27 Practical Aspects of T&D Load Forecasting 1017
27.1 The First Step in T&D Planning 1017
27.2 Weather Normalization and Design Criteria 1018
27.3 Selection of a Forecast Method 1030
27.4 Application of Spatial Forecast Methods 1039
Bibliography and References 1054

Contents xix

28 Balancing Reliability and Spending 1055
28.2 The Fundamental Concepts 1058
28.3 Optimizing Reliability Cost Effectiveness 1063
28.4 CERI - A Practical Method to "Bootstrap" Reliability Improvement 1078
28.5 Required Tools and Resources for Reliability Planning 1102
28.6 "Equitableness" Issues in Reliability Optimization 1106
28.7 Approaches to Setting and Planning Reliability Targets 1113
28.8 Asset Management 1117

29 Objectivity, Bias, and Accuracy in Planning 1127
29.1 Introduction and Purpose of this Chapter 1127
29.2 Objective Evaluation, Proponent Study, or Simply Poor Work? 1129
29.3 Ways that Bias Makes Its Way into a T&D Planning Study 1132
29.4 The "Rules" Used to Bias Planning Studies in an Unseen Manner 1135
29.5 Areas Where Bias or Mistakes Are Often Introduced into a Study 1140
29.6 Examples of Bogus, Proponent, and Masked Studies 1148
29.7 Guidelines for Detecting, Finding, and Evaluating Bias 1159
29.8 Summary and Conclusion: Forewarned is Forearmed 1184
References 1188

30 Key Points, Guidelines, Recommendations 1189
30.2 On Distribution Systems 1189
30.3 On Utilities and Utility Practices 1193
30.4 On Planning Well 1199
References 1206

Index 1207

1
Power Delivery Systems

1.1 INTRODUCTION

Retail sale of electric energy involves the delivery of power in ready to use form to the final
consumers. Whether marketed by a local utility, load aggregator, or direct power retailer,
this electric power must flow through a power delivery system on its way from power
production to customer. This transmission and distribution (T&D) system consists of
thousands of transmission and distribution lines, substations, transformers, and other
equipment scattered over a wide geographical area and interconnected so that all function in
concert to deliver power as needed to the utility's customers.
This chapter is an introductory tutorial on T&D systems and their design constraints.
For those unfamiliar with power delivery, it provides an outline of the most important
concepts. But in addition, it examines the natural phenomena that shape T&D systems and
explains the key physical relationships and their impact on design and performance. For this
reason experienced planners are advised to scan this chapter, or at least its conclusions, so
that they understand the perspective upon which the rest of the book builds.
In a traditional electric system, power production is concentrated at only a few large,
usually isolated, power stations. The T&D system moves the power from those often distant
generating plants to the many customers who consume the power. In some cases, cost can
be lowered and reliability enhanced through the use of distributed generation - numerous
smaller generators placed at strategically selected points throughout the power system in
proximity to the customers. This and other distributed resources - so named because they
are distributed throughout the system in close proximity to customers - including storage
systems and demand-side management, often provide great benefit.
But regardless of the use of distributed generation or demand-side management, the
T&D system is the ultimate distributed resource, consisting of thousands, perhaps millions,
of units of equipment scattered throughout the service territory, interconnected and
1

2 Chapter 1

operating in concert to achieve uninterrupted delivery of power to the electric consumers.
These systems represent an investment of billions of dollars, require care and precision in
their operation, and provide one of the most basic building blocks of our society - widely
available, economical, and reliable energy.
This chapter begins with an examination of the role and mission of a T&D system -
why it exists and what it is expected to do. Section 1.3 looks at several fundamental
physical laws that constrain T&D systems design. The typical hierarchical system structure
that results and the costs of its equipment are summarized in sections 1.4 and 1.5. In section
1.6, a number of different ways to lay out a distribution system are covered, along with their
advantages and disadvantages. The chapter ends with a look at the "systems approach" -
perhaps the single most important concept in the design of retail delivery systems which are
both inexpensive and reliable.

1.2 T&D SYSTEM'S MISSION

A T&D system's primary mission is to deliver power to electrical consumers at their place
of consumption and in ready-to-use form. The system must deliver power to the customers,
which means it must be dispersed throughout the utility service territory in rough proportion
to customer locations and demand (Figure 1.1). This is the primary requirement for a T&D
system, and one so basic it is often overlooked - the system must cover ground - reaching
every customer with an electrical path of sufficient strength to satisfy that customer's
demand for electric power.
That electrical path must be reliable, too, so that it provides an uninterrupted flow of
stable power to the utility's customers. Reliable power delivery means delivering all of the
power demanded, not just some of the power needed, and doing so all of the time. Anything
less than near perfection in meeting this goal is considered unacceptable - 99.9% reliability
of service may sound impressive, but it means nearly nine hours of electric service
interruption each year, an amount that would be unacceptable in nearly any first-world
country.

2011
WINTER PEAK
3442 MW

N

Ten miles

Shading indicates
relative load
density. Lines
show major roads
and highways.

Figure 1.1 Map of electrical demand for a major US city shows where the total demand of more than
2,000 MW peak is located. Degree of shading indicates electric load distribution. The T&D system
must cover the region with sufficient capacity at every location to meet the customer needs there.


Table 1.1 Required Functions, or "Mission Goals" for a Power Delivery System
1. Cover the utility's service territory, reaching all consumers who wish to be
connected and purchase power.
2. Have sufficient capability to meet the peak demands of those energy
consumers.
3. Provide satisfactory continuity of service (reliability) to the connected
energy consumers.
4. Provide stable voltage quality regardless of load level or conditions.

Beyond the need to deliver power to the customer, the utility's T&D system must also
deliver it in ready-to-use form - at the utilization voltage required for electrical appliances
and equipment, and free of large voltage fluctuations, high levels of harmonics, or transient
electrical disturbances (Engel et al., 1992).
Most electrical equipment in the United States is designed to operate properly when
supplied with 60 cycle alternating current at between 114 and 126 volts, a plus or minus
five percent range centered on the nominal utilization voltage of 120 volts (RMS average
of the alternating voltage). In many other countries, utilization standards vary from 230 to
slightly over 250 volts, at either 50 or 60 cycles AC.1 But regardless of the utilization
voltage, a utility must maintain the voltage provided to each customer within a narrow
range centered within the voltages that electric equipment is designed to tolerate.
A ten percent range of delivery voltage throughout a utility's service area may be
acceptable, but a ten percent range of fluctuation in the voltage supplied to any one
customer is not. An instantaneous shift of even three percent in voltage causes a perceptible,
and to some people disturbing, flicker in electric lighting. More important, voltage
fluctuations can cause erratic and undesirable behavior of some electrical equipment.
Thus, whether high or low within the allowed range, the delivery voltage of any one
customer must be maintained at about the same level all the time - normally within a range
of three to six percent - and any fluctuation must occur slowly. Such stable voltage can be
difficult to obtain, because the voltage at the customer end of a T&D system varies
inversely with electric demand, falling as the demand increases, rising as it decreases. If this
range of load fluctuation is too great, or if it happens too often, the customers may consider
it poor service.
Thus, a T&D system's mission is to meet the four goals in Table 1.1, and of course,
above all else, to achieve these goals at the lowest cost possible and in a safe and
esthetically acceptable manner.
1.3 RELIABILITY OF POWER DELIVERY
Reliability of service was always a priority of electric utilities, however the tone of that
focus, and the perspective on reliability began to change in the 1990s when the power
industry saw a growing emphasis on reliability of customer service. In the early part of the
electric era (1890 - 1930) most electric utilities viewed interruptions of service primarily as
interruptions in revenue - when outages of equipment occur and the utility cannot provide
1
Power is provided to customers in the United States by reversed alternating current legs (+120
volts and -120 volts wire to ground). This provides 240 volts of power to any appliance that needs
it, but for purposes of distribution engineering and performance acts like only 120 volt power.

4 Chapter 1

electric service, it earns no money (Abbott, 1895). Industry studies of outages and
interruptions during this period were almost entirely based around the "loss of revenue" that
the utility would experience from poor reliability, and reliability was managed from that
perspective. To some extent, the view of utilities then was the same as a cable or broadcast
TV network might have today: when we aren't distributing, we aren't earning. Just as a TV
network knows that when its broadcast is interrupted its revenues (from advertising) are cut,
so electric utilities in the first third of the 20th century knew that when equipment was out,
they weren't earning money. Resources and actions to avoid outages were justified and
managed from the standpoint that, to an extent, they were profitable because they kept the
revenue stream coming into the utility's coffers.
During the 1930s through the 1960s, electric power came to be viewed as an essential
and needed service, and reliability came to be viewed as an obligation the utility had to its
customers. Most utilities, including their upper management, adopted a "public
stewardship" attitude in which they viewed themselves and their company as having an
obligation to maintain power flow to their customers. However, the computer and data
collection technologies needed to collect and manage quantitative data on customer-level
reliability were not available at that time (as they are today). A widespread outage overnight
might be reported to management the next morning as "we had a lot of customers, maybe as
many as 40,000, out for quite a while, and had about sixteen crews on it all night." No, or
very limited, reporting was done to state regulatory agencies.2 As a result, most utilities
used only informal means based on experience and good intentions, and sometimes
misguided intuition, to manage reliability.
Several changes occurred during the period 1970 through 2000 that led to more
quantitative emphasis on reliability. First, electric power increasingly became "mission
critical" to more and more businesses and homes. Equally important, it became possible to
measure and track reliability of service in detail. Modern SCAD A, system monitoring,
outage management, and customer information systems permit utilities to determine which
customer is out of service, and when, and why. Reports (see footnote 2, below) can be
prepared on individual outages and the reliability problems they cause, and on the aggregate
performance of the whole system over any period of time. Managerial study of past
performance, and problems, could be done in detail, "slicing and dicing" the data in any
number of ways and studying it for root causes and improved ways of improving service.
Thus, the growing capability to quantitatively measure and study reliability of service
enabled the industry to adopt much more specific and detailed managerial approaches to
reliability.
Simultaneously, during the 1980s and 1990s, first-world nations adopted increasing
amounts of digital equipment. "Computerized" devices made their way into clocks, radios,
televisions, stoves, ovens, washers and dryers and a host of other household appliances.
Similarly, digital systems became the staple of "efficient" manufacturing and processing
industries, without which factories could not compete on the basis of cost or quality.
This dependence on digital systems raised the cost of short power interruptions. For
example, into the mid 1970s, utilities routinely performed switching (transfer of groups of
between 100 and 2000 homes) in the very early morning hours, using "cold switching"; a

2
By contrast, today management, and regulators, would receive a report more like "a series of weather-
related events between 1:17 and 2:34 interrupted service to 36,512 customers, for an average of 3
hours and 12 minutes each, with 22,348 of them being out for more than four hours. Seven crews
responded to outage restoration and put in a total of 212 hours restoring service. Fifteen corrective
maintenance tickets (for permanent completion of temporary repairs made to restore service) remain
to be completed as of 9 AM this morning."


Table 1.2 Summary of Power Distribution Reliability
Outages of equipment cause interruptions of service.
In most distribution systems, every distribution equipment outage causes service interruptions
because there is only one electrical path to every consumer.
Reliability measures are based on two characteristics:
Frequency - how often power interruptions occur
Duration - how long they last
Equity of reliability - assuring that all customers receive nearly the same level of reliability - is
often an important part of managing reliability performance.
Drops in voltage (sags) often have the same impact as complete cessation of power.

branch of a feeder would be briefly de-energized as it was disconnected from one feeder,
and then energized again through closing a switch to another feeder a few seconds, or
minutes, later. Such short, late night/early morning interruptions caused few customer
complaints in the era before widespread use of digital equipment. Energy consumers who
even noticed the events were rare; analog electric clocks would fall a few seconds behind,
that was all. Like the clocks, devices in "pre-digital" homes would immediately take up
service again when power was restored.
But today, a similar "cold switching" event would disable digital clocks, computers, and
electronic equipment throughout the house, leaving the homeowners to wake up (often late
because alarms have not gone off) to a house full of blinking digital displays. Surveys have
shown homeowners consider even a "blink" interruption to cause them between seven and
fifteen minutes of inconvenience, resetting clocks, etc. To accommodate this need, utilities
must use "hot switching" in which a switch to the new source is closed first, and the tie to
the old source is opened. This avoids creating any service interruption, but can create
operating problems by leading to circulating (loop) currents that occasionally overload
equipment.
This gradual change in the impact of brief interruptions has had even a more profound
impact on business and industry than on homeowners. In the 1930s, the outage of power
to a furniture factory meant that power saws and tools could not be used until power was
restored. The employees had to work with ambient light through windows. Productivity
would decrease, but not cease. And it returned to normal quickly once power was
restored. Today, interruption of power for even a second to most furniture factories
would immediately shut down their five-axis milling machines and CAM-assembly
robots. Work in progress would often be damaged and lost. And after power is restored
the CAM assembly systems may take up to an hour to re-boot, re-initialize, and re-start:
in most businesses and industries that rely on digital equipment, production is almost
always interrupted for a longer period than the electrical service interruption itself.
Thus, by the 1990s, consumers (and thus regulators) paid somewhat more attention to
service interruptions, particularly short duration interruptions. But it would be far too
simplistic to attribute the increasing societal emphasis on power reliability only to the use of
digital equipment. The "digital society" merely brought about an increasing sensitivity to
very short interruptions. In the bigger sense, the need for reliable electric power grew
because society as a whole came to rely much more on electric power, period. Had digital
equipment never been invented, if the world were entirely analog, there would still be the

6 Chapter 1

same qualitative emphasis on reliability. Specific attention to "blinks" and short-term
interruptions might receive somewhat less attention, and longer duration interruptions a bit
more, than what actually occurs today. But overall, reliability of power would be just as
crucial for businesses and private individuals. For these reasons, the power industry began
to move toward quantitative, pro-active management of power system reliability: set
targets, plan to achieve them, monitor progress, and take corrective actions as needed.
The most salient points about power system reliability are summarized in Table 1.2.
Chapters 4,7, 8, 14, 21, 23, and 28 will discuss reliability and its planning and management
in much further detail.

1.4 THE "NATURAL LAWS OF T&D"

The complex interaction of a T&D system is governed by a number of physical laws
relating to the natural phenomena that have been harnessed to produce and move electric
power. These interactions have created a number of "truths" that dominate the design of
T&D systems:

1. It is more economical to move power at high voltage. The higher the voltage, the
lower the cost per kilowatt to move power any distance.
2. The higher the voltage, the greater the capacity and the greater the cost of
otherwise similar equipment. Thus, high voltage lines, while potentially
economical, cost a great deal more than low voltage lines, but have a much
greater capacity. They are only economical in practice if they can be used to
move a lot of power in one block - they are the giant economy size, but while
always giant, they are only economical if one truly needs the giant size.
3. Utilization voltage is useless for the transmission of power. The 120/240 volt
single-phase utilization voltage used in the United States, or even the 250
volt/416 volt three-phase used in "European systems" is not equal to the task of
economically moving power more than a few hundred yards. The application of
these lower voltages for anything more than very local distribution at the
neighborhood level results in unacceptably high electrical losses, severe voltage
drops, and astronomical equipment cost.
4. It is costly to change voltage level - not prohibitively so, for it is done throughout
a power system (that's what transformers do) - but voltage transformation is a
major expense which does nothing to move the power any distance in and of
itself.
5. Power is more economical to produce in very large amounts. Claims by the
advocates of modern distributed generators notwithstanding, there is a significant
economy of scale in generation - large generators produce power more
economically than small ones. Thus, it is most efficient to produce power at a
few locations utilizing large generators.3

3
The issue is more complicated than just a comparison of the cost of big versus small generation. In
some cases, distributed generation provides the lowest cost overall, regardless of the economy of
scale, due to constraints imposed by the T&D system. Being close to the customers, distributed
generation does not require T&D facilities to move the power from generation site to customer.


6. Power must be delivered in relatively small quantities at a low (120 to 250 volt)
voltage level. The average customer has a total demand equal to only 1/10,000 or
1/100,000 of the output of a large generator.

An economical T&D system builds upon these concepts. It must "pick up" power at a
few, large sites (generating plants) and deliver it to many, many more small sites
(customers). It must somehow achieve economy by using high voltage, but only when
power flow can be arranged so that large quantities are moved simultaneously along a
common path (line). Ultimately, power must be subdivided into "house-sized" amounts,
reduced to utilization voltage, and routed to each business and home via equipment whose
compatibility with individual customer needs means it will be relatively quite inefficient
compared to the system as a whole.

Hierarchical Voltage Levels

The overall concept of a power delivery system layout that has evolved to best handle these
needs and "truths" is one of hierarchical voltage levels as shown in Figure 1.2.
As power is moved from generation (large bulk sources) to customer (small demand
amounts) it is first moved in bulk quantity at high voltage - this makes particular sense
since there is usually a large bulk amount of power to be moved out of a large generating
plant. As power is dispersed throughout the service territory, it is gradually moved down to
lower voltage levels, where it is moved in ever smaller amounts (along more separate paths)
on lower capacity equipment until it reaches the customers. The key element is a "lower
voltage and split" concept.
Thus, the 5 kW used by a particular customer - Mrs. Rose at 412 Oak Street in
Metropolis City - might be produced at a 750 MW power plant more than three hundred
miles to the north. Her power is moved as part of a 750 MW block from plant to city on a
345 kV transmission line to a switching substation. Here, the voltage is lowered to 138 kV
through a 345 to 138 kV transformer, and immediately after that the 750 MW block is

• 345 kV transmission I ] switching substation
•" 138 kV transmission Q substation
— 12.47 kV primary feeder • service transformer
120/240 volt secondary

Figure 1.2 A power system is structured in a hierarchical manner with various voltage levels. A key
concept is "lower voltage and split" which is done from three to five times during the course of power
flow from generation to customer.

Chapter 1

split into five separate flows in the switching substation buswork, each of these five parts
being roughly 150 MW. Now part of a smaller block of power, Mrs. Rose's electricity is
routed to her side of Metropolis on a 138 kV transmission line that snakes 20 miles through
the northern part of the city, ultimately connecting to another switching substation. This 138
kV transmission line feeds power to several distribution substations along its route,4 among
which it feeds 40 MW into the substation that serves a number of neighborhoods, including
Mrs. Rose's. Here, her power is run through a 138 kV/12.47 kV distribution transformer.
As it emerges from the low side of the substation distribution transformer at 12.47 kV
(the primary distribution voltage) the 40 MW is split into six parts, each about 7 MW, with
each 7 MVA part routed onto a different distribution feeder. Mrs. Rose's power flows along
one particular feeder for two miles, until it gets to within a few hundred feet of her home.
Here, a much smaller amount of power, 50 kVA (sufficient for perhaps ten homes), is
routed to a service transformer, one of several hundred scattered up and down the length of
the feeder. As Mrs. Rose's power flows through the service transformer, it is reduced to
120/240 volts. As it emerges, it is routed onto the secondary system, operating at 120/240
volts (250/416 volts in Europe and many other countries). The secondary wiring splits the
50 kVA into small blocks of power, each about 5 kVA, and routes one of these to Mrs.
Rose's home along a secondary conductor to her service drops - the wires leading directly
to her house.
Over the past one hundred years, this hierarchical system structure has proven a most
effective way to move and distribute power from a few large generating plants to a widely
dispersed customer base. The key element in this structure is the "reduce voltage and split"
function - a splitting of the power flow being done essentially simultaneously with a
reduction in voltage. Usually, this happens between three and five times as power makes its
way from generator to customers.

1.5 LEVELS OF THE T&D SYSTEM

As a consequence of this hierarchical structure of power flow from power production to
energy consumer, a power delivery system can be thought of very conveniently as
composed of several distinct levels of equipment, as illustrated in Figure 1.3. Each level
consists of many units of fundamentally similar equipment, doing roughly the same job, but
located in different parts of the system so as to "cover" the entire utility service territory.
For example, all of the distribution substations are planned and laid out in approximately
the same manner and do roughly the same job. All are composed of roughly similar
equipment doing the same job. Some substations might be "larger" than others in both
physical and capability terms - one could have four 50 MVA transformers and associated
equipment, another only one 7 MVA transformer. But both perform the same function for
their area of the system, taking incoming power from sub-transmission, lowering voltage,
splitting the power flow, and routing it onto distribution feeders for delivery in the
neighborhoods around them. These constitute a "level" of the system, because all the power
delivered everywhere flows through one such substation; in every part of the utility system,
there is a "local" substation whose function is to provide the power for the neighborhoods
around it. Together, these substations constitute the "substation level" of the system. Their
service areas fit together in a mosaic, each covering its piece of the service territory.

4
Transmission lines whose sole or major function is to feed power to distribution substations are often
referred to as "sub-transmission" lines.


Transmission

Service/Secondary

Customer

Figure 13 A T&D system consists of several levels of power delivery equipment, each feeding the
one below it.

10 Chapter 1

Likewise, the feeders the substations route power into are all similar in equipment
type, layout, and mission, and all service transformers to which those feeders route power
are similarly serving the same basic mission and are designed with similar planning goals
and to similar engineering standards.
Thus, power can be thought of as flowing "down" through these various levels, on its
way from power production and the wholesale grid to the energy consumers. As it moves
from the generation plants (system level) to the energy consumers, the power travels
through the transmission level, to the sub-transmission level, to the substation level, onto
and through the primary feeder level, and onto the secondary service level, where it finally
reaches the customer. Each level takes power from the next higher level in the system and
delivers it to the next lower level in the system. In almost all cases each flow of power is
split into several paths at or shortly after the transition down to the next level.
While each level varies in the types of equipment it has, its characteristics, mission, and
manner of design and planning all share several common characteristics:

• Each level is fed power by the one above it, in the sense that the next higher level
is electrically closer to the generation.

• Both the nominal voltage level and the average capacity of equipment drops from
level to level, as one moves from generation to customer. Transmission lines
operate at voltages of between 69 kV and 1,100 kV and have capacities between
50 and 2,000 MW. By contrast, distribution feeders operate between 2.2 kV and
34.5 kV and have capacities somewhere between 2 and 35 MW.

• Each level has many more pieces of equipment in it than the one above. A system
with several hundred thousand customers might have fifty transmission lines, one
hundred substations, six hundred feeders, and forty thousand service
transformers.

• As a result, the net capacity of each level (number of units times average size)
increases as one moves toward the customer. A power system might have 4,500
MVA of substation capacity but 6,200 MVA of feeder capacity and 9,000 MVA
of service transformer capacity installed.5

• Reliability drops as one moves closer to the customer. A majority of service
interruptions are a result of failure (either due to aging or to damage from severe
weather) of transformers, connectors, or conductors very close to the customer, as
shown in Figure 1.4.

Table 1.3 gives statistics for a typical system. The net effect of the changes in average
size and number of units is that each level contains a greater total capacity than the level
above it - the service transformer level in any utility system has considerably more installed
capacity (number of units times average capacity) than the feeder system or the substation
system. Total capacity increases as one heads toward the customer because of non-
coincidence of peak load (which will be discussed in Chapter 3) and for reliability purposes.

5
This greater-capacity-at-every-lower-level characteristic is a deliberate design feature of most power
systems, and required both for reliability reasons and to accommodate coincidence of load, which will
be discussed in Chapter 3.


Ten-year Service Interruption
Statistics by Level of Cause
15

io

E >
—

Oo

Generation Trans. Subs. Feeders Service
Level of the Power System

Figure 1.4 Ten years of customer interruptions for a large electric system, grouped by level of cause.
Interruptions due to generation and transmission often receive the most attention because they usually
involve a large number of customers simultaneously. However, such events are rare whereas failures
and interruptions at the distribution level create a constant background level of interruptions.

Table 13 Equipment Level Statistics for a Medium-Sized Electric System
Level Voltage Number Avg. Cap. Total Cap
of System kV of Units MVA MVA
Transmission 345, 138 12 150 1,400
Sub-transmission 138, 69 25 65 1,525
Substations 139/23.9, 69/13.8 45 44 1,980
Feeders 23.9, 13.8 227 11 2,497
Service Trans. .12, .24 60,000 .05 3,000
Secondary/Service .12, .24 250,000 .014 3,500
Customer .12 250,000 .005 1,250

Figure 1.5 A network is an electrical system with more than one path between any two points,
meaning that (if properly designed) it can provide electrical service even if any one element fails.

12 Chapter 1

The Transmission Level

The transmission system is a network of three-phase lines operating at voltages generally
between 115 kV and 765 kV. Capacity of each line is between 50 MVA and 2,000 MVA.
The term "network" means that there is more than one electrical path between any two
points in the system (Figure 1.5). Networks are laid out in this manner for reasons of
reliability and operating flow - if any one element (line) fails, there is an alternate route and
power flow is (hopefully) not interrupted.
In addition to their function in moving power, portions of the transmission system - the
largest elements, namely its major power delivery lines - are designed, at least in part, for
stability needs. The transmission grid provides a strong electrical tie between generators, so
that each can stay synchronized with the system and with the other generators. This
arrangement allows the system to operate and to function evenly as the load fluctuates and
to pick up load smoothly if any generator fails - what is called stability of operation. (A
good deal of the equipment put into transmission system design, and much of its cost, is for
these stability reasons, not solely or even mainly for moving power.)

The Sub-Transmission Level

The sub-transmission lines in a system take power from the transmission switching stations
or generation plants and deliver it to substations along their routes. A typical sub-
transmission line may feed power to three or more substations. Often, portions of the
transmission system - bulk power delivery lines, lines designed at least in part for stability
as well as power delivery needs - do this too, and the distinction between transmission and
sub-transmission lines becomes rather blurred.
Normally, sub-transmission lines are in the range of capacity of 30 MVA up to perhaps
250 MVA, operating at voltages from 34.5 kV to as high as 230 kV. With occasional
exceptions, sub-transmission lines are part of a network grid - they are part of a system in
which there is more than one route between any two points. Usually, at least two sub-
transmission routes flow into any one substation, so that feed can be maintained if one
fails.6

The Substation Level

Substations, the meeting points between the transmission grid and the distribution feeder
system, are where a fundamental change takes place within most T&D systems. The
transmission and sub-transmission systems above the substation level usually form a
network, as discussed above, with more than one power flow path between any two parts.
But from the substation on to the customer, arranging a network configuration would
simply be prohibitively expensive. Thus, most distribution systems are radial - there is only
one path through the other levels of the system.
Typically, a substation occupies an acre or more of land, on which the various necessary
substation equipment is located. Substation equipment consists of high and low voltage
racks and busses for the power flow, circuit breakers for both the transmission and
distribution level, metering equipment, and the "control house," where the relaying,
measurement, and control equipment is located. But the most important equipment - what
gives this substation its capacity rating, are the substation transformers, which convert the
6
Radial feed — only one line ~ is used in isolated, expensive, or difficult transmission situations,
but for reliability reasons is not recommended.


incoming power from transmission voltage levels to the lower primary voltage for
distribution.
Individual substation transformers vary in capacity, from less than 10 MVA to as much
as 150 MVA. They are often equipped with tap-changing mechanisms and control
equipment to vary their windings ratio so that they maintain the distribution voltage within
a very narrow range, regardless of larger fluctuations on the transmission side. The
transmission voltage can swing by as much as 5%, but the distribution voltage provided on
the low side of the transformer stays within a narrow band, perhaps only ± .5%.
Very often, a substation will have more than one transformer. Two is a common
number, four is not uncommon, and occasionally six or more are located at one site. Having
more than one transformer increases reliability - in an emergency, a transformer can handle
a load much over its rated load for a brief period (e.g., perhaps up to 140% of rating for up
to four hours). Thus, the T&D system can pick up the load of the outaged portions during
brief repairs and in emergencies.
Equipped with from one to six transformers, substations range in "size" or capacity from
as little as five MVA for a small, single-transformer substation, serving a sparsely
populated rural area, to more than 400 MVA for a truly large six-transformer station,
serving a very dense area within a large city.
Often T&D planners will speak of a transformer unit, which includes the transformer
and all the equipment necessary to support its use - "one-fourth of the equipment in a four-
transformer substation." This is a much better way of thinking about and estimating cost for
equipment in T&D plans. For while a transformer itself is expensive (between $50,000 and
$1,000,000); the bus work, control, breakers, and other equipment required to support its use
can double or triple that cost. Since that equipment is needed in direct proportion to the
transformer's capacity and voltage, and since it is needed only because a transformer is
being added, it is normal to associate it with the transformer as a single planning unit - add
the transformer, add the other equipment along with it.
Substations consist of more equipment, and involve more costs, than just the electrical
equipment. The land (the site) has to be purchased and prepared. Preparation is non-trivial.
The site must be excavated, a grounding mat (wires running under the substation to protect
against an inadvertent flow during emergencies) laid down, and foundations and control
ducting for equipment must be installed. Transmission towers to terminate incoming
transmission must be built. Feeder getaways - ducts or lines to bring power out to the
distribution system - must be added.

The Feeder Level

Feeders, typically either overhead distribution lines mounted on wooden poles or
underground buried or ducted cable sets, route the power from the substation throughout its
service area. Feeders operate at the primary distribution voltage. The most common primary
distribution voltage in use throughout North America is 12.47 kV, although anywhere from
4.2 kV to 34.5 kV is widely used. Worldwide, there are primary distribution voltages as low
as 1.1 kV and as high as 66 kV. Some distribution systems use several primary voltages -
for example 23.9 kV and 13.8 kV and 4.16 kV.
A feeder is a small transmission system in its own right, distributing between 2 MVA to
more than 30 MVA, depending on the conductor size and the distribution voltage level.
Normally between two and 12 feeders emanate from any one substation, in what has been
called a dendritic configuration — repeated branching into smaller branches as the feeder
moves out from the substation toward the customers. In combination, all the feeders in a

14 Chapter 1

power system constitute the feeder system (Figure 1.6). An average substation has between
two and eight feeders, and can vary between one and forty.
The main, three-phase trunk of a feeder is called the primary trunk and may branch into
several main routes, as shown in the diagram. These main branches end at open points
where the feeder meets the ends of other feeders - points at which a normally open switch
serves as an emergency tie between two feeders.
In addition, each feeder will be divided, by normally closed switches, into several
switchable elements. During emergencies, segments can be re-switched to isolate damaged
sections and route power around outaged equipment to customers who would otherwise
have to remain out of service until repairs were made.
By definition, the feeder consists of all primary voltage level segments between the
substations and an open point (switch). Any part of the distribution level voltage lines -
three-phase, two-phase, or single-phase - that is switchable is considered part of the
primary feeder. The primary trunks and switchable segments are usually built using three
phases, with the largest size of distribution conductor (typically this is about 500-600 MCM
conductor, but conductor over 1,000 MCM is not uncommon, and the author has designed
and built feeders for special situations with up to 2,000 MCM conductor) justified for
reasons other than maximum capacity (e.g., contingency switching). Often a feeder has
excess capacity because it needs to provide back-up for other feeders during emergencies.
The vast majority of distribution feeders worldwide and within the United States are
overhead construction, wooden pole with wooden crossarm or post insulator. Only in dense
urban areas, or in situations where esthetics are particularly important, can the higher cost
of underground construction be justified. In this case, the primary feeder is built from
insulated cable, which is pulled through concrete ducts that are first buried in the ground.
Underground feeder costs from three to ten times what overhead does.
Many times, however, the first several hundred yards of an overhead primary feeder are
built underground even if the system is overhead. This underground portion is used as the
feeder getaway. Particularly at large substations, the underground getaway is dictated by
practical necessity, as well as by reliability and esthetics. At a large substation, ten or 12

fj substation — primary trunk
Three miles
• closed switch — lateral branches
o open switch

Figure 1.6 Distribution feeders route power away from the substation, as shown (in idealized form -
configuration is never so evenly symmetric in the real world) for two substations. Positions of
switches make the system electrically radial, while parts of it are physically a network. Shown here are
two substations, each with four feeders.


three-phase, overhead feeders leaving the substation mean from 40 to 48 wires hanging in
mid-air around the substation site, with each feeder needing the proper spacings for
electrical insulation, safety, and maintenance. At a large-capacity substation in a tight
location, there is simply not enough overhead space for so many feeders. Even if there is,
the resulting tangle of wires looks unsightly and, perhaps most important, is potentially
unreliable - one broken wire falling in the wrong place can disable a lot of power delivery
capability.
The solution to this dilemma is the underground feeder getaway, usually consisting of
several hundred yards of buried, ducted cable that takes the feeder out to a riser pole, where
it is routed above ground and connected to overhead wires. Very often, this initial
underground link sets the capacity limit for the entire feeder - the underground cable
ampacity is the limiting factor for the feeder's power transmission.
The Lateral Level
Laterals, short stubs or line segments that branch off the primary feeder, represent the final
primary voltage part of the power's journey from the substation to the customer. A lateral is
directly connected to the primary trunk and operates at the same nominal voltage. A series
of laterals tap off the primary feeder as it passes through a community, each lateral routing
power to a few dozen homes.
Normally, laterals do not have branches, and many laterals are only one- or two-phase -
all three phases are used only if a relatively substantial amount of power is required, or if
three-phase service must be provided to some of the customers. Normally, single- and two-
phase laterals are arranged to tap alternately different phases on the primary feeder, as
shown below, in an attempt by the distribution planning engineer to balance the loads as
closely as possible.
Typically, laterals deliver from as little as 10 kVA for a small single-phase lateral to as
much as 2 MVA. In general, even the largest laterals use small conductors (relative to the
primary size). When a lateral needs to deliver a great deal of power, the planner will
normally use all three phases, with a relatively small conductor for each, rather than employ
a single-phase and use a large conductor. This approach avoids creating a significant
imbalance in loading at the point where the lateral taps into the primary feeder. Power flow,
loadings, and voltage are maintained in a more balanced state if the power demands of a
"large lateral" are distributed over all three phases.
Laterals (wooden poles) are built overhead or underground. Unlike primary feeders and
transmission lines, single-phase laterals are sometimes buried directly. In this case, the
cable is placed inside a plastic sheath (that looks and feels much like a vacuum cleaner
hose), a trench is dug, and the sheathed cable is unrolled into the trench and buried. Directly
buried laterals are no more expensive than underground construction in many cases.
The Service Transformers
Service transformers lower voltage from the primary voltage to the utilization or customer
voltage, normally 120/240 volt two-leg service in most power systems throughout North
America. In overhead construction, service transformers are pole mounted and single-
phase, between 5 kVA and 166 kVA capacity. There may be several hundred scattered
along the trunk and laterals of any given feeder; since power can travel efficiently only up
to about 200 feet at utilization voltage, there must be at least one service transformer
located reasonably close to every customer.
Passing through these transformers, power is lowered in voltage once again, to the final
utilization voltage (120/240 volts in the United States) and routed onto the secondary
system or directly to the customers. In cases where the system is supplying power to large

16 Chapter 1

commercial or industrial customers, or the customer requires three-phase power, between
two and three transformers may be located together in a transformer bank and
interconnected in such a way as to provide multi-phase power. Several different connection
schemes are possible for varying situations.
Underground service, as opposed to overhead pole-mounted service, is provided by
padmount or vault type service transformers. The concept is identical to overhead
construction, with the transformer and its associated equipment changed to accommodate
incoming and outgoing lines that are underground.
The Secondary and Service Level
Secondary circuits, fed by the service transformers, route power at utilization voltage within
very close proximity to the customer, usually in an arrangement in which each transformer
serves a small radial network of utilization voltage secondary and service lines, which lead
directly to the meters of customers in the immediate vicinity.
At most utilities, the layout and design of the secondary level is handled through a set of
standardized guidelines and tables, which are used by engineering technicians and clerks to
produce work orders for the utilization voltage level equipment. In the United States, the
vast majority of this system is single-phase. In European systems, much of the secondary is
three-phase, particularly in urban and suburban areas.
What Is Transmission and What Is Distribution?
Definitions and nomenclature defining "transmission" and "distribution" vary greatly
among different countries, companies, and power systems. Traditionally, three types of
distinction between the two were made:
By voltage class: transmission is anything above 34.5 kV; distribution is
anything below that.
By function: distribution includes all utilization voltage equipment, plus all
lines that feed power to service transformers.
By configuration: transmission includes a network; distribution is all the
radial equipment in the system.
Generally, all three definitions applied simultaneously, since in most utility systems, any
transmission above 34.5 kV was configured as a network, and did not feed service
transformers directly, while all distribution was radial, built of only 34.5 kV or below, and
did feed service transformers. Substations - the meeting places of transmission lines
(incoming) and distribution lines (outgoing) - were often included in one or the other
category, but were sometimes considered as separate entities.
Today, the terms are evolving to somewhat different meanings. 'Transmission" is
becoming synonymous with "wholesale level grid" while "distribution" means the "retail"
or "service to native load" level. Chapter 30, section 2, provides a discussion of modern
interpretations and their use.

1.6 UTILITY DISTRIBUTION EQUIPMENT
The preceding section made it clear that a power delivery system is a very complex entity,
composed of thousands, perhaps even millions, of components which function together as a
T&D system. Each unit of equipment has only a small part to play in the system, and is only
a small part of the cost, yet each is critical for satisfactory service to at least one or more
customers, or it would not be included in the system.


T&D system planning is complex because each unit of equipment influences the
electrical behavior of its neighbors, and must be designed to function well in conjunction
with the rest of the system under a variety of different conditions, regardless of shifts in the
normal pattern of loads or the status of equipment nearby. While the modeling and analysis
of a T&D system can present a significant challenge, individually its components are
relatively simple to understand, engineer, and plan. In essence, there are only two major
types of equipment that perform the power delivery function:
• transmission and distribution lines, which move power from one
location to another
• transformers, which change the voltage level of the power
Added to these three basic equipment types are two categories of equipment used for
a very good reason:
• protective equipment, which provides safety and "fail safe" operation
• voltage regulation equipment, which is used to maintain voltage within an
acceptable range as the load changes. This monitoring and control
equipment is used to measure equipment and system performance and feed
this information to control systems so that the utility knows what the system
is doing and can control it, for both safety and efficiency reasons.

Transmission and Distribution Lines

By far the most omnipresent part of the power distribution system is the portion devoted to
actually moving the power flow from one point to another. Transmission lines, sub-
transmission lines, feeders, laterals, secondary and service drops all consist of electrical
conductors, suitably protected by isolation (transmission towers, insulator strings, and
insulated wrappings) from voltage leakage and ground contact. It is this conductor that
carries the power from one location to another.
Electrical conductors are available in various capacity ranges, with capacity generally
corresponding to the metal cross section (other things being equal, a thicker wire carries
more power). Conductors can be all steel (rare, but used in some locations where winter ice
and wind loadings are quite severe), all aluminum, copper, or a mixture of aluminum and
steel. Underground transmission can use various types of high-voltage cable. Capacity of a
line depends on the current-carrying capacity of the conductor or the 'cable, the voltage, the
number of phases, and constraints imposed by the line's location in the system.
The most economical method of handling a conductor is to place it overhead, supported
by insulators, on wooden poles or metal towers, suitably clear of interference or contact
with persons or property. However, underground construction, while generally more costly,
avoids esthetic intrusion of the line and provides some measure of protection from weather
(it also tends to reduce the capacity of a line slightly due to the differences between
underground cable and overhead conductor). Suitably wrapped with insulating material in
the form of underground cable, the cable is placed inside concrete or metal ducts or
surrounded in a plastic sheath.
Transmission/sub-transmission lines are always three-phase - three separate conductors
for the alternating current - sometimes with a fourth neutral (unenergized) wire. Voltage is
measured between phases - a 12.47 kV distribution feeder has an alternating current
voltage (RMS) of 12,470 volts as measured between any two phases. Voltage between any
phase and ground is 7,200 volts (12.47 divided by the square root of three). Major portions

18 Chapter 1

of a distribution system - trunk feeders - are as a rule built as three-phase lines, but lower-
capacity portions may be built as either two-phase, or single-phase.7
Regardless of type or capacity, every electrical conductor has an impedance (a
resistance to electrical flow through it) that causes voltage drop and electrical losses
whenever it is carrying electric power. Voltage drop is a reduction in the voltage between
the sending and receiving ends of the power flow. Losses are a reduction in the net power,
and are proportional to the square of the power. Double the load and the losses increase by
four. Thus, 100 kilowatts at 120 volts might go in one end of a conductor, only to emerge at
the other as 90 kilowatts at 114 volts at the other end. Both voltage drop and losses vary in
direct relation to load - within very fine limits if there is no load, there are no losses or
voltage drop. Voltage drop is proportional to load - double the load and voltage drop
doubles. Losses are quadratic, however - double the load and losses quadruple.

Transformers

At the heart of any alternating power system are transformers. They change the voltage and
current levels of the power flow, maintaining (except for a very small portion of electrical
losses) the same overall power flow. If voltage is reduced by a factor of ten from high to
low side, then current is multiplied by ten, so that their overall product (voltage times
current equals power) is constant in and out.
Transformers are available in a diverse range of types, sizes, and capacities. They are
used within power systems in four major areas: at power plants, where power which is
minimally generated at about 20,000 volts is raised to transmission voltage (100,000 volts
or higher); at switching stations, where transmission voltage is changed (e.g., from 345,000
volts to 138,000 volts before splitting onto lower voltage transmission lines); at distribution
substations, where incoming transmission-level voltage is reduced to distribution voltage
for distribution (e.g., 138 kV to 12.47 kV); and at service transformers, where power is
reduced in voltage from the primary feeder voltage to utilization level (12.47 kV to 120/240
volts) for routing into customers' homes and businesses.
Larger transformers are generally built as three-phase units, in which they
simultaneously transform all three phases. Often these larger units are built to custom or
special specifications, and can be quite expensive - over $3,000,000 per unit, in some cases.
Smaller transformers, particularly most service transformers, are single-phase - it takes
three installed side by side to handle a full three-phase line's power flow. They are
generally built to standard specifications and bought in quantity.
Transformers experience two types of electrical losses - no-load losses (often called
core, or iron, losses) and load-related losses. No-load losses are electrical losses inherent in
operating the transformer - due to its creation of a magnetic field inside its core - and occur
simply because the transformer is connected to an electrical power source. They are
constant, regardless of whether the power flowing through the transformer is small or large.
Core losses are typically less than one percent of the nameplate rating. Only when the
transformer is seriously overloaded, to a point well past its design range, will the core losses
change (due to magnetic saturation of the core).
Load-related losses are due to the current flow through the transformer's impedance and
correspond very directly with the level of power flow - like those of conductors and cables
they are proportional to current squared, quadrupling whenever power flow doubles. The
result of both types of losses is that a transformer's losses vary as the power transmitted
through it varies, but always at or above a minimum level set by the no-load losses.

7
In most cases, a single-phase feeder or lateral has two conductors: the phase conductor and the neutral.


Switches

Occasionally, it is desirable to be able to vary the connection of line segments within a
power delivery system, particularly in the distribution feeders. Switches are placed at
strategic locations so that the connection between two segments can be opened or closed.
Switches are planned to be normally closed (NC) or normally open (NO), as was shown in
Figure 1.6.
Switches vary in their rating (how much current they can vary) and their load break
capacity (how much current they can interrupt, or switch off), with larger switches being
capable of opening a larger current. They can be manually, automatically, or remotely
controlled in their operation.

Protection and Protective Equipment

When electrical equipment fails, for example if a line is knocked down during a storm, the
normal function of the electrical equipment is interrupted. Protective equipment is designed
to detect these conditions and isolate the damaged equipment, even if this means
interrupting the flow of power to some customers. Circuit breakers, sectionalizers, and
fused disconnects, along with control relays and sensing equipment, are used to detect
unusual conditions and interrupt the power flow whenever a failure, fault, or other
unwanted condition occurs on the system.
These devices and the protection engineering required to apply them properly to the
power system are not the domain of the utility planners and will not be discussed here.
Protection is vitally important, but the planner is sufficiently involved with protection if he
or she produces a system design that can be protected within standards, and if the cost of
that protection has been taken into account in the budgeting and planning process. Both of
these considerations are non-trivial.
Protection puts certain constraints on equipment size and layout - for example in some
cases a very large conductor is too large (because it would permit too high a short circuit
current) to be protected safely by available equipment and cannot be used. In other cases,
long feeders are too long to be protected (because they have too low a short circuit current
at the far end). A good deal of protective equipment is quite complex, containing sensitive
electro-mechanical parts (many of which move at high speeds and in a split-second manner)
and depending on precise calibration and assembly for proper function. As a result, the cost
of protective equipment and control, and the cost of its maintenance, is often significant -
differences in protection cost can make the deciding difference between two plans.
Voltage Regulation
Voltage regulation equipment includes line regulators and line drop compensators, as well
as tap changing transformers. These devices vary their turns ratio (ratio of voltage in to
voltage out) to react to variations in voltage drop - if voltage drops, they raise the voltage;
if voltage rises, they reduce it to compensate. Properly used, they can help maintain voltage
fluctuation on the system within acceptable limits, but they can only reduce the range of
fluctuation, not eliminate it altogether.
Capacitors
Capacitors are a type of voltage regulation equipment. By correcting power factor they can
improve voltage under many heavy loads (hence large voltage drop) cases. Power factor is
a measure of how well voltage and current in an alternating system are in step with one
another. In a perfect system, voltage and current would alternately cycle in conjunction with

20 Chapter 1

one another — reaching a peak, then reaching a minimum, at precisely the same times. But
on distribution systems, particularly under heavy load conditions, current and voltage fall
out of phase - both continue to alternate 60 times a second, but during each cycle voltage
may reach its peak slightly ahead of current - there is a slight lag of current versus voltage,
as shown in Figure 1.7.
It is the precise, simultaneous peaking of both voltage and current that delivers
maximum power. If out of phase, even by a slight amount, effective power drops, as does
power factor - the ratio of real (effective) power to the maximum possible power (if voltage
and current were locked in step).
Power engineers refer to a quantity called VAR (Volt-Amp Reactive) that is caused by
this condition. Basically, as power factors worsen (as voltage and current fall farther apart
in terms of phase angle) a larger percent of the electrical flow is VARs, and a smaller part is
real power. The frustrating thing is that the voltage is still there, and the current is still there,
but because of the shift in their timing, they produce only VARs, not power. The worse the
power factor, the higher the VAR content. Poor power factor creates considerable cost and
performance consequences for the power system: large conductor is still required to carry
the full level of current even though power delivery has dropped, and because current is
high, the voltage drop is high, too, further degrading quality of service.
Unless one has worked for some time with the complex variable mathematics associated
with AC power flow analysis, VARs are difficult to picture. A useful analogy is to think of
VARs as "electrical foam." If one tried to pump a highly carbonated soft drink through a
system of pipes, turbulence in the pipes, particularly in times of high demand (high flow)
would create foam. The foam would take up room in the pipes, but contribute little of value
to the net flow - the equivalent of VARs in an electrical system.
Poor power factor has several causes. Certain types of loads create VARs - in simple
terms loads which cause a delay in the current with respect to voltage as it flows through
them. Among the worst offenders are induction motors, particularly small ones as almost
universally used for blowers, air conditioning compressors, and the powering of conveyor
belts and similar machinery. Under heavy load conditions, voltage and current can get out
of phase to the point that power factor can drop below 70%. In addition, transmission
equipment itself can often create this lag and "generate" a low power factor.

VOLTAGE VOLTAGE

}Time f f-> Time

> Time i 1
- 4 * Time

CURRENT CURRENT

Figure 1.7 Current and voltage in phase deliver maximum power (left). If current and voltage fall out
of phase (right), actual power delivered drops by very noticeable amounts - the power factor falls.


Capacitors correct the poor power factor. They "inject" VARs into a T&D line to bring
power factor close to 1.0, transforming VAR flow back into real power flow, regaining the
portion of capacity lost to poor power factor. Capacitors can involve considerable cost
depending on location and type. They tend to do the most good if put on the distribution
system, near the customers, but they cost a great deal more in those locations than if
installed at substations.

1.7 T&D COSTS

A T&D system can be expensive to design, build, and operate. Equipment at every level
incurs two types of costs. Capital costs include the equipment and land, labor for site
preparation, construction, assembly and installation, and any other costs associated with
building and putting the equipment into operation. Operating costs include labor and
equipment for operation, maintenance and service, taxes and fees, as well as the value of the
power lost to electrical losses. Usually, capital cost is a one-time cost (once it's built, the
money's been spent). Operating costs are continuous or periodic.
Electrical losses vary depending on load and conditions. While these losses are small by
comparison to the overall power being distributed (seldom more than 8%), they constitute a
very real cost to the utility, and the present worth of the lifetime losses through a major
system component such as a feeder or transformer can be a significant factor impacting its
design and specification, often more than the original capital cost of the unit. Frequently, a
more costly type of transformer will be selected for a certain application because its design
leads to an overall savings due to lower losses, or a larger capacity line (larger conductor)
will be used than really needed due to capacity requirements, purely because the larger
conductor will incur lower losses costs.
Cumulatively, the T&D system represents a considerable expense. While a few
transmission lines and switching stations are composed of large, expensive, and purpose-
designed equipment, the great portion of the sub-transmission-substation-distribution
system is built from "small stuff' - commodity equipment bought mostly "off the shelf to
standard designs. Individually inexpensive, they amount to a significant cost when added
together.
Transmission Costs
Transmission line costs are based on a per mile cost and a termination cost at either end of
the line associated with the substation at which it is terminated. Costs can run from as low
as $50,000/mile for a 46 kV wooden pole sub-transmission line with perhaps 50 MVA
capacity ($1 per kVA-mile) to over $1,000,000 per mile for a 500 kV double circuit
construction with 2,000 MVA capacity ($.5/kVA-mile).

Substation Costs
Substation costs include all the equipment and labor required to build a substation,
including the cost of land and easements/ROW. For planning purposes, substations can be
thought of as having four costs:

1. Site cost - the cost of buying the site and preparing it for a substation.

2. Transmission cost - the cost of terminating the incoming sub-transmission
lines at the site.

22 Chapter 1

3. Transformer cost - the transformer and all metering, control, oil spill
containment, fire prevention, cooling, noise abatement, and other
transformer related equipment, along with typical buswork, switches,
metering, relaying, and breakers associated with this type of transformer,
and their installation.

4. Feeder buswork/getaway costs - the cost of beginning distribution at the
substation, includes getting feeders out of the substation.

Often, as an expedient in planning, estimated costs of feeder buswork and getaways are
folded into the transformer costs. The feeders to route power out of the substation are
needed in conjunction with each transformer, and in direct proportion to the transformer
capacity installed, so that their cost is sometimes considered together with the transformer
as a single unit. Regardless, the transmission, transformer, and feeder costs can be estimated
fairly accurately for planning purposes.
Cost of land is another matter entirely. Site and easements or ROW into a site have a
cost that is a function of local land prices, which vary greatly, depending on location and
real-estate markets. Site preparation includes the cost of preparing the site (grading,
grounding mat, foundations, buried ductwork, control building, lighting, fence, landscaping,
and access road).
Substation costs vary greatly depending on type, capacity, local land prices, and other
variable circumstances. In rural settings where load density is quite low and minimal
capacity is required, a substation may involve a site of only several thousand square feet of
fenced area, a single incoming transmission line (69 kV), one 5 MVA transformer, fusing
for all fault protection, and all "buswork" built with wood poles and conductor, for a total
cost of perhaps no more than $90,000. The substation would be applied to serve a load of
perhaps 4 MW, for a cost of $23/kW. This substation in conjunction with the system around
it would probably provide service with about ten hours of service interruptions per year
under average conditions.
However, a typical substation built in most suburban and urban settings would be fed by
two incoming 138 kV lines feeding two 40 MVA, 138 kV to 12.47 kV transformers, each
feeding a separate low side (12.41 kV) bus, each bus with four outgoing distribution feeders
of 9 MVA peak capacity, and a total cost of perhaps $2,000,000. Such a substation's cost
could vary from between about $1.5 million and $6 million, depending on land costs, labor
costs, the utility equipment and installation standards, and other special circumstances. In
most traditional vertically integrated, publicly regulated electric utilities, this substation
would have been used to serve a peak load of about 60 MVA (75% utilization of capacity),
which at its nominal $2,000,000 cost works out to $33/kW. In a competitive industry, with
tighter design margins and proper engineering measures taken beforehand, this could be
pushed to a peak loading of 80 MVA (100% utilization, $25/kW). This substation and the
system around it would probably provide service with about two to three hours of service
interruptions per year under average conditions.

Feeder System Costs

The feeder system consists of all the primary distribution lines, including three-phase trunks
and their lateral extensions. These lines operate at the primary distribution voltage - 23.9
kV, 13.8 kV, 12.47 kV, 4.16 kV or whatever - and may be three-, two-, or single-phase
construction as required. Typically, the feeder system is also considered to include voltage
regulators, capacitors, voltage boosters, sectionalizers, switches, cutouts, fuses, any intertie


transformers (required to connect feeders of different voltage at tie points, as, for example,
23.9 and 12.47 kV) that are installed on the feeders (i.e., not at the substations or at
customer facilities).
As a rule of thumb, construction of three-phase overhead, wooden pole crossarm type
feeders of normal, large conductor (about 600 MCM per phase) at a medium distribution
primary voltage (e.g., 12.47 kV) costs about $150,000/mile. However, cost can vary greatly
due to variations in labor, filing and permit costs among utilities, as well as differences in
design standards and terrain. Where a thick base of topsoil is present, a pole can be installed
by simply auguring a hole for the pole. In areas where there is rock close under the surface,
holes have to be jack-hammered or blasted, and cost goes up accordingly. It is generally
less expensive to build feeders in rural areas than in suburban or urban areas. Thus, while
$150,000 is a good average cost, a mile of new feeder construction could cost as little as
$55,000 in some situations and as much as $500,000 in others.
A typical distribution feeder (three-phase, 12.47 kV, 600 MCM/phase) would be rated at
a thermal (maximum) capacity of about 15 MVA and a recommended economic (design)
peak loading of about 8.5 MVA peak, depending on losses and other costs. At
$150,000/mile, this capacity rating gives somewhere between $10 to $15 per kW-mile as
the cost for basic distribution line. Underground construction of three-phase primary is
more expensive, requiring buried ductwork and cable, and usually works out to a range of
$30 to $50 per kW-mile.
Lateral lines - short primary-voltage lines working off the main three-phase circuit - are
often single- or two-phase and consequently have lower costs but lower capacities.
Generally, they are about $5 to $15 per kW-mile overhead, with underground costs of
between $5 to $15 per kW-mile (direct buried) to $30 to $100 per kW-mile (ducted).
Cost of other distribution equipment, including regulators, capacitor banks and their
switches, sectionalizers, line switches, etc., varies greatly depending on the specifics of each
application. In general, the cost of the distribution system will vary from between $10 and
$30 per kW-mile.

Service Level Costs

The service, or secondary, system consists of the service transformers that convert primary
to utilization voltage, the secondary circuits that operate at utilization voltage, and the
service drops that feed power directly to each customer. Without exception these are very
local facilities, meant to move power no more than a few hundred feet at the very most and
deliver it to the customer "ready to use."
Many electric utilities develop cost estimates for this equipment on a per-customer
basis. A typical service configuration might involve a 50 MVA pole-mounted service
transformer feeding ten homes, as shown in Figure 1.8. Costs for this equipment might
include:

Heavier pole & hardware for transformer application $250
50 kW transformer, mounting equipment, and installation $750
500 feet secondary (120/240 volt) single-phase @ $2/ft. $ 1,000
10 service drops including installation at $ 100 $1,000
$3,000

This results in a cost of about $300 per customer, or about $60/kW of coincident load.

24 Chapter 1

lateral (primary) secondary

customer service transformer

Figure 1.8 Here, a service transformer, fed from a distribution primary-voltage lateral, feeds in turn
ten homes through secondary circuit operating at utilization voltage.

Maintenance and Operating Costs

Once put into service, T&D equipment must be maintained in sound, operating function,
hopefully in the manner intended and recommended by the manufacturer. This will require
periodic inspection and service, and may require repair due to damage from storms or other
contingencies. In addition, many utilities must pay taxes or fees for equipment (T&D
facilities are like any other business property). Operating, maintenance, and taxes are a
continuing annual expense.
It is very difficult to give any generalization of O&M&T costs, partly because they vary
so greatly from one utility to another, but mostly because utilities account for and report
them in very different ways. Frankly, the author has never been able to gather a large
number of comparable data sets from which to produce even a qualitative estimate of
average O&M&T costs.8 With that caveat, a general rule of thumb: O&M&T costs for a
power delivery system probably run between 1/8 and 1/30 of the capital cost, annually.

The Cost to Upgrade Exceeds the Cost to Build

One of the fundamental factors affecting design of T&D systems is that it costs more per
kilowatt to upgrade equipment to a higher capacity than to build to that capacity in the
original construction. For example, a 12.47 kV overhead, three-phase feeder with a 9 MW
capacity (336 MCM phase conductor) might cost $120,000/mile to build ($13.33 per kW-
mile). Building it with 600 MCM conductor instead, for a capacity of 15 MVA, would cost
in the neighborhood of $150,000 ($10/kW-mile).
However, upgrading an existing 336 MCM, 9 MW capacity line to 600 MCM, 15 MVA
capacity could cost $200,000/mile - over $30 per kW-mile for the 6 MW of additional
capacity. This is more expensive because it entails removing the old conductor and
8
For example, some utilities include part of O&M expenses in overhead costs; others do not. A few
report all repairs (including storm damage) as part of O&M; others accumulate major repair work
separately. Still others report certain parts of routine service (periodic rebuilding of breakers) as a
type of capital cost because it extends equipment life or augments capacity; others report all such
work as O&M, even when the rebuilding upgrades capacity or voltage class.


installing new conductor along with brackets, crossarms, and other hardware required to
support the heavier new conductor. Typically, this work is done hot (i.e., with the feeder
energized), which means the work must be undertaken with extreme care and following a
number of safety-related restrictions on equipment and labor.
Thus, T&D planners have an incentive to look at their long-term needs carefully and to
"overbuild" against initial requirements if growth trends show eventual demand will be
higher. The cost of doing so must be weighed against long-term savings, but often T&D
facilities are built with considerable margin (50%) above existing load to allow for future
load growth.
The very high cost per kW for upgrading a T&D system in place creates one of the best
perceived opportunities for DSM and DG reduction. Note that the capital cost/kW for the
upgrade capacity in the example above ($33/kW) is nearly three times the cost of similar
new capacity. Thus, planners often look at areas of the system where slow, continuing load
growth has increased load to the point that local delivery facilities are considerably taxed as
areas where DSM and DG can deliver significant savings.
In some cases, distributed resources can reduce or defer significantly the need for T&D
upgrades of the type described above. However, this does not assure a significant savings,
for the situation is more complicated than an analysis of capital costs to upgrade may
indicate. If the existing system (e.g., the 9 MW feeder) needs to be upgraded, then it is
without a doubt highly loaded, which means its losses may be high, even off-peak. The
upgrade to a 600 MCM conductor will cut losses 8,760 hours per year. Losses cost may
drop by a significant amount, enough in many cases to justify the cost of the upgrade alone.
The higher the annual load factor in an area, the more likely this is to occur, but it is often
the case even when load factor is only 40%. However, DSM and in some cases DG also
lower losses, making the comparison quite involved, as will be discussed later in this book.

Electrical Losses Costs

Movement of power through any electrical device, be it a conductor, transformer, regulator,
or whatever, incurs a certain amount of electrical loss due to the impedance (resistance to
the flow of electricity) of the device. These losses are a result of inviolable laws of nature.
They can be measured, assessed, and minimized through proper engineering, but never
eliminated completely.
Losses are an operating cost

Although losses do create a cost (sometimes a considerable one) it is not always desirable to
reduce them as much as possible. Perhaps the best way to put them in proper perspective is
to think of T&D equipment as powered by electricity - the system that moves power from
one location to another runs on electric energy itself. Seen in this light, losses are revealed
as what they are — a necessary operating expense to be controlled and balanced against
other costs.
Consider a municipal water department, which uses electric energy to power the pumps
that drive the water through the pipes to its customers. Electricity is an acknowledged
operating cost, one accounted for in planning and weighed carefully in designing the system
and estimating its costs. The water department could choose to buy highly efficient pump
motors, ones that command a premium price over standard designs but provide a savings in
reduced electric power costs, and to use piping that is coated with a friction-reducing lining
to promote rapid flow of water (thus carrying more water with less pump power), all toward
reducing its electric energy cost. Alternatively, after weighing the cost of this premium

26 Chapter 1

equipment against the energy cost savings it provides, the water department may decide to
use inexpensive motors and piping and simply pay more over the long run. The point is that
the electric power required to move the water is viewed merely as one more cost that had to
be included in determining what is the lowest "overall" cost.
It is the same for power - takes power to move power. Since electric delivery equipment
is powered by its own delivery product, this point often is lost and losses are viewed
entirely as a negative factor - a failure on the part of the system and its planners. However,
losses are the energy required to power the system in its operation, and therefore just one
other factor to be engineered among many other factors, and their costs, that can be traded
against one another in the optimization of its design. In order to do its job of delivering
electricity, a T&D system must be provided with power itself, just like the water
distribution system. Energy must be expended to move the product. Thus, a transformer
consumes a small portion of the power fed into it. In order to move power 50 miles, a 138
kV transmission line similarly consumes a small part of the power given to it.
Initial cost of equipment can always be traded against long-term losses costs. Highly
efficient transformers can be purchased to use considerably less power to perform their
function than standard designs. Larger conductors can be used in any transmission or
distribution line, which will lower impedance and thus losses for any level of power
delivery. But both examples here cost more money initially - the efficient transformer may
cost three times what a standard design does; the larger conductor might entail a need for
not only large wire, but heavier hardware to hold it in place and stronger towers and poles
to keep it in the air. In addition, these changes may produce other costs - for example, use
of larger conductor not only lowers losses, but a higher fault duty (short circuit current),
which increases the required rating and cost for circuit breakers. Regardless, initial
equipment costs can be balanced against long-term losses costs through careful study of
needs, performance, and costs to establish a minimum overall (present worth) cost.

Load-related losses

Flow of electric power through any device is accompanied by what are called load-related
losses, which increase as the power flow (load) increases. These are due to the impedance
of the conductor or device. Losses increase as the square of the load - doubling the power
flowing through a device quadruples the losses. Tripling power flow increases the losses by
nine.
With very few exceptions, larger electrical equipment always has a lower impedance,
and thus lower load-related losses, for any given level of power delivery. Hence, if the
losses inherent in delivering 5 MW using 600 MCM conductor are unacceptably large, the
use of 900 MCM conductor will reduce them considerably. The cost of the larger conductor
can be weighed against the savings in reduced losses to decide if it is a sound economic
decision.

No-load losses

"Wound" T&D equipment - transformers and regulators - have load-related losses as do
transmission lines and feeders. But they also have a type of electrical loss that is constant,
not a function of loading. No-load losses constitute the electric power required to establish a
magnetic field inside these units, without which they would not function. Regardless of
whether a transformer has any load - any power passing through it at all - it will consume a
small amount of power, generally less than 1 % of its rated full power, simply because it is
energized and "ready to work." No-load losses are constant, 8,760 hours per year.


Given similar designs, a transformer will have no-load losses proportional to its capacity
- a 10 MVA substation transformer will have twice the no-load losses of a 5 MVA
transformer of similar voltage class and design type. Therefore, unlike the situation with a
conductor, selection of a larger transformer does not always reduce losses, because while
the larger transformer will always have lower load-related losses, it will have higher no-
load losses.
Again, low-loss transformers are available, but cost more than standard types. Lower-
cost-than-standard, but higher-loss, transformers are also available, and often are a good
investment for back-up and non-continuous use applications.

The costs of losses

The electric power required to operate the T&D system - the electrical losses — is typically
viewed as having two costs, demand and energy. Demand cost is the cost of providing the
peak capacity to generate and deliver power to the T&D equipment. A T&D system that
delivers 1,250 MW at peak might have losses during this peak of 100 MW. This means the
utility must have generation, or buy power at peak, to satisfy this demand, whose cost is
calculated using the utility's power production cost at time of peak load. This is usually
considerably above its average power production cost.
Demand cost also ought to include a considerable T&D portion of expense. Every
service transformer in the system (and there are many) is consuming a small amount of
power in doing its job at peak. Cumulatively, this might equal 25 MW of power - up to 1/4
of all losses in the system. That power must not only be generated by the utility but
transmitted over its transmission system, through its substations, and along its feeders to
reach the service transformers. Similarly, the power for electrical losses in the secondary
and service drops (while small, these are numerous and low voltage, so that their
cumulative contribution to losses is noticeable) has to be moved even farther, through the
service transformers and down to the secondary level.
Demand cost of losses is the total cost of the capacity to provide the losses and move
them to their points of consumption. Losses occur whenever the power system is in
operation, which generally means 8,760 hours per year. While losses vary as the square of
load, so they drop by a considerable margin off-peak. Their steady requirement every hour
of the year imposes a considerable energy demand over the course of a year. This cost is the
cost of the energy to power the losses. Example: Consider a typical 12.47 kV, three-phase,
OH feeder, with 15 MW capacity (600 MCM phase conductor), serving a load of 10 MW at
peak with 4.5% primary-level losses at peak (450 kW losses at peak), and having a load
factor of 64% annually. Given a levelized capacity cost of power delivered to the low side
bus of a substation of $10/kW, the demand cost of these losses is $4,500/year. Annual
energy cost, at 3.50 /kWh, can be estimated as:

450 kW losses at peak x 8,760 hours x (64% load factor)2 x 3.50 = $56,500

Thus, the losses' costs (demand plus energy costs) for this feeder are nearly $60,000
annually. At a present worth discount factor of around 11%, this means losses have an
estimated present worth of about $500,000. This computation used a simplification -
squaring the load factor to estimate load factor impact on losses - which tends to
underestimate losses' costs slightly. Actual losses costs probably would be more in the
neighborhood of $565,000 PW. If the peak load on this feeder were run up to its maximum
rating (about 15 MW instead of 10 MW) with a similar load factor of 64%, annual losses
cost would increase to (15/10)2 or $1,250,000 dollars. This roughly two-to-one cost ratio is

28 Chapter 1

Figure 1.9 Cost of power delivery varies depending on location. Shown here are the annual capacity
costs of delivery evaluated on a ten-acre basis throughout a coastal city of population 250,000. Cost
varies from a low of $85/kW to a high of $270/kW.

common for feeders in many utility systems, which are: (1) designed with a considerable
margin between expected peak load and their actual thermal (maximum continuous)
capacity, for reasons of reliability margin (so that there is capacity to pick up customers on
a nearby feeder should it go out of service), and (2) made large for economics reasons -
usually losses are too expensive at close to the thermal limit of any conductor. This typical
feeder would include perhaps four miles of primary trunk at $150,000/mile and thirty miles
of laterals at $50,000/mile, for a total capital cost of about $2,100,000. Thus, total losses
costs are on the order of magnitude of original cost of the feeder itself, and in cases where
loading is high can approach that cost. Similar loss-capital relations exist for all other levels
of the T&D system.

Total of T&D Costs

Table 1.4 shows the cost of providing service to a "typical" residential customer in an
example power system. These figures are representative of all systems, but costs, practices,
and accounting systems vary so that these are not general values applicable to all utilities.

Table 1.4 Cost of Providing Service to a Typical Residential Customer
Level Cost Components Cost
Transmission 4 kW x 100 miles x $.75/kW mile $300
Substation 4kWx$60/kW $240
Feeder 4 kW x 1.5 miles x $10/kW-mile $60
Service l/10th of 50 kVA local service system $300
Total Initial cost (Capital) $900
All Operations, Maintenance, and Taxes (PW next 30 years) $500
All Cost of electrical losses (PW next 30 years) $700
Estimated cost of power delivery, 30 years, PW $2,100


1.8 TYPES OF DISTRIBUTION SYSTEM DESIGN
There are three fundamentally different ways to lay out a power distribution system used by
electric utilities, each of which has variations in its own design. As shown in Figure 1.10,
radial, loop, and network systems differ in how the distribution feeders are arranged and
interconnected about a substation.
Radial Feeder Systems
Most power distribution systems are designed to be radial, to have only one path between
each customer and the substation. The power flows exclusively away from the substation
and out to the customer along a single path, which, if interrupted, results in complete loss of
power to the customer. Radial design is by far the most widely used form of distribution
design, accounting for over ninety-nine percent of all distribution construction in North
America. Its predominance is due to two overwhelming advantages: it is much less costly
than the other two alternatives and it is much simpler in planning, design, and operation.
In most radial plans, both the feeder and the secondary systems are designed and
operated radially. Each radial feeder serves a definite service area (all customers in that area
are provided power by only that feeder). Many radial feeder systems are laid out and
constructed as networks, but operated radially by opening switches at certain points
throughout the physical network configuration (shown earlier in Figure 1.6), so that the
resulting configuration is electrically radial. The planner determines the layout of the
network and the size of each feeder segment in that network and decides where the open
points should be for proper operation as a set of radial feeders.
A further attribute of many radial feeder system designs, although not essential, is the
use of single-phase laterals. Throughout North America, most utilities use single- and two-
phase laterals to deliver small amounts of power over short distances, rather than use all
three phases. These laterals are also radial, but seldom, if ever, end in a switch (they just
end). There are some utilities, particularly urban systems in Europe, Africa, and Asia, that
build every part of the radial distribution system, including laterals, with all three phases.
Each service transformer in these systems feeds power into a small radial system around
it, basically a single electrical path from each service transformer to the customers nearby.
Regardless of whether it uses single-phase laterals or not, the biggest advantages of the
radial system configuration, in addition to its lower cost, is the simplicity of analysis and

Radial Loop Network

Figure 1.10 Simplified illustration of the concepts behind three types of power distribution
configuration. Radial systems have only one electrical path from the substation to the customer, loop
systems have two, and networks have several. Arrows show most likely direction of electric flows.

30 Chapter 1

predictability of performance. Because there is only one path between each customer and
the substation, the direction of power flow is absolutely certain. Equally important, the load
on any element of the system can be determined in the most straightforward manner - by
simply adding up all the customer loads "downstream" from that piece of equipment.
Before the advent of economical and widely available computer analysis, this alone was
an overwhelming advantage, for it allowed simple, straightforward, "back of the envelope"
design procedures to be applied to the distribution system with confidence that the resulting
system would work well. The simplicity of analysis and confidence that operating behavior
is strictly predictable are still great advantages.
Because load and power flow direction are easy to establish, voltage profiles can be
determined with a good degree of accuracy without resorting to exotic calculation methods;
equipment capacity requirements can be ascertained exactly; fault levels can be predicted
with a reasonable degree of accuracy; and protective devices - breaker-relays and fuses -
can be coordinated in an absolutely assured manner, without resorting to network methods
of analysis. Regulators and capacitors can be sized, located, and set using relatively simple
procedures (simple compared to those required for similar applications to non-radial
designs, in which the power flow direction is not a given).
On the debit side, radial feeder systems are less reliable than loop or network systems
because there is only one path between the substation and the customer. Thus, if any
element along this path fails, a loss of power delivery results. Generally, when such a
failure occurs, a repair crew is dispatched to re-switch temporarily the radial pattern
network, transferring the interrupted customers onto another feeder, until the damaged
element can be repaired. This minimizes the period of outage, but an outage still occurred
because of the failure.
Despite this apparent flaw, radial distribution systems, if well designed and constructed,
generally provide very high levels of reliability. For all but the most densely populated
areas, or absolutely critical loads (hospitals, important municipal facilities, the utility's own
control center) the additional cost of an inherently more reliable configuration (loop or
network) cannot possibly be justified for the slight improvement that is gained over a well-
designed radial system.
Loop Feeder Systems
An alternative to purely radial feeder design is a loop system, which has two paths between
the power sources (substations, service transformers) and each customer. Such systems are
often called "European" because this configuration is the preferred design of many
European utilities. Equipment is sized and each loop is designed so that service can be
maintained regardless of where an open point might be on the loop. Because of this
requirement, whether operated radially (with one open point in each loop) or with closed
loops, the basic equipment capacity requirements of the loop feeder design do not change.
Some urban areas in Europe and Asia are fed by multiple hierarchical loop systems: a
100+ kV sub-transmission loop routes power to several substations, out of which several
loop feeders distribute power to service transformers, which each route powers through a
long loop secondary.
In terms of complexity, a loop feeder system is only slightly more complicated than a
radial system - power usually flows out from both sides toward the middle, and in all cases
can take only one of two routes. Voltage drop, sizing, and protection engineering are only
slightly more complicated than for radial systems.
But if designed thus, and if the protection (relay-breakers and sectionalizers) is also built
to proper design standards, the loop system is more reliable than radial systems. Service


will not be interrupted to the majority of customers whenever a segment is outaged, because
there is no "downstream" portion of any loop.
The major disadvantage of loop systems is a higher capacity cost than purely radial
distribution. A loop must be designed to meet all power and voltage drop requirements
when fed from either end. It needs extra capacity on each end, and the conductor must be
large enough to handle the power and voltage drop needs of the entire feeder if fed from
either end. This makes the loop system inherently more reliable than a radial system, but the
larger conductor and extra capacity increase its cost.
Distribution Networks
Distribution networks are the most complicated, most reliable, and in very rare cases also
the most economical method of distributing electric power. A network involves multiple
paths between all points in the network. Networks provide continuity of service (reliability)
far beyond that of radial and loop designs: if a failure occurs in one line, power instantly
and automatically re-routes itself through other pathways.
Most distribution networks are underground systems, simply because they are employed
mostly in high density areas, where overhead space is not available. Rarely is the primary
voltage level a network, because that proves very expensive and often will not work well.9
Instead, a "distribution network" almost always involves "interlaced" radial feeders and a
network secondary system - a grid of electrically strong (i.e., larger than needed to just feed
customers in the immediate area when everything is functioning) conductor connecting all
the customers together at utilization voltage. In this type of design, the secondary grid is fed
from radial feeders through service transformers, basically the same way secondary is fed in
radial or loop systems. The feeders are radial, but laid out in an interlaced manner - none
has a sole service area, but instead they overlap, two or more feeding alternate transformers
into a secondary network, as shown in Figure 1.11. While segments from two feeders
always run parallel in any part of the system, the same two feeders never overlap for all of
their routing. The essence of the interlaced system (and a design difficulty in any practical
plan) is to mix up feeders so that each feeder partially parallels quite a few other feeders.
Thus, if it fails, it spreads its load over quite a few other feeders (Figure 1.12).
At a minimum, distribution networks use an interlacing factor of two, meaning that two
feeders overlap in any one region, each feeding every other service transformer. But such a
system will fail when any two feeders are out of service. Interlacing factors as high as five
(five overlapping feeders, each feeding every fifth consecutive service transformer) have
been built. Such systems can tolerate the loss of any three feeders (the other two in any area
picking up the remaining load, although often very overloaded) without any interruption of
customer service. If an element fails, the power flow in the elements around it merely re-
distributes itself slightly. So slightly, in fact, that a real problem can be determining when
failures occur. If the damage is not visible (most networks are underground systems) and no
alarm or signal is given by monitoring equipment, the utility may not know a failure
occurred until months later, when a second failure nearby puts a strain on the system or
causes an outage.
Networks are more expensive than radial distribution systems, but not greatly so in
dense urban applications, where the load density is very high, where the distribution must
be placed underground, and where repairs and maintenance are difficult because of traffic

9
Particularly if a feeder network is created by networking feeders out of different substations, this
puts feeder paths in parallel with transmission between substations, which often results in
unacceptable loop and circular flows and large dynamic shifts in load on the distribution system.

32 Chapter 1

service transformers

Feeder A
Feeder B

Figure 1.11 To obtain an interlacing factor of 2, two feeders are routed down each street, with
alternating network transformers fed from each.

and congestion, networks may cost little more than loop systems. Networks require little
more conductor capacity than a loop system. The loop configuration required "double
capacity" everywhere to provide increased reliability. A distribution network is generally
no worse and often needs considerably less capacity and cost, if it is built to a clever design
and its required capacity margins are minimized.
Networks have one major disadvantage. They are much more complicated than other
forms of distribution, and thus much more difficult to analyze and operate. There is no
"downstream" side to each unit of equipment in a network or loop system. This complicates
load estimation, power flow analysis, and protection planning. It makes maintenance and
restoration more difficult in many cases. Loadings, power flow, and fault current and
protection must be determined by network techniques such as those used by transmission
planners.

NORMAL FEEDER #2 FAILED
• • • •

NORMAL FEEDER #2 FAILED

Figure 1.12 Top, a non-interlaced feeder system experiences the loss of one feeder, and all
transformers in the lower right part of the system are lost - service is certain to be interrupted. Bottom,
the same system, but interlaced. Loss of the feeder is a serious contingency, but can be withstood
because the feeder losses are distributed in such a way that each transformer out of service is
surrounded by transformers still in service.


However, more involved calculation methods than those applied to transmission may be
required, because a large distribution network can consist of 50,000 nodes or more - the
size of the very largest transmission-level power pool. Distribution network load flows are
often more difficult to solve than transmission systems because the range of impedances in
the modeled circuits is an order of magnitude wider.
In densely populated regions, such as the center of a large metropolitan area, networks
are not inherently more expensive than radial systems designed to serve the same loads.
Such concentrated load densities require a very large number of circuits anyway, so that
their arrangement in a network does not inherently increase the number of feeder and
secondary circuits, or their capacity requirements. It increases only the complexity of the
design. But in other areas, such as in most cities and towns, and in all rural areas, a network
configuration will call for some increase (in kVA-feet of installed conductor) over that
required for a radial or loop design. The excess capacity cost has to be justifiable on the
basis of reliability.
Networks and types of network designs are discussed further in Chapter 19, section 6.

Large-Trunk vs. Multi-Branch Feeder Layout

Most distribution systems are radial, and that approach is most appropriate for the vast
majority of situations. For that reason this book focuses primarily on radial system design.
Figure 1.13 illustrates two different ways to lay out a radial distribution system, and
illustrates a basic concept of distribution planning: flexibility of design. Each of the two
configurations can be engineered to work in nearly any situation. Each has advantages and
disadvantages in some situations as compared to the other, but neither is always superior to
the other in terms of reliability, cost, ease of protection, and service quality in all situations.
Most planning engineers have a preference for one or the other - in fact, about 20% of
utilities have standardized on the large-trunk design as their recommended guideline while
another 20% prefer the multi-branch approach. Beyond showing that there are significantly
different ways to lay out a distribution system, this brings to light an important point about
distribution design: major differences in standards exist among electric utilities; as a result
comparison of statistics or practice from one to the other is often not completely valid.
These concepts and practices are discussed in greater detail in Chapters 13 and 14.

:i in t i n [II
r™ ' T f i li lr [ 1 r J[ 1 1
,1 i i J11 i 1 1 I f D-

' 1 1[in[ I I I [II
Figure 1.13 Two ways to route a radial feeder to 108 service transformers. Left, a "multi-branch"
configuration. Right, a "large trunk" design. Either approach can always be made to do the job.
Neither is a lower-cost or greater-reliability design under all conditions, although many utilities
standardize or institutionalize practices around only one approach, thus losing some flexibility of
approach in minimizing costs and improving reliability.

34 Chapter 1

Ten Miles

Figure 1.14 A power system is divided by substation service boundaries into a set of substation
service areas, as shown.

Substation and Feeder Service Areas

As mentioned earlier, in most power systems, each substation is usually the sole provider of
electrical service to the region around it - its service area. Similarly, feeders and
distribution networks also have distinct service areas. Usually, the service area for a
substation, feeder, or other unit of equipment is the immediate area surrounding it, and
usually these service areas are contiguous (i.e. not broken into several parts) and exclusive -
no other similar distribution unit serves any of the load in an area. An example, Figure 1.14
shows a map of substation service areas for a rectangular portion of a power system. Each
distribution substation exclusively serves all customers in the area containing it.
Cumulatively, the customers in a substation's or feeder's service territory determine its
load, and their simultaneous peak demand defines the maximum power the substation must
serve. Within a power system, each individual part, such as a substation or service
transformer, will see its peak load at whatever time and in whatever season the customers in
its service area generate their cumulative peak demand. One result of this is that the peak
loads for different substations often occur at different seasons of the year or hours of the
day. But whenever the peak occurs, it defines the maximum power the unit is required to
deliver. Peak demand is one of the most important criteria in designing and planning
distribution systems. Usually it defines the required equipment capacity.
Dynamic Service Area Planning
By making switching changes in the distribution system, it is possible to expand or shrink a
substation or feeder's service area significantly, increasing or decreasing its net load, or
keeping is loading constant over time as the demand in a region gradually grows. Switching
to "balance loading" is an important element of T&D planning, as illustrated in Figure 1.15,
which shows a very typical T&D expansion situation. Two neighboring substations, A and
B, each have a peak load near the upper limit of their reliable load-handling range. Load is
growing slowly throughout the system, so that in each substation annual peak load is
increasing at about 1 MW per year. Under present conditions, both will need to be upgraded
soon. Approved transformer types, required to add capacity to each, are available only in 25
MVA or larger increments, costing $500,000 or more.


Before After

transferred load

Figure 1.15 Load in both substations is growing at about 1 MW per year. Each substation has
sufficient capacity to handle present load within contingency criteria (a 25% margin above peak) but
nothing more. By transferring load as shown, only one substation has to be reinforced with an
additional (25 MVA) capacity, yet both end up with sufficient margin for another ten years' growth.
Service area shifts like this are how expansion costs are kept down in spite of the fact that equipment
like transformers is available only in large, discrete sizes.

Both substations do not need to be reinforced. A new 25 MVA transformer and
associated equipment are added to substation A, increasing its ability to handle a peak load
by about 20 MVA. Ten MW of substation B's service area is then transferred to A. The
result is that each substation has 10 MW of margin for continued load growth - further
additions are not needed for 10 years.
This type of planned variation in service areas is a major tool used to keep T&D
expansion costs low, a key element in building a reliable, economical schedule of
expansion. Optimization of this particular aspect of planning can be a challenge, not
because of any inherent difficulty in analysis, but simply because there are so many parts of
the system and constraints to track at one time. This is one reason for the high degree of
computerization in distribution planning at many utilities. Balancing the myriad
requirements of many substations and their design and operating constraints is an ideal
problem for numerical optimization techniques.

1.9 THE SYSTEMS APPROACH AND TWO-Q PLANNING

One complication in determining the most economical equipment for a power system is that
its various levels — transmission, substation, and distribution - are interconnected, with the
distribution, in turn, connected to the customers. This means that the best size and
equipment type at each level and location in the system is a function not only of the local
load, but of the types of equipment selected for the other levels of the system nearby, their
locations and characteristics, as well as the loads they serve. Thus, in general, T&D
equipment is so interconnected that it is impossible to evaluate any one aspect of a system's
design without taking many others into account.
For example, consider the question of substation spacing - determining how far apart
substations should be, on average, for best utilization and economy. Within any utility
system, if substations are located farther apart, there will be fewer of them, reducing the

36 Chapter 1

cost of buying and preparing substation sites, as well as reducing the cost of building such a
large number of substations. However, with fewer substations, each substation must serve a
larger area of the system and will have a larger load, and thus require a larger capacity,
meaning it must have more or larger transformers. Overall, this does tend to reduce the
overall cost of the substation level, because there is an economy of scale to substations: one
100 MVA site is less costly than two 50 MVA sties, etc.
But that is not the end of the cost considerations; the aforementioned interrelationships
mean transmission impacts must be considered. Larger substations will also require a larger
amount of power to be brought to each one, which generally calls for a higher sub-
transmission voltage. Yet, there will be fewer sub-transmission lines required (because
there are fewer substations to which power must be delivered). All these aspects of layout
are related - greater substation spacing calls for larger substations with bigger transformers,
and a higher transmission voltage, but fewer lines are needed - and all together can create
better economies of scale if spacing is "optimized."
Furthermore, there is yet another set of interrelated impacts on the downstream side of
the substation. The feeder system is required to distribute each substation's power through
its service area, moving power out to the boundary between each substation's service area
and that of its neighbors. Moving substations farther apart means that the distribution
system must move power, on average, a greater distance. Distributing power over these
longer distances requires longer and more heavily loaded feeders. This in turn increases
voltage drop and can produce higher losses, all of which can increase cost considerably.
Employing a higher distribution voltage (such as 23.9 kV instead of 13.8 kV) improves
performance and economy, but regardless it costs more to distribute power from a few large
substations than from many smaller substations which will be closer together.
The major point: All of these aspects of system design are interconnected: (1) substation
spacing in the system, (2) size and number of substations, (3) sub-transmission voltage and
design, and (4) distribution feeder voltage and design. One of these factors cannot be
optimized without close evaluation of its interrelationship with the others. Therefore,
determining the most cost-effective design involves evaluating the transmission-substation-
feeder system design as a whole against the load pattern, and selecting the best combination
of transmission voltage, substation transformer sizes, substation spacing, and feeder system
voltage and layout. This economic equipment sizing and layout determination is based on
achieving a balance between two conflicting cost relationships:
Higher voltage equipment is nearly always more economical on a per-MW basis.
Higher voltage equipment is available only in large sizes (lots of MW).
In cases where the local area demands are modest, higher voltage equipment may be
more expensive simply because the minimum size is far above what is required - the utility
has to buy more than it needs. How these two cost relationships play against one another
depends on the load, the distances over which power must be delivered, and other factors
unique to each power system, such as the voltages at which power is delivered from the
regional power pool and whether the system is underground or overhead.
Figure 1.16 illustrates the difference that careful coordination of system design between
levels of the power system can have in lowering overall cost. Shown are the overall costs
from various combinations of T&D system layout for a large metropolitan utility in the
eastern United States. Each line connects a set of cost computations for a system built with
the same transmission and distribution voltages (e.g., 161 kV transmission and 13.8 kV
distribution) but varying in substation sizes (and hence, implicitly, their spacing).

In all cases, the utility had determined it would build each substation with two equally
sized transformers (for reliability), with none over 75 MVA (larger transformers are too
difficult to move along normal roads and streets, even on special trailers). Either 161 kV or
69 kV could be used as sub-transmission, either 23.9 kV or 13.8 kV could be used as
distribution voltage. Any size transformer, from 15 MVA to 75 MVA, could be used,
meaning the substation could vary from 30 MVA to 150 MVA in size. (Peak load of such
substations can normally be up to 75% of capacity, for a peak load of from 23 to 100 MW.)
Substation spacing itself is implicit and not shown. Given the requirement to cover the
system, determining transmission voltage, distribution, and substation size defines the
system design guidelines entirely.
Overall, the ultimate lowest cost T&D system guidelines are to build 120 MVA
substations (two 60 MVA transformers) fed by 161 kV sub-transmission and distributing
power at 23.9 kV. This has a levelized cost (as computed for this utility) of about $179/kW.
(Levelized and other time-versus-money concepts will be discussed in Chapter 5). In this
particular case, a high distribution voltage is perhaps the most important key to good
economy - if 13.8 kV is used instead of 23.9 kV as the primary voltage, minimum
achievable cost rises to $193/kW.
The very worst design choices plotted in Figure 1.16, from an economic standpoint,
would be to build 25 MVA substations fed by 161 kV sub-transmission and feeding power
to 23.9 kV feeders ($292/kW). This would require many small substations, each below the
effective size of both the transmission and distribution voltages used. Overall, 161 kV and
23.9 kV are the correct choices for economy, but only if used in conjunction with a few,
large substations. If substations are to be 25 MVA, then 69 kV and 13.8 kV do a much
more economical job ($228/kW), but still don't achieve anything like the optimum value.
The most important point: Achieving economy in power delivery involves coordinating the
interactions, performance, and economies of the multiple system levels. Chapters 11-19
cover the issues and techniques of coordinated multi-level planning.

300

V*
O
O

161kV&13.8kV
Ul
Q.
69kV & 23.9kV
"5

1 69kV&13.8kVX
<0
O 200
o

50 100 150
SUBSTATION CAPACITY - MVA

Figure 1.16 Overall cost of T&D system depends on the balanced design of the sub-transmission,
substation, and feeder level, as described in the text. Cost can vary by significant margins depending
on how well performance and economy at various levels of the system are coordinated.

38 Chapter 1

Two-Q Planning
Figure 1.16 illustrates rigorously cost-conscious planning and engineering from within
the traditional utility paradigm of focusing on capacity and assessing options on the basis
of cost per kW served or cost per kVA of capacity. A complication of modern, non-
traditional distribution planning is that reliability is often a key design element (see
section 1.3). Planners are challenged to reduce overall cost while achieving designs that
achieve targets for both capacity and reliability. Planning methods and concepts that deal
simultaneously with a distribution system's quantity (demand and capacity) and quality
(continuity of sufficient service - reliability) are labeled by the author as Two-Q methods.
Traditional distribution planning and engineering was predominately "single-Q."
Reliability was addressed through the use of standards and was not a specific, targeted
level of performance. Cost was minimized with respect to meeting capacity needs. The
most economical design alternative from the capacity standpoint (as discussed above and
illustrated in Figure 1.16) may not be the most reliable option, nor the alternative that
offers the best "bang for the buck" from the standpoint of reliability. (The various
systems depicted in Figure 1.16 do not all have the same level of reliability.)
In general, due to the layout and behavior of radial distribution systems, lower voltage
primary feeder systems have slightly better reliability statistics than higher voltage
systems, other things being equal. Thus, a 12.5 kV primary distribution system will
usually have slightly better reliability than a 25 kV system designed to serve the same
area. A planner choosing between those two options would need to determine:
if the difference in reliability between these different voltage options is important
and worth any cost differential
if any savings due to the lower cost of the higher voltage system (assuming that is
the case) can be used to buy reliability improvement equipment such as reclosers
and automation that will improve that option's reliability so that it is equal or
superior to that offered by the lower-voltage option
The basic concept behind all Two-Q methods is to extend the planning analysis and
evaluation, and cost minimization approaches, to cover the additional dimension of
reliability from a performance standpoint. This means they plan and design systems to
numerical reliability targets in addition to, as traditional methods do, planning and
designing to numerical demand targets. Two-Q methodology does not mean any specific
analytical method, but includes all methods that take this approach.
Figures 1.17-1.19 show the Two-Q equivalent of the "single-Q" analysis shown in
Figure 1.16. Two-dimensional analysis produces manifolds (surfaces) in a three
dimensional space of cost versus reliability versus capacity, rather than lines on a plane
(cost versus capacity). Figure 1.17 shows the results for one system type (69 kV and 13.8
kV). Figure 1.18 shows a second type of system (161 and 23.9 kV) added, providing data
to assemble Figure 1.19's identification of situations where one or the other is least
costly. This analysis is the Two-Q equivalent of the traditional cost-minimization system
approach covered in Figure 1.16. With regard to Two-Q planning:
Reliability is an issue in planning: Systems built to equivalent engineering
standards (all of those depicted on the broad lines in Figures 1.17 and 1.18) do not
give the same performance with respect to reliability.
Achieving targeted levels of reliability means designing to constant reliability
targets, not constant design standards.


Figure 1.17 Surface depicting the costs for a 161 - 13.8 kV system design standard for various
substation capacities and optimized to meet various levels of SAIDI (system average intemption
duration index) targets. The broad line represents the same set of systems as the 69+13.8 kV line in
Figure 1.16, standardized designs in which substation size varies. As shown these various options
are not of constant reliability level, but vary from near 100 minutes to about 165 minutes SAIDI
annually, with larger substation systems having worse reliability.

Figure 1.18 A second surface, depicting the costs for the 138 - 23.9 kV system design standard for
various substation capacities and optimized to meet various levels of SAD1 (system average
interruption duration index) targets. The most lightly shaded surface area shows the portion of the
138-23.9 kV surface that has higher costs than 69-13.8 kV designs.

MARCEL DEKKER, INC.
270 Madison Avenue, New York, New York 10016

40 Chapter 1

0 50 100 150
SUBSTATION CAPACITY - MVA

Figure 1.19 Projection of the lowest cost areas of the two surfaces leads to this information,
(heavily shaded area) depicting situations where the 69-13.8 kV design paradigm is less costly than
the 161-23.9 kV one.

Planning that includes reliability is considerably more complicated than traditional
planning that minimizes cost/kW.
Reliability can be managed in company with cost and traditional design goals.
Two-Q types of techniques are becoming standard in the power industry, as more and
more utility commissions regulate utility performance with respect to both profit and
customer service quality. Various procedures and analytical methods exist to address
reliability and produce the types of information shown in Figures 1.17 through 1.19. These
will be discussed in much more detail in Chapters 23 and 28.
Are Two-Q Methods Reliability Improvement
or Cost-Reduction Techniques?
For the most part, Two-Q methods can be viewed as cost reduction methods for modern
electric delivery utilities, rather than as reliability improvement methods. Traditional
electric distribution planning methods and practices achieved largely satisfactory levels of
customer service reliability through the use of design, equipment, and operating standards.
Adherence to these standards was expected to provide "good reliability," and in the vast
majority of cases it did. But use of rigid standards was not always optimal from a cost
standpoint. Standards would sometimes lead to facilities that were "overbuilt" from the
standpoint of reliability, and often a standard would call for more expense than minimally
necessary. By contrast, Two-Q eschews rigid standards for equipment and design, in favor
of rigid numerical standards (quantitative targets) for reliability. This requires more
complicated analysis and planning methods, but usually leads to lower cost designs to
achieve the targeted levels of reliability. Two-Q spends less in situations where that cost
reduction will not affect reliability. The majority of applications of Two-Q are for reducing
the cost of achieving satisfactory, and rather traditional, levels of reliability. For this reason
it is perhaps most proper to view Two-Q as predominately a cost reduction methodology.
Two-Q methods can and are also used in situations where reliability must be augmented,
either because the traditional design is not working out, or because much better reliability
than traditional levels is needed. In either case, it again minimizes the cost to achieve the
required reliability. Therefore, again, it is a cost-minimizing technique.
Of course, Two-Q, by achieving designs that meet quantitative reliability targets, does
help produce more consistent reliability performance and assure targeted levels are more
likely to be reached, so it is also legitimately a way of improving reliability beyond that
given by traditional approaches.


1.10 SUMMARY OF KEY POINTS

A transmission and distribution system moves power from a utility's power production and
purchase points to its customers. The T&D system's mission of delivering power to the
customers means that it must be composed of equipment spread throughout the service
territory, arranged locally so that capacity is always in proportion to local electrical demand,
with the facilities in each neighborhood sized and configured both to fit well into the whole
and to serve the local needs. In the authors' opinion, engineering a system to meet this
challenge is seldom easy, but engineering a system to do so at the minimum possible cost is
always extremely challenging.
A T&D system is composed of several interconnected, hierarchical levels, each of
which is required for completion of the power delivery task. These levels are:
• transmission
• sub-transmission
• substation
• feeder
• secondary
• customer
To a certain extent, the burden of power delivery, and costs, can be shifted from one
level to another through changes in the specifications of equipment, lay-out standards, and
design of the various levels. For example, costs can be pared at the substation level by using
fewer, larger substations, but this means feeders in each substation area will have to carry
power farther, and perhaps carry more power per feeder as well, increasing feeder costs.
Low overall design cost is achieved by balancing these factors. The performance and
economy, and thus the design, of a power system are dominated by a number of constraints
due to physical laws, and further shaped by a number of practical considerations with
regard to equipment, layout, and operating requirements. The more important of these are
listed in Table 1.5, and discussed below.
Three Basic Requirements Dominate T&D Planning
The T&D system must cover ground
This is the rule about T&D - ultimately the electric system must "run a wire" to every
customer. A significant portion of the cost of a distribution system, perhaps 25%, is due to
this requirement alone, independent of the amount of peak load, or energy supplied, or
reliability required. Although this requirement does not create the majority of costs, the
author considers it the single most important aspect of T&D because it is most responsible
for dictating the character of the system requirements and for shaping the constraints and
challenges facing power delivery planners.
The T&D system must have sufficient capacity to meet peak demand
The T&D system must be designed not to meet average loading conditions, but to handle
peak demand levels, even if those last only a few hours a year. The requirement for
capacity to meet peak demands, above and beyond the minimum levels needed just to
"cover ground," is responsible for about 25-40% of the cost of most T&D systems, but
generally does not contribute much to design challenges or constraints.

42 Chapter 1

Table 1.5 One-Page Summary of Chapter 1

A T&D system must cover ground.
A T&D system must have sufficient capacity to meet peak demand.
A T&D system must achieve reliability on the order of 99.975% or better.
Six "natural laws" shape the character of T&D systems and their equipment.
1. It is more economical to move power at high voltage than at low voltage.
otherwise similar equipment.
3. Utilization voltage is useless for the transmission of power.
4. It is costly to change voltage level.
advocates of modern distributed generators notwithstanding, there is a
significant economy of scale in generation - large generators produce
power more economically than small ones. Thus, it is most efficient to
produce power at a few locations utilizing large generators.
6. Power must be delivered in relatively small quantities at low (120 to 250
volts) voltage levels.
A power distribution system consists of multiple "levels": sub-transmission,
substation, feeder, and service.
The systems approach - planning and engineering the four levels of the system
as a whole - is a key element of good planning; a power delivery system's parts
cannot be viewed in isolation, particularly if one desires to truly minimize cost.
Electrical equipment is available only in certain discrete sizes. Planners must
build their systems with transformers, breakers, cables, and other "building
blocks" that are available only in discrete sizes such as 15, 25,40, and 60 MVA.
Dynamic service area re-assignment through switching is a power tool for
maintaining high utilization of facilities while accommodating the fact that only
discrete equipment sizes are available, ultimately reducing cost if used well.
The cost per kW to upgrade existing facilities is usually greater than the cost
per kW to build it initially. Often it makes economic sense to install more
capacity than needed in order to avoid upgrade costs later on.
Peak demands are non-coincident - they do not all occur at the same time.
Electrical losses can create a significant cost, often greater over the lifetime of
the equipment than the cost of the equipment and its care, even when these long-
term losses costs are discounted using valid time-value-of-money methods.

Reliability was traditionally obtained through standards that specified
contingency margins and design features.

Reliability is increasingly achieved in modern systems by "designing" to
specific targets using reliability-based engineering methods.


Electrical connectivity must be maintained at very high levels of reliability
Reliability on the order of 99.975% or better - less than two hours out of service per year -
is typically required in most distribution systems built in first- and second-world countries.
This requires protection, sectionalizing, and control equipment, along with redundant
capacity, and operations resources. All told, this requirement for high reliability is
responsible for roughly 50% of the cost of most T&D systems. It is thus the most
important cost element in modern power systems and also creates design and cost-reduction
challenges.
Six Natural Laws of Power Delivery
The physical phenomena and the natural laws that govern their behavior, along with the
realities of power delivery needs, produce six "laws" or fundamental truths that shape T&D
system layout and performance.
1. It is more economical to move power at high voltage. The higher the voltage, the
lower the cost per kilowatt to move power any distance.
otherwise similar equipment. Thus, high voltage lines, while potentially
economical, cost a great deal more than low voltage lines, but have a much
greater capacity. They are only economical in practice if they can be used to
move a lot of power in one block - they are the giant economy size, but while
always giant, they are only economical if one truly needs the giant size.
3. Utilization voltage is useless for the transmission of power. The 120/240 volt
single-phase utilization voltage used in the United States, or even the 250
volt/416 volt three-phase used in "European systems" is not equal to the task of
economically moving power more than a few hundred yards. The application of
these lower voltages for anything more than very local distribution at the
neighborhood level results in unacceptably high electrical losses, severe voltage
drops, and astronomical equipment cost.
4. It is costly to change voltage level - not prohibitively so, for it is done throughout
a power system (that's what transformers do) - but voltage transformation is a
major expense, which does nothing to move the power any distance in and of
itself.
advocates of modern distributed generators notwithstanding, there is a significant
economy of scale in generation - large generators produce power more
economically than small ones. Thus, it is most efficient to produce power at a
few locations utilizing large generators.
6. Power must be delivered in relatively small quantities at low (120 to 250 volts)
voltage levels. The average customer has a total demand equal to only 1/10,000
or 1/100,000 of the output of a large generator.
Multiple, Interconnected Levels
A power distribution system consists of four identifiable "levels:" sub-transmission,
substation, feeder, and service. Together, these comprise a highly interconnected
system with the electrical and economic performance at one level heavily dependent on
design, siting, and other decisions at another level. To a certain extent the T&D system

44 Chapter 1

must be planned and designed as a whole; its parts cannot be viewed in isolation,
particularly if planners desire to minimize cost.
Discrete Equipment Sizes
In many cases equipment is available only in certain discrete sizes. For example, 69
kV/12.47 kV transformers may be available only in 5, 10, 20, and 22 MVA sizes.
Usually, there is a large economy of scale - the installed cost of a 20 MVA unit being
considerably less than two times that for a 10 MVA unit.
Dynamic Service Area Assignment Is a Powerful Planning Tool
Linearity of expansion costs by and utilization in the face of upgrades that are only
available in discrete sizes is obtained by arranging service areas. Given that a substation
can be built in only one of several discrete sizes, the planner obtains an economical
match of capacity to load by varying service areas. When reinforced with new capacity
additions, a substation might pick up load of surrounding substations, effectively
spreading the capacity addition among several substations.
Cost to Upgrade Is Greater than Cost to Build
For example, one mile of a 12.47 kV, three-phase feeder using 600 MCM conductor (15
MW thermal capacity) might cost $150,000 to build, and one mile of 9 MW feeder (336
MCM conductor) might cost only $110,000. But the cost to upgrade the 336 MCM
feeder to 600 MCM wire size at a later date would be about $130,000, for a cumulative
total of $240,000. Therefore, one aspect of minimizing cost is to determine size for
equipment not on the basis of immediate need, but by assessing the eventual need and
determining whether the present worth of the eventual savings warrants the investment
in larger size now.
Peak Demands Are Non-Coincident
Not all customer loads occur at the same time. This has a number of effects. First, peak
load in different parts of the utility system may occur at different times. Second, the
system peak load will always be less than the sum of the peak loads at any one level of
the system - for example in more power systems the sum of all substation peak loads
usually exceeds system total peak by 3-8%. This diversity of peak loads means that
considerable attention must be paid to the pattern and timing of electric load if
equipment needs (and consequently costs) are to be minimized.
Losses Cost Can Be Significant
In some cases, the present worth of future losses on a line or heavily loaded transformer can
exceed its total capital cost. In most cases, this does not happen. But in most cases, the
present worth of losses is greater than the difference in cost between most of the choices
available to the planner, meaning that losses are a major factor to be considered in
achieving overall least-cost design.
Reliability Was Traditionally Obtained Through Standards that
Specified Contingency Margins and Design Features
T&D reliability was traditionally "designed into" distribution systems by the use of
design, equipment, and operating "standards" — actually not standards in any legal or
regulatory sense, but rather hard guidelines set up by the utility. These specified
specific ways to lay out and design facilities, defined what type and how equipment was


to be used, called for specific amounts of redundant capacity and specified how it was to
be used, and laid out operating policies and rules. They assured that satisfactory levels
of reliability were achieved (if not, then the "standards" were raised).
Reliability Is Increasingly Achieved by "Designing" to Specific
Targets Using Reliability-Based Engineering Methods
Standards-driven reliability design, the traditional approach, can achieve good reliability
performance, but such an approach often "overbuilds" or applies equipment and
configuration in less than an optimally useful manner. In some cases much more cost
effective use of funds can be accomplished by shifting to reliability-based standards
(numerical targets) and designing to those rather than employing hard and fast
guidelines (standards). Such planning is more complicated and costly but results in both
more assurance that reliability targets will be met and a savings in cost.

REFERENCES AND BIBLIOGRAPHY

A. V. Abbott, Electrical Transmission of Energy, Van Nostrand, New York, 1895.

R. Brown, Electric Distribution System Reliability, Marcel Dekker, New York, 2002.

J. C. Das, Power Systems Analysis, Marcel Dekker, New York, 2002.

M. V. Engel et al., editors, Tutorial on Distribution Planning, IEEE Course Text EHO 361-6-PWR,
Institute of Electrical and Electronics Engineers, Hoes Lane, NJ, 1992.

L. Philipson and H. L. Willis, Understanding Electric Utilities and De-Regulation, Marcel Dekker,
New York, 2000.

E. Santacana et al., Electrical Transmission and Distribution Reference Book, ABB, Raleigh, NC.

Consumer Demand and
Electric Load

2.1 THE TWO Qs: QUANTITY AND QUALITY OF POWER

Electric consumers require power, whether delivered from the utility grid or generated
locally by distributed sources, in order to help accomplish the uses for which they need
energy. Their need for electric power, and the value they place upon its delivery to them,
has two interrelated but fundamentally separate dimensions. These are the two Qs: quantity,
the amount of power needed, and quality, the most important aspect of which is usually
dependability of supply (reliability of power supply, or availability as it is often called). The
relative importance of these two features varies from one consumer to another depending
on their individual needs, but each consumer finds value in both the amount of power he
obtains and its availability as a constant, steady source that will be there whenever needed.
This chapter discusses demand and use of electric power as seen from the consumer's
standpoint: the utility's job is to satisfy consumer needs as fully as possible within
reasonable cost constraints. Cost is very much an important aspect to consumers too, so
both the utility and the consumer must temper their plans and desires with respect to power
and reliability based on real world economics. Energy consumers do not get everything they
want, only what they are willing to pay for. Utilities should not aim to provide flawless
service, which would be prohibitively expensive, but instead aim to provide the highest
level possible within economic constraints of the consumers' willingness to pay.
This chapter begins, in section 2.2, with a discussion of electric load, the quantity of
usage of a consumer, and its "cause," consumer demand for the end-use of electrical
appliances. That section continues with how demand varies as a function of consumer type
and end-use, and how power demand is represented in electric system studies using load
curves and load duration curves. Section 2.3 then discusses reliability and availability as
seen by the consumers and ways this can be characterized and studied. Section 2.4 covers
concepts about variation in customer needs and the value customers obtain from electric
service. Section 2.5 briefly reviews Two-Q analysis and planning concepts and their
application to consumer load analysis. Finally, section 2.6 provides a summary of key
points.
47

48 Chapter 2

Quantity and Quality of Power Are Both
Important Aspects of Consumer Value
The amount of power used and the dependability of its availability for use as the consumer
wants are both important attributes in determining the value of the electric power to that
consumer. The value of each, quantity and quality, is somewhat independent of the value
attached to the other. In order to demonstrate this, it is worth considering two nearly
ubiquitous appliances, which happen to represent opposite extremes of valuation in each of
these dimensions: the electric water heater and the personal computer.
A typical residential storage water heater stores about 50 gallons of hot water and has a
relatively large connected load compared to most household appliances: about 5,000 watts
of heating element controlled by a thermostat. Its contribution to coincident peak hourly
demand in most electric systems is about 1,500 watts. The average heater's thermostat has
it operating about 30% of the time (1,500/5,000) during the peak hour.
Despite its relatively high demand for quantity, this appliance has a very low
requirement for supply quality. A momentary interruption of electric flow to this device -
on the order of a minute or less - is literally undetectable to anyone using it at that moment.
Interruptions of up to an hour usually create little if any disruption in the end-use the device
provides, which is why direct load control of water heaters is an "economy discount" option
offered by many utilities. Similarly, the water heater is not critically dependent on other
aspects of power quality: it will tolerate, and in fact turn into productive heat, harmonics,
spikes, and other "unwanted" contamination of its supply.
By contrast, a typical home computer requires much less quantity, about 180 watts. But
unlike the water heater, it is rendered useless during any interruption of its power supply.
Just one second without power will make it cease operation. Battery backup power
supplies, called uninterruptible power supplies, can be used to avoid this problem, but at an
increase in the computer's initial and operating cost. In addition, a computer is more
sensitive to harmonics, voltage dips, and other power quality issues than a water heater.

Space
High heating
i Water
heater

Refrigerator

TV,
Computer

Low
Low High
Need for Continuity of Service

Figure 2.1 Electrical appliances vary in the amount of electricity they demand and the level of
continuity of service they require to perform their function adequately.


Figure 2.2 shows these two appliances, and several other household devices, plotted on
a two-dimensional basis according to the relative importance of quantity and continuity of
service in determining their successful operation. Traditionally, "demand" has been viewed
only as a requirement for quantity of power, but as can be seen the true value of electric
service depends on both dimensions of delivery. Recognition of this fact is the great driver
in moving the electric industry toward "tiered rates," "premium rates," and other methods
of pricing power not only by amount and time of use, but reliability of delivery.
Such evaluation, in two dimensions, is particularly important in many distributed
generation (DG) planning situations, because DG's emphasis can be tailored, from site
designs that provide mostly additional reliability and voltage support (quality) to those that
provide mostly capacity (quantity) with low quality, if that is what is needed. A majority of
the DG units that are already installed worldwide were tailored to provide quality. They
were designed and put in place as "back-up" generation units and are seldom operated.
They provide improved quality of power at their site, not additional capacity.

2.2 QUANTITY OF POWER DEMAND: ELECTRIC LOAD

The term load refers to the electrical demand of a device connected to and drawing power
from the electric system in order to accomplish some task, e.g., opening a garage door, or
converting that power to some other form of energy, such as a light. Such devices are called
appliances, whether they are a commonly regarded household item, e.g., a refrigerator,
lamp, garage door opener, paper shredder, electric fence to keep cattle confined, etc. To the
consumer, these appliances convert electricity into the end product. Consumer desires for
the end products of these devices is the root cause of the load - the reason it exists at all.
But the electric service planner can view this relation from the other direction: a particular
appliance (e.g., a heat pump) is a device for transforming a demand for a specific end-use
(warm or cool air) into electric load.
Electrical loads are usually rated or measured by the amount of power they need, in
units of real volt-amperes, called watts. Large loads are measured in kilowatts (thousands of
watts) or megawatts (millions of watts). Power ratings of loads and T&D equipment refer to
the device at a specific nominal voltage. The electric load, measured in watts as seen by the
system, is very much a function of the characteristics of these electric machines. Their load
- the electrical burden they place on the system as they operate - is what the distribution
system sees and what it exists to serve. The schedule of loads — when an appliance is
operating — is a function of consumer schedules and demand. Lights create load when
homeowners and business people need lighting. Televisions create a load when viewers are
watching them. Dishwashers use power when turned on.
But the load is also a function of the devices themselves and can vary depending on the
size and type of device. Lighting is a good example. Illumination can be provided by
incandescent lights, fluorescent tubes, halogen lights, various types of vapor, or LED lights.
For the same purpose, each would have different electric loads and characteristics. A table
lamp providing 1100 lumens of light would create a load of 75, 62, or 18 watts depending
on if it was using an incandescent, halogen, or compact fluorescent bulb.
Constant Power, Constant Current, and Constant Impedance Loads
The load of an electric device is rated at its nominal voltage, but many electric appliances
and devices have an electric load that varies as the supply voltage is varied. Generally,
loads are grouped into three categories depending on how their demand varies as a function
of voltage, constant power (demand is constant regardless of voltage), as a constant current
(demand is proportional to voltage), or as a constant impedance (power is proportional to

50 Chapter 2

voltage squared). The load at a particular consumer or in a particular area might be a
mixture of all three.
Incandescent lighting, resistive water heaters, stovetop and oven cooking loads, and
many other loads are constant impedance loads. On a feeder with a 7.5% voltage drop
from substation to feeder end (the maximum permitted under typical "range A" voltage
drop criteria - see Chapters 20 and 22) a constant impedance load will vary by up to
14.5% depending on where it is located on the feeder. The same set of incandescent lights
that create 1 kW of load at the feeder head would produce only .855 kW at the feeder
end.
Electric motors, regulated power supplies, and some other loads are constant power -
basically as the voltage drops, the devices make up for that loss by drawing more current,
so the voltage times current product remains constant. Thus, 1 kW constant power is 1 kW
at 1.0 per unit voltage, at .97 PU, and at .925 PU (i.e., at the maximum voltage drop
permitted under typical range A voltage drop limits of 7.5%). In addition to actual constant
power loads, all loads downstream of a tap-changing transformer appear to the power
system as relatively constant power loads and are generally modeled in this manner.
There are a small number of devices that are constant current loads. They draw power in
proportion to the voltage. Thus, a load that is 1 kW at 1.0 PU voltage is .95 kW at .95 PU
and 1.1 kW at 1.1 PU. Some types of welding units, smelting, and electroplating processes
are constant current loads. However, constant current loads constitute a small minority of
electric devices. Constant current loads are sometimes used in distribution studies, however,
because they represent the aggregate behavior of a mixture of constant impedance and
constant power loads. Figure 2.2 shows the load versus voltage characteristics for 1 kW
loads of all three types.
Electric loads are almost always represented as at nominal voltage. Constant impedance,
power, and current loads of 1,000 watts all create exactly that level of demand at 1.0 PU
voltage, although they would differ at other voltages; the three loads depicted in Figure 2.2
are all 1.0 kW loads at 1.0 P.U. voltage, but differ in load level at other voltages because of
their different voltage sensitivities.

Constant Impedance

Constant Current

.95 .96 .97 .98 .99 1.0 1.01 1.02 1.03 1.04 1.0S

Voltage - PU

Figure 2.2 Actual demand for power created by "one kilowatt" of each of the three types of load, as a
function of voltage supplied to them.


Given a feeder with a 7.5% voltage drop from substation to feeder end, a constant
impedance load will vary by up to 14.5% depending on where it is located on the feeder.
The same constant impedance load that creates 1 kW of demand at the feeder head would
produce only .855 kW at the feeder end. By contrast, a 1 kW constant power load will
produce 1 kW of demand anywhere on the feeder. Incandescent lighting, resistive water
heaters, and electric stoves and ranges and many other household loads are constant
impedance loads. Induction motors, controlled power supplies, as well as loads downstream
of any tap-changing transformer in a power system appear to the power system as relatively
constant power loads and are generally modeled in this manner. Very few loads are
constant current, although a 50/50 mixture of power and impedance loads looks very much
like and can be modeled as "constant current."
There are situations where it is quite important to model the voltage sensitivity of load
correctly. Detailed modeling of load characteristics can become quite complicated as shown
in Figure 2.3. There are basically nine elements to the electric load at a node: three types of
load for each of three phases. Generally, when modeled in detail, loads are represented with
a type and load represented at nominal (1.0 PU) voltage. Thus, a 1 kW constant power load
is referred to as "1 kW, constant power" and constant impedance load that produces 1 kW
at 1.0 PU is referred to as a "1 kW, constant impedance load."
From a practical standpoint, it is not necessary to model constant current loads.
However, the distinction between constant power and constant impedance loads is
important.1 Since voltage drop on a feeder depends very much on the load, the manner of
representing load is critical to computing voltage drop accurately.
For example, a three-phase 12.47 kV line built with 336 MCM phase conductor (a
typical suburban overhead distribution trunk) can serve a 10.72 MW constant impedance
load at a distance of two miles, while staying just within "range A" 7.5% voltage drop
limits. The "10.72 MW constant impedance load" will create a demand of only 9.36 MW
due to the voltage drop (9.36 MW = (1 - .075)2 x 10.72 MW). Thus, the feeder stays within
voltage criteria if serving a "10.72 MW constant impedance load." However, voltage drop
rises to 8.59 volts, more than a volt beyond what the criteria permit if the load is a 10.72
MW constant power load.
The cost of writing and putting into place software to represent loads as constant power,
constant impedance, or any mixture thereof in distribution engineering tools (load flows) is
minor, but the cost of obtaining good data for such a model is not. Usually, the data are
estimated, in one of two ways:
1. All loads, except special ones, are designated as a default mixture of power
and impedance loads. In the absence of measured results to the contrary, the
recommended general rule of thumb is to use a 60/40 constant
power/constant impedance mixture for summer peak loads, and a 40/60 split
for winter peak loads.

2. The proportion of constant power and constant impedance loads at each
location in the power system is estimated based on a spatial end-use
appliance model.

1
A 50/50 mix of constant voltage and constant power loads represents a constant current load within
4% over the range .9 to 1.1 PU.

52 Chapter 2

D = RBCn

D = QBCn x(v/V)

= WBCn x(v/V)2

Figure 23 Shown here, the load between the two end nodes consists of three types of loads -
constant power, constant current, and constant impedance, designated as P, Q, and W respectively.
Conceivably each could have a distinctly different power factor. So whether entered as load and
power factor, real and imaginary load, or magnitude and angle, this load model requires six data input
values for this one load. From a practical standpoint, without loss of real accuracy, the model can be
simplified and data requirements reduced to twelve values, by deleting the constant current
representations and representing any constant current loads as a mixture of the other two types.

The load of most individual consumers, and thus in any small area, is a mixture of
types. Industrial areas are most typically closest to constant power loads (about 80/20
constant power/constant impedance). Residential areas with very strong summer-peaking
loads usually have a 70/30 split of constant power/constant impedance load, while
winter-peaking residential areas, or those summer-peaking areas with low summer load
densities per household (little air conditioning), tend to have a mix closer to 30/70
constant power/constant impedance loads. Commercial areas are usually about 50/50 or
60/40 constant power/constant impedance loads. In rural and residential areas of
developing countries, and commercial areas in economically depressed regions, loads
generally tend to be about 20/80 constant power/constant impedance.
Linking Load-Voltage Characteristic Models to End-Use Models
Often, the best way to study the load characteristics in a region is to use the end-use load
approach that will be covered in Chapter 3, section 3.3. End-use appliance categories are
used to identify what portion of a consumer's demand has what type of electric voltage
sensitivity. For example, electric water heaters and incandescent lighting, both purely
resistive loads, are constant impedance loads. Heat pumps and air conditioners, being motor
loads, are constant power, etc. A dishwasher (heating elements and a motor) is a cross
between the two. Few loads outside the industrial sector are ever constant-current loads.
(For that matter, few industrial loads are constant current, usually only certain types of
facilities in which electric current is used for driving chemical processes like electro-
plating, electrolysis, etc., in what is essentially a large, controlled-current process.)
Transient Load Characteristics
The models referred to here and used in virtually all planning studies are steady state
models of load behavior. They represent the long-term characteristics of the load. For
example, in the long term an induction motor is pretty much a constant power load; reduce
voltage to it by 3% and it will draw about 3% more current in order to try to maintain its
mechanical output. But in the instants following a change in voltage, there may be a
transient load response, in which the motor or other device would react differently for a
moment. For example, a motor connected to a large rotating mass - rotational inertia -


might not instantly react with a 3% current increase when voltage is reduced by 3%. It
might gradually react over a few moments, eventually reaching a 3% current increase. In
other cases, "constant power" electronic circuits will in fact draw more current for an
instant if voltage is increased, before transitioning to a lower level of current draw. These
types of transient load models are often indispensable for stability and voltage collapse
analysis, but are not needed in day-to-day expansion planning and optimization studies.

Electric Power Is Bought for End-Uses
No consumer wants electric energy itself. Consumers want the products it can provide - a
cool home in summer, hot water on demand, compressed air for manufacturing, electronic
robotic factory control, cold beer in the 'fridge, and football on color TV. Electricity is only
an intermediate means to some end-use. These different goals and needs are called end-
uses, and they span a wide range of applications. Some end-uses are unique to electric
power (the authors are aware of no manufacturer of natural-gas powered TVs, stereos, or
computers), but for many other end-uses, electricity is only one of several possible energy
sources (water heating, home heating, cooking, clothes drying). In many other end-uses,
electricity is so convenient that it enjoys a virtual monopoly, even though there are
alternatives, e.g., gasoline-powered refrigerators and natural gas for interior lighting and for
air conditioning.
Each end-use is satisfied through the application of appliances or devices that convert
electricity into the desired end product. For example, with lighting a wide range of
illumination devices are used, including incandescent bulbs, fluorescent tubes, sodium
vapor, high-pressure monochromatic gas-discharge tubes, and, in special cases, lasers. Each
type of lighting device has differences from the others that give it an appeal to some
consumers or for certain types of applications. Regardless, each requires electric power to
function, creating an electric load when it is activated. Similarly, for other end-uses, such as
space heating, there are various types of appliances, each with advantages or disadvantages
in initial cost, operating efficiency, reliability and maintenance, noise and vibration, or other
aspects. Each produces an electric load when used to produce heat for a home or business.

Load Curves and Load Curve Analysis
The electric load created by any one end-use usually varies as a function of time. For
example, in most households, demand for lighting is highest in the early evening, after
sunset but before most of the household members have gone to bed. Lighting needs may be
greater on weekends, when activity lasts later into the evening, and at times of the year
when the sun sets earlier in the day. Some end-uses are quite seasonal. Air-conditioning
demand generally occurs only in summer, being greatest during particularly hot periods and
when family activity is at its peak, usually late afternoon or very early evening.
The result of this varying demand for the products of electricity application is a
variation in the demand for power as a function of time. This is plotted as a load curve,
illustrated in Figure 2.4, a diagram that shows how load varies as a function of time.
Typically, the value of most interest to the planners is the peak load (maximum amount
of power that must be delivered). The load curve defines, directly or indirectly, the
capacity requirements for equipment; the minimum load and the time it occurs; the total
energy, area under the curve that must be delivered during the period being analyzed; and
the load value at the time of system peak load.

54 Chapter 2

10
3?
CD

Mid Noon Mid Mid Noon Mid
Time of Day Time of Day

Figure 2.4 End-use demand. Right, average demand for BTU of cooling among houses in the
author's neighborhood in Gary, NC on a typical weekday in early June. Left, lighting lumens used by
a retail store on a typical weekday in early June.

Summer and Winter Residential Peak Day Load Curves

Summer „ Winter

HOUR HOUR
Figure 2.5 Electric demand for each class varies hourly and seasonally, as shown here, with a plot of
average coincident load for residential users in central Florida.

Other utilities - 9%

Industrial - 11%

Commercial - 22%

Cooling - 50%

Residential - 58%
Lighting - 17%
Water Heat - 12%
Cooking - 9%
Refrig. - 6 %
Other - 6%

Figure 2.6 Electric peak demand of a utility in the southeastern United States, broken down by
consumer class, and within the residential class by contribution to peak for the major uses for which
electricity is purchased at time of peak by the residential class.


Residential
400

Time of Day

Commercial
150

Time of Day

Industrial
150

Time of Day

Figure 2.7 Different consumer classes have different electrical demand characteristics, particularly
with regard to how demand varies with time. Here are summer peak day load curves for the three
major classes of consumer from the same utility system.

56 Chapter 2

Load curve shape is a function of consumer demand, or when homeowners and
businesses are using power. Thus, load varies by time of day and time of year. Figure 2.5
shows daily load curves for a utility system in Florida, illustrating how electric load varies
over a day, and how winter and summer days vary in profile. The difference in shape and
peak level (winter is about 5% higher than summer in this case) is due to varying consumer
needs.

Consumer classes
Different types of consumers purchase electricity. About half of all electric power is used in
residences, which vary in the brands and types of appliances they own and their daily
activity patterns. Another fourth is consumed by commercial businesses, both large and
small, that buy electricity, having some similar end-uses to residential consumers (heating
and cooling and illumination), but that have many needs unique to commercial functions
(cash register/inventory systems, escalators, office machinery, neon store display lighting,
parking lot lighting). Finally, industrial facilities and plants buy electricity to power
processes such as pulp heating, compressor and conveyor motor power, and a variety of
manufacturing applications. As a result, the load on an electric system is a composite of
many consumer types and appliance applications. Figure 2.6 shows this breakdown of the
peak electrical load for a typical power system by consumer class and end-use category
within one class.
Consumer class load curves
Electric utilities usually distinguish between consumers based on the type of activity they
perform - residential, commercial, industrial - those categories often broken into various
subclasses. Usually pricing, or the rates charged for electricity, is different for each class.
Load is often tracked for each class, as shown in Figure 2.7.

Details of Load Curve Measurement

Demand and demand periods
"Demand," as normally used in electric load analysis and engineering, is the average value
of electric load over a period of time known as the demand interval. Very often, demand is
measured on an hourly basis as shown in Figure 2.8, but it can be measured on any interval
basis - seven seconds, one minute, 30 minutes, daily, and monthly. The average value of
power during the demand interval is given by dividing the kilowatt-hours accumulated
during the demand interval by the length of the interval. Demand intervals used for
measuring and defining peak demand vary among power companies, but those commonly
used in collecting data and billing consumers for "peak demand" are 15, 30, and 60
minutes. Generally, if no period is explicitly defined, demand is on an hourly basis.
Load curves may be recorded, measured, or applied over a specific period of time. For
example, a load curve might cover one day. If recorded on an hourly demand basis, the
curve consists of 24 values, each the average demand during one of the 24 hours in the day,
and the peak demand is the maximum hourly demand seen in that day.
Load data are also gathered and used on a monthly basis and on an annual basis in
various activities within a utility. In some cases, load curve data may be recorded on an
hourly basis, but not really represent demand. This is the case with discrete sampling of a
type that will be discussed in Chapter 3, for example load curves sampled once an hour for
only a second each time. The measurement captures not the total energy — average usage -
during the hour, but only that for a brief period at the top of each hour.


250

200

2 150
Q
O 100

50

12 Noon 12
Hour

Figure 2.8 Demand on an hourly basis for one day and the actual load curve (solid black line)
metered on a minute by minute basis, for a feeder segment serving 53 homes.

Hourly Load Curve Re-ordered as load duration curve

Hour of the Day Hour of the Day

Figure 2.9 The hourly demand samples in a 24-hour load curve are reordered from greatest
magnitude to least to form a daily load duration curve.

100
80

60

40

20

2800 8760
Hours per Y e a r

Figure 2.10 Annual load duration curve for a commercial site with a 90 kW peak demand, from an
reliability study of two distributed generation (DG) units of 40 and 60 kVA capacity respectively. As
shown here, the demand exceeds 40 kW, the capacity of the smaller of two units, 2,800 hours a year.
During those 2,800 hours, adequate service can be obtained only if both units are operating.

58 Chapter 2

Load factor
Load factor is the ratio of the average to the peak demand. The average load is the total
energy used during the entire period (e.g., a day, a year) divided by the number of demand
intervals in that period (e.g., 24 hours, 8,760 hours). The average is then divided by the
maximum demand to obtain the load factor:
kWh/Hrs
Load Factor _ (2.1)
Peak Demand

Average Demand kW
(2-2)
Peak Demand kW

kWh in period /2 3)
(Peak kW in period) x (Hours in period)

Load factor gives the extent to which the peak load is maintained during the period under
study. A high load factor means the load is at or near peak a good portion of the time.

Load Duration Curves
A convenient way to study load behavior is to order the demand samples from greatest to
smallest, rather than as a function of time, as in Figure 2.9. The two diagrams consist of the
same 24 numbers in a different order. Peak load, minimum load, and energy (area under the
curve) are the same for both.

Annual load duration curves

Most often, load duration curves are produced on an annual basis, reordering all 8,760
hourly demands (or all 35,040 quarter hour demands if using 15-minute demand intervals)
in the year from highest to lowest, to form a diagram like that shown in Figure 2.10. The
load shown was above 26 kW (demand minimum) for all 8,760 hours in the year. It was
never above 92 kW, its peak, but was above 40 kW for 2,800 hours.
Load duration curve behavior will vary as a function of the level of the system. The load
duration curves for small groups of consumers will have a greater ratio of peak to minimum
than similar curves for larger groups due to coincidence of load effects (see Chapter 3).
Those for very small groups (e.g., one or two consumers) will also have a pronounced
"blockiness," consisting of plateaus - many hours of similar demand level (at least if the
load data were sampled at a fast enough rate). The plateaus correspond to combinations of
major appliance loads. The ultimate "plateau" load duration curve would be a load duration
curve of a single appliance, for example a water heater that operated a total of 1,180 hours
during the year. This appliance's load duration curve would show only two load levels: no
load and full load, without any values in between.


:: Lo-densRes.

= Residential

H Apartments
===
i Retail

9 Offices

• Hi-rise

= Lgt.lnd.

= Med.lnd.

• Hvylnd.

SPRINGFIELD LIGHT & POVER CO.
Service Territory and
Distribution Substations

Figure 2.11 Left, map showing types of consumer by location for a small city. Right, map of electric
demand for this same city.

Spatial Patterns of Electric Demand
An electric utility must not only produce or obtain the power required by its consumers, but
also must deliver it to their locations. Electric consumers are scattered throughout the utility
service territory, and thus the electric load can be thought of as distributed on a spatial basis
as depicted in Figure 2.11. Just as load curves show how electric load varies as a function of
time (and can help identify when certain amounts of power must be provided), so spatial
load analysis helps identify where load growth will be located and how much capacity will
be needed in each locality.
The electric demand in an electric utility service territory varies as a function of location
depending on the number and types of consumers in each locality, as shown by the load
map in Figure 2.11. Load densities in the heart of a large city can exceed 1 MW/acre, but
usually average about 5 MW per square mile over the entire metropolitan area. In sparsely
populated rural areas, farmsteads can be as far as 30 miles apart, resulting in load density as
low as 75 watts per square mile. Chapter 25 and Table 25.1 give more details in typical load
densities in various types of areas. Regardless of whether an area is urban, suburban, or
rural, electric load is a function of the types of consumers, their number, their uses for
electricity, and the appliances they employ. Other aspects of power system performance,
including capability, cost, and reliability, can also be analyzed on a locational basis, an
often important aspect of siting for facilities, including DG.

2.3 ELECTRIC CONSUMER DEMAND FOR QUALITY OF POWER
As mentioned in this chapter's introduction, a central issue in consumer value of service
analysis is matching availability and power quality against cost. T&D systems with near
perfect availability and power quality can be built, but their high cost will mean electric
prices the utility consumers may not want to pay, given the savings an even slightly less
reliable system would bring. All types of utilities have an interest in achieving the correct
balance of quality and price. The traditional, franchised monopoly utility, in its role as the

60 Chapter 2

"electric resource manager" for the consumers it serves, has a responsibility to build a
system whose quality and cost balances its consumers' needs. A competitive retail
distributor of power wants to find the best quality-price combination, only in that way will
it gain a large market share.
While it is possible to characterize various power quality problems in an engineering
sense, characterizing them as interruptions, voltage sags, dips, surges, or harmonics the
consumer perspective is somewhat different. Consumers are concerned with only two
aspects of service quality:
1. They want power when they need it.
2. They want the power to do the job.
If power is not available, neither aspect is provided. However, if power is available, but
quality is low, only the second is not provided.

Assessing Value of Quality by Studying the Cost of a Lack of It
In general, consumer value of reliability and service quality are studied by assessing the
"cost" that something less than perfect reliability and service quality creates for consumers.
Electricity provides a value, and interruptions or poor power quality decrease that value.
This value reduction (cost) occurs for a variety of reasons. Some costs are difficult, if not
impossible, to estimate. An hour-long interruption that began fifteen minutes from the end
of a crucial televised sporting event like the Superbowl, or the end of a "cliffhanger" prime
time movie, would garner considerable complaint.
But often, very exact dollar figures can be put on interruptions and poor power quality:
food spoiled due to lack of refrigeration; wages and other operating costs at an industrial
plant during time without power; damage to product caused by the sudden cessation of
power; lost data and "boot up" time for computers; equipment destroyed by harmonics; and
so forth. Figure 2.12 shows two examples of such cost data.

20

10

0 15 30 45 60 75 90
Interruption Duration - Minutes Total Harmonic Voltage Distortion - %

Figure 2.12 Left, cost of a weekday interruption of service to a pipe rolling plant in the southeastern
United States, as a function of interruption duration. Any interruption costs at least $5,000, the cost to
unload material in process, bring machinery back to "starting" position and restart it. In addition to
that minimum interruption cost there is a "delay duration" cost that is the product of the interruption
time and the plant's fixed hourly operating cost (about $7,500/hour). At right, present worth of the
loss of life caused by harmonics in a 500 horsepower three-phase electric motor installed at that same
industrial site, as a function of harmonic voltage distortion.


Customer Value of Quality Utility Cost of Providing Quality Sum of Costs

Low High Low High Low High
Quality Quality Quality

Figure 2.13 Concept of value-based planning. The consumer's cost due to poorer quality (left) and
the cost of various power delivery designs with varying levels of quality (center) are computed over a
wide range. When added together (right) they form the total cost of quality curve, which identifies the
minimum cost reliability level (point A).

A A A A A
A A A A A A A A A

V V V V V V V V v v v V V V V

Figure 2.14 Output of a 5.2 volt DC power supply used in a desktop computer (top) and the
incoming AC line voltage (nominal 113 volts). A voltage sag to 66% of nominal causes power supply
output to cease within three cycles.

200

I 150

50

1
10' 101 102 103 10" 107
Time in Cycles

Figure 2.15 The original CBEMA curve as often depicted, showing curves of voltage deviation
versus period of deviation, with the sag shown in Figure 2.14 plotted (black dot).

62 Chapter 2

101 102 103
Time in Cycles

Figure 2.16 CBEMA 96 voltage sag profile specifies tolerance of dips to 70% for periods of less
than 30 cycles (500 ms).

Value-Based Planning
To be of any real value in utility planning, information of the value consumers put on
quality must be usable in some analytical method that can determine the best way to
balance quality against cost. Value-based planning (VBP) is such a method: it combines
consumer-value data of the type shown in Figure 2.12 with data on the cost to design the
T&D system to various levels of reliability and power quality in order to identify the
optimum balance. Figure 2.13 illustrates the central tenet of value-based planning. The cost
incurred by the consumer due to various levels of reliability or quality and the cost to build
the system to various levels of reliability are added to get the total cost of power delivered
to the consumer as a function of quality.2 The minimum value is the optimum balance
between consumer desire for reliability and aversion to cost. This approach can be applied
for only reliability aspects, i.e., value-based reliability planning, or harmonics, or power
quality overall. Generally, what makes sense is to apply it on the basis of whatever qualities
(or lack of them) impact the consumer - interruptions, voltage surges, harmonics, etc., in
which case it is comprehensive value-based quality of service planning.
Cost of Interruptions
The power quality issue that affects the most consumers, and which receives the most
attention, is cessation of service, often termed "service reliability." Over a period of several
years, almost all consumers served by any utility will experience at least one interruption of
service. By contrast, a majority will never experience serious harmonics, voltage surge, or
electrical noise problems. Therefore, among all types of power quality issues, interruption
of service receives the most attention from both the consumers and the utility. A great deal
more information is available about cost of interruptions than about cost of harmonics or
voltage surges.
Voltage Sags Cause Momentary Interruptions
The continuity of power flow does not have to be completely interrupted to disrupt service:
If voltage drops below the minimum necessary for satisfactory operation of an appliance,
power has effectively been "interrupted" as illustrated in Figure 2.14. For this reason many

Figure 2.13 illustrates the concept of VBP. In practice, the supply-side reliability curves often have
discontinuities and significant non-linearities that make application difficult.


consumers regard voltage dips and sags as momentary interruptions - from their
perspective these are interruptions of the end-use service they desire, if not of voltage.
Much of the electronic equipment manufactured in the United States, as well as in many
other countries, has been designed to meet or exceed the CBEMA (Computer and Business
Equipment Manufacturer's Association) recommended curves for power continuity, shown
in Figures 2.15 and 2.16. If a disturbance's voltage deviation and duration characteristics
are within the CBEMA envelope, then normal appliances should operate normally and
satisfactorily. However, many appliances and devices in widespread use will not meet
this criterion at all. Others will fail to meet it under the prevailing ambient electrical
conditions (i.e., line voltage, phase unbalance, power factor, and harmonics may be less
than perfect).
The manner of usage of an appliance also affects its voltage sag sensitivity. The voltage
sag illustrated in Figure 2.14 falls just within the CBEMA curve, as shown in Figure 2.15.
The manufacturer probably intended for the power supply to be able to withstand nearly
twice as long a drop to 66% of nominal voltage before ceasing output. However, the
computer in question had been upgraded with three times the standard factory memory, a
second and larger hard drive, and optional graphics and sound cards, doubling its power
usage and the load on the power supply. Such situations are common and means that power
systems that deliver voltage control within recommended CBEMA standards may still
provide the occasional momentary interruption.
For all these reasons, there are often many more "momentary interruptions" at a
consumer site than purely technical evaluation based on equipment specifications and T&D
engineering data would suggest. Momentary interruptions usually cause the majority of
industrial and commercial interruption problems. In addition, they can lead to one of the
most serious consumer dissatisfaction issues. Often utility monitoring and disturbance
recording equipment does not "see" voltage disturbances unless they are complete cessation
of voltage, or close to it. Many events that lie well outside the CBEMA curves and
definitely lead to unsatisfactory equipment operation are not recorded or acknowledged. As
a result, a consumer can complain that his power has been interrupted five or six times in
the last month, and the utility will insist that its records show power flow was flawless. The
utility's refusal to acknowledge the problem irks most consumers more than the power
quality problem itself.
The SEMI F47 Standard
Somewhat similar to the CBEMA curve in concept and application is the SEMI
(Semiconductor Equipment and Materials International) standard F47, which defines an
expected voltage sag tolerance of equipment and facilities used in semiconductor
electronics fabrication. It was created to focus attention on shorter-duration sags The
standard is designed to apply to the actual fabrication equipment itself, such as film
deposition and photolithography equipment, ancillary systems such as surface preparation
and cleaning, and automated inspection and testing equipment, too. The standard specifies
that equipment meeting it tolerate dips in voltage to 50% of nominal for 200 ms, to 70% of
nominal for 500 ms, and to 80% of nominal for a full second. Figure 2.17 shows this
standard as most typically plotted. Note that the standard's performance requirements are
step functions, and that it does not specify performance for sag periods shorter than the 200
ms or for sags to below 50% of nominal voltage level. An allied SEMI standard, F42,
covers testing of equipment for compliance with the standard. It includes definitions of the
metrics involved and descriptions of how to measure them, and of compliance and how to
assess equipment on a pass-fail basis relative to the standard.

64 Chapter 2

0 100 200 300 400 500 600 700 800 900 1000

Time in Milliseconds
Figure 2.17 The SEMI F47 standard for voltage sag tolerance of equipment for semiconductor
electronics fabrication specifies greater tolerance of voltage sags of short duration (less than 200 ms)
than the CBEMA 96 standard.

o

A A A A A A

v v v v v v

" v V V v v v v v

Figure 2.18 Two ways that voltage can temporarily fall to 66% of nominal voltage. Top, the diesred
result. The waveform remains a sine wave. Middle, the voltage is clipped. Bottom, the voltage sags.
The net effect on equipment of the two bottom waveforms will be similar (a cessation of operation or
freezing of mechanic components), but the diagnosis and correction of the two problems will be far
different. Many standards and some power system test equipment do not "see" or record such
distinctions.


Nature of the Voltage Sag
Figure 2.18 highlights an important distinction for voltage sag studies and standards, that of
the exact form of the reduction in voltage. At the top is the voltage sag from Figure 2.14, a
drop to 66% of nominal but with a shape that is still a sine wave. In the middle, the voltage
is clipped at 66%. Many standards and definitions do not address the nature of the
waveform shape, and much of the testing methods power quality engineers may use do not
explicitly respond to, measure, or record it. Any particular piece of electrical equipment
may react differently to the top and bottom supply profiles. While the difference in these
two waveforms will rarely make a difference in practice (equipment disturbed by one will
likely be disturbed by the other) T&D engineers should keep in mind that subtle differences
like these and other waveform-related issues are often undefined and unmeasured with
respect to power quality and the equipment they use. For that reason alone there is often a
bit of uncertainty in any practical power quality mitigation project.
The bottom waveform in Figure 2.18 shows yet another "undefmable" issue. This sag
drops linearly to 33% and then returns linearly to the nominal voltage level. It is difficult to
characterize with respect to its duration or voltage limit since it has no single definitive
value.
Frequency and Duration of Interruptions Both Impact Cost
Traditional power system reliability analysis concepts recognize that service interruptions
have both frequency and duration (See Chapter 4). Frequency is the number of times during
some period (usually a year) that power is interrupted. Duration is the amount of time
power is out of service. Typical values for urban/suburban power system performance in
North America are about two interruptions per year with about 100 to 120 minutes total
duration. Both frequency and duration of interruption impact the value of electric service to
the consumer and must be appraised in any worthwhile study of consumer value of service.
A number of reliability studies and value-based planning methods have tried to combine
frequency and duration in one manner or another into "one dimension." A popular approach
is to assume all interruptions are of some average length (e.g., 2.2 interruptions and 100
minutes is assumed to be 2.2 interruptions per year of 46 minutes each). Others have
assumed a certain portion of interruptions are momentary and the rest of the duration is
lumped into one "long" interruption (i.e., 1.4 interruptions of less than a minute and one 99-
minute interruption per year). Many other approaches have been tried (Patton, 1968,
Horton, et al., 1989, Rau, 1994). But all such methods are at best approximations, because
frequency and duration impact different consumers in different ways. No single
combination of the two aspects of reliability can fit the value structure of all consumers.
Figure 2.19 shows four examples of the author's preferred method of assessing
interruption cost, which is to view it as composed of two components, a fixed cost (Y
intercept) caused when the interruption occurred and a variable cost that increases as the
interruption continues. As can be seen in Figure 2.19, consumer sensitivity to these two
factors varies greatly. The four examples are:
1. A pipe rolling factory (upper left). After an interruption of any length, material
in the process of manufacturing must be cleared from the welding and polishing
machinery, all of which must be reset and the raw material feed set-up to begin
the process again. This takes about 1/2 hour and sets a minimum cost for an
interruption. Duration longer than that is simply a linear function of the plant
operating time (wages and rent, etc., allocated to that time). Prior to changes
made by a reliability study, the "re-setting" of the machinery could not be done
until power was restored (i.e., time during the interruption could not be put to

66 Chapter 2

Pipe Rolling Plant Insurance Claims Office

20 5

o 4
15 o
o
x 3
10
35 2
o
5 o

0 15 30 45 60 75 90 0 15 30 45 60 75 90
Interruption Duration - minutes Interruption Duration - minutes

Chemical Process Plant Residence
400 40

300 30

200 § 20
o
100 10

0 15 30 45 60 75 90 0 15 30 45 60 75 90

Figure 2.19 The author's recommended manner of assessing cost of interruptions includes evaluation
of service interruptions on an event basis. Each interruption has a fixed cost (Y intercept) and a
variable cost, which increases as the interruption continues. Examples given here show the wide range
of consumer cost characteristics that exist. The text gives details on the meaning of solid versus
dashed lines and the reasons behind the curve shape for each consumer.

use preparing to re-start once it was over). The dotted line shows the new cost
function after modifications to machinery and procedure were made so that
preparations could begin during the interruption.
2. An insurance claims office (upper right) suffers loss of data equivalent to
roughly one hour's processing when power fails. According to the site
supervisor, an unexpected power interruption causes loss of about one hour's
work as well as another estimated half hour lost due to the impact of any
interruption on the staff. Thus, the fixed cost of each interruption is equivalent
to about ninety minutes of work. After one-half hour of interruption, the
supervisor's policy is to put the staff to work "on other stuff for a while,"
making cost impact lower (some productivity); thus, variable interruption cost
goes down. The dotted line shows the cost impact of interruptions after
installation of UPS on the computer system, which permits orderly shutdown in
the event of an interruption.


3. An acetate manufacturing and processing plant (lower left) has a very non-linear
cost curve. Any interruption of service causes $38,000 in lost productivity and
after-restoration set-up time. Cost rises slowly for about half an hour. At that
point, molten feedstock and interim ingredients inside pipes and pumps begins
to cool, requiring a day-long process of cleaning sludge and hardened stock out
of the system. The dotted line shows the plant's interruption cost function after
installation of a diesel generator, started whenever interruption time exceeds
five minutes.
4. Residential interruption cost function (lower right), estimated by the author
from a number of sources including a survey of consumers made for a utility in
the northeastern United States in 1992, shows roughly linear cost as a function
of interruption duration, except for two interesting features. The first is the fixed
cost equal to about eight minutes of interruption at the initial variable cost slope
which reflects "the cost to go around and re-set our digital clocks," along with
similar inconvenience costs. Secondly, a jump in cost between 45 and 60
minutes, which reflects inconsistencies in human reaction to outage time on
questionnaires. The dotted line shows the relation the author uses in such
analysis, which makes adjustments thought reasonable to account for these
inconsistencies.
This recommended analytical approach, in which cost is represented as a function of
duration on a per event basis, requires more information and more analytical effort than
simpler "one-dimensional" methods, but the results are more credible.

Interruption Cost Is Lower If Prior Notification Is Given
Given sufficient time to prepare for an interruption of service, most of the momentary
interruption cost (fixed) and a great deal of the variable cost can be eliminated by many
consumers. Figure 2.20 shows the interruption cost figures from Figure 2.19 adjusted for
"24 hour notification given." Some industries can mitigate the cost of power interruptions if
given even one day's notice; others can't.

Cost of Interruption Varies by Consumer Class
Cost of power interruption varies among all consumers, but there are marked distinctions
among classes, even when cost is adjusted for "size" of load by computing all cost
functions on a per kW basis. Generally, the residential class has the lowest interruption cost
per kW and commercial the highest. Table 2.1 gives the cost/kW of a one-hour interruption
of service by consumer class, obtained using similar survey techniques for three utilities in
the United States:

1. A small municipal system in the central plains

2. An urban/suburban/rural system on the Pacific coast

3. An urban system on the Atlantic coast

68 Chapter 2

Table 2.1 Typical Interruption Costs by Class for Three
Utilities - Daytime, Weekday (dollars per kilowatt hour)
Consumer Class Utility: 1
Agricultural 3.80 4.30 7.50
Residential 4.50 5.30 9.10
Retail Commercial 27.20 32.90 44.80
Other Commercial 34.00 27.40 52.10
Industrial 7.50 11.20 13.90
Municipal 16.60 22.00 44.00

Pipe Rolling Plant Insurance Claims Office

20

o 4
15 o
o
x 3
10
*
o 2
o

0 15 30 45 60 75 90 0 15 30 45 60 75 90

Chemical Process Plant Residence
40

o 300 30
«»
I 20
o
10

0 15 30 45 60 75 90 0 15 30 45 60 75 90

Figure 2.20 If an interruption of service is expected consumers can take measures to reduce its
impact and cost. Solid lines are the interruption costs (the solid lines from Figure 2.19). Dotted lines
show how 24-hour notice reduces the cost impact in each case. Compare dotted lines in this figure to
those shown in Figure 2.19.


Cost Varies from One Region to Another
Interruption costs for apparently similar consumer classes can vary greatly depending on
the particular region of the country or state in which they are located. There are many
reasons for such differences. The substantial difference (47%) between industrial costs in
utilities 1 and 3 shown in Table 2.1 is due to differences in the type of industries that
predominate in each region. The differences between residential costs of the regions shown
reflect different demographics and varying degrees of economic health in their respective
regions.
Cost Varies Among Consumers within a Class
The figures given for each consumer class in Table 2.1 represent an average of values
within those classes as surveyed and studied in each utility service territory. Value of
availability can vary a great deal among consumers within any class, both within a utility
service territory and even among neighboring sites. Large variations are most common in
the industrial class, where different needs can lead to wide variations in the cost of
interruption, as shown in Table 2.2. Although documentation is sketchy, indications are the
major differing factor is the cost of a momentary interruption — some consumers are very
sensitive to any cessation of power flow, while others are impacted mainly by something
longer than a few cycles or seconds.
Cost of Interruption Varies as a Function of Time of Use
Cost of interruption will have a different impact depending on the time of use, usually being
much higher during times of peak usage, as shown in Figure 2.21. However, when adjusted
to a per-kilowatt basis, the cost of interruption can sometimes be higher during off-peak
than during peak demand periods, as shown. There are two reasons. First, the data may not
reflect actual value. A survey of 300 residential consumers for a utility in New England
revealed that consumers put the highest value on an interruption during early evening
(Figure 2.21). There could be inconsistencies in the values people put on interruptions (data
plotted were obtained by survey).

Table 2.2 Interruption Costs by Industrial Sub-Class for One
Hour, Daytime, Weekday (dollars per kilowatt)
Class $/kW
Bulk plastics refining 38
Cyanide plant 87
Weaving (robotic loom) 72
Weaving (mechanical loom) 17
Automobile recycling 3
Packaging 44
Catalog distribution center 12
Cement factory 8

70 Chapter 2

<* 25

§ 20
*3
Q.
15

I 10
15
5
V)
*- °
O
O 0
Mid. Noon Mid.
Time of Day

Figure 2.21 Cost of a one-hour interruption as a function of when it occurs, as determined by surveys
and interviews with 300 consumers of a utility in the northeastern United States, was determined on a
three-hour period basis. A high proportion of households in this survey have school age children at
home and thus perhaps weighed interruption costs outside of school hours more than during school
hours. However, in general, most households evaluate their interruption cost as higher in early
morning and early evening than at during the day.

However, there is a second, and valid, reason for the higher cost per kilowatt off-peak:
only essential equipment, such as burglar alarms, security lighting, and refrigeration is
operating - end-uses that have a high cost of interruption. While it is generally safe to
assume that the cost of interruption is highest during peak, the same cannot be assumed
about the cost per kilowatt.

Recommended Method of Application of Consumer Interruption Cost Data
A good analytical approach to value-based planning of power delivery includes assessment
of consumer costs using functions that acknowledge both a fixed cost for any interruption,
no matter how brief, and a variable cost as a function of duration. It should acknowledge
the differences in value of service among the different consumer classes, and the time
differences within those consumer classes.
Ideally, reliability and service quality issues should be dealt with using a value-based
reliability or power-quality planning method with consumer interruption cost data obtained
through statistically valid and unbiased sampling of a utility's own consumers, not with
data taken from a reference book, report, or technical paper describing data on another
utility system. However, in many cases for initial planning purposes the cost and time of
data collection are not affordable.
Figure 2.22 provides a set of costs of typical interruption curves that the author has
found often match overall consumer values in a system. It is worth stressing that major
differences can exist in seemingly similar utility systems due to cultural and economic
differences in the local consumer base. These are not represented as average or "best" for
use in value-based planning studies, but they are illustrative of the type of cost functions
usually found. They provide a guide for the preparation of approximate data from screening
studies and initial survey data.


Industrial Commercial Retail

a>150 «200
CO

J=

50

0 15 30 45 60 75 90 15 30 45 60 75 90

Commercial Non-Retail Residence

w150 High sensitivity
»150
CO

T5 100
o
Low sensitivity
8 50

v>
O e
0—
O 0 15 30 45 60 75 90 15 30 45 60 75 90

Figure 2.22 Typical interruption cost characteristics for consumer classes.

Cost in Figure 2.22 is given in terms of "times normal price." The cost of a power
interruption to nearly any business is usually much more than it would have paid for the
power had it been available. It is worth noting that in many surveys and studies of
interruption cost, the cost per kW of interruption is on the order often to one hundred times
the normal price (rate) for a kWh. Generally, if a utility has low rates its consumers report a
lower cost of interruption than if it has relatively higher rates. No reliable data about why
this correlation exists has been forthcoming.3
Cost of Surges and Harmonics
Far less information is available on consumer costs of harmonics and voltage surges as
compared to that available on the cost of interruptions. What data are made available in
publications, and most of the reported results and interpretations in the technical literature,
were obtained in very consumer-specific case studies, most done on a single consumer

3
It could be that value of continuity is worth more in those areas where rates are high (generally, more
crowded urban areas). However, the author believes that a good part of this correlation is simply
because in putting a value on interruptions, respondents to surveys and in focus groups base their
thinking on the price they pay for electricity. Given that a typical residential consumer uses roughly
1,000 kWh/month, they may simply be valuing an interruption as about "one-tenth of my monthly
bill."

72 Chapter 2

basis. As a result there is very little information available on average or typical costs of
voltage surge sensitivity and harmonics sensitivity.
Few consumers suffer from voltage dip (voltage problems lasting minutes or hours),
surge, harmonics, electrical noise, and similar problems. For this reason most studies of
cause, effect, and curve are done on a single-consumer or specific area basis. Here, results
reported in technical literature are the best guide.

End-Use Modeling of Consumer Availability Needs
The consumer class, end-use basis for analysis of electric usage, discussed in more detail in
Chapter 4, provides a reasonably good mechanism for study of the service reliability and
power quality requirements of consumers, even if originally developed only for analysis of
demand and DSM needs. Reliability and power quality requirements vary among
consumers for a number of reasons, but two characteristics predominate analytical
considerations:
1. End-usage patterns differ. The timing and dependence of consumers'
need for lighting, cooling, compressor usage, hot water usage,
machinery operation, etc., varies from one to another.
2. Appliance usage differs. The appliances used to provide end-uses will
vary in their sensitivity to power quality. For example, many fabric and
hosiery manufacturing plants have very high interruption costs purely
because the machinery used (robotic looms) is quite sensitive to
interruption of power. Others (with older mechanical looms) put a
much lower cost on interruptions.
End-use analysis can provide a very good basis for detailed study of power quality
needs. For example, consider two of the more ubiquitous appliances in use in most
consumer classes: the electric water heater and the personal computer. They represent
opposite ends of the spectrum from the standpoint of both amount of power required and
cost of interruption. A typical 50-gallon storage electric water heater has a connected load
of between 3,000 and 6,000 watts, a standard PC a demand of between 50 and 150 watts.
Although it is among the largest loads in most households, an electric water heater's ability
to provide hot water is not impacted in the least by a one-minute interruption of power. In
most cases a one-hour interruption does not reduce its ability to satisfy the end-use demands
put on it.4 On the other hand, interruption of power to a computer, for even half a second,
results in serious damage to the "product." Often there is little difference between the cost
of a one-minute outage and a one-hour outage.
It is possible to characterize the sensitivity of most end-uses in most consumer classes
by using an end-use basis. This is in fact how detailed studies of industrial plants are done
in order to establish the cost-of-interruption statistics, which they use in VBP of plant
facilities and in negotiations with the utility to provide upgrades in reliability to the plant.
Following the recommended approach, this requires distinguishing between the fixed cost
(cost of momentary interruption) and variable cost (usually linearized as discussed above)
on an end-use basis.

4
Utility load control programs offer consumers a rebate in order to allow the utility to interrupt power
flow to water heaters at its discretion. This rebate is clearly an acceptable value for the interruption,
as the consumers voluntarily take it in exchange for the interruptions. In this and many other cases,
economic data obtained from market research for DSM programs can be used as a starting point for
value analysis of consumer reliability needs on a value-based planning basis.


Figure 2.23 A simulation-type load forecast method’s end-use model (see Chapters 25 and 27) is
m d f e to handle “two-dimensional” appliance curves, a shown here for a residential electric water
oiid s
heater. The electric demand curve is the same data used in a standard end-use model of electric
demand. Intemption cost varies during the day, generally low prior to and during periods of low
usage and highest prior to high periods of use (a sustained outage prior to the evening peak usage
period would result in an inability to satisfy end-use demand).

Figure 2.24 The daily load curve for single family homes segmented into four interruption-cost
categories. High-cost end-uses in the home (darkest shading) are predominantly digital appliances
(alarm clocks, computers) and home entertainment and cooking systems. Lowest cost items are
deferrable loads like water heating and dishwashing, which if delayed an hour have little impact on
household activity patterns.

.. ..... -

74 Chapter 2

Residential Class 2
t 251
e

c
0
c 5
v)
0
0 0
Mid. Noon Mid.
Time of Day

Figure 2.25 Total interruption cost by hour of the day for a one-hour outage. based on the load curve
and cost category data in Figure 2.2 1.

Figure 2.26 Map of average reliability needs computed on a 10-acre small area grid basis for a port
city of population 130,000, using a combination of an end-use model and a spatial consumer
simulation forecast method, of the type discussed in Chapters 25 and 27. S h d n g indicates general
level of reliability need (based on a willingness-to-pay model of consumer value).


A standard end-use model used to study and forecast electric demand can be modified to
provide interruption cost sensitivity analysis, which can result in "two-dimensional"
appliance end-use models as illustrated in Figure 2.23. Generally, this approach works best
if interruption costs are assigned to appliances rather than end-use categories. In
commercial and industrial classes different types of appliances within one end-use can have
wildly varying power reliability and service needs. This requires an "appliance sub-
category" type of an end-use model. Modifications to an end-use simulation program to
accommodate this approach are straightforward (Willis, 2003), and not only provide
accurate representation of interruption cost sensitivity, but produce analysis of costs by time
and, if combined with the right type of simulation-based spatial forecast algorithm, location,
as shown in Figures 2.24 through 2.26.

2.4 THE MARKET COMB AND CONSUMER VALUES
Numerous surveys and customer focus meetings conducted by utilities, consultants, and
research firms (including the author's firm) indicate that electric consumers differ widely in
their need for, and their willingness to pay for, reliability of service. Consumers also differ
in exactly what "reliability," and in a broader sense "quality," means to them, although
availability of service, quick and knowledgeable response on the part of their supplier, and
power of a usable nature are always key factors in their evaluation.
As a result, the electric power demand marketplace can be likened to a comb as shown
in Figure 2.27, a series of small niches, each constituting only a small portion of the
customer base. Each niche represents customers with varying needs and interpretations for
reliability. Niches vary both in what type of reliability consumers need (some are sensitive
mostly to frequency of outage, others care greatly about duration).
A big dimension of variability among consumers is willingness to pay for reliability.
There are a few customers who need very high levels of reliability and who will pay
heavily for it. There are others who do not need reliability, only power, and are motivated
purely by lowest cost.

Customers who put Customers reasonably Customers who require
little value on reliability satisfied by the industry's high reliability and are
and are very unwilling traditional reliability and price willing to pay a higher
to pay for it combinations. price in return.

Motivated by Price Motivated by Service Quality

Figure 2.27 The electric marketplace can be likened to a comb: composed of many small niches,
each made up of customers who have a different cost-value sensitivity to reliability of service. Even
those who put a high value on reliability may differ greatly in how they define "good reliability," one
reason why there are no broad market segments, only dozens of somewhat similar but different niches.

76 Chapter 2

c
ro

Cost
Quality
Figure 2.28 Ultimately, the value electric power consumers see in their electric power supply is
based on a "value volume" of three dimensions: quantity, quality, and cost.

Cost Is King
Quantity and quality both have value for the electric consumer. But so does the consumer's
money and as a result no factor in the decision about energy source and energy usage is
more important to most consumers than cost. The three dimensions of power - quantity,
quality, and cost - form a "value volume" shown in Figure 2.28. The importance of each in
determining what is "best" will vary from one consumer to another.
Electricity Pricing Has Traditionally Been Based Only on Quantity
Traditionally, the price of power in the electric utility industry was based solely upon the
quantity used. A consumer would pay more if he or she used more power and less if the use
was less. Except in rare and specially treated cases, an electric utility would offer a "one-
size-fits-all" level of quality. Customers got what was available, whether or not they needed'
and would have paid for higher quality, or whether they would have preferred a discount
even if it meant somewhat lower quality.
Offering Variable Levels of Reliability
Traditionally, electric utilities have selected equipment and designed and engineered their
systems based on engineering standards that were aimed at maintaining high levels of
power system equipment reliability. These standards and methods, and the logic behind
them, were recognized as a proxy for designing the system to provide good customer
reliability: design the system so it seldom fails and the customers will seldom be without
power. This assumption was basically true, at least as implemented by most utilities. But
this paradigm was not without problems. First, it led to a culture in which taking care of the
equipment, not maintaining customer service quality, became the inherent goal. Second,
this often cost more than was necessary to achieve satisfactory customer service quality -
some systems ended up "overbuilt." These impacts will be discussed later in the chapters on
modern reliability-based planning and management methods.
But the relevant point is that a traditional utility's customers had little option but to
purchase a level of reliability corresponding to this set of standards (Figure 2.29). Only

<D
o

Lo Hi
Reliability
Figure 2.29 Generally, the cost of providing reliability in a power system is a non-linear function of
reliability level: providing higher levels carries a premium cost, as shown by the line above. While
reliability varies as a function of location, due to inevitable differences in configuration and equipment
throughout the system, the concept utilities used was to design to a single level of reliability for all
their customers.

<0
O

Lo Lo Hi
Relicfcility Reliability

Figure 230 A competitive market in power will recognize the demand for various levels of reliability
and providers will have to broaden the offerings they provide (left) until the range of options available
to the customer covers the entire spectrum of capabilities, as shown at the right. Actual options offered
will be even more complicated because customers will vary in the type of reliability they wish to
"buy" - making sure that all interruptions are of short duration may be more important to some
businesses and making sure there are few of them to others, etc.

78 Chapter 2

large industrial customers who could negotiate special arrangements had any ability to
bypass this "one-size-fits-all" approach to reliability that utilities provided.
The traditional way of engineering the system and pricing power has two
incompatibilities with a competitive electric power marketplace. First, its price is cost-
based, a central tenet of regulated operation, but contrary to the market-driven paradigm of
deregulated competition. Second, any "one-size-fits-all" reliability price structure will
quickly evaporate in the face of competitive suppliers. Someone among the various
suppliers will realize there is a profit to be made and a competitive advantage to be gained
by providing high-reliability service to consumers willing to pay for it. While others will
realize that there are a substantial number of customers who will buy low reliability power,
as long as it has a suitably low price. The concept that will undoubtedly emerge (in fact, is
already evolving at the time this book is being written) in a de-regulated market will be a
range of reliability and cost options. The electric utility market will broaden its offerings, as
shown in Figure 2.30. This will be a slow, but steady process, taking well over a decade to
pervade the industry.
These offerings will be created by artful use of power delivery system capabilities and
three types of "distributed resources" — storage (UPS), automation, and distributed
generation (DG). All of these distributed resources, but particularly DG, will be regarded
simultaneously from two perspectives. First, this new marketplace will create an
opportunity for DG, which can be tailored to variable reliability needs by virtue of various
design tricks (such as installing more/redundant units). Second, DG is a "threat" that may
force utilities to compete on the basis of reliability, because it offers an alternative, which
can easily be engineered (tailored), to different reliability-cost combinations. But either
way, power delivery "wire utilities" will have no choice but to compete in the reliability
arena in one way or another, even if restricted to continuing to operate as regulated utilities.

2.5 TWO-Q ANALYSIS: QUANTITY AND QUALITY VERSUS COST
The amount of power used and the dependability of its availability for use are both key
attributes in determining the value of the electric power to each consumer. The values
attached to both quantity and quality by most consumers are linked, but somewhat
independent of the value attached to the other. The "two-dimensional" load curve illustrated
previously in Figure 2.23 is a Two-Q load curve, giving demand as a function of time in
both Q dimensions: amount and quality of power needed.
In order to demonstrate the Two-Q concept, two nearly ubiquitous appliances, which
happen to represent opposite extremes of valuation in each of the two Q dimensions, can
be studied: the electric water heater and the personal computer. A typical residential
storage water heater stores about 50 gallons of hot water and has a relatively large
connected load compared to most household appliances: about 5,000 watts of heating
element controlled by a thermostat. Its contribution to coincident peak hourly demand in
most electric systems is about 1,500 watts. The average heater's thermostat has it
operating about 30% of the time (1,500/5,000) during the peak hour.
Despite its relatively high demand for quantity, this appliance has a very low
requirement for supply quality. A momentary interruption of electric flow to this device -
on the order of a minute or less - is literally undetectable to anyone using it at that
moment. Interruptions of up to an hour usually create little if any disruption in the end-
use the device provides, which is why direct load control of water heaters is an "economy
discount" option offered by many utilities. Similarly, the water heater is not critically
dependent on other aspects of power quality. It will tolerate, and in fact turn into
productive heat, harmonics, spikes, and other "unwanted" contamination of its supply.

By contrast, a typical home computer requires much less quantity, about 180 watts. But
unlike the water heater, it is rendered useless during any interruption of its power supply.
Just one second without power will make it cease operation. Battery backup power
supplies, called uninterruptible power supplies, can be used to avoid this problem, but at an
increase in the computer's initial and operating cost. In addition to its sensitivity to supply
availability, a computer is also more sensitive to harmonics, voltage dips, and other power
quality issues than a water heater.
Figure 2.31 shows these two appliances and several other household energy uses plotted
on a two-dimensional basis according to the relative importance of both quantity and
quality. This Two-Q load plot is a type of two-dimensional demand and capability analysis
that treats both Qs as equally important. Traditionally, "demand" has been viewed only as a
requirement for quantity of power, but as can be seen the true value of electric service
depends on both dimensions of delivery. Recognition of this fact has been the great driver
in moving the electric industry toward "tiered rates," "premium rates," and other methods
of pricing power not only by amount and time of use, but by reliability of delivery. Either or
both dimensions in the plot can be interpreted as single-dimensional profiles for quantity or
quality, as the case may be, providing useful information in the analysis of consumer needs
and how to best meet them (Figure 2.32).

Space
High heating
Water
heater

or
<D
DC
<5
o Refrigerator
Q.

TV,
Computer
CO

O

Low
Low High
Need for Continuity of Service
Figure 231 Electrical appliances vary in the amount of electricity they demand and the level of
continuity of service they require to perform their function adequately. Water heaters and computers
represent opposites in these regards; one requires a lot of load but has a low need for continuity of
service (the water heater), the other is the opposite. Scales shown here are qualitative. Other typical
household appliances are also shown.

80 Chapter 2

I 0)
x ©

Low High
Energy

Low High
Load

Figure 2.32 A Two-Q diagram of a consumer's needs can be interpreted in either of two
dimensions to identify the total (area under the curve) and range of needs in either Q direction,
producing Q profiles of the importance of demand or reliability to his particular needs.

Two-Q Planning and Engineering
Two-Q analysis using these same two dimensions also applies to the design of utility power
systems. This section will only summarize Two-Q planning and engineering so that an
appreciation of its capabilities, and how load forecasting and analysis interacts with them,
can be understood. The essence of Two-Q planning is to add cost as a third dimension and
then "optimize" either rigorously or through trial, error, and judgement, until the most
effective cost to obtain both Q target levels is obtained. Figures 1.17 through 1.19
illustrated that application in Chapter 1.
Figure 2.33, shows a similar type of plot based on a series of design studies, which were
used to form the manifold (surface) shown. It indicates what the cost/kW would be if peak
load or reliability performance levels were changed from current levels. This manifold
gives a rough idea of the cost of building and operating a modified version of the existing
system. (Only a rough idea, because the manifold was interpolated from a total of only eight
design studies. Once this has been used as a rough guide in planning target designs, specific
studies then refine the plan and selection of specific detailed design elements.) A specific
distribution planning problem (expressed here as a certain density of electric load) has been
set constant resulting in the plane shown running from front to back through the Two-Q
cube.
Figure 2.34 shows a second manifold plotted along with the first. As was illustrated with
another system in Figures 1.17-1.19, this represents the characteristics of a different type of
system, 25 kV being used on a targeted basis for primary distribution in the area, serving
large loads and for "express" service to areas where very good reliability is a challenge with


Figure 233 Three-dimensional Two-Q plot of a power system’s capabilities, in which cost has been
added as the third consideration. The manifold (surface) shown represents the capabilities of a 138
kV112.47 kV system optimized for various combinations of load density and SAIDI.

Figure 2.34 Left, the capabilities of a 24kV distribution system are shown by a second manifold.
Right, comparison of the two manifolds results in a Two-Q diagram that shows where each system
approach is most cost effective. The results shown are f o a real study, but one wt a number of
rm ih
real-world constraints. As a result the shape and results shown are not generalizable, although the
approach is.

82 Chapter 2

12.47 kV. The plot shows that the 25 kV design offers lower costs at higher load densities
and is more expensive at lower load levels. To the right is shown a diagram developed
from this data, which shows which Two-Q situations are best served by 12.47 kV or 25
kV systems. Inspection of the shapes of the manifolds in Figures 2.33 and 2.34 will
indicate to the reader the importance of accurate load and reliability targets for planning.
The slopes of each manifold are steep in places - small errors in estimating the demand
for either Q will generate substantial displacements in the cost estimates used in planning.
Thus good forecasts are important.
Further, the distance between the manifolds is small in many places, and the angle of
divergence from where they meet is small - thus small errors in estimating the demand
for either Q will yield substantial mistakes in selecting the optimal approach. The reader
should note that this sensitivity is not a characteristic of the Two-Q method - the figures
above are merely illustrating the cost and decision-making realities of distribution
planning, sensitivities that would be there regardless of the method used to engineer
reliability and demand capability into the system. Therefore, the lesson for the forecaster
is that accurate projections of future load levels are crucial, both for achieving good
reliability and cost. While this concept is fairly obvious to any experienced T&D planner,
the Two-Q approach can quantitatively measure, analyze, and help manage these issues.

2.6 CONCLUSION AND SUMMARY
The whole purpose of electric utility systems is to serve the electric demands of individual
households, businesses, and industrial facilities. Two useful mechanisms for evaluation of
consumer requirements and planning for their provision are end-use analysis and Two-Q
planning, both of which provide a useful basis for the study and analysis of both quantity
and quality. Table 2.3 summarizes key points from this chapter.
In addition to having a demand for electric energy, electric consumers value availability
and quality of power too, seeking satisfactory levels of both the amount of power they can
obtain and the availability of its supply. Each consumer determines what is satisfactory
based on specific needs. An interruption that occurs when a consumer does not need power
is not of concern. Harmonics or surges that make no impact on appliances are not
considered of consequence. Among the most important points in consumer value analyses
are:
1. Interruption of service is the most widely affecting power quality problem.
A majority of electric consumers in most utility systems experience
unexpected interruptions at least once a year.
2. Voltage dips and sags create momentary interruptions for many consumers
that are indistinguishable in impact on their needs with short-term
interruptions of power flow.
3. Analysis of consumer needs for power quality is generally done by looking at
the cost which less than perfect quality creates, not the value that near-
perfect quality provides. Thus, availability of power is usually studied by
analyzing the cost incurred due to interruptions and determining the costs
created by those interruptions assesses harmonics.
4. Power quality costs and values vary greatly depending on consumer class,
time of day and week, region of the country, and individual characteristics.



End-uses are the only reasons consumers buy power. They want the products power
provides: a comfortable room temperature, cold milk, images on a television screen,
stamped metal, or data retained for sub-millisecond retrieval.
Quantity and quality - the two Qs are both important aspects of every consumer's
demand. Everyone prefers to have no restrictions on either one, but every consumer has
definite limits on the amount of each that he will demand when he must pay a fair price for
them.
Cost is king: despite consumer demand for both quantity and quality, the vast majority of
consumers still are very cost conscious with respect to electric power.
Consumer classes - residential, commercial, industrial, etc. - are a good way of
distinguishing salient characteristics of loads and of categorizing consumers into sets of the
same overall behavior, because consumers within a class have similar needs and usage
patterns.
The market comb concept - viewing the consumer base as composed of many niches each
with different values for quality and quantity — is a useful tool in many marketing and
reliability planning functions.
Electric load is the quantity of power. It is the result of consumer usage of electrically
powered appliances and equipment to produce the end-uses discussed above.
Three types of voltage-related behavior characterize loads: constant power, constant
current, and constant impedance. Most consumers have a mixture of the three types.
Load Duration Curves are a convenient but approximate way to study load behavior
with respect to equipment and system needs over long periods of time (8,760 hours).
Load curves and load curve analysis provide an accurate but often involved means
to study electric load in detail with respect to system and equipment means.
End-use modeling of load curves - studying load over time broken down by
consumer class and end use application - is the most detailed way to study a
consumer's demand.
Spatial patterns of electric load - where and how it is distributed throughout the
utility service territory - is a very important aspect of distribution planning.
Reliability is the most important characteristic of quality in most power delivery planning.
Frequency and duration of interruptions are both important in determining the
impact of service interruptions on customers.
Consumer cost of reliability is usually studied by assessing the cost impact of power
system outages on the consumer - their cost of doing without power.
Value-based planning (VBP) involves balancing the utility's cost of spending more
to improve reliability against the consumers' cost of dealing with interruptions.
Voltage sags can cause momentary interruption of the end-use product itself, often
to the extent that their impact on the consumer is the same as short-duration
interruptions
End-use load curve modeling of quality provides a reasonably good mechanism for
study of the service reliability and power quality requirements of consumers.

84 Chapter 2

5. Price concerns often outweigh quality concerns. Throughout the late 1980s
and early 1990s, several independent studies indicated that roughly 30-40%
of commercial and industrial consumers are willing to pay more for higher
power quality, or something similar, a statistic that seems quite believable
and realistic. This led to considerable focus on improving power system
reliability and quality. However, it is worth bearing in mind that these same
studies indicate that 20-30% are quite content with the status quo, while the
remaining 30-40% would be interested in trading some existing quality for
lower cost.

REFERENCES
P. F. Albrecht and H. E Campbell, "Reliability Analysis of Distribution Equipment Failure Data," EEI
T&D Committee Meeting, January 20, 1972.

R. N. Allan et al., "A Reliability Test System for Educational Purposes - Basic Distribution System
Data and Results," IEEE Transactions on Power Systems, Vol. 6, No. 2, May 1991, pp. 813-821.

R. E. Brown and J. R. Ochoa, "Distribution System Reliability: Default Data and Model Validation,"
paper presented at the 1997 IEEE Power Engineering Society Summer Meeting, Berlin.

"Guide for Reliability Measurement and Data Collection," EEI Transmission and Distribution
Committee, October 1971, Edison Electric Institute, New York.

W. F. Horton et al., "A Cost-Benefit Analysis in Feeder Reliability Studies," IEEE Transactions on
Power Delivery, Vol. 4, No. 1, January 1989, pp. 446-451.

Institute of Electrical and Electronics Engineers, Recommended Practice for Design of Reliable
Industrial and Commercial Power Systems, The Institute of Electrical and Electronics Engineers,
Inc., New York, 1990.

A. D. Patton, "Determination and Analysis of Data for Reliability Studies," IEEE Transactions on
Power Apparatus and Systems, PAS-87, January 1968.

N. S. Rau, "Probabilistic Methods Applied to Value-Based Planning," IEEE Transactions on Power
Systems, November 1994, pp. 4082-4088.b

H. L. Willis, Spatial Electric Load Forecasting, Second Edition, Marcel Dekker, New York, 2002

Electric Load, Coincidence,
and Behavior

3.1 INTRODUCTION

This chapter discusses how electric load "looks" when measured and recorded at the
distribution level. Load behavior at the distribution level is dominated by individual
appliance characteristics and coincidence - the fact that all customers do not demand their
peak use of electricity at precisely the same time. For this reason alone, accurate load
studies on the distribution system require considerable care. In addition, proper
measurement and modeling of distribution loads can be quite difficult for a variety of
reasons related to how coincidence phenomena interact with load metering and numerical
analysis methods. Successful analysis of load data at the distribution level requires
procedures different from those than typically used at the system level. This chapter
addresses the reasons why and the measurement and analytical methods needed to
accurately represent and handle distribution load curve data.
The bulk of this chapter is divided into two major sections. In 3.2, coincident and non-
coincident load curve behavior are discussed - the actual behavior of distribution load, as it
appears it completely and accurately measured. Section 3.3 then looks at how some
popular measuring systems and analytical methods interact with coincident/non-coincident
behavior, with respect to distortions in measurement that may occur. Finally, section 3.4
reviews key points and presents some guidelines for accurate load curve measurement.
3.2 PEAK LOAD, DIVERSITY, AND LOAD CURVE BEHAVIOR
Almost all electric utilities and electric service companies represent the load of
consumers on a class by class basis, using smooth, twenty-four hour peak day load curves
like those shown in Figure 3.1. These curves are meant to represent the "average
behavior" or demand characteristics of customers in each class. For example, the system
whose data are used in Figure 3.1 has approximately 44,000 residential customers. Its
analysts will take the total residential customer class load (peaking at about 290 MW) and
divide it by 44,000 to obtain a 'typical residential load' curve for use in planning and
engineering studies, a curve with a 24-hour shape identical to the total, but with a peak of

85

86 Chapter 3

Summer and Winter Residential Class Peak Day Load Curve

Summer Winter
300

5200
5

12 12 12 12 6
HOUR HOUR

Summer and Winter Residential Peak Day Customer Load Curves

Summer Winter

12 6
HOUR

Figure 3.1 Utilities often represent the individual load curves of consumers with smooth,
"coincident" load curves, as shown here for the residential load in a utility system in Florida. Top,
summer and winter demand of the entire residential class in a southern US utility with 44,000
residential consumers. Bottom, representation of "individual" consumer load curves. Each is
1/44,000 of the total class demand.

22 kW
One household

Hour
Figure 3.2 Actual winter peak day load behavior for an individual household looks like this,
dominated by high "needle peaks" caused by the joint operation of major appliances.


6.59 kW (1/44,000 of 290 MW). This is common practice and results in the "customer
class" load curves used at nearly every utility.
Actually, no residential customer in any utility's service territory has a load curve that
looks anything like this averaged representation. Few concepts are as important as
understanding why this is so, what actual load behavior looks like, and why the smooth
representation is "correct" in some cases but not in others.
The load curve shown in Figure 3.2 is actually typical of what most residential customer
load looks like over a 24 hour period. Every residential customer's daily load behavior
looks something like this, with sharp "needle peaks" and erratic shifts in load as major
appliances such as central heating, water heaters, washer-dryers, electric ranges, and other
devices switch on and off.

Appliance Duty Cycles

The reason for the erratic "needle peak" load behavior shown in Figure 3.2 is that a
majority of electric devices connected to a power system are controlled with what is often
called a "bang-bang" control system. For example, a typical 50 gallon residential electric
water heater holds 50 gallons of water, which it keeps warm by turning on its heating
elements anytime the temperature of the water, as measured by its thermostat, dips below a
certain minimum target value. Thus switched on, the heating elements then create a
demand for power, which heats the water, and thereby the temperature begins to rise. When
the water temperature has risen to where it meets the thermostat's high target temperature
setting, the elements are turned off by that thermostat's relay. The on-and-off cycling of the
demand and the consequent impact on water temperature are shown in Figure 3.3. The
water heater cycles on and off in response to the thermostat - bang it is on until the water is
hot enough, then bang it is off and remains so until the water is cold enough to cause the
thermostat to cycle the elements back on.

Ti me — >

Figure 33 Top curve indicates the water temperature inside a 50-gallon water heater whose
thermostat is set at 172 degrees over a period of an hour. The thermostat cycles the heater's 4 kW
element on and off to maintain the water temperature near the 172 degree setting. Resulting load is
shown on the bottom of the plot.

88 Chapter 3

This bang-bang control of the water heater creates the actual demand curve of the water
heater. Over a 24-hour period, it is a series of 4 kW demands, as shown in the top row of
Figure 3.4. There is no "curve" in the sense of having a smooth variation in demand, or
variation in demand level, beyond an up and down variation between 4 kW when the water
heater is on, and 0 when it is not.

Aggregation of Individual Demand Curves into Coincident Behavior

Daily load curves for several residential water heaters are shown in Figure 3.4. Most of the
time a water heater is off, and for a majority of the hours in the day it may operate for only
a few minutes an hour, making up for thermal losses (heat gradually leaking out of the
tank). Only when household activity causes a use of hot water will the water heater
elements stay on for longer than a few minutes each hour. Then, the hot water use draws
heated water out of the tank, it being replaced by cool water piped in to re-fill the tank,
which lowers the temperature in the tank significantly, requiring the element to remain on
for a long time thereafter to "manufacture" the replacement hot water. During that hour, its
duty cycle - the percent of the time it is on - might be high, but the rest of the day it is quite
low.

B C
Cust R1-37747QW Cust. R1-67548BU
Feb. 6,1994 Feb. 6, 1 994 Feb. 7, 1 994
5 kU 5 kW

•o
a
o
_j

Hour Hour Hour

Cust. R1-37747QW and Five water heaters 50 Water Heaters
10kW Cust. R1-67548BU 25 kW
Feb. 6,1994

Hour Hour Hour

Figure 3.4 Curves A and B are water heater loads at neighboring houses on the same day (Feb. 6),
displaying similar but not identical behavior. Curve C is the second household's water heater curve for
Feb. 7, showing that random differences and slight shifts in usage that occur from day to day. Curve D
shows the sum of the two water heaters' curves for Feb. 6. For the most part, there is little likelihood
that they are both on at the same time, but occasionally around peak time they do coincide.
Aggregating 50 water heaters (F) results in a smoother load curve, one where load peaks during peak
periods of water heater usage, and the average peak time of the water heater curve, is longer (wider) -
meaning there is more likelihood that many overlap, and hence add to a high peak value.


Figure 3.4 shows several water heater loads, as well as plots representing the sum of the
load for a number of water heaters over a 24 hour period. It illustrates several important
points. First, all water heaters exhibit the same overall on-off behavior, but differ slightly as
to the timing of the cycles. Second, if enough load curves are added together, their sum
begins to look more like a smooth curve (similar to Figure 3.1) than the typical blocky
individual device curve.
Most large household loads operate in a similar manner. Air conditioners and heaters
operate by control of thermostats. So do refrigerators and freezers and electric ranges. As a
result, the household's total load curve (Figure 3.2) is the sum of a number of individual
appliance on-off curves that each look something like those at the top of Figure 3.4.
Duty Cycle Load of Air Conditioners and Heat Pumps
Figure 3.5 shows the ambient (outside) temperature and the electric load of an air
conditioner set to 68°F inside temperature over a 24 hour period. It illustrates a slightly
more complex behavior and will help to examine appliance duty cycle performance in
slightly more detail than the preceding section.
Note that, like the water heater, the air conditioner unit shown here cycles on and off. As
outside temperature increases, and the burden of removing heat from inside the house
becomes greater, the AC unit stays on longer every time it turns on, and stays off for a
shorter time. During the hottest period of the day, its duty cycle (percent of the time it is on
during the hour) is above 90%. The particular unit shown in Figure 3.5 was selected for this
house based on a 35 degree difference criteria - at 100% duty cycle it can maintain a 35°F
difference between outside and inside. At the 100°F ambient temperate reached, it is
providing a 32°F difference for the 68°F thermostat setting, near its maximum.
Note also that here, unlike in the case of the water heater, the amount of load (the height
of the load blocks when the unit is on) varies over the day. Connected load of the AC unit
increases as the ambient temperature increases.
The mechanical burden of an air conditioner or heat pump's compressor will be higher
when the temperature gradient the unit is opposing is higher (due to the nature of the
system's closed cycle design causing higher internal pressures and hence more pumping
load). When the pressure is higher, the electric motor must work harder, and hence draws
more power (i.e., has a higher load). The net result is that an air conditioner's operating
load increases as outside temperature gets warmer, and a heat pump's load increases as
outside temperature gets colder. Thus, during the heat of mid-day, the AC unit shown in
Figure 3.5 not only stays on most of the time, but has a load about 10% higher than it
would when ambient is cooler. Most AC and HP units, and many other mechanical
pumping loads, display this characteristic.
Coincident Load Behavior
Suppose one were to consider one hundred homes, for example one hundred houses served
by the same segment of a distribution feeder. Every one of these homes is full of equipment
that individually cycles on and off like appliances just described - water heaters, air
conditioners, heaters, and refrigerators that all are controlled by thermostats.
Thus, each household will have an individual daily load curve similar to the erratic,
choppy daily load curve shown in Figure 3.2, although each will be slightly different,
because each home has slightly different appliances, is occupied by people with slightly
different schedules and usage preferences, and because the times that individual thermostats
activate appliances are usually random, happening at slightly different times throughout the
day for each appliance.

90 Chapter 3

12

Figure 3.5 Daily cycle of THI (temperature-humidity-illumination index) and air conditioner load as
it cycles between on and off under thermostatic control (see text). As THI rises throughout the
morning, demand for cooling increases, and the air conditioner's duty cycle (% of time the unit is
operating) increases, until at peak it is operating all but a few minutes in every hour.

Two Homes Five Homes
22kW

Hour Hour

Twenty Homes One Hundred Homes
15 kW 15 kW
4-J

3
u
"v.
•o
O

Hour Hour

Figure 3.6 Daily load curves for groups of two, five, twenty, and one hundred homes in a large
suburban area. Note vertical scale is in "load per customer" for each group. Peak load per customer
decreases as the number of customers in the group becomes larger. This is coincidence of peak load.

Thus, one particular household might peak at 22 kVA between 7:38 AM and 7:41 AM,
while another peaks at 21 kVA between 7:53 AM and 8:06 AM, while another peaks at 23
kVA between 9:54 AM and 10:02 AM. These individual peaks are not additive because
they occur at different times. The individual peaks, all quite short, do not all occur
simultaneously. They are non-coincident.
As a result, when one adds two or more of these erratic household curves together, the
load curve needles usually interleave, much as the overlapping teeth of two combs might
interleave. As yet more customer load curves are added to the group, more and more
"needles" interleave, with only an occasional situation where two are absolutely coincident
and add together. The load curve representing the sum of a group of households begins to
take on the appearance of a smoother curve, as shown in Figure 3.6, in exactly the same
way and for the same reasons that the water heater load curves in Figure 3.4 did.
By the time there are twenty-five homes in a group, a smooth pattern of load behavior
begins to emerge from within the pattern of erratic load shifts, and by the time the loads for
one hundred homes have been added together, the curve looks smooth and "well-behaved."
This is the non-coincidence of peak loads. Individual household demand consists of load
curves with erratic swings as major appliances switch on and off at the bequest of their
thermostats and other control devices. Distribution equipment that serves only one customer
- service drops or a dedicated service transformer for example - sees a load like this.
But distribution equipment that serves large numbers of customers - a distribution
feeder for example - is looking at many households and hundreds of appliances at once.
Individual appliance swings do not make a big impact. It is their sum, when many add
together, that causes a noticeable peak for the feeder. This happens when many duty cycles
overlap.
Peak load per customer drops as more customers are added to a group. Each household
has a brief, but very high, peak load - up to 22 kW in this example for a southern utility
with heavy air-conditioning loads. These needle peaks seldom overlap, and under no
conditions would all in a large group peak at exactly the same time.1 As a result, the group
peak occurs when the combination of the individual load curves is at a maximum, and this
group peak load is usually substantially less than the sum of the individual peaks, as can be
seen by studying Figure 3.6.
This tendency of observed peak load per customer to drop as the size of the customer
group being observed increases is termed coincidence and is measured by the coincidence
factor, the fraction of its individual peak that each customer contributes to the group's peak.
observed peak for the group
C = coincidence factor = ^ ,. ,. ..—: r—
I, (individual peaks) v(3.1)
'
Typically, the value of C for large groups of customers is between .50 and .25. The
coincidence factor, C, can be thought of as a function of the number of customers in a
group, n:
observed peak load of a group of N customers
C(n) = 2 (their individual peaks) (3.2)
where n is the number of customers in the group,
and 1 < n < N = number of customers in the utility
1
Except under certain types of emergency operations during outages, due to something called "cold
load pickup." This occurs when all appliances have been without power during a long outage and
power is restored. All want power, creating a larger than normal demand and 100% coincidence.

92 Chapter 3

C(n) has a value between 0 and 1 and varies with the number of customers in identical
fashion to how the peak load varies, so that a curve showing C(n) and one showing peak
load as a function of the number of customers are identical except for the vertical scale.
Distribution engineers sometimes use the inverse of C to represent this same
phenomenon. The diversity factor, D, measures how much higher the customer's individual
peak is than its contribution to group peak.
D = diversity factor = I/coincidence factor (3.3)
The coincidence curve shown in Figure 3.7 is typical of residential peak load behavior
in the United States - C(n) for large groups of customers is typically between .5 and .33,
and may fall to as low as .2. Coincidence behavior varies greatly from one utility to another.
Each utility should develop its own coincidence curve data based on studies of its own
customer loads and system.
Estimates of equipment loads are often based on load data from coincidence curves.
Very often, T&D equipment sizes are determined by using such coincidence curves to
convert load research data (or whatever data are available) to estimates of the equipment
peak loads. For example, the "coincident peak" for a group of customers served by a feeder
segment, transformer, or secondary line can be estimated from individual ("instantaneous")
peak load data, as
Group peak for n customers = C(n) x n x (average individual peak load) (3.4)
The peak load for a transformer serving eight houses, each with an estimated individual
peak load of 22 kW would be
Transformer peak load = C(8) x 8 x 22 kW (3-5)
Using a typical value of C(8), which is about .6, gives 106 kW as the estimated peak load
for the transformer.

22 o

cc
LU

O
^
a
a:
o
O
O
5 10 100 1000
NUMBER OF CUSTOMERS IN THE GROUP

Figure 3.7 The peak load per customer drops as a function of the number of customers in a group.
Although load characteristics vary from one utility system to another, this curve is representative of
typical behavior. Generally, C(2) is about .85 to .8 and C(large number) is between .5 and .3.


60

LU
CL

§
o:
cc

5 10 100 1000
NUMBER OF CUSTOMERS IN THE CROUP

Figure 3.8 The peak period as a function of the number of customers in a group. Larger groups
generally have longer peak load periods.

Often, reliable estimates of the individual customer peak loads (the 22 kW figure in
these examples) may not be available. In this example, that would mean the planner did not
know the 22 kW value of individual peak load but instead had only the 'contribution to
peak1 value of 7.9 kW/customer. The transformer load can still be estimated from the
coincidence curve as
Transformer peak load = C(N)/C(8) x 8 x 7.9 kW (3.6)
where N>»8
Here, using .36 as C(N) gives the same 106 kW
Peak period lasts longer for larger groups of customers. The erratic, up and down load
curve of a single customer may have a very high peak load, but that peak lasts only a few
minutes at any one time, although it is reached (or nearly reached) several times a day. By
contrast, the load curve for one hundred homes reaches its peak only once during a day but
stays at or near peak much longer - perhaps for over an hour. Figure 3.8 shows the peak
load duration - the longest continuous period during the day when the peak load is at or
within 5% of maximum - for various numbers of customers.
Coincident Load Curve Behavior Can Be Viewed
as the Expectation of a Single Household's Load
What does the smooth, average residential customer curve, one computed from Figure 3.1
as described earlier, indicate? As stated, it is the average, coincident customer load curve,
1/44,000 of the load from 44,000 households. While no customer within the entire utility
service territory has an individual load curve that looks anything like this curve, the
application of a smooth load curve computed in this way has two legitimate interpretations
on a per customer basis:
1. The curve is an individual customer's contribution to system load. On the
average, each customer (of this class) adds this much load to the system. Add
ten thousand new customers of this type, and the system load will increase by
ten thousand times this smooth curve.

94 Chapter 3

2. The curve is the expectation of an individual customer's load. Every customer
in this class has a load that looks something like the erratic behavior shown on
Figure 3.1, but each is slightly different and each differs slightly from day to
day. The smooth curve gives the expectation, the probability-weighed value of
load that one would expect a customer in this class, selected at random, to have
at any one moment, as a function of time. The fact that the expectation is
smooth while actual behavior is erratic is a result of the unpredictable
randomness of the timing of individual appliances.
The behavior discussed above is infra-class coincidence. Commercial and industrial
customers have similar behavior to that discussed above for residential loads, but the shape
of the their coincidence curves may be different. Qualitatively the phenomenon is the same,
quantitatively it is usually quite different.
One can also speak of coincidence between classes. Figure 3.1 showed average or
typical coincident curves for residential, commercial, and industrial customer classes in a
summer peaking utility in the southern United States. These various classes peak at
different times of the day: commercial in mid-afternoon, residential in late afternoon or
early evening. A substation or feeder serving a mix of customer types will not see a peak
equal to the sum of the individual residential and commercial peaks, but instead a lower
peak, because of the diversity, or lack of coincidence, between the classes' peak loads.
Coincidence and its effects on planning will surface again and again throughout this
book, and we will make reference to coincidence, and discuss its impact on planning when
we discuss how to determine feeder and transformer capacity; what loads to use in feeder
voltage drop and losses studies; how to site and size substations; which load curves are
appropriate for certain special applications; and how to forecast load growth. Therefore, it
is vital that the reader understand the concept and phenomenon of load coincidence, both
within each customer class and among the classes
3.3 MEASURING AND MODELING LOAD CURVES
The manner in which load curve data are collected, recorded, analyzed, and represented can
produce a dramatic effect on what the resulting load curves look like and the perceived
values of peak load and coincidence. The apparent load curve shape can change depending
on how the load data are measured and how periodically these measurements are recorded.
Sampling rate and sampling method both have a big impact on what load curve data
look like, how accurate they are, and how appropriate they are for various planning
purposes. Sampling rate refers to the frequency of measurement - the number of times per
hour that load data are recorded. Sampling method refers to exactly what quantity is
measured - instantaneous load or total energy used during each period.
Most load metering equipment measures the energy used during each period of
measurement (called demand sampling). If the load is measured every fifteen minutes, the
equipment measures and records the total energy used during each 15 minute period, and
the resulting 96 point load curve gives the energy used during each quarter hour of the day.
Plotted, this forms a 24 hour demand curve. This method of recording is sampling by
integration - the equipment integrates the area under the load curve during each period and
stores this area, even if the actual load is varying up and down on a second-by-second basis.
By contrast, discrete instantaneous sampling measures and records the actual load at the
beginning of each period. Essentially, the equipment opens its eyes every so often, records
whatever load it sees, and then goes to sleep until the beginning of the next period. Some
load monitoring equipment uses this type of sampling.


Period integration Discrete instantaneous
22 kW 22 kW

Hour Hour
Figure 3.9 Load for a single household (the load curve in Figure 3.1) sampled on a fifteen minute
interval by both period integration (left) and discrete instantaneous sampling (right). Neither curve is
fully representative of actual behavior, but the period integration is more accurate for most purposes.
The discrete sampling is nearly worthless for planning and load study applications.

Discrete instantaneous sampling often results in erratically sampled data that
dramatically misrepresents load curve behavior. The right side of Figure 3.9 shows a data
curve for a single home, produced by a discrete instantaneous sampling technique applied
on a fifteen-minute basis to the individual household data first shown in Figure 3.2. This
"load curve" is hardly representative of individual, group, or average load behavior. It does
not represent hourly load behavior in any reasonable manner. This data collection error
cannot be corrected or counteracted with any type of subsequent data processing.
What happened is that the very instant of measurement, which occurred once every
fifteen minutes, sometimes happened upon the very moment when a sharp load spike of
short duration was at its peak. Other times, the sampling stumbled upon a moment when
energy usage was low. This type of sampling would yield a good representation of the load
behavior if the sampling were more rapid than the rapid shifts in the load curve. But
sampling at a fifteen minute interval is much too slow — the load curve can shoot back and
forth from maximum to minimum several times within fifteen minutes. As a result, it is just
random luck where each period's sampling instant happens to fall with respect to the actual
load curve behavior — and a portion of the sampled load curve data is essentially randomly
selected data of no real consequence or value.
Automatic load metering equipment seldom uses discrete instantaneous sampling.
Almost all load recorders use demand sampling (period integration). However, many
sources of load data do start with instantaneous discrete sampled data, particularly because
human beings usually do use this approach when gathering data.
For example, data written down by a technician who reads a meter once an hour, or by a
clerk who takes hourly load values off a strip chart by recording the reading at each hour
crossing tick mark, have been sampled by this method. So, too, have data that have been
pulled from a SCADA monitoring system on an hourly basis by programming a system
macro to sample and record the power flow at each bus in the system at the top of each
hour.
In all such cases, the planner should be aware of the potential problem and correct for it,
if possible. The technician taking hourly values off the strip chart can be instructed to

96 Chapter 3

estimate the average of the plotted value over the entire hour. The SCADA system should
make several readings separated by several minutes and average them.
To be entirely valid, load curve data gathered using discrete instantaneous sampling
must satisfy the Nyquist criteria, which requires that the load be sampled and recorded at
least twice as often as the fastest load shift occurring in the measured load.2 When
measuring a single customer, whose load can shift up and down completely within a few
minutes, this means a sampling rate on the order of a minute or less must be used. If
taken at one minute intervals, such data are accurate, but even at ten minute intervals, a
rate often considered very fast for load research purposes, discrete sampling will produce
very poor results when applied to individual or small groups of customers.
As was discussed earlier, larger groups of customers have smoother and more well-
behaved load curves. Thus, the problem discussed here is not as serious when the load
being metered is for a feeder or substation as when the load being measured is for only a
few customers, although it can still produce noticeable error at the feeder level. At the
system level it seldom makes any difference.
A telltale sign that discrete sampling data errors are present in load data is erratic load
shifts in the resulting load curves, like those shown on the right side of Figure 3.9.
However, lack of this indicator does not mean the data are valid. Hourly, half hour, or even
fifteen minute load research data gathered via discrete instantaneous sampling should be
used only when absolutely necessary, and the planner should be alert for problems in the
resulting load studies.
Sampling by integration, by contrast, always produces results that are valid within its
context of measurement. For example, hourly load data gathered by period integration for
an individual customer will accurately reflect that customer's average energy usage on an
hour by hour basis. Whether that sampling rate is sufficient for the study purposes is
another matter, but the hourly data will be a valid representation of hourly behavior.
Figure 3.10 shows the daily load curve for one household (that from Figure 3.2) demand
sampled (period integration) at rates of quarter hour, half hour, one hour, and two hours.
None give as accurate a representation of peak load and peak period as the original curve
(which was sampled at one minute intervals), but each is an accurate recording usage
averaged to its temporal sampling rate. Such is not the case with discrete sampling, where
the recorded values may be meaningless for any purpose if sampled at a rate below the
Nyquist rate for the load behavior being measured.
Figure 3.11's curve of peak load versus sampling rate bears a striking resemblance to the
peak coincidence curve plotted in Figure 3.7, an interesting potential source of error, which
will be discussed later in this chapter. As the sampling period of demand sampling is
lengthened, the observed peak load drops, as shown in Figure 3.11. The sharp needle peaks
are being averaged with all behavior over longer periods, reducing the measured "peak
demand."

2
The author has taken slight liberties with the interpretation of the Nyquist criteria (also called the
sampling theorem), but from a practical standpoint the description is more than sufficient to explain
application to load curve measurement. A more theoretical presentation of load curve sampling
rates can be found elsewhere (Willis, Vismor, and Powell, 1985).

22 kW 22 kW
15 Min. 30 hin.

10 kU 10 KW

Hour Hour
Figure 3.10 Daily load shape for the household load curve in Figure 3.1 demand sampled (i.e., using
period integration) on a 15, 30,60, and 120 minute basis.

22

oc
LU

Cfl
ID
O
^s.
Q
<r
o

0 30 60
LENGTH OF SRMPLING PERIOD - MINUTES

Figure 3.11 Observed peak load is a function of time period of sampling for the single household's
daily load curve, as a function of the sampling period. Only high sampling rates can see the needle
peaks and thus correctly sample their full behavior.

98 Chapter 3

10 kW 10 kW
15 Min,

10 kW

Hour Hour
Figure 3.12 Daily load shape for 100 homes demand sampled with a 1,15, 30, and 60 minute period.
Little error is introduced by using a one hour sampling rate as opposed to 1 minute, because the basic
coincident load behavior of 100 households' load is inherently smooth, not erratic and full of needle
peaks as in the single household case. Note vertical scale is the same in all four plots in this figure.

10

i
DC
111
E
O
H
(f)
•3
U
•--.
a
a:
o

60 120
LENGTH OF SRMPLING PERIOD - MINUTES
Figure 3.13 Measured per customer coincident peak load for a group of 100 homes, as a function of
the sampling period used with period integration. For any sampling period less than 60 minutes, the
measured peak load varies little, indicating that the load curve does not need to be sampled more often
than hourly.


High Sampling Rates Are Needed When
Sampling Small Groups of Customers
In Figure 3.12 sampling rate is again varied, this time for a load curve representing the sum
of loads from 100 homes. The measured curve varies insignificantly as sampling rate is
increased. The smooth, coincident load curve for a large group of customers does not need
to be sampled at a high rate in order to be accurately recorded. High sampling rates are
needed only when studying the non-coincident load behavior of small groups of customers.
Figure 3.13 shows the measured peak load of this group of 100 homes as a function of
the demand period used in measurement. Clearly, little error is introduced with sampled
periods up to one hour in length. Again, high sampling rate is not needed when measuring
aggregate load behavior for large groups of customers.
Comparison of the 60 minute sampled curves from Figures 3.10 and 3.12 shows that
they are very similar. An individual customer curve demand sampled at a slow rate yields a
good estimate of the coincident curve shape. In essence, measuring an individual customer's
load at a lower rate with period integration gives a picture of that customer's contribution to
overall coincident load behavior. Depending on the goals of the load measurement, this can
be a positive or negative aspect of using the lower sampling rate.
Determining the Required Sampling Rate
Recall that the measured peak load versus sampling period curve for a single customer
(Figure 3.11) bears a striking resemblance to the actual peak load versus number of
customers coincidence curve (Figure 3.7). A good rule of thumb is: the demand sampling
rate required to measure the load for a group of customers is that which gives the same
measured coincident peak/customer when applied to an individual customer curve. For
example, suppose it is desired to sample the load behavior of a group of five homes to
determine peak loading, daily load curve shape, and so forth. Note that in Figure 3.7, C(5) =
.8, giving a coincident load of .8 x 22 kW = 17.6 kW per household for a group of five
homes.
As Figure 3.11 shows, when applied to an individual household load curve, a sampling
period in the range of 10 minutes (6 demand samples/hour) should give a peak of 18 kW.
Therefore, one can conclude that a group of five homes needs to be sampled and analyzed
at least at this rate to be represented accurately.
Similarly, one can determine the smallest number of customers for which a particular
sampling rate provides accurate data. Suppose that half hour data are to be used in a load
research study. Figure 3.11 shows that this can be expected to record a peak load of 10.9
kW when applied to a single customer load curve. This is equivalent to a coincidence factor
of 10.9 kW/22 kW, or .49, which is equal to C(25). Thus, half hour data are accurate (in this
utility) for studies of distribution loads and loading on any portion of the system serving 25
customers or more. It is not completely valid for studies on equipment serving smaller
groups of customers. In a similar way, a plot of sampling period versus number of
customers can be determined, as shown in Figure 3.14.
This example is qualitatively generalizeable but the values shown are not valid for
systems other than the one used here - average individual household needle peak of 22 kW,
coincident behavior as plotted here, and so forth, vary from one utility to another and often
by area within utility systems. These data were taken from a utility system in the Gulf Coast
region of the United States, and the results given are valid for that system. Quantitative
results could be far different for systems with different customer types, load patterns, and
coincidence behavior (EPRI, 1990).

100 Chapter 3

60
30
15

u_ i w
~l tr
Q O> t>

o cc
2^ LLl
UJ Q-
I
1 |
1 5 10 100 1000
Iog sea Ie

Figure 3.14 Rule-of-thumb plot of sampling period (for integrated sampling) versus number of
customers in the group.

Estimating Coincidence and Sampling Requirements

Often, a planner does not have even a few samples of high resolution (very short sampling
period) load data from which to make the type of analysis described above. In that case, the
author recommends that the coincidence curve (Figure 3.7) and the peak versus sampling
rate curve (Figure 3.11) be estimated as best as can be done from whatever data are
available. For example, suppose only fifteen minute period load curve data are available.
Added together in blocks of two and four periods, this gives thirty and sixty minute
sampled curves, whose peak loads can be used to provide the peak versus sampling rate
curve in the 30 to 60 minute sampling period portion of the curve.
Estimates of the absolute individual household peak load can often be derived from
knowledge of the types and market penetrations of appliances or by the planner's judgment
to give an estimated value for the Y intercept of the coincidence curve. Measured peaks on
distribution equipment such as service transformers and feeder line segments provide detail
on the peak loads of groups of customers of various sizes. Analysis of sampling period
versus number of customers in the group can then be carried out as outlined earlier.

L. I W
°Q O> 5
xo o

5 10 100 1000
log scale

Figure 3.15 Example of estimating the sampling rate versus peak load relation from incomplete data,
as explained in the text. The solid line indicates areas of the relationship that can be determined
exactly based on available data; the dashed line shows the estimated portion. As a general rule, the
curve passes through the origin, the point (1, 1), a fact useful in estimating the curve shape.



Most individual-consumer load curves are volatile, with sharp "needle peaks" and
erratic shifts in load as major appliances such as central heating and water heaters cycle on
and off.
"Bang-bang" control systems cycle major appliances and equipment on and off in
response to thermostats or pressure monitors, etc., of cycle equipment.
Duty cycle is the percent of the time that an appliance or device operates in response to its
cycling control.
Load curves for groups of customers are smoother - less volatile - because the needle
peaks and valleys interleave
Interleaving of needle peaks and valleys in a group means that as more customers are
added to the group the peak load per customer drops and the duration of any one peak
period increases.
Distribution equipment that serves larger numbers of customers sees smoother daily load
curves.
Coincidence factor, C(N), is a measure of the simultaneity of peak demands of a group of
N customers. Its inverse is diversity factor. Peak load for the group equals N x individual
peak demand x C(N). Typical values of C(N) for N > 100 are in the range of .20 to .50.
The coincident load curve can be viewed as the expectation of a customer's non-
coincident curve.
Rapid sampling of a load curve is necessary to "see" non-coincident load curve
behavior. Fully capturing all the volatility may require sampling electric load every
minute throughout the day.
Sampling by demand period involves measuring and recording the total energy in each
uniform period (e.g., 15 minutes, 60 minutes).
Discrete period sampling, by contrast, measures the load only at the beginning of each
period, perhaps for only one second every fifteen minutes.
The two sampling methods differ in what they "see" as their sampling rate is varied:
If demand period sampling is applied to a very volatile non-coincident load curve at
a relatively slow sampling rate (slower than the curve's back-and-forth swings) it
"sees" - records - a curve that is very much like the coincident version of the load
curve. The data may be useful as an approximate measure of fully coincident load
curve shape and peak.
If discrete period sampling is applied to a very volatile non-coincident load curve at
a relatively slow sampling rate (slower than the curve's back-and-forth swings) it
"sees" - records - only corrupted data, which is utterly worthless for any purpose
even if it looks like a valid load curve.
High sampling rates are needed only when studying the non-coincident load behavior of
individual or very small groups of customers. The smooth, coincident load curves of a
large group of energy consumers do not need to be sampled at a high rate in order to be
accurately recorded.
Many algebraic and mathematical methods for data analysis "smooth out" load
curves - remove any quick shifts in load cycles - as an unintended consequence of their
application. Analysis of non-coincident load curves often produces smooth curves that are
not valid representations of the actual (non-coincident) load curve behavior.

102 Chapters

The approximate nature of the number of customers versus sampling period relation is
estimated from these data, as shown in Figure 3.15. While not of the quality of accurate,
high-sample-rate data, the resulting curve is the best available under the circumstances and
can be used, with judgment, to make decisions about the legitimate uses of whatever load
data are available.

3.4 SUMMARY

Distribution load behavior is dominated by coincidence, the fact that peak loads do not
occur simultaneously. Individual customer load curves are erratic, formed of fast, sharp
shifts in power usage. As customers are combined in groups, as when the planner analyzes
load for groups served by equipment such as service transformers, laterals, and feeders, the
erratic load curves add together with the sharp peaks intermingling and forming a smoother
curve. As a result, when comparing the load curves for different size groups of customers,
as the number of customers increases, the group load curve becomes smoother, the peak
load per customer decreases, and the duration of the peak increases. Also important is the
manner in which the load curves are measured. Where possible, particularly when working
with small groups of customers, data should be collected and analyzed using the period
integration method, and the sampling period should be quite short, generally less than 15
minutes. Table 3.1 summarizes key points from this chapter.

REFERENCES

H. L. Willis and T. D. Vismor, and R. W. Powell, "Some Aspects of Sampling Load Curves on
Distribution Systems," IEEE Transactions on Power Apparatus and Systems, November 1985, p.
3221.

Electric Power Research Institute, DSM: Transmission and Distribution Impacts, Volumes 1 and 2,
Palo Alto, CA, August 1990, EPRI Report CU-6924.

4
Power System Reliability

4.1 INTRODUCTION

"Reliability" as normally applied to power distribution means continuity of service to the
utility's customers. It is a critical element of customer service, one of the two Qs. A reliable
power system provides power without interruption. Almost anything less than perfection in
this regard will garner complaints that the utility's system is "unreliable." An availability of
99.9% might sound impressive, but it means eight and three quarters hours without electric
service each year, a level of service that would be considered unsatisfactory by nearly all
electric consumers in North America.
This chapter reviews distribution reliability concepts in general and introduces the key
components used in reliability analysis. It begins with a discussion of equipment outages
and service interruptions, and the two aspects of reliability that measure both - frequency
and duration of failure. The relationship between equipment outages and interruptions - the
extent of the service interruption caused by an outage - is presented and identified as one of
the key concepts for planning reliability of distribution systems. The most popular
reliability indices are discussed, along with their usefulness to the distribution planner and
the differences in their application among electric utilities. Finally, various ways of setting
reliability standards and criteria for distribution are covered. Planning methods to design
distribution systems to meet these or other reliability-related criteria are discussed
elsewhere in this book (see Index). Table 4.1 provides a glossary of standard reliability-
related terms. Regardless, "reliability" has only a single meaning throughout this book:
Reliability is the ability of the power delivery system
to make continuously available sufficient voltage, of
satisfactory quality, to meet the consumers' needs.
There are four important aspects to this definition that planners need to keep in mind when
laying out and optimizing their system.
Customer focused. In the modern business world, very few terms are as misused and
overused as "customer focused," but this is one case where it is apt. The focus of a
planner's reliability analysis must be on the service customers receive. Ultimately,
this is all that matters. Evaluation of and focus on reliability measured with respect to
the utility's system and equipment is not sufficient. The utility must assure itself that

103

104 Chapter 4

Table 4.1 Reliability Terms and Definitions

Availability: the fraction of the time that service is available. The (steady-state) probability that power
will be available.
Clear: to remove a fault or other disturbance-causing condition from the power supply source or from
the electrical route between it and the customer.
Cut set: in reliability analysis, a set of components which when removed from service causes a
cessation of power flow.
Dependability: used here, it means reliability of performance over the short-term, or with respect to
only a certain schedule: "Although available only during the day, the solar-power generator proved a
dependable source of power from 9 to 5 every day."
Durability: the ability to maintain high levels of reliability or dependability while in service over long
periods of time. Two units of equipment can be equally reliable, or dependable, when initially
installed, but the more durable will be in much better shape after ten years, etc.
Duration: the total elapsed time of an interruption or outage, as the case may be.
EEAR (Expected Energy at Risk): A calculated index based on system configuration; equipment
type, capacity, and condition; and load duration data that predicts the percent of annual energy at risk
of not being delivered for a planned system and load.
Event: a specific change in equipment status, load, or operating state or in environmental and ambient
conditions.
Extent: the amount of the system, usually measured as the number of customers, rendered out of
service by an event or failure.
Failure: a change in status of equipment or a system from performing its function as expected to no
longer performing its function completely.
Failure rate: the average number of failures of a component or unit of the system in a given time
(usually a year).
Frequency: how often interruptions or outages, as the case may be, occur.
Interruption: a cessation of service to one or more customers, whether power was being used at that
moment or not.
Instantaneous interruption: an interruption lasting only as long as it takes for completely
automatic equipment to clear the disturbance or outage. Often less than one second.
Momentary interruption: an interruption lasting only as long as it takes automatic but manually
supervised equipment to be activated to restore service, usually only a few minutes, sometimes
less than fifteen seconds.
Planned interruption: an interruption of power due to a planned outage, of which customers
were informed reasonably far in advance.
Temporary interruption: an interruption manually restored, lasting as long as it takes.
Interruption duration: the time from cessation of service until service is restored.
Interruption frequency: the average number of times interruption of service can be expected during
some period (usually a year).
Mean time between failures (MTBF): the average or expected time a unit of equipment is in service
between failures.
Mean time to repair (MTTR): the average or expected time to repair an outaged component or unit
of the system once failed.

Power System Reliability 105

Table 4.1, continued
Minimal cut set: a cut set (of components) that does not contain any other cut set as a subset.
Misoperation: incorrect operation of the system whether due to human error, the incorrect operation
of a relay or control system, or a binding or broken part in a device. This flawed operation may cause
an interruption of customer service, as with the inadvertent opening of a breaker. Regardless, the
equipment result is called "an outage." Even though it may not be damaged, the unit is out of service.
Outage: the failure of part of the power supply system - a line down, transformer out of service, or
whatever else is intended to be in operation but is not - whether due to unexpected or planned
circumstances.
Expected Outage - an equipment outage that was anticipated in some sense. Application of this
differs from utility to utility, but generally this means something more definite than "Well, we
knew it would fail eventually." Often, this is applied to equipment failures that were anticipated
and for which preparations had been made: "Diagnostics indicated that cable failure was
imminent but not which section would fail. We had switched in a backup cable to continue
service when failure occurred, and went ahead and let the failure happen so as to identify the
bad section of cable."
Forced Outage - an equipment outage that was not scheduled in advance, even if initiated by
the utility. The equipment may not be damaged but usually something went wrong: "The unit
was observed to be smoking slightly so it was withdrawn from service. It turned out just to be a
slight oil leak and the unit was repaired and returned to service the next day."
Scheduled Outage - an equipment outage that was scheduled in advance and initiated by the
utility's actions, as for example when work crews remove equipment from service for
maintenance, etc.
Unexpected Outage - an equipment outage that was not anticipated. Usually, an unexpected
outage is due to equipment failure, an accident, or damage of some type.
Reliability: the ability of the power delivery system to make continuously available sufficient voltage,
of satisfactory quality, to meet the consumers' needs.
Repair time: the time required to return an outaged component or unit of the system to normal
service. Related in a cause and effect way, but not necessarily the same as, interruption duration.
Reporting period: the period of time over which reliability, interruption, and outage statistics have
been gathered or computed.
Restoration: return of electric service after an interruption, because of repair of the outage causing the
interruption or because of re-switching of supply source.
Restoration time: the time required to restore service to an interrupted customer or portion of the
system. Also, in T&D operations, the time required to return equipment to service in a temporary
configuration or status (i.e., not completely repaired) as for example an overhead feeder put back into
service with stubbed poles and slack spans temporarily in place. It has been restored to service, but not
repaired.
Service: supply of electric power in sufficient amount and quality to satisfy customer demand.
Storm: a weather-related event or series of events causing outages who consequences meet specific
criteria for "storm," usually measured by number of customers out of service, amount of repair work
needed, or customer-hours of interruption, and reported as a separate category of interruptions.
Switching time: the time required to operate devices to change the source of power flow to a circuit,
usually to restore service where interrupted due to an equipment outage.
Trouble: anything that causes an unexpected outage or interruption including equipment failures,
storms, earthquakes, automobile wrecks, vandalism, operator error, or "unknown causes."

106 Chapter 4

consumers see good reliability from their point of view when looking back into the
system, regardless of equipment issues.
Continuous availability. Perhaps the most important aspect of reliability is that
power most be continuously available. Basically, the utility must maintain an always
unbroken chain of power flow from a source of power to each consumer.
Sufficient power. This chain of power flow must be able to deliver enough power
that it does not constrain the consumer's ability to have as much as he needs
whenever he needs it.
Satisfactory quality. Voltage, frequency, and other aspects of the power made
available must be suitable to meet the consumer's needs.
This is what it really means to design and operate a "reliable" power delivery system.
Figure 4.1 illustrates this, based on a diagram (Figure 2.1) from Brown (2002). Reliability -
whatever guise it takes - is a part of "Power Quality" in the larger sense. Power quality in
turn is part of customer satisfaction, the real measure of a utility's good customer focus.
However, many of the other factors contributing to customer satisfaction, such as the
quality of billing and response to customer inquiries, are generally beyond the venue of the
planner.

Customer Satisfaction
Price Responsiveness

Power Quality

Reliability
—< ' !*««

Availability
Interruption Duration

Harmonic Distortion
Public Image Community Role

Figure 4.1 One can look at consumer needs and reliability-engineering concepts as contained within
one another, as shown. Generally, "reliability planning" focuses on reliability, which includes
availability as an important element. Aspects of power quality not related to reliability are generally
addressed only tangentially, as they normally work out in systems designed to provide sufficient
reliability of capacity, as the equipment and units used in the system have been designed so that will
occur.


4.2 OUTAGES CAUSE INTERRUPTIONS
The single most important distinction to make in distribution reliability analysis is between
two words that are often used interchangeably: outage and interruption. They have far
different meanings. "Outage" means a failure of part of the power supply system - for
example a line down, transformer out of service, or a breaker that opens when it shouldn't.
"Interruption" means a cessation of service to one or more customers. Interruptions are
almost always caused by outages, and this leads many people to confuse the two words and
their meanings. However, to understand distribution reliability properly and, more
importantly, set criteria and design a system to achieve high reliability, the distinction
between outage (the cause of service problems) and interruption (the result) is critical.
Electrical service to a customer is interrupted whenever equipment in the line of supply to
that customer fails or is otherwise out of service. Outages cause interruptions.
Frequency and Duration
Two different aspects of reliability receive attention in any type of power distribution
reliability analysis, whether of outages or interruptions. These are the frequency (how often
something occurs) and duration (how long it lasts). With respect to interruption, frequency
refers to the number of times service is interrupted during a period of analysis - once a
decade, two times a year, five times a month, or every afternoon. Duration refers to the
length of these interruptions; some last only a few cycles, others for hours, even days.
Frequency and duration are used with regard to outages, too.
Extent
Frequency and duration are important factors in reliability analysis and design, but there is a
third factor equally important to the distribution planner because it is the one over which the
planner has considerable control, particularly in the design phase. The extent of an outage -
how many customers are interrupted by the outage of a particular line or unit of equipment
- is the key to the outage-interruption relationship. The layout of a distribution system
greatly influences just how many customers are interrupted when outages occur. Consider
for a moment two hypothetical distribution systems, A and B. During a particular year, both
experience the same number of distribution failures. In both, it takes just as long for repairs
to be made and service to be restored. But suppose that in system A, the average equipment
outage cuts off service to 150 customers, while in system B, it cuts off service to only 75. In
this case, system B's customer interruption statistics will be only half of A's, even though
their equipment-level failure rates are the same. The extent to which outages spread
customer interruptions through a distribution system is very much under the control of the
distribution planner - among other things it is a function of how the distribution system is
laid out. Various types of distribution designs and layouts behave differently in terms of
extent of outages. In general, a system laid out to minimize extent of outages from any
failure will be more expensive than one that is not, but for the planner this means that
consideration of configuration and its influence on reliability is just one more alternative to
be considered in planning. In many cases a higher-reliability layout may be less expensive
than any other way of achieving the particular reliability goal. Most distribution systems use
some form of radial feeder and secondary layout, which means that a single failure
anywhere in the electrical path between substation and customer will cause service
interruptions to all customers downstream. (There are distribution layouts so robust that
one, two, or even three equipment failures will not lead to a single customer interruption.
That capability is due to a secondary network properly combined with an interlaced primary
radial feeder set as discussed in Chapters 1 and 19.) One way to improve reliability in a

108 Chapter 4

distribution system is to lay out the feeder trunk and branches and arrange the protection
(fusing and sectionalization) so that outages have a low extent, i.e., so that most failures do
not interrupt service to very large numbers of customers.
Control of extent through layout design
Figure 4.2 and Table 4.2 show two different configurations for feeder layout to reach the
same customer points. Although simplified, this example illustrates well both the concept
and the typical magnitude of differences in costs and reliability between layouts.
Configuration B gains a 6.76% improvement in customer interruptions over configuration
A, even though it has a 4.8% higher expected rate of equipment outages. (Expected outages
are 4.8% higher because configuration B has 4.8% more miles of feeder length.) Also quite
important, the worst number of interruptions any customer on the feeder can expect (i.e., its
worst case customer) is 12.6% less.
The reason for the difference is due to variations in the extent of outages that can be
expected on the feeders. In configuration A, if the second trunk feeder segment fails,
service to 300 customers is interrupted. That trunk is a mile long and expected to fail once a
year. By contrast, in configuration B, the maximum extent of any outage is 200 customers,
for either of the two major branches, which have expected failures of .5 per year (for one
mile total). This single difference in extent reduces expected customer outages by 100
customer interruptions per year. Configuration B has slightly longer branches to its most
outlying nodes (1.18 miles final run instead of 1.0, which increases outage rate slightly, but
overall its layout results in a net decrease of 76 customer interruptions per year (6.76%
less), due only to its different layout. Configuration B has better reliability from the
customer standpoint than configuration A because, on average, fewer customer
interruptions occur whenever an outage happens.

B

261.8
300

150
11

.5

D- 100 200
D- 100 200
.5 .5

150

1.18
300 261.8

One mile

Figure 4.2 Simplified illustration of two ways a radial feeder can be laid out to serve the same 600
customers and how configuration alters reliability. Each node represents 100 customers (identical in
both feeder layouts). Both feeders have a failure rate of once per year per mile. A fuse protects the rest
of the feeder from outages behind every node. Numbers shown at the middle of each segment are the
expected failures for that segment. Numbers shown at nodes are the total customer interruptions/year
expected at that node assuming the protection works.


Table 4.2 Differences Between the Feeder Layouts A and B in Figure 4.2
Statistic A B Difference
Miles of conductor 5.00 5.24 +4.8%
Feeder failures/year 5.00 5.24 +4.8%
Customer interruptions/year 1,200 1,124 -6.8%
Maximum interruptions/customer 3.00 2.62 -12.6%
Estimated relative cost 1.00 1.03 + 3.0%

The difference in outage extent between configurations A and B really impacts only the
frequency of outages. The average customer on feeder configuration B is 4.8% less likely to
be interrupted. The expected total duration of customer interruptions - the number of
customer hours of expected customer interruptions annually - is also lower by 4.8%, but
only as a consequence of there being fewer customer interruptions in the first place.
Changes in layout and fusing that reduce the expected extent of outages work on only the
frequency of customer interruptions. They do little if anything to reduce duration of
interruption, or to shorten the time to repair or restore service (assumed as one hour in all
cases in Figure 4.2).
Changing the layout and protection scheme for a feeder to lower extent of outages is
often the distribution planner's chief means to reduce frequency of interruptions. Other
means of improving customer reliability, such as providing additional contingency margin,
arranging multiple re-switching zones, and automation to reduce duration, do not reduce
frequency of interruptions.

Types of Interruptions
Interruptions can be classified by their duration. Table 4.3 lists typical interruption
definitions and averages from the author's survey of usage throughout the power industry.
Practice and interpretation vary from utility to utility. In the author's opinion these
definitions are somewhat unsatisfactory. The names imply that they relate to interruption
duration, yet they are defined by equipment and operating factors, such as whether an
interruption is restored automatically or manually. This, along with differences in exactly
how the definitions are interpreted, means that application, for example the use of
"momentary interruption," varies greatly from utility to utility. In addition, the typical
definitions shown in Table 4.3 strike the author as too lengthy to match their names.
"Instantaneous" as used by many utilities encompasses interruptions of up to 20 seconds. In
the author's opinion, "instantaneous" implies no more than a second. Similarly,
"momentary" is used for durations of up to three minutes, whereas a minute seems more
appropriate.
Perhaps more important, these distinctions mean nothing to electric consumers, so they
do little to help a utility improve service from the customer's perspective. A four-minute
interruption is a four-minute interruption, whether restored automatically (making it
momentary) or manually (making it a "temporary" interruption).

110 Chapter 4

Table 4.3 Traditional Type Classification for Interruptions
Type Definition
Instantaneous An interruption restored immediately by completely automatic equipment,
or a transient fault that causes no reaction by protective equipment.
Typically less than 15 sec.
Momentary An interruption restored by automatic, supervisory, or manual switching at a
site where an operator is immediately available. Usually less than three
minutes.
Temporary An interruption restored by manual switching by an operator who is not
immediately available. Typically, thirty minutes.
Sustained Any interruption that is not instantaneous, momentary, or temporary.
Normally more than an hour.

Beyond the distinctions shown in Figure 4.3, interruptions are classified into scheduled
interruptions (the utility scheduled the interruption for maintenance purposes) or forced
interruptions (the interruption was not expected and not scheduled). Implicit in these
definitions, and the most important factor in the author's mind, was whether customers
were informed reasonably far in advance of the details of any "scheduled interruption." The
combination of the distinctions with the duration classifications creates such subcategories
as "scheduled temporary outage," etc.
Every planner should pay close attention to the definitions used by his company and by
other utilities to which his system's reliability is compared. What appear to be identical
indices can vary because of differences in how interruptions are defined. Many utilities do
not include scheduled outages, storms, and other "non-failure-related" interruption in their
reporting procedure.1 Others leave momentary interruptions out or report them as a special
category.
Such differences in definitions, interpretations, and reporting practices can frustrate
distribution planners who try to find some real meaning in comparisons of their company's
reliability experience to that of other utilities. Worse, these differences often cause
confusion and frustration among customers, regulators, and industry standards
organizations. A very common situation that leads to frustrated customers is the following:
a disgruntled factory manager calls his electric supplier to complain about six interruptions
that have occurred during the last five weeks. The electric utility's representative stops by
the next day with operating statistics that show that no interruption in the plant's service
occurred during the last six months.

In the past, many utilities remove some "storms" from reliability statistics, arguing that outages
and interruptions caused by hurricanes, tornadoes, and earthquakes should be removed from
statistics that are meant to reflect long-term trends in satisfying normal reliability needs. There is
some merit in this concept, but the removal of "storm-caused" interruptions has been carried to
extremes, such as a utility which claimed that all line-down failures during the three months
following a hurricane were due to stress put on overhead lines by the hurricane's high winds. As a
result, statistics for that year were the best ever recorded. What makes more sense, and is a
recommended practice, is for a utility to separate "normal weather" and "extreme weather"
performance statistics. The first indicates the quality of its operations with respect to system design,
maintenance, and normal operations. The second indicates how it manages storm response,
restoration, and rebuilding, activities that some utilities (Progress Energy in North Carolina) have
shown can be managed with far above average results due to focus and careful planning.

Table 4.4 Duration Limits Used to Define Instantaneous, Momentary, and
Temporary Outages at Eight Utilities
Utility Instantaneous Momentary Temporary
1 not used < 5 minutes 5 min. to 45 min.
2 < 15 sees. 15 sec. to 1 min. 1 min. to 45 min.
3 not used up to 30 seconds 30 sec. to 30 min.
4 < 15 seconds 15 sec. to 2 min. 2 min. to 1 hour
5 < 60 seconds 60 sec. to 5 min. 5 min. to 2 hours
6 not used < 1 minute 1 min. to 52 min.
7 < 1 minute not used 1 min. to 2 hours
8 "cleared by inst. relay" up to 2 minutes 1 min. to 1 hour

Both factory manager and utility representative are "correct," but the utility's definition
counted only temporary and sustained interruptions, while the factory manager was
concerned because interruptions lasting less than a minute could lock up robotic equipment
and bring other computerized operations to a standstill. At the present, this leads to
disgruntled customers. As the power industry switches to open access and competition
between suppliers, it will lead to customers who switch to another supplier. This is one
reason reliability will become more important in distribution planning and operations.
Another point every distribution planner must bear in mind when working with
reliability statistics is that utilities which report a certain type of interruption statistic, such
as number of momentary operations, vary widely in how they interpret that and other
definitions, as shown by Table 4.4, which lists interruption definitions from eight utilities in
North America.
4.3 RELIABILITY INDICES
In order to deal meaningfully with reliability as a design criterion for distribution, it is
necessary to be able to measure it and set goals. A bewildering range of reliability indices
or measures are in use within the power industry. Some measure only frequency of
interruption, others only duration. A few try to combine both frequency and duration into a
single value, which proves to be a nearly impossible task. Some measures are system
oriented, looking at reliability averaged over the entire customer base. Others are customer
or equipment oriented, meaning that they measure reliability only with regard to specific
sets of customers or equipment. To complicate matters further, most utilities use more than
one reliability index to evaluate their reliability. The large number of indices in use is
compelling evidence that no one index is really superior to any other. In fact, the author is
convinced that no one index alone is particularly useful.
The basic problem in trying to measure reliability is in how to relate the two quantities,
frequency and duration. Are two one-hour interruptions equivalent to one two-hour
interruption, or are two one-hour interruptions twice as bad as the one two hour
interruption? Most people conclude that the correct answer lies somewhere in between, but
no two distribution planners, and no two electric consumers for that matter, will agree on
exactly how frequency and duration "add up" in importance. For one thing, the importance
of frequency and duration vary tremendously from one electric consumer to another. There
are some industries where a five minute interruption is nearly as damaging to productivity

112 Chapter 4

as a one-hour interruptions - their computer and robotic equipment requires hours to re-set
and re-start, or they have a long production process so sensitive to power fluctuations that
they have to start over at the beginning if there is even a slight interruption. For these
customers five five-minute outages are much more serious than a single one-hour outage.
But there are other customers for whom short outages cause no significant problem, but
who experience inconvenience during a sustained interruption: factories or processes where
production is only significantly reduced if equipment has time to "cool down." Interruption
of power to a heater or tank pressurizer for a minute or two is not serious, but sustained
interruption allows a cool-down or pressure drop that results in reduced production.
Frequency of interruption has becoming increasingly emphasized by utilities and their
customers, partly because of the widespread use of digital clocks and computerized
equipment. Even momentary interruptions cause blinking digital displays and disable
robotic and automatic digital equipment. Although this "digital" economy has contributed
to some of the increased emphasis and is often cited as a justification for a very high
reliability power system, the author believes frequency of interruption has grown in
importance for other reasons, as will be discussed in Chapters 21 and 28.
The amount of power interrupted is also an important factor. Some reliability
calculations and indices are weighted proportionally to customer load at the time of
interruption, or to the estimated energy (kWh) not delivered during the interruption.
The four most popular indices used in reliability analysis make no distinction of
customer size. They treat all customers alike regardless of peak demand, energy sales, or
class. These indices are SAM and CATFI, which measure only frequency, and SATDI and
CTATDI which measure only duration. Usually these four are used in conjunction with one
another - four numbers that give a rough idea of what is really happening to reliability
system-wide. These four indices, and most others, are based on analysis of customer
interruptions during some reporting period, usually the previous month or year. All count a
customer interruption as the interruption of service to a single customer. If the same
customer is interrupted three times in a year, that constitutes three customer interruptions. If
one equipment outage caused simultaneous interruption of service of three customers, that,
too, is three customer interruptions.
The "S" in SATFI and SAUDI means these two indices average their interruption
statistics over the entire customer base (the system), the "C" in CAIFI and CTAIDI means
those indices refer to only customers who experienced interruptions. Customers who had
uninterrupted service during the period are precluded from CTAIDI and CAIFI analysis.
System Average Interruption Frequency Index (SAIFI) is the average number of
interruptions per utility customer during the period of analysis. This is just the total number
of customer interruptions that occurred in the year divided by the total number of customers
in the system,
0 . T_ number of customer interruptions ,. ,.
SAM= ——: : : 1 (4.1)
total customers in system
Customer Average Interruption Frequency Index (CAIFI) is the average number of
interruptions experienced by customers who had at least one interruption during the period.
While SAIFI might state that there where 1.3 interruptions per customer, perhaps half of the
customers experienced no outage. This means the rest had an average of 2.6
outages/customer. CAIFI tries to account for this by considering as its denominator only
those customers that had at least one interruption.
_ number of customer interruptions .. _.
~ number of customers who had at least one interruption


If there is a large difference between CAIFI and SAM, it means outages are concentrated
in only certain parts of the system or on certain customers. This could be due to poor
design, poor maintenance, differences in the weather among areas of the system, or just
plain bad luck, and further investigation will be required to determine cause and cure.
System Average Interruption Duration Index (SAIDI) is the average duration of all
interruptions per utility customer during the period of analysis. Here, the total customer-
minutes of interruption are added together and divided by the total number of customers in
the system.
S ATDT - sum Qf me durations of all customer interruptions .. ,,,
total customers in system
Customer Total Average Interruption Duration Index (CTAIDI) is the average total
duration of interruptions among customers who had at least one interruption during the
period of analysis.
_ sum of the durations of all customer interruptions
~ number of customers who had at least one interruption '
This particular computation, total number of customer hours out of service divided by the
number of customers who had at least one outage during the period, was traditionally called
CAIDI, customer average interruption duration index, without the "T." However, recently
many planners in the power industry have tended to compute CAIDI as the total number of
customer hours of interruption divided by the total number of customer interruptions, which
yields just the average duration of a customer interruption.
sum of the durations of all customer interruptions
CAIDI = number of customer interruptions (4.5)
The difference between CTAIDI and that definition of "CAIDI" is that a customer who has
three outages is counted only once in the denominator of CTAIDI, but three times in the
denominator of "CAIDL" This term, "CAIDI" is an example of how an incorrect popular
use of a term can change its meaning. The value now associated with "CAIDI" - average
duration of an interruption - is actually a statistic, not an index. Regardless, the more
commonly used definition of "CAIDI" now corresponds to the average customer
interruption duration. Note that CTAIDI/CAIFI = CAIDI.
If there is a large difference between CTAIDI and SAIDI, it means service outages are
concentrated on only part(s) of the system, with certain customers receiving many and some
others receiving few. This could be due to poor design, poor maintenance, aging
infrastructure in some areas, or just plain bad luck. Further investigation will be required to
determine cause and cure.
Momentary Average Interruption Frequency Index (MAIFI) is the average number of
momentary (and sometimes instantaneous) interruptions per utility customer during the
period of analysis. Typically, if a utility distinguishes momentary interruptions as a separate
class, then they are not included in the SAM and CAIFI statistics and MAIFI may be
computed separately as
number of customer momentary interruptions
MAIFI = totaj customers in system (4-6^

In such a situation, total interruptions (from the customer standpoint) are probably best
estimated by the sum: SAM + MAM.
Generally, interruptions are defined as momentary if they last less than some time limit,
the MAIFI threshold or the momentary interruption time limit. They are not counted into

114 Chapter 4

SAJQFT, but into MAIFI instead, and their duration is not included in SAIDI. Interruptions
with a duration beyond this limit are defined as sustained interruptions, and counted in the
SAIFI index instead of MAIFI, and their duration is included in SAIDI statistics. The
momentary interruption time limit used by utilities varies widely within the power industry,
from as little as one minute to long as half an hour. Five and fifteen minute momentary
interruption line limits are the most common. Chapter 14 will discuss how rapid switching
can affect reliability measures by "moving" interruptions from SAIFI and SAIDI counts to
the MAIFI count by reducing restoration times below the MAIFI threshold.

Load Curtailment Indices
In addition, a number of indices try to relate reliability to the size of customer loads in order
to weight the interruption of a large load more than a small load. These usually work with
kV A-minutes of outage using
Customer load curtailment = duration of outage x kVA unserved (4.7)
Customer Average Load Curtailment Index (CALCI) is the average interruption kVA
duration interrupted per affected customer per year.
sum of all customer load curtailments
~~ number of customers who had at least one interruption
Often, a very important statistic is the interruption rate and duration for the worst-case
customer or customers. While the system as a whole could have good reliability, a small
group of customers could still receive substantially lower reliability. CTATDI and CAIFI do
not always highlight such a situation, because the worst-case statistics could be mixed in
with hundreds, even tens of thousands, of other interruptions. For this reason, some utilities
record and use the worst case situations for their system evaluations.
Maximum Individual Customer Interruption Frequency (MICIF) is the maximum
number of interruptions experienced by any customer during the period. In some cases, to
average the impact of extreme or unusual cases, MICIF will be computed as the average of
the worst n customers with regard to number of outages during the period (n is typically
between 12 and 100).
Maximum Individual Customer Interruption Duration (MICID) is the maximum total
interruption time experienced by any customer during a period. Similar to MICIF, this
index is often averaged over the worst n customers in order to remove the impact of unusual
circumstances.
Usually, the same customer(s) is not involved with both MICIF and MICID, because
frequency and duration are tied to very different processes. The customer with the
maximum duration time in a year may have had only one or two outages, but be in a very
isolated location that takes quite some time to reach in order to effect repairs.

Availability Indices
Roughly 75% of utilities in North American also calculate and report monthly and annually
the availability of power: the percent of time, percent of customers, or fraction of demand
for which power was available. These statistics tell almost nothing about the system,
because they are always very close to 1.0 and the difference between acceptable (e.g.
99.97%) and non-acceptable performance (99.94%) is negligible as measured. Availability
indices expressed in this manner are popular chiefly because they look very impressive to
the uninitiated - as stated in the introduction of this chapter, a 99.9% availability statistic

looks very good, and this number exceeds the "very best" that can be done in many other
endeavors, but with respect to power it is unacceptable to most customers and industries.
Indices such as SAIDI, SAM, CTAIDI, and CAM are much more useful at tracking
reliability and indicating what and where improvement is needed.
Using Reliability Indices
Duration as given by SAIDI and CTAIDI and frequency as given by SAM, MAM, and
CAM are very different aspects of reliability, related separate aspects of the utility system
and its operation. Frequency, whether defined by SAM, CAM, or some other index,
measures how often failures occur. It is mostly a function of the causes of the outages and
their extent - the type and condition of T&D equipment and the layout of the system.
Duration, on the other hand, is almost exclusively a function of the utility's organization,
management, and resources for T&D repair.2 Generally, the planner has a great deal more
influence over layout and specification of equipment, factors that relate to frequency, not
duration. Therefore, distribution planners tend to focus mostly of SAM, CAM, MAM,
and MICH7. SAM is the simplest reliability measure, the most intuitively understandable,
and the one which holds the least counterintuitive surprises. It is perhaps the best single
indicator of a distribution system's relative "health" as regards equipment and design -
number of customers interrupted is dependent on the layout of the system and the condition
of its equipment more than it is on the utility's management of the restoration process.
By contrast, SAIDI and CTAIDI and duration indices in general are not so much a
measure of the condition of the distribution system and equipment as they are a measure of
how quickly the utility can restore service once a failure occurs. Doubling the repair crews
available to restore service and spacing them out over the system so that on average they are
closer to potential trouble spots will reduce repair times, and thus SAIDI and CTAIDI, but
it will do nothing to reduce SAM and CAM.
CAM and CTAIDI are only superficially similar to SAM and SAIDI. The point of
these two statistics is to distinguish how evenly distributed interruptions are over the
system, or how concentrated in only a few areas they may be. Note that for any one system
and period

SAM < CAM SAIDI < CTAIDI CAM > 1.0
Note that the ratio of SAM to CAM and SAIDI to CTAIDI is the same; each pair has the
same numerator, and they differ only in (the same two) denominators.
SAIDI/CTAIDI = SAM/CAM = fraction of customers who
experienced at least one
interruption in the period
Therefore, CTAIDI versus SAIDI tells the planner nothing more than CAM versus SAM.
Although many utilities use both, it actually provides no more information than using only
one, either CTAIDI or CAM.

For example, the duration indices can be cut dramatically by doubling the number of repair crews
standing by, and spacing them throughout the system so that the probability is high one or another
will be close to every outage. Time to reach outages and restore service will be cut, leading to a
reduction in SAIDI and CTAIDI. However, frequency of outages and the SAIFI and CAIFI indices
will be unaffected.

116 Chapter 4

(A
3
O
</>

I
Q_
3
L
it.
0)
+•«
c
65 70 75 80 85 90 95
YEAR

Figure 43 Three-year moving average of the annual number of interruptions for a region of a utility
in the central United States. This area grew from open farmland into a developed metropolitan suburb
during the period 1962 to 1970, during which time interruption rate rose due to the problems common
to areas where new construction is present (digging into lines, initial high failure rates of new
equipment and connections). Twenty years later, interruption rate began to rise again, a combination
of the effects of aging cable (direct buried laterals in some residential areas) and trees, planted when
the area was new, finally reaching a height where they brush against overhead conductor. The
anomaly centered around 1982 was due to a single, bad storm.

en 300

c
Z
200

0100
o
a
(53
1_
O

< n
Jan. July Jan. July Jan. July Jan.
1987 1Q88 1989
Month
Figure 4.4 Average hours out of service per interruption on a monthly basis for a rural utility in the
northern United States varies due to random fluctuations, differences in the number of days in each
month, and changes in weather. It also appears that average duration of interruption (dotted line), is
slowly increasing over time, but further study is necessary before this conclusion can be confirmed.


Analysis Using Reliability Indices
Reliability indices are applied in historical analysis to evaluate operating experience and
reveal trends or patterns, expose problems, and indicate how and where reliability can be
improved. They are used in predictive analysis to evaluate how well proposed solutions are
likely to solve the identified problems. Figure 4.3 shows frequency of customer interruption
(SAIFI) for one suburban operating district of a metropolitan utility on an annual basis over
a twenty-eight year period. The operating district was open farmland (i.e., with no trees) up
to 1962, when suburban growth of a nearby metropolitan region first spilled into the area,
with much construction and the typically high interruption rates characteristic of
construction areas (trenchers digging into underground lines, early equipment failures of
new distribution facilities, etc.).
Over the following ten years, as the area plotted in Figure 4.3 gradually filled in with
housing and shopping centers, and construction and system additions stabilized, frequency
of interruption dropped. Twenty years after the area first began developing, interruption rate
again began to rise because of two factors: gradually increasing cable failure rates due to
age and the effects of trees, planted when the area first developed, now reaching a height
where they brush against bare conductor on overhead lines.
Another obvious comparison to look at is the average duration of an outage, which is
simply the total duration of all interruptions divided by the total number of interruptions
that occurred during a period. The average duration is also given by the ratio SAIDI/SAIFI
or CTAIDFCAIFI. Average duration of interruption provides an indication of how the
types of outages vary or how the utility is managing the entire process from trouble call to
service restoration, as shown in Figure 4.4. Here, the increase in outage duration during
winter for a utility in the northern United States is clearly visible (winter storms stress repair
resources and make travel to repair sites lengthier). A possible upward trend in overall
duration is also discernible. Without additional study, it is not possible to determine
whether this is due to an increase in the number of outages, a change in type of outages
occurring (perhaps "more serious" outages are occurring more often as equipment ages), or
to a less effective (slower) repair process.

4.4 COMPARISON OF RELIABILITY INDICES AMONG UTILITIES

Comparison of SAIDI, SAIFI, and other reliability indices among different utilities can be
an exercise in frustration for the distribution planner, because supposedly similar standard
definitions such as SAIFI vary tremendously in practice from one utility to another, as
alluded to in section 4.3. The definitions and formulae for the various indices are nearly
always the same from utility to utility, but the characteristics of the component
measurements and data sources that input into the formulae are different. As a result,
comparison of SAIFI values between two utilities may show a large discrepancy when there
is no significant difference in actual customer interruption levels. Despite this difficulty,
and the questionable utility gained by such comparisons, most distribution planners are
asked to perform such studies from time to time.
The usual reasons that utilities compare reliability statistics are to determine how well
they are doing versus the rest of the industry and to prove that their system is better than
average. They may also be interested in determining what might be a reasonable target for
their reliability criteria. Table 4.5 gives SAIDI and SAIFI statistics for six utilities based on
what the author believes to be fairly equivalent definitions and reporting methods. If
nothing else, this table shows that reliability results differ tremendously because of real
differences in service territory, climate, and demographics.

118 Chapter 4

Table 4.5 Average Annual SAIFI and SAIDI for Six Utilities 1988-1992
Utility Service Area Climate SAIFI SAIDI
1 Dense urban area Hot summers, bitter winters .13 16
2 Urban/suburban Hot nearly year round 2.7 34
3 Suburban & rural High isochronic (lightning) 2.0 97
4 Urban & rural Mild year round 2.1 122
5 Rural, mountainous Mild seasons 1.1 168
6 Rural, mountainous Bitter winters 2.3 220

Differences in How Reliability Indices Are Computed
The values shown in Table 4.5 were adjusted, as carefully as possible, to reflect a common
basis of definitions, measurement, and reporting. The author believes they are roughly
equivalent. However, as mentioned earlier, terms like SAIDI and SAIFI are in wide use
throughout the power industry but often with differences in exact interpretation and
reporting of their component parts. For one thing, as was discussed earlier (Table 4.4), there
are differences in how utilities define the types of interruptions they include in their
reliability indices. This can materially affect the resulting values.
In addition, there is no commonality in how the customer count used in the denominator
of most reliability indices are defined and ascertained. Many utilities consider a master-
metered apartment building or shopping mall to be a single "customer," while a few count
the individual units or stores as customers to get a more accurate assessment of interruption
impact. Regardless of this difference, there are other variations in customer count. Most
utilities define the number of customers system-wide as the average number during the year
or use best estimates on a monthly basis. Others use the number as of January 1st or some
other date. Policy on counting seasonal customers (summer beach homes only served part
of the year) varies too. The important point is that slight variations in system reliability
indices, due to growth, seasonal change, and simple differences in how "customers" are
counted, are common and can account for differences of up to 5% in SAIDI and SAIFI
statistics.
In addition, utility practice differs widely when estimating how many customers any
particular outage actually interrupts, something that makes statistics like CTATDI and
CAIFI vary greatly from utility to utility. Some utilities use very comprehensive databases
and distribution management systems (DMS) to determine precisely how many customers
each equipment outage takes out of service. But many utilities use approximate methods to
estimate the customer count of each particular outage because they lack the data or
resources to perform such analysis accurately.
For example, one utility in the midwest United States employees a reliability assessment
procedure that assumes every feeder outage knocks out service to an average number of
customers, in this case 1764 customers (the average customer per feeder over their entire
system). Thus, every instance of a locked-out (completely out of service) feeder is assumed
to have interrupted service to exactly 1764 customers, even though the number of
customers per feeder varies from less than 600 to more than 2400. Furthermore, their
procedure assumes that every partial feeder outage knocks out power to 294 customers (one
sixth of the customer count on an average feeder). Over a large system and a full year such
approximations may make little difference in the computed SAIDI, SAIFI, and other


indices. On the other hand, they may; and they do introduce a significant approximation
error when computing indices over short periods or within small regions.
Another difference in reporting practices is whether momentary interruptions caused by
"cold switching" during restoration and maintenance are counted. Sometimes, to avoid
closing a loop between substations or feeders, it is necessary to isolate a portion of a feeder
first before switching it to a new feeder or substation source. Proper organization of
distribution operations and personnel can reduce this interruption to as little as five seconds,
but it does cause customer interruptions. Some utilities count these interruptions as
momentary or temporary interruptions, but many don't.
Load estimation procedures for the various curtailment indices such as CALCI are
another factor that varies tremendously in practice. Some utilities calculate the curtailed
energy based on average load (annual billing divided by 8760 hours) for the customers
involved, not the actual load at time of outage. Others try to estimate the actual demand
using hourly load models for various customer types.
Estimated Durations of Interruptions
The single widest variation in reliability reporting practices relates to the definition of the
duration of interruptions. Nearly every utility has a good record of when service is restored
to each interrupted customer, but all but a few (those with real-time monitoring equipment
in the field) estimate the time the interruption began. Breaker operations that lock out a
feeder or a whole substation usually are recorded by SCADA or on-site equipment, so the
timing is known precisely. But these outages constitute only a small part of distribution
failures. There is no similar recording mechanism to identify the precise time when a line
fuse operated or when a lateral line fell. To compute the duration of the outage and the
interruptions it caused, this time of occurrence has to be estimated by some means.
Many utilities define the "start" of each interruption as the time when it is first reported,
even though it had to have occurred before the first customer called in to report the trouble.
Some utilities try to account for that delay by adding a minute or three minutes to the time
when they recorded the call. However, no one wants to use a procedure that lengthens the
duration of the outage data, making the duration indices worse, as that policy does. A very
few utilities still apply a practice with the opposite effect, one common in the 1930s,
defining the "start time" for interruptions as the moment when each is relayed to the
distribution dispatcher, which can be from two to fifteen minutes after the trouble call was
first received from the customer.
Thus, anyone comparing statistics between two or more utilities expect noticeable
differences to exist in their statistics due only to definitions and recording differences. If
one wants to adjust these data to account for these differences so that they can be compared
on a valid basis, they will have to be prepared to do a great deal of additional work, and
they should not be surprised if this cannot be done well.
Setting Reliability Targets
Most utilities set reliability targets for their operations and manage planning, engineering,
and operations to achieve those targets. However effectively, and economically, this is done
is often subject to question, and practices at many leave much room for improvement. But
the process starts with the determination of the reliability target - the desired reliability
performance of the system. Utilities base their decision on the needs of their customers, the
business advantages they see of one reliability level over another, and in many cases
considerable "help" from the regulatory process. Chapters 20 and 21, covering criteria for
planning of a power system, will discuss methods for determining reliability targets.

120 Chapter 4

4.5 BENCHMARKING RELIABILITY
Benchmarking is a process of comparing a person, group, or company's performance to its
peers for purposes of identifying needed areas of improvement or discovering ways to
improve performance. Many electric and gas utilities "benchmark" themselves against other
utilities in order to determine "how good they are" as part of programs or initiatives aimed
at improving customer and stockholder performance. Reliability, as measured by SAIDI,
SAEFI, and other indexes, is nearly always a feature in these comparisons. Money, as
measured by profit margin, debt/equity ratio, and/or capital and O&M expenses per
customer, MW, or whatever, is also an important aspect.
Benchmarking projects are usually initiated as the first step in an internal initiative by
the utility to improve its performance. Reliability as seen by the customers and customer
service and satisfaction are always key factors of comparison. But money is always at the
heart of any such study; even if the focus is on improving reliability and customer service,
concerns about expenses and improving the bottom line (profitability) will be very much on
executives' minds. Figure 4.5 shows the relationship that benchmarking projects have with
respect to follow-on activities and results.
Benchmarking, particularly of reliability statistics and operational practices, is fraught
with pitfalls, beginning with the difficulties mentioned earlier created by subtle differences
in the interpretation and application of definitions, measuring systems and accounting
practices, and reporting periods, etc. This makes it very difficult if not impossible to
completely "equivalence" two utilities and their operating histories — in the author's opinion
about the best that can be done is to obtain comparisons within a 10% range.
Types of Benchmark Studies
Benchmarking is done for a variety of reasons. But while the need and motivation may
differ from project to project, in all cases the utility is concerned about its performance,
usually the concerns covering a combination of customer service (e.g., reliability, customer
satisfaction, public safety) and financial (debt/equity ratio, profit margin, O&M spending,
etc.) issues. Almost all benchmark studies fall into one of the four categories, summaries of
which are given in Table 4.6.

:
100

^i-

32 Suburban-Rural Utilities

Figure 4.5. Here, one particular utility (darkly shaded bar) has been "benchmarked" with respect to
SAIDI against 31 other utilities. A bar chart is often used to display ranking data as it conveniently
communicates both where the utility ranks (somewhere in the lower half in this case) and gives a good
idea of how wide the spread in performance is among the group (quite a lot, in this case).

Power System Reliability

on O
•^ (U
c en en
•*—»

c S
(D
o £ S
•1 "1 (U
O -4-*
t^
oo en ap
2
o> oo eu ^ c c o -3 O
a efi Jrt '•• >~> ctf
s s (U V ,^ c
*"N 3 d 00 •
t: o i a a
e g g
"Z3
G
C
<L>

.* o £ G C|—i
&
ll 0 c 
sa
t*H H >

I l f i eu I I
H o & o
I -3 ex t;
«
cu cx .5
t! cx
0 g
21
>-. cx
cd
cu .5
c

en
. r
Q .
f-H
D
& S~*. >— I
o -a
T3 en
^ Basic statistics (SAI]

Usually limited to wi
5^
i SAIFI) and financial

Processes, technolog;
Basic indices (SAID
and complex financh

dissemination of stati
organizational issues
ratios (O&M budget

human resource and

on areas of expected
ce

ratios (e.g., new
u
o

debt/revenue)

improvement
ta
customer)

en .5
c<3 __, T3 '3 u
<D 1 2 C 1r? c o
_M
Q § <L> rt
•8 .2 g
r Low. Often limited t

subdivide performam

Tracks basic perform;
procedures, organizat
Level of Detai

to obtain statistics fo

financial ratio indices
Comprehensive. Ma

areas and look at cosi

Often goes into great

statistics, usually not
resource efficiency.

about technology,
publicly reported
documents, etc.

used.

<D
(U CX *T™1 ca 3 C j_,
00 & •S G 2 c
2 8 S
en
(U
•%
a
X
-s 33 S
§ T3
a
S c~* "9
g w
C en §8
•c S
^ --H ^
Their Purpos

1> O
2 c ^ en £ £ -fi
E 0 o. fe -a
•a 2 •° 'Z G >>
« "^ 1
a ^
Nature

u .ts W ^ c
c ±i
CX en ex u ^
§ *
C3 C3
OO "3 CX CX i- -Z3 "3
OOeC" £
& oo O «t- °- s T3 "C
S ^>
0
w>
0
-G c3 en o £<£ f °* C
t3
"
en rt
fcn g l)
_ -° -d
J3 -5 Z. X)
5S C 5^ -^ "O 'S
1 § flj£
c$ ;>•* T3
^ .*« e2 "° u G
0 rt •t 1?
.._,
« **
O ~r-t
t-i
en r-" Z3 o> gl
OJ 5 a ^
<L> «J T3
S S -2 -S § S, o 6 ^
•3 3 CX cx JL en
>, S en _g CX O T3 ^2
3
0 3 «* oj
V) 1oo
H e
^ S -0
J-* n r"
H 3 i O
TT
tU
*^T
_Q J—
O ^ jD
^ ^
*i 1 § 1
!=> S S
^
ol u,
U
J 2 flj ±3
h -C en
C

1 1 a^
U S *-<-i C _ 4 (
l-i
D S ^T3 o e
~>
e
«• s 5
_. O
CX -i-" t*-i
.O Cfl
en
3
^S j^js
 -^ *~ t (
en y •—
— o — H cxo £' X) CXJS £ 22 «3 o
r- o o> o
i 3
««»
^T 00
C
"c3
p2 en
O
3
en
- a. 2
8.
•a >> §
03
CX
ID
0
_g
G
85 c3 O
H H a; O £
O
U

122 Chapter 4

Position or ranking benchmark projects
What might be called position or ranking studies compare a utility to a group of its peers
(similar utilities in terms of size and other salient characteristics) or the industry as a whole
and rank the utility in terms of its position. This type of study might determine that a utility
was in the top 18th percentile with respect to SAIDI and the 53rd percentile with respect to
O&M spending per mile of line, etc. Figure 4.5 shows this type of a ranking for a group of
32 utilities. Given the aforementioned fuzziness of any benchmark study, such studies are
never conclusive, but they do give some indication of how the utility is performing; Figure
4.5 makes it clear that, even if the data are off by 10% or more, the overall conclusion (that
a number of other utilities are noticeably outperforming this utility) is valid.
Ranking projects are sometimes done at the request of one of the utility's executives
who is curious or concerned that the utility may not be performing as well as possible, and
thus requests an objective study of how the company is doing relative to the rest of the
industry. But this type of study is most often undertaken as a defensive project in response
to criticisms or suggestions that improvement is necessary, usually from a regulator or
intervener, but perhaps in the press or from stockholders.
Defensive projects are nearly always done in something of a crisis environment, with
considerable executive sponsorship and attention. Often, such projects have a very narrow
span, limited to the areas of criticism (i.e., if maintenance practices were criticized, then
only maintenance practices are benchmarked). The project's purpose is merely to rank the
utility against its peers and show (hopefully) this it is doing well compared to other utilities.
There is seldom any consideration given to the possibility that all utilities in the identified
peer group might be performing poorly compared to the best that could be done.
Regardless, there is almost always a firm belief within the utility, and an expectation, that
the study will refute the external criticism and prove that the criticism is unfounded. This
can bias the study's objectivity and results.
Gap analysis benchmark projects
Where a ranking study focuses on utilities and looks at many other utilities and determines
"position within the pack," a gap analysis focuses on the performance level, looking mainly
or only at any performance gaps. One can view these projects as differing in perspective:
ranking projects focus mainly on the horizontal aspect of Figure 4.5. Gap analysis projects
focus on the vertical scale and its aspects.
A gap analysis for the utility in Figure 4.5 would find the three or four best performers
in its peer group and focus on the SAIDI gap between the utility being benchmarked and
these utilities (a difference of about 80-90 minutes). At some point, this type of evaluation
always moves to a bang-for-the-buck basis. Instead of looking at SAIDI alone, it would
look at O&M spending versus SAIDI and seek to identify as top performers those utilities
that spend the least to obtain the most performance in this category.
Process benchmarking projects
In many cases, a gap analysis project will go into considerable detail in analyzing the
performance gaps between a utility and the best performing utilities in each case and
include a study of how and why the performance difference exists. Projects that move to an
examination of the methods used by the leaders in the peer group, and that try to diagnose
and prescribe methods to make improvements, are process benchmarking. They focus on
the means to achieve the performance, and will try to ferret out the organizational structures
or technologies that the top performers are using to enable them to achieve their "best in


class" level. Often, a very profitable aspect of these studies is to also evaluate the lowest
performers, to see what they don't have and don't do.
This process level of investigation can be viewed as a second stage to a gap analysis, a
stage which identifies the means through which the utility can hopefully replicate the top
performance level and close that gap. ("In this category, all three top performers in our peer
group had three common characteristics - a de-centralized service engineering structure, a
discretionary reliability projects budget for each division, and mobile computing integrated
with the regional TCOMS. By contrast, none of the lower quartile utilities had any of these
characteristics.")
Regardless, the process benchmarking studies and compares the technology (physical
equipment, tools and machinery used), procedures (specific methods of performing tasks),
and processes (method of grouping and fitting together procedures and communication
among various tasks). It will include a look at organization (how functions and activities are
identified, grouped, and assigned), communication (how information and priorities are
disseminated through an organization), and responsibility and accountability.
The best way to gain this detailed information on processes is to talk with the top
performers. Surprisingly, given the apparent level of competition often implied by the
benchmarking process, electric utilities are rarely reluctant to talk with another utility, and
most are quite willing to share detailed information about how they attain their good
performance. Simply calling and asking often garners an invitation to visit and discuss their
practices.
Continuous improvement benchmarking
As part of initiatives to improvement performance, many utilities set up formal programs of
measuring and tracking reliability and performance on a periodic basis. This internal
benchmark process usually focuses on basic performance statistics (SAIDI, SAIFI, cost of
unexpected maintenance) rather than cost-specific measures as done in gap analysis. The
process generally both tracks performance over time and against the specific targets set in
the utility's "self-improvement program." It may also compare one part of the company to
another (e.g., north versus south operating divisions, etc.).
The better projects of this type also track the continuing performance of the utility's top
performing peers (who are not stationary targets — they may have their own improvement
programs).3 Generally, the results of these benchmark studies are widely disseminated
within an organization, one of their purposes being to keep people focused on the
improvement targets and let them see what their progress has been.
Driving Forces and Purposes of Benchmark Studies
Benchmarking projects are undertaken when a utility is concerned about its performance,
whether because it must justify its performance level to an external agency or because it
wants to improve it or both. Figure 4.6 summarizes the basic causes, flow, and effects of
benchmarking projects. Emphasis will of course vary depending on the utility's specific
situation and strategy.
As mentioned earlier, often a utility is compelled by external criticism or regulatory
requirements to defend its performance level, which often leads to a "defensive study"
whose purpose is to show the utility's performance is on par with its peers. However, a

3
In fact, the author's experience is that the top performing utilities are generally the utilities that are
least satisfied with their present performance and most committed to improvement, and that they are
usually top performers in driving continuous improvement.

124 Chapter 4

Figure 4.6 Benchmarkjng, often centered around reliability indices and performance, is a key aspect
of utility programs aimed at improving corporate financial performance.

majority of benchmarking projects are part of a larger process of driving improvement in an
organization. These improvement projects generally begin with a very visible initiative
sponsored by an executive, widely publicized throughout the company and given
considerable priority, and eventually they lead to both a one-time “quantum leap” effort to
make big improvements in a short time and to a continuous improvement process
institutionalized within the utility.
One-Time “Quantum Leap” Programs
Many utilities initiate a “self improvement program” aimed at quickly increasing financial
performance andor improving customer satisfaction, service reliability, or both. These are
usually driven from the top down, often initiated and quite visibly championed by a top
executive. Money is always the key factor in these initiatives, and in fact financial
performance improvement has been at the heart of every such initiative in which the author
has acted as consultant or advisor
One-time initiatives generally set aggressive goals for improvement in spending and
financial ratios (improvements of 50% in some numbers are not uncommon) and have
relatively short-duration focus (one to two years). However, it is recognized that reliability
will degrade unless an initiative to improve finances also pays attention to and maintains a
commitment to good customer service quality. Thus the initiative takes the form of “we
plan to improve [these particular financial measures] while making sure that customer
service quality as measured by [specific reliability or customer satisfaction indices] is kept
at satisfactory levels.”
Ranking studies are among the first steps in these types of internal performance
improvement initiatives. Even if the executives are certain that performance needed to be
improved, they will ask for this study in order to conclusively document to employees that
the utility needs to improve its performance. Gap analysis will be used as well as a guide in
determining how much improvement can be expected, and process benchmarking will be


used to determine how to make the improvements and, through further study, how long it
can be expected to take. Usually, the role of benchmarking in this process is to
• Provide documentation for management to use to justify the need for
improvement. ("We're second quartile and need to improve to first.")
• Identify the nature of the improvement challenge for strategic planning
and for planning of the improvement program. ("Detailed gap analysis
indicates we can improve by 31% if we can emulate the performance
Big State Electric currently has.")
• Identify the processes and technology steps that might be employed.
("Clearly, a GIS system on the IT side, and an incentivized
performance based contracting paradigm on the supply and services
management side, are key factors in obtaining high performance.")
• Determine and institutionalize the actual metrics ("We're going to
measure reliability as SAIDI with respect to this project, and this is
exactly how we will gather data and compute i t . . . ")
• Set targets ("We will drive SAIDI down from our current 188 minutes
to 150 by June 2005.")
These one-time self-improvement initiatives are usually created and managed as an
extraordinary activity in the true sense of the word - an extra activity with a temporary
mandate and goal, something quite apart from ordinary, continuing processes.

Continuous Improvement Programs
The highest levels of performance come from a system that is never satisfied with its
performance, but instead continuously strives to improve it. To do this over the long term,
the utility must institutionalize improvement — set up a system that is part of its basic
processes and daily operation, and make the desire for improvement part of its basic
culture. Frankly, this culture change is often the single biggest challenge utilities face.
Internal benchmarking is a key factor in any such continuous program. It needs to be
limited to a few key performance indicators (no more than five) and be routinely measured
and reported (monthly, quarterly). Reasonable levels of internal comparison set up to create
competition among divisions is usually an effective means to spur improvement. Thus,
benchmarking of one function and one operating division against others is often an
important continuing need.

Internal dissemination of external benchmark data
Whether used in a one-time initiative or as part of a continuous process, benchmarking
projects can provide a utility with hard data about where improvement is possible and how
to effect that improvement. One value of benchmarking is just proving it can be done. The
explicit demonstration that higher levels of performance are possible because top
performers are doing it quiets skeptics of any improvement program. This "demonstration"
of the gap also provides management with indications of how much improvement can be
expected.
Thus, at the end of the initial benchmarking process, the utility will know where it needs
to and can improve performance, even if the comparisons upon which this conclusion are
based are fuzzy (partly or mostly qualitative). It will have an idea of how, where, and why

126 Chapter 4

to implement changes to drive that improved performance. If it then implements these
changes with focus, the use of internal periodic (e.g., quarterly) benchmarking, and a good
management process that provides positive and negative feedback based on the
benchmarks, the utility's performance will steadily get better.
A good deal of the value that flows from benchmarking against "peers" can come from
dissemination of the information throughout the company, and its use as part of a program
to create a will to improve. A key factor here is the identification of and dissemination
within the company's workforce of its "peer group" -just who are they being compared to?
Picking known industry leaders as the peer group sets high standards but lets employees
know that there is some pride expected in being among the best. Targeting a utility that is
considered a rival creates a spirit of competition.
Here, those managing the benchmark and continuous improvement process need to be
mindful of the difference they can create between the peer group used in the early analysis
stages and that used "publicly" within the company during the continuous improvement
process. Objective analysis in the early stages of a benchmark project might determine
conclusively that the company best matches the "losers" category (lowest quartile) among
other electric or gas utilities, something that can and should be conceded and used in
identifying the nature of the challenge the utility confronts in moving to the front of the
pack. But the identification throughout the company of bottom-tier utilities as the identified
"peer group" in a well-publicized continuous improvement program can be self-defeating,
creating poor morale. The use of a few well-selected leaders as the target can have a much
better effect on morale. This is not a subtle point, but one often neglected.
Inevitably, there will be resistance to any effort to drive change and grumbling and
objections. Management should anticipate these objections and have answers ready: "We
realize that Big State Electric has slightly less rural distribution that we do, but comparison
of our S ATDI to theirs is valid because . . . "
Internal dissemination and use of internal benchmarks
As mentioned earlier, a utility can accurately and dependably benchmark itself against
itself. A continuing program of periodic measurement and reporting of key performance
indicators is essential to any meaningful program of improvement. There are several issues,
but for the most part this is straightforward; key factors are measured on a monthly or
quarterly basis and communicated widely (posted on bulletin boards, put on IntraNet
newsletters). This will show the workforce how and if their efforts are meeting goals.
But the utility must have some system in place to apply corrective action if goals are not
being met. Generally, as part of the improvement initiative, it will set up a group whose
function is to not only track performance, but analyze processes and identify where
deviations from expected performance are occurring, bringing this to management's
attention. This core "program compliance" group is an important element of driving
change, and the internal benchmark process is its eyes and ears.
Benchmarks of one part of a company with another can create spirit of competition for
improvement. Posting these results is a way to reward those who improve with recognition
among their coworkers and to put pressure on poor performers to work harder. Internal
benchmarking can fairly easily compare similar groups or functions (i.e., north and south
operating divisions, etc.) in a "fair" way.
However, it can be difficult to compare different types of departments or functions with
others on an absolute scale. How does one benchmark the efficiency and improvement in
the Fleet Management department with that of the T&D Planning and Engineering
department, for example? These two departments have very different types of processes
with very different needs and skill sets involved. Neither is necessarily more important or


difficult than the other; they are just different. This is a real issue (it is difficult). But it
creates an additional issue of perception, for even if done well, inevitably some people will
object to the benchmark results on the basis that it is difficult and may not have been done
well.
A practice that works very well to resolve this "apples to oranges" comparison issue is
to judge internal divisions on the basis of a department's industry position - status relative
to its peer departments in other utilities. This takes additional work in the benchmarking
study, but has the added benefit of turning much of the competitive spirit outward, rather
than inward against peers within the company. Thus, Fleet Management is benchmarked
against fleet management performance (safety, cost/vehicle, etc.) across the industry.
Similarly, so is T&D Planning and Engineering and so is the preventive maintenance
program, human resources, and so forth. Divisions at the top of their peer group are
rewarded; those at the bottom are urged to improve more.
Caveats and Pitfalls in the Benchmarking Process
Good benchmarking — valid comparisons and effective measures to guide improvement — is
not easy to do. Superficial benchmarking studies, which might be useful as a general,
qualitative guide can be done by quickly. For example, FERC Form 1 data or statistics from
a published source such as McGraw Hill's Electrical World Directory of Utilities can be
assembled with a spreadsheet into a ranking of various statistics and ratios (O&M budget
per mile of feeder, etc.). But such uncorrected and unadjusted benchmark comparisons
provide little if any value, and on occasion can point the wrong way. There are a number of
barriers that must be overcome to create truly useful benchmarks.
Differing definitions
The single biggest challenge facing a benchmark study is differing definitions. This impacts
benchmarks of all areas of a utility. There are numerous examples, but here are several:
What is transmission and what is distribution? Utilities (and state regulators) differ
in how they define T and D. Therefore, performance measures (e.g., contribution
to SATDI) and expenses (O&M) that are attributed to 'T" at one utility may be
identified as "D" at another. That this is clearly an issue is quite easy to establish.
The O&M budgets reported by utilities separately for'T" and "D" have as a set a
greater standard deviation than the sum of T&D O&M budgets, and there is an
inverse correlation between O&M spent on T and on D. Figure 4.7 shows this by
plotting T and D O&M spending as reported by 25 medium sized public utilities.
As mentioned in section 4.4, the exact definition of various reliability terms and
indices, such as "interruption" and SAIFI, differs among utilities. Therefore,
reported SAIDI at one utility may not exactly correspond to SAIFI as reported by
another; one might include all interruptions over one minute in its SAIFI count,
another only all interruptions over 15 minutes in duration. This requires work to
identify (one must obtain the actual definitions in use) but can be nearly
impossible to correct in the data comparisons (one needs the raw interruption
statistics in order to determine how the other utility's definition would have
counted them).
What constitutes the "field force" in counts of number of employees and cost of
resources for repair and restoration varies greatly from one utility to another. In
some utilities, personnel assigned to the stores and warehousing functions, as well

128 Chapter 4

"c 4 - ^
g A
u ^.•^ ^
.Q
'jZ o A j. •*
*- O
w
Q * *

o 9-
° il * ,
O)
C
T3
03 11
Q.
W

n .

0 1 2 3 4 5
Spending on 'Transmission"

Figure 4.7 Maintenance spending on items classified as transmission and distribution among 25
utilities in one peer group. A negative correlation exists between the two measures, meaning much of
the variation in either measure is due to definitional differences among utilities.

as service estimators and inspectors, record and survey clerks, and meter readers
are considered part of the field force. At other utilities, the "field force" will
include only that part of the work force actually assigned to field construction and
repair work, and in a few cases only those who are journeyman or above line
workers. This definitional discrepancy can result in more than a 2:1 difference in
reported labor costs for field force functions.

What constitutes Planning and what constitutes Engineering vary greatly among
electric utilities. Like with T and D, a strong inverse correlation exists between
"P" and "E" comparisons such as number of full-time equivalent employees per
million dollars of capital spending, etc. For example, one utility in the Midwest
has 46 planners for a T&D capital budget averaging about $200 million annually.
Another has only 28, but a T&D capital budget that is slightly larger. Yet the two
utilities have very comparable processes - one simply defined "planning" as
consisting of a greater portion of the entire plan-engineer-design process than the
other.
Inverse infrastructure correlations
In addition to the inverse correlations of data that benchmark teams must overcome to
achieve consistency and validity in their results, there is a more subtle type of correlation
between the quality of the data and the performance of the utilities they are measuring. As
an example, a clear inverse correlation exists between quality of data and performance in
the area of interruption reporting. A computerized trouble call and outage management
system (TCOMS) enables a utility to detect and respond to outages and storm damage
much more quickly and in a more efficient way than a utility that is still using a traditional
"wall map and notepad" process. Opinions and examples vary, but even a mediocre OMS


can lead to a 10-15 minute improvement in SAIDI over traditional "paper" outage
management processes. The best TCOMSs do much better.
But a TCOMS also does a much better job of tracking the actual start time, the true
extent (customers affected), and the actual restoration time of every outage, and of
recording all of this for later reporting. By contrast, traditional systems underestimate
duration and extent. Many utilities see, upon installing their first computerized TCOMS,
that their reported SAIDI and SAIFI values increase substantially. The author's experience
is that this difference averages about 30%, with some utilities seeing nearly a 50% increase.
Thus, the top performers in terms of managing trouble and outages are going to be those
utilities who have a TCOMS. Yet these same utilities will, on average, report about 30%
more outages for any given amount of trouble or bad weather than utilities using inferior
processes.
Figure 4.8 shows the raw SAIDI statistics from Figure 4.5 adjusted by the author for
these types of factors. SAIDI from utilities without any rigorous documented OMS were
multiplied by 1.5 while those with systems were adjusted by factors from 1.35 to 1.05,
depending on the quality of their TCOMS. While this adjustment system is only
approximate, the results are certainly more useful to a utility wanting to really understand
its situation than the raw values.
A general observation: better performing utilities report more of their problems.
Similarly, benchmark project participants should be aware that in general better-performing
utilities have better reporting in general and that this means that statistics on negative
aspects of operation are more complete and more dependable among top performers than
among low performers. Thus, benchmark analysis teams should expect to find similar
inversely correlated results in many areas of utility data. For example, one omnipresent
symptom of poor safety practices is poor accident and injury reporting systems: thus a lack
of reported accidents does not necessarily mean no accidents or a safety and quality process
worthy of emulation.

250

32 Suburban-Rural Utilities

Figure 4.8 Raw SAIDI data from Figure 4.5 "corrected" for reporting quality differences by the
author (see text). Position of every utility in the plot has stayed the same (the rightmost bar in this
figure and Figure 4.5 refer to the same utility, etc.). Large differences in performance still exist, and
the specific utility being benchmarked against its peers (darkly shaded bar) looks no better with
respect to the industry than before, but the data are more relevant to additional study and analysis.

130 Chapter 4

"Degree of difficulty"
Another barrier benchmarking projects must overcome is distinguishing the impact of
unavoidable differences in service territory or environmental factors from the impact of
manageable differences in processes and technology. For example, industry operating
reports leave no doubt that the SAIDI recorded at Central Maine Power (CMP) is roughly
four times that reported by San Diego Gas and Electric (SDG&E). Does this mean that
SDG&E is greatly outperforming CMP in terms of system design, protection,
sectionalizing, and its trouble call, dispatching, restoration, and repair processes? Not
necessarily. These two utilities, at opposite "corners" of the United States, are a contrast in
many SAIDI- and SAIFI-related factors of a type beyond the control of the utility. CMP
sees severe winters with extreme ice storms while San Diego enjoys some of the mildest
weather in the nation. CMP's customer basis is very dispersed, with 95% of its customers
being in rural areas, throughout which there is a limited road system (in many regions this
gives only one way in and out of an area). By contrast San Diego's service territory is
largely urban and rural with a nearly ubiquitous road grid that makes moving repair
resources around a storm-damaged system or during adverse weather much easier. These
and other differences in the "degree of difficulty" of maintaining system connectivity make
it difficult to draw valid conclusions about the relative quality of these two companies'
operations processes based on operating statistics alone.
Figure 4.9 shows the data from the previous figures normalized by the author to "level
the playing field" with respect to factors like dispersion of the customer base, difficulty of
terrain and road network, and adverse weather's impact on field-crew travel times, etc.
Again, as with the "corrections" applied in Figure 4.8, these adjustments are approximate
although done with a sound and unbiased procedure. Regardless of a remaining ±5% range
of uncertainty, this increases the utility of the data for benchmarking and performance
monitoring.

250

Figure 4.9 SAIDI data shown in previous figures after both correction for reporting differences
(Figure 4.8) and normalization for service territory and weather differences beyond the control of the
utilities. The data are now more useful for identifying which utilities have the best processes in place
to manage equipment outages and customer service interruptions. Darkly shaded bar is the utility
being benchmarked.


The final data actually show that the benchmarked utility (darkly shaded bar) is a worse
performer than shown by the raw data; the raw data (Figure 4.5) showed that it was
outperforming seven of the thirty-two utilities in its peer group. These adjusted and
normalized data show that it is outperforming only five. More important to the utility's
improvement process, the final data identify two clear top performers (only one of whom
was among the top performers found with the raw data).
Focusing on both good and bad performers
Due to the many barriers to completely valid comparison - different definitions, inverse
correlations, and "degree of difficulty" problems - it can be a mistake to focus only on the
top performers in a benchmarking study. While one needs to identify and study the top
performers, and the "gap" they identify is important to many aspects of improvement
planning, a lot can be learned by looking at the other end of the spectrum. There are often
clear patterns and trends among the poor performers that indicate what not to do, and
sometimes this is more demonstrable than at the top end of the spectrum.
For example, both of the two top performers in Figure 4.9 have a policy of temporarily
assigning some of their repair and restoration resources (crews with bucket and derrick
digger trucks) to troubleshooting early in a storm restoration process. The additional
information and quicker diagnosis of the extent, location, and type of damage done to the
system is considered more than worth the short-term loss of resources that can be dedicated
to repair and restoration. Utilities practicing this approach believe it leads to better early
planning of the entire restoration scheme and optimization of the use of resources,
ultimately improving the whole storm restoration process.
But does this policy really improve performance? After all, this is a group of only two
utilities, even if they are top performers in this group. The fact that none of the bottom ten
performers in the group practices this policy seems to indicate that, yes, this may be a factor
distinguishing good from bad practice.

Reliability of service is one of the major factors electric consumers weigh in perceiving
how well their electric supply system is doing its job. Whether that service is provided by
an electric utility system or a distributed generator, the planners of the supply system must
set appropriate reliability goals and plan, engineer, and operate it to achieve targeted levels
of customer service reliability.
Equipment outages cause customer service interruptions. Outages create more problems
if they are frequent, if the have a lengthy duration, and if the extent of the interruptions they
cause is great. Various reliability indices can be applied to combine frequency, duration,
and extent into single-valued measures of reliability. Usually a utility uses one or more of
these indices to track its customer service performance over time, by location, and with
respect to function.
Most utilities set reliability targets and manage their processes and performance to
achieve these targets. Benchmarking - the process of comparing performance to peers in
order to identify areas of improvement - is a useful tool for utilities that need to increase
their customer service quality.
Table 4.7 provides a one-page summary of key points from this chapter.

132 Chapter 4


Reliability is the ability of the power delivery system to make voltage, of a satisfactory quantity and
quality, available to meet the consumers' needs.

Equipment outages cause customer service interruptions. Many people often use the term
"outage" to refer to a customer who is out of power, but strictly speaking, "outages" occur to
equipment and the resulting loss of service by a customer is an "interruption."
Power interruptions have three aspects: frequency, or how often interruptions occur; duration,
which is how long the interruptions last; and extent - how many customers are out of service due to an
outage.
Reliability is improved in a system by efforts aimed at reducing all three aspects, frequency,
duration, and extent.
frequency is reduced by reducing the frequency of equipment outages through use of good
equipment, good maintenance and inspection, and sound vegetation management.
duration is reduced through system plans that provide alternate feeds and sources and fast
restoration, including good field operations for manual switching and, increasingly,
automation for near instantaneous switching.
extent of the interruptions caused by any outage is minimized by good combination of
configuration and protection-sectionalizer planning.

Reliability indices are attempts to produce single-valued measures of reliability for a power system or
a portion of it. There are many, all are computed based on different formulae using frequency,
duration, and extent of power interruptions. Extent is usually normalized by computing an index on a
per customer of customer base basis. The basic problem in trying to measure reliability is in how to
relate the two quantities frequency and duration.
The two most widely used reliability indices are SAIFI (System Average Interruption Frequency
Index), which is the average number of interruptions customers in a utility see in a year, and SAIDI,
which is the average total duration of outages seen by those customers in a year. Reliability indices are
applied in historical analysis to evaluate operating experience and reveal trends or patterns, expose
problems, and indicate how and where reliability can be improved. They are used in predictive
analysis to evaluate how well proposed solutions are likely to solve the identified problems
Differences in how reliability indices are computed, or how frequency, duration, and extent are
defined and measured, often make comparison of reported reliability results among utilities
meaningless. "SAIDI" as measured at one utility might be far different - perhaps as much as 35% -
than as defined and measured at another.
Benchmarking is a process of comparing a group or company's (reliability) performance to its peers
for purposes of identifying needed areas of improvement or discovering ways to improve
performance. Many electric and gas utilities "benchmark" themselves against other utilities in the
industry by comparing not only reliability results, but processes and procedures.
Reliability targets - quantitative goals based on indices like SAIDI and SAIFI are used by most
utilities as part of programs that manage their activities and performance to achieve satisfactory
customer service.
Reliability targets are set (determined) by any of several methods, from arbitrary decree by
management to comprehensive risk-based balancing of the cost of improving reliability versus the
cost of the consequences of customer service interruptions.
Utilities that monitor and manage reliability well often report operating results which are worse
than those that do not. The better processes they use do produce improved performance. However,
they also identify and track problems so much more comprehensively that a host of outages and
interruptions not fully counted in the past are now counted and recorded.


REFERENCES

P. F. Albrecht, and H. E. Campbell, "Reliability Analysis of Distribution Equipment Failure Data,"
EEI T&D Committee Meeting, January 20,1972.

R. Allan, et al., "A Reliability Test System for Educational Purposes - Basic Distribution System Data
and Results," IEEE Transactions on Power Systems, Vol. 6, No. 2, May 1991, pp. 813-821.

R. Billinton and J. E. Billinton, "Distribution System Reliability Indices," IEEE Transactions on

R. Billinton and R. Goel, "An Analytical Approach to Evaluate Probability Distributions Associated
with the Reliability Indices of Electric Distribution Systems," IEEE Transactions on Power Delivery,
PWRD-1, No. 3, March 1986, pp. 245-251.

R. E. Brown, Electric Distribution System Reliability, Marcel Dekker, New York, 2002.

J. B. Bunch, H. I. Stalder, and J. T. Tengdin, "Reliability Considerations for Distribution Automation
Equipment," IEEE Transactions on Power Apparatus and Systems, PAS-102, November 1983, pp.
2656-2664.

"Guide for Reliability Measurement and Data Collection," EEI Transmission and Distribution
Committee, October 1971, Edison Electric Institute, New York.

W. F. Horton et al., "A Cost-Benefit Analysis in Feeder Reliability Studies," IEEE Transactions on

Industrial and Commercial Power Systems, The Institute of Electrical and Electronics Engineers, Inc.,
New York, 1990.


Systems, November 1994, p. 4082.

A. J. Walker, "The Degradation of the Reliability of Transmission and Distribution Systems During
Construction Outages," Int. Conf. on Power Supply Systems. IEEE Conf. Pub. 225, January 1983, pp.
112-118.

H. B. White, "A Practical Approach to Reliability Design," LEEE Transactions on Power Apparatus
and Systems, PAS-104, November 1985, pp. 2739-2747.

W. Zhang and R. Billinton, "Application of Adequacy Equivalent Method in Bulk Power System
Reliability Evaluation," paper presented at the 1997 IEEE Power Engineering Society Summer
Meeting, Berlin.

Economics and Evaluation of Cost

5.1 INTRODUCTION

A major attribute of planning in almost all endeavors is reduction of cost. Criteria on
service quality and standards must be met and guidelines must be followed, but within those
limits the planner's goal is to minimize the cost. Every alternative plan contains or implies
certain costs: equipment, installation labor, operating, maintenance, losses, and many others
as well. Alternatives vary not only in the total cost, but often, equally important, when the
costs are incurred - how much must be spent now, and how much later?
Traditionally, electric utilities have been given a monopoly franchise for electric service
in a region, which carried with it both an obligation to serve and a requirement to work
within a regulated price structure. Regulated prices are based on cost and regulated utility
planning on cost minimization - the utility can expect to make a reasonable return on its
investment and recover all its costs, but it must work to reduce its costs as much as possible.
Thus, in traditional utility planning, expansion needs were defined by the obligation to
serve, and planning goals were defined by a cost-based pricing environment. Utility
planning focused on finding the lowest cost alternative to serve all the customers. In
addition, as the sole provider of electric service the utility assumed a "resource portfolio
management" function for its customers, meaning that it had an obligation to determine and
implement the least-cost use of energy efficiency and other resources, which led to
integrated resource planning.
In a de-regulated power industry, this situation will change very little for the distribution
planner. The local distribution company (LDC) will still be regulated, and prices for
delivery (if not power) will still be cost-based. Instead of an obligation to serve, the LDC
will have an obligation to connect, or more specifically an obligation to provide sufficient
capacity. Under retail wheeling, some other company's power may be flowing through the
distribution system to the customers, but the distribution company is still required to
provide sufficient capacity and to do so at the lowest possible cost.
This chapter looks at costs, costing analysis, and cost minimization concepts for
planning purposes. Entire books have been written on the subject of engineering economics
(i.e., cost analysis and comparison), and the purpose of this chapter is not to duplicate or
summarize them. Instead, its focus is on what costs represent and how they are to be used in
planning. The chapter begins with a look at cost and the various types of cost elements that

135

136 Chapter 5

are included in an analysis, in section 5.2. Section 5.3 examines the time value of money
and the application of present worth analysis methods to planning. Section 5.4 looks at the
time value of money and future versus present costs. Section 5.4 concludes with a look at
an important aspect of planning, particularly in a competitive marketplace: price-related
variation of costs with respect to location, timing, and level of the system
5.2 COSTS
"Cost" is the total sacrifice that must be expended or traded in order to gain some desired
product or end result. It can include money, labor, materials, resources, real estate, effort,
lost opportunity, and anything else that is given up to gain the desired end. Usually, such a
combination of many different resources and commodities is measured on a common basis
- such as money - by converting materials, equipment, land, labor, taxes and permits,
maintenance, insurance, pollution abatement, and lost opportunity costs to dollars, pounds,
marks, yen, or whatever currency is most appropriate. In cases where all the elements of
cost can be put onto a common basis, the subsequent planning can be done in a single-
attribute manner, the goal being to minimize the monetary cost and the cost reduction
basically trading the cost of one item against another to find the best overall mix. However,
in rare cases, some costs cannot be converted to money - for example rarely can esthetic
impact or other "intangibles" be converted to monetary cost. In such cases, cost reduction
must be done as multi-attribute planning and cost minimization, involving more
complicated planning methods which are, frankly, often not entirely satisfactory.
Initial and Continuing Costs
A distribution substation, feeder, or any other item has both an initial cost to create it and
put it in place, and a continuing cost to keep it in operation, as illustrated in Figure 5.1. The
initial cost includes everything needed to put the substation in place for its initial operation,
and may include a very comprehensive list including engineering, permits, surveying, land,
legal fees, site preparation, equipment, construction, testing, inspection, certification,
incidental supplies and labor, and insurance. Continuing costs entail keeping it in operation
- inspection and maintenance, routine supplies and replacement parts, taxes, insurance,
electrical losses, fuel, and perhaps other expenditures.

Initial cost

Continuing costs

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Figure 5.1 A new substation's costs are broken into two categories, the initial cost - a one-time cost
of creating the substation - and continuing annual costs - those required to keep it in operation.

Economics and Evaluation of Cost 137

The initial cost is incurred only once, often over a period of several months or years
during which the item (e.g., substation or transmission line) is built, but it is usually
considered as a single budget expense item allocated to a specific year in the utility plan.
The continuing costs persist as long as the substation exists or is in operation. Usually,
continuing costs are recorded and studied on a periodic basis - daily, monthly, or annually -
with yearly analysis generally being sufficient for most planning applications.
Fixed Versus Variable Costs
The cost of a substation, feeder, or other element of a power system can also be viewed as
composed of fixed and variable costs. Fixed costs are those that do not vary as a function of
any variable element of the plan or engineering analysis. For example, the annual cost of
taxes, insurance, inspection, scheduled maintenance, testing, re-certification, and so forth
required to keep a 100 MVA substation in service do not vary depending on its loading -
whether it serves a peak load of 5, 32, 47, 60, or 92 MVA these costs would be the same.
Similarly, the no-load (core) losses of its transformers create a cost that does not vary as a
function of load, either. By contrast, the transformer's load-related (copper) losses do vary
with peak load - the higher the substation's loading the higher the losses costs. These are a
variable cost, varying as a function of the amount of load served. Some types of
maintenance and service costs may be variable, too - highly loaded transformers need to be
inspected and stress damage repaired more often than similar but lightly loaded ones.
Figure 5.2 illustrates fixed and variable costs.
In many engineering economic evaluations, fixed costs include all costs except variable
costs of operation such as fuel and losses. However, there are a sufficient number of
exceptions to this rule so that planners should both consider what they will designate as
"fixed" and "variable" in a study, and also check any data or study upon which they build to
make certain the meaning of those terms in that work is consistent with their own.

Fixed costs

Variable costs

Figure 5.2 The new substation's costs can also be viewed as composed of both fixed costs - those
that are constant regardless of loading and conditions of its use - and variable costs, which change
depending on load, conditions, or other factors in its application as part of the power system. In this
example, the variable costs increase nearly every year - reflecting higher losses costs as the load
served gradually increases due to load growth.

138 Chapter 5

Variable
Costs

None 25 50 75 100
Capacity To Be Installed - MVA

Figure 5.3 Definition of "fixed" and "variable" costs depends on the context. Although often the
capital cost of a substation is labeled a "fixed" feature in engineering cost studies, here it consists of
both fixed and variable costs: planners have four capacity options to consider for a substation - 25, 50,
75, and 100 MVA - representing installation of from one to four transformers. In this planning context
the cost of the substation can be viewed as composed of a "fixed" component consisting of those
items common to all four options and a "variable cost" that depends on the installation of capacity
from 25 to 100 MVA, as shown.

As an example of how "fixed" and "variable" costs can change depending on the
planning context, consider a 100 MVA substation - one composed of four 25 MVA
transformers and associated equipment. After it is built and in place, its initial construction
cost is always considered "fixed" in any study about how it might be utilized, expanded, or
otherwise changed with regard to the plan. Variable costs include only those that change
depending on its load or utilization: if more load is transferred to it so that it serves a higher
load, its losses go up; if low-side voltage is regulated more tightly, its LTC maintenance
costs will probably increase, etc.
But in the initial planning stage, perhaps the planners considered the alternatives of
building it with one, two, three, or four transformers in place. In that situation, its "fixed"
cost within the planners' minds might have included only those costs common to all four
alternatives - land, site prep, and so forth. In addition, the planners would consider a
variable capacity cost, which changes as a function of adding one, two, three, or four
transformers (Figure 5.3). This is in fact how some substation planning optimization
programs accept cost data as input, as will be discussed in Chapter 24.
Sunk Cost
Once a cost has been incurred, even if not entirely paid, it is a sunk cost. For example, once
the substation has been built (all 100 MVA and four transformers of it) it is a sunk cost,
even if ten years later the utility still has the substation on its books as a cost, paying for it
through depreciation or financing of its initial cost.
Embedded, Marginal, and Incremental Cost
Embedded cost is that portion of the cost that exists in the current planned system or
configuration or level of use. Depending on the application, this can include all or portions


200

150

125

u 100

3 n
O 50
z

10 20 30 40 50 60
Load-MVA

Figure 5.4 Hourly cost of losses for power shipped through a substation transformer as a function of
the load. Currently loaded to 45 MVA, it has an embedded losses cost of $1.73/MVA and a marginal
cost of losses at that same point of $3.70/MVA. The incremental cost of losses for an increase in load
to 55 MVA averages $5.90/MVA.

of the initial fixed cost and all or parts of the variable costs. Often, the "embedded" cost is
treated as a fixed cost in subsequent analysis about how cost varies from the current
operating point. Marginal cost is the slope (cost per unit) of the cost function at the current
operating point (Figure 5.4). This point is usually (but not always) the point at which
current embedded cost is defined.
Incremental cost is the cost per unit of a specific jump or increment - for example the
incremental cost of serving an additional 17 MVA from a certain substation or the
incremental cost of losses when load on a feeder decreases from 5.3 to 5.0 MVA. Marginal
cost and incremental costs both express the rate of change of cost with respect to the base
variable, but they can differ substantially because of the discontinuities and non-linearities
in the cost relationships. Therefore, it is important to distinguish correctly and use the two.
In the example shown in Figure 5.4, the marginal cost has a slope (cost per unit change) and
an operating point (e.g., 45 MW in Figure 5.4). Incremental cost has a slope and both
"from" and "to" operating points (e.g., 45 MVA to 55 MVA in Figure 5.4) or an operating
point and an increment (e.g., 45 MW plus 10 MVA increase).
Revenue Requirements
In order to operate as a business, an electric utility must take in sufficient revenue to cover
its continuing operating costs, pay for its equipment and system, cover its debt payments
(loans and bonds), and provide earnings for its owners (shareholders), as shown in Figure
5.5. These revenues are the charges its customers must pay. Minimizing the total amount of
revenue the utility needs is one way to keep customer charges at a minimum, a method used
by a majority of utilities.
Minimum Revenue Requirement (MRR) planning is aimed at keeping customer bills as
low as possible — it seeks to minimize the amount of money the utility must collectively
charge its customers in order to cover its costs. While cost reduction nominally contributes
to revenue requirements reduction, often particular planning decisions not only incur or

140 Chapter 5

avoid costs, but also shift them from one column of the balance sheet to another. It is here
where attention to revenue requirements can often reduce revenue even if not costs.
While similar in some cases, particularly capital expansion planning, revenue
minimization, and cost minimization will lead to slightly different decisions with regard to
selection of alternatives and timing of expenses. Capital expansion decisions - such as
when and how to build a new substation - require capital which is often financed, while
most continuing operating costs are not. For example, returning to the question posed
earlier in this chapter about whether to build a new feeder with larger conductor than the
minimum necessary in order to reduce losses, the capital cost of building with the larger
conductor may often be a financed cost to the utility - an expense which it carries on its
books and which it pays as interest.
Thus, the present worth of a "present expense" (e.g., adding the larger conductor today)
has a future element to it (finance costs over some number of future years). In order to reach
the correct decision about whether the feeder should be built in this manner, the utility's
planning evaluation must include in its assessment the correct evaluation of such costs. This
can be done by adjusting the present worth factors used to compare "present" to "future"
costs so that they are compatible with the utility's structure, as illustrated in Figure 5.5. If a
utility has very high interest payments for its capital, it means it wants to raise the present
worth factor for its capital investments, because the benefits (losses savings) of an
investment made today (a larger feeder) must "pay for" the capital financing of the
investment.

Dividends
Debt
Fixed costs
Depreciation

Other items

.— Q>
3 5.
O" ^
Operations
J2 </>
3 £

Variable
II
operating 58
Other items costs o

Fuel

Figure 5.5 Generally, the pricing of electric power (certainly electric power delivery) is based on
cost, since it is a regulated business. The total revenue required by a utility includes what is needed to
cover all costs of operations, power purchased or generated, capital, debt, etc. Minimum Revenue
Requirements planning seeks to make all decisions with regard to spending based on minimizing the
total requirement for revenue (rightmost arrow).


5.3 TIME VALUE OF MONEY
Any utility planner must deal with two types of time versus money decisions. The first
involves deciding whether a present expense is justified because it cancels the need for a
future expense of a different amount. For example, suppose it has been determined that a
new substation must be built in an area of the service territory which is the site of much new
customer growth. Present needs can be met by completing this new substation with only
one transformer, at a total initial cost of $1,000,000. Alternatively, it could be built with two
transformers - twice as much capacity - at a cost of $1,800,000. Although not needed
immediately, this second transformer will be required within four years because of
continuing growth. If added at that time, it will cost $1,284,000 - a reflection of the
additional start-up cost for a new project and of working at an already-energized and in-
service substation rather than at a "cold" site. Which plan is best? Should planners
recommend that the utility spend $800,000 now to save $1,284,000 four years from now?
A second and related cost decision involves determining if a present expense is justified
because it will reduce future operating expenses by some amount. Suppose a new feeder is
to be built along with the new substation. If built with 336 MCM conductor at a cost of
$437,000, the new feeder will be able to satisfy all loading, voltage drop, contingency, and
other criteria. However, if built with 600 MCM conductor, at a total cost of $597,000, it
will lower annual losses costs every year in the future by an estimated $27,000. Are the
planners justified in recommending that $160,000 be spent on the larger, based on the long-
term continuing savings?
These decisions, and many others in distribution planning, involve comparing present
costs to future costs, or comparing plans in which a major difference is when money is
scheduled to be spent. To make the correct decision, the planner must compare these
different costs in time on a sound, balanced basis, consistent with the electric utility's goals
and financial structure.
Common Sense: Future Money Is Not
Worth as Much as Present Money
Few people would happily trade $100 today for a reliable promise of only $100 returned to
them a year from now. The fact is that $100 a year from now is not worth as much to them
as $100 today. This "deal" would require them to give up the use of their money for a year.
Even if they do not believe they will need to spend their $100 in the next year, conditions
might change.
A utility has similar values, which means that when money has to be spent it can have a
value of its own - in general the later, the better. For example, suppose the planning process
has determined that a new feeder must be built at a certain site within the next five years in
order to serve future load. It will cost $100,000 to build, regardless of whether built this
year, next year, or five years from now. If there are no other reasons to build it any earlier,
the best decision is to wait as long as practically possible - in this case for five years. One
hundred thousand dollars spent five years from now is future money, not present money,
and while it is the same number of dollars it is not worth as much to the company at the
moment:

The value of money available only in the future is less than
the value of that same amount of money available today.

Of course, there could be sound reasons why it will benefit the utility to build the feeder
earlier. Perhaps if added to the system now it will help lower losses, reducing operating cost

142 Chapters

over the next five years by a significant amount. Perhaps it will improve reliability and
service quality, reducing the risk of customer outages and penalty repayments under
performance-based contracts. Thus, the sooner it can be built, the more the utility will save.
Or perhaps if built a year from now, during a predicted slump in the local construction
industry, the utility believes it can negotiate particularly advantageous rates from the
contractor it will hire to do the construction. Or perhaps some of the cable and equipment to
be used is expected to rise in price over the next few years, so that if the utility delays even
a year, the feeder will eventually cost a bit more than $100,000.
There are often many possible reason why it might be prudent to build earlier than
absolutely necessary. A planner's job is to balance factors like these to determine how they
contribute to the decision of when to build an item or actually commit to an expense and to
identify what alternative with regard to timing has the lowest overall cost (or, alternately,
the highest benefit) to the company.
Present Worth Analysis
Present worth analysis is a method of measuring and comparing costs and savings that
occur at different times on a consistent and equitable basis for decision-making. It is based
on the present worth factor, P, which represents the value of money a year from now in
today's terms. The value of money at any time in the future can be converted to its
equivalent present worth as

Value today of X dollars t years ahead = X x P1 (5.1)
where P is the present worth factor
For example, suppose that P = .90, then $100 a year from now is considered equal in value
to today's money of
100 x (.90) = $90
and $100 five years from now is worth only

$ 100 x(.90)5 = $59.05
Present worth dollars are often indicated with the letters PW. Today's $100 has a value
of $100 PW, $100 a year from now is $90 PW, $100 five years from now is $59.05 PW,
and so forth.
Alternately, the present worth factor can be used to determine how much future money
equals any amount of current funds, for example, to equal $100 present worth, one year
from now the utility will need
$1007.90 = $111.11
and two years from now, one would need $100/(.90)2 = $123.46 to equal a present worth of
$100 today.
A continuing annual future cost (or savings) can be converted to a present worth by
adding together the PW values for all future years. For example, the present worth of the
$27,000 in annual losses savings discussed earlier in this chapter can be found by adding
together the present worths of $27,000 next year, $27,000 the year after, and so forth
00

PW of $27,000/year = Z ($27,000 x P1) (5.2)
t=l


= $27,000 x (Z
t=l
30
= $27,000 x
t=i
= $258,554

Discount rate
Present worth analysis discounts the value of future costs and savings versus today's costs
and savings, as shown in Figure 5.6. The discount rate used in an analysis, d, is the
perceived rate of reduction in value from year to year. The present worth factor is related to
this discount rate:
P(t)=l/(l+d) t (5.3)
where d = discount rate
and t = future year
If d is 11.11%, it means that a year ahead dollar is discounted 11.11% with respect to
today's dollar, equivalent to a present worth factor of P = (1/1.111) = .90. Therefore,
$111.11 a year from now is worth $111.11/1.1111 = $100. A decision-making process
based on the values o f d = 11.11% and PWF = .90 would conclude that spending $90 to
save $100 a year from now was a break-even proposition (i.e., there is no compelling
reason to do it), while if the same $100 savings can be had for only $88, it has a positive
value. A higher discount rate - a lower present worth factor - renders the decision-making
process less willing to trade today's costs for future costs. If the discount rate is doubled, to
22.2%, P drops to .82. Now, $88 spent today to obtain a $100 cost reduction a year from
now is no longer viewed as a good investment: that $88 must save at least $88/1.222 = $108
a year from now to be viewed as just break even.

1.00

.90

.80

.70

? '6°
» .50
«
> .40

.30
.20

.10

7 8 9 10 11 12 13 14 15 16 17 18 19 20
Year
Figure 5.6 Present worth analysis discounts the value, or worth, of a dollar the further it lies in the
future. Shown here is the value of one dollar as a function of future year, evaluated at a .90 present
worth factor (11.1% discount rate).

144 Chapters

Present worth analysis does not say "no"
to truly essential elements of a plan
Power system planners, with a commitment to see that service is always maintained, often
fall into a trap of wanting to build everything as early as practicable so as to have plenty of
capacity margin. As a result, they come to view present worth analysis, and any other
aspect of planning that may say "delay," as negative.
But this is not the case. If applied correctly, present worth analysis never says "no" to
essential expenses - for additions or changes which must be done now. For example,
suppose a hurricane has knocked down ten miles of critically needed line. Then there is no
question about the timing of this expense. It must be done as soon as other conditions
permit, and present worth analysis is not even appropriate to apply to this decision.
Present worth analysis is used to evaluate and rank alternatives only when there is a
difference in the timing of expenses - the substation could be built this year or next year -
or when costs and savings that occur at different times must be balanced - more spent today
will lower tomorrow's costs. It is an essential element of keeping cost as low as possible
and it should be applied in all situations where alternatives differ with respect to when
expenses will be incurred. The present worth factor is the tool which makes these
comparisons valid within the utility's value system.
How Are Present Worth Factors
and Discount Rates Determined?
There is no one rule or set of guidelines that rigorously defines what contributes to the
present worth factor (or its companion, the discount rate) as used by all utilities. There is a
rigorous financial basis for discount rate, briefly discussed below. But as used by utilities,
these numbers often include other factors, and there is no inviolate formula like Ohm's law
that lays out completely and with no exceptions how to compute the discount rate and the
PWF.
Quite simply, the best advice for planners is to treat PWF simply as a number to be used
in planning. This does not mean they should blindly accept it without an understanding of
its basis. Instead, the point is that one should not assume it has only the purely financial
basis one sees explained in business finance tests. Discount rate is often determined through
careful analysis and computation, but sometimes it is an empirical estimate "that just works
well," a number the utility's upper management has come to regard as prudent. Thus, the
recommended perspective is that PWF is a decision-making tool that allows planners to
compare future expenses to present ones.

For planning purposes, the present worth factor
should be regarded simply as a value that sums
up all the reasons why a company would prefer
to spend money tomorrow rather than today.

There can be many reasons why a utility may wish to limit spending today in favor of
spending tomorrow, or in rare instances wish to spend now instead of in the future. These
reasons all influence the selection of a value for the present worth factor. Priorities vary
greatly from one company to another and also change over time. Among them is one reason
planners often find difficult to accept: "We just don't have the money, period." In most
utilities, present worth factors and discount rates are determined by a utility's financial
planners and based on the company's requirements for financing and use of money. The
methods applied to fix these requirements quantitatively and their relation to PWF are


beyond the scope of this discussion and available in other texts (IEEE Tutorial on
Engineering Economics). However, it is worth looking at some of the primary influences on
present worth factor and how and why they impact it.
Interest rate
The fact that it takes more future dollars to equal a dollar today is often attributed to interest
rate: a person who has $100 today can invest it at the prevailing interest rate, i, so that a
year from now it will be (1+i) times as much. Thus, $100 invested today at an annual rate of
5% interest, will be worth $105 a year from now and $128 five years from now. If the
prevailing interest rate is 5%, then it is cost-effective to spend $100 only if it will save or
return a value a year from now that exceeds $105. Otherwise it would be better to simply
leave the money invested and drawing interest. A PWF of .952 can be applied in present
worth analysis as described earlier and will lead to decisions that reflect this concern.
Interest rate, financial risk, and expected gain
Exactly what interest rate does one use to compute the discount rate? At any one moment,
there are many interest rates being quoted in the financial community: the prime rate, rates
quoted for mortgages, rates offered on new cars to buyers with good credit ratings, rates
offered to new car buyers with poor credit ratings, credit card rates, new furniture financing
rates, etc. In general, interest rates reflect the level of risk that creditors see in a person,
institution, or business model. A low element of risk means a relatively low interest rate,
and vice versa. Ideally (theoretically) creditors adjust their interest rates to take into account
the risk inherent in each loan or investment, with a goal of assuring that their expected
return is positive and about the same from all loans.1
Financial analysts regard discount rate and PWF as reflections of the cost of capital.
Suppose that a particular creditor, a bank, can loan $100 today and it will be repaid $110
next year, but that the risk, due to all causes, means that it expects only an average 96% net
repayment on loans of this type (defaults and other problems absorbing the other 4%). The
expected gain is only $110x.96-$100 = $5.60. In this situation, it would be foolish for
this creditor to spend $100 now to avoid an expense of $104 next year. Better to invest the
money, earn an expected $5.60, and pocket $1.60 of profit after paying the $104 cost next
year. Since a creditor applies this same logic over many loans and investments, his portfolio
pretty much earns the expected gain (5.6% in this case).
Basically, the discount rate a borrower uses in planning is related to the cost of capital
he "sees" in any situation. The company that took the 10% loan in the paragraph above
would look at the situation like this. In order to spend $100 today to save $104 next year, it
will have to borrow another $100, repaying $110 a year from now, meaning it loses $6.
Better to defer the expense for a year. Similarly, the new car buyer with a poor credit rating
should use a discount rate in his decision-making (with respect to his new car) that reflects
the high interest rates he will be charged. And a utility with a solid bond rating (and thus
able to borrow money at a much lower rate than many other businesses) should use that

1
From the creditors' standpoint, if they can earn more expected return by making one type of loan or
investment rather than another, they will - it is the best use of their money. Therefore, creditors
quote rates for various types of loans and invest money in various ventures and companies with the
same overall expectation of growth (profit) from each. For most small (personal, small business)
loans, banks and lending institutions use tables and statistics to group borrowers into risk categories.
For large loans and investments, they make detailed assessments of both risk and the probability of
various scenarios of not being fully repaid. All of this is ultimately reflected in the rates that a person
or business sees offered to it for credit.

146 Chapters

lower interest rate. The discount rate used in any situation would reflect the cost of capital
for the situation.
Interest rate, risk, and cost of capital alone do not explain PWF
While theoretically the discount rates and present worth factors used by any business should
be based on its cost of capital, in practice the present worth factor used by many utilities
clearly represents something more than just its cost of capital, because present worth factor
as applied by most electric utilities is nearly always greater than what corresponds to the
prevailing capital costs alone. For example, at the time of this writing, the inflation-adjusted
interest rate on safe, long-term investments is about 3.0%, yet most utilities are using a
present worth factor of about .93, equivalent to a 7.5% interest rate. The difference, 7.5%
versus 3.0% in this case, is attributable to other factors, including conservatism and risk
avoidance. Although it may prove difficult or impossible for a distribution planner to
determine and quantify all of these reasons, it is worth understanding the other factors that
often go into the determination of a high PWF.
Inflation
Inflation is not one of the factors normally taken into account by the present worth factor,
although this is often misunderstood and neophytes assume it is a part of the analysis.
Inflation means that what costs a dollar today will cost more tomorrow - 3% annual
inflation would mean that same item or service will be priced at $1.03 a year from now.
While inflation needs to be considered by a utility's financial planners, distribution planners
can usually plan with constant dollars - by assuming that there is no inflation.
The reason is that inflation raises the cost of everything involved in distribution
planning cost analysis. If inflation is 3%, then on average a year from now equipment will
cost 3% more, labor will cost more 3% more, as will paperwork and filing fees and legal
fees and taxes and replacement parts and transportation and everything else. More than that,
over time inflation will call for similar adjustments in the utility's rates and in the value of
its stock, its dividends, and everything associated with its expenses and financing (including
hopefully the planners' salaries). It affects nearly every cost equally, at least to the extent
that costs can be predicted in advance.2 From the standpoint of planning, inflation makes no
impact on the relative costs of various components, so it can be ignored, making the
planning process just a little bit easier.
Some utilities do include inflation in their planning and their present worth analysis. In
such cases, an increment to account for inflation is added to the discount rate. Given a 5%
interest rate and a 3% inflation rate discussed above, this would mean that the planners'
present worth factor might be

2
In actuality, planners and executives know that inflation will not impact every cost equally, but
small variations in individual economic sectors have proven impossible to predict. In cases where
there is a clear indication that certain costs will rise faster or slower than the general inflation rate,
that difference should be taken into account. This was mentioned at the start of this section when a
possible rise in cost of materials for the candidate feeder that could be built early was discussed as
one reason that might lead the utility to decide to build it early.


Present worth analysis would now compare today's investment measured in today's dollars
against tomorrow's investment measured in tomorrow's inflated dollars. This type of
accounting of future costs must sometimes be done for budgeting and finance estimation
purposes.3
While inflation must be included in the analysis,
planning is best done with "constant dollars"
However, while "inflated dollars" present worth analysis sometimes has to be done to
estimate budgets, it is rarely an effective planning tool. Planning and decision-making are
facilitated when expenses and benefits are computed on a common basis - by measuring
costs from all years in constant dollars. Is plan A, which calls for $1.35 million in expenses
today, worth more than plan B, which costs $1.43 million three years from now? If both are
their respective present worth measured in constant dollars, then it is clear that plan A is the
less costly of the two. But if inflation were a factor in the present worth analysis, then one
has to do a further adjustment of those numbers to determine which is best (at 3% inflation,
$1.35 million in three years will equal $1.48 million in then inflated currency, so plan B at
$1.43 million three-year-ahead inflated dollars is the better plan). Beyond this, errors or
unreasonable cost estimates for future projects are much more difficult to catch when
expressed in non-constant dollars. For these reasons, present worth analysis in constant
dollars is strongly recommended for all planning, even if non-constant analysis may need to
be done for some other purposes.
Earnings targets
The present worth factor for a dollar under the utility's control should be higher than 1/(1 +
interest rate), because the utility must be able to do better with its earnings than the
prevailing interest rate. If a utility cannot better the prevailing interest rate with its own
investment, then it should liquidate its assets and invest the results in earning that interest
rate. Frankly, it won't get the chance to do so: its shareholders will sell their stock, take
their money elsewhere, and invest it in companies that can beat the prevailing interest rate
through their investment of their stockholders' money.4
Therefore, the goal of any investor-owned utility must be to use its money to earn more
than it could by other, equivalently safe means. As a result, while the prevailing interest rate
may be 5%, the utility's financial planners may have determined that a 12% earnings
potential on all new expenditures is desirable. Rather than a PWF of .952 (5% interest) the
utility would use a PWF of .892 (12%) to interject into the decision-making process its
unwillingness to part with a dollar today unless it returns at least $1.12 a year from now.5
This is one tenet of MRR planning, that all investments must return a suitable "revenue"
earning.

3
For example, when estimating a future budget, the costs often do have to be put in dollars for the
years being estimated, and those are inflated dollars.
4
Planners from municipal utilities may believe that this does not apply to their company, but that
is not necessarily true. If a municipal electric department cannot "earn" more from its electric
system investment than other businesses could, it is costing the city and its voters money, designing
a system and a "business" that needs to be subsidized. If such subsidized operation is the policy of
the city government, the planners should make the most of it and try to get as sound a return as
possible on what money is spent.
5
This concept strikes many planners as bizarre, yet it is completely valid and is in fact only common
sense from a business perspective. If a company expects to earn 9% return on its money, then as a rule

148 Chapters

"Internal" monetary risk
One hundred dollars invested with the promise of $108 payback in a year can look like a
good investment (at least when prevailing interest rates are only 5%), but it is a good
investment only if there is a very small likelihood that things will go so wrong that the
principal will be lost. Such cases are rare, particularly if investments are made prudently.
But there are other types of risks, similar in basic nature, that a utility faces every day. For
example, suppose that the utility spends $100 on system expansion in order to save $108 a
year from now. However, shortly after this spending is complete, a severe storm hits the
utility system, causing widespread damage. The utility may now desperately wish that it
had that $100 to pay for repairs and rebuilding of its system, work that it simply has to have
done. It has no choice in the circumstances but to borrow the money needed at short-term
interest rates, which might be 12%. In retrospect, its expenditure of $100 to save $108 a
year later would look like a poor choice.
In practice, the present worth factor used by a utility, at least the implicit PWF that is
really applied in its decisions, often includes a bias or margin to account for this type of
"bird in the hand" value of money not spent. By raising the PWF from 8% to 10% or even
12%, the utility would be stating that, yes, perhaps $108 is the year-ahead earnings goal, but
simply breaking even with that goal is not enough to justify committing the company's
resources: today's money will be committed only when there are very sound reasons to
expect a better than minimum return.
Planning errors
In addition, the present worth factor often implicitly reflects a sobering reality of planning -
mistakes cannot be avoided entirely. The author has devoted considerable effort to the study
of the distribution planning process itself, particularly with regard to how accurate it is and
how and why planning mistakes or non-optimal plans come about (Willis and Northcote-
Green). Under the very best realistic circumstances, even the finest planning methods
average about 1% "planning error" for every year the plan is extended into the future.
Distribution expansion projects that must be committed a full five years ahead will turn out
to spend about 5% more than could be arranged in retrospect, if the utility could somehow
go back and do things over again, knowing with hindsight exactly what is the minimum
expenditure it needs to just get by.
Adding 1% - or whatever is appropriate based on analysis of the uncertainty and the
planning method being used - to the PWF biases all planning decisions so that they reflect
this imperfection of planning. This makes the resulting decision-making process a little
more reluctant to spend money today on what appears to be a good investment for
tomorrow, unless the predicted savings includes enough margin over the element of risk to
account for the fact that the planning method is simply wrong.
Spending limitations
Raising the discount rate (lowering the PWF) that is used in decision-making tends to
reduce the amount of money spent today with respect to any view of expenses in the future.
Utilities that are hard-pressed for cash often raise the discount rate used in their decision-
making for this reason alone, far out of proportion to their actual cost of capital as an
expedient way to limit capital spending. This is discussed further in Chapter 6, section 6.4.

it should never invest in anything that is projected to earn or repay less than that rate. Since some of
the unavoidable elements of its expenses may not have the potential to earn this rate, the expected
return on the rest may have to be even higher.


A Planner's Perspective: Present Worth Factor Is
a Decision-Making Tool, Not a Financial Factor
The real point with the foregoing discussion is that present worth analysis is a decision-
making tool, one used to determine not only which alternative is best but when money
should be spent. It can embrace some or all of the factors discussed above, as well as others.
But from the planner's perspective, there is no "naturally right" number to use as the
present worth factor, one that should be used regardless of his or her company executive's
opinions. Present worth factor is merely a value, a planning tool, that is set by priorities,
goals, willingness to take risk, etc. In a very real sense, one can say that it is determined by
the results it will give: that is actually how many utilities settle on the PWF (or discount
factor) that they will use - the "right" one is the one that gives the budget impact and
"deferred spending" results they want.

Present worth factor is a planning tool. It is used to
evaluate and rank alternatives based on when they call
for expenditures and leads to prioritization of options.

A relatively low PW factor means that the planning process will be more likely to select
projects and plans that spend today in order to reduce costs tomorrow. As the PWF is
increased, the planning process becomes increasingly unwilling to spend any earlier than
absolutely necessary unless the potential savings are very great. A very high PW factor will
select plans that wait "until the last moment" regardless of future costs.
Two utilities with sound management but different situations might select very different
present worth factors for their planning, factors reflecting their circumstances, as
summarized in Table 5.1. This table shows the inputs and resultant analysis of PWF values
used as planning guidelines in the mid-1990s by a large investor-owned utility in the
northwestern United States and a municipality in the southwestern United States.
Example Present Worth Analysis
Returning to the time value of money decisions that started this section, the first question
was whether to build a new substation now with two transformers or to wait four years to
add the second unit. The second unit can be added while the substation is being built (now)
for only $800,000, even though it is not really needed for four years. But if the utility waits
to add it four years from now, it will cost $1,284,000 at that time. The higher cost is higher
then because of the cost of starting up a separate project for it and working at an energized

Table 5.1 Discount Rate "Computation" for Two Utilities
Factor IOU Muni Comment
Prevailing interest rate 5.7% 5.1% Municipal bond rate is lower than lOU's.
Inflation factor Both do constant dollar planning.
Earnings target 5.5% Municipal has no need for "earnings."
Risk 1.5% 3.0% Political cost more serious than financial.
Planning error 1.0% 1.5% This IOU has the better planning method.
"We just don't have funds" 5.0% This municipality is nearly "broke."
Total discount 13.7% 14.6%
Equivalent PWF 87.9% 87.3%

150 Chapters

substation rather than a "cold" site as it is now. Evaluated with a 90% PWF (discount rate
of 11 %) the two scenarios become
Present worth of spending $800,000 on the substation now = $800,000
Present worth of spending $1,284,000 four years from now = $842,400

The answer is yes, spending the money in year one is justified by the "savings" in present
worth, of roughly 5%.
Often, it is useful to determine if the decision is particularly sensitive to the value of the
PWF in use. Would a slight change in the perceived value of future money have changed
the recommendation? This can be determined by looking at what the present worth factor
would have to be for the two alternatives to appear equal in PW:

Present worth decision limit = (cost in year ti/cost in year t2)l/(L2't^ (5.4)

= (800,000/1,284,000)1/4

= .888, equivalent to a discount rate of 12.6%
In this case, while a present worth decision limit of .888 seems quite close to the .90
present worth factor used in the analysis, a comparison of the corresponding discount rates
- 11.1% versus 12.5% - shows that they are substantially different. The gulf between these
values indicates that at least one of the factors discussed earlier in this section (risk,
planning error, etc.) would have to change noticeably for the appropriate PWF to change
enough to affect this decision. Thus, this recommendation seems sound: the substation
should be built with both transformers in place.
Looking at the second planning decision that headed this section, is a $160,000 expense
justified on the basis of a continuing $27,000 annual savings? Using the same 90% PWF
(discount rate of 11%) the two costs are:

Present worth of spending the $160,000 this year = $160,000
Present worth of savings of $27,000/yr. for 30 years = $258,600

So again the answer is yes, the feeder's greater initial cost is more than justified by the
continuing savings in losses. It should be built with the larger conductor.
Determining the present worth decision limit in this case is slightly more complicated,
but can be accomplished iteratively in a few steps using a calculator or a spreadsheet. In this
case it is .832, equivalent to a discount rate of 20.2%. Thus, this is a very strong
recommendation.
Comprehensive Present Worth Example
In general, actual planning problems are more complicated with regard to the
interrelationship of long-term and initial costs than either of the two examples above. Table
5.2 and Figure 5.7 show more cost details on the two-versus-one substation transformer
decision given at the beginning of this section and analyzed on the previous page. However,
this analysis shows the type of cost and time details that actual planning situations must
address. In the case shown here, the decision is being made now (in year zero), even though
the substation will actually be built four years from now (the decision must be made now
due to lead time considerations).


Table 5.2 Comparison of Yearly Expenses by Category for an Eight Year Period
Alternative A Alternative B
Build initially with two Build initially with one
Year Load-MW Capital O&M&T Losses Capital O&M&T Losses
0
1 20 20
2 370 290
3 12.0 1010 52 54 690 44 39
4 15.5 110 119 92 99
5 18.5 110 130 92 122
6 21.3 110 142 80 92 147
7 23.8 110 155 562 101* 164*
1400 492 600 1642 317 666
Total = 2492 Total = 2633
Taxes and losses for the transformer added in year eight (alternative B) are pro-rated for that year,
since it is assumed here that it is installed just prior to summer peak (six months into the year).


1.0 1.0

. 75

i
x I .50 .50

.25 .25

•1 Capital O&M&T Krfi Losses

Figure 5.7 Yearly expenses for the two alternative plans for the new substation (constant dollars).

Here as before, alternative A calls for building the new substation with two 25 MVA
transformers at a total cost of $1,400,000. The substation is actually now completed in year
three, in time to be on line by year four. Alternative B defers the second transformer by four
years (installation deferred from year 3 to year 7). As before, there is a four-year shift in
some of the expenses, and the addition of the second transformer four years later is
estimated to cost $642,000 versus $400,000, as before.
However, this analysis provides much more detail on all of the expenses, and their
timing. The decision involves other costs than just initial capital, which while of secondary
importance, should be considered in any complete analysis. For example, no-load losses,
taxes, and O&M will be greater during the first four years in Alternative A than in
Alternative B. Table 5.2 and Figure 5.7 compare the expenditure streams for the eight-year

152 Chapters

period for both alternatives, including capital; operations, maintenance, and taxes
(O&M&T); and losses.6
Note that the capital for substation construction in either alternative is not spent entirely
in the year the substation is completed. A small amount is spent two years earlier (for
permits, the site itself, and clearing and preparation of the site) and some equipment is
purchased in the following year so that work can begin on schedule.
Comparing differences in the two alternatives, note that the second transformer in
Alternative A increases O&M&T costs during the first four years of operation because it
has value (taxes must be paid on its additional value) and it must be maintained and
serviced even if not heavily loaded. The difference in losses costs is more complicated.
Initially, losses are higher if two transformers are installed because that means twice the no-
load losses, and initially either one or two transformers would be lightly loaded so that there
is no significant savings in load-related losses over having just one transformer. But by year
six the two-transformer configuration's lower impedance (the load is split between both
transformers) produces a noticeable savings in reduced losses.
Table 5.3 and Figure 5.8 show the various costs for Alternatives A and B from Table 5.2
and Figure 5.7 converted to PW dollars using a present worth factor of .90 (equivalent to p
= 11.11%). Note that based on the present worth of capital costs, Alternative A is rated the
lower present worth cost ($1,054,000 versus $1,067,000), in keeping with the preliminary
evaluation done earlier. The margin of present worth savings shown in Table 5.4 ($13,000)
appears to be smaller than the $21,200 difference computed several paragraphs above,
largely because in this evaluation all the costs and savings are pushed three years into the
future (the substation is built in year three, not year zero), and thus both costs and savings
have been lowered by another four years of present worth reduction (i.e., .9 = .73).
However, more important is the fact that this more complete analysis shows that
Alternative B has a lower present worth by $51,000 PW - a margin in the other direction,
and one nearly four times larger than the difference in present worth based on capital alone.
What changed the balance in favor of Alternative B was the continuing fixed operating
costs - maintenance, taxes, and no-load losses - which are higher when two transformers
are installed.
The present worth difference between these alternatives, $51,000, is slightly more than
3% of the total PW value of either alternative. Considering the straightforward nature of
this analysis, and the amount of cost common to both alternatives, it is unlikely that the
analysis of this difference is in error by anything like this amount. Therefore, this is very
likely a dependable estimate: it is likely that Alternative B's PW costs really will be lower
over the long run.
Looking at the sensitivity of this decision to present worth factor, the two alternatives
would evaluate as equal at a present worth factor of .937, equal to a discount rate of 6.77%.
Present worth factor would have to be higher (i.e., the discount rate lower) for the future
savings to outweigh the increased initial costs. Valid PWF values less than .92 are
extremely uncommon, and therefore the recommendation to build the substation with one
transformer seems quite dependable, regardless of the exact financial considerations.

6
The table and figure cover the only period during which there are differences that must be analyzed in
the decision-making. At the end of the eight-year period, the two alternatives are the same - either
way the substation exists at that point with two 25 MVA transformers installed. Present worth
thereafter is essentially the same.


Table 53 Comparison of Yearly Present Worth by Category for an Eight-Year Period
Build initially with two Build initially with one
Year PWF Load-MW Capital O&M&T Losses Capital O&M&T Losses
0 1.0
1 .90 18 18
2 .81 300 235
3 .73 12.0 736 38 39 503 32 28
4 .6
6 15.5 72 78 60 65
5 .59 18.5 65 77 54 72
6 .53 21.3 58 75 43 49 78
7 .48 23.8 53 74 269 48* 78*
1054 286 343 1068 243 322
Total = 1684 Total = 1633

Taxes and losses for the transformer added in year eight (alternative B) are pro-rated for that year,
since it is assumed here that it is installed just prior to summer peak (six months into the year).


1.0

.75 *»?
xa
£ o

> I -50
i! .25
Ii
co .25
I

0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8

Capital O&M&T Losses

Figure 5.8 Present worth of expenses associated with the two alternative substation plans given in
Table 5.3 and Figure 5.7.

Table 5A Use of Present Worth Decision Limit Analysis to Identify the Implicit Present
Worth Factor Actually at Work in a Utility's Approval Process
Recommended Project PWDL Equiv. Disc. Rate Disposition
Upgrade Eastwood substation .40% 150% Approved
Build new Knoll Wood #6 feeder .48% 108% Approved
Upgrade Echo-Wouton 69 kV line .51 96% Approved
Split bus & upgrade, Keen River subs. .58 72% Approved
Breaker upgrades, River Bend station .59 69% Approved
Add third 25 MVA transf. at Pine St. .60 66% Delayed
Larger conductor, new Knoll Wood # 5 feeder .75 33% Rejected
Reconductor Downtown #2 feeder .80 25% Rejected
Wharton subs, enhancement plan .87 15% Rejected

154 Chapter 5

Implied Present Worth Factor
A frustrating reality for many distribution planners is that occasionally projects that justify
themselves on the basis of present worth analysis are rejected by management nonetheless,
purely for budgetary reasons. For example, perhaps using the utility's PWF and approved
method of analysis, a decision is made that a new substation should be built with two new
transformers, while one would satisfy immediate needs, because the higher initial cost is
more than made up for by the long-term present worth of the savings. Yet the
recommendation may be rejected by management because "there simply isn't enough
money in the budget." Implicitly, management's approval process is using a lower present
worth factor (e.g., higher discount rate) than the official or planning present worth factor.
Often situations like this are a reality which cannot be avoided - budgetary restrictions
beyond those accounted for in the present worth factor can be forced on the utility by
unexpected or unavoidable conditions. In such cases, planners may wish to analyze the
present worth factor decision limit of projects that are approved and rejected to determine
what the implied present worth factor being used in the approval process actually is.
For example, the projects listed in Table 5.4 were decided upon as shown in 1993 by the
municipal utility whose "discount rate computation" is given in column three of Table 5.1.
While officially the discount rate in effect was 14.6% (PWF of .873), management was
implicitly using a PWF of less than .60, equivalent to a discount rate greater than 66%.
Knowledge of the payback that is expected in any project can help planners in not wasting
their or their management's time with recommendations that have little chance of approval.
Levelized Value
Often, it is useful to compare projects or plans on the basis of their average annual cost,
even if their actual costs change greatly from one year to the next. This involves finding a
constant annual cost whose present worth equals the total present worth of the plan, as
illustrated by Figure 5.9. Levelized cost analysis is a particularly useful way of comparing
plans when actual costs within each vary greatly from year to year over a lengthy period of

Alternative A - 32 Years
1.0
c
0 £
= «.75
1 1 Levelized cost, years 3-32
<» e.50

I 1.25

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 181920 21 22 23 24 25 26 27 28 29303132
Year
I Capital O&M&T CD Losses

Figure 5.9 Cost of alternative A from Figure 5.7 over a 32-year period that includes its first 30 years
of service. Present worth of all costs sums to slightly more than $3,000,000. Levelized over years
three through 30 - those in which the substation is planned to be in service - this represents an annual
cost of nearly $350,000 per year, which can be viewed as the annual cost of providing the substation
service.


time, or when the lifetimes of the options being considered are far different. It also has
applications in regulated rate determination and financial analysis. In general, the present
worth, Q, levelized over the next n years is
Levelized value of Q over n years = Q (d x (l+d)n )/((l+d)n - 1) (5.5)
Thus, the steps to find the levelized cost of a project are: (1) find the present worth of the
project, (2) apply equation 5.5 to the present worth. As an example, using Alternative A's
calculated total present worth of $1,684,000 from Table 5.3, that value can be levelized
over the seven-year period in the table as
= $1,684,000 (.1111(1.1111)7/(1. l l l l 7 - ! )
= $358,641
By contrast, Alternative B has a levelized cost over the same period of $347,780. If the
utility planned to "pay for" Alternative A or B over the course of that seven-year period, the
payments it would have to make would differ by a little more than $10,800 per year.
Lifetime levelized analysis
Rather than analyze planing alternatives over a short-term construction period - for
example the seven-year period during which the schedule of transformer installations may
vary - planners or budget administrators often wish to look at a new addition to the system
over the period in which it will be in service or be financed.
For example, in the two-or-one transformer substation planning problem presented
earlier, regardless of whether Alternative A or B was chosen, the substation would be
planned on the basis of providing a minimum of 30 years of service, and it would be carried
as a depreciating asset on the company's books for 30 years or more (i.e., the company is
basically financing it over that period). Therefore, planners might wish to both analyze cost
over 30 years and compare alternative plans on that same basis.
Figure 5.9 shows projected costs for Alternative A over a 32 year period, which includes
the first through thirtieth years of service for the substation. The total present worth of the
substation's costs over the period is just over $3 million (construction plus 30 years of
O&M&T plus losses), which represents a levelized cost of $348,500/year. By contrast,
Alternative B (the less expensive of the two construction schedules, as determined earlier)
has a levelized cost $342,200/year. It saves an average of $5,900/year or about $500/month
in terms of a 30-year budget perspective.

Comparing cost over different equipment lifetimes
Very often the lifetimes or periods of cost being considered for two or more options are not
equivalent, and yet it is desired to compare their costs on a valid basis. Levelized lifetime
cost is a good way to do so. Table 5.5 compares the annual costs of providing 1000 lumens
of illumination (about the equivalent of a standard 75 watt incandescent light bulb) for 2500
hours per year, using incandescent, halogen, or compact fluorescent lamps, which have
expected lifetimes of three, five, and six years respectively. The initial costs, the annual
operating costs, and the lifetimes of these three options all differ substantially. In each case,
initial and continuing annual energy costs are evaluated in Table 5.5 on a present worth
basis over the lifetime of the particular lamp; then the levelized cost over that period is
determined. This permits a comparison that identifies the compact fluorescent as having
less than half the annual cost of the incandescent.

156 Chapter 5

Table 5.5 Lifetime Analysis of Illumination
Incandescent
Year PWF Capital Energy Total PW
1 .90 $2.00 $23.00 $25.00
2 .81 $23.00 $23.00
3 .73 $23.00 $23.00
Total Present Worth
Levelized over 3 years = $23.74/yr.
Halogen
1 .90 $8.00 $13.00 $21.00 $18.90
2 .81 $13.00 $13.00 $10.53
3 .73 $13.00 $13.00 $9.48
4 .66 $13.00 $13.00 $8.53
5 .59 $13.00 $13.00 $7.69
Total Present Worth $55.11
Levelized over 6 years = $ 14.95/yr.
Fluorescent
1 .90 $17.00 $8.00 $25.00 $22.50
2 .81 $8.00 $8.00 $6.48
3 .73 $8.00 $8.00 $5.83
4 .66 $8.00 $8.00 $5.25
5 .59 $8.00 $8.00 $4.72
6 .53 $8.00 $8.00 $4.25
Total Present Worth $49.04
Levelized over 6 years = $11.62/yr.

Suppose that the analysis in Table 5.5 had compared all three options over a 30-year
period. That would have required a series of ten incandescent bulbs - one every three years
- six halogen, and five compact fluorescent lamps. Each of the sections of Table 5.5 would
have extended for 30 years, with appropriate PWF, computations, and sums, but each
would have arrived at exactly the same levelized cost per year for each of the three lamps.
An important point is that each was evaluated over a different period of time - its lifetime,
yet the results are directly comparable. The results, however, are identical to those that
would have been obtained from analysis of all over the same, longer, period.
Benefits and Costs Must Be Discounted by Similar Methods
One point often not appreciated by planners is that one must discount both benefits and
costs by the same discount factor in any study. Many people accept this when both benefit
and cost are measured in dollars, but do not realize that it is just as necessary when benefit is


Table 5.6 False Analysis Showing that Power Bought at
100/kWh for 30 Years Actually Costs Only 3.210/kWh
Year PWF kWH Bought Total Paid PW of Cost
0 1 10000
,0,0 $0,0
1000 $0,0
1000
1 09
.0 10000
,0,0 $100,000 $000
9,0
2 0.81 10000
,0,0 $100,000 $81,000
3 07
.3 10000
,0,0 $0,0
1000 $290
7,0
4 06
.6 10000
,0,0 $100,000 $65,610
5 05
.9 10000
,0,0 $100,000 $909
5,4
6 05
.3 10000
,0,0 $0,0
1000 $53,144
7 04
.8 10000
,0,0 $100,000 $47,830
8 04
.3 10000
,0,0 $100,000 $307
4,4
9 03
.9 10000
,0,0 $0,0
1000 $872
3,4
10 0.35 10000
,0,0 $100,000 $34,868
11 0.31 10000
,0,0 $100,000 $31,381
12 02
.8 10000
,0,0 $0,0
1000 $823
2,4
13 02
.5 10000
,0,0 $100,000 $25,419
14 02
.3 10000
,0,0 $0,0
1000 $287
2,7
15 0.21 10000
,0,0 $100,000 $20,589
16 0.19 10000
,0,0 $100,000 $18,530
17 0.17 10000
,0,0 $0,0
1000 $16,677
18 0.15 10000
,0,0 $100,000 $15,009
19 0.14 10000
,0,0 $100,000 $13,509
20 0.12 10000
,0,0 $100,000 $12,158
21 0.11 10000
,0,0 $100,000 $10,942
22 0.10 10000
,0,0 $100,000 $9,848
23 00
.9 10000
,0,0 $0,0
1000 $8,863
24 00
.8 10000
,0,0 $100,000 $7,977
25 00
.7 10000
,0,0 $100,000 $7,179
26 00
.6 10000
,0,0 $0,0
1000 $,6
641
27 00
.6 10000
,0,0 $100,000 $5,815
28 00
.5 10000
,0,0 $100,000 $5,233
29 00
.5 10000
,0,0 $0,0
1000 $4,710
30 00
.4 10000
.0.0 $100,000 $.3
429
Total 3 , 0 , 0
00000 $961,848

Price, per kilowatt hour = $ 6 , 4 / 0 0 0 0 0
91883,0,0
= 3.210/kWh

158 Chapters

measured in some other form. This is easiest to see by first observing the "absurd example"
shown in Table 5.6. That shows the annual costs and present worth costs to buy 1,000,000
kWh annually over a 30 year period at a constant price of 100 per kilowatt hour (a total cost
of $100,000 per year, each year). It uses an 11% discount rate (.9 PW factor).
What is the average price paid for power? Clearly, it is 1007 kWh. Yet the table can be
used to "show" that power costs only 3.2 leY kWh. One uses the total 30-year discounted
cost for buying the power ($961,848) and divides this discounted cost by the total amount
of power bought, 30,000,000 kWh, to get 3.2107 kWh.
This is an absurdity, of course, because every kWh bought is priced at 100. The mistake
made is in the use of discounted cost but not discounted benefit - in this case the "benefit"
is kWh, which needs to be discounted, too. Any PW benefit-cost analysis is valid only when
both cost and benefit are equivalently discounted. Here, either cost must not be discounted
or benefit must be discounted, too. In either case, one would then get an evaluated cost of
100/kWh. The sheer absurdity of this example makes this point quite clear, but this type of
mistake passes muster with many planners when made just as blatantly but in a less obvious
case. In particular, this mistake is often made when dealing with "negawatts" (energy use
avoided) due to DSM expenditures or to the price/cost of power from a wind turbine.
Chapter 29 discusses that last example in more detail.

5.4 VARIABILITY OF COSTS
Given that T&D planners have sufficient lead time and flexibility of selection in equipment,
site, and routing, and can optimize T&D arrangements without severe limitation, the cost of
moving power as perceived in their planning is usually fairly linear with respect to both the
amount and the distance power is to be moved as illustrated in Figure 5.10: double the
amount of power moved and the cost will double, move it twice as far and the cost will
double. Thus, from a long-term or "Green Field" perspective, the marginal and incremental
costs of delivering electric power are largely unconstrained and linear.
However, once a system is in place, its structure is defined, its fixed costs are sunk, and
its embedded cost structure cannot be avoided. Its design affects the capability and costs of
local power delivery. Its limitations constrain how much and where power can be delivered.
Its capacity limits impose an exponential relationship between the amount of power moved
and the losses costs. The existing T&D system creates constraints with respect to expansion
and modification plans: marginal and incremental costs may no longer be so well behaved
and linear. Cost of delivery varies within a T&D system depending on the time of delivery,
the level of the system through which power is delivered, the location of delivery, and on
any system or equipment contingencies.
Temporal Variation
The cost per kilowatt for both power and power delivery varies with season, day of week,
and time of day. The cost of the power itself rises during peak periods because of increasing
production cost (expensive generation is used only when all less expensive generation has
already been committed). Delivery cost rises because of increased losses.
While the embedded fixed costs of equipment and operations do not change during peak
conditions, the delivery cost rises nonetheless because of the non-linearity in losses costs
and the increased production cost at peak times. A typical residential load curve may have
an annual load factor of .55 and a minimum annual load that is only 20% of the annual peak


1.5

5 1.0

i
•K -5

0 1 2 3 4 5 6 7 8 9 10
Peak Annual Load - MVA

Figure 5.10 Present worth of building and operating one mile of 12.47 kV feeder with any of four
different line types, as a function of peak load carried over the next 30 years (annual load factor of
.60). Given that a planner has freedom to select which of several line types to build, as shown here, the
locus of "minimum cost solutions" is fairly linear over a wide range, approximating a linear power
versus cost relationship (dotted line).

Mid. Noon 6 Mid. Noon 6 Mid.
Hour of the Day Hour of the Day

Figure 5.11 Coincident demand for a feeder (left) and losses cost (right) on an hourly basis for its
annual peak day. While load varies by a factor of 3:1 during the day, losses vary by nearly an order of
magnitude.

Table 5.7 Cost of Losses Per Kilowatt Delivered on a Feeder as a Function of the Operating
Conditions

System Load Losses Price of Value of Losses Cost
Situation kW kW Power (cents) Losses per KW delivered
Minimum (April, Sun., 3 AM) 1200 7.9 5.0 $.40 .3
Average load (8760 hours) 3300 59.4 6.8 $4.03 1.2
Average during a contingency 4900 130.9 6.8 $8.87 1.8
Peak load (July, Tues., 5 PM) 6000 196.4 7.5 $14.33 2.5
Peak during a contingency 7600 315.1 7.5 $23.63 3.1

160 Chapters

demand. In such a situation, I2R losses at peak are 25 times their value at minimum load
conditions and roughly five times their average value over the year. Second, the cost of
energy to make up losses is greater during peak, because the cost per kWh of power being
pushed into the system is higher. Table 5.7 gives computation of losses cost on a feeder
under average, minimum, and peak conditions. Figure 5.11 illustrates the behavior of losses
cost over a typical day.
Contingency Variations
Economy of operation is normally a concern in the design of T&D systems only with
respect to "normal" conditions, and no attempt is made to assure economy during
contingency or emergency situations. During contingencies, when equipment is out of
service and temporary switching or other measures have been implemented to maintain
service, load levels, losses, and loss-of-life costs are quite high. Table 5.7 shows that during
contingency operation of a feeder (when it picks up half of the load on another feeder)
losses costs rise substantially.
Level of the System
Power delivered to customers at normal utilization voltage - 120/240 volts - travels through
all levels of the T&D system on its way to the customer. If delivered at a higher voltage
level, some levels are avoided and no cost is associated with these lower levels. Therefore,
power is typically priced less if delivered in bulk at higher voltage, and even lower if the
consumer owns or pays for the receiving-end power system equipment (substation, control,
relaying and breakers, and metering, etc.). As an example, in Figure 5.12, power delivered
at subtransmission voltage (69 kV) to a large industrial customer would not travel through a
distribution substation, feeder, service transformer, or secondary system. As a result, its cost
would be 6.4 cents per kilowatt hour, not the full 8.2 cents per kilowatt hour charged at the
120/240 volt level. Distribution costs were avoided and thus the power is priced in
accordance with just the transmission-only costs that were incurred.
Spatial Variation of T&D Cost
Cost to deliver power varies depending on location within a power system. Variations are
both due to "natural" reasons - it is difficult to deliver power to some locations like the tops
of mountains, etc., and due to "system characteristic" reasons: no matter how a delivery
system is laid out, some consumers will be close to a substation and others will be farther
away from it. In some sense, those farthest from it are most expensive to serve, those
closest to it less expensive to serve. One must bear in mind that traditional and established
regulatory opinion holds that such variations do not automatically justify proportional
pricing of electric service.
In general, the price charged by a local distribution company will be based on the
average cost to serve all consumers and applied across the local customer base without
spatial variation. There are exceptions, however; the consumer who chose to live on top of
the nearly inaccessible mountain might have to pay a premium for service to a "hard to
reach area." However, at the transmission (wholesale) level, spatial pricing is often applied.
Certain pricing mechanisms (locationally based marginal costing, etc.) do vary price by
location based on how cost to deliver power varies from one locale to another.
Regardless, cost varies on a spatial basis, and there is often merit in studying its
variations in order to understand how to reduce cost. Figure 5.13 shows the cost of power
delivery throughout a small coastal town, evaluated on a high resolution (1/8 square mile)
spatial basis. Here, as seems typical, embedded cost of power delivery varies by a factor of


Transmission

Service/Secondary/

Customer

Figure 5.12 The cost of power depends on the effort and facilities used to deliver it. Thus, price
varies depending on the level of the system from which it is delivered, as shown here. Prices indicate
the average cost/kWh.

162 Chapter 5

Figure 5.13 Shading indicates annual delivery cost of electric power on a location-dependent basis
for a small coastal city.

Figure 5.14 Maps of incremental T&D delivery cost for the region in Figure 5.13 under two different
plans. Left, a proposed new power line is routed along a ROW near the coast. Right, it is not built
because of environmental concerns. Either way, subsequent expansion costs vary spatially (a
characteristic of all T&D systems), but the locations of high- and low-cost areas work out differently.
The specific locations of where power delivery is most expensive are often a function of planning and
engineering decisions, not inherent difficulties associated with the terrain or the area's isolation etc.

to CB Hi t U» ruiEtr k«u o7 th. i-«nir«Jf«atiT«.

MARGINAL
TfcD COST OF
SERVICE
AVG. $294/kW

Shading indexes
rahlivilotd
density. Lines
show mtjornadf
andhi^wvys.

Figure 5.15 Both T&D and DSM costs vary spatially, depending on a variety of factors. Shown here
are maps of the cost of expansion of T&D and DSM within a small city. Quite obviously, the cost-
effectiveness of DSM (the ratio of these two) also varies spatially.


three among locations. Cost varies for a number of reasons: equipment in one part of the
city is highly loaded, in other parts lightly loaded; areas for the substation incur higher
losses; load density/fixed cost ratio varies from area to area.
Marginal and incremental costs of power delivery or T&D expansion also vary on a
spatial basis, often by a factor often to one, as shown in Figure 5.14. High local cost levels
are usually due to tight constraints on existing equipment capacities and land availability -
for example a lack of any more room to expand a substation regardless of what is done.
Spatial cost variations and integrated resource planning
Many aspects of integrated resource cost and benefit analysis vary geographically, whether
the IRP is done on a resource value basis (traditional IRP) or a customer value basis (retail
energy services marketing). The high local costs that occur in some areas of a power system
are one reason why "geographically targeted" IR programs are often stressed by regulators,
for even if energy efficiency, renewable resources, and other integrated resources are not
cost-effective with respect to average T&D costs, they may have high benefit/cost ratio in
the high cost areas.
The marginal and incremental costs of energy efficiency and renewable resources vary
geographically, too, in much the same way as T&D costs. Usually DSM costs vary over a
slightly narrower range than T&D costs, but as shown in Figure 5.15, the variations can be
significant nonetheless. One reason DSM costs vary geographically is that the location of
the target customer types varies. In addition, the market penetration of target appliances will
vary - for example some residential neighborhoods have high proportions of electric water
heaters, others do not.
Climatological differences within a utility's service territory can have an impact on the
distribution of weather-related appliances' loads. For example, a very high market
penetration of residential air conditioning can be found in many low-lying, coastal urban
areas. Since air conditioners in these homes run at near 100% duty cycle during peak
periods, every load controller will produce a very high reduction in coincident load. As a
result, a reasonable market penetration of A/C load control can make a very noticeable
impact on local distribution system peak loads. Yet less than 100 miles away, at higher
elevations, only a small proportion of homeowners may have A/C units and those may not
produce a significant contribution to local distribution peaks. As a result, it would cost a
great deal (many installations, many customers signed up with corresponding incentive
payments) to "buy" any significant load control reduction. Such different areas of weather
impact are called micro-climates, and will be discussed in more detail in Chapter 27. An
important point for planners to keep in mind is that micro-climates have two effects: they
change the market penetration of electric appliances and therefore the demand, which
affects the need for capital investment (initial, fixed cost), and they change load curve
shapes, which affect losses (continuing, variable costs).
5.5 CONCLUSION
All cost evaluation for power delivery planning should be based upon economic evaluation
that is consistent over the range of all alternatives. Usually, this means that present and
future costs must be compared. Present worth analysis, levelized costing, or some similar
method of putting present and future value on a comparable basis will be used. While
methods and standards vary from one company to another, the important point is that
planners apply their company's method consistently and correctly. Table 5.8 provides a
one-page summary of key points in this chapter.

164 Chapters

Cost is the total sacrifice or commitment of resources that must be expended or traded in order to gain
some desired product or end result. Typically at electric utilities, all costs for any endeavor or project
are all translated to a common basis - money.
Initial and continuing costs are, respectively, the cost of preparing or initiating an item, for example
equipment, site, and construction costs for a substation, and continuing or periodic costs for
maintaining it in operation or effect, such as maintenance and electric losses costs.
Fixed and variable costs are, respectively, those costs that will not vary as a function of business
conditions or use of a facility (such as taxes based on capital value) and those that are a function of
usage, commitment, or some other controllable aspect of the facility or endeavor (electric losses).
Costs fall into four categories: fixed initial costs, fixed continuing costs, variable continuing costs
and (rare but not unknown) variable initial costs.
Time value of money is a decision-making tool that puts less value on an amount of money, whether
a cost or a revenue, if it will occur far in the future. $100,000 ten years from now may be valued at
only $35,000.
Time value of money decisions fall into two categories. The first involves deciding whether a present
expense is justified because it cancels the need for a future expense of a different amount. The second
involves determining if a one-time present expense is justified because it will reduce continuing or
periodic future costs.
Present worth (PW) analysis is a time-value-of-money technique for comparing costs and savings
that occur at different times on a consistent basis. It is a decision-making tool to determine when
money should be spent. Although theoretically PWF and DR should be based solely on the financial
risk involved, in practice there is no uniform way that utilities determine the factors used in planning.
PW factor (PWF) is the multiplier that relates value of money in one year to its value a year later.
Value in year N+l = value in year N x PWF.
Discount rate (DR) is the amount value is discounted per year. PWF = 1/(1+DR)
Net present value (NPV) is the value of all future monies involved, all appropriately discounted over
time.
From the planner's perspective, there is no "naturally right" number to use as the present worth
factor. For planning purposes PWF should be regarded simply as a value that sums up all the
reasons why a company would prefer to spend money tomorrow rather than today.
Implied present worth factor - a value that can often be "reverse engineered" by analyzing actual
approval and rejection decisions for projects at a utility. It often shows that spending decisions are
actually being made with a much higher aversion to risk and spending than the "official" PWF
indicates would be the case.
Minimum revenue requirement (MRR) planning is aimed at keeping customer bills as low as
possible - it seeks to minimize the total amount of money the utility must collectively charge its
customers in order to cover its costs.
Levelized analysis is a method of applying time-value-of-money analysis to initial and future costs to
develop an equivalent constant annual cost.
Lifetime cost is the total cost (usually a PW sum of current and all future prices) over the lifetime of a
device or facility.
Benefits and costs must both be discounted or present-worthed by similar amounts in any analysis
that makes decisions or comparisons on the basis of benefit/cost ratios.


REFERENCES

A. V. Abbott, Electrical Transmission of Energy, Van Nostrand, New York, 1895.

T. W. Berrie, Electricity Economics and Planning, Peter Peregrinus, London, 1992.


D. G. Newman, Engineering Economic Analysis, Engineering Press, Inc., 1983.

New York, 2000.

M. P. Reeser, Introduction to Public Utility Accounting, American Gas Association and the Edison
Electric Institute, 1997.

M. Shahidehpour and M. Alomoush, Restructured Electrical Power Systems - Operation, Trading,
and Volatility, Marcel Dekker, New York, 2001.

H. L. Willis and J. E. D. Northcote-Green, "Comparison of Several Distribution Planning
Techniques," IEEE Transactions on Power Apparatus and Systems, January 1985, p. 1278.

H. L. Willis and G. B. Rackliffe, Introduction to Integrated Resource T&D Planning, ABB
Guidebooks, Raleigh, NC, 1994.

World Bank, Guidelines for Marginal Cost Analysis of Power Systems, Energy Department paper
number 18,1984.

Evaluation, Prioritization,
and Approval

6.1 DECISIONS AND COMMITMENTS

One of the chief goals of "Planning" - spelled with a capital P to denote the function and
departments) within a utility rather than just the process - is to determine the capital
projects needed to meet the utility's future needs and then to initiate the process leading to
their execution. Basically, Planning is in charge of the decision-making process with regard
to expansion - it determines if the system will have a deficiency unless reinforced, decides
when that reinforcement must be made, and then recommends what should be done.
The if aspect of planning - determining if forecasted needs will exceed capability -
requires load forecasts, risk analysis, and planning guidelines, topics that will be covered in
later chapters. But the decision about what and when to build is based on evaluation of
alternatives, and that feeds the decision-making process which selects and approves certain
projects for investment and execution and rejects others.
This chapter looks at evaluation, comparison, prioritization, and approval methods used
to assess and approve projects and programs for power delivery utilities. It is primarily
intended as a tutorial on the context and basis behind spending decisions that flow out of
utility planning, and is written in a very serial fashion. It begins in section 6.2 with a look at
the steps in the project approval process - evaluation, comparison, ranking and
prioritization, and approval. Sections 6.3 through 6.6 then look at the major evaluation-
ranking-approval paradigms that planners can use and discuss the advantages,
disadvantages, and application of each, using examples to draw attention to and contrast the
characteristics of each approach with respect to the others. Section 6.7 concludes with a
summary of key points and lessons learned. Table 6.1 lists key terms.

6.2 EVALUATION, COMPARISON, PRIORITIZATION, AND APPROVAL
Evaluation is a process of appraising a unit of equipment, a project, a program, an
alternative, or some other item which is involved in a yes/no or either/or decision. What the
"appraisal" involves, how it is determined, and how and why it is applied in the decision-
making process all depend on a number of factors which are dictated both by the nature of
the item being evaluated, the type of decision being made, and their context. Usually,
most if not all of the evaluation is quantitative, resulting in numerical assessment of

167

168 Chapter 6

Table 6.1 Glossary of Evaluation and Prioritization Terms
Term Meaning
Approval metric The measure used to appraise projects, programs, equipment purchases,
operating alternatives, or other actionable options to determine if they will be
given authorization.
Approval threshold The minimum (or maximum) value of the approval metric below which (above
which) authorization will be denied.
Attribute A factor or issue considered in evaluation and used as the factor in determining
rank or approval. This is usually cost. Regardless, an attribute is optimized in
some sense - the planner's goal is to minimize it, or maximize it, as the case
may be (compare to criteria, below). Comparison and ranking become very
difficult if there is more than one attribute involved in a decision.
Authorization Permission to spend money and commit resources to the
purchase/construction/implementation/use of the equipment/facility/
project/program/operating policy that has been submitted for approval.
Comparison To judge two or more projects against one another with respect to one or more
metrics.
Criteria Factors important in the comparison and decision-making. Unlike an attribute
(see above) criteria are not optimized, but merely has satisfy certain minimum
requirements with respect to each criterion. Service voltage level of a plan is an
example - any plan mat provides voltage within guidelines (e.g., between 114
and 126 volts on a 120 voltage scale) would be considered satisfactory in that
criterion's regard.
Evaluation (1) The process of analyzing and measuring a project, program, unit of
equipment, operating alternative, or other actionable decision against one or
more metrics.
(2) The result of the evaluation process, a report or data set giving the evaluated
measure of the item.
Metric A "measuring stick" used to rate or assess a project, alternative, or other
decision result. Many evaluations include more than one metric applied
simultaneously.
Multi-attribute Characteristic of a difficult class of planning problems in which the goal is to
optimize more than one attribute at a time, for example minimizing cost and
maximizing reliability. Ultimately, one must be traded for the other, leaving no
single, clear "winner" among alternatives.
Pareto curve A plot showing the set of alternatives in a multi-attribute decision, all of which
are "optimal" in the sense they represent an optimized tradeoff among the two
attributes.
Prioritization The process of ranking a set of projects or alternatives with respect to some
metric of need or importance.
Project An undertaking involving capital investment in new facilities or modifications
or enhancement of existing facilities.
Ranking The process of putting a set of projects or alternatives into order based on their
measures with respect to one metric or to one evaluation.
Single-attribute Characteristic of a class of planning problems in which the major planning goal
involves optimizing only one attribute (such as cost), even if many criteria
(voltage, reliability, maintainability ) are involved.

Evaluation, Prioritization, and Approval 169

alternatives for one or more key factors (metrics). However, there are cases where
qualitative evaluation is done, ideally using legitimate grade-logic (fuzzy logic) methods to
assure validity of conclusions.
Evaluation is often done so that comparison of options can be done - so that different
projects, programs, equipment, facilities, or alternative ways of accomplishing a goal or
mission can be judged against one another. One of the key aspects of setting up a good
evaluation process is to assure that it will establish a "level playing field" among the
alternatives so that any subsequent comparison is valid.
Prioritization is the process of ranking a number of projects or alternatives on the basis
of their merit with respect to some value system for which they have been evaluated and
can be compared on a uniform basis. Such ranking may be only relative (e.g., alternatives or
projects are rated relative to one another by simply listing them in order of best to worst), or
it could be absolute, in which they are ranked against a threshold of minimum acceptable
value: some exceed the threshold and are "acceptable" while others fail the test.
All of this work - evaluation, comparison, and ranking - is part of an approval process
that accepts projects that meet certain requirements and rejects those that do not.
"Approval" as used here means to authorize, commit to, or give permission to proceed on a
particular item.
Thus, evaluation is done so that projects can be compared and prioritized, to be either
cast against an approval threshold to see if they have enough merit for approval or ranked
against one another so those with greater merit gain approval at the expense of those with
less in a merit-based competition for limited funds. Regardless, the entire evaluation-
comparison-prioritization-approval process is most easily understood by looking back from
its ultimate goal; the utility wants to assure that sound decisions are made with respect to
using its resources (money, people, facilities) to accomplish its mission.
Evaluation and Comparison
The process of evaluation includes any testing, examination, modeling, analysis, and
subsequent interpretation needed to appraise a unit of equipment, project, program, or
policy with respect to one or more salient aspects of its performance and costs. The product
of the process is the assessment resulting from this process, usually a report or information
for a report. The evaluation process must meet three requirements set by the decision-
making context within which it fits:
1. Completeness. The evaluation must cover everything important to the decision-
making process. For example, if property taxes are part of the costs being
considered in the decision-making process (investor-owned utilities have to pay
property taxes on their facilities) and if taxes would vary from one option to
another,1 then the taxes for every alternative need to be evaluated.
2. Accuracy. The evaluation process must provide accurate evaluation, sufficient
for the decision-making process. All analytical methods have limits. If the
analysis method used to estimate lifetime cost of different options for a new
substation is accurate to only 5%, then that method is not really contributing to

1
Taxes might vary for several reasons. Alternatives might vary with respect to where a facility, such
as a new substation, would be located. Some candidate locations might be within a municipal area
with a higher tax rate than at other locations. And some alternatives might not have taxes - a
decision to outsource all maintenance

170 Chapters

good decision-making among alternatives that are evaluated as costing with
only 3% of one another.
3. Valid comparison. The evaluation process must assure that comparison of all
evaluated alternatives will be valid, that it establishes a "level playing field"
with respect to the various options being considered. In some cases, this
requirement can be used to reduce the work effort required, as an all-
encompassing analysis may not be required. For example, if all the alternatives
for a new substation call for the same equipment, but vary only in where it will
be located (i.e., the substation site), then long-term service and maintenance
costs need not be considered - all options have the same equipment which will
need the same service over their lifetimes. But if options include a DSM load
reduction plan which would avoid building the substation by reducing the load,
then the maintenance costs (and a host of other items) need to be evaluated in
order to identify all salient differences among the options.
In addition to these three characteristics driven by the required quality of the process's
result (and conserving money and time) there are two practical matters that any evaluation
process must satisfy.
4. It must be efficient - timely and economical. The fact that a utility's
engineering, planning, rates, forecasting, and corporate strategy departments can
study something in great depth and intricate detail does not mean that such
effort is needed in all cases. Planners' goals should be to always meet the three
requirements listed above while using the most economical evaluation processes
possible. At times that will mean that approximations or omissions of irrelevant
factors are perfectly fine, simply the process.
5. The process must be traceable - it must leave a fully documented "trail" from
hard input data (hopefully indisputable fact) to final assessment. This is both
good practice whenever large amounts of money or value are concerned, but
also prudent given the regulatory requirements for most utilities and the need
they create for the utility to justify all spending decisions.

Single- Versus Multi-Attribute Comparison
Attributes are characteristics of a planning decision that are to be optimized (minimized,
maximized). By far the most common attribute in electric utility planning is cost. By
contrast, criteria are characteristics that are important, but which must only meet certain
requirements. One does not sacrifice any of an attribute to improve them if they are already
within an acceptable range-
Service voltage is a good example of a criterion. It is a nearly universally applied
criterion in distribution planning - the voltage at each consumer's service entrance must fall
within a narrow range of acceptable values, but as long as it does, it is completely
satisfactory. No utility should ever sacrifice more of the attribute it seeks to optimize
(money, reliability, or whatever) in order to improve a criterion (voltage, harmonic content,
etc.) that is already in a satisfactory range. With very few exceptions, electric utility
planning involves many more criteria than attributes.
The evaluation process does not treat criteria and attributes differently, but the rest of the
approval process does. In fact, attributes and criteria do not exist for the evaluation phase,
and the evaluation process makes no distinction between the two. At that stage in the
process every factor is the same, an aspect of performance or a feature of each alternative to


be assessed as to its specific value (numerical or grade). Evaluation determines the value of
the various factors associated with its scope of analysis (voltage, reliability, cost, etc. ). A
factor becomes an attribute when the utility planners decide to use it as a figure of merit in
ranking or prioritizing alternatives or to use it as an approval threshold.
Single-Attribute Planning
Single-attribute planning situations involve only one attribute, the single factor that is to be
optimized by selecting those options that minimize (or maximize, as the case may be) that
characteristic. As stated earlier, cost is almost always the attribute, while the criteria consist
of requirements for service quality and equipment performance that define what qualifies as
a viable alternative - the planner's goal is to minimize cost while meeting various
requirements, and the evaluation and comparison phase appraises and ranks various options
with regard to: (1) if they meet the criteria requirements and (2) how they "score" on the
attribute. If ranking is done, options are compared and ranked according to the attribute
(e.g., lowest cost option first, etc.). Criteria are irrelevant to that process. Single-attribute
planning problems are straightforward to optimize, although they may not always be simple
due to technical complexities. Optimization is straightforward because there is only one
factor that must be optimized - the option that scores the best in this regard is the clear
optimum alternative. When two or more attributes are involved, the situation and the
identification of the "optimum" alternative are not nearly as clear.
Multi-Attribute Planning
Multi-attribute planning situations involve two or more attributes that are to be
simultaneously "optimized." This happens whenever planners or their management want
to "maximize" more than one factor in their decisions. As an example, a three-attribute
situation develops in some distribution automation system projects revolving around the use
of data communications. One important goal is to maximize data throughput of a system, its
ability to move masses of data, such as meter reads, quickly. Another is to minimize query
and alarm response time. Very often, improving one means making design/protocol
changes that sacrifice the other. And the only way to solve such compromises without
sacrificing either is to spend more money, and minimal cost is always the third important
attribute.
But by far the most common multi-attribute situation seen by power distribution
planners is when their utility wants to minimize cost and maximize reliability, two factors
that are at fundamentally at odds with one another: ultimately improving reliability means
spending more money. One must be traded for the other. Low cost means less reliability.
More reliability means more spending.
Multi-attribute situations are not optimizable in the same sense as single-attribute
problems, for the simple reason that in most cases no single "best" alternative exists. Figure
6.1 shows a Pareto curve that lays out the "optimal options" for planners. The example
shown is from a plan developed by the author working with utility planners for an electric
utility in the western United States and represents the best possible options for enhancing
the reliability of a system serving about 110,000 consumer delivery points in a suburban-
rural region. Every point on the Pareto curve represents an "optimum" combination of
reliability and cost. Any particular level of reliability (e.g., 90 minutes) can be obtained for
no less than the cost indicated by the curve ($12.5 million in the case of 90 minutes).
Similarly, the curve can be interpreted as telling planners just how effective any particular
budget level (e.g., $12.5 million) can be made - that amount can be used in many ways but
the greatest reliability it can "buy" is a 90-minute SAIDI. Similarly, $18 million could be
spent in many ways, but the best reliability it can buy is a SADDI of 72 minutes.

172 Chapter 6

± o
<^
</)
2
TJ
2

UX ^
III

5 10 15 20 25
Program Cost (millions of dollars)

Figure 6.1 A Pareto curve shows the best options a utility has in planning the expansion and
augmentation of the distribution system serving a region with 110,000 consumer delivery
points. The utility can spend between one and twenty-five million dollars and obtain a system
with an expected SAIDI between 160 and 80 minutes. Historically, SAIDI in this region has
averaged 122 minutes, which means the utility must spend at least $7 million to keep reliability
where it has been as the consumer base grows.

Identification of spending "causes"
Figure 6.1 provides an insight into a useful product of multi-attribute analysis and one
which often justifies the additional effort required over single attribute studies. The region
in this example had historically averaged a 122 minute SAIDI. Figure 6.1 indicates that
maintaining this level of reliability in the plan will require spending of at least $7 million.
Whatever budget the utility finally decides upon, it can use this analysis to legitimately
segment what it spends into two categories:
Growth expansion, which includes all spending to meet future load growth, basically
requires $7 million. This is money being spent to provide for all "new" consumer
demand with traditional levels of service quality. This spending fits nearly all
definitions of "expansion cost" as applied by various accounting, business, rate
analysis, cost-of service, and regulatory functions within the utility.
Reliability augmentation includes anything over $7 million spent on the region's
T&D facilities. Thus, if the utility ultimately authorizes a $12 million plan, it means
that $5 million of that is for improving service reliability for all consumers in the
region, new and old alike, and can be put into appropriate accounting and
justification columns for that reason.

Working with Pareto curves
The curve's shape in Figure 6.1 deserves some discussion. Pareto curves are usually shown
in textbooks and references as smooth curves with no inflection points or sharp changes in
slope. However, the curve in Figure 6.1 shows a realistic case, with two clear break points
where the slope changes, dividing the curve into three distinct "zones." Each zone

represents the optimized application of a somewhat different approach to distribution
design, as shown in Figure 6.2. For this utility in this area, its best approach when laying
out a system aimed at an expected SAIDI in the 130-160 minute range is to use a type of
system design that might be characterized as "basic suburban radial feeder" approach.
Increased spending with this approach means using more circuits, better sectionalization
and more complicated protection schemes, and more robust equipment.
As Figure 6.2 indicates, this basic approach begins to provide diminishing returns as one
approaches 120 minutes SAIDI. To efficiently go much below 130 minutes or over $4
million in budget for this region, planners need to "shift" to a type of loop/rollover scheme
they commonly apply in more densely populated regions, a "design paradigm" that will
emphasize multi-branch feeder designs (see Chapters 13-15), install many alternative
switching points, both open and closed, and use rollover switching at key feeder tie points.
This approach provides better cost effectiveness (less cost for any level of reliability) in the
region from about 105 minutes to 130 minutes SAIDI.
For SAIDI design below 100 minutes, a type of targeted automation, using remote
monitoring and switching in a highly prioritized point by point application, obtains more
bang for the buck in both retro-fit applications (augmentation of the existing system) and
new construction. A fourth design paradigm used by the utility, an underground secondary
network approach used in metropolitan core areas, is off the scale of Figure 6.2 and not
applicable for this example.
Such analysis of different alternatives for three different design paradigms or
approaches was carried out to produce the Pareto curve in Figure 6.1, and only the portion
of each of the three curves that is lower than the other two is shown there. The composite
curve, with three segments, represents "the best that can be done" and is the needed tool for
multi-attribute decision-making.

o
<o Suburban radial

go
3 •*
C i-

•o
£

x
UJ
Targeted automation

5 10 15 20 25

Figure 6.2 The Pareto curve in Figure 6.1 represents the best of three different approaches - design
paradigms - each selected to be applied in only the region of cost and reliability combinations where it
is truly better than the other two.

174 Chapter 6

Optimizing the ratio of attributes as a first step
Ultimately, in any multiple-attribute situation, someone will have to determine how much
of one desired quantity (e.g., reliability) is worth how much of another (money), thereby
defining one point on the Pareto curve as "best." However, before doing that, there is an
"optimization" step that is often performed to assure that the curve really does represent the
best possible combinations. This step involves improving the ratio of the attributes; in this
case that means either maximizing the reliability that can be obtained per dollar spent or
minimizing the cost of attaining any given level of reliability; the two are equivalent. This is
done by focusing on the process or system for which the attributes were evaluated. In this
case it means looking at the system design guidelines and the planning process and their
effectiveness in translating money into reliability.
The dotted line in Figure 6.3 represents the reliability-versus-cost curve when using the
utility's original equipment loading, layout, protection, and design paradigms - the best it
could do before review and revision of its planning and engineering approaches. There was
nothing fundamentally wrong with those guidelines, but a few had not been significantly
revised in decades, most were quite conservative, and none had ever been updated or
optimized for "bang for the buck" using any type of rigorous, quantitative method.
The review and revision study led to changes in planning guidelines and priorities,
recognition of a new paradigm (targeted automation), and shifts in emphasis and guidelines
that ultimately moved the dotted line curve down to become the solid line, the same line as
in Figure 6.1. This revision improved the "reliability per dollar" ratio as much as possible.
Such optimization is why one can legitimately state that it is often possible to both cut costs
and improve reliability. As the arrow in Figure 6.3 indicates, the utility can spend about
$1.5 million less and still improve expected SATDI by about 6 minutes compared to its
previous capability. Alternately, it could choose to move "straight down," keeping spending
the same but picking up about 10 minutes of improvement. Or it could move horizontally,
keeping SATDI at 120 minutes but cutting spending by over $2.5 million.

o
<o

IS
CO
•D
0)
00

X
8.2
UJ

5 10 15 20 25

Figure 63 Dotted line shows the relationship prior to optimization, solid line shows improved
economic performance realizable with "optimized" planning guidelines and design rules. The arrow
indicates how a revised plan can both improve reliability and save money.


But ultimately, after the gain from this "ratio optimization" is taken, someone in the
decision-making/approval process must still decide which combination of cost and
reliability on the Pareto curve is "optimum" in terms of its two values. The Pareto curve
lays out the options available to the utility and shows the tradeoffs that must be accepted
between the two attributes, but it does not make it clear what that decision should be.
Methods to make multi-attribute decisions
Most often a multi-attribute decision is made by converting all but one of the attributes to
criteria. For example, by deciding that reliability will be set at a target of 116 minutes
SAIDI, the curve in Figure 6.3 now specifies a single selection of $8 million as the
"optimum" case. Similarly, someone in upper management could decree that "we will
spend no more than $9 million," in which case reliability can be optimized and the utility is
now going to "buy" a SAIDI of about 111 minutes.
In cases where this "convert an attribute to a criterion" cannot or is not used, a decision
is most often made on the basis of cost-effectiveness. For example, a comprehensive
analysis of consumer needs, regulatory opinion, strategy, rates, costs, and permitted profit
margin might lead the utility to decide that it can afford to spend up to $2.78 to "buy" a
customer-minute of SAIDI improvement, but no more. (This can also be viewed as a
requirement that at least 21.5 customer-seconds of avoided interruption time be obtained for
every dollar of spending on a reliability enhancement plan.)2 There are various ways of
determining such a reliability cost-effectiveness target covered in Chapters 21,23, and 28.
A $2.78 per customer-minute threshold would mean that this utility would be willing to
spend up to $305,800 per minute for SAIDI improvement in this region of 110,000
customers (110,000 x $2.78 = $305,800). Figure 6.4 shows a line whose slope is
$305,800/minute superimposed on the Pareto curve. The black dot indicates the "optimum"
point, where the curve's slope equals this cost-effectiveness threshold limit, at $9.8 million
and 109 minutes SAIDI. Dollars spent up to that point buy more reliability than dictated by
this cost-effectiveness target, dollars spent past that point buy less. Another way to look at
this is that in this region SAIDI improvements to lower than 109 minutes are simply too
expensive to justify the cost.
There are two other points on the Pareto curve that have this same $305,800/minute
slope, indicated by the Xs in Figure 6.4. Why would the utility chose the point shown rather
than either of these other two? The reason it did not choose the topmost point ($4.4 million)
is that the $5.4 million increment from $4.4 million to the $9.8 million of the second point
is also cost-effective. That increment "buys" 21 minutes, for an average of $276,190/minute
gained. The entire $5.4 million increment beats the threshold value, even if some of the
dollars in that block (those from $4.4 to $6.2 million), do not. So while the first point meets
the cost-effectiveness test, the decision to jump to the second point also meets that test.
By contrast, the third point does not meet this double requirement test. The additional
increment of spending to reach the third point from the second point, a $4.95 million
increase in spending, gains 15.5 minutes of expected SAIDI, for a cost-effectiveness of
$319,355/minute. As a block, that additional spending comes close to, but does not quite
meet, the cost-effectiveness threshold, falling about 3.5% short. No interim point makes a
better compromise, so the second of the three points is clearly the best solution in this case.

2
This value, $2.78/customer-minute, is used here only as an example of a cost-effectiveness target.
Readers should not interpret this as a value that is in any way representative of a general
recommended target.

176 Chapter 6

o
<o
(A

ss
en

X
01

5 10 15 20 25

Figure 6.4 Cost-effectiveness target is used to make a multi-attribute decision. Here, the utility's
previously set policy of spending $2.78/customer-minute of SAIDI improvement helps it identify the
optimum point- spending $9.8 million to reach an expected 109 minutes of SAIDI. Dotted line, dot
and Xs are explained in the text.

Different Decision-Making Paradigms

There are several fundamentally different ways that electric utilities can prioritize
alternatives and decide upon how they will allocate their resources among a large set of
candidate actions and activities they could undertake. These different approaches have
some slight differences in their evaluation, comparison, and selection processes, but for the
most part those functional elements are quite similar regardless of the decision-making
paradigm within which they work. For example, evaluation of a T&D project within any
paradigm requires assessment of how well it does its job (electric voltage and reliability
analysis), determination of the various elements of its cost (initial capital, construction
costs, taxes, operating and maintenance costs, etc.), and assessment of other salient factors
related to land required, permits, legal requirements, equipment, maintainability, and a host
of other issues. All of these facts about an alternative have to be assembled prior to any
determination of the project's merit and are needed by any decision-making framework.
The fundamental differences among these decision-making paradigms relate to how
they value and respond to the various evaluated factors. These various paradigms do differ
slightly in what they classify as criteria and what they use as attribute, in how they weight
one factor relative to others, and in how they treat money matters. But most important,
these paradigms vary in how they define "best" and in if and how they change their
evaluation of one alternative based on the evaluated merits and liabilities of other
alternatives that are available or that have also been accepted or rejected. The rest of this
chapter will examine the major decision-making paradigms and give examples of their
application and the different situations and perspectives each fits. Table 6.2 summarizes the
decision-making paradigms that will be covered.

CO
8
.0)

CO
i>
f
2 IS ^
.£ S CD

'*i
|"c JO
Q II

Ha
•§•• |
^£

S
To
1 1 co |

> o 2 «-
00"°^
•o ^ f |
w

Is 11
•0 N .2 x
r* >s O
a>
J: E

§f
81
g •*- c Q_ 2
ais| 1
i
•
.Q
1 1
^^
CD CD
1 i .1 -g s
C

S| S« T3 JS
.2 Q.
*^ ,-
1CD
CO "

t
s
•*••• i* T3 T3 *^ •fc c
CD CD Q. CD 'CO
o
o H H ifS £ US S 1s! a. CD

fs f5 £ N
•— 'E -i
§1 •a « .2 CD X) CO
l'o
l ;s 3
CD O
tl
CD Q.
O)
C CD
CD Q.
5 T5
.55 CD c
'^5 o .£ E
c
.55 o Q c? o x:
QJ ^3 " C
(0

i
£ CO ^— CO o *-*
"cB 3 .2
•- JO co -Q
JZ CD ic CD to == ^ CO 'C
^ CO O

•Q Q. co XI Q. CO
_CD CO § CD CO
5 -| s^
•o '= "E C c CD 3
. * « 'B
3
® i-ffl CO ~
5

ill
v O" CO ^.•8 co
3 CD TJ
CO

i co -a E
CO C 2
< «0 CO
CO T3
co c S
< CO
E
CO
I ll| 1 « 1
I
CD
S
c
1 • &'co {2 o w Q.
0) I
>

pportunities. Highest ROI
= ** CD
u
?1|
§ o=
£^E
o "55 S
E —. 3
1 ai
R.i
CO •>-
CO

To
8 ll1 «^
Minimizing long-1
NPV of costs is t

tuations and for "t
Money to pay foi

Dm petit ion with otr

ets the investment
egulated or unregi
to that limit every

pportunity to inves
uitable for restrict*

rofit margin. Ever
alient Character)

the revenues will
connect a new c

CD
As often applied:
Intent: Spend no

jck" optimization.
(0 CO
OT
>,
8
1 CO
« :*= CO
CD

'co
3
CO i •^ oi •^ CN CO DC CO 'co X) Q. 0 Q O O>

"a CD
CO o> 'S -o
H C C CO CD .C
TJ
CO *>
CD 0 c C co
§ "~ TJ CD VS .S> ^5
sz CD
obligation to serve) is just
on optional projects (thos
each project or evaluated

building for new customei

needed to satisfy the utilit

Money is invested to mak
Regulated utility. A certai

revenue they will provide.

only when benefits outwe
Optional spending. Spen
Regulated utility. Minimiz

Ask every dollar spent "w
justified by the additional
long-term or lifetime cost

amount of new spending <

have you done for me?"

CO
(0

O)

i
a
program.

5
money.

a
costs.

_o a.
£ . .Q
•o 00 CD
c o CO CO
CB u 8 o>
C w "55 •—
CO O
CO CO

X CD
'55 ^
"CD
E
2 c" g CO
CD
5 xi
s 1 1
E
T3 CD *^
en TS
C
CO 0
O
<N •g O CO
CO51 CO S
CD
C
Ov •0 CO
5 •;
S
H
§
£
1
CO
CD
•z.
1
m
CD
—
O "§
0.

178 Chapters

6.3 TRADITIONAL REGULATED UTILITY LEAST-COST PLANNING
Regulated Rates, Lowest Costs, and Rigorous Standards
The traditional electric utility approach to planning and budgeting is often called regulated
utility least-cost planning. This was the paradigm under which nearly all utilities operated
during most of the 20th century. A regulated electric utility holds a monopoly franchise on
providing electric service in its territory and accepts in return for that an attendant
obligation to serve all consumers who wish to buy electric power in the region.
A regulated utility's rates - the prices it charges consumers for electricity and/or electric
delivery - are determined by a regulatory process that is part and parcel of the legal system
that grants the monopoly franchises. Traditionally, this process provided the utility with
near certainty that it would earn an acceptable, targeted, level of profit if it practiced
reasonable control of its costs. The rates were determined by: (1) taking the utility's
projected costs, (2) adding in a "reasonable" profit margin, and (3) allocating this total
amount of money over the total amount of electricity it expected to sell to obtain the rate.
For example, the sum of all of Big State Electric Company's annual costs might have
been $1 billion dollars. Permitted profit might have been $88 million. This total, $1.088
billion, would be allocated among the expected sales; for the sake of argument assume they
are 10.88 billion kWh. This would result in an allowed rate of lOtf/kWh ($1.088 billion
divided by 10.88 billion kWh). This is a great simplification but illustrates the operative
concepts. Important "complications" to this process that the reader should be aware of are
here are:
• Profit was permitted only on capital investment. The utility makes nothing on
operating costs, per se. In the example above, the utility is making an 8.8% profit
as judged against its total revenues. However, it is earning this profit as a return
on its investment, which perhaps is only $800 million capital in its system,
working capital, and other assets. That is an 11% return on its $800 million. A
utility might have more capital invested than its annual revenues or less,
depending on a number of circumstances. But the point is return on investment
(ROI) on the rate base (the total value treated as "invested") is what determines
the profit margin.
• Rates were not the same for all consumers. In most situations, residential
consumers would get a "break" and pay something less than the average -
perhaps only nine cents per kilowatt hour, while commercial and industrial
consumers might pay a bit more for each kilowatt hour of their service. This is
called cross-subsidization of rate classes.
• Consumers who bought large amounts of power - for example industrial plants
buying multi-MW blocks from the transmission level - would receive a discount
on their price, not because of the volume they bought, but because buying at the
transmission level meant the utility did not need to buy distribution equipment to
serve them and thus would not allocate distribution costs to their rates.
As mentioned above, this regulatory, franchise-granting authority expects that the utility
will exercise reasonable control over its costs. In nearly any of its myriad forms, the
regulatory process has some means of oversight - a review and approval process - of the
utility's practices with respect to spending policy as well as its major spending decisions.
The utility is expected to practice "least-cost spending," to plan (evaluate and compare
alternatives), and commit to (approve and execute) projects and activities so that its costs to

§
CO

Q.
a
CO
o
u_

1
c
<o

2
s ex
/
*< <L>
1 Q)
X'- (A " • OT •> / 8^ ^
<o r^ —
O fg-8 i XI
"° «
"x
Is «>
+•• — — •*-•
3
0) > 42 1
c ^
3
•sr 1
lalyze for a

om standar
all circuits,

deviations

•o

**
O) O ^ CO CO «^3
'5* CD O £
5J.E
CO M=
s I
"E
c
CO
Q.
a?"o -Q
_ Q) flj CO
-2^w=
"S 5 « g
1* -*•
^ H • • i
• • • •
•t—
X
V

jj »^i cO k. < * suojiejAdp ON
CO u.
^'o

4 en

180 Chapters

provide its services are as low as possible. The utility has some incentive to invest as much
as possible in its system (the more it invests, the larger the rate base and the more it earns,
since it receives the "guaranteed" rate of return on every dollar in the rate base). However,
it must adhere to least-cost practices, and is subject to oversight so that it must spend as
little as necessary. The evaluation, prioritization and approval system that most utilities
developed to work within this framework was a standards-driven, project-oriented, least-
cost planning/approval process. Its general nature is diagrammed in Figure 6.5.
Standards-driven. What has to be done and what doesn't to assure quality of service
and consistency and efficiency in their operations by adhering to a set of standards -
guidelines for their planners, engineers, operators, and managers - that define what
types of equipment they will use, how that equipment will be utilized, how it will put
together into a delivery system, and how that system will be operated and cared for
once built.3
These standards drive the utility's planning process, although that is often not
recognized even by long time engineers involved in this type of planning. Typically,
"planning" at a utility operating under this approach is done on an annual basis,
looking several years ahead, so that if changes are needed, the utility has sufficient
lead time to make them by the time they are needed. The first step in this annual
planning process involves looking at expected changes over the next few years and
determining if any of these changes will result in a deviation from standards. Each
deviation triggers a "planning study" which verifies the need and determines a least-
cost solution. If that solution requires additions or expenditures, it leads to a project.
Early in each year's cycle, the utility's planning process will produce a projection
of changes expected over the next few years:
• a load growth forecast
• a list of any alterations to the system mandated by local
governments (substations that must be moved to make way for
new roads, etc.)
• new interconnection points with neighboring utilities
• any equipment being withdrawn from service
Appropriate technical analysis will be done to determine what impact if any these
changes will have on the system. Deviations from standards that that analysis
exposes will trigger projects aimed at correcting them. For example, if projected load
growth causes the expected peak load at a particular substation to exceed the
maximum utilization (ratio of peak load to rating) permitted within the utility's
guidelines, that will trigger a planning study to determine the "best" way to correct
the situation. All corrective actions taken as a result that lead to spending are
projects.
Project-oriented. Projects are specific capital investment undertakings that add to or
modify the utility's existing system. Each project's raison d' etre is to correct some

3
The word "standard" is used here in the traditional context of its use throughout the industry, as a
guideline or specification for the performance of a feature(s) of part of the utility system or its
apparatus. The author generally tries to avoid using the word in that context and instead uses
"guideline," because the word "standard" has recently been interpreted to mean a legally binding
requirement, such as the National Electric Code is in many places.


potential deviation seen in the future system. Projects are planned and evaluated on a
host of factors, all but one of which are criteria. Again, these criteria factors each
must satisfy its appropriate standards as the utility applies them, but none are
optimized. The single attribute is cost, which is minimized.
Thus, a planning study would consider a number of alternative ways to correct
the higher-than-permitted loading at the substation discussed above. Various
alternatives would be assessed, each for the various relevant criteria - voltage,
equipment loading and standards, etc. - and for total cost.
Least-cost means minimizing cost, which is done within this paradigm on a project
by project basis. Among viable alternatives (those whose evaluated factors fully
meet all their relevant criteria guidelines) for correcting the "overload" at this
substation, the one with the lowest cost will be selected. Cost is generally defined as
total lifetime cost — initial cost plus maintenance, taxes, etc., and any other
continuing or periodic costs over the lifetime of the unit, or over some lengthy period
of time (e.g., 30 years), computed while acknowledging the time value of money and
"brought back" to the present through present worth or a similar adjustment (see
Chapter 5).
Often the monetary factor actually minimized is revenue requirement, which is
usually applied in planning as a type of annual cost interpretation of total lifetime
cost from a regulated rate recovery perspective. It is very close to annualized cost but
includes all carrying charges and other factors. Revenue requirement gets down to
the essence of why the utility minimizes costs: to minimize the price it has to charge
its customers; to cover the total cost of the proposed changes at the substation, the
utility will have to take in so much money each year. That figure, the revenue
requirement associated with the project, is what it tries to minimize.
This is a single-attribute planning process designed to minimize cost on a project-by-
project basis. An important point and a major contrast with several of the other paradigms
discussed in this chapter is that this approach puts no constraint or control on total spending.
While the cost of each project is minimized, this process in its purest form will approve any
number of projects as long as each individually meets two requirements:
1. It is needed - it corrects a deviation from standards.
2. It is least-cost - among all candidate projects that correct the deviation it has
the lowest evaluated cost.
Traditionally (through the early 1980s), the fact that this approach had no "built in brake"
on the total budget was not a major burden to most utilities. The "guaranteed" rate of return
built into the traditional regulatory framework under which utilities operated provided a
sound enough financial foundation that each could find funds (i.e., borrow money) for
whatever spending this paradigm justified. The system worked and worked well, and the
traditional paradigm fit it perfectly. But for a host of reasons that will be addressed later in
this chapter, utility financial needs began to change by the early 21st century and most
began to be concerned about controlling the overall budget total. Most utilities responded
by changing their standards so they called for less equipment (higher utilization). Several
modified versions of this approach were developed that included measures aimed at
reducing spending. Section 6.5 will go into these in more detail and show evaluation
methods tailored to meet modern needs, and Chapters 23 and 28 will give examples of the
problems continued use of the traditional paradigm suffered in this new environment.

182 Chapter 6

Example Application
In this example, there is considerable steady load growth expected in a part of the utility
service area which will call for additions from time to time in order to both extend circuits
to new areas of the system (those previously without electric service) and to handle the
increased amounts of load. In one part of this area, planners have determined that
something must be built to extend service to new homes and businesses built in an area
where substantial load is expected very soon and long-term growth prospects are high. This
new feeder can be built in any of several types of layout, using conductor and fittings of
various capacities, as shown in Table 6.3. The numbers given are from an actual planning
case for a utility in the Mississippi valley area of the US. The conductors and layouts
discussed are of the type covered in Chapters 13 through 17.
Further details about exactly what each alternative is and how and where it varies from
other alternatives are unimportant to this discussion. What is relevant to this example is that
there are alternatives, and that they differ in both capability to handle future load growth
and in present cost. The table shows the alternatives sorted by initial cost. It gives for each
the peak demand level it can handle, whether it meets all criteria guidelines at that loading,
its capital cost (initial cost), and the total 30-year ahead PW cost, which includes that capital
cost plus the present value of losses, estimated O&M, taxes, etc., all calculated using the
methods covered in Chapter 5. If, to accommodate future load growth, an upgrade in
capability or equipment replacement is needed during the 30-year period, the PW cost of
that is included in the alternative's total PW cost. The comment column gives pertinent
details of each alternative.
The utility has to do something. Option 0, do nothing, will fail to live up to the utility's
obligation to serve all electric consumers in its service territory. The barest feeder it can
build, Option 1, using minimum wire sizes throughout, etc., would cost $2.25 million,
would carry the initial peak demand level, but would not meet at least one or more of the
many standards the utility has in place to assure service quality and proper use and lifetime
of equipment. Option 2 costs only $100,000 more initially, and is the lowest initial cost

Table 63 Example New Feeder Problem Alternatives Evaluated on the Basis of Traditional Least-
Cost Present Worth Value System at 11% Discount Rate -$xlOOO
Feeder Peak Cap. Meets Yearl PW
Option MVA Criteria? Capital Cost Comment
0 Do nothing 0 No - - Does not meet utility's obligation to serve.
1 4/0 large trunk 2.0 No $2,250 $6,750 Must be rebuilt for load growth in year 2.
2 4/0-4/0 branch 2.3 Barely $2,350 $6,530 Must be rebuilt for load growth in year 2.
3 336-4/0 branch 4.0 Barely $2,775 $5,935 High losses. Partial rebuilds in years 7 & 12.
4 336 large trunk 4.4 Suffices $2,850 $5,875 Must be rebuilt for load growth in year 8.
5 636 large trunk 5.8 Suffices $3,115 $5,728 Reinforcement needed in year 14.
6 795 large trunk 7.4 Yes $3,550 $5,694 Reinforcement needed in year 17.
7 636-336 branch 8.2 Yes $3,600 $5,685 Partial rebuild in year 17.
8 795-336 branch 9.1 Yes $3,800 $5,754 Partial rebuild in year 24. Lowest PW cost.
9 795-636 branch 9.8 Yes $3,875 $5,791 Lower losses don't justify larger conductor.
10 11 13 large trunk 12. Yes $3,900 $5,796 Lower losses don't justify larger conductor.
11 11 13-633 branch 14. Yes $4,250 $5,901 Lower losses don't justify larger conductor.


option that technically meets all standards criteria. It is the lowest initial cost among viable
options, although it has a host of less than ideal characteristics, including high losses costs
at peak, only marginal service voltage levels, and barely adequate sectionalization.
But Option 2 does meet the criteria, so it is a viable option. However, just two years will
pass before portions of it will require reconductoring in order to accommodate continued
load growth in the area. Such rebuilding is quite expensive. Spending more up front (Option
3) pushes out the date when rebuilds are needed to year 7, and cuts the amount that must be
done then. While option 3 costs $425,000 more than option 2 initially, it saves much more
than that amount in PW cost, indicating that from a least-cost standpoint, it is preferable to
Option 2.
In a similar way, one can move down the table from option to option. Each has a higher
initial cost than the option directly above it, money spent buying capacity and robust
capability which mean that the feeder can perform its function unaided by additions for a
longer period of future load growth. It also means that losses costs are less, and perhaps that
reliability is better, too. Present worth cost continues to drop with each escalation in initial
design and cost, until Option 7 is reached, at a present worth cost of $6,750,000. This
option initially costs fully $1,350,000 more than Option 2, but it cuts present worth cost by
$1,065,000. It has the lowest "long term" cost and is the option the traditional paradigm
picks to build in this case.
Why Was this Paradigm So Successful?
This basic standards-driven, least-cost approach to evaluation, comparison, and approval
selection, perhaps augmented by small additions or changes that do not change the
fundamental nature of its process, was used by virtually every electric utility in the United
States throughout most of the last two-thirds of the 20th century. It is still used today by
many utilities, although in most cases in some more modified form that attempts to limit
total spending, and that does change its nature.
The reason for its popularity for so long was that it worked very well during the period
when the power industry was completely regulated, during which utilities were treated by
investors and Wall Street as a somewhat special class of stock, for which concerns about
high debt-equity ratios were not quite as relevant as for non-regulated businesses. That
paradigm assured the utility that it would fulfil its obligation to serve. It assured that all
money spent was spent "wisely" - note that every dollar spent under this paradigm was, in a
real sense, justifiable. Furthermore, every dollar not spent (costs of options rejected as "not
worth it in the long run") was not justifiable. The utility had achieved a type of optimized
spending.
Further, the reliance on the standards guidelines to both drive the process and assure
good service quality and equipment lifetimes was an important factor, particularly when
this paradigm was first developed by the industry. It is important to remember that
engineering a system to work well was both quite difficult and very expensive in a period
before the advent of digital computers, data monitoring and communications systems, and
photocopiers. Setting high standards that were carefully and meticulously engineered, and
then applying them in a nearly cookie-cutter approach, led to good service and was efficient
for the times.
Prioritization and Approval of New Construction
At many utilities using the traditional approach, projects involving service extension of
additions to support "new load" are evaluated with a special "business case" process
designed to assure that only an appropriate amount of money is spent on the extension or
additions. Under the utility's regulated rate structure, all money it invests in its system and

184 Chapters

equipment is put into its rate base. Revenues sufficient to pay for everything in the rate base
and to assure it a reasonable return on that investment are collected from all customers.
Therefore, when a traditional regulated utility builds new facilities to serve new
customers, it ought to make certain that it does not spend more on the new facilities than is
warranted by the revenues these customers will pay, based on its rates. For the sake of
simple example, suppose that the utility has a total investment of $6.5 billion in capital in its
system and takes in $3.25 billion in revenues annually, this amount, equal to half its capital
investment, being sufficient to pay all its operating costs and leave it money enough to pay
the charging charges and other costs on its capital and give it sufficient profit. If a new mall
and associated office park is being built, and if the businesses there are expected to pay a
total of $120,000 per year in T&D delivery charges, then the utility cannot spend more
than $240,000 on new facilities and extension of its system to the mall-office part site and
keep investment and revenues consistent with its present investment, revenues, and the rates
built around them.
If the utility spends exactly $240,000, then its rate base and revenues grow in proportion
and its rates suffer no inflationary or deflationary pressure (assuming that its operating costs
for the new load also grow in proportion). If the utility has to spend more, the effect is to
increase rates (the amount now in the rate base is more than the amount covered by
revenues, so rates ought to go up in the next rate adjustment). If the utility spends less, rates
would likewise be under downward pressure.
Therefore, a process of evaluation based on the expected revenues is performed on new
service projects. It "backs out" an amount of spending that cannot be exceeded from the
expected revenues of the new consumer load. The evaluation determines expected annual
revenue based on the consumer type(s) and corresponding demand and energy sales
expected and the utility's rate(s). Capital costs, and sometimes maintenance and operating
costs for all equipment and facilities called for by the planned project for the new service
extension, are used to determine costs. The evaluation then determines if the proposed costs
are covered by the expected revenue. If not, the project is not approved. If so, it is usually
"pre-approved" subject to a final authorization by management. Approaches used by
different utilities vary in the details, but all are similar in their overall nature.
A "blank check" for new additions
What happens in most cases when this approach is used is that planners use this evaluation
process as kind of "blank check." If this "business case evaluation" says they can spend up
to $240,000 on the project, they will, even if they could get quite satisfactory new service
extended to the new load for only $100,000. Planners always have a number of as-yet-
unfunded additions or refurbishments that they have not, as yet, been able to justify. They
will work some of these into the example project until its cost is $240,000.
That approach and that attitude is completely appropriate for a traditional, regulated
electric utility operating in a financial environment where sufficient capital to meet the
traditional paradigm's "unrestricted" limit on budget is always available. First, it meets the
regulatory rules set out within which the utility must operate. Second, spending $240,000 in
this example creates no pressure in the utility's finances to change its rates, up or down.
But that approach is not consistent with the priorities of utilities operating with a more
business focus and for which sufficient, low-cost capital isn't always available. It is also not
consistent with the priorities of a utility under a rate freeze or subject to performance based
rates or trying to maximally improve reliability without escalating rates. Those utilities
usually stop the use of this type of business case review of new construction and instead
take the attitude that: (1) the new future revenues already belong to them, and (2) the goal
for new extension planning is to minimize the capital spent on extension, period.


6.4 THE BENEFIT/COST RATIO PARADIGM
Evaluating and Deciding upon Optional Activities
The traditional standards-driven, least-cost paradigm provides a simple and effective way
for an electric utility to "minimize the cost" of responding to its franchise's obligation to
serve. A key aspect of this paradigm in the context of its normal application is that the
projects it evaluates and approves are basically mandatory. While there are decisions about
exactly what will be done in each case, the fact that something must be done is beyond
question - the utility's obligation serve requires that the utility extend service and meet its
guidelines.
But utilities also face situations from time to time where they must evaluate, compare,
and decide upon projects that are optional, in the sense that they are not needed to meet
minimum standards but have some benefit that is deemed valuable. Most often, this occurs
when the projects will reduce utility or customer cost in some form. Often benefit/cost
analysis is the basis of the appraisal and decision process for such activities. It makes
decisions on project merit based on the ratio of benefit to cost (B/C), or by computing their
difference, the net benefit (B - C).
Among these two approaches, that using the ratio is most often applied to utility
decision-making because it keeps the scale of benefit in perspective. A candidate project
might cost $50 million but potentially save $51,000,000. Its net benefit would be $1
million, which might sound good. But its B/C ratio of only 1.02:1 reveals that it is barely
better than break-even. Money spent on this project is not being used in a highly effective
manner. Further, this very likely is a project that could turn out to be a loser if any number
of small factors or issues of context change - two percent is not much buffer against the
unexpected.
Thus, benefit/cost analysis, using a computed ratio of effectiveness, is most often
applied when evaluating utility projects which are optional in some sense, those not
required by the obligation to serve or the requirements of the utility's standards, but instead
by possible benefits that outweigh costs. Projects or actions with high ratios have benefits
that far outweigh their costs and are approved. Those with lesser ratios do not, and are not.
Planners most often encounter this type of paradigm when they are asked to help evaluate
special or "extraordinary" maintenance projects, or demand-side management (DSM)
programs. This section will look at two examples, one for maintenance, one for DSM.
Refurbishment-Maintenance Example
Table 6.4 shows a set of options for a project to refurbish a primary distribution feeder in a
50-year-old area of the system, where a combination of equipment aging and very slow
load growth (.3% per capita increase annually) have gradually led to low capacity margins,
above-average failure rates, and high annual interruption times for consumers in the area.
The present feeder system, though old, fully adheres to all the utility's standards. In the
strictest sense, none of the initiatives listed in Table 6.4 has to be done for the utility to meet
its design standards. This means that the "normal" planning process, which is triggered by
deviations from standards, will not recognize this situation as one needing study or
planning.
Eventually, the slow load growth in the area will create a situation where project future
utilization rates will be too high, and that will trigger the planning process and projects to
add capacity to bring it up to standards. But equipment in this area is old now, and failure
rates are high. As a result, the utility's operating costs in this area of its system are among

186 Chapter 6

Table 6.4 Reftirbishment Alternatives for Service Quality Refurbishment of an Aging Feeder in a
Stable Developed Area at 11% Discount Rate - $ x 1000
Feeder Capital O&M&A Failure Total
Option Cost PW Risk PW Comments
Base Do nothing $1,480 $1,520 $3,000 Tolerate failures and poor service
1 Fix only serious $450 $1,090 $732 $2,272 Repair only the most critical items
2 Rebuild plan A $600 $980 $543 $2,123 Rebuild trunk, repl. switches & regul.
3 Rebuild plan B $775 $865 $405 $2,045 "A" + pole insp./repl.; lowest PW
4 Rebuild plan C $1,050 $830 $120 $2,000 "B" + service trans. inspVreplt.
5 Total rebuild $2,250 $800 $0 $3,050 Complete inspecton + replacement

the highest in the company, because unexpected failures mean unexpected costs for
dispatching repair crews, often outside of normal business hours and at overtime pay rates.
In addition, some old equipment requires more routine service than newer equipment of the
same type, and spare parts are relatively costly because of their rarity compared to newer
equipment.
Table 6.4 shows the costs of various rebuild and replacement options. Its columns list,
from left to right, the option, a brief description of it, and its capital cost. The fourth column
is the PW of the O&M&A - operations and maintenance and augmentation for load
growth. This includes the cost of all expected (normal) maintenance and service for the
area, the cost of unexpected failures, and any augmentation needed due to future rebuilding,
looking as far as 20 years ahead. Column five is a failure risk (the PW of an expected value-
based consumer interruption cost "penalty" used in planning at this utility) due to those of
higher-than-average risk for future failures. The final column gives comments on the
option. Both future O&M&A costs and failure risk were calculated from the same
probabilistic analysis of equipment age and condition, load curves, circuit switching and
protection, and failure likelihood based on equipment statistics for this and other utilities
(see, for example, the IEEE "Gold Book"). They include a gradually escalating failure rate
farther into the future.
The high repair and service costs the utility sees and the penalty used to represent
customer inconvenience mean that a business case might be made for spending money to
refurbish this feeder, replacing selected equipment, and going ahead now and upgrading
capacity while this work is done, so that it will not need upgrading in a few years as load
slowly grows. The least-cost paradigm discussed in section 6.3 can be applied to select
from these options by picking the option with the least PW cost. Option 4 is the lowest. It
spends $1.015 million on a fairly extensive inspection and replacement of equipment,
reducing the PW of future O&M&A by $.65 million and PW failure risk by $1.4 million.
Its present worth is two million dollars, less than any other option and one million dollars
less than the cost of "doing nothing."
Table 6.5 shows these same options evaluated on the basis of benefit/cost ratio. The total
PW savings is the benefit and initial cost is the cost used in the B/C ratio. The analysis
shows that options 1-4 all have benefit/cost ratios greater than one. Basically, any of these
is a "good idea" in the sense that the spending buys a savings. But note that the analysis
also shows that option 3 (2.02 B/C) is a "better idea" in terms of ratio than option 4 (1.78).
While option 4 has the lowest overall present worth, option 3 gains more savings per dollar
invested than option 3. Option 2 gains still more. The decision about which option is best


Table 6.5 Refurbishment Alternatives for Service Quality Refurbishment of an Aging Feeder in a
Stable Developed Area - from Benefit/Cost Ratio Perspective
Feeder Capital O&M Failure Cost for Benefit for B/C
Option Cost PW Risk B/C Analysis B/C Analysis Ratio
Base Do nothing - $1480 $1,520 - - -
1 Fix only serious $450 $1090 $685 $450 $1,225 2.72
2 Rebuild plan A $600 $980 $560 $600 $1,460 2.43
3 Rebuild plan B $850 $865 $405 $850 $1,730 2.02
4 Rebuild plan C $1,150 $830 $120 $1,150 $2,050 1.78
5 Replace everything $2,250 $800 $0 $2,250 $2,200 0.98

depends on how one defines "best" and will be dealt with in more detail later, in Chapters
17-19, and 23-28 The important point here is that this type of evaluation context is often
applied by utilities when they are deciding upon optional projects because it addresses the
issue of whether the savings (the only reason for doing these projects, which are not
mandated by the obligation to serve) outweigh the costs of an optional activity.
Although a benefit/cost ratio greater than 1.0 indicates that savings outweigh costs, most
utilities insist that B/C ratio be considerably better before they will commit to optional
projects. This is part of the "philosophy" of choosing the B/C paradigm; optional spending
has to pay back at a higher rate than mandatory spending, or it should not be done. A ratio
of 2:1 might be appropriate for this decision, indicating option 3, not option 4, is selected.
Payback Period
An allied way of looking at benefit and cost is to ask "when do I get my money back?"
What is the payback or break even period, the time that must pass until savings equal costs.
The benefits computed in Table 6.4 are based on savings in reduced O&M and failure risk
being distributed over the subsequent thirty years. A 1.1 benefit/cost ratio indicates that
over that period, benefits outweigh the initial cost by just 10%. Of course, the benefits are
computed using present worth factor of .9 in that example. A dollar saved in year 29 is
worth only 4.70 compared to a dollar spent in year zero. For this reason, fully 72% of the
benefit computed in such a situation comes from those savings realized in the first decade.
But still, the total savings over all 30 years adds up to only 1.1 times the costs. A little bit of
analysis shows that it will take more than 18 years for the accumulated present worth of
savings to equal the initial cost.
Figure 6.6 shows the "payback period" associated with various benefit/cost ratios when
computed both using the time value of money (present worth) and without discounting over
time. Here, as in the example above, all cost is incurred in year zero and the benefits are
distributed evenly over a thirty-year period from years 0 through 29. The reader might be
surprised that the curve computed with discounted future savings lies below the one
computed without discounting (indicating payback is sooner). This is because one cannot
directly compare the two curves since any plan has a different B/C ratio depending on
whether it is computed using discounting or straight dollars. A plan with a B/C of 1.1
spends a dollar to save 11.50 annually for the next 30 years (the 30-year present worth of
11.50 annually is $1.10). The undiscounted B/C ratio of this plan is 3.45. Payback for this
plan, when evaluated using time discounting of savings, occurs at a bout 18.5 years. When
computed without discounting, it occurs at a bit less than 9 years.

188 Chapter 6

•
"; benefits computed using present worth
25
N
— _ _ — benefits computed using actual dollars

20

•o
o
"S 15
CL
c
0)
ill 10
.*
(0
0)
m

1 2 3
Benefit/Cost Ratio

Figure 6.6 Break even period associated with various B/C ratios computed over a 30 year period,
where costs are incurred in year zero and the same level of savings is realized in each of years 0
through 29.

Table 6.6 Benefit Cost Analysis of Energy Efficiency and Load Control Alternatives -
Utility in New England (1987)
DSM Cust. ($ x 1000) B/C
Alternative Class Savings Cost Ratio
Insulation blankets on water heaters resid. $3,172 $6,756 2.13
Direct water heater control resid. $2,025 $3,868 1.91
Direct water heater control comm. $697 $1,243 1.78
Direct water heater control ind. $1080 $1,555 1.44
Building shell upgrades comm. $6720 $9,010 1.34
Efficient interior lighting upgrade program resid. $8,496 $10,640 1.25
High-efficiency motor upgrade program indus. $1,336 $1,623 1.22
Efficient interior lighting upgrade program comm. $2,650 $3,214 1.21
Insulation blankets on water heaters comm. $2,040 $2,326 1.14
Efficient exterior lighting upgrade program comm. $4,200 $4.445 1.06
$32,416 $44,680 1.38


Viewed in this way, the utility's decision to insist on a 2.0 benefit/cost ratio can be seen
as a requirement that it see a relatively "short-term" back pay period on any optional
spending. Break even in this situation, with a discounted B/C of 2.0, occurs in year 5. A
ratio of 1.5 would mean it would take nearly twice as long.
Demand-Side Management Example
Table 6.6 shows the benefit/cost analysis of ten demand-side management alternatives as
evaluated on a 30-year basis by a utility in New England. Costs and benefits for each option
were computed over a 30-year period using a PW factor of .90. Unlike the previous
example, here both benefits and costs are spread over the 30-year period (options such as
lighting had periodic costs for illuminating device replacement, etc.). These options have
B/C ratios that vary from more than two-to-one to just over one-to-one. If the utility
invested in all of these DSM options, it would spend a total of $32.4 million to achieve a
projected gain of $44.7 million, for an overall benefit cost ratio of 1.38. Payback period
would be 8 years, (Figure 6.5 cannot be applied to this example because costs as well as
benefits are spread over the 30-year period. That figure is limited to cases where all costs
are initial costs, at time zero.)
In this case, the utility selected only the top three programs for implementation. Added
together ($11,876,000 PW in savings, $5,894,000 PW in costs) this three-option DSM
program had a projected benefit-cost ratio of 2.04 and a payback period of five years. This
insistence on the higher B/C ratio not only achieves the goal of "getting a lot" from any
optional spending, but has a public relations appeal: the utility can say "a dollar spent on
DSM saves two."
Resource cost effectiveness tests
"Negawatt" programs often require methods of evaluation unique to their applications, for
both regulatory and common sense reasons. Usually, these approaches apply a B/C analysis
either directly (B/C) or indirectly (B - C). There are several well-defined ways that demand-
side management (DSM) alternatives are evaluated, the two most formally identified and
used methods being the RIM test (Rate Impact Measure) and the Total Resource Cost
(TRC) measure.
These and other evaluation and ranking methods for DSM vary in exactly whose costs
and whose benefits are included in the analysis and how the net benefit is computed. For
example, are the utility's costs for equipment and systems included? Almost always. But
what about the customers' costs? Perhaps a program requires the customer to buy
something, initially or later (e.g., the utility gives away the first set of "efficient" light bulbs
but expects the homeowner to pay for replacements from then on). Another point is whether
costs and savings of non-participants in a DSM program should be included. If a peak
reduction plan (e.g., load control) is really effective, it might reduce the peak demand in a
region enough to have a noticeable impact on price of energy at peak. Everyone, including
homeowners and business owners who do not participate in the program, benefits as a
result. Should their savings in power costs be included? Should they bear some of the cost
of the program (through changes in their rates) because of this potential savings to them.
DSM analysis is full of "philosophical" questions like this.
In most regulatory frameworks, the benefits of a project, program, change in policy, or
other step that could be taken by the utility are the avoided costs. These are the power
system costs which are not incurred by the utility as a result of the decision being
considered and are usually framed against some other scenario. For example, when
evaluating energy efficiency applications, the avoided costs are the incremental costs over
the future planning period (in this case 30 years) that would be required to provide

190 Chapters

equivalent service were the energy efficiency not implemented. These might include fuel
expected to not be burned, new facilities not needed to be built, etc. These represent a
savings. The savings might also include any savings that customers see, including possibly
that due to lower energy costs, better use of resources, etc. But such savings are often not
included.
On the other end of the B/C ratio, the costs are the expenses associated with the
proposed energy efficiency program or plan and may include the costs borne directly by the
participants as well as those paid for by the utility and/or government subsidization. These
costs include such items as the technology (materials and installation), maintenance, and
administrative expenses. In the case of the analysis of a new facility or option for T&D
expansion, the savings might be the change in present worth of future expenses in
equipment, O&M&T, and losses, and the cost would be the present worth of all future costs
associated with instituting the plan.
Very often, exactly what is to be included in both benefits and costs is prescribed by law
or regulation. Again, the two major types of cost-effectiveness tests for DSM options are
the Total Resource Cost (TRC) test and the Rate Impact Measure (RIM) test. Usually a
state regulatory authority will mandate which is to be used. Sometimes, only programs that
look good when evaluated both ways will be considered positively.
Total resource cost (TRC) test
The TRC method is an analytical test which evaluates all of the direct costs and benefits to
society associated with a specific resource. Under TRC, a resource option is cost-effective
when the present worth of benefits over the planning period exceeds the present worth of
costs using an appropriate discount rate. The allocation of costs and benefits between the
utility and its participants is not a consideration when applying the TRC test - only the total
of utility+customer costs and benefits. Most often, the appropriate discount rate for the TRC
test is deemed to be the utility's discount rate.
In some jurisdictions, the TRC test is referred to as the "All Ratepayers Test." This test
identifies if and how the same end-use results can be provided more efficiently with some
integrated resource plan, considering the costs and benefits of the program to the utility and
to ratepayers, taken together.
Program Benefits in an IRP are the benefits directly accruing to both electric
utility and its customers, including avoided capital and operating savings.
Indirect benefits, such as decreased environmental impact or conservation of
resources accruing to society or tax incentives accruing to the utility, are not
included.
Participant Benefits are the incremental costs avoided by the electric utility
customers due to the IRP. As included in cost-effectiveness evaluation, these
benefits do not include the electricity purchase costs avoided, nor do they give
credit for any improvement in comfort levels.
Program Costs include both the electric utility's costs related to the program
and the total of all participant costs - if a utility pays only half of the cost of
energy efficiency upgrades, the customers' half must be included in the All
Ratepayers Test.
Participant Costs include those incurred by the electric utility's customers
which would not have been incurred but for program participation, and as
normally defined include only those costs incurred to save electricity.


Basically, the TRC or All Ratepayers Test looks at the sum of both utility- and
customer-side costs for achieving the desired end-uses of electricity - warm homes in
winter, lighting, etc. It assesses as "good" any IRP that reduces this total cost to achieve
those ends, even if that program redistributes costs so much that one side (e.g, the customer,
or alternately the utility) sees its costs increase. Such situations are viewed merely as cases
for pricing or rate adjustments that will redistribute the savings equitably. For example, if a
new energy efficiency program means that customers have to shoulder a considerable cost
even though the utility's costs go down substantially, then the utility may have to subsidize
those customer costs, through incentives and rebates, to shift the economics to the point that
both utility and customer see a portion of the savings.
A variant of the TRC test is the Societal Cost Test which includes in the evaluation the
effects of externalities - costs or issues outside of the utility/customer combination
considered by the All Ratepayers Test. This is based on analysis of the economic efficiency
of the use of society's resources as a whole to produce the desired products of energy usage
(end-uses) as a whole. Usually, regulatory guidelines will give very specific definition to
what is included in this analysis and how benefits and costs are to be computed. An
integrated resource program satisfies the Societal Test if the total cost of the program to
society is less than the total value of the resources saved by avoiding electricity production.
Costs include all resources used to produce energy savings, regardless of who pays such
costs. Benefits include all resources saved by avoiding production of electricity, including
environmental benefits and any other social benefits external to the transaction between the
utility and its customers. Transfer payments, such as tax incentives, are not included in this
calculation. The Societal Test may use a discount rate different from the utility's to the
extent of reflecting an interest rate more applicable to public projects.
Rate impact measure (RIM) test
Unlike the TRC test, the RIM test considers net lost revenues caused by reduced sales of
electricity as part of the "costs" in its evaluation. As a result, this test quantifies the
expected change in rates or customer bills that may result from an integrated resource plan.
A DSM program that reduces sales of electricity (e.g., conservation) reduces the revenues
the utility receives to cover its costs. As a result, the utility may have to raise its price per
kWh for the remaining sales so that it can collect enough revenue to cover its costs. This
means that while the DSM program reduced the use of electricity, and perhaps reduced the
total electricity bills (amount times rate) for those customers who participated in the
program, it may have raised the bills of those who did not participate, because no change
was made in their energy usage while their cost went up.
Basically, the RIM test is an analysis of the extent to which the average rate for
electricity is altered by implementation of a demand-side management or integrated
resource program. A program will satisfy the Rate Impact Test if the present worth revenue
requirements per kWh after implementation of the program are equal to or less than the
present value revenue requirements per KWh before implementation of the program,
assuming that the end-use remains unchanged. This basically means that the rate charged
for electricity cannot increase.
Comparison of TRC and RIM tests
The TRC test focuses on minimizing the cost of electricity service while the RIM test
focuses more on minimizing the price of electricity. These are similar but different goals,
their difference being due to the fixed costs of electricity and the need for a utility to pass
them on to its customers. If a DSM program cuts kWh sales significantly, electric rates -
cost per kilowatt - will increase. There are fewer kW-hours over which the utility can

192 Chapters

Table 6.7 Components of TRC and RIM Cost-Effectiveness Tests

Item TRC RIM

Program Costs Paid by the Utility X X
Program Costs Paid by the Participant X
Program Administrative Costs X
Incentives Paid to Participants by the Utility X
Lost Revenues Due to Sales X
Increased Fuel Costs (Fuel Substitution Programs) X X

spread its fixed costs. Thus, potentially a DSM program that cuts overall kWh usage by
15% might result in an increase in rates; for the sake of argument assume that is 5%. The
TRC test would rate this result as positive, because overall electric costs of participants in
the program have gone down: .85 x 1.05 < 1.0. The RIM test would view the result as
unfavorable - the price of electricity went up by 5%; energy consumers who are not
participating in the program would see their costs rise.
The foregoing example is representative of how and why these two tests can disagree
about the desirability of a program. But they often agree qualitatively about a program
although its B/C ratio may differ as evaluated by one compared to the other. In general,
integrated resource plans that satisfy both tests are considered quite good. Those that meet
only one or the other are often rated as marginal and unacceptable.
A comparison of the costs considered under the TRC and the RIM cost-effectiveness
testing methodologies is shown in Table 6.7. As can be seen, the TRC and the RIM cost-
effectiveness test methods utilize some of the same cost information but, in essence, have
different cost contexts. As a result, the choice of evaluation technique is likely to produce
different B/C ratios (particularly since the utility's avoided costs serve as the benefits
portion in both methods). For example, a program that results in a large reduction in KWh
sales or requires significant financial incentives to achieve a target market penetration may
fail the RIM test while passing the TRC test. On the other hand, a program that requires a
major investment by the program participants may result in the opposite conditions. It may
not be accurate to attempt a complete generalization in this regard. Programs that focus less
on energy conservation, and more on peak demand reductions, may tend to favor the RIM
test. Conversely, programs that focus more on energy conservation, and less on peak
demand reduction, may tend to favor the TRC test.
The various methods of conducting cost-effectiveness testing should be considered as
different perspectives or various means of assessing any plan or integrated resource
program. Each such measure provides some alternative insight to the feasibility of a
program. Thus, the final selection of DSM programs is perhaps best made by including the
results from different cost-effectiveness measures in the decision-making process.
Other benefit and cost contexts for DSM
The TRC and RIM tests are the two most popular and most traditionally institutionalized
"approval tests" used in DSM planning, but there are many others. Note, for example, that
neither the TRC or RIM test includes all of the factors listed in Table 6.7. Beyond that,
there are other factors not listed in the table that can be included in a DSM evaluation. For
this reason there are simply too many slight variations on the basic B/C theme to go into in


detail here. However, the important point is that DSM is nearly always viewed as an
optional activity to be evaluated on the basis of its benefits and costs. The B/C approach, in
some form, is nearly always applied, typically with both benefits and costs evaluated using
PW or another time value of money discount method, because of the considerable
difference in the timing of costs and benefits in most DSM programs.
How B/C Evaluation and Prioritization Fit Within a Utility Processes
Benefit-cost evaluation and prioritization is usually part of a process that is "outside" the
normal T&D planning process, in the sense that the projects covered are considered
optional, not part of the mainstream requirements mandated by standards to meet the
obligation to serve. Most of these tend to be special initiatives or one-time programs such as
DSM, automation, or special maintenance/replacement projects - all undertakings that are
potentially worthwhile, but none absolutely necessary to meet the utility's obligation to
serve.
Figure 6.7 shows the salient aspects of how benefit/cost ratio evaluated projects are
typically adjudged and worked into the utility's system. An important point is that typically
the initiative or programs creating these candidates for funding are not part of the planning
process and its cycles. As shown below, projects evaluated in this manner "start" from a
different point than the "standards deviation" trigger point for traditional (standards
mandated) projects. They proceed through a different planning and evaluation process,
largely independent of the normal process. They may be one-time efforts (as are most
extraordinary maintenance efforts) or a continuing but independent planning process (as
were most utility DSM programs in the 1980s and 1990s).
Typically, after projects make it through their own evaluation and "approval" process,
they are part of a budget approval made en masse or on an item-by-item basis by executive
management. In a slim majority of cases, optional programs are treated as shown below, the
B/C process merging into the mainstream process after evaluation and
prioritization/selection, at the authorization stage. However, at some utilities the optional
budget is treated as a separate process entirely with its own authorization at the executive
level. The important points for planners to keep in mind are that:
• Typically, projects evaluated with B/C ratio methods are "optional" in the sense
that they are not being used to meet standards or for additions needed to meet
the basic obligation to serve. Their sole raison d' etre is to save money or
improve efficiency, etc.
• B/C analysis is inappropriate for regulated utility T&D planning because its
treatment of spending as "optional" is not consistent with the utility's obligation
to serve. This can be "fixed," but there are better approaches.
• Most often, these optional projects are initiated, planned, and evaluated outside
of the normal T&D planning process, or from within it as special, one-time and
very infrequent, perhaps unique, initiatives.
• Optional spending is typically expected to provide very good return on
investment. A 2:1 B/C threshold, when PW value and cost are considered, is
often considered the minimum acceptable.
• The evaluation and approval process for optional projects may merge with the
mainstream process after projects have made it through the B/C prioritization
and the first level of approval. They generally flow into the overall budget and
mix with "normal" projects at a point where all have been approved except for a
final, overall approval of the total budget.

o
i-N &O
so
co nj
>•> O C
:§ S .SP

-a g - s
C O 05
.2 13 S
=6-20
2 | a.
2 "2 o
^ _.
£1
8 § o.
o. c 9-

=§ * J
*-§
i
"M 05 ctf

Hi

& . l-H
C/3
,
N c/T
<.•«! ! ™ <-T ti O 1 / ^ O 1> i £ §i W N W
1O
*^* i y» *«. » _*—' j ' *f— • *__.',
o| 8 1 '~, % ' ^ 3 2*
2% « Q. m V3
_O
O

x S
<
"§1 iS |j | /
S '£" ~ 5 0
u. _O
c« C
Ofl A siSt •c
LU *Y^ 2« i w
?
1
o
o
°
N jfr
/ -o" § c
<fr

v «. « — »* i0
" ""^x^Z^ZZ-LL "5
^«,.»»«^ O T » m ^^^«= ^ ms »^ » i™4mr^ I M
^ .

|« s


6.5 INCREMENTAL BENEFIT/COST EVALUATION
Section 6.3 showed that the traditional electric utility investment planning approach has no
built-in control to explicitly limit total spending. All projects deemed necessary by the
utility's obligation to serve and standards are basically "pre-approved" by the system. Each
will have its cost minimized, but each will be done in some form. There is no explicit
control on the total budget, no built-in feature that will say "Enough! We just can't spend
this amount of money." Figure 6.8 shows the salient relationships in this traditional
paradigm. Basically, the engineering case (standards) drive the budget. The arbitrary feature
in this paradigm is the engineering standards, which dictate the budget that comes out of
this planning, evaluation, prioritization, and approval system
While this traditional approach may seem natural and even necessary to anyone who has
long worked in the electric utility industry, it does not fit how an increasing number of
business leaders want to run their utilities. Very simply, they consider any planning and
management system that does not have an explicit ability to control spending as
unacceptable. What they want is something more like Figure 6.9, a system in which the
business case will drive the process, or what might be called an asset management approach
(see Chapter 28, section 28.8). This, of course, will seem unacceptable to hard-line
traditionalists among utility planners: it means than an arbitrary, fixed budget drives the
"standards" that the system can achieve. Cut the budget too much and it lowers the utility's
standards too far.
As elsewhere in this chapter, the reasons or the merits of these various paradigms will
not be discussed or debated here. (Readers interested in the reasons should see Chapter 26,
section 26.2, "Balance among priorities - the business utility.") However, it is worth
identifying what needs motivated utility executives to shift from the traditional approach to
a business driven approach. Traditionally, a utility could depend on the financial framework
around the regulatory process to provide all the money required by the process illustrated in
Figure 6.8. Thus, the fact that the process did not have explicit controls on the total was not
a major concern: whatever the budget, the utility could handle it without major problems.

ENGINEERING irowth BUSINESS
CASE & Change CASE

Engineering
& Service
Quality
Standards

Figure 6.8 Traditionally, utility budgets were largely engineering driven. Rigorous (and inflexible)
standards and long-term present worth cost minimization defined budget needs, which drive the
utility's subsequent business planning. Not surprisingly, most engineers and planners are particularly
fond of the process and the distribution planning paradigm which grew up around it.

196 Chapter 6

BUSINESS Growth ENGINEERING
CASE .& Change CASE

'Business Service
Best Plan
Defined" Standards
that Meets
Budget Maximizing
Budget
Process

Figure 6.9 Many utilities have shifted to a business-driven structure, in which business needs define
the available budget and planning seeks to use this as effectively as possible. Traditional planning and
prioritization approaches have difficulty adjusting to this situation if budgets have been cut by
noticeable amounts. Although this paradigm treats some traditionally rigid standards as "elastic," it
obtains both better reliability and better economy from the limited budget.

But today, even though many electric utilities still operate within a regulatory
framework and still have an obligation to serve, they face very untraditional financial
constraints and concerns. Many know that the money to pay for a budget dictated by the
traditional standards-driven process may not "be there" when they need it. It is an
oversimplification to say that this is because of rate freezes or other limits on the revenues
utilities can obtain. While that is certainly true for many utilities, a wider and more
fundamental change has been at work. Investors and regulators began to treat power
delivery utilities more like "normal" businesses than they had in the past. The periodic rate
increases, the high debt-to-equity ratios, and the significant debt-to-revenue ratios that can
develop under the traditional paradigm all gave rise to concerns about utility spending
levels. Many utility executives, and investors in utilities, began to doubt that fundamental
assumption on the traditional paradigm about borrowing capital: "whatever it is, it can be
put into the rate base." Beyond this, the investment community began to react negatively to
any business that did not have the checks and balances in place to drive its budget where it
wanted it. For these reasons, in the 1990s many utilities began to scale back spending
significantly.
Addressing Spending Constraints With the Traditional Paradigm
The most common way of modifying the traditional standards-driven least-cost paradigm so
it produces less capital spending is to "lower" the planning guidelines (standards) that
trigger it. The revised guidelines mandate fewer additions to the system and that means less
spending. For example, suppose that in order to cut its spending, Big State Electric changes
its substation transformer utilization guideline from "peak demand cannot exceed 75% of
nameplate rating" to "peak demand cannot exceed 90% of nameplate rating." As a
consequence of this change, a number of things happen, all of which result in less spending.
1. There will be fewer instances where substations need added capacity to handle
load growth. A substation that might have been rated as "overloaded" under the
old guideline (projected peak load three years from now = 85 MVA, rating =
100 MVA) now is considered only adequately utilized.

2. Upgrades that are needed will be slightly smaller and less costly. If load at this
same substation eventually grows, it will reach a point (84 MVA) where it will
trigger the planning process for a project to add capacity at the substation. When
that happens, the utility will consider less costly options than under the old
guidelines. Only 1.11 (1/.90) rather than 1.33 MVA (1/.75) of capability need be
added for every projected MVA of peak demand growth over the (now higher)
rating limit.
3. One-time windfall capability gain. When the utility revises its guidelines, every
substation transformer in the system gains capability. Suppose that Big State
Electric had 9,000 MVA of installed transformer capability. Under its old
guidelines those transformers could carry a maximum of 6,750 MVA. Under the
revised ratings (90%), these units can carry 8,100 MVA. The utility instantly
gained 1,350 MVA in capability. Much of this released capacity will be in areas
of the system where there is no load growth, so it will be of doubtful value to
the utility. But some of it will be harnessable over the next decade to absorb
load growth.
This one-time windfall is really only a "bigger picture" view of (1), above.
But what it means is that the utility's spending on substation capacity additions
as its load grows takes a considerable drop for a number of years, during the
time it takes for this released capacity to be used. A similar impact occurs on all
other areas of system expansion costs where guidelines were revised.
As a result of all these changes, Big State Electric's overall spending drops considerably
in the first few years after the standards are revised, and then slowly rises to a sustainable
but lower level than before, as shown in Figure 6.10. For a few years, the utility can achieve

280

240
<
| 200

| 1 0
| 6

1 c 120
23
k— 5
5? 8°
3<
I 40

-10 -5 0 +5 +10
Year Prior to and After Re-Rating

Figure 6.10 A utility with 9000 MVA in load (year 0) and 2% annual load growth has to add
transformer capacity each year corresponding to the solid line in order to satisfy a 75% loading
guideline (240 MVA in year zero). Upping that guideline to 90% in year zero results in a sustainable
reduction in transformer requirements of 16.6% (dotted line) to only 198 MVA needed in year zero.
But the re-rating creates additional capability throughout the system, an additional 15% in all existing
transformers. Roughly half of this will be usable within the next decade, a little less being found
employable each year. Overall cost reduction is therefore something like that shown by the dashed
line, with a huge initial savings as the utility begins mining the capability created by the re-rating.

198 Chapters

can achieve a very significant reduction in spending. However, it cannot sustain this level of
spending using this paradigm (and may not be able to sustain it using any paradigm). While
the spending dictated by this paradigm will never reach the same level it would have
remained at under the old guidelines, it will rise greatly from the level of the first few years,
which is usually "too much." A few utilities attack this problem by "going back to the well
again" and revising their guidelines a second time, which merely delays the inevitable.
Chapters 23 and 28 will discuss this (using similar examples to that discussed here and in
Figure 6.10) in more detail.
There can be serious consequences for service reliability and system security that can
develop when using this approach. But it is much too much of a simplification to say that
revision of the standards causes reliability problems. In fact, this is usually not the case. The
standards and planning guidelines that were in use at most utilities up through the 1980s
were so robust that modest and in some cases substantial revisions could be made without
any major impact on service reliability.
Detailed discussion of the impact on reliability, why and how it occurs, and how serious
it might become is beyond the context of this discussion on evaluation and prioritization
mechanisms (although it will be covered in Chapter 28). But the heart of the matter with
respect to reliability is that the standards-driven, least-cost paradigm relies on robust,
almost overbuilt, standards to assure reliability. It has no mechanism within it to detect if
reliability problems are about to occur and no means to address them if they do. Thus, there
is no systematic way built into the system to assure reliability remains at acceptable levels.
Another adjustment that can be made within the traditional paradigm that can drive
some capital spending reductions is to change the time value of money used in the
paradigm's PW analysis. The discount rate can be raised to 20%, 30%, or even higher. Each
increase reduces the perceived value of future savings and biases the decision-making
process in favor of alternatives with lower initial costs. However, there are three problems
this approach creates:
1. This mechanism does not have an ability to directly control spending. By trial
and error a utility can find a discount rate that "works out" to the specific
spending limit it wishes to implement, but it is not a simple process to do so
and this approach does not have a simple rule so that it can be "aimed" at a
specific spending target.
2. This change may not make enough difference. The paradigm's goal and the
planning mechanism are still aimed at balancing long- and short-term costs.
Truly deep budget cuts need to be addressed by looking at how necessary the
various projects are today, not at how today's costs weigh against tomorrow's
savings. (This will be discussed at length in a few paragraphs.)
3. Using a high discount rate is inappropriate. Higher discount rates mean future
savings are less unimportant in any evaluation. That is not true: The utility's
financial values preclude it from affording long-term savings, but this is not
the same as saying it does not want them. (The importance of this distinction
will be discussed later in this chapter.)
One advantage of this adjustment is that it has no direct impact on service reliability. The
utility is still implementing its standards, and if unrevised or if revised only modestly, then
it is still building traditional or near-traditional levels of reliability into its system. However,
this often cuts capital but increases long-term operating costs. For example, it will spend
less for large wire and low loss equipment to reduce electrical losses, etc.


Incremental Benefit/Cost Analysis
A somewhat better idea than changing discount rates or revising standards within the
traditional paradigm might be to change to a benefit/cost ratio analysis for all projects,
mandatory or optional. For a variety of reasons, this doesn't work well. But a variation on
that idea does, and forms the core of what are called "business driven" planning and
prioritization schemes. This section will look at the reasons simple B/C ratio doesn't work
well and how the correction of those problems leads to a workable prioritization system
with a mechanism to drive spending to any specific level.
Suppose a utility wants to do the two distribution projects listed in Tables 6.3 and 6.4,
both the construction of a feeder in a growing area of its system in order to meet its
obligation to serve new customers and the refurbishment of a feeder in the older area of the
system in order to improve reliability and lower its long-term O&M costs. The traditional
paradigm would select the least-cost PW options for both projects, spending a total of $4.65
million in capital, that being the total sum of the least-cost PW figures for the two projects,
$3.6 and $1.05 million respectively.
But what if the utility does not have $4.65 million. Suppose that it has decided to cut
back on spending by 20%, to only $3.72 million. What should planners do? Planners at
traditional utilities would select to build only option 7 for project 1, at $3.6 million, since
that is "required" to meet the obligation to serve and has a cost that is also "justified by the
business case" based on the additional revenue the utility will receive from the new load.
Project 2 would get no funding, or perhaps would be given the left-over $120,000, so that
some small amount of work could be done.

Table 6.8 Examples: New Feeder Problem Alternatives and Refurbished Feeder, Both Evaluated on
the Basis of Benefit/Cost Ratio at 11% Discount Rate for Thirty Years - $ x 1000
Project Description Expected Yearl O&M&T Failure Total Total B/C
Option of Work SAIDI Cost per Year Cost PW "Benefit" Ratio
1 0 Do nothing 8760 $L,173,531 $1,173,531
1 1 4/0 large trunk 105 $2,250 $4,500 $13,934 $20,684 $1 ,155,097 513.4
1 2 4/0-4/0 branch 16 $2,350 $4,180 $2,010 $8,540 $1 ,167,341 496.7
1 3 336-4/0 branch 6 $2,775 $3,160 $670 $6,605 $1 ,169,701 421.5
1 4 336 large trunk 4 $2,850 $3,025 $402 $6,277 $1 ,170,104 410.6
1 5 636 large trunk 3 $3,115 $2,613 $268 $5,996 $1 ,170,650 375.8
1 6 795 large trunk 1.5 $3,550 $2,144 $67 $5,761 $1 ,171,320 329.9
1 7 636-336 branch 1.05 $3,600 $2,085 $7 $5,639 $1 ,171,392 325.4
1 8 795-336 branch 1 $3,800 $1,954 $0 $5,754 $1 ,171,577 308.3
1 9 795-636 branch 1 $3,875 $1,916 $0 $5,791 $1 ,171,615 302.4
1 10 1113 large trunk 1 $3,900 $1,896 $0 $5,796 $1 ,171,635 300.4
1 11 11 13-633 branch 1 $4,250 $1,651 $0 $5,901 $1 ,171,880 275.7

2 0 Do nothing 9.82 $1,480 $1,520 $3,000 -
2 1 Fix only serious 5.25 $450 $1,090 $732 $2,272 $1,178 2.62
2 2 Rebuild plan A 4.15 $600 $980 $543 $2,123 $1,477 2.46
2 3 Rebuild plan B 3.35 $775 $865 $405 $2,045 $1,730 2.23
2 4 Rebuild plan C 1.70 $1,050 $830 $120 $2,000 $2,050 1.95
2 5 Total rebuild 1.00 $2,250 $800 $0 $3,050 $2,200 0.98

200 Chapter 6

This approach is exactly what many utilities took in the 1990s, as they reduced spending
but continued to use a traditional project approval paradigm. The result was that what
capital there was went to new construction. Older areas of the system received less attention
than desirable, and a gap developed between reliability in older and newer areas of the
system. In some systems, a difference of more than two-to-one developed between the
interruption frequencies and durations seen in older and newer areas of the system. This
"non-equitable" service quality concerned many utilities, yet they couldn't get past the fact
that they had to spend on new customer connections, and that they had to cut back on
spending, too.
Table 6.8 gives B/C ratio analyses for the two feeder projects used earlier in Tables 6.3
and 6.4. It lists data from those tables for both projects, along with some new data and some
computations of the benefit/cost ratio. Here (fourth column from the left) the actual SAIDI
value for all options is given for both projects, a value calculated using a probabilistic
analysis method along with projected equipment failure rates based on current age and
condition and recent operating history (see Chapters 4,7, and ^23) for every alternative.
Also given (sixth column) is the failure cost corresponding for each option's SAIDI,
computed for the new feeder project (it was not included in Table 6.3) as well as the
refurbishment project (it was included in Table 6.4). This value is a "planning penalty" used
in reliability-based planning and evaluation to represent the value of the utility's obligation
to serve, or if one wishes to view it as this, the negative value to the customer of the
interruption (see Chapter 2). Use of this type of failure risk penalty function is one
technique that planners can use to make certain that any type of benefit/cost ratio
prioritization, which to some extent treats everything it considers as "optional," will always
respond to an obligation to serve requirement. Note that the failure cost for Project 1's "do
nothing" option is extremely high at $1.17 billion. That value represents the cost of that
option's 8,760 hour "interruption" of power for the new customers in that area. The cost of
"doing nothing" is totally unacceptable.
Also shown (three rightmost columns) are the total PW and total benefit of each project
(the $1,173,351 failure cost for "nothing" minus the sum of that option's failure and
O&M&T costs) and the B/C ratio (same as the capital cost shown in column 5).
But the B/C ratio is no real help in making a decision in this case. The B/C ratios of all
of the options for Project 1 are roughly 100 times that of any of the options for Project 2.
This is perhaps as it should be (project 1 is responding to an obligation to connect the new
loads), but this particular B/C ratio analysis is no help in deciding among these projects. But
the only procedural way to apply the information in Table 6.8 is to use it to maximize the
benefit one could buy, which would mean spending all of the available budget on project 1,
again buying option 8 (which the traditional paradigm also selected).
Table 6.9 shows the B/C ratios recomputed based on the ratio of incremental benefit to
incremental cost, what is called incremental benefit/cost ratio (iB/C). The incremental
benefit is an option's additional benefit over the next-least costly option to it. Its
incremental cost is its additional cost over buying each option instead of its next-least-
costly option instead. Sometimes this ratio is referred to as the marginal benefit/cost ratio
(mB/C) rather than the incremental benefit/cost ratio
An option's incremental B/C ratio measures the effectiveness of a decision to select that
particular option instead of a less expensive option. For example, for project 1, option 4 has
an iB/C of 5.4. Money spent on buying option 4, instead of option 3, yields a savings of
$5.4 for every dollar spent. By contrast, the decision to buy option 5, over option 4, has an
iBC of 2.0. All of Project 2's options are rated in this same step-by-step, incremental-
decision manner.


Table 6.9 Example Projects Evaluated on the Basis of Incremental Benefit/Cost Ratio
Project Description Yearl O&M&T Fail, Total Increm. Increm. iB/C
Option of Work Cost per Year Cost PW Benefit Cost Ratio
1 0 Do nothing $1173531 $1173531
1 1 4/0 large trunk $2,250 $4,500 $13,934 $20,684 $1.15 mil $2,250 513.4
1 2 4/0-4/0 branch $2,350 $4,180 $2,010 $8,540 $12,244 $100 122.4
1 3 336-4/0 branch $2,775 $3,160 $670 $6,605 $2,360 $425 5.6
1 4 336 large trunk $2,850 $3,025 $402 $6,277 $403 $75 5.4
1 5 636 large trunk $3,115 $2,613 $268 $5,996 $546 $265 2.1
1 6 795 large trunk $3,550 $2,144 $67 $5,761 $670 $435 1.5
1 7 636-336 branch $3,600 $2,085 $54 $5,739 $72 $50 1.4
1 8 795-336 branch $3,800 $1,954 $0 $5,754 $185 $200 .93
1 9 795-636 branch $3,875 $1,916 $0 $5,791 $38 $75 0.5
1 10 1113 large trunk $3,900 $1,896 $0 $5,796 $20 $25 0.8
1 11 11 13-633 branch $4,250 $1,651 $0 $5,901 $245 $350 0.7

2 1 Fix only serious $450 $1,090 $732 $2,272 $1,178 $450 2.6
2 2 Rebuild plan A $600 $980 $543 $2,123 $299 $150 2.0
2 3 Rebuild plan B $775 $865 $405 $2,045 $253 $175 1.4
2 4 Rebuild plan C $1,050 $830 $120 $2,000 $320 $275 1.2
2 5 Total rebuild $2,250 $800 $0 $3,050 $150 $1,200 0.1

Table 6.10 Example Project Options Sorted in Order of Incremental Benefit/Cost Ratio Value
Project Description Year 1 O&M&T Failure Total Increm. Increm. iB/C Cumul.
Option of Work Cost per Year Cost PW Benefit Cost Ratio Cost
1 1 4/0 large trunk $2,250 $4,500 $13,934 $20,684$ 1,1 55,097 $2,250 513 $2,250
1 2 4/0-4/0 branch $2,350 $4,180 $2,010 $8,540 $12,244 $100 122 $2,350
1 3 336-4/0 branch $2,775 $3,160 $670 $6,605 $2,360 $425 5.6 $2,775
1 4 336 large trunk $2,850 $3,025 $402 $6,277 $403 $75 5.4 $2,850
2 1 Fix only serious $450 $1,090 $732 $2,272 $1,178 $450 2.6 $3,300
1 5 636 large trunk $3,115 $2,613 $268 $5,996 $546 $265 2.1 $3,565
2 2 Rebuild plan A $600 $980 $543 $2,123 $299 $150 2.0 $3,715
1 6 795 large trunk $3,550 $2,144 $67 $5,761 $670 $425 1.6 $4,140
1 7 636-336 branch $3,600 $2,085 $54 $5,739 $72 $50 1.4 $4,190
2 3 Rebuild plan B 775 865 405 2045 $253 $175 1.4 $4,365
2 4 Rebuild plan C 1050 830 120 2000 $320 $275 1.2 $4,640
1 8 795-336 branch $3,800 $1,954 $0 $5,754 $185 $200 0.9 $4,840
1 10 1113 large trunk $3,900 $1,896 $0 $5,796 $20 $25 0.8 $4,865
1 11 11 13-633 branch $4,250 $1,651 $0 $5,901 $245 $350 0.7 $5,215
1 9 795-636 branch $3,875 $1,916 $0 $5,791 $38 $75 0.5 $5,290
2 5 Total rebuild 2250 800 0 3050 $150 $1,200 0.1 $6,490

202 Chapter 6

Optimizing Based on Incremental Benefit/Cost Ratio
Table 6.10 shows the same project options as in Table 6.9, with the same evaluation iB/C,
but with the project options mixed and sorted in descending by iB/C. A rightmost column
has been added to Table 6.10, showing cumulative cost. For the top row it is only the cost
of the top project option, for the second row it is the cost of that second option and the first,
for the third the cost of that third option and the two above it, etc. Listed in this way, iB/C
evaluation results are easy to use to optimize the overall customer reliability obtained from
any capital budget level. One merely moves down the table, "buying" each project option
until the budget is exhausted. To optimally apply the limited budget of only $3.720 million
for this example, the utility's planners need only move down the table until the next listing
cumulative cost column exceeds this amount, thus the optimal allocation of the budget is as
follows:
1. Decide to spend $2,250,000 on project 1 for option 1, money which buys
an average of $513 in benefit for each dollar spent, the most effective use of
money in the entire table.
Total committed spending at this point is $2,250,000. Total benefit
purchased at this point is $1,155,097.
2. Decide to spend an additional $100,000 on project 1 to "upgrade" from
option 1 to option 2, money which buys an average of $122 in benefit for
each dollar spent, the most effective use of money among project options as
yet unselected.
purchased at this point $1,167,341.
option 2 to option 3, money which buys an average of $5.6 in benefit for
yet unselected.
Total committed spending at this point $2,775,000. Total benefit
yet unselected.
5. Decide to spend $450,000 on project 2 for option 1, money which buys an
average of $2.6 in benefit for each dollar spent, the most effective use of
money among project options as yet unselected..
Total spending is now $3,300,000. Total benefit purchased so far is
$1,171,282.
yet unselected.
Total committed spending is now $3,565,000. Total benefit purchased
so far is $1,171,828.


yet unselected.
purchased so far is $1,172,127.

At this point, the next item on the list costs $670,000, an amount that would put the total
spending at $4,140,000, considerably above the mandated $3,720,000 limit. Thus, the
process of selection and spending ends. In essence, this process ultimately decided upon:

Project 1, Option 5, costing $3,115,000, providing benefit of $1,170,650,000
Project 2, Option 2, costing $600,000, providing benefit of $1,477,000
For totals of total cost of $3,715,000, total benefits of $1,172,127,000

for an aggregate benefit/cost ratio of 315.51.
In a very practical sense, this is an "optimal" solution to this planning problem. First,
this plan responds to the obligation to serve new load - it builds a feeder extension into the
load growth area that is sufficient to serve that demand. Equally important, it maximizes the
total benefit (reliability improvement and O&M cost reduction) gained from the total spent.
No other allocation of any amount of money less than or equal to $3,720,000 among the
available project options can buy so much benefit or achieve as high a net B/C ratio. The
benefit purchased with this plan, $1,172,127,000, is not only more than that gained if option
7, project 1 is selected alone, it is greater than the benefit that can be purchased by spending
any amount on Project 1 (see Table 6.8). Figure 6.11 shows the Pareto curve for this case, a
plot of the cumulative cost versus cumulative benefit as one moves down Table 6.10. Slope
of the curve between points is the iB/C ratio.

$1,180

$1,175

.0
i $1,170 -

$1,165

• $1,160

$1,155

$1,150
$2,000 $3,000 $4,000 $5,000 $6,000 $7,000
Capital Spending - $ x 1000

Figure 6.11 Pareto curve for the two-feeder budget-allocation problem in Tables 6.8 - 6.10, showing
"bang for the buck" options the utility planners had. Points on the curve indicate project options.

204 Chapter 6

An effective evaluation tool for large or small problems
This example, involving only two feeder projects, was simple to the point that a planner
who wished to balance reliability and cost could determine what to do merely by carefully
inspecting the various options for each project and considering what would be gained or
lost by sacrificing spending on one for spending on the other. However, in actual utility
planning situations there are dozens, perhaps hundreds, of projects mixed into
consideration. Any such heuristic approach to planning is very unlikely to both be efficient
and find an optimal, or even a particularly good, budget allocation.
Why incremental benefit/cost ratio evaluation works well
The easiest way to see why iB/C prioritization performs so well in this situation is to look at
why B/C evaluation did not. Looking back at Table 6.8, note that all of the options listed for
project 1 have very high B/C ratios. Any of them has a B/C ratio far superior to even the
best option for project 2.
But Table 6.9 tells a much different story. Project 1's option 1, which is basically a
decision to build the barest minimum possible to serve the new demand, has a tremendously
high payback of over 500:1. Option 2, which one can interpret from the data in Tables 6.3
and 6.9 as a decision to add to option 1 those items needed to "barely " meet requirements,
has a very high payback of 122:1. All other decisions to spend more on project 1 have
much, much lower benefit-to-cost ratios, ratios of the same order of magnitude as project
2's refurbishment options. Fundamentally, this is because options for project 1, beyond
option 2, are like project 2's options: they spend money to improve the reliability of a
feeder that already "barely meets" requirements.
The reason that all of the options for project 1 looked good in Table 6.8's B/C
evaluation is that every option included option 1's decision to "do something." The benefit
of that one decision "masked" all other spending on project 1, making every option look
good. Not just in this example, but in many real-world situations, one part of a project or
one aspect of a program will have so much benefit that it will mask inefficiencies in other
parts if the whole is evaluated from only a B/C standpoint. This "inefficiency masking" of
incremental spending occurs in many situations, not just those that use a reliability penalty
factor as in this example or in just those that involve new construction alternatives.
An absurd example will make it quite easy to see and understand why B/C evaluation
alone can lead to incorrect conclusions about the value of a project alternative. Suppose a
planner created yet another option for project 1, called option 2-G. "G" in this case stands
for "gold plated" and means that in addition to selecting option 2 at $2,350,000, an amount
equal to that, another $2,350,000, would be spent gold-plating various metal fixtures and
brackets on poles along this feeder. As a result, option 2G, at $4.7 million, would be the
most expensive option in Table 6.8. The gold plating would do nothing to improve the
reliability, lower the O&M costs, or extend the lifetime of the feeder. With twice the cost,
but the same benefit, option 2-G's benefit/cost ratio would be exactly half of option 2's. But
that would be 247.4. And therefore, if listed in a B/C table like Table 6.8, option 2-G
would appear to be a worthwhile alternative, much more valuable that any option for
project 2, even though it throws away over two million dollars. By contrast, if evaluated
and listed in an iB/C table, it would be clear that this option has no effectiveness at all,
compared to option 2's.
Incremental benefit/cost analysis, as applied here, separates project alternatives into
incremental decisions. This isolates "very high payback" decisions from other decisions and
permits a planner's evaluation to look at the efficacy of money spent after that one high-
payback decision has been made.


Ratios less than 1.0 mean ineffective use of money
Incremental benefit/cost ratio prioritization and selection's application shares one rule with
benefit/cost ratio evaluation: never fund any project with a ratio less than one. That means
the option's spending is buying less benefit than the cost. In the foregoing examples, a
budget limit meant that the budget limit was reached before the list of worthwhile project
options (those with iB/C ratios greater than one) was exhausted. However, in every case
where the spending limit is not reached, spending should continue only to where iB/C is
equal to or less than 1.0. Options below $1.0/$1.0 are ineffective - money should not be
spent regardless of whether it is available.
Unconstrained incremental benefit/cost evaluation will minimize PW cost
Note that in Table 6.10, if the budget were not constrained, iB/C would minimize total PW
cost. Without a spending limit, iB/C would proceed down the table until iB/C ratio is less
than or equal to 1.0. Thus, ultimately it would select option 7 for project 1 and option 4 for
project 2, minimizing long-term PW cost. These are the same options that the traditional
paradigm selected.
In cases where incremental benefit/cost evaluation is
set up to look at the same costs in the same way as
the traditional paradigm does, it will allocate money
in the same way if the budget is not constrained.
Thus, iB/C ratio evaluation and prioritization is not really new or different in what it wants
to do, if there is money to do it. It is just a way of doing what has always been done, but
with a method that can respond well to budget constraints.
iB/C evaluation can be viewed as "minimizing the pain" of budget cuts
There is another way to apply iB/C and a correspondingly slightly different perspective on
"what it does" that is often useful in explaining its application to people who are not
inexperienced planners. Instead of starting at the top of a table such as 6.10 and working
down until the spending limit is reached, one can start at the point on the list where iB/C is
less than or equal to 1.0 and work upwards, "de-selecting" options until the budget limit is
reached. In effect, this means starting with the budget amount and allocation from the
traditional paradigm and removed elements of that plan that are the least effective use of
money until the budget has been reduced enough to meet the spending limit.
Incremental benefit/cost evaluation can be viewed as
a way to "minimize the impact" of budget cuts.
To apply this method to Table 6.10, one would start with project 2, option 4, since it is
the lowest item on the list with an iB/C > 1.0. Everything above and including this line is
something that the traditional (unconstrained) paradigm would have selected to fund. But if
the total spent, $4.64 million, is too much, then one can begin working up the table, "de-
selecting" or removing items.
Thus, if the budget target is the same $3.72 million used earlier, option 4 is "de-
selected" in favor of option 3 for that project, reducing spending to 4.37 million. Since that
is too much, option 3 is de-selected in favor of option 2 for project 2. Spending is still too
much, so option 7 for project 1 and then option 6 for project 1 are both de-selected in turn
until the resulting total spending amount, $3.715 million, fits the budget limit.
It really makes no difference in cases like this, where benefit and cost are both measured
in dollars, which approach one takes: starting at the top and working down, or starting
where iB/C equals one and working up. The exact same plan is ultimately reached. But
almost all iB/C methods take the first approach and work down the table from the top. The

206 Chapter 6

reason is that a "down the table" method is more flexible. That approach must be used in
situations where benefit is not measured in dollars, for example when "benefit" is measured
as the number of SAIDI minutes reduced or the increase in customer satisfaction rating, or
some other non-monetary feature. Those types of planning problems are situations were
outside of the context of the traditional paradigm, but can be optimized by the right type of
iB/C analysis. They will be discussed later in this section.
Incremental Benefit/Cost Evaluation: A Multi-Attribute Process
The iB/C approach illustrated in the foregoing example implements a multi-attribute
planning evaluation (see section 6.2). The two attributes are benefit, here a combination of
reliability penalty reduction, O&M cost reduction, and minimization of long-term
expansion costs, and cost, which is just initial capital cost. As attributes, both are values that
planners want to optimize (the former maximized, the latter minimized). But one can be had
only at the expense of the other. Planners can balance one against the other but have only
the choices shown by the Pareto curve in Figure 6.11.
In that example, the goal was to get the most from a predetermined budget, $3,720
million. With the figure for one attribute fixed, a single solution now existed - the plan
selected above. This two-attribute evaluation, prioritization, and selection method would
have worked just as well if the utility's goal had been to minimize the cost of achieving a
certain amount of benefit, or reliability, as will be demonstrated later in this section.
Incremental benefit/cost evaluation can "target" either attribute
Incremental benefit/cost analysis could have been applied to the previous example in an
application where reliability, rather than the budget, was the "fixed" or targeted attribute. In
such an example, the process would identify the minimum budget required to achieve a
particular level of reliability. From the standpoint of the mechanics used to apply the
incremental benefit/cost ratio evaluation and selection methodology, the situation would be
exactly the same. The ranking and selection process would be identical. Each project option
would evaluated for the relevant factors - cost and reliability in this case. Then the project
options would be sorted by decreasing incremental benefit to cost ratio. Several of
reliability-targeted examples will be given later in this section.
Out-of-series sorting of alternatives
Table 6.10 reveals a complication that develops in some iB/C methodology applications.
Project 1's options 10 and 11 are both listed before its option 9. The situation arose in this
case because the money spent in upgrading from option 9 to 10 is more efficient in terms of
buying reliability than that spent upgrading option 8 to 9. This is invalid from the standpoint
of any application. The actual rating of 10 ahead of option 9 is not valid in the sense that
one could not select 10 like this: Option 10's incremental benefits and costs assumed that a
decision to spend whatever option 9 costs had already been made.
This particular instance creates no problem from a practical standpoint because this out-
of-series listing both occurs farther down the table, below the spending limit, and below
where iB/C drops below 1.0. It is a moot point. But there can be cases where a situation like
that occurs in a portion of the table that will be used for approval.
In such cases, the less effective of the two options (the one farther down the table, with a
lower iB/C) is removed. Option 9 in this case. The incremental costs of the remaining
option (10 in this case) must be re-computed (from option 8) and its position in the table
recalculated. For example, the incremental benefit and incremental cost for option 10 are
now 58 and 100, respectively, for an iB/C of .58. This moves it, putting it below option 11,


creating a second out-of-series listing. Removing option 10 in a similar way leaves option
11 as the sole remaining jump from option 8, with a benefit of 303 and a cost of 450, for a
benefit/cost ratio of .67.
Dealing with "left-over" discrete increments
The two feeder projects used in the foregoing examples utilize actual project descriptions
from a utility budgeting case. However, there were only two projects involved in the
example, and as a result any one option was a sizeable part of the entire example's budget.
For this reason, any budget limit was likely to land in the middle of a big "all or nothing"
increment. In Table 6.10's example, the average increment (project option size) is just
slightly less than $406,000, more than 6% of the total. For example, if the budget limit had
$3,500,000, the optimization process would have bought down through project 2, option 1.
At that point, it would have lacked $65,000 of being able to buy the next item on the list. It
would have to stop approval of projects and leave $200,000, fully 5.7% of the budget,
unallocated.
In some cases when this happens, planners can spend the left-over increment on buying
part of the next option on the list. For example, if the next option is a repair or maintenance
program, usually that can be partially funded since it is composed of many small work
orders. But a partial-funding approach does not work in most cases and would not work in
this example. The next option is actually a change in feeder layout from a previous option -
a decision in which it is one or the other and for which complete funding would no doubt
have to be made.
Planners can also sometimes skip the next option on the list if it is too big to buy and
move further down to find the next affordable item to buy. Again, that often works, but not
in this situation. In fact, here nothing really works well. The left-over $200,000 is just not
applicable among the options listed.
This situation is seldom serious in utility budget applications, where there are dozens,
often hundreds, of projects in the list. The resulting "granularity" of the sorted project list
tables and Pareto curves is very fine: what is left over is likely to be only a very small
fraction of the total.
Budget-Constrained Planning Using iB/C
One application of iB/C is for budget-constrained planning, situations like the example used
earlier in which a predetermined spending limit has been put in place, usually because that
limit was determined by a financial planning method external to the T&D planning process.
Regardless of the how or why of the spending limit, the traditional standards-driven, least-
cost paradigm has no mechanism to respond well to such a limit, as was discussed in
section 6.2 By contrast, incremental benefit/cost evaluation and ranking will allocate a pre-
determined budget among candidate projects in the most effective manner possible. But it
must use a second attribute in its evaluation, an evaluated factor that, when compared to
cost, is used to define "effectiveness" of the money being spent. Budget-constrained
planning refers to situations where the second attribute is also measured in money. The
most common example of budget-constrained planning is when "benefit" is long-term
savings and "cost" is initial (capital) cost. Basically, planners want to both maximize long-
term cost and minimize how much they spend now. As shown earlier, iB/C applied in this
situation will:
• Make the same choices the traditional paradigm would, if budget is not constrained
• Minimize the negative impact of budget cuts
• Maximize the benefit gained from what budget is available

208 Chapter 6

The second and third items listed above are two interpretations of the same characteristic -
optimization of budget effectiveness. For this reason, iB/C is a useful choice for a utility
that faces budget cuts, for whatever reason, regardless of whether its values have changed
or not (see below)
Optimizing reliability gained per dollar spent using IB/C
In the late 20th and early 21st century, many electric utilities began to place increased
emphasis on customer service quality, specifically the reliability of service seen by their
customers. Reliability of service had traditionally been a priority of electric utilities, but an
increasing amount of explicit attention was paid to it beginning in the late 1990s, along with
management toward reliability targets.
There were many reasons for the increased emphasis, among them a recognition on the
part of utility executives and planners that budget cuts might impact reliability of service
unless measures were put in place to focus on maintaining reliability at satisfactory levels.
But the modern emphasis on reliability is more than a reaction to and concern about budget
reductions and their potential impact. It is the result of gradually changing importance of
reliability of service to meet the needs of modern energy consumers. Part of this is certainly
the modern dependence on digital equipment for entertainment, convenience, and business
applications, but another part is simply a growing reliance on electric power in a "just in
time" world that permits little tolerance for delays and inefficiency caused by service
interruptions.
Incremental benefit/cost ratio evaluation and prioritization can be applied to optimize
"bang for the buck" in cases where the "buck" is reliability - SAIDI, SAM, or a
combination of both or some other reliability index. Table 6.11 shows the two feeder
examples from earlier in this chapter evaluated on the basis of "SAIDI per capital dollar"
of improvement. Evaluated in this way, one option in each project stacked out of series and
was removed. Coincidentally, it happened to be option 3 in both projects. Both were
removed from the list. Options 9, 10, and 11 for project 1 contribute no reliability
improvement and were also eliminated from the list (they would have sorted to the bottom
of the list, with an iB/C ratio of zero). Figure 6.12 shows the resulting Pareto curve.

Table 6.11 Feeder Project Options Evaluated on the Basis of SAIDI Improvement
Project Description Incremental Incremental iB/C SAIDI Plan's Cost
Option of Work Gain - Cust. Hrs. Capital $x 1000 Ratio (Hours) So Far $xlOOO
1 1 4/0 large trunk 12,117,000 $2,250 5,385 51.5 $2,250
1 2 4/0-4/0 branch 124,600 $100 1,246 12.5 $2,350
1 4 336 large trunk 16,800 $500 34 7.3 $2,850
2 1 Fix only serious 8,190 $450 18 4.7 $3,300
2 2 Rebuild plan A 1,980 $150 13 4.1 $3,450
2 4 Rebuild plan C 4,416 $425 10 2.7 $3,875
1 5 636 large trunk 1,400 $265 5 2.3 $4,140
1 6 795 large trunk 2,100 $425 5 1.6 $4,565
1 7 636-336 branch 140 $50 3 1.6 $4,615
1 8 795-336 branch 560 $200 3 1.4 $4,815
2 5 Total rebuild 1,254 $1,200 1 1.0 $6,015


15

$0 $1,000 $2,000 $3,000 $4,000 $5,000 $6,000 $7,000
Capital Cost-Sx 1000

Figure 6.12 Pareto curve for SAIDI versus capital spending on the two-feeder areas from previous
examples.

Using a monetary penalty function instead of reliability index values
Table 6.11's example is the simplest possible reliability-basede optimization, focused on
only one aspect of reliability (SAIDI - duration). Usually an electric utility desires to
minimize both SAIFI and SAIDI, although usually one or the other is given somewhat
higher priority than the other, depending on the utility's customer needs and its regulatory
agency's preferences. Although SAIDI and SAIFI are related, they are somewhat
independent (driving down SAIFI generally reduces SAIDI, but actions aimed at SAIDI do
not generally drive down SAIFI). Thus there are really three attributes: cost, SAIDI, and
SAIFI. While it is possible to set up a "three-attribute" framework for decision-making,
such methods are much more intricate and complicated than two-attribute iB/C, and it often
proves very difficult to drive them to a clear-cut solution.
Instead, the reliability attributes are combined into one attribute by translating both into
dollars using reliability penalty functions. Hence, a "failure penalty" function of the type
used earlier (Tables 6.4, 6.8 and 6.9), which converts customer interruption minutes into
dollars. The nature and rationale behind such functions will be covered later (see Chapters
23 and 28). The costs computed are not costs actually paid by the utility. Their calculation
and use is merely a mechanism to force iB/C to optimize reliability in a situation where
there is more than one "savings" or "benefit" being bought by the spending that is being
optimized. From the standpoint of this discussion, the important point is that the functions
convert the SAIDI and SAIFI value for each project option to a dollar "equivalent."
The basic SAIDI formula used in this example is typical of most reliability penalty
functions, a linear function of all factors including demand customers and peak demand

Penalty $ = $2 x average peak coincident demand per customer (6.1)
x customer count x expected SAIDI x 9.57

The scalar 9.57 converts a constant annual cost over the next thirty years to its PW value
at .90 PWF. Peak coincident demand is 5.0 kW per customer. Feeder 1 has 1,400 consumer

210 Chapter 6

Table 6.12 Feeder Project Options Evaluated on the Basis of SAIDI Improvement $ x 1000
Project Description SAIDI Capital "SAIDI Incremental iB/C
Option of Work Hrs. Cost Penalty" Benefit Cost Ratio
1 1 4/0 large trunk 105 $2,250 $14,068 $1,159,597 $2,250 5,154
1 2 4/0-4/0 branch 16 $2,350 $2,144 $11,924 $100 119.2
1 4 336 large trunk 4 $2,850 $536 $1,608 $500 3.21
2 1 Fix only serious 5.3 $450 $904 $784 $450 1.74
2 2 Rebuild plan A 4.2 $600 $715 $189 $150 1.26
2 4 Rebuild plan C 1.7 $1,050 $292 $423 $450 0.94
1 5 636 large trunk 3 $3,115 $402 $134 $265 0.51
1 6 795 large trunk 1.5 $3,550 $201 $201 $35 0.46
1 7 636-336 branch 1.4 $3,600 $188 $13 $50 0.27
1 8 795-336 branch 1 $3,800 $134 $54 $200 0.27
2 5 Total rebuild 1.0 $2,250 $172 $120 $1,200 0.10

delivery points while feeder 2 serves 1,800. The $2/kW figure is rather high but not
extreme. Most functions use a value between .50 and $5 per kW. The "expected SAIDI"
value used is produced by a reliability-based analysis using one of the probabilistic
techniques covered in Chapter 23. It is the average SAIDI expected over the long-run,
allowing for weather variations, bad and good luck, and all factors.4
Table 6.12 shows these SAIDI penalties computed for the example feeders and used to
optimize the table. The ordering of options is identical to that shown in Table 6.11. As
before, the same options proved out-of-series and were removed.

Use of a penalty function like this
produces the same prioritizjation results
as using the actual reliability values.

Simultaneous Focus on Two or More Reliability Indices
Usually T&D planners wish to work down both SAIDI and SAIFI, although as stated above
the relative priority of one to the other will vary depending on circumstances. Regardless,
some attention to both is warranted. Table 6.13 shows the example feeders with penalty
functions applied to both SAIDI and SAIFI, each converted to dollars. The two are then
added into one reliability penalty, reductions of which are the incremental benefit computed
for each option. Here, a more useful penalty function has been applied to both SAIFI and
SAIDI. The SAIDI function used is similar to that applied above, but has one hour
subtracted from the SAIDI value and a threshold at zero.

4
In these examples, no failure rate escalation (worsening reliability due to aging) is used. SAIDI is
expected to be a stationary value throughout the next 30 years. This is a reasonable assumption for
feeder 1. It will be newly installed - thus even thirty years from now it will not be "old" in the sense
of equipment being beyond their design lifetimes. However, this is not an entirely valid assumptions
for feeder 2, whose projects are being done because elements of it are already quite old. Escalation
will be dealt with later in this section.

The SAIDI function used is the same as equation 6.1, with one exception
x customer count x (SAIDI - 1) x 9.57
whenever SAIDI > 1.0
= $0 whenever SAIDI ^ 1.0
Similarly, the SAIFI function takes the same approach. It is linear with respect to all factors,
but includes a subtracted value and is set to zero below a certain threshold.
x customer count x (SAIFI - 1) x 9.57
whenever SAIFI >1.0
= $0 whenever SAIFI <L 1.0
Here, 1.0 is subtracted from the various options' SAIDI and SAIFI values. All calculated
function values less than zero are set to zero. This is basically a "trick" used to fine-tune the
optimization to provide the results the planner wants - consistency of results across the
whole customer base.
An electric utility desires more than just a low system-wide SAIDI. It wants to achieve
some degree of consistency, roughly the same quality of service everywhere. In actual
utility budgeting applications, where there are dozens, perhaps hundreds of feeders
involved, there will usually be several that are real "bargains" in the sense that buying
SAIDI improvement there is very inexpensive relative to other feeders.
Unrestrained, an iB/C optimization will spend to drive down SAIDI where it can find
bargains - after all that is the whole purpose of optimizing on bang for the buck. It will
spend at those locations to drive SAIDI down to very low levels, far below the utility's
target level. As a result, the utility could end up with a SAIDI average that was acceptable,
but areas that were much lower and much higher than its average.
This particular "trick" reduces this tendency by giving no credit to improvements that
push SAIDI below the 1.0 hour nominal threshold level. The iB/C process will move on to

Table 6.13 Feeder Project Options Evaluated on the Basis of both SAIDI and SAIFI $ x 1000
Project Description SAIDI SAIFI Capital "SAIDI "SAIFI Incremental iB/C
Option of Work Hrs. Events Cost Penalty" Penalty" Benefit Cost Ratio
1 1 4/0 large trunk 105 31.00 $2,250 $13,934 $6,029 $1,153,568 $2,250 512.7
1 2 4/0-4/0 branch 16 8.90 $2,350 $2,010 $1,588 $16,366 $100 163.7
1 4 336 large trunk 4 3.20 $2,850 $402 $442 $2,753 $500 5.51
2 2 Rebuild plan A 4.2 2.30 $600 $543 $336 $2,188 $600 3.65
2 4 Rebuild plan C 1.7 0.94 $1,050 $120 $0 $759 $450 1.69
1 5 636 large trunk 3 2.00 $3,115 $268 $201 $375 $265 1.42
1 6 795 large trunk 1.5 1.10 $3,550 $67 $20 $382 $435 0.88
1 7 636-336 branch 1.4 0.78 $3,600 $54 $0 $33 $50 0.67
1 8 795-336 branch 1 0.56 $3,800 $0 $0 $54 $200 0.27
2 5 Total rebuild 1.0 0.56 $2,250 $0 $0 $120 $1,200 0.10
1 9 795-636 branch 1 0.56 $3,875 $0 $0 $0 $75 0.00
1 10 11 13 large trunk 1 0.56 $3,900 $0 $0 $0 $25 0.00
1 11 11 13-633 branch 1 0.56 $4,250 $0 $0 $0 $350 0.00

212 Chapter 6

other feeders, even if, in actuality, money buys slightly less reliability there: it gives more
of what the planners really want.
The relative values of the penalty value for SAIDI and SAIFI ($2 and $3 respectively)
were assigned in order to provide the desired bias in spending on one or the other.
Determination of these penalty values, both relative to one another and in absolute terms, is
covered in Chapters 23 and 28. The fact that the SAIFI penalty value, $3, is greater than
that used for SAIDI, $2, does not mean that SAIFI is more important here than SAIDI (it
isn't). The reader must remember that these penalty values are multiplied times actual
SAIDI and SAIFT values that have different scales. SAIFI is usually measured on a slightly
lower numerical scale than SAIDI, and thus a larger penalty scalar is required to make its
computed penalties equivalent to those of SAIDI so the two are relatively equal in
importance to the optimization.

Adding O&M Costs into the Reliability Optimization Analysis
Since the reliability penalty functions put SAIDI and SAM into dollar measures, the
benefit of improving them can be mixed in with other monetary benefits, such as reductions
in O&M costs due to better design and equipment. Table 6.14 shows PW O&M cost (from
Tables 6.3 and 6.4) added into the benefits computed for all options in Table 6.13. Capital
spending is now judged against its impact on achieving a combined goal of reducing
SAIDI, reducing SAIFI, and reducing long-term O&M costs. To optimize any give budget,
planners start at the top and work down the list, "buying" options until they reach their
budget limit. Whatever amount they have will be optimized in terms of maximizing the
combined SAJDI-SATFI-O&M reduction it provides.
Note that O&M costs are treated here as a benefit - with reductions in O&M added into
the "benefit." One can try to set up this evaluation so that they are treated instead as a cost,
combined with capital costs into a total PW cost which is used as the "cost" in the iB/C
evaluation. Although iB/C can be applied in this manner, it proves messy and unsatisfactory
to set up (the reason being that part of the "cost" is now an item (long term costs) which one
really wants to drive down with present spending). But it can be made to work, after a
fashion.

Table 6.14 Feeder Project Options Evaluated and Optimized in Order of Capital Spending to Reduce
SAIDI, SAIFI, and Long-Term Operations and Maintenance Costs $ x 1000
Project Description SAIDI SAIFI Capital "SAIDI "SAIFI O&M&T Incremental iB/C
Option of Work Hrs. Events Cost Penalty" Penalty" Cost Benefit Cost Ratio
1 1 4/0 large trunk 105 31.00 $2,250 $13,934 $6,029 $4,500 $1,149,068 $2,250 510.7
1 2 4/0-4/0 branch 16 8.90 $2,350 $2,010 $1,588 $4,180 $16,686 $100 166.9
1 4 336 large trunk 4 3.20 $2,850 $402 $442 $3,025 $3,908 $500 7.8
2 2 Rebuild plan A 4.2 2.30 $600 $543 $336 $980 $2,688 $600 4.5
1 5 636 large trunk 3 2.00 $3,115 $268 $201 $2,613 $787 $265 3.0
2 3 Rebuild plan B 3.4 1.86 775 $413 $227 $865 $353 $175 2.0
2 4 Rebuild plan C 1.7 0.94 $1,050 $120 $830 $556 $275 2.0
1 6 795 large trunk 1.5 1.10 $3,550 $67 $20 $2,144 $851 $435 2.0
1 7 636-336 branch 1.4 0.78 $3,600 $54 $2,085 $92 $50 1.8
1 8 795-336 branch 1 0.56 $3,800 $0 $1,954 $185 $200 0.9
1 11 11 13-633 branch 1 0.56 $4,250 $0 $1,651 $303 $450 0.7
2 5 Total rebuild 1.0 0.56 $2,250 $0 $800 $150 $1,200 0.1


But the real reason to keep O&M as a benefit, not a cost, is that that reduced O&M cost
is a benefit of capital spending and should be treated so. The only reason to spend capital is
to improve service quality or reduce future costs. An evaluation and prioritization like that
in Table 6.14 provides for that.
Adding DSM Or Other Optional Programs Into The Evaluation
The final iB/C evaluation and prioritization example given here throws DSM options into
the set of considerations and demonstrates the flexibility which can be applied to the
spending prioritization's span of considerations. Here, the DSM examples from Table 6.6
have been used, combined with the SAIDI-SAIFI-O&M example from Table 6.14. Those
DSM examples from Table 6.6 had costs and benefits representing programs applied to a
fairly large region and 70,000 customers. In Table 6.15, they have been scaled to
application over the 3,200 customers in the two feeder areas covered by this and previous
examples, simply by multiplying all numbers in Table 6.6 by 3,200/70,000.5
Note that each DSM option is treated as if its costs and benefits were incremental
values. In essence, they are: DSM options here do not build on one another - none are
optional decisions building on top of an earlier, interim step (although, in some cases, DSM
can be set up and studied in that manner). Here, the decision to take a particular DSM
program is an incremental decision to spend that program's cost to get that program's gain.
The resulting list of feeder and DSM options sorted by iB/C is shown in Table 6.16.
Several DSM programs made their way to the top half of the list and would be among
programs funded in all situations except where budget was severely restricted. Note that
project 1, option 7 and project 2, option 4 were the minimum PW cases for their respective
example feeders. Thus, buying down the list through project 1, option 7, for a total of
$4,873,000, achieves what one can call an "unconstrained" case. At that point both feeders
have achieved the same selected option as the traditional paradigm would have selected.
Interestingly, only two DSM options, those with B/C ratios near 2.0, were selected prior
to the final feeder options needed to achieve this "unconstrained" level. This somewhat
verifies the rule of thumb given earlier that DSM and other "optional" programs need to
have a B/C ratio far above 1.0 to justify their selection. While not general, this is a common
result: DSM and other truly optional activities typically need a B/C ratio of around two to
one to qualify for selection when compared to other utility spending even if capital is not
heavily restricted.
Summary and Important Points About Incremental Benefit/Cost Analysis
Incremental benefit/cost evaluation and prioritization is a relatively straightforward method
to apply. No exotic computer programs or systems are needed. With proper preparation and
procedure, electronic spreadsheets and good record-keeping are all that are necessary to
apply it well.
The incremental benefit/cost evaluation and prioritization method applies in two
situations that are mechanically very similar. In both, spending is balanced against
reliability. But the motivation and focus is quite different.
Budget-constraints in place. Here, budget levels have been cut to levels below the
total that the traditional paradigm will spend. As mentioned in section 6.3, that

5
The feeder and DSM options used here are all taken from actual utility planning cases, but the feeder
and DSM cases come from different utilities using slightly different accounting and cost estimation
methods. The author has attempted to adjust all to the same basis. While only approximate, the
results are representative of how such multi-dimensional planning "optimizations" typically turn out.

214 Chapter 6

Table 6.15 DSM Option Benefits and Costs from Section 6.3 Scaled to the
Two-Feeder Example Area $ x 1000
DSM Program Class Cost Benefit B/C
Insulation blankets on water heaters resid. $136 $290 2.13
Direct water heater control resid. $87 $166 1.91
Direct water heater control comm. $30 $53 1.78
Direct water heater control ind. $46 $67 1.44
Building shell upgrades comm. $288 $386 1.34
Efficient interior lighting upgrade resid. $364 $456 1.25
High-efficiency motor upgrade indus. $57 $70 1.21
Efficient interior lighting upgrade comm. $114 $138 1.21
Insulation blankets on water heaters comm. $87 $100 1.14
Efficient exterior lighting upgrade comm. $180 $191 1.06

Table 6.16 Feeder Project Options Evaluated and Optimized in Order of Capital Spending for S AIDI,
SAIFI, O&M Reductions, and DSM Payback $ x 1000
Project Description Incremental iB/C Cumulative Cumulative
Option of Work Benefit Cost Ratio Benefit Cost
1 1 4/0 large trunk $1,149,068 $2,250 511 $1,149,068 $2,250
1 2 4/0-4/0 branch $16,686 $100 167 $1,165,753 $2,350
1 4 336 large trunk $3,908 $500 7.8 $1,169,662 $2,850
2 2 Rebuild plan A $2,688 $600 4.5 $1,172,349 $3,450
1 5 636 large trunk $787 $265 3.0 $1,173,137 $3,715
DSM Insulation blankets on WH $290 $136 2.13 $1,173,427 $3,851
2 3 Rebuild plan B $353 $175 2.0 $1,173,779 $4,026
2 4 Rebuild plan C $556 $275 2.0 $1,174,335 $4,301
1 6 795 large trunk $851 $435 2.0 $1,175,186 $4,736
DSM Direct water heater control $166 $87 1.91 $1,175,352 $4,823
1 7 636-336 branch $92 $50 1.8 $1,175,444 $4,873
DSM Building shell upgrades $386 $288 1.34 $1,175,951 $5,237
DSM Efficient interior lighting $456 $364 1.25 $1,176,407 $5,601
DSM High-efficiency motor $70 $57 1.21 $1,176,476 $5,659
DSM Efficient interior lighting $138 $114 1.21 $1,176,614 $5,772
DSM Insulation blankets on WH $100 $87 1.14 $1,176,713 $5,860
DSM Efficient exterior lighting $191 $180 1.06 $1,176,904 $6,040
1 8 795-336 branch $185 $200 0.9 $1,177,089 $6,240
1 11 11 13-633 branch $303 $450 0.7 $1,177,392 $6,690
2 5 Total rebuild $150 $1,200 0.1 $1,177,542 $7,890


standards-based least-cost project-oriented paradigm does not have a mechanism
to control that spending total. By contrast, incremental benefit-cost evaluation and
prioritization can be set up and operated so it does have such a mechanism.
When applied for this purpose, iB/C is part of a prioritization and selection
process that has values little changed from those of the traditional paradigm.
"Savings" in the evaluation phase is set up to include long-term cost reductions
(due to both decreased O&M and decreased future capital spending) from a base
do-nothing or do-as-little-as-possible case. If savings includes only long-term cost
reductions, iB/C will select precisely the same projects as the traditional approach
would if the budget it is given is unconstrained. But unlike the traditional
approach, an iB/C will respond well to spending constraints, trading capital
spending reductions for long-term savings in the most effective way possible to
meet the budget constraint, and assuring utility management that it will achieve the
mandated budget level.
In this respect, then, iB/C can be viewed as little more than a "constraint
tolerant" form of the traditional paradigm. The utility is not really trying to do
anything substantially different than in the past, just do it with less spending than
in the past. It is true that service quality (reliability), in the form of a service
reliability cost penalty, is typically included in the benefits evaluation for budget-
constrained applications. But that is used in this context as a tool to control
unforeseen drops in reliability due to the reduced spending while still addressing
values and goals that are mostly traditional.
Specific focus on reliability improvement. By contrast, a utility that wants to focus
on reliability as a deliverable and targetable key performance metric must change
from traditional planning and prioritization methods simply because that paradigm
could not explicitly address reliability of service as a "design" goal. Incremental
benefit/cost ratio evaluation will optimize "bang for the buck" when set up and
used as outlined earlier in this section, so it is a natural selection when a utility
needs to improve reliability or make choices about how to spend money,
particularly in tight budget situations as for example when it faces rising failure
rates due to old equipment and infrastructure. In such cases, the "bang" is
reliability or a combination of reliability and other improved factors important to
planners. Savings, due to reductions in long-term costs, may be included as one of
those factors. Thus, the actual evaluation of benefits may be similar to that used in
budget-constrained planning. But the motivation and the focus will be different,
and usually this is reflected in the details of implementation. 6
Figure 6.13 depicts the overall process in which iB/C is used for the basis of selection
and approval of projects. This process is not that different from the traditional process on
the front end. There is an annual planning cycle in which system data and information about
new construction is used to produce a load forecast. That is used as the basis for a "scan" of
the entire system using analytical tools to detect deviations.
The first difference occurs during this scan of the system. Whereas the traditional
paradigm invariably used only a load-flow methodology to detect deviations from standards

6
'Tricks" for tailoring reliability optimization goals such as subtracting a basement target level (the
one hour in equations 6.2 and 6.3) and the use of both SAIDI and SAIFI in reliability evaluation are
generally applied only when reliability is the real motive. Budget-constrained planning usually will
work well enough for budget reduction purposes if a simple penalty for increases in SAIDI is used.

216 Chapter 6

(0
o
Q. (A O>
a §-£
C *•• (A
re o o
oc.9J.-o
CO

"co

x —
•o

D o
c
re
(0
o
GO re re Q.
c O
o re
s CJ
QQ

O)
c O
'E ^ /
c
CO
IE Ak

k
V 1
I

o /
w N
0 V) •o
o re
Q
O "re
o
O) 0) re
Q
c 1 'o O o
'E Q.
& •o n
c CD
<u o re
w
CO *-• 3
re re
Q_ •o 1
Q. H—
re
"co o
pm
^
1 c


(voltage, loading, etc.), this process looks at reliability and electric performance and
assesses them only against service quality targets. Generally, the "scan" of the system
involves a load flow, as in the traditional paradigm, plus some sort of reliability assessment.
A probabilistic analysis method (see Chapter 23) is best here, but table-driven reliability
evaluation based on categories of situation works well in cases where data or computing
resources are not available for the more comprehensive approach, but despite this, there is
not much substantial difference in the annual planning cycle: a forecast is done, the updated
model of the system is analyzed, and circuits and facilities with deviations are transferred
on to a process of determining how to "solve" the problems identified.
The planning study phase follows a project-oriented approach like the traditional
paradigm: each circuit or facility with a deviation is analyzed to determine what can be
done to "fix" the situation. The various options are analyzed for cost, savings, and service
quality. However, no test is done for sufficiency or to see if an option "passes" any
standard. All of the options evaluated are put into the list. (By comparison, in the traditional
paradigm depicted in Figure 6.5 only those that fully meet standards are put on the list.)
It is in the prioritization and approval stage where this process deviates substantially
from the traditional paradigm. In the traditional paradigm, this final stage leading to
authorization is still project-oriented. Options are reviewed by the process on a project-by-
project basis and the least cost for each is selected. The process guarantees that one option
will be selected for each project.
By contrast, here the final stage is portfolio-based. Options for each project are mixed
into one large group, and the entire set of options is sorted, best to worst. The best options
from the entire set are then selected. There is no built-in certainly that one option will be
selected from each project - some may receive nothing. But there is an assurance that a
specific target will be met - either a reliability goal or a budget limit. The money spent will
be spent in the most effective way possible overall.
Strategic Planning Analysis at the Executive Level
The iB/C-facilitated planning-prioritization-approval process provides a capability that has
no counterpart within the traditional paradigm and its process, a formal way for executive
management to study the relationship among cost, long-term savings, and reliability in an
effective way: one that lets them make informed decisions about budget cuts or reliability
targets. The entire options list, not just selected options, is used to form a Pareto curve
which is the basis for this planning. While Pareto curves for next year, and other future
years, might be slightly different that this current year's curve, they are reasonably
stationary over time and this year's will provide a useful guide. And, of course, it is
possible to perform this type of evaluation and planning looking several years ahead,
providing a more strategic look for long-range executive planning. Thus, management has
the capability to see in advance roughly what will happen to reliability, to long-term O&M
costs, to future expansion costs - to whatever - as it determines its budget.
Furthermore, once a budget or a reliability target has been set the point on the Pareto
curve is identified and the incremental benefit/cost ratio from that target (the slope of the
curve at that point) can be used as a short-cut in work throughout the utility. This target
iB/C ratio can be communicated throughout the organization, to planners, engineers, and
designers. It provides a useful benchmark for any project or option a planner is considering
for any evaluation being performed: if it does not come close to or better than this target,
then there is no chance of its gaining approval. This both streamlines and focuses the entire
organization.

218 Chapters

iB/C as the Tactical Tool of an "Asset Management" Strategy
The "portfolio" approach to assessment and allocation fits well in the "asset management"
strategy and organizational structure adopted by many electric and gas utilities (along with
many other businesses). "Asset management" is applied in the utility industry in a
bewildering array of meanings, to the extent that the term is nearly meaningless without
clarification by the user as to what it means in his or her context.
To some utilities, "asset management" refers to the processes, technology, and
procedures they use to manage and service the physical assets they own and operate -
maintenance planning and scheduling systems, repair and service practices, etc. Other
utilities use the term to describe a type of organizational structure in which responsibility
and accountability for performance are assigned on the basis of physical equipment
(circuits, substations), as opposed to assigning them on the basis of job function, geographic
area, level of the system or type of equipment, etc.
But an asset management approach to business, and the definition of the term as used
throughout this book, means the adoption of physical asset management principles which
are based on concepts borrowed from the investment management field. This means that
how a company considers the value of its physical property, how it uses that property, how
it cares for or neglects some or all of that property, and how it invests in new and existing
property are based on an overall business perspective that has:
1. A top-down-driven structure with a (hopefully) coherent organizational focus
on maximizing financial and corporate goals. Among these goals is nearly
always a commitment to high profit margin, but concerns about stock price,
risk indices, and other measures of financial health may be as or even more
important.
2. A portfolio approach to the management of assets and investment. This means
that decisions on any one investment or asset are made not just on the basis of
its situation or case, but partly on the basis of the status and needs of other
assets and investments of the company.
3. Minimization of the risk of failing to meet the corporate goals, accomplished
by diversifying (spreading investment) over the portfolio in order to mitigate
known vulnerabilities and minimize financial as well as scenario risk.
Clearly the iB/C method discussed here fits well within this environment. To begin, it
fits a top-down, strategy-driven business structure: through its Pareto curve evaluation it
gives strategic decision-makers a tool to measure and project the T&D impacts of their
financial decisions, so they can better fit them into its overall strategic planning. The
method is a natural fit for a portfolio approach. And to an extent, with its probabilistic
approach, it addresses many of the physical risk-related concerns of an electric utility
business.
However, "asset management" generally goes beyond looking at "risk" and
"uncertainty" in terms of physical asset performance. This approach to business governance
also means looking at uncertainties and risks beyond the context of the T&D failures
considered by "bang for the buck" analysis. At the top, an asset management utility often
uses either a probabilistic or scenario approach (or both) to study known uncertainties (e.g.,
variations in weather with its consequent affect on energy and demand, and hence revenue),
as well as to try to limit exposure to risks that are harder to analyze if not articulate
(technological obsolescence, possible changes in public buying habits, likelihood of war).


6.6 PROFIT-BASED PLANNING PARADIGMS

It is unlikely that T&D systems will ever operate in an environment where the basic
decision-making process is made within a purely profit-based value system, as is done in
purely de-regulated industries. However, even for a regulated utility, some ancillary and
optional services and equipment may be treated as profit-making enterprises. For
completeness, it is worth examining how the profit-motive paradigm makes decisions about
investment and spending.
A purely profit-motivated paradigm focuses on trying to maximize the return on the
investment. Electric utilities certainly have this as one priority, too, but they are working
within a regulatory framework and are subject to a number of constraints and the precepts
of the traditional regulated utility commitments. While there are variations on this theme, in
general the profit paradigm has several specific differences from traditional utility planning
paradigms. These and other major factors affecting planning may not be identified
explicitly, but the key differences that change how planning decisions are made are:
1. Serving all the demand is not necessarily in the business's best interests.
Serving the most profitable portion is.
2. Standards are defined by whatever quality and service characteristics yield
the highest profits.
3. Profit is the goal. Specifically, the goal is usually to maximize the
return/investment ratio or the return-to-cost ratio.
A basis for many profit-driven corporate value systems is a belief that if a good return on
investment cannot be found among the opportunities involved in a project, other potential
avenues of investment are available. Thus, the third goal given above is not to maximize the
profit, but to maximize the profit as a percentage of investment: Earning $10 million profit
on a $50 million investment is much better than earning $17 million profit on a $100
million investment.
Example from Outside the Electric Utility Industry
Consider the options an airline faces in deciding what size commercial jet to buy and put
into service on a shuttle route between two cities like Chicago and Tulsa. Analysis of
demand, competition, and other factors might establish that while demand varied by season
and day of the week, normally between 100 to 150 people would want to travel this route
on any particular flight.

Table 6.17 Profit Analysis of an Airliner for Shuttle Service -$ x 1000
Type of Passenger Purchase Annual Annual Annual Annual Percent
Airliner Seats Price Op. Cost Cost Revenues Profit Profit
Duojet 75 70 $7,000 $2,100 $2,975 $3,340 $365 12%
Convex 90 85 $7,277 $2,550 $3,460 $4,056 $597 17%
Duojet 100 100 $7,815 $3,000 $3,977 $4,772 $795 20%
Europlane 2JR 125 $8,776 $3,750 $4,847 $5,560 $843 17%
Convex 180 150 $10,221 $4,500 $5,778 $5,726 $10 0%
Convex 220 225 $12,910 $6,750 $8,364 $5,726 -$2,576 -31%
SonicBus 300 300 $17,271 $9,000 $11,159 $5,726 -$5,371 -48%

220 Chapter 6

Table 6.17 shows the airliners available, which range from 70- to 300-seat capacity,
along with their initial costs (a 300-seat airliner is more expensive than a 150-seat airliner,
but not twice as expensive), annual operating costs (much more linear as a function of size),
and total annual cost (a function of capital costs plus operations costs).
Any airliner with less than 100 seats will always travel fully loaded. Thus, below the
100-seat limit, revenues will be proportional to number of seats. But any plane with more
than 100 seats will fly only partially loaded at times. Additional seats over 100 on the plane
will earn money, but only on some flights. The 300-seat airliner will never fly with more
than 50% of a load, or generate more than half its potential for revenue.
The 125-seat Europlane 2JR generates the most profit, $843,000 per year. However, the
100-seat Doujet 100 generates the highest percentage profit, 20% as opposed to the
Europlane's 17%. Most businesses would buy the smaller aircraft, the assumption being
that the money not invested by buying the smaller plane can be invested in another jet-route
combination similarly earning 20%, not 17%.
Obviously, many more factors are part of any actual airline-purchase decision. But the
point is that an option far below peak capacity typically looks very good to an unregulated
business looking to maximize profit margin. The fact that customers may have to be turned
away at peak periods is of tertiary importance in this decision. Airlines routinely both turn
away customers because their flights are full and overbook flights. It is worth noting that
both practices are business decisions the airline makes because, ultimately, those decisions
maximize its profit. The airline could trade the smaller plan to buy a bigger plane. It could
stop overbooking flights. It does both because those decisions make money.
Profit Analysis of a Feeder Example
The concept of deliberately planning for less than peak capacity is antithetical to planners
who have worked their entire career in an industry where falling short of capacity is deemed
a sure sign that the planners failed in their mission. However, completely unregulated
businesses have no obligation to serve. To a great extent, they pick and choose their
customers (their market shares and target consumers) based on studies to determine if they
can make a profit selling to them. This is equivalent to a utility building circuits only in
areas that it believes make a positive business case and only to a size and reliability that
maximizes its profit. The unwillingness of many utilities to extend service to "unprofitable"
areas is one reason that the industry became regulated on an obligation to serve basis (see
Philipson and Willis, 1998).
Table 6.18 shows the example new feeder options from earlier in this section evaluated
on a purely profit-motivated basis in a manner similar to the airliner example discussed
above. Interim computations required to develop the numbers shown is not given here, nor
are all of the data needed. The point here is only to demonstrate the concept and results, not
the details, of computing business costs, etc.
The capital, O&M&T, and failure penalty costs from Table 6.14 are used, allocated on a
annualized basis with capital having a charging charge added. Here, it is assumed the utility
refunds the failure penalty costs used earlier to its customers, so they are an actual cost.
Revenues are projected from an 8760-hour analysis of electric delivery services (i.e., the
revenues shown are an appropriate annual fee for delivery and do not include the price of
the energy, just the delivery charges).
Interestingly, the traditional paradigm's choice, option 7, has the highest amount of
profit (a coincidence). However, it is not the option with the highest profit margin and not
the option most profit-motivated businesses would build. What is interesting is that, even


Table 6.18 Profit Analysis of a Distribution Feeder - $ x 1000
Project Project Initial Annual Annual Annual Annual R.O.I
Option Description Cost Cap. O&M&T Failure Costs Revenue Profit %
1 1 4/0 large trunk $2,250 $243 $470 $2,086 $2,799 $1,051 -$1,748 -77.7%
1 2 4/0-4/0 branch $2,350 $254 $437 $376 $1,067 $1,051 -$16 -0.7%
1 3 336-4/0 branch $2,775 $300 $330 $175 $805 $1,051 $246 8.9%
1 4 336 large trunk $2,850 $308 $316 $88 $712 $1,051 $339 11.9%
1 5 636 large trunk $3,115 $337 $273 $49 $659 $1,051 $392 12.6%
1 6 795 large trunk $3,550 $384 $224 $9 $617 $1,051 $434 12.2%
1 7 636-336 branch $3,600 $389 $218 $6 $613 $1,051 $439 12.2%
1 8 795-336 branch $3,800 $411 $204 $0 $615 $1,051 $436 11.5%
1 9 795-636 branch $3,875 $419 $200 $0 $619 $1,051 $432 11.2%
1 10 1 1 13 large trunk $3,900 $422 $198 $0 $620 $1,051 $431 11.1%
1 11 11 13-633 branch $4,250 $459 $173 $0 $632 $1,051 $419 9.9%

with the very substantial failure penalty function used here taken into consideration as an
actual cost, this profit-margin maximizing paradigm picks a rather lean feeder, option 5, as
the best result. This is the same option that the 20% budget constrained case in Table 6.10
picked, one requiring noticeably less up-front investment than the traditional paradigm
selected.
Failure cost was a big factor in determining the result here. The reader can confirm that
if failure costs are removed from consideration completely, the maximum "simple" solution
for profit margin becomes option 1, an extremely lean feeder that would give very poor
service quality during peak periods. The failure cost penalty has to be doubled (the scalars
in equations 6.2 and 6.3 raised to $4 and $6 respectively) before the traditional option, y, is
selected (and even then it only "wins" over option 6 by 0.04% profit margin). Of course,
many other considerations enter into business decisions than just those discussed here. But
the point is that a profit-focused paradigm provides significantly different incentives and
leads to quite different results than either the traditional or budget-constrained utility
paradigms discussed earlier.
The example given here demonstrates one reason why it is unlikely that T&D systems
and their operation as a delivery network will ever be fully de-regulated. What is in the best
interests of a profit-motivated business would not necessarily be in the best interests of the
public, when sufficient electric capacity is viewed as a near-essential item for proper
functioning of society and the common good. Regulation provides both an assurance that
such needs will be addressed and a mechanism for the utility to recover the costs required to
meet these goals.
6.7 SUMMARY, COMMENTS, AND CONCLUSION
Evaluation of merits, ranking of options according to merits, and selection of options to be
pursued based on these merits are a part of every utility planning process. However, there a
number of different valuation-prioritization-selection systems, each more appropriate than
others for certain environments and/or planning when certain issues are more relevant to the
decision process. Every utility, regardless of its situation, must select some means of
evaluation, prioritization, and approval for resources (money, labor, skill) it must put into

222 Chapter 6

performing its work. Of course, there are many variations on the themes discussed here, but
for the most part the evaluation-prioritization-approval paradigms presented in this chapter
cover the range of methods available to and in use by utilities.
Evaluation, Prioritization, and Selection: Cogs in the Machine
Evaluation, prioritization, and approval/allocation processes at a utility are not strategic
tools, and they need not be business case based. All of them work within a business that
has, at some executive level, been studied and developed on a business-case basis. Its goals
and priorities have been set, and someone has already decided how to organize and set up
the company to achieve those goals.
Thus, an important point, and one often unrecognized by many T&D planners: these
processes do not have to evaluate options on a business case basis to be business related or
fit within a "business driven" organization. Evaluation-approval processes are expeditious
tools that musty?/ inside the organization. They are part of the tactical mechanism of T&D
planning, whose job is to help implement that strategy. What is important is that the
evaluation-prioritization-approval process be coherent with and fully supportive of the type
of strategy and goals the organization has.
An ideal evaluation-prioritization-approval system
does not need to perform "business-based" evaluation
of project and program alternatives. It merely has to
lead to decisions consistent with and supportive of the
company's business strategy. The ideal evaluation-
prioritization-approval system is the simplest and most
economical system that will lead to decisions that fully
support the company's business strategy and goals.

Therefore, the ideal evaluation-prioritization-approval system is the simplest and most
economical system that will lead to decisions that fully support the company's business
strategy and goals.
For example, the traditional least-cost, project-oriented paradigm did not evaluate
options on a regulated-utility "business case" basis.7 Instead, it efficiently led to evaluation
of options and selection of projects that were consistent with a traditional regulated utility's
business goals. Similarly, other approaches fit organizations with other business strategies
and goals. For each company, the T&D evaluation and prioritization system selected for use
should be efficient, simple, quick to apply, yet compatible with and supportive of the
strategy, goals, and managerial approach that has been selected by company management.
The Traditional Standards-Driven,
Least-Cost, Project-Oriented Paradigm
Traditionally, electric utilities used a standards-driven, least-cost, project-oriented
evaluation and selection process to determine both how much capital they should invest and
how it should be allocated. That paradigm minimized long-term costs, using a time value of
money approach to balance initial capital investment against long-term need and savings.

7
Even the new-service evaluation (section 6.2) does not really perform a business case analysis. It
merely applies business case related formulae to "back out" of the particular project being studied,
the maximum that the utility's business case will support.


T&D planning is a forward-looking forecasting and decision-making process that determines if a
system will have a deficiency unless reinforced, decides when that reinforcement must be made, and
then recommends what should be done.
Evaluation is a process of appraising a unit of equipment, project, program, alternative, or some
other item which is involved in a decision.
Prioritization is the process of ranking a number of projects or alternatives on the basis of their
merit with respect to some value system, for which they have been evaluated and can be compared
on a uniform basis.
Attributes are characteristics of a planning decision or design that are to be minimized or
maximized. By far the most common attribute in electric utility planning is cost.
Criteria are characteristics of a planning decision or design that are important, but which must only
meet certain minimum requirements, such as voltage level.
Attribute value is never sacrificed to improve any criteria beyond its minimum satisfactory level.
Single-attribute planning involves situations where there is only one factor that is to be minimized.
Most, and all traditional, T&D planning was single-attribute planning directed at minimizing cost.
Multi-attribute planning involves situations where there are two factors - attributes - which
planners or their utility wish to minimize, such as cost and reliability or cost and environmental
impact.
Pareto curves show the results of analysis for multi-attribute situations, for example showing the
cost versus reliability options that a utility faces with respect to expanding or operating its system.
Paradigms are fundamentally different value systems for planning. The fundamental differences
among these decision-making paradigms relate to how they value and respond to the various
evaluated factors.
The traditional planning paradigm is a project-oriented, standards-driven approach to meeting a
utility's obligation to serve which seeks to achieve the least cost on a project basis.
It does not assess risk or attempt to minimize risk in any formal manner.
It does not include any mechanism to directly limit spending.
Benefit/cost ratio analysis and prioritization is a paradigm traditionally used by utilities for
evaluation of and decisions on programs and opportunities that are optional but perhaps prudent.
Incremental benefit/cost analysis (or marginal benefit/cost analysis) evaluates items based on the
ratio of incremental benefit to incremental cost.
Budget-constrained planning uses incremental benefit/cost analysis in a project-oriented, portfolio-
based approach to meeting a utility's obligation to serve that seeks to minimize the need for spending
to meet a customer-service quality target.
It does not directly assess risk but can be used in combination with probabilistic methods to
minimize certain risks.
It includes a direct mechanism to directly limit spending.
A purely profit-motivated paradigm focuses on trying to maximize the return on the investment
and usually underbuilds capacity in order to achieve utilization rates that are higher than the load
factor.
No one paradigm is ideal, or even applicable, in all situations. The traditional paradigm is not
inappropriate because the others are more recent developments. If it fits the current situation, it
should be retaining. Shifting paradigms only makes sense when condition shifts drive that change.

224 Chapter 6

This paradigm was ideally suited to the historical electric industry regulatory and financial
structure. First, this paradigm minimizes long-term cost. A long-range, least-cost
perspective was appropriate given the expected lifetimes of the expensive physical assets
that most utility capital went to purchase.
Second, the paradigm tended to be investment intense. "Building for the long term"
means spending more capital today to reduce future O&M costs, and as traditionally
applied this paradigm usually recommends spending the maximum amount of capital that
can possibly be justified on the basis of long-term payback. This was a positive virtue to
mid 20th century utilities, who made the regulated profit margin from every dollar in the rate
base and who worked within a regulated industry whose mechanism protected them from
most of the vulnerabilities their high debt/equity ratios would have caused non-regulated
businesses.
Third, it uses only a project orientation. Each project's evaluation is de-coupled from
the rest; it requires no integration or coordination with other projects. Thus, from beginning
to end this paradigm has no macro perspective or aggregate analysis of spending or
prioritization. Thus, it can be applied to as many projects as needed, in parallel if need be,
with no increase in complexity or cycle time and only a linear increase in the serial
management "traffic" in the final approval stage. Modern utility planners should recognize
this was of tremendous import to utilities in the eras before photocopiers, centralized
databases and computerized spreadsheets, and analytical tools.
But this paradigm had two disadvantages, both being at most minor inconveniences
through most of the mid-20th century, but increasingly of major concern to utilities as the
new century dawned. First, the least-cost, project-oriented structure has no mechanism to
assess, control, and integrate spending and spending decisions on a strategic basis. The
same virtue that makes it easy to manage also renders it incapable of meeting the needs of a
cost-reducing business or a business that wants to manage itself on a "portfolio" basis.
Also, the long-term perspective used by this paradigm creates increased business risk
from several standpoints. As mentioned in section 6.3, the high debt/equity ratios created by
an investment-intense approach make a business sensitive to changes in the economy,
weather (and the shifts in revenue it creates for a utility), regulation, and a host of factors
whose future vagaries are unpredictable. For a host of reasons, the business performance of
conventional, vertically integrated electric utilities in the mid- and late-20111 century tended
to be far less sensitive to these factors than that of the performance-regulated "wire
companies" of the 21st century.
And, finally, the traditional utility long-range perspective of least-cost selection and the
long-range debt it created to "build for the future" made electric utilities very sensitive to
technological obsolescence of equipment and debt. What was not considered in the vast
majority of utility long-range PW planning case was that while durable electrical equipment
might be fully functional after thirty years' of service, it might not be competitive.
Technological obsolescence would mean that equipment and systems developed during that
time would grossly outperform it.
Frankly, this was not a concern to a fully regulated utility in the traditional (mid- to late-
20th century) industry framework. The equipment they were considering would last 30
years. The equipment being assessed would, regardless of its potential obsolescence,
perform to its specifications during its entire lifetime. Therefore, if the evaluation of its
performance and costs, fit to need, etc., was accurate, its evaluated business case over the
next 30 years would be accurate. The fact that a better option might develop during that
period, and that consequently some shorter-term option might actually be preferable, was
irrelevant, because the system allowed for no competition. The equipment would go on the
system, its cost would go in the rate base, the debt created by borrowing to buy it would go


on the utility's books, and the utility's customer would pay for it all under regulated rates
that "guaranteed" the utility a profit.
The generation plant stranded-cost issue in power industry stems from this effect. Every
stranded electric utility generating plant was justified by the traditional paradigm - the
least-cost option to satisfy need, taking a long-term perspective. But many power plants are
stranded, because while still performing just as well as initially intended, they cannot
compete against superior, modern technology.8 Similarly, many other utility decisions made
on equipment and capacity look less than optimal in the light of technological progress in
the fields of equipment durability and lower maintenance, automation, and improved
materials.
But despite these shortcomings, the traditional, standards-driven, least-cost, project-
oriented evaluation and prioritization paradigm has some features that are not to be ignored,
and it is satisfactory for certain situations: those where there is no shortage of capital to
invest, where reduction hi tomorrow's operating costs is important today, where little
uncertainty exists about future conditions, and where the utility's financial risk is
reasonably covered by the regulatory process. Outside of situations like that, its
disadvantages generally outweigh its advantages, which is why all de-regulated business
and many business-focused utilities look elsewhere for a planning approach to help them
decide how to invest in their system.

Typical Modifications Made Within the Traditional Paradigm
Many modified versions of this traditional paradigm have been developed by utilities in
order to "fix" real or perceived problems with its performance as a decision-making tool.
As mentioned in section 6.3, among the most common have been modifications to the time
value of money factors (discount rate, present worth factor) used, or use of much shorter
evaluation periods (the author has seen utilities use as little as 8 years, instead of the
traditional 30). This does address much of the methods sensitivity to long-range
uncertainties, as alluded above. However, it does nothing to address its inability to cut costs
in a way that responds to present urgent needs. This, which might be called the "short-term"
version of the traditional paradigm, is appropriate as a response to concerns about long-term
uncertainty, but it is still not appropriate for either budget-constrained situations or for
utilities that want to maximize the effectiveness of money they spend today in fixed
reliability problems.
In addition to these changes, the author has seen many heuristic additions or
modifications to the traditional paradigm made by utilities facing tough budget cuts or
worried about how to apply reliability enhancement budgets well. Many of these methods
are labeled with impressive names and become part of significant "initiatives" to drive
change inside a utility. Most fail because the method is ineffective in meeting the planning
needs of the organization.

This is not an issue created by de-regulation. All that affected was who "got screwed." De-
regulation meant that utility stockholders assumed ownership of generating plants that could not
compete in a de-regulated marketplace. Had de-regulation not occurred, the "efficiency" gap
between those plants and modern technology would have still existed, but it would have been
consumers who absorbed it, as the regulatory process forced them to buy power form those units at
their (higher than modem technology) production cost. The regulatory process was perhaps partly to
blame for this, but the real problem was the long-range cost-reduction approach used by utilities and
regulators alike and its lack of regard for the vulnerability a 30-year debt on an asset has on its being
"worn out" by technological progress, not service wear or lifetime.

226 Chapter 6

Benefit/Cost Ratio Evaluation and Approval
From time to time, nearly all organizations evaluate investments or changes in procedure
that are optional within the context of their purpose or functions, in the sense that these
possible activities are not mandatory or essential for the continued operation of their
business. Instead, these optional activities might produce a benefit such as reduced
operating costs, lower plant-shutdown rates, smoother and more predictable production line
operation, or some other similar improvement.
Evaluation and appraisal of the value of such optional projects can be done objectively
and reasonably from the standpoint of most businesses by comparing the benefits to be
gained against the cost to be incurred. This can be done in absolute (subtracting cost from
benefits to get gain, sometimes referred to as B - C evaluation), but is most often done by
comparing the ratio of benefit to cost (B/C evaluation). Prioritization is usually done using
ratio evaluation. Absolute evaluation, while completely valid, is not as useful for ranking
different proposed projects as to their relative merit as an investment made for gain,
because it does not rate an optional project as to how advantageous it is on that basis.
For example, in Table 6.6, direct commercial water heater control, with a B/C ratio of
1.78, has a far lower absolute gain (cost $30,000, benefits $53,000), than residential interior
lighting upgrades ($364,000 cost, $456,000 benefits) simply because it is a program that
will cost less than one-tenth of the larger program's cost. Yet each dollar invested in it
provides nearly three times the savings of that larger program. Ratio analysis reveals this
higher payback and is thus suitable for prioritization where the goal is to maximize the
"bang for the buck" among alternatives for optional spending or to determine if the
evaluated incremental benefit of one alternative over another looks like it is worth whatever
incremental cost or risk might be involved.

Payback Period Benefit/Cost Evaluation

Often, benefit/cost analysis is done on a time period basis in which the analysis is used to
rank options by how long it will take before the option's accumulated savings or benefits
first outweigh its cost. Such "payback period" evaluation is suitable for ranking and
decision-making: projects with a shorter payback period are superior in that they pay back
at a faster rate. In most cases, ranking a number of options by payback period or B/C ratio
will put the options in the same order of merit.
Yet planners and decision-makers need to keep in mind that payback period and B/C
analysis are not just different ways of measuring precisely the same thing. The timing of
benefits and costs affects both types of analysis, but in different ways and in ways that are
difficult to generalize. For example, the residential lighting upgrade program referred to
above (and in Table 6.6) has a shorter payback period than one might believe from the data
presented in this chapter, because it is periodic: light-bulbs and tubes last only three to five
years, so its evaluated B/C ratio of 1.78 applies over a few years.
Such programs often pay back their investment early in year two, but their savings will
not last without continued investment in high-efficiency tubes, etc., when those equipments
fail. By contrast, building shell upgrades (also in Table 6.6) is a program with a very large
one-time cost (and very small, periodic maintenance costs) which produces a "perpetual"
annual savings in energy costs. It often takes more than a decade for such programs to
break even, whether judged on a PW or actual dollars basis. The author recommends
evaluating and ranking optional spending options on the basis of both.


Incremental Benefit/Cost Evaluation,
Prioritization, and Portfolio-Based Approval
Incremental Benefit/Cost evaluation (iB/C) treats various options for mandatory and
optional projects or programs as progressive options, using a stacked or progressive-
spending approach to determine how much should be spent on each project or program.
Incremental evaluations means that various plans or sizes or other distinguishing feature
options are sorted on the basis of "cost" and then the incremental spending from one to the
other is assessed against the benefit gained over the next less expensive option.
iB/C evaluation is suitable for a portfolio approach to budget allocation, in which it is
desired to allocate money over a set of possibly very diverse types of projects, on the basis
of maximizing the overall gain (benefit, savings). In a portfolio approach, the ultimate
merit of a particular project or spending option depends on its own merits as well as the
merit of other candidates for spending and the total budget to be allocated. Thus, a project
that might be funded in some situations would not be funded in other situations, depending
on what its competition for funding was and how much money was available overall. This
is in marked contrast to the traditional paradigm: a project that is assessed as "good
enough" is good enough regardless of what other funding requests are on the table or how
much money there is.
iB/C used for budget-constrained prioritization (BCP)
iB/C is usually part of a multi-attribute evaluation process in which reliability (or something
like it) and cost are being balanced against one another. Often, there is deficiency of one or
the other of the attributes, a deficiency which is a major concern, perhaps the overriding
concern, for the decision-makers at the utility. When this deficiency is capital, or a
limitation of labor, skills, time, or some other important resource, iB/C is applied with a
budget-constrained focus in order to help decide where and how to trim spending or
resource allocation to fit within the available budget.
Usually, in these situations, the utility retains most if not all of its traditional values and
priorities (minimizing of long-term cost, adherence to a set of specification-based
standards). iB/C, applied on a portfolio approach, is employed because the utility can no
longer buy as much of these traditional targeted values as it needs.
iB/C used for reliability-based spending prioritization
iB/C is also appropriate for and often used in situations where, whether there is a seriously
inhibiting capital spending limit or not, the utility has really undergone a shift away from its
traditional values Budgets may be tight, but utilities in this category have shifted away
from investing greatly to minimize apparent long-term costs because they are not confident
long-term perspectives on future trends, market and regulatory conditions, and technical
obsolescence are viable. Also, many utilities have eschewed specification-based standards
in favor of service quality standard (reliability target) driven planning.
Thus, while mechanically identical to budget-constrained prioritization (same
procedures, etc.), iB/C is being used in a far different environment in these situations, one
that is usually much more of a true multi-attribute value environment. Often, the utility's
goal is to achieve a particular reliability level (the target level) as quickly or with the least
possible investment or effort as possible. It is here that one sees often innovative
applications of involved failure penalty functions and the other "tricks" alluded to in section
6.5.
In a few cases, a utility's planning may be done to a budget constraint, but one that is
positive: the amount to be spent is greater than the traditional budget. Usually, these

228 Chapter 6

situations are due to a special initiative. Special funding is allocated for reliability
improvements. It is desired to spend this amount to get the most improvement in today's
reliability possible. Combined with the "normal" budget, the amount of money to be spent
is more than the traditional paradigm would specify. (Best results from this "extraordinary"
budget allocation are obtained when it is combined with the normal spending budgets and
the two blended in application: "all projects should be treated as reliability projects."
Regardless of the precise focus, iB/C is also applied in these situations on a portfolio
basis, determining not only what projects or programs are to be funded, but to what level.
Overall "bang for the buck" is optimized in the sense that no higher level of benefit (e.g.,
customer service reliability) can be bought for the amount of money to be spent, and/or no
lesser amount of money can buy the level of reliability attained.
"Asset management" approaches
iB/C's multi-attribute nature and "portfolio" approach mean it fits well into a utility being
run with an "asset management" approach (described in section 6.5). However, "asset
management" generally goes beyond looking at "risk" and "uncertainty" in terms of
physical asset performance. Normally, at the strategic level, business planners will look at
uncertainties and risks to the business itself. This will include assessment and formation of
tactics to reduce sensitivity to a variety of risk-creating uncertainties. iB/C fits into this
environment well, but is not the total answer. Usually, some sort of scenario-based
approach is used to look at a number of "best" and "worst" cases for the utility with regard
to the economy, weather, etc. Often, the Pareto curve of T&D system performance versus
cost is a very useful tool in these scenario studies.
Profit-Based Paradigms
Completely de-regulated businesses have no obligation to serve all consumers. They are
free to pick and choose their customers as they choose, or at least to try to by shaping their
offerings and the market niches they target. However, they face competition for all
customers and have no monopoly franchise, regulatory oversight, or regulated rate. Except
in cases where an unregulated competitive business is "buying" market share, or views the
pool of its investment opportunities as limited, or has some other non-standard motivation,
it will generally try to maximize its percentage profit margin - its return on investment.
The profit-based paradigm is almost always inappropriate for use by regulated electric
utility (i.e., wire company) planners, but it is interesting to compare it to the traditional
least-cost paradigm. In a sense, the two are at opposite ends of the decision-making
procedure spectrum. The least-cost approach assumes that there is an endless supply of
funds for whatever needs to be spent. It evolved under values systems in which the profit
margin (the utility's regulated profit margin) was constant regardless of the amount
invested. By contrast, the profit-based approach assumes there is an endless supply of other
opportunities in which to invest. Profit margin generally falls as one invests beyond a
certain point in any one opportunity, and the business's decision-making focus is to try to
influence overall profit margin by selection of where and what is invested.
There Is No Universal "Truth" to Utility Decision-Making
Every one of the paradigms discussed in this chapter is valid, each only within the context
and purposes for which it is suited. None, particularly not the traditional industry paradigm
used by virtually all electric utilities for decades, is "correct" or absolute with respect to
what decisions should be made. Frankly, it is very difficult if not impossible for some
planners and managers imbued with years of experience within the traditional paradigm to


recognize this. Many insist that a shift to another paradigm and the set of values it fits will
be a step backwards, a "lowering of standards." But that is not the case. The traditional
paradigm fit the historical industry structure and value system well — it was evolved to that
purpose. But taken out of its context, that least-cost, project-oriented approach really is a
bizarre process. It has no structural mechanism to limit spending - quite literally, it cannot
say no to a worthy project. It builds for the very long run, and while exactly what "long
run" means can be varied within the paradigm by adjustment of discount rates, etc., it is
basically a tool that allocates money based on long-term need. Similarly, each of the other
paradigms fits a value system within which it makes perfect sense and functions very
coherently, but it is "true" only within that type of system.
Regardless, some type of evaluation, prioritization, and budget allocation method has to
be used. The job of modern planners is to pick the method that will be most appropriate for
their utility, depending on the economic situation around it, its financial health and
constraints, the type and age of its T&D system, its customer base demographics and
preferences, its load growth, its regulatory environment, and the skills and experience of its
IT systems and human resources. The challenge for T&D planners is to recognize what they
need and to pick the right approach.
The Larger Context of Uncertainty and Risk
There is a larger context around the evaluation and prioritization paradigm used that utility
planners should look at and learn from. Unregulated businesses generally incorporate much
more business risk assessment into their overall strategic planning. While it is an
oversimplification to say that regulation ever gave any utility a "guaranteed" rate of return,
regulation greatly reduces the risk of an organization basic business case encountering
problems severe enough that it cannot survive (Philipson and Willis, 1998). While
unregulated businesses apply a profit-based paradigm as described earlier, the more
successful ones apply it along with a very comprehensive analysis of business risks and
uncertainty. In many cases, the majority of the business's planning is aimed not at selecting
or optimizing investment opportunities, but in assessing and setting up the business to
minimize risk due to a host of reasons - changes in the economy, in the cost of money, in
public attitudes or fashions, in technological obsolescence. Usually, such risk is assessed
with a combination of probabilistic and multi-scenario planning techniques, some quite
sophisticated.
Regardless, electric utilities generally have very limited experience in this type of risk
analysis. That inexperience showed during the period from the late 1980s through to the
early part of the 21st century when many large investor-owned utilities jumped into a variety
of unregulated businesses, usually in the energy, service, or technology sectors. While there
were many successes, there were many more failures, including several which drove the
parent company into such financial distress that it put its regulated business operations at
jeopardy - several large investor-owned electric utilities had to seek mergers or acquisitions
to maintain the financial wherewithal to continue their operations.
Even though electric power delivery utilities are and will remain regulated for the
foreseeable future, all have moved a few steps closer to operating in an "unregulated"
environment. Few remain vertically integrated. Most have to buy the power they deliver in
a competitive environment. Some have to sell it in one. All face some competitive pressures
(DG, alternative energy sources). Many are regulated on a performance basis. Some have
been bitten severely by technological obsolescence. And all face the fact that because some
of their peers did get into unregulated business activities in the past, the investment
community no longer sees "utilities" as the same low-risk investment area it once did.

230 Chapter 6

For all these reasons, utility planners need to expand the context of their planning scope
and consideration. Regardless of which evaluation and prioritization method they pick, and
how they implement it, the larger picture is that it must fit within a business strategy
development and planning framework that performs more of what was, traditionally,
unregulated business planning, including setting spending limits in a top-down manner
based on business risk and financial models. Generally, that means one or the other of the
approaches based on iB/C works best.

REFERENCES AND BIBLIOGRAPHY

New York, 1998.

H. L. Willis, Spatial Electric Load Forecasting, Second Edition, Marcel Dekker, New York, 2002.

World Bank, Guidelines for Marginal Cost Analysis of Power Systems, Energy Department paper
number 18, 1984.

7
Equipment Ratings, Loading,
Lifetime, and Failure

7.1 INTRODUCTION

T&D systems are an interconnection of thousands of building blocks, components arranged
and interconnected together to form a working power delivery system. A good deal of the
efficiency, economy, and reliability of the resulting system has to do with all the system
aspects of this interconnection - how equipment is selected, engineering, and arranged to
work together. Most of the chapters in this book focus on the planning and engineering of
those system concerns.
But the individual components are important themselves for two reasons. First, each
must do its job within the broader context of the system network: there are investment and
reliability issues associated with that mission. Second, if left in service long enough, every
component will eventually fail, usually unexpectedly.
This chapter looks at equipment, failure, and the how failure is the direct consequence of
equipment usage. As equipment is left in service it ages and deteriorates in various ways,
which eventually lead to its failure. Heavy loading, poor maintenance, or inappropriate
usage can exacerbate these deterioration rates. But, conversely, low loading and
conservative maintenance can make for a poor return on the investment in the equipment.
Thus, aging, usage, and the care of equipment in a system need to be managed.
Chapter 8 will examine the impact of equipment failures and failure modes and rates on
the behavior and reliability of a utility system. This chapter focuses on individual units of
equipment and what ratings, age, deterioration and damage, and failure mean in that
context. It begins in section 7.2 with a discussion of equipment ratings and lifetime. Often
the one is based on an expected or inferred level of the other. Regardless, loading and
expected lifetimes of key materials are inexorably linked, but fuzzily defined. Section 7.3
then looks at aging, deterioration, and damage and discusses some of their consequences.
This has particular relevance to the discussion on reliability given in Chapters 4, 23 and 28.
Section 7.4 summarizes the various measures that utilities take to extend or manage
equipment lifetime. Section 7.5 concludes with a summary of key points.

231

232 Chapter 7

7.2 CAPACITY RATINGS AND LIFETIME
Almost all electric equipment is rated in terms of its capacity to handle power. Very often
this rating is given in terms of, or directly derives from, a maximum rated current level.
Conductor and cable is rated for certain current levels. For example a common type of 336
MCM ACSR is rated at 540 amps. A particular transformer might be rated at 50 MVA.
Transformers and many other devices are rated in terms of MVA, for example a
"nameplate" rating of 50 MVA. There is a pragmatic basis for conductors and cables being
rated by current and transformers, regulators, and other devices being rated by MVA. First,
a transformer is designed and intended to operate at a specific voltage, while conductors
and cable are not. A 336 MCM conductor can be installed on just about any voltage of line,
the power (at 540 amps) it delivers varying depending on voltage. Cables, though also rated
by insulation level (e.g., 15 kV class), are intended to operate at any voltage up to that
rating. Again, a specified current level best meets their rating needs.
Transformers and similar electrical machinery do not have a wide "applied voltage"
range. A 50 MVA 138/12.47 kV transformer is designed to operate at that voltage. It can
handle a maximum of 210/2315 amps on the high/low sides. While it could be rated for
current, the specific voltage applicability implies that current has a direct MVA
interpretation, so that is used.

What Capacity Ratings Mean
The primary purpose of ratings is to provide a basis for identification of and comparison
among equipment, not to provide recommended operating limits. A "50 MVA" transformer
is rated at that level because it satisfies the definitions of a particular guideline ("standard"
or definition) of how to rate transformers. There are several points any planner must
understand about this "rating":
Very likely (almost certainly) a different guideline (standard) would result in a
slightly different rating - 51.3 MVA, 49.7 MVA - there is no way to tell except to
evaluate the unit with respect to that other standard. There are many standards for
transformer rating, IEEE and IEC being only two widely used ones.
The rating is a valid way of comparing the unit against other units rated under the
same standard. A 50 MVA will always have very nearly 25% more capacity than a
40 MVA unit, regardless of the "standard" used.
The rating is not a recommended loading. Of course, those groups of engineers in
charge of the various standards have assured their usefulness by arranging that
rating value bears some similarity to actual usage capability. But a 50 MVA unit
could easily be capable of actually serving only 45 MVA in service, or 55 MVA,
depending on a variety of environmental issues, load patterns, and other factors.

Loading, Rating, and Lifetime
All capacity ratings are linked in some sense to the concept of "lifetime." Often, this
linkage is incomplete, not directly applicable to what a planning or operating engineer may
think of when using the world "lifetime," or so nebulous that many people would disagree
with the previous sentence. But the fact is that capacity ratings — particularly when
interpreted as a capacity limit - are always in some way associated with the concept of "end
of life." This may be only in a terminal sense - put more current over an ACSR conductor
than its maximum rating for even a few hours and it will anneal and fail soon thereafter.

Equipment Ratings, Loading, Lifetime, and Failure 233

More generally, capacity ratings, particularly for transformers, motors, regulators, and
other wound devices, have their basis in deterioration rate, the concept being that loading
causes deterioration which shortens remaining life and enough deterioration means the unit
has exhausted its life. A widely used (IEEE) rating has a simple interpretation. The rated
capacity is the loading that will result in gradual deterioration of the transformer's internal
insulation (due to the heat from electrical losses to a defined point (50% of its mechanical
strength) over a certain amount of time. Other transformer loading standards work in much
the same way, conceptually, although they vary on definitions, including the amount of
time, the amount of deterioration, and what these terms mean. But regardless, almost all
capacity ratings are based on a relation to "life" in some sense.
Loading Versus End of Life
Given enough time, the deterioration due to the heat generated by losses will result in
deterioration of insulation strength which will cause the failure of a transformer, motor, or
regulator, usually in a manner that damages it beyond repair. Loaded to a higher loading,
the unit will deteriorate faster and very likely reach a point of failure sooner. Nearly all
power engineers both accept this as qualitatively accurate and use it in their work.
However, "end of life failure" is not that simple. One of the failure modes for equipment
does correspond exactly to this concept. A transformer or motor operates for a period of
time (usually many years) all the while suffering from gradual deterioration of its insulation
strength due to the heat from its losses. At some point the insulation has lost so much
strength that it fails electrically, a fault develops, and the unit "dies."
Many transformers and motors "die" in exactly this way. But only a portion. Many other
types of failures occur. There is a constant exposure to "damage events" that every unit
suffers. A passing bird streams excrement onto the unit's bushings and the resulting fault is
not cleared quickly enough. A vandal holes the unit with a bullet in just the wrong spot.
Lightning strikes the unit and destroys it. Failure from these and other external causes,
unrelated to insulation strength, is a constant possibility. Given enough time, any unit left in
service would "die" due to some cause.
But perhaps through-faults are the major contributor to transformer failure. A major
fault passing through a transformer shakes it severely. Large power transformers can "ring
like a bell" and move by several inches. The transient mechanical forces acting on bracing
and insulation inside them are tremendous. Many power engineers, including the author,
believe that through-faults are the predominant cause of transformer failure. Although
infrequent, any transformer will see a number if left in service long enough. Each
successive fault weakens the unit's structure, bracing and fittings, leaving it slightly
"looser." Eventually, with its insulation's mechanical strength weakened due to gradual
deterioration from the heat of losses, the insulation experiences a through-fault whose
mechanical transient leads to its electrical failure. An internal fault occurs. The unit "dies."
Thus, any "lifetime" associated with a "rating" has to be treated for what it is, a
number based on a standard, a number that has only a very limited meaning with respect to
actual expected service life. Even if the capacity rating associated with it correlated exactly
with the unit's actual usage, that lifetime would have only limited usefulness, because the
unit is exposed to many other possible failure cases, and the "insulation strength" being
used as a proxy for lifetime is at jeopardy from a failure mechanism not taken into account
by the rating standard. Table 7.1 gives failure causes that the author believes represent an
industry cross-section for the failure of transformers and motors. These are based on
experience and should be viewed as qualitative.

234 Chapter 7

Table 7.1 Causes of Failures for Transformers and Motors
Cause Transformers Motors
Long-term deterioration of insulation 50% 70%
Through-fault 40%
External damage 10% 10%
Mechanical damage/wear by connected load - 20%

A Conceptual Electrical Equipment Lifetime-Failure Model
The author's conceptual model of electrical equipment deterioration and failure, particularly
as applied for transformers, is shown in Figure 7.1. Three factors associated with
lifetime/failure are involved, explained with respect to transformers below:
A background failure likelihood, shown as a base possibility (dotted horizontal
line) that "unwelcome events" can happen at any time - lightning strikes,
external physical damage, unknown causes, etc.
A slow, linear deterioration in internal strength caused by service, shown as a
straight (dashed) line. In the real world, this is usually a stair-step function. For
example, each successive through-fault degrades a transformer's strength.
However, for planning, one knows nothing about when such events will occur
so this is modeled as a gradual deterioration. This susceptibility or "looseness"
is added to the background damage likelihood (dotted line), so that the dashed
line starts from the height of the dotted line.
A gradual degradation in strength of insulation, an exponential decay as the core
insulation degrades in mechanical strength due to heating from losses (solid).
This follows the exponential model of insulation strength deterioration
(Arrhenius theory of electrolytic dissociation) which will be discussed in detail
later in this chapter.
A deterministic representation of this model (the plotted lines in Figure 7.1, without the
probability distributions) views the unit's "net strength" as the gap between the dashed line
and the solid line. As insulation strength degrades and susceptibility grows, the gap
narrows, increasing the likelihood of a "life ending failure." At some point (A), when the
gap becomes negative, the unit is running on "borrowed time" and failure is likely. The
probability distributions attempt to illustrate the actual concept as the author uses it, which
is that all three factors are probability distributions.1
The likelihood of failure in the probabilistic application of this concept is given by the
amount of overlap of the insulation-strength distribution and the combined through-fault +
damage event distribution. This has proved to be a useful model, but one must remember
that it neglects a number of details.

1
A transformer will have a maximum through-fault level that it will see at the bus at its location,
perhaps 10,000 amps or more. But many through-faults will be caused by faults far out the
distribution circuit, resulting in far less than the maximum fault current. Thus, a variation in the
impact of damage events is to be expected. That variation can be analyzed and used in predicting
lifetime.


1.00

c
o
(A
™
(0
Q.

O
O
.C
•*-•
O)

Time

Figure 7.1 Simple conceptual model of transformer failure. See text for details.

Thus, as modeled, equipment lifetime is predominantly a function of four factors. The
first is the design of the unit. A unit built stronger, heavier, or of a more robust design will
last longer.
The other three factors relate to how the unit is used. These are factors that the planner
can vary in his planning (at least in a study) to determine the unit's expected service
lifetime. First, the level of exposure it will have to damage from external elements. A
transformer installed outdoors, in a harsh weather area (lightning, etc.), and near a road
(possibility of automobile accidents) will have a lower expected lifetime, than one installed
indoors and completely protected. Second, loading greatly affects deterioration rate, so
lower loading levels lengthen expected lifetime, and higher loadings shorten it. Third,
through-faults erode transformer strength, so a unit that is placed into a system and operated
so it will see faults only infrequently, and only of rather modest fault levels, will last much
longer than one exposed to more, and higher level, faults.
Looking at Rating Models and Standards in More Detail
There are also other "ratings" for many types of equipment. As stated above, insulated
cable has a rated maximum operating voltage. A transformer will have a maximum rated
short-circuit current. There are many other ratings which will not be discussed here. The
rating of most importance to a planner is the capacity that the equipment can carry, whether
that is stated in current or MVA. The actual basis for that rating is generally based on a
standard - a consistent and fully defined basis for defining the rating, specifying conditions
and other terms that are important. Thus, the 336 MCM conductor is rated at 540 amps
because that is the current that results in a maximum conductor temperature of 75 degrees
C, in steady state conditions of 25 degrees C with a wind speed of 1.4 miles per hour. In a
very similar way conceptually, the transformer is rated at 50 MVA under a very specific
standard of conditions.
There are so many variations, exceptions, and details involved in transformer ratings
that this discussion will only summarize them. Transformer ratings typically assume an
ambient temperature (usually about 30°C), an average winding temperature rise of 65°C (a

236 Chapter 7

few older "55°C rise" transformers were designed to that lower temperature), and an
additional hot spot rise of 10°C or 15°C above that.
Transformers are intended to operate under "normal circumstances" at these
temperatures on a "continuous basis." A traditional standard (and one still used) specifies
that the maximum power rating of the transformer, its "rated capacity," is the level of power
it can deliver in continuous service at that load for 7.42 years before its insulation is
degraded by deterioration driven by heating from losses to 50% of its original tensile
strength. Formulae exist to compute the expected time to 50% degradation for any other
loading pattern, not just a constant load. For example, an example later in this chapter will
show that a 50 MVA transformer serving a typical utility daily load curve that fluctuates up
to a 50 MVA peak each day, and not a constant 50 MVA load, has an expected time to 50%
insulation tensile degradation of 39.3 years, not 7.42 years.
Regardless of whether it is 39.3 years, 7.42 years, or some other value, the calculated
time to 50% degradation is sometimes referred to as the unit's "lifetime." However, this is
not a recommended use of the word "lifetime." The time period can be more properly
thought of as "insulation half-life." One reason is that a unit will not necessarily fail at 50%
insulation strength. In-service lifetime can be and usually is noticeably longer, but will
occasionally be shorter (see Figure 7.1). Individual units will vary greatly in service - some
lasting longer and some lasting shorter times, for a variety of reasons. Experience has
proven to the author that generally:
A calculated lifetime ofX years means that the unit is
very likely to last X years in service, likely to last up to
twice that long, but unlikely to last three times that long.
A proposed industry standard has an expected "normal" lifetime of 2.77 times insulation
half-life (transformers meeting the 7.42 criteria would actually last about 20.6 years before
insulation failure due to deterioration). This proposed standard fits the author's rule (above)
and is probably indicative of lifetime due to insulation failure. However, the reader is
cautioned to keep in mind the failure causes shown in Table 7.1 and the lifetime-failure
model shown in Figure 7.1: a majority of transformers die from other causes, or have at
least some other contributing cause to their demise beyond just insulation tensile-strength
degradation. The "2.77 half-lifes rule" is an upper bound on expected lifetime under the
best of conditions.
Why Ratings?
Regardless, while insulation half-life loading level is not an actual lifetime, it provides a
useful basis for study of many lifetime-related issues. A unit rated 60 MVA under this
standard will be able to handle roughly 20% more loading than a 50 MVA unit over a
similar service lifetime, whatever that might be. A 50 MVA unit serving 40 MVA will last
considerably longer than if it is serving 50 MVA, but not nearly as long as a 60 MVA unit
would serving that same load. This is useful information for utilization decision-making,
although planners must remember that there are considerations, not addressed here, that
also affect lifetime (i.e., the number of through-faults).
How Loading Level Affects "Lifetime"
The example 50 MVA transformer discussed above has an insulation half-life of 7.4 years
when serving a continuous load of 50 MVA. A similar amount of insulation deterioration
would occur over a different period of time if the loading level were different. The expected


half-life of the insulation is expressed by the Arrhenius theory of electrolytic dissociation,
which is

where L is the lifetime in hours and T is the temperature reached by the winding hot spot.
Constants for this equation are given in Table 7.2. These are based on lab tests and have
been determined for both power transformers and distribution transformers. Figure 7.2
shows a plot of half-life versus hot spot temperature for various types of transformers.
Using 110°C as the hot spot (30°C ambient temperature + 65°C core rise + 15°C hot
spot rise) and the KI and K2 values in Table 7.2 in equation 7.1 gives a computed insulation
half-life of
L = ir)(6972-15/(273+110))~13-391 n o^
= 65,020 hours
= 7.4 years
Since the electrical losses heat generated by a transformer is related to loading squared,
and heat losses of the unit are proportional to temperature rise, one can convert equation 7.1
to an equation based on MVA loading, as:
L = 10 (K1/(303+8°(MVA/R))-)-K2 (7 3)

where MVA is the load actually carried by the unit and R is its rated MVA. Using equation
7.3 with various levels of load shows that a 50 MVA unit carrying 50 MVA will result in
63,613 hours insulation half-life, whereas one carrying 52 MVA, a loading of only 4%
higher, will have an expected lifetime of only half that amount. Similarly, operating the unit
at a 4% reduced level, 48 MVA, will give in an expected lifetime of 14.8 years. A load of
46 MVA - an 8% de-rating" from the 50 MVA nameplate rating - gives an insulation half-
life of 30 years, again when serving a constant level of load 8760 hours a year.
Ambient Temperature Matters a Great Deal
One can compute the effect of a lower or higher ambient temperature on expected
insulation half-life or rating that corresponds with a specific lifetime. The examples above
and standard rating/lifetime computations assume an ambient temperature of 30°C (86°F),
a temperature far below that normally associated with summer peak load periods in most
utilities. Computing equations 7.1 and 7.3 with 38°C (100°F) means that: (1) lifetime is cut
by more than half or (2) the load being served must be reduced by more than five percent,
to 47.4 MVA. Similarly, a cooler ambient, for example 0°C in winter, permits a much
higher loading with no loss of expected insulation half-life.

Table 7.2 Values of K, and K, for Equation 7.1 to Compute Insulation Half-Life
Description K, K, Rise
Power Transformer (55°C Insulation) 6972.15 -14.133 95°C
Power Transformer (65°C Insulation) 6972.15 -13.391 110°C
Distribution Transformer (55°C Insulation) 6328.80 -11.968 95°C
Distribution Transformer (65°C Insulation) 6328.80 -11.269 110°C

238 Chapter 7

Distribution Transformer(55°CInsulation)
Power Transformer(65°C Insulation)

100 120 140 160 180 200
Hot Spot Temp(eC)

Figure 7.2 Expected insulation half-life of transformers as a function of hot spot temperature: plots of
equation 7.1, for various types of transformers. Typically transformers are designed for hot spot
temperatures of 95°C (55°C rise types of insulation) or 110°C (65°C rise insulation). Transformer life
drops exponentially as hot spot temperature rises.

Dynamic Loading and Lifetime
All of the above discussion referred to situations where a transformer is loaded to a certain
level, for example 42.28 MVA, and left at that loading permanently. While a few power
transformers in industrial service see such constant loading patterns, very few others do.
Most power transformers, and nearly all in utility service, serve a load that varies
temporally, by time of day, day of week, and season of year. Chapters 2 and 3 discuss load
variation over time and typical loading patterns seen on utility systems. In such cases where
the load varies with time, the situation with respect to temperature build-up and
deterioration rate is not nearly as simple to calculate as it is for a constant loading level.
Figure 7.3 illustrates why. The plots show the loading and peak core temperature of two
different 50 MVA transformers that are "cold," serving no load, up to the time they both
experience a step function in loading to 60 MVA. Both transformers are designed so that,
given enough time, a 50 MVA loading will raise their temperature to 110°C, an 80°C rise
above ambient, their rating. Thus, a 60 MVA loading will generate (60/50)2 as many losses
and eventually raise the temperature of concern in each to about a 115°C rise, an
"overload."
But it takes time to "cook" a large transformer, often a good deal of time. In effect, the
windings with their electrical losses are a large heating element trying to raise the
temperature of a very big assembly (the transformer core) inside a large metal tub (the case)
containing a great mass of oil. This whole assembly weighs many tons, and even a number
of kilowatts of losses take quite awhile to work the temperature up to really high levels.
In the case of these two units used in Figure 7.3's example, both are 50 MVA units with
identical windings generating identical losses when transforming 60 MVA. But one
transformer has a heavier core and a larger case that contains a much larger amount of oil. It
takes longer to heat this more massive cover up to any particular temperature than its does
for its less massive cousin.

120

9 ioo
0>
80
DC
o>
60
I
g. 40
|
" 20

0
4 6 8 10 12
Time (hours)

120

9 100

<2 80
DC

I 60
4-»

2
0)
CL 40
0)
20

0 2 4 6 8 10 12
Time (hours)

Figure 13 Temperature rise for two 50 MVA transformers, both unloaded and "cool" prior to being
loaded to 60 MVA at t = 0. Top, a unit with a time constant of 4-hours. The dotted line shows the
initial rate of temperature rise (see text). Bottom, a similar transformer, but with greater "thermal
inertia" giving it a 6-hour time constant.

From the moment the 60 MVA loading begins, temperature begins creeping upward in
both units. In each, the temperature will asymptotically approach a value of 115°C.
However, in the more massive unit the rate of rise will be slower, since the same amount of
heat has more mass through which it must diffuse. For either unit, and for any other
electrical equipment that is heating to some asymptotic value (or cooling in a similar way)
the temperature reached at any time is given by:
T = Ax(l-l/e c r / C ) ) (7.4)
where A is the asymptotic target
T is the time since the heating began
C is the temperature-rise time constant

240 Chapter 7

120

9 100
a>
.<2 80
IT
Q)
^ 60

a
CD
20

0
4 6 8 10 12
Time (hours)

Figure 7.4 Temperature when the smaller of the two 50 MVA transformers is loaded to 60 MVA
after first being at 50 MVA for a long period of time.

A physical interpretation of this formula is simple: At any instant, the rate of
temperature rise is such that if continued for a period of one time constant, the unit would
reach the asymptotic value. The lighter of the two transformers in this example has a time
constant of 240 minutes, the heavier unit a time constant of 360 minutes. As a result, when
the 60 MVA loading starts, temperature starts rising toward 115°C in both units, at a rate of
.48 and .32 degrees rise per minute respectively for the smaller and larger units. (These
figures correspond simply to 115°C/240 minutes and 115°C/360 minutes.) But the rate of
rise does not stay at that rate. Slightly over two minutes later, the smaller unit has already
reached a 1°C rise, and its rate of increase has now dropped slightly, to 114°C/240 minutes.
When its temperature eventually reaches an 80°C rise, the rate has decreased to only
.15°C/minute (computed as (115 - 80)°C/240 minutes).
Figure 7.3 shows temperature as a function of time for both units. The less weighty of
the two transformers (that with the 4-hour time constant) reaches a 25°C rise by the end of
the first hour, 73°C rise after four hours, and 108°C at the end of twelve hours. It takes 284
minutes (4.7 hours) for it to reach the 80°C rise level of the standard. By contrast, the
heavier unit, with its 6-hour rise time, reaches only a 18°C rise by the end of the first hour,
and takes 426 minutes to reach the 80°C level. During the time discussed they have yet to
reach the 80°C rise limit; these units are not under high degrees of thermal stress. In fact,
the earlier part of those 284- and 426-minute periods causes little deterioration. As the
temperature continues to creep upward deterioration accelerates. If the 60 MVA loading is
continued long enough, temperature will rise to 115°C and the deterioration rate will
stabilize at .004 percent per hour, a rate at which insulation half-life is only 1909 hours.2

2
Beyond the truly dynamic heat-flow factors at work in the discussion above, at really high (more
than two times) loadings, a number of other factors put a transformer at much higher levels of
jeopardy than mere deterioration of insulation would suggest. For example, if the hottest spot
reaches a temperature that boils the transformer oil, the "bubble path" through the oil created as a
result can cause a fault, and catastrophic failure may result).


A more realistic example, in the sense that it occurs much more often than an unloaded
transformer suddenly seeing a load above its rating, is shown in Figure 7.4. Here, the less
massive of the two units in Figure 7.3 was already at stable thermal operation at its full 50
MVA load when load was increased to 60 MVA at time zero. Again, temperature
asymptotically heads toward 115°C. But temperature was at 80°C prior to this time, so
deterioration above the standard's rate (equivalent to a 7.142 year half-life) begins to occur
immediately.
Key points from this example
This introduced two factors every planner should keep in mind when dealing with
equipment lifetime and loading issues. First, dynamic effects can be critical, particularly
when dealing with periods in which temperature rise might not reach critical levels. For
example, multi-hour "overloads" of 20%, as shown above, cause no great deterioration of
either unit.
Second, transformer characteristics vary significantly, often because of simple choices
made in the design of the units. Differences like that shown between the two 50 MVA units
used in the example above - one can tolerate the "overload" for about 5 hours, the other for
over 7 hours without significant deterioration - are common. Much larger differences than
this can exist among units that may have the same "rating."
A third point was not illustrated well by the examples above, but needs to be
emphasized. The Arrhenius equation (7.1) is a steady state equation. Its use in dynamic
situations is only approximate. While it seems logical that a combination of equations 7.1
and 7.4 could be used to calculate temperature rise and from it deterioration as a function of
time during dynamic loading periods such as shown in Figures 7.3 and 7.4, such a
calculation would be only approximate. The numbers obtained above for deterioration were
computed in this way on a minute-by-minute basis, and are accurate only within about 10%.
During periods when the load is changing rapidly, and consequently the core temperature is
changing rapidly, too, any method based only on this approach will tend to underestimate
stress on the transformer. This is particularly true if the asymptote is more than 20°C above
rated maximum rise.
Loss of Life from Emergency Loadings
Suppose that the 60 MVA load in Figure 7.4 is continued for 24 hours. Equation 7.1 can be
used to determine that during 24 hours of stable steady-state operation at 60 MVA (i.e., 24
hours after the temperature reached the asymptote) the unit would lose 24/1909 or about
1.25% of its total insulation half-life. Since the temperature in this example was quite cool
for the first four or so hours of operation, deterioration would be a bit less than this. In fact,
in this example, a minute-by-minute analysis for the 24 hour period using equations 7.1 and
7.3 predicts that the unit would lose about 7/8 percent of its insulation half-life during that
one-day period. This is a bit lower than the actual would be, because of the approximate
nature of the equation when applied to dynamic loading (see above). Such an overload will
most likely cause almost exactly a 1 % loss of life.
The concept of loss of life for brief overloads is an accepted way of studying and
characterizing loading versus lifetime issues for transformers and some other electrical
equipment. In truth, a transformer has no "maximum load." But any load it carries implies a
certain rate of deterioration and stress that will shorten expected lifetime.
Permitted emergency loadings are often determined on the basis of a 1% loss of life. In
the section above, the 60 MVA loading, a 20% overload, resulted in the use of 1% of the
insulation tensile half-life in that single 24-hour period. The specific definition of "life"

242 Chapter 7

involved is the insulation tensile half-life, but unlike other cases, the life to which this can
be applied is actual expected life of the unit. As stated above, the insulation half-life
(equation 7.1) is not a prediction of the expected service life of a unit and often one cannot
determine a basis for comparison. But half-life is assumed to be proportional or
representative of actual life in a limited sense, even if one does not know what that is. So a
1% loss of a calculated life of 7.42 years is 1% loss of the actual life, whatever that may be,
even if one doesn't know what it is.
This is perhaps the cause of the confusion that leads some people to try to interpret
insulation half-life as actual service life. Again, insulation half-life is not a prediction of
actual expected life in service. But a 1% calculated reduction in it is indicative of a 1.5%
loss in actual useful life of the unit, although that is probably not known (or knowable) with
certainty.
Loss of Life Tables
Generally, power equipment in utility service will see really high loadings — those
substantially above its nameplate rating - on very rare occasions which can be characterized
as emergencies. For example, two identical 50 MVA transformers might be installed at a
substation and, following typical industry practice of the late 20th century, loaded so that
both see a peak load of only 40 MVA each. When either is out service for some reason, the
other may have to serve up to 80 MVA, a 60% overload. Operation for about two hours at
this loading will remove more than 1% of its life. But emergency loadings are never
maintained for long. The peak period lasts only a few hours (that shown in Figure 7.4 is
typical). Field crews quickly perform switching to transfer some of the load to nearby
substations. Other crews work on repairs.
It is typical to talk of the "loss of life" of a transformer due to the high loads it serves
during such an emergency. Tables have been worked out by both computation and
experimentation that give the expected loss of life for a particular combination of prior load,

.25 .50 .75 1.0 .25 .50 .75 1.0
Percent Above Nameplate Rating Percent Above Nameplate Rating

Figure 7.5 Loading and time combination that result in the loss of 1 % of transformer life. Left,
typical curve of "emergency service" ratings. For example, the unit can serve four hours at about 40%
overload while losing 1% of expected life. Right, data plotted with a log scale for time. At zero
overload level (its rated level) the unit will serve about 1300 hours with loss of "1% of its rated life"
(1/50 of 7.4 years). (Source: Electric Transmission and Distribution Engineering Reference Book,
Table 10(a)).


(i.e., was the transformer at full rated load prior to the emergency loading?), amount of
emergency loading, and period of that loading. For example, a standard industry
transformer table states that loading an oil-filled power transformer to 200% of its
nameplate rating for only V* hour will shorten its expected lifetime by 1%.3 This would be
an extreme emergency loading situation, in the sense that 200% of nameplate is among the
highest that would ever be contemplated for a transformer. And this is close to the shortest
duration period that would really do any good. More typically, a utility might decide to load
a unit up to 160% of nameplate rating for two hours, a combination that also gives 1% loss
of life. Regardless, that data and others available in industry references are used as guides
on how far to "push" a transformer in service (Figure 7.5).
The "life" referred to in these tables and in the application of such guidelines as
discussed above, is the same "insulation half-life" of the Arrhenius theory, but the concept
can be applied to any expected lifetime: whatever that lifetime is, one forfeits 1% of it to
ride through the emergency.
Cyclical Loadings and Lifetime
For the vast majority of transformers in electric utility service, the load varies over time,
rising and falling as a function of the customer demand level. Figure 7.6 shows the
temperature and loading through a typical day for a 50 MVA transformer in service at a
utility in the Midwestern U.S. Although loading hits 50 MVA for one hour of the day (5-6
PM), the transformer never reaches 80°C and the deterioration rate never reaches a rate
equal to a 7.42-year half-life.
In fact, for much of the 24-hour period, the temperature rise is quite low and the loss of
lifetime is insignificant. In the early morning hours the deterioration rate is so low that if
continued, it would give an insulation half-life of over 500 years. Averaged over the entire
day, the deterioration rate corresponds to an insulation half-life of 39.3 years.
Here, then, is the "secret" of utility transformer service longevity. At a constant load of
50 MVA, the transformer has a calculated insulation half-life of 7.42 years. But when
serving the cyclical load with a peak of 50 MVA, as shown in Figure 7.6, it has over four
times that calculated "lifetime." That lifetime can be interpreted to mean that this unit is
very likely to last 39.3 years in such service, and likely to last far longer than that.
Qualitatively then, a transformer rating determined using the "7.42 year lifetime" standard
is similar to the peak load rating an electric utility wants in order to "nearly guarantee" a
satisfactorily long life, 40 years, in service.
Interestingly, by varying the values used in the computations, one can determine that if
this unit were asked to serve this same load curve shape, but with a higher peak load of 55
MVA (10% higher), then its average daily rate of deterioration (serving this same cyclical
load every day) would be the same as when serving a constant 50 MVA load. Insulation
half-life would be 7.42 years. Again, it would likely last more than 7.4 years in service, but
probably not 22.2 years.
The formulae and concepts used above can be applied to compute insulation half-life as
a function of peak load level for any load curve shape. Figure 7.7 shows half-life versus
peak load level for this load curve shape (dotted line). Also shown (the solid line) is a more
realistic situation: this same daily load curve shape but the peak value is used for only the

3
Electrical Transmission and Distribution Reference Book, fifth edition, page 114, Table 10(a) -
Permissible Short-time Transformer Loading, Based on Reduced Life Expectancy.

Chapter 7

m o m
T- O
o O O
o o o

Oo-

o o o o
00 CD -^ CM

^ 12

CM
CO
CM
CM
CM

CM
O
CM
CO
O5 «
o

o 3
_,. 0> CO <fi O
^ .c CD
•«- *" tn J1
CO O
"t; w
o T3 >-.

CM O

o

O)
•c .S
CO
•a 3 C
H —
cc 03 O
Is* O

CD
> ex
m "«? f~-
c «n

co
Cd ""*
CM q -a
J .2
c <B
_J , J— ^ £•
« ^
O
in
O
^t 0
CO
0
CJ
0
T-
II
b cs


1,000-

100

10
-J o>
•o o
fl) ^,
S
o
a.

.01
40 60 80 100 120 140
Annual Peak Load as 100% of Nameplate

Figure 7.7 Solid line - Insulation half-life versus peak load as a percent of nameplate, for the less
massive of the two 50 MVA transformers used in earlier examples, when serving an annual load curve
(8760 hours) with a typical utility load factor (63%), where the peak day (load curve in Figure 7.4) is
reached on only 37 days of the year, other days being lighter loadings. Serving a peak load of
nameplate rating gives an insulation half-life of 72 years. Dotted line - half-life if the peak load level
is reached every day.

peak 37 (10%) days in the year, with the rest of the days having lower peaks, matching a
typical utility usage pattern of 63% annual load factor. Just this reduction in the number of
peak days in the year which hit the peak loading of 50 MVA results in an increase in
insulation half-life to about 75 years.
The Crux of Wound-Equipment Lifetime and Loading Issues
Regardless, for transformers and other wound electrical equipment, there is no firm "rating"
above which a unit cannot be loaded. Any loading level has associated with it a period over
which the unit is expected to retain sufficient strength (or materials, or dielectric strength, or
whatever), based on some (or several) standards, which are specific definitions of how one
relates loading to lifetime, and what those terms mean. Often, this period of time until
deterioration has reached a certain defined point is referred to as "lifetime," but it is critical
for planners and engineers to keep in mind that it is not an actual prediction of "service
life."
The owner-operator of electrical equipment faces a choice with respect to loading
equipment. If he or she picks a conservative loading level, thermally driven deterioration
will be slow, and the unit could last for decades. During that time, long exposure to damage
by other causes - through-faults, weather, vandalism - could be the primary cause of
eventual failure.
Alternately, the unit can be used to serve a much higher load. If so applied, deterioration
will be faster, expected service lifetime of materials will be lower, and the unit can not be
expected to last as long. Exposure to through-faults, damage, and weather will be slighter,
and perhaps not as significant a consideration over the shorter period. And, used in this
way, the unit will do more work - it is serving a higher load so its utilization, whether
viewed on a per unit basis or per dollar invested basis, is higher.

246 Chapter 7

The optimal decision with regard to loading is thus a balancing act. Higher loads
accelerate lifetime but mean the unit provides more value now. Lower loads preserve the
unit for more future years of service, but mean that it provides slightly less value each year,
now and in the future.
7.3 AGING, DETERIORATION, AND DAMAGE
All electrical equipment "ages," in the sense that it gets older every day. But often
chronological age, the time since manufacture, is of little consequence. Service age - time
in electrical service - is the key factor and definition of "age" for most electrical equipment.
Cables, conductor, transformers, and many other kinds of equipment have very long "shelf
lives," so date since manufacture is not normally an issue with respect to age.
"Age" means time in service. During every year of service the equipment is exposed to
weather, sunlight, and potential damage from animals, plants, stray bullets, and vandalism,
etc. Furthermore, and often most important, various materials and components in
transformers, breakers and other equipment will deteriorate as a consequence of electrical
service. Deterioration means the (usually) gradual but steady reduction of some necessary
physical or electrical characteristic(s) of the equipment, be it dielectric (insulation) strength,
mechanical strength, current carrying capability, or some other needed quality of
performance for the unit to do its job.
At some point, the degradation in a needed characteristic leads to failure. This failure
creates an inability of the equipment to perform its role in the system. That is what "failure"
means. A remote capacitor switch may fail to open or close on command. It has failed. It is
not performing its function.
But usually, failure of an electrical component has a variety of other consequences. Of
great concern to the electric utility, it may cause an interruption of customer service, either
directly (a line falls cutting the path of power flow) or indirectly (a line faults and a fuse
cuts the path of power flow).
The failure might end the life of the equipment involved, because the unit is not
repairable or is not worth repairing. The unit (or what is left of it) ceases to be an asset and
might now be a liability. Many power engineers think of equipment failure only in this
manner - that failure means the end of life and the "destruction" of the unit or its value.
That is not the case, and failure can only be properly studied and managed by understanding
that. Failure does not necessarily mean that the equipment's life has ended, or even that it
causes an interruption, although that is often the case. A failure means specifically that the
unit has ceased to perform its function, whatever that was. That said, throughout all
discussion of lifetime and equipment here and in Chapter 8, the word "failure" alone will
mean an event that ends the unit's life.
Finally, a failure might cause a catastrophic event - perhaps not just ending the unit's
life, but starting a fire that destroys other equipment and property around it. Such cases are
rare but usually spectacular and memorable, and expensive. Table 7.3 summarizes these
types of failures by their consequences. In each type, the unit "failed" but in a manner that
had different consequences. Types 1 and 2 include effects only on the equipment itself (it
doesn't work, it has no value). Types 3 and 4 include external effects (customers are
affected, facilities are affected).
These failure types are not completely exclusive. Clearly, type 2 implies type 1 - a unit
that has failed so it has no remaining lifetime is not doing its job. However, not all failures
cause interruptions (consequence 3). Many failures do not lead to interruptions (example:
failures of tap changers on transformers). It is possible for a type 3 failure to occur without
a direct functional failure. For example, a remote switch can open simply through a false or
mistaken signal being sent to it. There has been a "failure" (actually, a "misoperation," see


Table 7.3 Types of Failure Consequences - One or More or All May Occur Simultaneously
Number Type Consequence Example Cure
1 Functional Equipment ceases Lightning arrestor fails, Repair or ignore it and live
to do its job. without the function.
2 Life ending Equipment ceases Lateral disconnect faults. Replace it or remove and do with-
to have any value. out if replacement unnecessary.
3 Interruption Equipment ceases Overhead line falls Restore service by switching
to do its job. to ground. if possible, then repair/replace.
4 Catastrophic Equipment fails, Substation transformer Restore service through switching.
causing fire and explodes during fault, Clean up the mess, and repair
damage to facilities, causing oil fire. or replace or do without.

Table 4.1) and an interruption has resulted, but the equipment is functional and has not
"died." Type 4 failures, although spectacular, may not cause outages. A transformer fire at
a substation might result in the unit itself and nearby buswork being destroyed, but
automatic equipment might isolate the entire portion of the substation and switch the load
instantly to another bus, so that no customer service interruption occurs.
For many types of equipment, the first three types of failure listed, functional, life-
ending, and interruption, always occur simultaneously. A clamp on an overhead line that
breaks (1) ceases to do its job, (2) ends its useful life (since clamps cannot be repaired) and,
(3) causes an interruption of service. Regardless, it is always useful and occasionally quite
important to keep these four types of failure consequences in mind when studying
equipment condition, aging effects, system reliability, and maintenance management.
Lifetime and "Final Failure"
At some point, a unit of equipment reaches a point where it has no useful remaining life.
Almost invariably, electrical equipment in utility service reaches this point only when it
suffers a life-ending or catastrophic failure. Most electric utilities in the Americas use T&D
equipment on a "run to failure" basis. Equipment is very rarely withdrawn from service
because it is old and has little remaining life left. (By contrast, utilities in Europe and Japan
sometimes remove major equipment from service at the end of periods of about 40 years).
All four failure-consequence types listed in Table 7.3 are of interest to a utility. However,
with respect to planners, types 2 and 3 predominate. Type 2 relates directly to equipment
lifetime - the length of time the equipment will return a benefit (perform a function) in
return for the utility's investment in it. That is a key factor in optimizing return on
investment and improving utilization of both investment and equity. Type 3 failures - those
that cause interruptions - are clearly an issue in managing reliability of customer service.
Lifetime ends when a unit is no longer worth retaining, either because it cannot be
repaired, because it is not worth repairing, or because it does not perform well enough to
meet current needs. This last issue is not failure related, and applies mostly to computer and
automation equipment, due to obsolescence, which will be discussed later in this chapter.
Deterioration and Damage
From the moment a unit enters service, it begins to deteriorate in a number of ways, some
or all of which are critical to its function. For example, paint begins to weather due to wind,
rain, temperature extremes, and sunlight. Insulation loses mechanical strength due to the

248 Chapter 7

high temperatures caused by electric losses. Metal fatigues and tight assemblies loosen
slightly due to cyclical expansion and weather-induced stresses, and perhaps due to
magnetic transients caused by fault currents. Wood poles and assemblies rot. Metal cases
and supports corrode and weaken. Moving parts wear and critical tolerances widen. Contact
surfaces pit and burn. Insulation "trees" or otherwise loses dielectric strength.
Eventually, if left in service long enough and if the deterioration is not corrected, the
unit will lose enough of some vitally needed characteristic that it will fail. Any or all of the
types of failure consequences listed in Table 7.3 can result. At some point, the deterioration
will cause a life-ending failure.
Deterioration can follow any number of trends, illustrated in Figure 7.8. Various types
of physical processes involved create different types of cumulative trends. Some types of
deterioration accelerate over time, so that the unit's strength "drops of a cliff over a short
period of years, after many good years of service (trend B in Figure 7.8). This is often the
case in the breakdown of voltage-withstand strength in some equipment - as it weakens the
process of deterioration is accelerated. Other characteristics degrade more slowly as
deterioration takes place (trend A): essentially the rate of deterioration is proportional to the
amount of the characteristic left. This is the process that applies to the loss of life of
insulation strength in a transformer. Still other components and materials, such as paint
wearing off a case and some types of corrosion, deteriorate at an essentially constant rate
(C).
In effect, the various components in a transformer or a breaker are in a "deterioration
race" to be the first failure to end the unit's life, as illustrated in the Figure 7.9. There is
value in keeping this "deterioration race" concept in mind. Often planners or operators
focus only on a single characteristic (e.g., core insulation strength of a transformer)
neglecting the other ways the unit could fail (bushings could fail, radiators could rupture, a
loose core could succumb to a violent through-fault shock).
Worn-Out Equipment
It is worth considering that a good design for a complicated unit will invariably include
deterioration rates for its various components such that all reach failure at about the same
time. If this doesn't make sense to the reader, he or she should consider that it would hardly
make economic sense to put a "300 year" case on a transformer that had a 20-year core.
Better to take some cost out of the case and apply it to a stronger core and to similarly
engineer trade among all components until no one component is expected to cause a "pre-
mature" failure.
It is not possible to predict failures and deterioration rates so precisely that equipment
engineers can precisely manage this process. But to the extent it can be engineered, and
based on experience and feedback over time from utilities interested in cost reduction,
almost all electrical equipment has had its designs refined in this way. There are no
particularly weak or rapidly deteriorating parts, and nothing more than needed is spent on
any parts to the extent they are "unnecessarily strong."4

4
Auto manufacturers have applied this principle well. Most modern cars are designed so that neither
engine, transmission, suspension, nor any other major component is the weakest link, expected to be
the source of early or frequent warranty problems. By design, all these components have roughly
similar service lifetimes, well beyond any warranty period or even the period that most owners really
expect them to run "trouble-free."


O
•a
0)
•o
o
o>

10 20 30 40 50 60
Time - years
Figure 7.8 Deterioration of a necessary characteristic in an electrical component can follow any of
several types of trends. Line (A) shows an exponentially decreasing rate of deterioration, (B) an
exponentially increasing rate of deterioration, and (C) a linear deterioration, typical of certain types of
corrosion and rot.

a
O

o
0)
3
W *-
s*
5 o
Failure threshold

Time

Figure 7.9 The various subsystems and components in a unit of equipment can be considered to be in
a "race" to be the cause of eventual failure. A utility can extend the life of equipment and avoid an
incipient failure by servicing the most-deteriorated components to restore them. Here, dotted lines
show the effect of servicing two components that tests have determined are nearest to failure among
six different components of one unit. This service will very likely extend remaining life, but planners
must bear in mind that the rest of the unit is still "aged" and that it has not been restored to "as new"
condition.

250 Chapter 7

This principle of design has an interesting consequence. Equipment that has deteriorated
to the point that one significant element of it is near failure may be just about as near failure
in every other characteristic, too. Simply put, it is worn out. Thus, equipment in which the
most-deteriorated components need to be serviced or repaired to keep it in service may not
be worth servicing, simply because some other failure will shortly occur. A unit that has
been serviced for only the most-critical deterioration problem may be of only limited value
when looked at from an asset management perspective, and further service work may not be
cost justifiable. This will be discussed in much more detail in Chapters 8 and 28.
Damage from External Causes
Deterioration of materials and equipment is the primary lifetime factor for electrical
equipment. However, many electrical equipment failures are caused by damage from
external sources, as opposed to the action of environmental factors (sunlight, wind, rain) or
electrical usage. Damage can also be caused by through-faults, which can be interpreted as
either an external or internal cause.
Regardless, damage tends to be the result of a physical event such as a bullet hitting an
insulator, a squirrel or bird faulting an insulator, a tree falling on a line, lightning striking a
line and vaporizing a conductor, the root of a tree growing into a pad-mounted switchbox,
an automobile hitting a pole. The damage caused by the event may cause any of the types of
failure covered in Table 7.3. Most often, a damage-related failure results in all of 1,2, and 3
types of failure.
Damage is almost always physical, as opposed to material. For example, a pole hit by a
car may have a chunk knocked out of it, be split up the center, and have attachments near
the top loosened. A line hit by lightning may have strands literally vaporized or melted
away. A tree branch falling on a line might twist and weaken a nearby cross arm. By
contrast, material damage caused by deterioration seldom occurs due to such physical
events. Rare exceptions do occur and usually involve water as a contributing factor: a leak
permitting water into the oil of a transformer will accelerate deterioration of the insulation.

Failure threshold

TO 20 30 40 5
Years in Service

Figure 7.10 Here, the mechanical strength of a pole begins to deteriorate on the day it is installed.
After a number of years in service it is hit by a car, slightly damaging it and reducing its strength
slightly. Years later it is hit much harder, doing much damage and slightly accelerating its rate of
deterioration due to rot (a common consequence of damage to a pole). It fails several years later.


Table 7.4 Categories of Equipment Aging Impact and Their Meaning
Category Meaning and Impact
1. Chronological age (CA) Age since construction. Certain materials deteriorate over time
due to natural causes most directly associated with chronological
age (paint chalking and corrosion, hardening of rubber and
gasket materials, etc.).
2. Cumulative service stress (CSS) The cumulative effect of the time the unit has been energized or in
use and the load (mechanical, electrical) it has served in that time.
3. Abnormal event stress (AES) The cumulative impact of severe events generally not considered
as "normal service." This includes through-faults for
transformers, storm and auto-accident stress for poles, etc.
4. Technical obsolescence (TO) Digital and data communications equipment, as well as other "hi-
tech" equipment, can become "too old" by virtue of not being
compatible with new systems and equipment.

One vexing problem for power delivery utilities is that equipment damage may not
result in an immediate failure, but instead exacerbate the deterioration of the equipment, in
effect greatly shortening its life but remaining undetected. For example, an automobile may
hit a pole but not actually knock it over or break it. However, the damage may be such that
the pole will fall in the next storm, or simply succumb to rot years earlier than it otherwise
would (Figure 7.10).
This has a partly positive aspect: the pole did not fall over and cause an interruption
when the automobile hit the pole. Service was maintained. However, the concern a utility
has for this situation is two-fold. First, a good deal of such undetected damage may exist
unless inspection is thorough. Even then it may not detect all or most of it. Failures due to
such damage tend to occur during storms. The utility may not be notified that the pole was
hit and should be inspected. Poles, lines, and other equipment that were damaged but not
destroyed often "come down" all at once, in the next storm, adding to the burden the utility
faces due to all the failures caused by damage from that storm.
Damage can be considered as an effect of aging. Time in service creates exposure to
damage. As a result, damage is an inevitable consequence of equipment being in service.
Over time it is unavoidable, although its consequences can be minimized and its effects
mitigated by proper selection of equipment types, materials, and design.
Categories of Aging and Deterioration
Table 7.4 summarizes four categories of aging impact on electrical equipment. The first
three cause deterioration that inevitably leads to failure of one type or another. The fourth
category, technical obsolescence, should be considered an aging factor in many equipment
decisions. Spare parts and factory engineering support may not be available for older
equipment. This leads to higher maintenance costs (the utility may have to produce
replacement parts in its own machine shop) and poorer operating reliability (a lack of
factory service support degrades the quality of diagnosis and inspection that a utility can
obtain for problem equipment), creating a situation very similar to that caused by
deterioration.
Table 7.5 lists six types of deterioration created by aging and indicates the most typical
primary (X) and contributing (x) causes from among the four given in Table 7.4. Generally,
the terms "aged," old" and "aging" are used to refer to some non-specific combination of all

252 Chapter 7

four causes, with the understanding that the equipment is not as useful or dependable as
similar equipment of lesser age.
Chronological age
Certain materials such as paint, paper and fabrics, rubber and synthetic gaskets and seals,
and insulation deteriorate over time. Time since manufacture alone is the direct cause of
this type of deterioration. Thus, seals and gaskets begin to harden, and paint to chalk, as
soon as a transformer or breaker is built, progressing at much the same rate regardless of
whether the unit is installed and in service or kept in inventory in an equipment yard. A
creosote pole will dry out and begin the very slow process of chemical breakdown of its
preservative whether or not it is stored in an equipment yard or put in the ground. It will,
however, begin to rot at the ground line only after being put in the ground particularly if
the soil around it is moist and of the right pH. As the pole example in the last sentence
illustrates, very often the rate of deterioration over time is exacerbated or mitigated by
ambient conditions.
Transformers, poles, and cables stored outside and subjected to summer heat, winter
cold, direct sunlight, and ambient moisture will deteriorate faster than if kept in sealed,
temperature-controlled, and dehumidified warehouses. Deterioration due to chronological
aging can be cut to nearly nothing in some cases if the equipment is keep in a controlled
environment. Generally, "chronological aging" rates and average lifetimes assume that
the equipment is installed in typical ambient service conditions (i.e., poles and
transformers are kept outside whether in service or not).
In many cases, the deterioration that would occur due to time alone is accelerated by the
stresses caused by putting the equipment in service. Hardening of gaskets and seals and
deterioration of paper insulation will occur at a faster rate when a transformer is operating
(heat generated by electrical losses will accelerate the deterioration processes).

Table 7.5 Types of Deterioration or Damage Caused by Aging
Type of Caused By
Deterioration CA CSS AES TO Comments
Corrosion X Chemical decomposition, or combination with
oxygen or other ambient elements, until the
material loses its required mechanical or
electrical strengths or qualities.
Dielectric loss x X Various deterioration mechanisms (treeing,
contamination) lead to the loss of electrical
withstand strength.
Shrinkage/hardening Paper, rubber, and synthetic gaskets and seals
harden and or shrink with age, losing their ability
to keep out moisture or contain pressure.
Wear X X Mechanical components lose tolerance and bind,
or do not hold with the same bond they once did.
Moisture retention X Water is gradually absorbed into a material,
degrading its mechanical or electric strength.
Performance compatibility X Technical obsolescence leads to inability to
perform to modern levels or to interact well with
other equipment in the system.


In some cases however (i.e., paint chalking and some types of corrosion) the rate of
deterioration does not depend on whether the unit is in service. And in a few situations,
putting a unit in service will reduce the rate of chronological deterioration. For example,
energizing a power transformer being kept for standby service creates a low level of heating
(due to no-load losses) that "cooks" any moisture build-up out of the unit, slowing that rate
noticeably.
Cumulative service stress
Certain types of deterioration occur only, or primarily, due to the use or operation of the
unit for its designed purpose, and are proportional to the time or the cumulative level of use
in service, not chronological time. There are three primary types of processes involved:
1. Electromagnetic field stress established when a cable or unit of equipment is
placed in service can lead to degradation of dielectric strength and in some cases
promote corrosion. Essentially, applying voltage to the device produces a stress
that will eventually break down insulation and accelerate some chemical
processes (particularly if the unit is not grounded well or cathodically protected).
This type of stress depends only on the voltage. It does not depend on the amount
of current (power) the unit is handling.
Every unit of electrical equipment - transformer, cable, bell insulator - is
always designed to withstand a nominal voltage, that which it can withstand
safely and dependably over a long period of time (decades). But given enough
time, the electromagnetic stress from that voltage will lead to various types of
insulation breakdown, which can include treeing, attraction of impurities that
build flashover paths, etc. Eventually, over a century (post insulators) or only a
period of several decades (UG cable) the insulation will fail for this reason alone.
In most cases the deterioration rate due to this cause is very voltage sensitive.
Raising voltage even slightly will increase stress a good deal. Thus, higher
voltages due to operating the unit at the high end of its nominal range, or to
misoperation, switching surges, or lightning, can accelerate this deterioration,
sometimes quite dramatically. Similarly, lowering the voltage on equipment that
is suspected of being weak (old cables) to the low end of the permissible scale
(e.g., .95 PU) can often reduce stress, deterioration, and failure rates significantly.
And ,of course, voltage that reaches too high a level - above the tolerable range -
leads to immediate failure.
2. Wear. Parts that move against one another, in devices such as tap changers,
capacitor switches, circuit breakers, load break switches, etc., wear. The
movement gradually erodes material in the moving junction between parts,
loosening tolerances and often scouring smooth bearing surfaces. Such wear can
lead to binding of moving parts and failure to operate or operation slower than
necessary (breakers).
In addition, "wear" can be applied to a type of deterioration that occurs in
parts that are not technically moving against one another. The most common
form of this is in overhead conductors and their associated splices, terminations,
and support brackets. Conductor "moves" in the sense that it sways in the wind,
and expands and shrinks, and sags more or less, with temperature (and loading).
Over decades, this movement can lead to loosening of the bond between clamps
and conductor, or within splices or termination fittings, and even to cracking of
the conductor material itself. Eventually something breaks and the line falls.

254 Chapter 7

This gradual deterioration is accelerated greatly by aeolian vibration. Wind
blowing across the conductor sets in motion a resonant vibration, much like
blowing on a harp string might cause it to hum. This vibration stresses the bonds
in splices, brackets, and clamps and can fatigue the metal strands of the
conductor, over a period of only months in extreme cases. Various types of
vibration dampers are available to fit to conductors to stop such vibration, but
these are usually fitted only to known or suspected cases of heavy vibration. Most
overhead spans experience a slight amount of this vibration, and over decades it
leads to a degradation of mechanical strength in the conductor and its fittings.
3. Heat stress. Higher temperatures accelerate many of the physical and chemical
mechanisms involved in materials deterioration, to the point that heat can be
considered electrical equipment's worst enemy. In electrical equipment like
transformers, regulators, and motors, heat is generated by electrical losses in the
equipment. The higher temperatures created as a result cause deterioration of
materials in insulation, gaskets, seals, the transformer oil itself, and in some
cases, the metal in conductors and/or mechanical components. Similarly heat
affects all electric equipment, "aging" or ruining the functionality of various
materials or their components. Parts that should be soft and expanded, so they
seal well, shrink. Parts that should insulate lose dielectric strength. Metal
expands, binding and wearing, or at very high temperatures anneals so that it
becomes brittle and of low strength - essentially a different type of alloy.
In almost all cases, the rate of deterioration of a component or material is a
function of the temperature, generally increasing exponentially with temperature
up to some maximum-tolerable temperature beyond which the material fails
immediately. The relationship is exponential - a device might be able to operate
for decades at 80°C, for several years at 90°C, for a week at 100°C, and for only
a few hours at 110°C.
In addition, temperatures that reach certain levels can cause internal physical
changes in some types of electrical equipment, changes that promote
misoperation. For example, in a power transformer, extreme heating of the
windings (due to losses from very high loadings) can cause hot spots on the core
that are hot enough to create gassing or boiling of the oil. The gas introduced into
the oil by the bubbling contaminates the oil's ability to act as an insulating
medium, and a "bubble path" can provide a route for flashover, leading to
immediate and catastrophic failure.
In mechanical devices such as breakers, load tap changers, and motors, high
temperatures can cause swelling of mechanical parts, leading to binding (bearing
tolerances are reduced to the point that the device will not operate) and
misoperation, and/or high rates of wear. In overhead conductor, high enough
temperatures (typically caused when losses generate heat sufficient to create a
100°C rise above ambient) will anneal the conductor material. As a result it
hardens, becomes brittle, and loses its elasticity and mechanical strength.
Vibration from wind and the natural expansion of heating and cooling from
diurnal temperature variations then quickly lead to minute cracking and
mechanical failure and the conductor breaks and falls to the ground.
The deterioration caused by high temperature is cumulative - i.e., a period of
high temperature will cause deterioration that will not be "undone" when
temperatures fall. A device will suffer gradual deterioration of its components,
equal to the cumulative stress of high temperature over its service life to date.


Since heat (the cause of high temperatures) is due to load (the load current causes
losses that create the heat), power engineers often associate deterioration rate
directly with load: higher levels of load lead directly to higher rates of
deterioration. The term "loss of life" refers to the rate of equipment lifetime loss
associated with a particular loading (heat and temperature) level for the
equipment.
Many of the chronological aging processes discussed earlier are accelerated if the device
sustains high temperatures for long periods of time (as is common for many electrical
devices). Thus, chronological aging and cumulative service stress is not, in practice,
completely independent.
Abnormal event stress
Over its service life, any particular unit of power system equipment is likely to see a
number of "events" which lie outside the normal conditions expected in service. These
include electrical events such as through-faults, switching surges, and lightning strikes
and/or harsh mechanical events such as automobiles striking the device (i.e., a pole or pad
mounted device), high ice loadings from freak winter storms that mechanically stress
overhead conductors, and similar situations. Although these are the inevitable result of a
long time in service, these events tend to be rare, unpredictable, and are thus treated as
"abnormal."
Auto accidents and storm damage are a prime cause of wood pole and metal
structure failure, and somewhat unavoidable given that these structures are
often placed on public easements that parallel roadways. However, most of
these events are reported (many result in immediate failure of the pole or
equipment at the top of it) so the utility is at least aware of the event and can
inspect and repair any damage done.
Ice loadings are another stressful mechanical factor, but one that is more
difficult to track. During severe winter weather, ice will accumulate on
overhead conductors. Given extreme conditions, several inches of ice can
accumulate, weighing many hundreds of pounds per span. The cumulative
weight of conductor and ice can exceed the strength of the line, and it will
part, falling to the ground. However, in the vast majority of winter storms,
icing does not lead to failure - the weight is tolerated without any damage or
deterioration.5 When the ice melts after the storm, the conductor is, if not
"good as new," as good as it was before the storm.
But in between these two situations — outright failure on one hand and
no damage done on the other — is a narrow range of "damage-done" cases that
are particularly vexing to utilities. The ice loading on a particular span of
overhead conductor might reach a level that does not cause immediate failure,
but is sufficient to stretch the conductor to its elastic limits, or over-stress
clamps, splices, etc. This leads to accelerated deterioration and failure soon

5
Conductors, particularly those with steel reinforcement (ACSR - Aluminum Clad Steel Reinforced
Conductor) are designed to have high mechanical strength to deal with periodic severe ice loadings.
In some areas of the world (e.g., parts of Saskatchewan) the distribution utility uses an entirely steel
wire in its OH distribution, sacrificing the lower resistance of aluminum for the superior ice loading
capability of pure steel wire.

256 Chapter 7

afterward. The utility has no way of knowing if any portion of its overhead
lines were so affected, and if so, which portions.
Through-faults are the most stressful event routinely seen by many
transformers from the standpoint of gradual deterioration leading to failure. A
device downstream of the transformer experiences a fault, and for a brief time
until a circuit breaker clears the fault, the transformer sees current flow
through it that is from three to fifty times normal maximum. Heating caused
by this high current is usually not the major impacting process - in most cases
the fault current lasts but a few cycles and little heat energy is generated.
Instead, the most common damaging effect of through-faults is the
magnetic field created by the very high current levels and its shock impact on
internal components. Transformers (and motors and voltage regulators as
well) are designed so that the current flow through them causes an intense
magnetic field - required for their operation. But a through-fault multiplies
that field strength in direct proportion to its greater current level, creating a
tremendous magnetic field and compressing or pulling apart nearby
components, etc.
The magnetic force from a severe fault can create tremendous force. If
high enough, this force can literally rip connections lose inside the
transformer, leading to immediate and perhaps catastrophic failure. But
typically, in a well-planned power system, fault currents are limited by design
and operation to levels not so severe that they lead to such problems. Still, the
mechanical impulse from a "tolerable" through-fault will make a large power
transformer, weighing many tons, shake and "ring" as if it had been dropped
several feet and has an overall effect on it similar to such abuse. The
cumulative impact of a number of such severe shocks over a transformer's
lifetime can be a significant loosening of the core stack, and stretching and
twisting of windings and connections - a general weakening of the
mechanical integrity of the unit.
The cumulative impact of through-faults is thought by many power
equipment experts to be the leading cause of power transformer failure. There
are recorded cases where transformers (and motors and voltage regulators)
fail when a through-fault occurs, generally when such faults occur during
peak loading conditions (when the transformer is already quite hot from high
loadings and thus under a good deal of stress). Previous events have
weakened the overall integrity of the unit until this one last through-fault is
sufficient to cause failure.
However, more often than not, failure does not occur during or
immediately after a through-fault. Given the probability of occurrence (peak
loads occur less than 5% of the time, most through-faults happen during off
peak times or when conditions in the transformer are of low stress level. As a
result, the unit tolerates the fault, but the damage is done, and the unit
subsequently fails at some later time - perhaps hours, days, or even weeks
afterward when it is exposed to peak load stresses.
Few utilities have dependable, easy to access records on the through-faults
experienced by major transformers during their lifetimes. As a result, one of
the most useful records of cumulative stress needed to determine present
condition is unavailable to engineers and planners trying to deal with aging
equipment and the estimation of its remaining lifetime.


Lightning strikes are a severe-stress event that often leads to immediate
failure of a device, almost a certainty in cases where the device is struck
directly. Lightning is a pure current source, with the current flow being from
ten to one hundred times the normal fault levels seen in power systems. A
lightning strike can cause an immense "through-fault-like" failure or lead to a
voltage flashover that causes other problems.
Failure modes due to lightning strikes are complex, and not completely
understood, but failure occurs mainly due to heat. Although very brief, the
incredible magnitude of the lightning current, often over 500,000 amps,
creates heat - trees hit by lightning explode because the water in them is
instantly vaporized. Similar impacts occur in electrical devices. In other cases
the magnetic shock of high current does the damage.
There is considerable evidence that a good deal of damage can be done to
electric equipment by indirect lightning strikes. A lightning strike to the
ground near a substation can cause high voltages and/or erosion of grounds. A
strike in the ground may also travel to a nearby UG cable or grounding rod,
burning it badly.
Lightning strikes to overhead lines often cause immediate failure, but can
instead cause a flashover which, due to breaker operation and an interruption,
leaves no outaged equipment. However, the equipment may be damaged to
the extent that it will fail in the next few weeks or months. In some utility
systems, there is a statistically significant increase in service transformer and
line failures after the annual "lightning season."
Abnormal stress events can occur from time to time in the life of any electrical
equipment. In some cases, the severity of one event leads to immediate failure, but often it
is the cumulative impact of numerous events, over years or decades, that gradually weakens
the device. Sometimes these events cause deterioration similar to that caused by time or
service, but most often the degradation in capability is different - through-faults loosen
transformer cores and twist internal fittings in a way that no amount of normal service does.
Technical obsolescence
Most types of equipment used in power systems are from "mature" technologies. Although
progress continues to be made in the design of transformers, breakers, and so forth, these
devices have existed as commercial equipment for over a century, and the rate of
improvement in their design and performance is incremental, not revolutionary. As a result,
such equipment can be installed in a utility system and be expected to "do its job" over its
physical lifetime - obsolescence by newer equipment is not a significant issue. A forty-
year-old transformer in good condition may not be quite as efficient, or have as low
maintenance costs as the best new unit, but the incremental difference is small and not
nearly enough to justify replacement with a newer unit. However, in most cases, if a unit is
left in service for a very long time, the issue of spare parts will become significant. Some
utilities have some circuit breakers that have been in service 50 years or more. Replacement
parts and fittings are no longer manufactured for the units, and have to be customized in
machine shops or scavenged from salvaged equipment.
But a noticeable segment of the equipment used in utility power systems does suffer
from technical obsolescence. The most often cited cases are digital equipment such as
computerized control systems, data communications and similar "hi-tech" equipment
associated with automation and remote monitoring and control. Typically anything of this
nature is eclipsed in performance within three to five years, often providing a utility with a

258 Chapter 7

new level of performance that it may wish to purchase. The older equipment still does its
job, but is vastly outperformed by newer equipment. Economic evaluation of the benefits of
the new systems can be done to determine if the improvement in performance justifies
replacement. Often it does.
But beyond performance issues alone, there are other factors that mitigate against this
equipment remaining in service for decades. First, since in some cases spare parts are made
for this type of equipment for only a decade or so after manufacture, there is no viable
prospect of leaving the equipment in place for many decades. Second, even if parts are
available, newer equipment requires different test and maintenance procedures. Qualified
personnel to make inspections and repairs become more difficult (and expensive) to find.
While digital equipment used in control and automation is the most obvious case of
"technical obsolescence," there is a much more dramatic and significant case that is causing
perhaps the most contentious issue facing the power industry as it moves toward full de-
regulation. The issue is a stranded asset, and the cause is technical obsolescence of
generating plants. Many utilities own generating units that are fifteen to twenty years old.
These units are not fully depreciated - when the units were built, the utilities borrowed
money to pay the expenses, financed over thirty year periods, and they still owe significant
amounts on those loans.
But the technology of generator design has improved considerably since those units
were built. Newer generators can significantly outperform those older units, producing
power for as much as a penny/kWh less and requiring less frequent and costly O&M and
fewer operators. In a de-regulated market, merchant generators can buy these units and can
compete "unfairly" with the utilities' older units. Thus, these older units are "stranded" by
the shift to de-regulation - the utility was promised cost recovery when they were
purchased, but now faces competition due to technical obsolescence and a drop in the sales
that were expected to provide revenues to pay off the loans.
In the T&D arena, technical obsolescence will continue to be an important factor in
control, automation, and metering areas. In addition, a high rate of technical progress will
continue to impact power system equipment areas such as distributed generation, power
electronics/power quality equipment, and home and building automation and control
systems.
"Old" depends on the expected or design lifetime
Electrical equipment can be designed to withstand any degree of deterioration, by building
it from a higher grade of materials that assures longer life (slowing deterioration rates or in
some cases giving immunity from deterioration). Heavier material (e.g., thicker paint) can
be used to allow more deterioration to occur before failure occurs or because harsher
conditions are expected. Additives can be included in oil to decrease deterioration rate,
extending service life of oil, etc. This results in a longer expected period of service time.
But such measures cost money, so ultimately equipment lifetime is somewhat a function
of cost. In general, most electrical equipment like transformers, breakers, and regulators is
designed to withstand the chronological, service, and abnormal events associated with about
"40 years" of service under typical ambient conditions and loading. Overhead conductors
and their ancillary equipment generally have a lifetime of 60 years or more (the required
mechanical strength pretty much assures immunity from electrical, thermal, and/or rapid
deterioration unless abused by very harsh service). Wood poles have a 40-75 year expected
lifetime, depending on quality and soil conditions. Steel structures typically last up to 100
years. Underground cable has perhaps the shortest expected lifetime among common types
of electrical equipment: although lead-covered paper-insulated cable has an expected


lifetime of up to 60 years if well cared for, most other types have lifetimes between 20 and
35 years.
Lack of proper care can accelerate aging effects
Lack of proper maintenance or inattention in setup and/or operation can, in some cases, lead
to excessive deterioration due to secondary impacts (e.g., neglected seals let in moisture
which leads to dielectric failure). The degree to which neglect leads to premature failure
depends on the type of device and the extent to which its design depends on periodic
maintenance.
Equipment with moving mechanical components tends to be most sensitive, that
sensitivity proportional to the frequency of mechanical operation. Thus, tap changers and
voltage regulators are most prone to early failure due to neglect. However, upon failure they
can often be repaired by performing the maintenance or replacement of parts that has been
deferred.
On the other extreme, most service transformers and residential power meters are
essentially no-maintenance devices - they can be installed and expected to last for a normal
lifetime without any maintenance. The first attention they get subsequent to installation is
when they fail to do their job for some reason. At this point they are usually replaced.
Equipment requiring little or no maintenance generally has only one "service mode":
replacement.
Neglect or stupidity in setup or operation is another matter entirely. A neglected aspect
of equipment use often leading to rapid deterioration is poor grounding. Equipment may not
be well grounded when installed, or the quality of the ground might degrade over time due
to corrosion of clamps, etc. In either case, a weak ground can cause higher than normal
and/or abnormal currents during "normal operation" accelerating the deterioration caused
by normal service stresses. More critical yet, poor grounds often exacerbate the impact of
lightning, switching surges, and through-faults to the point that the stress caused is two to
five times greater. Failure is nearly assured.
Other types of poor setup include mismatch of transformer impedance in multi-phase
banks composed of single-phase units, incorrect settings for regulators, and similar
inattention to details in equipment utilization.
7.4 MEASURES TO IMPROVE EQUIPMENT RELIABILITY AND LIFE
It is impossible to reverse aging, in humans or electrical equipment. Various measures can
be taken, however, to slow it down, mitigate its effects, or merely predict its effects more
precisely. All three have value to an electric delivery utility. Slowing deterioration extends
lifetime and cuts the probability of outage due to deterioration. Mitigating the effects
reduces the consequences of the failure. Predicting deterioration or failure more accurately
permits better management of impending failure and consequences in an effective and
economical manner. This section summarizes the steps a utility can take to reduce the
likelihood of failure of a key system component.
Replacement
Replacement with a new unit is an effective way of reducing the likelihood of failure of an
older unit. It is expensive when viewed from the capital-expense standpoint: the utility has
to purchase the replacement unit. However, it may be justifiable when examined on a life-
time cost basis, and it might actually be the least expensive option in cases where service or
repair would be expensive or ineffective or the consequences of a failure are extremely
expensive.

260 Chapter 7

One problem caused by replacement of really old equipment is that replacement models
of original design are no longer available, and the physical size or some other aspect of 'Tit"
of all possible replacements may be far different.
Often newer units are smaller, but this is not always the case. Regardless, the difference
in size may mean that the unit does not "fit" in some sense. Substations often have to be re-
designed, with foundations and bus-work rebuild to a different design to accommodate
replacements for 60-year-old breakers. This increases the cost of replacement and makes
other alternatives look attractive.
Retrofitting
In some cases "kits" exist which essentially "replace the guts" of an aged unit with new
equipment of vastly improved design and performance. This is called "retrofitting." As a
result, often the retrofitted unit is literally "better than new," although it is important to
realize that means better than a new unit of its original (usually obsolete) type. For example,
modules exist to replace the working mechanism of an older oil-filled medium voltage
circuit breaker with a vacuum-type breaker assembly. The retrofit module is designed to fit
the existing case, connections, and control plugs exactly. It performs as well or better than
new and requires less maintenance in the future. A retrofitted unit generally is as reliable as
a new unit.
Rebuilding or Refurbishment
A transformer, breaker, or other complicated electrical unit can usually be rebuilt to the
point where it is effectively "as good as new." This involves disassembly and rebuilding
with all possible worn or deteriorated parts and materials replaced. Often reassembly is
involved and a major reason the refurbishment makes big improvements. For example, a
rebuilt transformer may use the material from the original transformer core, ensuring it is
tightened: after years of usage it was likely to be loose, a liability during a though-fault.
Similarly, reassembly of some circuit breakers using a "factory" method is more likely to
result in truly "as new" assembly tolerance than refitting done in the field.
Usually this type of service can only be done at the factory or a rebuilding/refurbishment
service center. It is not done "on site": the unit must not only be removed from service but
shipped to a service center. It is also not substantially less expensive than buying a new unit
(by this the author means it is likely to be on the order of l/2 the cost of a new unit).
However, if done well the rebuilt unit has an identical lifetime (or nearly so) as compared to
a new unit.
Overhaul
Overhaul has different meanings at different utilities. However, it is generally considered to
mean, and the author considers that it means, the most comprehensive service that can be
done on a unit (other than possible retrofits) while keeping it at its location. Overhaul
includes replacement of all replaceable deterioratable parts such as gaskets, seals, and
fittings, and may include replacement of control wiring (insulation ages and cracks).
Mechanical parts subject to wear are generally replaced or at least inspected for tolerances
and adjusted if needed. Without exception the unit is completely tested, often with precision
test equipment (e.g., a breaker's operation is timed, etc.).
Overhauled units are often as reliable as new units, but will not be as durable. They
provide reliable service for a few years but they are again "old" much earlier than a new or
rebuilt unit would be. However, it is worth considering the comments made earlier in this
chapter under the heading "Worn-Out Equipment."


Major Service and Service
This can mean anything depending on how a utility defines it and how its operations and
service crews interpret it. However, usually it means the scheduled partial overhaul of units.
Not everything included in an overhaul is included. The deleted items may not need to be
done, or they may have been deleted due to cost. Generally, at utilities trying to reduce cost,
"major service" or "service" is targeted to include only those items that have lately proven
to be creating problems for the utility. It tends to reduce the incidence of problems in the
short term, although it does not reduce it to "like new levels."
Inspection and Repair as Required
A program that calls for inspection of equipment and immediate repair or service only of
items found needed will improve reliability. Equipment that might fail or malfunction is
often found before that occurs, and the situation corrected.
If done well, this often has a more profound effect on reliability than "major service"
because it generally includes a wider scope of application for the money. "Service" is often
scheduled for units thought to need it. Inspection often casts a wider net and applies service
resources to only those that need the repair. Often, "inspection and repair as required" is
more cost effective than scheduled service based on interval-since-last-service.
Inspection
Inspection alone will not improve reliability, but it will improve predictability of failures
and lifetime and permit better planning for contingencies and replacement budgets. Thus it
can have a good impact per dollar. Comprehensive inspection, if backed up by a program
of replacement, repair, overhaul, or service based on its results, lowers costs.
Condition Assessment
A detailed evaluation, often involving diagnostic tests along with analysis of results and the
pattern of the diagnostic evaluations, can be used to evaluate the condition of a transformer,
breaker, or underground cable. This alone does nothing to improve the reliability or extend
the life of the unit, in fact, there is some evidence that a few of the "extreme" diagnostic
tests accelerate some types of deterioration themselves.
However, condition assessment can be used to improve customer service reliability. It
identifies incipient failures, often weeks before they would occur. It identifies units which
are not likely to fail so that resources can be put elsewhere. It improves predictability of
future operating problems. It permits much better management of resources and reliability.
Condition monitoring
"Real time" inspection or diagnosis is often expensive but effective at tracking equipment
performance and identifying problems before they become serious. This does improve
reliability, because actions can be taken to both reduce stress (loading, voltage) if it appears
to be getting the better of the equipment and to take it out of service, avoiding a life-ending
failure. This also reduces service costs, although generally not enough to justify the cost of
the monitoring on that basis alone.
Condition modeling
Condition assessment data can be used, in conjunction with a computerized model of
equipment deterioration, loading, operating events, and other factors, to "predict" if and
when the equipment needs inspection or service or when it might possibly fail. Often this
"modeling" is fairly simple, merely prioritization of equipment kept in a maintenance and

262 Chapter 7

inspection database on the basis of time since last inspection; loading (if a transformer) or
operations (if a breaker); and (if available) operational data like nearby lightning strikes and
through-faults. A TLM (Transformer Load Management) program is a good example of a
widely used model in the industry that nearly always proves very effective in both reducing
failures and extending equipment lifetimes.
De-Rating to Preserve Remaining Lifetime
Since the major modes of deterioration and failure are often heat-related or other service-
stress created causes, one way to extend the lifetime of any electric equipment containing
insulated components is to ask it to do less work. The lifetime of transformers, motors,
regulators, and some other electrical devices is heavily dependent on the temperature at
which it operates, which in turn is a function of the load it serves. Reducing the load
reduces the temperature which reduces stress, thus extending life. The relationship between
loading and deterioration varies depending on equipment type, but this can be quite
effective at extending life with cables and transformers. Due to the quadratic relationship
between loading and losses and the exponential relationship between losses and stress, a
very modest reduction in loading makes a substantial extension of expected lifetime. As a
rule of thumb, an 8% reduction in loading will double expected lifetime (and an 8%
increase will halve lifetime). Thus, at the end of 20 years of serving a full 50 MVA peak
annual load, a "40-year life" transformer would have a further 40 years of expected service
if applied only to serve 46 MVA of peak demand.
Figure 7.11 shows a de-rating curve for a generic power transformer that has an
expected 40-year lifetime and has already seen various years of service at its full nameplate
load level. Calculations were actually based on a 138 to 12.47 kV 50 MVA oil-filled
transformer, serving a typical utility load pattern of 55% annual load factor, using the
equations 7.1 and 7.3. When new it needs no de-rating - its expected lifetime is 40 years.
After 20 years of service against a 50 MVA peak load, it must be de-rated to 46 MVA if
expected to last another 40 years. If it has served 30 years at 50 MVA peak, it must be de-
rated to 42 MVA peak load. De-rating both extends the expected insulation lifetime of the
equipment and reduces its expected failure rate by a similar amount: the two are equivalent.
Often, de-rating makes sense in a situation. It extends the lifetime (and value) of the
equipment and it reduces the probability of failure. However, it does not always make
sense. De-rating also reduces the value the utility is obtaining from the unit - the unit is
doing less work. In a few cases it can actually make economic sense to increase the load on
a unit and thus "burn it up" in only a few years. For example, increasing the load of this
same unit to 58 MVA, after 20 years of 50 MVA service, results in an expected remaining
service life of only 5 years. But the value provided by serving the higher load, and deferring
replacement of other additions to the system, may be more than worth the monetary value
of a 20-year-old transformer. This and similar decisions of usage and economic value
optimization will be discussed in subsequent chapters.

6
This is a simple and only approximate model based on equations 7.1 and 7.3 covered later in this
chapter and is meant to illustrate the type of reductions necessary to obtain meaningful extensions of
lifetime. Equations 7.1 and 7.3 were used to determine what loading level would extend the
remaining period of the 40-year life to 40 years. For example, at the end of 35 years of service
against a 50 MVA load, the unit has an expected 5 years left. A load reduction of 22.5%, to 38.5
MVA, cuts deterioration rate by a factor of eight, increasing the five years remaining to forty years.
This analysis neglects a number of factors including evidence that a unit that has survived 35 years
of service is statistically "better than average" and thus probably has more than 5/40 of its remaining
lifetime life left in it.

De-Rating That Must Be Applied To Provide
Expected Forty-Year Remaining Lifetime

100

0) (B
Q. Q. 60

« E
< 40

20

10 20 30 40
Time Already In Service - years

Figure 7.11 One way of obtaining additional lifetime from a transformer (and other wound devices)
is to reduce thermal stress on it by de-rating it. See text for details.

Figure 7.1 showed the conceptual model of transformer lifetime, which in later chapters
will be used to model certain aspects of transformer performance and asset management.
Figure 7.12 uses this to show several qualitative examples of how changes in one or more
parameters affects expected service lifetime of electrical equipment. The six illustrated
examples in Figure 7.12 and important points about each are:
1. This model represents an electrical unit as suffering from constant deterioration
of its insulation strength, a background damage likelihood, and a growing
susceptibility to through-faults and damage causing failure, due to the cumulative
effect of both damage and through-faults.
2. Loading affects the calculated lifetime because it reduces deterioration rate. But
the computed lifetime gained from using a unit at a lower loading may not buy
many years of actual service. Here, a lower loading results in a deterioration
curve that now takes 50% longer to reach the original vertical level of point A.
But extended lifetime means extended exposure to damage and through-faults for
a (possibly) already susceptible unit. Actual lifetime gain realized may be far
shorter than calculations indicate.
3. A harsher fault-duty environment at the transformer's location means a more
rapid deterioration. Here, the unit sees larger and/or more frequent through-faults.
4. A harsher physical environment at the transformer's location also raises its
susceptibility This does not normally accelerate the rate of increase in
susceptibility as does fault-duty harshness, but it raises susceptibility and lowers
expected life.

264 Chapter 7

1.00 Base Case 1.00 Lower Loading

Time Time

Harsher Fault Duty More Physical Damage
1.00 1.00

V)

Time Time

1.00 Heavier Duty Unit 1.00 Lack of Maintenance

Time Time

Figure 7.12 Conceptual model showing changes that affect lifetime.


Rating of equipment provides a basis for identification and comparison of units, not recommended
operating or loading limits. The rating is determined by evaluating the unit on a set of standard criteria
(a standard) that are used to assure uniformity in the ratings.
Loading is the actual utilization of a unit of equipment in service.
Lifetime is the time period associated with some specific decline or deterioration of a measurable
parameter such as material strength, as determined under specific test or rating conditions (a standard).
It is not a predicted or recommended service lifetime and may not relate in any meaningful way to
expected lifetime of the unit in service.
Service lifetime is that actual time that a unit remains in service before failing or wearing to the point
of having no ability to perform its task and having no value. Service lifetime and "lifetime" with
respect to any standard often have very little or no correlation.
A failure means that the unit has ceased to perform its function, whatever that is. The term "life-
ending failure" is often used for the event or condition that ends the unit's service life.
Loading is directly related to expected service lifetime. High utilization of a unit will increase heat
which will accelerate deterioration and eventual loss of useful remaining life.
End-of-senice-life actually occurs when a unit either fails in a way, or wears to a point, where it is
no longer repairable or worth repairing.
There is no firm "rating" above which a wound device like a transformer, motor, or regulator cannot
be loaded, at least briefly. Higher loadings just accelerate deterioration, but "lifetime" at very high
loadings may be quite short. By contrast, there are firm limits to the loading that switches, breakers,
and other equipment which handle current or mechanical stress through moving contacts or parts can
handle: operation above these limits leads to immediate failure.
A useful model of equipment lifetime is that a unit has "one lifetime" when new - for example forty
years times 8760 hours/year equals 350,400 hours of useful life. Actual usage will then "spend" these
350,400 hours at a rate faster or slower than one hour of life per hour of use, depending on loading,
other conditions, and character of usage.
Run-to-failure is the preferred usage plan for most electrical equipment: a transformer, breaker, or
pole is installed and used until a failure renders it not worth repairing. This "final failure" ends its life.
Four causes lead to failure of functionality and final failure: long-term deterioration of materials,
long-term deterioration of mechanical surfaces (wear), through-faults, and external damage. Most
"standards" evaluate a rating and lifetime only on the basis of deterioration alone.
Deterioration occurs due to heat, from electrical losses, or current leakage, due to voltage stress.
Both inevitably result from usage of the equipment for its intended purpose.
Small changes in loading can make large impacts on rated lifetime of some equipment. A 4%
increase in loading of a transformer running near its rating maximum temperature may halve its
expected lifetime.
The expected service lifetime of any unit can be estimated in a dependable manner based on
intended loading and information on the physical and electrical environment in which it will operate.
The expected service lifetime of a unit can be extended through comprehensive service including
perhaps rebuilding or replacement of parts. Rebuilt units often have much increased expectation of
remaining lifetime, but seldom have the same expected lifetime as new equipment.
Lack of proper maintenance or inattention in setup and/or operation can, in some cases, lead to
excessive deterioration or wear. This may lead to functional or complete service-lifetime-ending
failure.
Utilization decisions require juggling loading, expected lifetime, probability of damage, and cost. A
conservative loading level means deterioration will be slow and the unit could last for decades. But
that means poor economy of use and long exposure to damage by other causes - through-faults,
weather, vandalism - which will likely be the cause of eventual end of life.

266 Chapter 7

5. A "more robust unit," reinforced as appropriate (i.e., internally braced for
stronger through-faults, lightning protected, etc.), will be less susceptible over
time and thus has a higher expected service life.
6. Lack of maintenance results in an exponential acceleration of susceptibility to
failure. Even a unit in "good condition" in the sense that it has not had its internal
strength "eroded" by many through-faults or years of exposure to weather can go
bad in only a few years if not cared for properly.
The single most important point for utility planners and managers to keep in mind is that
all equipment installed on their electric system will eventually fail. This chapter used
transformers as an example for a discussion of ratings, lifetime, failures, and methods of
influencing lifetime. The same basic concepts apply to other electrical equipment, although
often the specific service-age relationships are different. But qualitatively, all equipment
has rating-lifetime behavior as discussed here. Key points brought up in this chapter are:
1. Equipment loading and expected service lifetimes are usually based on a
standard definition of how to calculate the effect of one on the other. A rated
capacity is really only the specific load associated with some calculated
deterioration rate for the equipment.
2. Failure is caused by deterioration due to age, service, infrequent but stressful
abnormal events, and damage from external sources.
3. Most electrical units can tolerate brief periods of high loading, until temperature
rises, without any undue deterioration.
4. Due to the efficient design of most equipment, a unit that is near failure from
one area of deterioration is probably fairly "shot" in other areas, too.
Table 7.6 gives a one-page summary of key points and concepts in this chapter.

FOR FURTHER READING
R. E. Brown, Electric Power Distribution System Reliability, Marcel Dekker, New York, 2002
P. Gil, Electrical Power Equipment Maintenance and Testing, Marcel Dekker, New York, 1998.
N. H. Malik, A. A. Al-Arainy, and M. I. Qureshi, Electrical Insulation in Power Systems, Marcel
Dekker, New York, 1998
New York, 1999.
E. Santacana, et al, Electric Transmission and Distribution Reference Book, fifth edition, ABB Inc.,
Raleigh, 1997.
W. Thue, Electrical Power Cable Engineering, Marcel Dekker, New York, 1999.

8
Equipment Failures and
System Performance

8.1 INTRODUCTION

T&D systems are an interconnection of thousands of building blocks, components arranged
and interconnected together to form a working power delivery system. When the system
fails to do its job of delivering power to all of its distributed consumer points, it is usually
because one or more of these building blocks has failed to fulfill its role in the system.
Many system performance problems are due to equipment failures.
Chapter 7 covered some of the basics of capacity ratings, expected lifetimes, and failure
modes associated with individual units of equipment — the components that make up this
system, v^ne can usefully speak about failure rates and probabilities only v/hen re/erriug to a
large number of units. This chapter will examine failure as seen in a large population of
equipment and the impact that varying failure rate or lifetime or other aspects of equipment
has on its reliable operation as part of the power system.
Section 8.2 begins with a look at equipment failure rates, aging, and how the two
interrelate over the large set of equipment in a typical distribution system to influence
system reliability. Section 8.3 then takes a look at failure rate and the distribution of ages in
the inventory of equipment actually in service. What does the age distribution of equipment
over the entire system look like as equipment ages, "dies," and is replaced by newer
equipment? How does that effect replacement budgets and failure rates of the whole
system? What impact do various age-management policies have? Section 8.4 concludes
with a summary of key points.
8.2 EQUIPMENT FAILURE RATE INCREASES WITH AGE
All equipment put into service will eventually fail. All of it. Some equipment will last
longer than others will, and often these differences in lifetimes will appear to be of a
random nature. Two apparently identical units, manufactured to the same design from the
same materials on the same day at the same plant, installed by the same utility crew on the
same day in what appear to be identical situations, can nonetheless provide very different
267

268 Chapter 8

lifetimes. One might last only 11 years, while the other will still be providing service after
40 years. Minute differences in materials and construction, along with small differences in
ambient conditions, in service stress seen by the units, and in the abnormal events
experienced in service, cumulatively lead to big differences in lifetimes.
In fact, about the only time one finds a very tight grouping of lifetimes (i.e., all of the
units in a batch fail at about the same time) is when there is some common manufacturing
defect or materials problem - a flaw - leading to premature failure.
Failures, Failure Rates, and Lifetime
Table 7.3 listed four consequences of failure and the allied discussion in Chapter 7 made an
important distinction that a failure does not necessarily mean that a unit has "died," only
that it has ceased to do its function. As there, that type of failure will be called a "functional
failure." The terms "failure" and "failed" as used here will mean an event that ends the
useful life of the device.
Failure rate is the annual rate of failure - the likelihood (if predicting the future) or
actual record (if analyzing the past) of failure as a portion of population. It is usually
expressed as a percentage (e.g., 1.25%) and/or applied as a probability (1.25% probability
of failure). Expected lifetime - how long equipment will last in service - and failure rate are
clearly linked. A set of units expected to last only a short time in service must have a high
failure rate - if they do not fail at a high rate they will last a long time in service and thus
have a long expected lifetime
However, expected lifetime and failure rate are not necessarily functionally related. For
example, a set of units that have an average expected lifetime of twenty years may not
necessarily have a failure rate of 5% annually. It may be that failure rate for the next 10
years is only 2%, and that it then rises rapidly near the end of the period. Regardless, as a
qualitative concept, one can think of failure rate and lifetime as inversely related. If set A of
equipment has an expected lifetime twice that of set B equipment, then it has a much lower
failure rate.
Quantitative Analysis of Equipment Failure Probabilities
The concepts of failure and expected lifetime can be applied to individual units of
equipment (Chapter 7). But failure rate really only makes sense when a large population is
involved. Application of a failure rate to a specific unit of equipment is appropriate only in
the context of its being a member of a large set. Although mathematically one can apply
probabilities to a single unit of equipment to obtain expectations, any real value comes only
when the information is applied to analyze or manage the entire population.
Figure 8.1 shows the classic "bathtub" lifetime failure rate curve. Qualitatively, this is
representative of equipment failure likelihood as a function of age for just about all types of
devices, both within the power systems industry and without. The curve shows the relative
likelihood of failure for a device in each year over a period of time.
This "bathtub curve" and its application are probabilistic. Curves such as Figure 8.1
represent the expected failure of devices as a function of age over a large population. This
curve cannot be applied to any one particular unit. It can only be applied over a large
population, for which one can predict with near certainty the overall characteristics of
failure for the group as a whole. Thus, based on Figure 8.1, which has a failure rate of 1.5%
for year 8, 1.5 out of every 1000 units that have lasted 8 years will fail in that 8th year of
service. One can also see that among those units that survived 35 years of service, the
average probability of failure in the next year is .075%, etc.

Equipment Failures and System Performance 269

10 20 30 40 50
Component Age (years)

Figure 8.1 The traditional bathtub failure rate curve, with three periods as discussed in the text.

Figure 8.1 shows three different periods during the device's lifetime. These are:
1. Break-in period. Failure rate for equipment is often higher for new
equipment. Flaws in manufacturing, installation, or application design lead
to quick failure of the device.
2. Useful lifetime. Once the break-in period is past, there is a lengthy period
when the device is performing as designed and failure rate is low, during
which the design balance of all of its deterioration rates is achieving its
goal. During this period, for many types of equipment, the failure rates due
to "natural causes" (deterioration due to chronological aging and service
stress) may be so low that the major causes of failure for the equipment set
are damage from abnormal events.
3. Wear-out period. At some point, the failure rate begins to increase due to the
cumulative deterioration caused by time, service stress, and abnormal
events. From this point on failure rate increases with time, reaching the
high double-digit percentages at some point.
Often, periodic maintenance, rebuilding or refitting, filtering of oil, etc., can "extend"
lifetime and lower expected failure rate, as shown in Figure 8.2 This concept applies
only to units designed for or conducive to service (i.e., this applies to breakers, perhaps,
but not overhead conductor). Here, the unit is serviced every ten years or so. Note
however, that failure rate still increases over time, just at a lower rate of increase, and that
these periodic rebuilds create their own temporary "infant mortality" increases in failure
rate.
Failure Rate Always Increases with Age
All available data indicate that, inevitably, failure rate increases with age in all types of
power system equipment. Figure 8.3 shows operating data on the failure rate of

270 Chapter 8

1.0

8 0.8
V.
0)
0.6
Periodic maintenance
(0
tt 0.4 or rebuilding/refit
o>

0 10 20 30 40
Component Age (years)
Figure 8.2 A bathtub curve showing the impact of periodic rebuilding and refitting.

underground equipment for a utility in the northeast United States, which is qualitatively
similar to the failure performance seen on all equipment in any power system.
Failure rate escalation characteristics
Figure 8.3 shows how the rate of increase of failure rate over time for different
equipment can exhibit different characteristics. In some cases the failure rate increases
steadily over time (the plot is basically a straight line). In other cases the rate of increase
increases itself - failure rate grows exponentially with a steadily increasing slope. In yet
other cases, the rate climbs steeply for awhile and then escalation in failure rate decreases
- a so-called "S" shaped curve. Regardless, the key factor is that over time, failure rate
always increases.
There are cases where failure rate does no-. Increase over time, where it is constant
with time (it just does not increase with age) or where it may actually go down over time.
But these rare situations are seen only in other industries and with equipment far different
than power system equipment. All electrical equipment sees sufficient deterioration with
time and service that failure rate is strictly increasing over time.
Eventually the failure rates become quite high
Figure 8.3 is actual data representing a large population in an actual and very
representative electric utility. Note that the failure rates for all three types of equipment
shown eventually reach values that indicate failure within 5 years is likely (rates in the
15-20% range). In some cases failure rates reach very high levels (80%). Failure in the
next year or so is almost a certainty. To power system engineers and managers who have
not studied failure data, these values seem startlingly high. However, these are typical for
power system equipment - failure rates do reach values of 15%, 25%, and eventually
80%. But these dramatically high values are not significant in practice, as will be
discussed below, because few units "live" long enough to become that old. What really
impacts a utility is the long period at the end of useful lifetime and beginning of wear-out
period when failure rate rises to two to five times normal (useful lifetime) rates.


0.5
CABLE SECTIONS 25-kV Solid
c5 0.4
CD

CD
25-kV Paper,
I 0.2
0> 15-kV Paper,
i 0.1

10 15 20 25 30
Equipment Age (years)

1.0
CABLEJOINTS 25-kV Solii
CD 0.8
>»
15-kV Soli
CD
_ 0.6
CD
15-kV Paper, 15-kV Paper-Solid
S 0.4
25-kV Paper-
CD
3 0.2
a
10 15 20 25 30

PA D-M OUNTBD TRA NSFORM ERS

0.00
10 15 20 25 30

Figure 83 Data on failure rates as a function of age for various types of underground equipment
in a utility system in the northeastern US. Equipment age in this case means "time in service."
Equipment of different voltage classes can have radically different failure characteristics, but in all
cases failure rate increases with age.

272 Chapter 8

Predicting Time to Failure
High failure rate and uncertainty make for a costly combination
As a conceptual learning exercise, it is worth considering how valuable exact knowledge of
when any particular device would fail would be to a power delivery utility. Suppose that it
were known with certainty that a particular device would fail at 3:13 PM on July 23rd.
Replacement could be scheduled in a low-cost, low-impact-on-consumers manner just prior
to that date. There would be no unscheduled outage and no unanticipated costs involved:
impact on both customer service quality and utility costs could be minimized.
It is the uncertainty in the failure times of power system equipment that creates the high
costs, contributes to service quality problems, and makes management of equipment failure
so challenging. The magnitude of this problem increases as the equipment ages because the
failure rates increase: there are more "unpredictable" failures occurring. The utility has a
larger problem to manage.
Failure time prediction: an inexact science
With present technologies, it does not seem possible to predict time-to-failure of an
individual equipment unit exactly, except in cases of expensive real-time monitoring such
as for power transformer DGA systems. In fact, capability in failure prediction for
equipment is about the same as it is for human beings. The following statements apply just
about as well to people as to electrical equipment:
1. Time-to-failure can be predicted accurately only over a large population (set of
units). Children born in 2003 in the United States have an expected lifetime of 77
years, with a standard deviation of 11 years. Similarly, service transformers put
into service in year 2003 at a particular utility have an average expected lifetime
of 53 years with a standard deviation of 9 years.
2. Assessment based on time-in-service can be done, but still leads to information
that is accurate only when applied to a large population. Thus, medical
demographers can determine that people who have reached age 50 in year 2003
have an expected average 31 years of life remaining. Similarly, statistical
analysis of power transformers in a system can establish that, for example, those
that have survived 30 years in service have an average 16 years of service life
remaining.
3. Condition assessment can identify different expectations based on past or existing
service conditions, but again this is only accurate for a large population. Smokers
who have reached age 50 have only a remaining 22 years of expected lifetime,
not 31. Power transformers that have seen 30 years service in high-lightning
areas have an average of only 11 years service life remaining, not 16.
4. Tests can narrow but not eliminate uncertainty in failure prediction of individual
units. All the medical testing in the world cannot predict with certainty the time
of death of an apparently healthy human being, although it can identify flaws that
might indicate a high likelihood for failure. Similarly, testing of a power
transformer will identify if it has a "fatal" flaw in it. But if a human being or a
power system unit gets a "good bill of health," it really means that there is no
clue to when the unit will fail, except that it is unlikely to be soon.
5. Time to failure of an individual unit is only easy to predict when failure is
imminent. In cases where failure is due to "natural causes" (i.e., not due to


abnormal events such as being in an auto accident or being hit by lightning),
failure can be predicted only a short time prior to failure. At this point, failure is
almost certain to be due to advanced stages of detectable deterioration in some
key component. Thus, when rich Uncle Jacob was in his 60s and apparently
healthy, neither his relatives nor his doctors knew whether it would be another
two years or two decades before he died and his will was probated. Now that he
lies on his deathbed with a detectable bad heart, failure within a matter of days
is nearly certain. (The relatives gather.)
Similarly, in the week or two leading up to catastrophic failure, a power
transformer usually will give detectable signs of impending failure: an
identifiable acoustic signature will develop, internal gassing will be high, and
perhaps detectable changes in leakage current will be present, etc. But a lack of
those signs does not indicate certainty of any long period before failure.
6. Failure prediction and mitigation thus depend on periodic testing as units get
older. Given the above facts, the only way to manage failure is to test older units
more often than younger units. Men over 50 years of age are urged to have
annual physical exams in order that possible system problems are detected early
enough to treat. Old power transformers have to be inspected periodically in
order to detect signs of impending failure in time to repair them.
7. Most electrical equipment gives some diagnosable sign of impending failure.
Temperature rise, changes in sound volume or frequency, leakage current,
changes in power factor - something — nearly always provides a factor which, if
noted, indicates failure is very near.
Table 8.1 summarizes the key points about failure time prediction.

Table 8.1 Realities of Power System Equipment Lifetime Prediction

1. Time to failure can be predicted accurately only over large populations and in
a probabilistic manner.
2. Past and present service conditions can be used to narrow the expected
uncertainty range, but "deterministic" models of remaining lifetime are
unrealistic and unreliable.
3. Testing provides accurate "time to failure" information only when it reveals
flaws that mean "failure in the near future is nearly certain."
4. Test data that reveal no problems do not mean the unit has a long lifetime
ahead of it. At most, they mean there is little likelihood of failure in the near
term, nothing more. At worst, the results mean only that the tests did not find
flaws that will lead to failure.
5. Periodic testing and inspection is the only way to assure accurate prediction of
time to failure.
6. "Good" test scores have no long-term meaning. All they mean is that the unit
is currently in good condition and unlikely to fail soon.
7. Testing needs to be done more often as a unit gets older.

274 Chapter 8

8.3 A LOOK AT FAILURE AND AGE IN A UTILITY SYSTEM
This section will examine the effects of failure rate escalation, high failure rates, and
various replacement policies on the average age and the average failure rate of a large
population of equipment in utility service. This is done by using a computerized program,
Transformer Population Demographics Simulator, written by the author. It computes
failures and lifetimes, and can simulate replacement policies for large sets of transformers.
The example cases given below designed to illustrate key points about the relationship
between failure rates, how many units are left, and how many must be replaced. Often the
results are interesting and counter-intuitive. For example, the extremely high failure rates
that an escalating trend eventually reaches (perhaps as high as 50% annually, see Figure
8.3) are real, but of little concern to a utility. The reason: few units get to be that age - the
vast majority fail long before they are that old.
A large electric utility may have 100,000 or more service transformers in the field and
over a thousand power transformers in operation on its system. Every year all of those units
age. Some fail and are replaced by newer units. The population's average age may increase
or decrease, depending on how many failed and at what ages. The "demographics" of this
population depend on the failure rates and the failure rate escalation curve for the
equipment.
Example 1: A Typical Failure Rate Escalation and Its
Impact on the Installed Equipment Base
Figure 8.4 illustrates a very simple example that will begin this section's quantitative
examination of failure, installed base characteristics, and overall impact on the utility. In
this example, the group of 100,000 service transformers is installed in one year, a rather
unrealistic assumption but one that has no impact on the conclusions that this example will
draw.
As a group, this set of 100,000 units has the statistical failure rate characteristic shown
in the top part of Figure 8.4. That plot gives the probability that an operating unit of any
particular age will fail in the next 12 months of service. In this case there is no high break-
in-period failure rate. The base rate during normal lifetime begins at 1.5% per year, rising to
2.5% by year 24, 6.6% by year 30, and to 9% annually by age 40. This curve is based upon
actual service transformer failure rate curves (see bottom of Figure 8.3).
The bottom diagram in Figure 8.4 shows, as a function of age, the percent of the
100,000 units installed in year zero that can be expected to remain in service each year, as
units fail according to the expectation defined by the top curve. In year 1, 1.5% of the units
fail, meaning 99% are in service at the beginning of year two. At the end of a decade, 85%
are still in service. The failure rate is initially 1.5%, increasing slightly above that value
each year. Despite this rise, only 15% of the units (ten times 1.5%) fail in the first decade.
The reason is that the number of units left to fail decreases each year - there are only
98,500 units left at the end of the first year, etc., so 1.5% is not exactly 1,500 in the second
year. The number of actual failures decreases slightly each year for the first decade, to a low
of only 1,440 failures in year ten, because the number of units remaining to fail drops faster
than the failure rate increases.
At the end of 20 years, 71% of the units remain, and at the end of thirty, only 53%
remain. The 50% mark is reached at 32 years. By the end of year 50, failure rate has
escalated to more than 15%, but only a paltry 10.3% of the units remain. Only 500 (.7%)
make it to year 60. Less than two are expected to be in service by year 70, when failure rate
has escalated to 50%. The average unit ends up providing 43.8 years of service before
failure.


Failures in intermediate years are the real
cause of this system's reliability problems
As Figure 8.4 shows, every year, as the units in this example grow older, their failure rate
increases. But every year, because many have already failed in previous years, there are
fewer units remaining to potentially fail in the next year. In this example, the number of
units that can be expected to fail in any year is the failure rate for that age times the number
remaining in that year. When does that value reach a maximum?
The left side of Figure 8.5 answers this question. That plot is the year-by-year product
of failure rate (left side of Figure 8.4) times the number of units remaining (right side of
Figure 8.5). As mentioned earlier, the failure rate is initially only 1.5% and it does not
increase much in the first few years. Thus, the number of units actually failing each year
drops slightly during the first decade as the product (failure rate x number of units
remaining) decreases slightly from year to year. But then, at about ten years, the annual
number of failures begins to rise. At this point, failure rate is increasing at a rate faster than
the number of remaining units is decreasing (i.e., the annual increase in the failure rate is
greater than the failure rate itself)-
The number of units actually failing each year peaks in year (age) 44, with 2,576
expected failures. Thereafter, even though the failure rate keeps increasing every year, the
number of failures actually occurring decreases, because there are fewer and fewer units left
each year, so that the net number of failures (failure rate times number of units left)
decreases from there on. From here on, the annual increase in the failure rate is not greater
than the failure rate itself.
Thus, the very high failure rates that develop after five or six decades of service make
little real impact on the utility's quality of service. Units that are 70 years old have a 50
percent likelihood of failing in the next year, but as shown earlier, there are only two units
out of every 100,000 that make it to that age - essentially an anomaly in the system.
Instead, the high impact failure levels that plague a utility are caused by transformers of
intermediate age.

1.00

S .10

10 20 30 40 50 60 70 10 20 30 40 50 60 70
Time - Years Time - Years

Figure 8.4 Left, failure rates as a function of age for a group of 100,000 service transformers. Right,
the percent remaining from the original group as a function of time if they fail at the failure rates
shown at the left as they age. After twenty years, 60% remain, but after fifty years only 12.5% are still
in service, and after 70 years, only 2 of the original 100,000 are expected to still be in service.

276 Chapter 8

2.5

= 2.0

10 20 30 40 50 60 70 10 20 30 40 50 60 70
Time - Years Years of Service

Figure 8.5 Left, number of failures occurring each year in Example 1's population that originally
numbered 100,000 units. The maximum is 2,576, in year 44, when the combination of escalating
failure rate and number of remaining units peaks. More than half the failures occur in the range
between 20 and 45 years in service. Right, a generalized result applicable to any population made up
of these units no matter how many and regardless of when they were installed. This curve is the
distribution of failure likelihood for one of these units, as a function of service age. It gives the
probability that a unit of this type will fail at a certain age. The area under this curve (including a small
portion of it beyond 70 years that is not shown) is 100%.

Failure-Count Diagrams
The right side of Figure 8.5 shows this same failure-count diagram with the scale changed
from a base of 100,000 units to just one unit. This diagram is basically a probability
distribution of when a unit can be expected to fail. The difference between this diagram and
a failure rate diagram (left side of Figure 8.4) is that the failure rate curve gives the
probability of failure for a unit as a function of its age, n, assuming it has lasted the
previous n-1 years. The diagram shown at the right of Figure 8.5 gives the likelihood of
failure in year n taking into account that the unit may not have lasted to year n. To
distinguish this curve from failure rate curves like that shown in Figure 8.4, it wi1! be called
& failure count curve, even if scaled as it is in Figure 8.5 to a percentage basis.
This is an interesting and very useful diagram, because it applies to any population
made up of this same type of transformer. The curve gives, in relative terms, how much
transformers of a particular age contribute to an installed base's failure rate. For any
population of this type of unit, no matter what the mix of ages - some installed last year,
some installed long ago - it is still those units that have reached 44 years of service that
contribute most to the system's problems. Units older than that fail with a higher likelihood
(left side of Figure 8.4) but there are too few to generate as high a total count.
Information like Figure 8.5, when developed for a specific situation (this diagram is for
only one particular transformer type as applied at one specific utility), is the foundation for
studies of proposed inspection, service, and replacement policies as well as various asset
management strategies. Application to strategic and tactical planning of that type will be
discussed later in this chapter.
Example 2: A More "Real World" Case
The preceding example provided useful insight into equipment failure characteristics, and
led to the failure count contribution curve (Figure 8.5), a very useful analytical tool. But to
see the full implications of failure rate escalation on a utility and understand how certain


replacement policies might reduce failure count, it is necessary to sacrifice Example 1's
simplicity for more realism. Specifically, one needs to look at a situation where failed units
are replaced, which is the fact of life in the "real world." Thus, Example 2 builds on
Example 1, using the same type of service transformers. It:
1. Assumes, as before, 100,000 units are installed in year 0.
2. Assumes a period of 70 years.
3. But assumes that failed units are replaced immediately with new
ones, keeping the overall count at 100,000.
4. And also assumes these replacement units follow the same failure
rate curve as the original units.
5. Looks at the entire population of transformers that results from
these assumptions, at the end of 70 years.
This is a more realistic example, as it represents what a utility has to do - keep the same
number of units in service, replacing failed units when they fail.1 It is still not completely
realistic because of the initial "build" of 100,000 units in year zero, but for the moment that
is not an issue. The important point in the overall equipment base and its interaction with
failures in this example is that replacement units can fail, too.
What does this installed equipment base look like with respect to age distribution of
units in service, average failure rate, and failure count? Figure 8.6 shows the distribution
of ages of the 100,000 units in the system after 70 years. As in Example 1, in year 1 (the
first year) 1,500 units failed, but here they were replaced with new units, which are a
year newer, thus at the end of the 70-year period, those "first year" replacements that
have survived are now 69 years old. But those 1,500 units did not all survive. They failed
with the same characteristic trend as the original set, meaning that 1.5%, or 22, failed in
their first year of service and were replaced in year 2. Their replacements are now 68
years old, assuming they lasted and did not fail in some interim year.
Furthermore, in year 2, 1,478 of the original 100,000 units failed (1.5% of the 98,500
original units remaining after year 1). Thus, a total of 1,500 replacement units were
installed in year 2 (1,478 + 22). Those replacements began to fail along the same trend as
the original units. Thus, in year 3 there were failures of units that had been installed in
years 0, 1, and 2, etc.
Eventually, as the population ages and its average failure rate rises, the utility sees the
annual replacement rate over the entire population rising to about 3,000 units a year, and
it finds itself replacing units of all ages. And the net result, when all of these
replacements and replacements for replacements, etc., are added up, is that the utility had
to install about 60,000 additional replacement units during the 70 year period. More than
half of the original units failed. And following the failure count contributions of Figure
8.5, most of those that failed were "middle aged transformers" - those in the 15- to 45-
year-old range.

1
The fact that this example assumes the system is created from scratch in year zero makes no impact
on the results given here. As shown earlier, by year 70 only 2 of the original units are left. The
population consists of units that have been replaced and in some cases failed and replaced again. As
a result after 70 years there is only an insignificant "start effect" involved in the data - the model's
results in year 70 represent a fairly stable look at what year-to-year operation for the utility would be
with respect to service transformers.

278 Chapter 8

50

£ o
c o
3O
*- C-
SJB
%S
E
D a
0>
Area under the curve
Z 2 equals 100,000 units

.OS

10 20 30 40 50 60 70
Age - Years
Figure 8.6 Distribution of ages of units in the Example 2 system, 100,000 units that have been
replaced as they failed over the last 70 years.

A key point: the failure contribution curve from Example 1 (Figure 8.5) applies to this
example as well. Again, as stated earlier, that curve applies to all populations made up of
this same type of unit. Readers who are uncertain of this should stop and reflect on this
fact before moving on. For any such population, this curve is a representation of the
relative contribution to failures of units as a function of their age. Thus, the failure
contribution curve is a very important tool in managing reliability and replacement
policies. It will be discussed in detail later in this section.
Figure 8.6 shows the resulting equipment base's distribution of transformer ages, after
70 years for this example. It has nearly an even distribution of transformers from age 0
(new) to 30 years of age. At about 35 years of age the count takes a rapid plunge - this is
the period (starting at about 35 years m service) during which the bulk of failure counts
occur (see Figure 8.5), and thus the age when a good deal of replacements had to be made.
In this system, at the end of 70 years, the average unit in service is 22 years old.
However, due to the escalation of failure rates as units age, those older than the average
contribute a good deal more to the average failure rate. The entire population has an
average failure rate of 3.15%, or more than twice that of new units. That figure corresponds
to the failure rate of a unit that is 29 years old (see Figure 8.4, top). Thus, while the average
age of this population is 22 years old, its failure rate is equal to that of a population made
up of 29-year-old units.
Other Example Cases
The author ran a number of other cases, increasing realistic representations of actual utility
operations, through the simulation. However, no outstanding additional conclusions useful
for managing transformer failures and lifetime (as will be discussed in the next section) are
revealed. Conclusions of these studies and the cases studied were:
2
Figure 8.5 can also be interpreted as giving the relative age of units when they are replaced by the
utility over the course of its annual O&M in each year.


Case 3: A simulation that has the transformers being added gradually, rather than
all in one year, results in a population little different than shown in Figure 8.6.
That plot is fairly representative despite that "all in one year" scenario. Any "end
effect" of modeling all the units as starting in one year is worked out of the model
by the end of any simulated period longer than 50 years.
Case 4: Growth of the population was modeled in several cases, where the utility
had to expand the number of transformers each year by an amount that varied
from 1 to 3% annually. Figure 8.7 shows the types of changes that result from
applying an annual growth rate to the population. Of course, the total number of
units is greater than 100,000 in these scenarios. These populations become
somewhat "younger" and there is a sloped rather than a flat distribution of age
over the 0- to 30-year-old period.
Case 5: A period of high growth, lasting ten years with a growth rate of 5%, was
modeled as having occurred in the past. The result is a "bulge" in the age
distribution around the time (years in the past) of that growth period, as shown in
Figure 8.8. In cases where the growth spurt occurred about 30-40 years ago, this
large population of "now failing units" (see Figure 8.5) results in a relatively high
failure rate for the entire population. Where it occurred more than 50 years ago,
the impact is minimal - most of the units added then have failed since then.
Replacement Policy Analysis
Suppose this utility decided to replace all units as they reach fifty years of age, even if they
appear to be in good condition. Looking at the data for Case 2 (Figure 8.6), it would have to
replace only about 75 units annually - not a tremendous cost, particularly considering the
units will have to be replaced pretty soon anyway (they will most likely fail in a few years
at that age). However, the impact on the overall failure rate would be insignificant, making
no real difference in the average age or failure rates for the total population. The real
contributor to the system's annual failure count comes from units that are aged 25 to 45
years, because there are so many of them. Replacement at age 50 gets to the units after too
many have failed.
Replacement of units at age 40 has a far different effect. First, nearly 1,000 units a
year have to be replaced, so the annual cost is roughly 12 times that of a 50-year
replacement policy. But a noticeable portion of the system's unexpected failure rate is
avoided. The average failure rate drops to 2.6% (from 3.1%), a reduction in unexpected
failures of nearly 20%, wrought by replacement on only 1% of the units in the system
annually. Average age of a unit under this policy falls from 22 years given earlier to less
than 18 years. The population's average failure rate of 2.6% under this policy is
equivalent to the failure rate of a 25-year-old unit.
Thus, run-to-failure usually makes economic sense
Whether a replacement policy makes economic sense or not is something for
management to weigh in its evaluation of how best to spend the (always-limited) monies
it has for system improvements. The marginal cost of replacement can be weighed
against the marginal gain in reliability obtained, as well as compared against the marginal
cost of similar reliability gains available through other avenues. What is most important
in the foregoing examples is that the analytical results of a rather simple analysis of age,
failures, and remaining units can provide the type of tool needed to support such
decision-making.

280 Chapter 8

High growth
Low growth
No growth (Case 2)

20 30 40 50 60 70
Age - Years
Figure 8.7 Distribution of ages of units in examples that included annual growth of the utility (and
thus its population of transformers).

50
High growth 15-25 years ago
XI w
"E § No growth (Case 2)
3o

•9 ° .5
E «
3 O)
Z O

.05

10 20 30 40 50 60 70
Age - Years

Figure 8.8 A period of high growth, here a 5% annual addition of units for ten years, results in a
"bulge" in the age distribution around the time of the growth spurt that lasts about 40 years, failures
gradually working it out of the population. All cases shown here assume replacement of units as they
fail. Note that the 5% for 10-year growth results in a greater area under the curve. The shape of the
base and 50-60 year ago curves are nearly identical.


Table 8.2 Key Points from Chapter 8
All equipment will eventually fail unless replaced while still having some useful lifetime left.
Failure rate increases with age: excepting a brief "infant mortality rate" when new, failure
rate is monotonically increasing over time.
The average failure rate of equipment in any population will be greater than the failure rate
for equipment that is the age of the average age of the population.
Different modes or types of failures affect most equipment: a transformer can fail due to
core, winding, bushing, or case failures. One can view deterioration in each of these areas as
being in a kind of "race" to see where failure occurs first.
A unit that is "worn out" in one failure mode area is probably near failure in other
modes, too. Sound, common sense design principles result in equipment in which all modes
reach failure at about the same time: core, windings, bushings all are designed with similar
service lives.
The very high failure rates of really old equipment actually make little impact on a utility's
service reliability, because there are very few such units left.
The slightly-higher failure rates of forty-year-old equipment typically create the greatest
reliability problem for a utility. The bulk of failures come not from very old equipment but
from "middle aged" equipment. There, failure rate times number of units of this age left in
service is usually the highest.
The bathtub curve model of failure likelihood, or a modified version of it, is useful in almost
all situations involving anticipation and management of equipment failure.
Time-to-failure can be predicted accurately only by using probablistic methods applied over
large populations and in an expectation-of-failure manner.
Uncertainty about remaining lifetime (time to failure) is the real factor shaping both poor
service quality and increased utility costs.
Past and present service conditions can be used to narrow the expected uncertainty range,
but "deterministic" models of remaining lifetime for individual units are unrealistic and
unreliable.
Testing and condition assessment provide accurate "time to failure" information only when
they reveal flaws that in essence mean "This unit is very likely to fail in the near future."
Good test results and assessed condition have no long-term meaning. They mean only that
the unit is currently in good condition and unlikely to fail in the near future.
Periodic testing and inspection is the only way to assure accurate prediction of time to
failure.
Inspection and testing needs to be done more often as a unit gets older.
Replacement policy analysis is a rather straightforward way to combine age, failures, test
results, and other data to determine if and how units should be replaced rather than left in
service.
Early replacement policy can be optimized to determine if and how equipment should be
replaced at some specific age (e.g., 40 years) or condition (e.g., meets IEEE category 4 for
transformers).
Run to failure is still the best cost-justifiable policy for most electric utilities and most
equipment, in spite of the "costs" associated with failures.

282 Chapter 8

All equipment installed in an electric system will eventually fail and need to be replaced. It
is mostly the uncertainty associated with failure that creates poor service quality and raises
utility costs: if a utility could accurately predict failure, it could use all the lifetime available
in its equipment but still avoid service interruptions due to failing equipment. Inspection,
testing, and condition modeling can reduce uncertainty but not eliminate it or even produce
meaningful results on a single-unit "deterministic" basis. Early replacement policies can be
worked out for equipment, but usually "run to failure" is the most economical approach.
Table 8.2 gives a one-page summary of key points made in this chapter.

REFERENCES
P. F. Albrecht and H. E. Campbell, "Reliability Analysis of Distribution Equipment Failure Data,"
EEI T&D Committee Meeting, January 20, 1972.
R. E. Brown, Electric Power Distribution Reliability, Marcel Dekker, New York, 2002.
J. B. Bunch, H. I. Stalder, and J. T. Tengdin, "Reliability Considerations for Distribution Automation
Equipment," IEEE Transactions on Power Apparatus and Systems, PAS-102, November 1983,
pp. 2656-2664.
EEI Transmission and Distribution Committee, "Guide for Reliability Measurement and Data
Collection," October 1971, Edison Electric Institute, New York.
P. Gil, Electrical Power Equipment Maintenance and Testing, Marcel Dekker, New York, 1998.
Systems, November 1994, pp. 4082-4088.
E. Santacana et al, Electric Transmission and Distribution Reference Book, fifth edition, ABB Inc.,
Raleigh, 1997.
A. J. Walker, "The Degradation of the Reliability of Transmission and Distribution Systems During
Construction Outages," Int. Conf. on Power Supply Systems. IEEE Conf. Publ. 225, January
1983, pp. 112-118.
H. B. White, "A Practical Approach to Reliability Design," IEEE Transactions on Power Apparatus
and Systems, PAS-104, November 1985, pp. 2739-2747.

9
Load Reach and
Volt-VAR Engineering

9.1 INTRODUCTION
This chapter reviews voltage behavior on distribution systems and voltage planning and
engineering concepts. Many distribution planners view voltage, along with loading, as the
factor they must engineer in laying out and designing an acceptable distribution system.
In both cases, voltage and loading, planners often come to view that their job is to see
that planned levels of each are well within requirements.
Certainly, voltage must be kept within standards. It is a key aspect of deliverable
service quality. The voltage provided to each customer must lie in that narrow range as
specified by the utility's standards, which in the United States normally correspond to
something close to Range A of ANSI standard C84.1-1989, which lists 126 to 110 volts
as the acceptable range for utilization voltage and 126 to 114 volts as the acceptable
range for service voltage-1 In addition, ve'tage- must be kept within proper range on the
system, too, otherwise electrical equipment will not operate properly or have long,
trouble-free service lifetimes.
Voltage Is a Resource to Be Used Well
But there is much more to voltage engineering than merely keeping voltage within the
specified range for delivery. Voltage is what moves power through a distribution system,
and voltage drop is the inevitable and unavoidable result of distributing power. A planner
can reduce voltage drop through any of many means covered in this chapter, but all cost
money, and many add complexity and additional cost to future operation of the system.
Therefore, voltage, and particularly voltage drop, is best viewed as a resource that
planners can use to move power through a radial distribution system, something to be
husbanded and used to optimal purpose in the layout and execution of a power delivery
plan. A planner who uses voltage drop well and who, most important of all, uses all of the
1
Range A (105% to 95% of nominal) specifies the limits under which most operating conditions
occur. Range B (105.8% to 91.7%, or 127 to 110 volts at the service entrance to 86.7% or 106 volts
minimum for utilization) applies to certain temporary or infrequent operating conditions.

283

284 Chapter 9

voltage range available will repeatedly outperform the planner who does not in terms of
creating workable feeder designs that meet all criteria and have low cost.
Failure to use all of the voltage range available is perhaps the most widespread "poor
practice" in the power delivery industry. If applicable voltage standards permit planners a
full 10% voltage drop (e.g., from 126 to 114 volts on a 120 volt scale), then that full
range should be used before money is spent to reduce voltage drop. But many utility
planners spend money to cut peak voltage drop to 8% or even 7%, convincing themselves
that they are doing a better job by providing improved voltage performance. Those
expenses aren't necessary.
But further, many planners simply fail to treat and engineer voltage drop as a limited
resource, something that must not only be kept within a limit, but also used well. Just
how planners at a utility decide to use the 10% voltage drop available to them has a lot to
do with how inexpensive the plan will be, how well it will perform, and how easy and
inexpensive it will be to upgrade it in the future.
This chapter looks at volt-VAR behavior on distribution systems from the dual
perspective of voltage as a deliverable quality that must meet standards and of voltage as
a resource which the artful planner can use well to minimize cost. It assumes the reader
has an understanding of basic electrical engineering and power systems concepts with
respect to voltage, current, and power factor on a distribution system and the ways to
compute them. Numerous texts cover these basics. Perhaps the best treatment of power
flow, voltage, and loading computations specifically for power distribution systems is in
Kersting and Grigsby (2001).
Section 9.2 begins this chapter with a look at the character of voltage and voltage drop
on typical distribution systems. This provides some insight into the situations normally
faced by distribution planners in attempting to optimize design. Section 9.3 then introduces
the concept of load reach and its use to assure optimum use of voltage and distance
capability in a distribution system. Power factor and volt-VAR control and correction
through the use of shunt capacitors are covered in somewhat more detail in section 9.4.
Section 9.5 is a brief discussion of the similarity between using DG (distributed generation)
for voltage support and shunt capacitors for VAR support in a distribution network. Key
points are summarized and highlighter1 in section 9.6. Some of the concepts introduced in
this chapter will not be fully developed or given in examples until later, in Chapters i 1-18.

9.2 VOLTAGE BEHAVIOR ON A DISTRIBUTION SYSTEM
In a radial or loop power delivery system, electrical power flows out and "down" from ;he
source (substation) to the consumers (load). Although phase angle is a big factor in
movement of power, as is described in any text on power flow, on a distribution system
phase is normally not a major factor, and voltage alone is viewed as associated with power
flow. By contrast, in a true network voltage situation, such as the planning of a high voltage
transmission grid, voltage phase-angle differences among parts of the system are a big part
of the "power flaw problem." In a radial or loop distribution system, phase angle (of the
voltage) does to some extent shift in relation to the decrease in voltage along a feeder, but
voltage phase differences between locations on a radial or loop feeder are not important to
most planning applications. There are exceptions, of course, and planners should stay alert
to phase during studies of switching feasibility.2

2
Large phase-angle differences across an open tie point indicate large tie currents and a possible
inability to open the switch, if closed, due to a current exceeding the switch rating. In fact, a point
distribution engineers and operators often forget is that 30% phase shifts occur due to different

Load Reach and Volt-VAR Engineering 285

3100
3144
7.6

7.2

, 6.8

0 1 2 3
Distance (miles)

Figure 9.1 Profile of voltage on a 12.47 kV (p-p) distribution feeder for a system in the central United
States. This particular display is given in actual phase to ground voltage (vertical scale) and distance in
miles out one particular pathway. It shows all three phases individually (here as solid, dashed, and
dotted lines, but more typically as different colors). Small numbers at points on the display refer to
node numbers in the system being analyzed.

Generally, except for the actions of voltage control devices like regulators placed on the
system at key points, voltage decreases with distance away from the source (the substation)
as well as with the amount of power moved. In any satisfactory system, this decrease is
relatively small, never more than eight to ten percent of the total voltage under normal
conditions.
Voltage on a feeder is often depicted using a voltage profile of voltage at points along
the circuit, as shown in Figures 9.1 and 9.2. The profile in Figure 9.1 shows a number of
features typical of distribution feeder circuit voltage profiles. First, this profile follows the
voltage from the substation out one particular pathway or the feeder - one specific set of
branches. Here, voltage is constrained by high and low limits (shown) that are associated
with both equipment ratings and the highest and lowest P.U. voltages that can be conveyed
to consumers. (These standards will be discussed more in Chapter 20.) Finally, as stated
above voltage generally decreases, or drops, as one moves away from the source, except at
locations that are regulated or boosted, as shown.
Voltage profiles can give voltage in terms of actual voltage (e.g., 7.2 kV p-g) as in
Figure 9.1, per unit, or on a utilization basis (120 volt scale in the U.S., 250 volts in
Europe), as shown in Figure 9.2. Displays of voltage on a per unit or utilization equivalent
basis have a disadvantage in that they do not show actual voltage. But for many planning
purposes they more than make up for that by permitting display of voltage along a path that

transformation can exist across useful tie points. A 12.47 kV feeder from a substation fed by 138 kV,
and a 12.47 kV feeder from a substation fed at 69 kV are likely 30 degrees apart because the 69 kV
line is fed upstream through a 138/69 kV transformation. Such ties are worth less for reliability and
load balancing than ties with very small phase differences, because load transfers can only be done
there through "cold switching" - momentarily interrupting service of the transferred portion of the
feeder as it is switched from one source to another.

286 Chapter 9

17200 17269 17281
125

17205
a 120 17268

o 115

5>110

0 1 2 3 4 5 6
Distance (miles)

Figure 9.2 Results of a "one-line" load flow analysis are displayed here in 120 volt equivalent form
for a 23.8/12.47 kV feeder and one of its secondary circuits. This per-unit display permits the planner
to view "voltage" with respect to a common measure across transformers. Here there are two
transformers. The first, about 2.4 miles out of the substation, is a 23.9/12.47 kV feeder conversion
point. The second is a service transformer (node 17294).

includes transformers. Figure 9.2 shows a circuit path that involves 34.5 kV, a
transformation to 12.47 kV, and finally a transformation to 208 volts (all voltages phase to
phase). Displays in per unit or on a utilization basis allow the planner to see the effects of
the entire pathway including all voltages and transformation, and keep the voltage display
in terms of its use for customer deliver. Displays in absolute voltage (e.g., Figure 9.1), on
the other hand, permit engineers to check actual expected voltage against cable, insulation,
and equipment voltage ratings. Most voltage engineering software tools (e.g., ABB's
CADPAD®, Milsoft's Windmils®) permit the planner to select the basis used for display.
Conductor Tapering and Voltage Drop
Generally, distribution circuits are laid out and engineered with conductor size selected to
be proportional to the load level of each pathway. Portions of the electrical pathways that
have high loadings are built with large conductor, those with smaller loadings have
correspondingly smaller conductor. Conductor size is matched to loading level through a
process of minimizing overall cost, as will be discussed in Chapter 11. This is just sound
engineering, and is almost a universal practice. The only exception of wide application is
that some pathways are reinforced for contingency capability, as will be discussed in
Chapters 13 through 15. A feeder trunk may have "oversized" conductor so that during an
emergency it can pick up additional load to maintain service even though other nearby
feeders are out of service.
Thus, in a well-designed distribution system, conductor size is roughly proportional to
loading level, as shown in Figure 9.3. One must say only "roughly" because typically a
utility has only about three to six "sizes" of conductor or cable available (Chapters 11 and
12 will discuss why and how those sizes can be selected to optimize the benefit from
conductor tapering.)


#2

4/0

636 MCM

4/0

*- #2
B

Figure 93 Conductor tapering. Top, wire size as designed for a typical multi-branch 12.47 kV (7.2
kV nominal phase to ground) feeder serving a 6 MVA peak load in a triangular service area, showing
conductor tapering in its design, except along the three-mile trunk, which is reinforced to "big wire"
size for contingency switching reasons. The myriad laterals stemming from these major three-phase
branches are not shown. Bottom, power flow along these branches during normal peak conditions,
varying from 6 MVA at the substation to very light loading at the ends of each branches. Points A and
B refer to Figure 9.4.

288 Chapter 9

Distance (miles)

Figure 9.4 Voltage profile along the feeder shown in Figure 9.3 for the pathways from the substation
to points A and B. Voltage profile to A shows a gradually decreasing voltage drop per mile because
the trunk, reinforced for contingency needs, is oversized versus normal flow needs. Voltage profile to
point B shows a nearly constant drop/mile.

As a result, a fairly common characteristic of radial feeder systems in North America,
and elsewhere where North American practices are followed, is that voltage drop per mile
is rather constant in a well-designed system, regardless of wire size or loading. Big wire
has big loadings and small wire has small loadings, and both thus have roughly the same
voltage drop per mile, since impedance is inversely related to conductor or cable size.
Figures 9.3 and 9.4 illustrate this concept.
This wire sizing can be optimized to achieve least-cost requirements. The two
exceptions to this rule are:
1. Room for growth. Planners know near-term load growth will require
larger capacity, and thus install larger wire now to accommodate futiue
growth.
2. Contingency capacity. Certain pathways are reinforced to be able to pick
up load if the primary feed pathway is disabled.
Both of these consideration, along with optimizing conductor tapering and ways to lay
out distribution systems so they are economical, reliable, and retain sufficient flexibility
for future growth, will be covered in Chapters 12 through 17.
Regardless, the result of this typical design practice is that the product of current times
resistance (I x R) throughout a system is roughly constant throughout a well laid out
distribution system. Voltage drop due to resistance is therefore rather constant per mile.
In "European" systems, which tend to consist of loop feeders feeding loop sub-feeders
(utilization loops), wire size of the loop in generally constant (each side of the loop is the
contingency path supporting the other side) (see Chapter 13). In this type of system,
voltage profile follows the pattern shown in Figure 9.4 for the pathway to point A, a
gradually decreasing drop per mile as one moves out to the substation.


R and X and Their Ratio
Every circuit has an impedance, a complex value composed of real (resistance, R) and
imaginary (reactive, X) components. R and X are proportional to distance, and are a
function of conductor or cable type and size, and the spacing of the conductors with respect
to one another, the neutral conductor, and ground. Current flow, voltage drop, and
impedance interrelate according to the natural laws of physics for electrical interaction as
described in Kersting and other texts. Electrical losses depend only on R, but loss of voltage
with power flow - voltage drop - depends on both R and X.
Resistance causes voltage drop under any situation where power flows, in any system
be it DC or AC. But for any AC circuit, reactance to power flow also creates voltage
drop. However, while reactance is somewhat like resistance in that it is a function of wire
size and type, it is mostly a function of conductor spacing. Larger phase spacings increase
the reactance, which increases the voltage drop due to reactance. Increase the cross-arm
spacing of phases from nine to ten feet on the pole design of a feeder, and reactance is
increased slightly. Voltage drop is a complex variable, the power factor of the load and
the power factor of the current flow on the feeder interact with R and X to create a
complex voltage drop. Poor power factor shifts the power flow so it is more sensitive to
impedance, worsening voltage drop. The net effect of this is that in most distribution
systems, voltage drop is a function of both R and X, but that X matters most on "big
wire" pathways and resistance matters most for small wire segments.
Every power engineer knows that voltage drop can be reduced on a circuit pathway by
employing larger (and usually more expensive) wire, which will have a lower value of R
and Z and therefore provide less voltage drop for any given level of power flow. But X
limits how well this trick works as one turns to ever bigger wire to solve voltage
problems. For example, R and X for #2 wire (180 amp thermal limit) on a typical
12.47 kV overhead feeder design are respectively 1.69 and .665 ohms mile. "Doubling"
the wire size to 4/0 conductor (340 amp thermal limit) cuts R by nearly two-thirds, to
.592 ohms, but only decreases reactance by 13%, to .581 ohms. Overall impedance is
reduced by 54%, perhaps a good return on the money spent on larger conductor.
This same trick will work once more. If one "doubles" the 4/0 conductor to 477 MCM
which has 670 amps thermal capacity - nearly twice its ampacity - R falls to .216, or
again by nearly 2/3. But reactance drops by only 26%, to .43 ohms. Still, impedance is
cut by 42%, to .48 ohms. Note, however, that reactance is now the larger element of
impedance. R/X was nearly 3:1 for the small conductor but only about 1:1 for this wire
size.
Thus, if one tries this "trick" again, the results are not nearly as useful. Doubling the
ampacity again by selecting 1510 MCM ACSR (1340 amps, or exactly twice the thermal
current capability) means that resistance drops to .072 ohms per mile, or again by nearly
2/3. But reactance falls to only .362 ohms, or by 16%, and impedance drops by only 24%.
R/X ratio is now 5:1, and the planner gets only half the improvement of the previous
doubling step.
Impedance is a function of frequency
Impedance of power equipment has both real and reactive components at all but zero
frequency (i.e., DC). For most power equipment, both the real and reactive components of
impedance depend on the frequency (Figure 9.5). This is seldom an issue for distribution
engineers as their designs are usually restricted to one frequency.

290 Chapter 9

.48

.40

.32

.-
<A

Resistance

.08

30 60 90 120
Frequency - Hz
Figure 9.5 Both R and X can vary as functions of the alternating voltage frequency. Here are
values for 636 MCM 57-strand ACSR overhead conductor. Resistance varies only slightly, but
reactance is very much a function of frequency.

Power Systems Are Designed to a "Constant Voltage" Paradigm
All modern power systems are designed within a constant voltage paradigm. At each
point of delivery, they are intended to provide voltage that does not vary as a function of
time, other loads, or the customer's load level. Most often, they succeed quite well in
these regards. For example, a residential consumer nearly anywhere in the United States
can draw from 1 to at least 10,000 watts of power out of the local power system, under
peak or off-peak conditions, and at any time. And despite all these varying factors, they
never see their service entrance voltage stray by more than about 10%. Most see far less
voltage swing than that. Even at a full 10% shift due to load, voltage change is six orders
of magnitude less than change in load (or current). Real power systems are very close to
being "ideal voltage sources."
A measure of how well this constant voltage ability ±s performed is embodied in
engineering calculations of voltage regulation, a measure often applied to transformers.
There, it is defined as the difference between the voltage provided by the transformer at full
load and at no load (when the input side voltage is held constant).
Percent regulation = (Vf- Vn)/Vf x 100% (9.1)
Where Vf is the voltage at the output side at full (maximum) rated load and Vnis the voltage
when no load is flowing.
Strictly speaking, voltage regulation as represented in nearly all power systems
textbooks is applied only to transformers, for which there is an often-quoted characteristic
value. However, the concept is useful in the design of power distribution systems. The
variation in voltage at a particular point between maximum and no-load load flow cases is
useful in analyzing potential problems and obtaining hints on where potential capability can
be found at low cost.
Voltage Regulation and Flicker Standards
Most utilities have standards that set out the maximum variation in voltage permitted at
any particular customer, often calling for only a 6% range of variation at any service


point, due to all normal operating causes. In addition, utilities often have "flicker"
standards which specify a limit on rapid change of voltage level. A large motor on the
same circuit can make a noticeable change in a consumer's voltage as it turns on and off.
Generally flicker standards are based on what people can perceive in the effects of
voltage change in lighting. Thus, a utility can have three levels of standards that apply,
for example: service entrance voltage will be maintained within a 10% range about
nominal, no customer will see a variation over time of more than 6%, and no customer
will see a change in voltage within a short period of more than 3%. Chapter 20 will
discuss these criteria in more detail.

9.3 LOAD REACH AND DISTRIBUTION CAPABILITY
The purpose of a distribution system is to distribute power by moving it from central
sources to many consumer points. Each circuit is responsible for moving power from one
location to others — for moving power over a distance(s). Voltage engineering includes
many metrics that are useful to planners in laying out, assuring satisfactory performance,
and minimizing the cost of a distribution system. Among these, measures of voltage and
voltage drop themselves are very valuable. The K-factor and other rules of thumb provide
useful insight into circuit capability. Voltage drop/mile and various losses measures
support engineering directed at honing performance and minimizing cost.
But none of these directly measure a system, circuit, or planning alternative against
this most basic power distribution need of moving power over distance. Load reach was
created to do exactly that. Load reach's application is covered in Chapters 11 through 15.
This section will only introduce it as a concept. It is a direct measure of the basic
performance of a power system as constrained by voltage drop and a concept extremely
useful in effective voltage planning and engineering.
Load Reach
Load reach is measured in units of distance - usually kilometers or miles. It is useful as a
planning tool because it measures the distance that the feeder system can move power. It
is useful as a concept because it focuses the planner's attention on the fact that economy
and efficiency must be achieved in a design with respect to the need to move power
certain distances, and not just on a per foot basis.
The load reach of a circuit or design is the distance that it can move a certain amount
of power before encountering the applicable steady-state voltage drop limit. Obviously,
the load reach of a circuit depends on the amount of power being moved. The reach of a
circuit limited to moving 5 MW will be farther than if it is transporting 6 MW. But there
are three specific load reach definitions that are not load-dependant that are extremely
useful to planners.
Thermal load reach
The thermal load reach of a circuit is the distance it can move its rated maximum amount
of power before encountering the applicable voltage drop limit. For example, a balanced
three-phase, overhead crossarm, 12.47 kV circuit with 336 MCM ACSR phase conductor
has a voltage drop of about 298 volts/mile, or 4.1% per mile, when moving its rated
ampacity limit of 530 amps. Thus, its thermal load reach evaluated against a 7.5%
voltage drop criteria is
Thermal load reach = .0757.041 = 1.8 miles

292 Chapter 9

Table 9.1 Thermal Load Reach Distances for
12.47 kV OH 3-P Line Type - Miles
Conductor Thermal Reach
#6 1.8
#2 1.8
4/0 1.8
336 1.8
636 1.6
795 1.5
1113 1.4
1354 1.2
2000 1.1

Table 9.2 Economical Ranges of 12.47 kV
OH Three-Phase Line Types Evaluated for
Capital and Losses Economics - Annual
PeakMW
Conductor Low High
#2 0 1.6
4/0 1.6 3.7
336 3.7 5.1
636 5.1 7.1
795 7.1 9.7
1113 9.7 12.4
1354 12.5 16.4
2000 16.4 21.2

Table 93 Economic Load Reach for 12.47
kV OH 3-P Line Type - Miles
Conductor Economic Reach
#6 3.6
#2 3.5
4/0 3.6
336 3.6
636 3.5
795 3.3
1113 2.7
1354 2.4
2000 2.2


Thermal load reach is nearly a constant
for all distribution conductor sizes
Interestingly, the thermal load reach of various sizes of a conductor of one type (e.g.,
ACSR) is very nearly a constant at any one voltage level. Table 9.1 shows the thermal load
reach of various sizes of distribution-range ACSR conductor at 12.47 kV, which varies only
slightly over a range of conductor from #2 to 636 MCM. If one intends to use conductor to
its utmost capacity, its ability to move power is constant regardless of conductor size: the
thermal load reach of any ACSR distribution wire size is roughly 1.75 miles at 12.47 kV.
Larger wire has a lower impedance and thus less voltage drop per mile when moving
any specific amount of power. But it has a higher ampacity, so a larger wire's thermal limit
is greater. The two almost exactly cancel out: Impedance and ampacity are inversely related
in this range of distribution conductor sizes. This means that their product is constant
regardless of wire size - at least in the distribution range. Figure 9.6 shows ampacity times
impedance for ACSR wire sizes from #2 to 1590 MCM. This product is inversely
proportional to load reach. As discussed earlier, as one leaves "distribution" wire size range
(#6 - 500 MCM) for truly large wire, R/X ratio becomes very much greater than one as
reactance begins to dominate resistance. As a result, as conductor size is increased, the
product increases as shown in Figure 9.6, and load reach falls as listed in Table 9.1.

Economical load reach
Few utilities lay out and plan their distribution systems to be loaded to the thermal limit of
conductor during normal operations. Most electric delivery utilities determine economic
loading ranges for each conductor in an inventory of several sizes they use. These define
preferred loading ranges for each conductor size, based on minimizing lifetime cost, which
is the sum of initial cost of the feeder and the present worth of the future losses associated
with moving power over a typical (usually 30 to 40 year) lifetime. Such economic loading
ranges call for far less peak current than the thermal limit of any conductor. The lifetime
cost of losses is so high that it leads the utility to spend a bit more up front, on large
conductor, to reduce their value. Such expense more than pays for itself. As a general rule
of thumb, for a utility with typical financial factors, the economical loadings for overhead
conductor used in the 15-kV range will be about 1/3 to 1/2 their thermal load limits.
Chapter 11 discusses exactly how to determine such economical loading ranges and gives
several examples and a very detailed discussion of methods to minimize "lifetime cost" of
conductor in this manner. Table 9.2 shows the economic loading ranges for distribution
conductor at 12.47 kV, as determined for one utility. Figure 9.7 shows and explains a plot
of the analysis that forms the basis for this table.
A key fact that emerges from such optimization of conductor loading is that the lifetime
cost per foot is almost linear with loading (see Figure 9.6). This linearity has substantial
implications which will be discussed later in this section and throughout this book. For the
moment, however, the focus is on load reach. Looking at Table 9.2, a 336 MCM three-
phase line has an economical loading range of from 3.7 to 5.1 MW as utilized by this
utility. Evaluated against a 7.5% voltage drop criteria, the load reach at the upper end of this
range, 5.1 MW, is 3.6 miles - this conductor, carrying 5.1 MW, will reach 7.5% voltage
drop in 3.6 miles. That distance is this conductor's economic load reach - the minimum
distance a planner can depend upon this 336 MCM line type to deliver power and stay
within service-quality voltage standards, when utilized according to Table 9.2.

294 Chapter 9

500 -,

400
o
c
re
0> 300
Q.
E
200

o
05
Q. 100
E

200 400 600 800 1000 1200 1400
Ampacity - Amps

t2 4/0 636 795 1192 1590
Wire Size

Figure 9.6 Thermal load reach is basically a distance-based measure inversely proportional to
ampacity times impedance, which is plotted here for wire sizes from #2 to 1590 MCM ACSR. Within
the distribution conductor range, the value varies little, but for larger conductor it increases linearly.

1.5
795/

1113

.5

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Peak Load - MW
Figure 9.7 Analysis of six three-phase line types for overhead 12.47 kV crossarm construction can be
used to identify the range over which each conductor size is most economical. Y-intercept for each
type is its initial construction cost per mile. Variable PW cost of losses over its life has been computed
for loads up to its thermal limit, assuming a 63% annual load factor and 90% power factor. Each
conductor (except 795 MCM, which is a redundant conductor type as will be discussed in Chapter
12), has a range of loadings in which it provides the lowest cost. Guideline tables (e.g., Table 9.2) can
be developed from this type of analysis to direct planners to always select the most cost effective
conductor. Of particular importance, a straight line (dotted) approximates very well the "best that can
be done" from this plot. This "linearized" cost characteristic is a key element of optimizing voltage
and cost in feeder systems as will be discussed at length in later chapters.


Economic load reach is nearly a constant
for all distribution conductor sizes
As was the case with thermal load reach, the economic load reach of conductors is also
very nearly the same regardless of conductor size, at least within the range of wire sizes
normally used in distribution. Until quite large conductor is reached and X/R ratios are
large, economical load reach of the conductors from Table 9.3 is 3.6 miles, or about twice
the thermal load reach. Table 9.3 shows economic load reaches for various size lines at
12.47 kV, as determined for one utility.
Emergency load reach
An important aspect of distribution planning is arranging contingency backup for key
elements of the feeder system, so that by re-switching open and close points, power flow
can be diverted around outaged portions of substations or feeders until they are restored
to service. In such situations, the portion of the feeder system that has "picked up"
additional customer demand is carrying more load than its normal design situation and
therefore voltage drop will be greater than under peak normal conditions. The 336 MCM
three-phase line discussed above might carry 7 MW instead of 5.1 MW. It would
therefore have roughly 40% greater voltage drop per mile. In addition, the maximum
longest pathway in the contingency-switched network would undoubtedly be greater than
normal. Lowest voltage drop could be far less than in a normal situation.
The emergency load reach of a circuit or configuration is the distance it can move
power and still conform to emergency or contingency-situation voltage requirements.
Most utiliti

Power distribution planning_reference_book__second_edition__power_engineering__23_

More Related Content

What's hot (20)

Viewers also liked (19)

Similar to Power distribution planning_reference_book__second_edition__power_engineering__23_ (20)

Recently uploaded (20)

Power distribution planning_reference_book__second_edition__power_engineering__23_