Professional Documents
Culture Documents
EL-2651
Formerly EL-81-16-LD
Research Project 1356-1
Prepared by
Principal Investigators
S. J. Kostyal
T. D. Vismor
R. Billinton
NOTICE
nr THIS REPORT ARE ILLEGIBLE. W
PORTlOk,e
has been
copy to p
ability.
Prepared for
603933' OF THIS
DISCLAIMER
auMJ'Pi Bead.
NOTICE
This report was prepared by the organization(s) named below as an account of work sponsored by the Electric
Power Research Institute, Inc. (EPRI). Neither EPRI, members of EPRI, the organization(s) named below, nor any
person acting on behalf of any of them: (a) makes any warranty, express or implied, with respect to the use of
any information, apparatus, method, or process disclosed in this report or that such use may not infringe private
ly owned rights; or (b) assumes any liabilities with respect to the use of, or for damages resulting from the use
of, any information, apparatus, method, or process disclosed in this report.
Prepared by
Westinghouse Electric Corporation
Pittsburgh, Pennsylvania
ABSTRACT
• Historical assessment
• Predictive assessment
The main objectives of this project were achieved through the development of two
reliability assessment models:
Each model was tested in a utility environment by the Duquesne Light Company and the
Public Service Electric and Gas Company of New Jersey. A survey of 56 diverse util
ities served as a basis for examining current distribution reliability assessment
practices in the electric power industry.
A three-volume final report documents the research performed. This handbook con
tains the computer programs and a practical guide to distribution reliability.
ni'/V
EPRI PERSPECTIVE
PROJECT DESCRIPTION
The decade of the 1950s should be recorded in history as the era when the major
improvements in distribution electric service reliability evolved. Fuse coordi
nation alone provided the single most significant improvement. Innovative relay
concepts, supervisory control, improved construction standards, and mobile radio
communications are other significant factors that have brought the quality of
electric service to a plateau where customers place a high level of dependence and
confidence on electric energy for almost all work functions.
While both utilities and the public are aware of and enjoy this improved service, it
is difficult to compare the improved service with service conditions even 20 years
ago. Both the energy applications and the energy requirements have changed.
Further, few records that quantified service reliability were kept before 1950.
Today's electric utilities spend considerable time, effort, and money monitoring
system performance as an "after-the-fact" activity. The historical data obtained
serve both as a means to analyze what went wrong (and correct it) and as a tool to
predict and control future service reliability.
Adequate, consistent, and timely data, then, are the key to electric system perfor
mance analysis. Distribution, unlike transmission and generation, has suffered
because of the lack of these data. Utilities recognized the importance of history
to predict the future, and in 1977 they issued a Request for Proposal to develop
improved methods to analyze historical performance and to predict reliability of
future systems. This handbook contains the reliability assessment methods and
easily used computer programs to implement the methods developed under RP1356.
PROJECT OBJECTIVES
The objectives of this project are to examine reliability evaluation from two
perspectives: historical and predictive assessments. Historical assessment is
examined from the perspective of reliability assessment methods currently used by
v
the industry and used to develop general outage reporting schemes suitable for a
wide variety of distributing utilities. Predictive assessment is examined in terms
of "risk taking"; i.e.. What are the consequences to implementing or not implement
ing certain system reinforcement strategies? The object is to produce a handbook of
reliability assessment methods and easily used computer programs implementing the
methods.
PROJECT RESULTS
The project objectives were accomplished. Two reliability assessment models were
developed and coded into computer programs: a historical assessment model (HISRAM)
and a predictive assessment model (PRAM). Both computer programs have been success
fully tested and executed by two participating utilities: Duquesne Light Company
and the Public Service Electric and Gas Company. Both programs are designed specif
ically for use by practicing distribution engineers.
EPRI Final Report EL-2018, Volume 1 (RP1359), contains the executive summary of this
project. Volume 2 presents the research undertaken to determine present utility
practices in reliability evaluation and the organizations within the utility that
are supported by or require reliability information. The theoretical and utility
background material on which the EPRI reliability assessment models were based is
included in this volume.
Volume 3 describes HISRAM and PRAM in detail. The discussion of HISRAM focuses on
the data requirements for each of the four levels of data analysis. The multilevel
reporting system is designed in a modular manner so that only data required to
support the desired level of analysis must be collected. The output reports are
formatted to display the analyzed data in terms of the standard Edison Electric
Institute system performance indexes.
HISRAM, PRAM, and the distribution reliability handbook add to the distribution
planner's repertoire of planning tools. The handbook is available from Research
vi
Reports Center, and the computer programs for HISRAM and PRAM are available from the
Electric Power Software Center.
vi i *•////
CONTENTS
Section Page
ix
Section Page
x
Section Pag
APPENDIX A GLOSSARY A-l
APPENDIX B GENERAL FORMULAE FOR CALCULATING SYSTEM INDICES B-l
APPENDIX C BIBLIOGRAPHY ON DISTRIBUTION SYSTEM RELIABILITY EVALUATION C-l
APPENDIX D SUPPLEMENTARY EXAMPLES D-l
xi" Xu
ILLUSTRATIONS
Figure Page
2- 1 Frequency of Use of Various PerformanceIndices asDetermined 2-5
by an EPRI RP 1356-1 Survey of United States Utilities
3- 1 Definitions of Terms Associated with Multiple Restoration 3-3
3-2 Tabulation of Interruptions by Duration andVoltage Class 3-11
3-3 Relationship Between HISRAM Data Requirements and Output
Capabilities 3-13
3-4 Summary of System Performance Indices 3-15
3-5 Joint Frequency Table 3-16
3-6 Cross-Association (Contingency) Table 3-17
3-7 Sample Listing of Incidents 3-18
3-8 Feeder Trouble List 3-19
3-9 Component Failure Statistics 3-20
3-10 Summary of Switching Operations 3-21
3-11 Sample Reporting Form (Level 4) 3-25
3-12 Guidelines for Keypunching Data Recorded on the Incident Report
ing Form 3-26
3- 13 Sample Reporting Form (Level 1) 3-27
4- 1 Sample Random Operating History for Transformer Discussed in
the Text 4-2
4-2 Average Operating History for Transformer with the Random
History Depicted in Figure 4-1 4-3
4-3 Two-State Component Representation 4-4
4-4 The Predictive Model Reduces the System Between Source and Load
to a Single Equivalent Component 4-6
4-5 Two-Component Series System 4-7
4-6 Illustrative Series Configuration 4-10
4-7 Simple, Radial Configuration for Illustrating Fai1ure-Modes-
and-Effects Analysis 4-12
4-8 Simple, Radial Configuration with an AlternateSupply Point 4-15
4-9 Two-Component Parallel Configuration 4-21
4-10 Sample Series - Parallel Configuration 4-22
4-11 One of the Two Parallel Supply Paths of Figure 4-10 4-22
xi i i
Figure Page
4-12 Two-State Weather Model 4-28
4-13 Average Performance Profile 4-28
4-14 Example Illustrating IPPS 4-36
4-15 Example Illustrating EPPS 4-38
4-16 Example Illustrating EIPS 4-39
4-17 Load-Point Indices 4-42
4-18 System Indices 4-43
4-19 Summary of Nodes with Worst Mean Failure Rates 4-44
4-20 Capital Cost Summary 4-45
D-l Looped system D-3
D-2 Looped system with automatic sectionalizing and manual ties D-7
D-3 Spot networks D-9
D-4 Substation with manual throwover switching D-l 2
xiv
TABLES
Tabl e Page
3-1 Incident List for Sample Feeder Serving a Total of 200 Customers
(1000 kVA total connected load) 3-2
3-2 Summary of Information Provided by HISRAM 3-22
3- 3 Data Requirements by Level 3-23
4- 1 Fai 1 ure-Modes-Effects and Analysis of Feeder Depicted in Figure
4-7 4-13
4-2 Bus Statistics for Feeder of Figure 4-7 4-14
4-3 Fai 1 ure-Modes-and-Effects Analysis for Modified Feeder Depicted
in Figure 4-8 4-16
4-4 Fai1ure-Modes-and Effects Analysis for Feeder of Figure 4-7,
Assuming Solidly Connected Laterals 4-17
4-5 Fai1ure-Modes-and-Effects Analysis for Feeder of Figure 4-7
with Imperfect Fusing 4-18
4-6 Summary of Fai 1 ure-Modes-and-Effects Analyses on Feeder of
Figures 4-7 and 4-8 4-20
4-7 Sample Problem Data 4-23
4-8 Sample Problem Data 4-26
4-9 Error Involved in Using Overall Failure Rate Rather than Weather-
Related Rates in Two-Component Parallel Systems 4-32
4-10 Sample Problem Data 4-33
4-11 Data Requirements 4-40
D-l Outage Data for Feeder Depicted in Figure D-l D-3
D-2 Downtime Array D-4
D-3 Analysis of Looped System of Figure D-l D-5
D-4 Summary of Operating Practices for Configuration Shown in
Figure D-2 D-7
D-5 Outage and Switching Data D-8
D-6 Summary of Reliability Calculations D-8
D-7 Outage Data for Figure D-3 D-9
D-8 Analysis of Series Service of Each Substation in Figure D-3 D-10
D-9 Outage Data and Switching Times for Figure D-4 D-12
D-10 Analysis of the Configuration of Figure D-4 D-l 3
xv-Y//
Section 1
SCOPE
This handbook views reliability from a much more restricted perspective which has
its roots in the fields of statistics and probability. Reliability evaluation in
volves observing outages on the distribution system, noting their causes and effects,
compressing this historical data into relevant summary information, and using this
information to improve the performance of existing systems and predict the perfor
mance of proposed systems. To this end, two approaches to reliability evaluation
are utilized:
Predictive reliability techniques are more involved and require more effort in data
collection. Realistic, simplifying assumptions are required for practical analysis
1-1
of distribution systems. Predictive techniques rely on two general classes of in
formation to estimate reliability: component failure statistics and system struc
ture. Each component is characterized by a failure rate and restoration time. Using
system structure and component performance data, predictive methodologies evaluate
the reliability of specific load points on the system.
A hierarchy of uses for data collected by reporting systems was compiled by the
Reliability Task Force of the Edison Electric Institute Transmission and Distribu
tion Committee in 1971 (1_) and is reproduced below. Additional discussion can be
found in Volume 2 (Section 2) of the EPRI RP 1356 Final Report. Reporting systems
may be used to:
1-2
• Provide a basis for individual companies to establish service con
tinuity criteria. Such criteria could then be used to monitor sys
tem performance and to evaluate general policies, practices, stan
dards and design.
1-3
ORGANIZATION OF THIS HANDBOOK
The technical background required to undertake reliability analyses varies with the
type of assessment (historical or predictive) and the complexity of the model. As
stated earlier, historical reliability assessment has its roots in statistics. In
addition to statistics, the field of probability weighs heavily as the basis for
predictive assessment. Some of this background can be gained by consulting the EPRI
1356-1 Final Report; other information can be found in the bibliography in Appendix
C.
This handbook is intended for the distribution engineer with little or no background
in probability and statistics. As mentioned above, rigorous mathematical treatments
can be found elsewhere. The approach used in this book is utilitarian; discussion
is based heavily on examples. The purpose here is to present historical assessment
methodologies in a way that will make the collection and reporting procedure much
more effective and efficient. It is also intended to reduce some of the more com
plex, yet fundamental, predictive techniques to a level that makes simple calcula
tions tractable and easily understood.
1-4
A glossary of terms can be found in Appendix A. Appendix C contains a bibliography
of reliability literature.
REFERENCE
1-5
Section 2
RELIABILITY MEASUREMENT
The ability to weigh the cost of various degrees of system reliability is contingent
upon the ability to make quantitative measurements of reliability. Providing
redundant service paths to each load in a distribution system intuitively appears to
increase distribution-system reliability. However, any analysis of costs versus
benefits requires more than an intuitive feeling of improved reliability. It requires
an answer to the question: How much is system reliability improved by this action?
TYPES OF INDICES
Generally speaking, the reliability of a distribution system, or any system for that
matter, can be described if two questions are answered:
• How long does it take to restore the system after a failure has
occurred?
Frequency and duration are the two fundamental aspects of failure which are addressed
in this section on measuring reliability. It is important to realize that there is
no single measure of reliability which may be used to answer these questions. The
emphasis placed on any one index of service continuity varies with application.
2-1
• Indices of component reliability (component failure statistics)—
indicators which reflect a measure of continuity of service provided
by components.
System indices compiled from historical interruptions on the distribution system have
been the most widely used measures of reliability. Although it is possible to esti
mate system indices (as is shown in Section 4, Predictive Reliability Assessment),
there is a general uneasiness about the additional data required to implement pre
dictive methods. As viewed here, predictive assessment treats the component as the
fundamental historical entity. By correctly combining component reliability parameter
the distribution system can be reduced to a single, equivalent component separating
any given load from its supply. Based on the load-point equivalents (load-point
indices), estimates of system indices can be computed. Although the specific equa
tions used to compute historical and predicted system indices differ, the basic con
cepts are the same.
SYSTEM INDICES
The Edison Electric Institute (EEI) (2_), the Institute of Electrical and Electronics
Engineers (IEEE) (_3), and the Canadian Electric Association (CEA) (4_) have suggested
a wide range of performance indices. These indices are generally yearly averages of
interruption frequency or duration. They attempt to capture the magnitude of dis
turbances by weighting the averages with the number of customers interrupted and/or
the amount of interrupted load. A few of the indices are summarized below.
2-2
§ The System Average Interruption Duration Index (SAIDI) is the
average interruption duration per customer served. It is determined
by dividing the sum of all customer interruption durations during a
year by the number of customers served.
2-3
served. It is found by dividing the annual load interruptions by
the total connected load.
A third category of indices uses both load and number of customers when computing
system statistics.
Figure 2-1 shows the degree to which the various indices are accepted by United
States utilities. Indices chosen for inclusion in Sections 3 and 4 of this handbook
are SAIFI, SAIDI, CAIDI, and ASAI. This reflects the predominance of customer-related
indices in widespread use among utilities. In addition, the two load-based indices
ALIFI and ALIDI are included in later discussion.
System indices have traditionally been used to evaluate past performance. Their role
in historical reliability assessment and illustrative examples are discussed in Sec
tion 3. However, as mentioned earlier, the section on predictive reliability
2-4
CO
o
ce
LU
CQ
Load-point reliability indices can be used to estimate the system indices produced by
historical models. Although similar indices of system behavior are used by both
approaches to reliability assessment, it is highly unlikely that numerical values
resulting from application of both models to the same system would be identical.
Historical assessment generally summarizes discrete incidents occurring at specific
locations over specific time periods; whereas, predictive assessment estimates the
long-term behavior of systems using component parameters that describe the central
tendency of an entire distribution of possible values. For example, it is possible
2-5
to compile historical indices for a specific feeder by recording only those interrup
tions which affect it. However, to estimate similar indices using predictive method
ologies requires incident data from many feeders for an accurate estimate of average
component failure parameters. These component statistics are then combined with
suitable probabilistic models and the feeder configuration to arrive at load-point
indices.
LOAD-POINT INDICES
Load-point indices are not generally used in historical assessment. However, pre
dicted values are useful in locating potential trouble areas or substandard circuits
Measures of load-point reliability used in Section 4 are defined below.
U1 /A (2-10)
eq eg' eq
2-6
• Steady-State Availability (A) is the probability of a load
being supplied at any time. It can also be expressed as the per
centage of demand time which is served.
REFERENCES
2-7
Section 3
Historical assessment has traditionally relied on system indices and various tabula
tions of descriptive characteristics collected for each incident to evaluate past
performance. Many kinds of data can be collected to fulfill many different goals.
Without careful definition of purpose, the time and capital spent on data collection
may outweigh benefits. The question which is often asked and which must be answered
by each utility on an individual basis is: What kinds of data are required to imple
ment a given reliability assessment strategy; that is, what data must be collected
to support a given purpose? This section on historical reliability assessment is
devoted not only to the calculation of performance indices, but also to a general
discussion of the design and use of an incident reporting system which best fulfills
the needs of a given utility.
Before delving into some sample calculations, it is useful to review some character
istics used to describe entries in the incident log. The time interval between the
onset of failure and the repair or replacement of outaged equipment is called restora
tion time. It is common practice among American utilities to restore as many cus
tomers as possible through switching operations before undertaking the time-consuming
task of replacing or repairing damaged equipment. This practice creates a potential
for associating several restoration steps with any given outage.
Figure 3-1 depicts a two-step restoration procedure. Each step consists of a series
of switching operations used to restore service to a block of customers before
3-1
Table 3-1
1ST RESTORATION
STEP
f
v.
r
50 J
TIME (HOURS)
'-------v------ ^ V
2 8
DURATION OF RESTORATION STEP
A, B: TOTAL CUSTOMER INTERRUPTION DURATION FOR CUSTOMERS AFFECTED BY 1ST AND 2ND
RESTORATION STEPS, RESPECTIVELY. A = 2 HRS, B = 2 + 8 = 10 HRS.
Now consider the incident list shown in Table 3-1. Interruption duration and number
of customers affected are shown for each step used in restoration. This example
3-3
illustrates the fundamental data required for calculating historical performance
indices as described below:
40 + 150 + 70 + 40
200
Note that two restoration steps are associated with the second inci
dent: 100 customers were without service for 2 hours, 50 customers
were without service for 10 hours. Each customer, on the average,
was without service for 4.55 hours during the year.
= 200(8760) - 910
(based on 8760 hours/year)
200(8760)
= .999481
This index indicates that 99.948% of demand (customer hours) was met.
Or interpreted another way, the probability that service was avail
able at any time during the year was .99948.
3-4
ni tft = total load interruptions
L total connected load
= 1.3
REPORTING SYSTEMS
The purpose of the remaining discussion of this section is to present some of the
considerations which should be taken into account in the design of a reporting
scheme and to present an example of this design (the Historical Reliability Assess
ment Model, HISRAM, developed under RP 1356-1). As a background for this discussion,
the following few paragraphs highlight some of the common properties of all informa
tion systems.
3-5
Information Systems
An information system is any structure which processes data (input) and produces
information (output). An information system is a filter which accepts input (raw
data) and generates output (information) germaine to decision-making. Because
historical reliability assessment is based on processing collected data, insight into
the nature of an effective reliability assessment system is gained by considering
attributes common to all information systems.
• accuracy,
t completeness,
• relevance,
• timeliness, and
• accessibility.
Completeness. Information systems are designed to aid users when they encounter
a specified set of decision-making situations. Completeness refers to the extent
to which available information actually satisfies the needs of system users.
Completeness does not imply large volumes of output. A complete information
system includes all output that is necessary to support design decisions and
only that necessary set of information. For example, a reliability information
system designed to support probabilistic models is incomplete if it cannot pro
duce component failure rates. The inability of such a system to produce cus
tomer billing information does not indicate incompleteness.
3-6
Relevance. Information systems produce output on a regular basis or when
queried by users. Relevance is associated with the degree to which system out
put matches user requests for information. Relevant output displays the infor
mation that the user has requested and only the information that the user has
requested. For example, an employee in the public relations department of a
utility may require the number of service interruptions on Feeder A during
1979. If the only information that the employee can request consists of a
listing of all distribution system interruptions in 1979, the relevance of the
available information is quite low.
• filtering,
• variance reporting,
$ modeling,
• interrogation.
3-7
Appropriate combinations of these dissemination techniques allow the development of
information systems that are both relevant and complete.
3-8
release. Automatic notification imples that the system stores data until some
predetermined time. At this point, data is processed and resulting information
is distributed to the appropriate users. The reliability information systems
of most utilities employ this method by issuing annual and subannual performance
summaries.
3-9
Input and Output of a Reliability Information System
Data requirements vary with the goals of the reporting system. However, a general
summary of the kinds of data which may be collected is useful. Description of an
incident (interruption or outage) generally requires data about precipitating events
and the corresponding action which was taken (restoration).
Output generated by the information system generally consists of incident lists and
performance tables. Classifications of output are summarized below:
3-10
EPAI TEST UTILITY
SAMPLE DATA INCIDENT TEST EON 1977 DATA
GENERAL TABULATION OF DATA
2.3 KV 0 0 0 0 0 0 0 0 0 0 0 0 0.00
3-11
4 KV 0 1 0 0 1 1 0 0 0 0 0 0 240.33
12 KV 0 0 0 0 0 0 0 0 0 0 0 0 0.00
13.2/23 K>/ 0 0 0 0 0 0 0 0 0 0 0 0 0.00
23 KV 0 0 0 0 0 0 0 0 0 0 0 0 0.00
23D - 4 KV 0 1 1 0 0 0 0 0 0 0 0 0 99.00
23T - 4 KV 0 0 0 0 0 0 0 0 0 0 0 0 0.00
23T - 23D KV 0 0 0 0 0 0 0 0 0 0 0 0 0.00
69 KV 0 0 0 0 0 0 0 0 0 0 0 0 0.00
138 KV 0 0 0 0 0 0 0 0 0 0 0 0 0.00
OTHER 0 0 0 0 0 0 0 0 0 0 0 0 0.00
total 0 2 10 0 0 0 0 0 0 183.80
Structure* •
In addition to the flexibility built into the structure of HISRAM, the program was
designed for portability. The major element in making HISRAM portable is a self-
contained data management system. This data manager only manipulates the specific
3-12
HISRAM OUTPUT REPORTS
DATA FROM
PERFORMANCE
OUTAGE REPORTING SYSTEM INDICIES
ADDITIONAL
RESTORATION SEQUENCE
DATA
CUSTOMER
FEEDER, LOCATION, COMMENTS
TOTALS
CONNECTED
SUBDIVISIONS PERFORMANCE LOAD
SUMMARY
TABLES
COMPONENT
CAUSE
3-13
POPULATIONS
WEATHER
WEATHER
CYCLE
COMPONENT FAILURE
(DEVICE TYPE)
ISOLATION SEQUENCE
(DEVICE TYPE)
IMPROPER ISOLATION
(DEVICE TYPE)
COMPONENT
RELIABILITY
PARAMETERS
Bigure 3-3. Relationship Between HISRAM Data Requirements and Output Capabilities.
data file required by HISRAM. It is not a generalized data base; however, it does
allow utilities that do not have a data base manager to implement HISRAM. For those
utilities which do have a data base manager, HISRAM is designed in a modular fashion
that allows calls to HISRAM1s data file to be conveniently replaced by calls to the
utility data base.
Output
Of course, each type of information has its own data requirements. Users must
evaluate the kinds of reliability information they require and the resources avail
able for data collection to get the maximum amount of information possible. To aid
in this evaluation, the list depicted above has been divided into the four levels
described earlier based on data requirements and is reproduced in Tables 3-2 and 3-3.
Input
The contents of Table 3-3, a listing of required data by level, are divided into two
general classifications:
3-14
EPRI TEST UTILITY
SAMPLE DATA INCIDENT TEST FOR 1977 DATA
GENERAL TABULATION OF DATA
NORTHERN 0 0 10 0 1 0 0 0 0 0 0 233.50
central 0 10 0 1 0 0 0 0 0 0 0 189.00
southern 0 10 0 0 0 0 0 0 0 0 0 74.00
TOTAL 0 2 10 0 0 0 0 0 0 183.80
6608 L 71 3
1700 G05010212647 WEATHER WRMWDY RN OTHER/UNKNWN CUTOUT NOT LISTED
6606MILLDAM RD
1 3HRS . OMlNS• 1 CUST. 5KVA
TIME PERIOD 1
1/ 1/77 TO 12/31/77
FEEDER TROUBLE LIST
NUMBER OF
FEEDER NO. INCIDENTS
L713 3
D675 2
V701 2
L 790 2
V216 2
V*33 2
V3C9 2
V573 2
D713 2
vacs 2
D681 2
0693 2
0920 2
V292
0676
0680
0683
0694
0704
0711
0714
0740
07r j
0 762
0769
0770
0803
0821
E162
E68S
L620
L622
L640
3-19
EPRI TEST UTILITY
SAMPLE DATA INCIDENT TEST FOR 1977 DATA
COMPONENT ANALYSIS
FAILING COMPONENTS
-- AVE. DURATION —---------------------- OUTAGE RATE - - --------------
TOTAL ----- NUMBER OF OUTAGES ------ N-PLAN PLAN. TOTAL NORMAL ADVERS. PLAN. TOTAL
COMPONENT QUANT. NORMAL ADVER. PLAN. TOTAL (HRS./OUTAGE) (OUTAGES/UNIT/YEAR)
1 TP10H CABLE 13540 0 0 0 0 0.000 0.000 0.000 0.000000 0.000000 0.000000 0.000000
2 TP10M SM.COM 34900 6 0 0 6 4.947 0.000 4.947 .000173 0.000000 0.000000 .000173
3 TP10H LG.CON 24620 0 0 0 0 0.000 0.000 0.000 0.000000 0.000000 0.000000 0.000000
A TP1UG CABLE 35330 2 0 0 2 5.775 0.000 5.775 .000057 0.000000 0.000000 .000057
TIME PERIOD 1
1/ 1/77 AVERAGE SWITCHING TIME .651 HRS.
12/31/77
ISOLATING COMPONENTS
----------R £‘ S T 0 R A T 1 ON------ ---------PRO T E C T I ON------
TOTAL TOTAL AVERAGE NU'iBER OF NUMBER OF PROBABILITY
NUMBER OF SWITCHING SWITCHING PROPER IMPROPER OF FAILURE
COMPONENT SWITCHINGS TIME(MIN) TIME(HR) OPERATIONS OPERATIONS (O-MODE)
3-21
Table 3-2
Levels
12 3 4
Li sting of Incidents
Including Location and Comments X X X X
Including Identification (e.g., serial number)
of Outaged Equipment X
Performance Indices
Tabulated for Entire System X X X X
Tabulated by User-defined Subsystems XXX
Performance Tables
Feeder Trouble List X X X X
3-22
Table 3-3
Levels
12 3 4
Reported Data
Restoration Steps (Date, Time, kVA, Customers) X X X X
Feeder X X X X
Location# 0 0 0 0
Comment 0 0 0 0
Subdivision of System XXX
Cause XXX
Weather X X
Outage Equipment X X
Identification (e.g., serial number) of Outaged Equipment# 0
Protection Device Which Operated to Clear Circuit of Outage X
Protection Device Which Should Have Cleared Circuit of Outage
(if different from above)+ 0
Protection Device Used in Switching (each restoration step)+ 0
User Defined X
3-23
HISRAM's primary source of data is the incident data recorded in the field (reported
data). Also, small amounts of data must be collected from secondary sources such as
circuit maps and the National Weather Service.
Once a decision has been made on the level of implementation, HISRAM may be used to
generate the corresponding reporting form. Figure 3-11 depicts an example of an out
age reporting form produced at level four. (Figure 3-13 shows a similar form pro
duced at level 1, for comparison.) Although the structure of this report is fixed,
the definition of subdivisions, causes, weather, equipment, protection devices, etc.,
is specified by each utility. Thus, each utility uses a unique form which reflects
its own descriptive qualities. In addition to the reporting form, HISRAM produces a
corresponding set of keypunching guidelines (Figure 3-12) for transferring data from
the reporting form to 80-column cards.
CONCLUDING REMARKS
For additional discussion of HISRAM, consult the User's Manual for the Historical
Reliability Assessment Model. Additional information on historical reliability
assessment may be found in the EPRI RP 1356-1 Final Report, and in the bibliography
contained in Appendix C.
3-24
< > *£PORT NUMCEB
COWENT
RESTORATION OF SERVICE
INTERRUPTED
DATE TINE FEEDER NO ) SWITCHING
( ) ( ) LOCATION ( ) OPERATION
RESTORED NO. OF RESTORED KVA LOAD RESTORE RESTORE
DATE TIME CUSTOMERS RESTORED COMPONENT COMPONENT
1 t
) ( ) < > ( ) ( ) ( )
2 ( ) ( ) ( ) ( ) ( ) ( )
3 < ) ( ) ( ) ( ) ( > ( )
4 ( ) ( ) ( } < ) ( > ( )
5 ( ) ( ) ( ) ( ) ( ) ( )
REFER TO SEC. 5
1 SUBDIVISIONS
( ) DIVISIONS
10100 NORTHERN 10200 CENTRAL 10300 SOUTHERN
C ) VOLTAGES
20100 2.3 KV 20200 4 KV 20300 12 KV
20400 1.2/23 KV 20500 23 KV 20600 23D - 4 KV
20700 23T - 4 KV 20800 23T - 23D KV 20900 69 KV
21000 138 KV 21100 OTHER
2 CAUSE
( ) CAUSE OF THE INCIDENT
1 DISORD/EMGCY 2 EXT.INTERFER 3 CONSTRUCTION
4 DELAYS 5 ERRORS - MAN 6 REQUEST
7 CUST.EQUIPMT 8 INCORRECT OP 9 THERM.-EQUIP
10 ELEC.CONDITN 11 MECH.CONDITN 12 OPER.CONDITN
13 M/0 CHANGE 14 P.S. FAILURE 15 testing
16 UNKNOWN 17 CONTAMINATE 18 WEATHER
19 OTHER 20 PLANNED
3 WEATHER
* FAILING COMP
( ) FAILING COMPONENT
< ) 10 NUPBER OF CONPONENT THAT FAILED
1 TP10H CABLE 2 TP10H SM.CON 3 TP10H LG.CON
4 TP1UG CABLE 5 TP1 SU6MAR. 6 TP1 POTHEAO
7 TP20H CABLE 8 TP20H SM.CON 9 TP20H LG.CON
10 TP2UG CABLE 11 TP2 SU6MAR. 12 TP2 POTHEAD
13 STEP TRANSF. 14 REGULATOR 15 sue TRANSF.
16 OTHER/UNKNWN
5 PROTECTION
( ) PROTECTIVE CONPONENT THAT ISOLATED THE INCIDENT
( > PROTECTIVE CONPONENT THAT FAILED TO ISOLATE THE INCIDENT,IF AN*
1 FUSE 2 RECLOSER 3 SECT '.ONALIZR
4 SECTL.SWITCH 5 CUTOUT 6 SUB BREAKER
7 SUPPLY DROP 8 OTHER/UNKWN
7 DEPARTMENT
( > DEPARTMENT
1 SO.*LD. 2 ss.*s. 3 PS.
4 T.*D. 5 MOT LISTED
1- 0 REPORT MA9ER
7-12 DATE - SECOND STEP
13-18 TINE - SECOND STEP
10-24 BUK3ER OF CUST. - SECOND STEP
25-30 KVA OF LOAD - SECOND STEP
31-30 PROTECTION SECOND STEP
37-42 PROTECTION SECOND STEP
43-48 DATE - THIRD STEP
40-54 TINE - THIRD STEP
55-00 NUTOER OF CUST. - THIRD STEP
01-00 KVA OF LOAD - THIRD STEP
07-72 PROTECTION THIRD STEP
73-78 PROTECTION THIRD STEP
CO CONTINUATION FLAG
1- 0 REPORT NUK3ER
7-12 DATE - FOURTH STEP
13-18 TINE - FOURTH STEP
10-24 NUH8ER Of CUST. - FOURTH STEP
25-30 KVA Of LOAD - FOURTH STEP
31-30 PROTECTION FOURTH STEP
37-42 PROTECTION FOURTH STEP
43-48 DATE - FIFTH STEP
40-54 TINE - FIFTH STEP
55-00 NUMBER OF CUST. - FIFTH STEP
01-00 KVA OF LOAD - FIFTH STEP
07-72 PROTECTION FIFTH STEP
73-78 PROTECTION FIFTH STEP
80 CONTINUATION FLAG
COMMENT CARD
1- 0 REPORT NUMBER
7-80 COMMENTS
1- 0 REPORT NUMBER
7-12 DIVISIONS
13-18 VOLTAGES
43-45 CAUSE
40-48 WEATHER
49-51 COMPONENT INVOLVED IN INCIDENT
52-03 ID OF COMPONENT INVOLVED IN INCIDENT
04-00 PROT. COMP. THAT CLEARED INCIDENT
07-09 PROT. COMP. THAT SHOULD HAVE CLEARED INCIDENT
70-'2 DEPARTMENT
3-26
( ) import Nufieta
COMHEWT
•ESTORATION Of SERVICE
IWTEaeUPTED
DATE TIME EEEDER MO .( )
( ) ( ) LOCATION ( )
RESTORED MO. Of RESTORED KVA LOAD
CATE TIME C'JSTCRTERS RESTORED
1 ( > ( ) < ) ( )
2 < 3 ( ) ( > ! )
3 ( 3 ( > ( ) ( )
4 ( ) ( > ( ) ( >
5 ( ) ( ) { ) ( )
3-27
Section 4
COMPONENT RELIABILITY
4-1
4 HOURS 6 HOURS 14 HOURS
TRANSFORMER OPERATING
TRANSFORMER FAILED
INSTALLATION 1ST FAILURE 2ND FAILURE 3RD FAILURE
The random failure history for a component can be described by average values. For
example, although it is difficult to predict time-to-failure, it is possible to cal
culate an average time-to-failure based on past performance. For the transformer
operating history shown in Figure 4-1,
= 1 + .5 + 1.5
3
= 8768 hrs
4-2
r = Mean Time to Restore (MUR)
= T - m
= 8 hrs
Figure 4-2. Average Operating History for Transformer With the Random
History Depicted in Figure 4-1.
The random cycle of failure and restoration may also be viewed from a slightly dif
ferent perspective. Figure 4-3 shows a two-state component representation; that is,
the component is either available (operating) or unavailable (failed). What are
the rates of transition between the two states? If m is the MTTF, then
is the failure (or hazard) rate. This is the rate of transition from the operating
state to the failed state. Similarly,
is the rate of transition from the failed state to the operating state; y is called
the restoration hazard rate. Also,
f = 1/T = 1/MTBF
4-3
A = 1/m
= 1/8760
u = 1/r
= 1/8
= .125 restoration/hr
f = 1/T
= 1/8768
Note that failure rate and failure frequency are approximately equal. Because they
are nearly equal, the terms "failure rate" and "failure frequency" are sometimes
used interchangeably. There are, however, conceptual differences, as illustrated
below.
1 _ 1
MTTF + MTTR MTBF
4-4
If MTTF » MTTR, then
1 1
MTTF + MTTR ~ MTTF '
Thus,
1 1
or A as f.
MTTF ~ MTBF
u = m = MTTF
(4-3)
A + y m + r MTBF
Because there are only two component states in this model, unavailability is:
A r MTTR
U = 1 - A = (4-4)
A + p m + r MTBF
MTTF
.999088 and
MTBF
MTTR
.000912
MTBF
A comparison between the component reliability parameters just presented and the
load-point indices introduced in Section 2 reveals some striking similarities. Load-
point indices are calculated by reducing the system between the source and a load
into a single equivalent component (Figure 4-4) whose characteristics describe the
reliability of service to the given load. These equivalent-component (load-point)
indices are completely analogous to the component parameters discussed above.
For example, A denotes the failure rate of a component. The analogous load-point
index is called mean failure rate (denoted A^ for "equivalent-component failure
rate") and represents the average rate of interruption to a given load. It is
important to realize that the distinction between failure frequency and failure
4-5
A
DISTRIBUTION
4-6
figure 4-4. The Predictive Model Reduces the System Between Source and Load to a
Single Equivalent Component.
rate remains in effect. Because mean time to restore is generally much less than
mean time to failure, it is assumed that the frequency of failure and the mean
failure rate are approximately equal. Thus, mean failure rate is used to estimate
system indices which involve frequency (refer to the sample calculations which ap
pear on pages 4-12 and 4-13).
Consider the series system of Figure 4-5. For the load to be served, both components
must operate properly. One technique for evaluating the reliability at the load is
to combine the two components into one equivalent component. Computationally, this
reduction is performed in the following manner.
4-7
Aeq A1 * A2 (4-6)
(4-7)
Ueq = U1 + U2 - Ut ■ U2
Xeq = X1 + X2 (4-8)
Vl + X2r2
(4-10)
eq A-J + A2
Since this equation, in effect, neglects the possibility of failure to the remaining
component after one component has already failed, it is an exact equation and not an
approximation in situations where this type of component behavior holds. Under
these conditions, however, component behavior is not independent, and the system
availabilities cannot normally be obtained by simple multiplication of the component
availabilities.
4-8
Consider the two-component system of Figure 4-5 with
A1 = 1.0 f/yr
x2 = 2.0 f/yr
r2 = 4.0 hrs
A2 = 0.999088 U2 = 0.000912
Now applying Eqs. 4-6 - 4-9, properties of the equivalent component are determined.
A = 0.998177
eq
U = 0.001823
eq
A =3.0 f/yr
eq J
r = 5.3358 hrs/failure
eq
If the MTTF of the equivalent component (system) is much greater than its MTTR, then
the average annual outage time can be approximated by
U' = A r (4-11
eq eq eq
which is equivalent to
Computing the restoration and annual outage times with the approximate Eqs. 4-10
and 4-11 yields
r = 5.333
eq
U' = 15.99
eq
4-9
Failure Modes and Effects Analysis of Series Configurations
Armed with Eqs. 4-8, 4-10, 4-12, and an understanding of the term "series" in
reliability applications, it is possible to attack many of the reliability problems
encountered in distribution systems. From the perspective of reliability analysis,
a group of n components are in series if failure of any member of the group
renders the remaining components nonfunctional. Consider Figure 4-6. Since a fault
on any of the six feeder segments causes the feeder breaker to deenergize the entire
circuit, all six components are considered to be in series during a reliability
analysis.
Eqs. 4-8, 4-10, and 4-12 describe the reduction of two series components to a single
equivalent component. They are a special case of the n-component reduction equa
tions :
n
A
eq
2 A.
1 (4-13)
i=l
n
2 Vi
i =l
r (4-14)
eq
iV'
4-10
n
u eq Airi (4-15)
i=l
The general relationship among A, r, and U'described in Eq. 4-11 remains valid;
that is,
U' = A r
eq eq eq
The phrase "failure modes and effects analysis," as applied in this report, simply
means that given a system configuration for study:
For hand calculations, the simplest and clearest approach to failure modes and
effects analysis is to display the process in tabular form. The following series
of problems illustrates this procedure.
Consider Figure 4-7. In this configuration, all switches are normally closed and
the customer load points A, B, and C are supplied from the primary main by fused
laterals. The feeder breaker and the substation supply bus are assumed to be fully
reliable. Component and switching data obtained from historical information are as
follows. For simplicity, all switching operations require the same amount of time:
• Switching - 0.5 hr
Load point A is affected by failures on the 2-mile, 3-mile, and 1-mile sections of
primary main and by failure of the 3-mile primary lateral from which it is served.
Failure of any of these components causes the breaker to open or a fuse to blow.
The mean failure rate, A , for load A is the sum of the failure rates of these
components.
4-11
2 MILES
3 MILES 1 MILE
=1.35 f/yr
The average annual downtime (U' ) is not simply the sum of individual component
downtimes as given by Eq. 4-15. Due to the existence of sectionalizing devices on
the main, load A will not be without service for the entire restoration time (3 hrs)
required by the 3-mile and 1-mile sections of primary main. By opening the appropri
ate switch and closing the breaker, downtime is limited to .5 hour. Thus, the average
annual downtime for load A is 1.55 hrs. as shown below.
=1.55 hrs/yr
4-12
Using Eq. 4-11,
^req^A ^eq^A^^eq^A
= 1.55/1.35
=1.15 hrs/f
These results are summarized in Table 4-1, which also contains a similar analysis
for loads B and C. Note that average restoration time, r^, is not simply a straight
forward sum of corresponding component values (Eqs. 4-11 and 4-14).
Table 4-1
Primary Main
2 m section 0.2 3.0 0.6 0.2 3.0 0.6 0.2 3.0 0.6
3 m section 0.3 0.5 0.15 0.3 3.0 0.9 0.3 3.0 0.9
1 m section 0.1 0.5 0.05 0.1 0.5 0.05 0.1 3.0 0.3
Primary Lateral
3 m section 0.75 1.0 0.75 __ __ __ __ __ __
2 m section — — —
0.5 1.0 0.5 --- --- —
1 m section — — —
ZZJ1 —- —- 0.25 1.0 0.25
Total Effect 1.35 1.15 1.55 1.1 1.86 2.05 0.85 2.41 2.05
It can be seen that load point C, despite being at the extremity of the primary main,
has the lowest failure rate due to its relatively short primary lateral. It has the
longest average restoration time, however, due to the fact that all restoration is by
repair rather than by isolation of the faulted section and restoration by switching
action. In the case of load point A, any failures on the primary main other than on
the initial 2-mile section involve restoration by switching rather than by repair.
4-13
Predictive Methods for Estimating System Indices
System indices for the sample feeder (Figure 4-7) can be estimated from the load-
point indices of Table 4-1 and the bus data of Table 4-2. As illustrated below,
system indices are weighted averages of predicted load-point indices.
Table 4-2
1.23
= 1.74
= 1.42
= .999802
4-14
total load interruptions
total connected load
= 1.18
= 1.80
Now, consider the effect of an alternate source on the load-point and system indices
just calculated; Figure 4-8 shows the modified feeder. Although failures on the
primary main still interrupt all three loads, downtimes at B and C are improved by
the ability to backfeed. For example, the switch nearer to the breaker can be opened
to isolate failures on the 2-mile section of primary main. Closing the normally
open alternate supply path provides for backfeeding to loads B and C. This analysis
assumes that a switching sequence which involves closing the tie switch and opening
an isolation device requires 1 hour; other switching sequences require .5 hour, as
before. Calculations are summarized in Table 4-3.
2 MILES
3 MILES 1 MILE
4-15
Table 4-3
Primary Main
2 m section 0.2 3.0 0.6 0.2 1.0 0.2 0.2 1.0 0.2
3 m section 0.3 0.5 0.15 0.3 3.0 0.9 0.3 1.0 0.3
1 m section 0.1 0.5 0.05 0.1 0.5 0.05 0.1 3.0 0.3
Primary Lateral
3 m section 0.75 1.0 0.75 — — —
--- — —
2 m section — — —
0.5 1.0 0.5 — — —
1 m section — —
—- —
0.25 1.0 0.25
Total Effect 1.35 1.15 1.55 1.1 1.5 1.65 0.85 1 .24 1.05
Based on these load-point indices and the bus data of Table 4-2, the system indices
are:
SAIFI = 1.23
SAIDI = 1.51
CAIDI = 1.23
ASAI = .999827
ALIFI = 1.18
AUDI = 1.50
Table 4-3 illustrates that the load point failure rates are not affected by the
ability to backfeed from an alternate configuration. This applies in all cases in
which the restoration of service is done manually. If automatic switching is used
and customer outage time is considered to be so short that the event is not classed
as a failure, then the overall failure rate will be reduced to a value closely
related to the primary lateral value. However, the ability to backfeed has a pro
nounced effect on the length of the interruption particularly for those customers
at the extremities of the primary main.
4-16
Effect of Solidly Connected Laterals
It can be seen from the cases studied that the load point failure rates are dependent
upon the components exposed to failure and the degree of automatic isolation of a
failed component in the network. This effect can be easily seen in the network of
Figure 4-7. If each lateral is solidly connected to the primary main, all load
points will have the same failure rate, as any fault will cause tripping of the
feeder breaker. The analysis in this case is shown in Table 4-4.
Table 4-4
Primary Main
2 m section 0.2 3.0 0.6 0.2 3.0 0.6 0.2 3.0 0.6
3 m section 0.3 0.5 0.15 0.3 3.0 0.9 0.3 3.0 0.9
1 m section 0.1 0.5 0.05 0.1 0.5 0.05 0.1 3.0 0.3
Primary Lateral
3 m section 0.75 1.0 0.75 0.75 1.0 0.75 0.75 1.0 0.75
2 m section 0.5 0.5 0.25 0.5 1.0 0.5 0.5 1.0 0.5
1 m section 0.25 0.5 0.125 0.25 0.5 0.125 0.25 1.0 0.25
Total Effect 2.10 0.92 1.925 2.10 1.39 2.925 2.10 1.57 3.30
SAIFI = 2.10
SAIDI = 2.35
CAIDI = 1.12
ASAI = .999732
ALIFI = 2.10
ALIDI = 2.49
4-17
Effect of Improper Operation of Protection Devices
The results shown in Table 4-1 illustrate the effect on the load point failure rates
of increasing the exposure to failure of the overall configuration. The results
shown in Table 4-1 illustrate the effect of perfect isolation arising from a failure
on a primary lateral. The probability associated with successful isolation of a
primary lateral fault depends upon the design of the protection coordination scheme
and on the operation and maintenance of the scheme. Table 4-5 summarizes the calcu
lations for the case in which the probability of successful isolation of a primary
lateral fault is 0.9.
Table 4-5
Primary Main
2 m section 0.2 3.0 0.6 0.2 3.0 0.6 0.2 3.0 0.6
3 m section 0.3 0.5 0.15 0.3 3.0 0.9 0.3 3.0 0.9
1 m section 0.1 0.5 0.05 0.1 0.5 0.05 0.1 3.0 0.3
Primary Lateral
3 m section 0.75 1.0 0.75 0.075 0.5 0.0375 0.075 0.5 0.0375
2 m section 0.05 0.5 0.025 0.5 1.0 0.5 0.05 0.5 0.025
1 m section 0.025 0.5 0.0125 0.025 0.5 0.0125 0.25 1.0 0.25
Total Effect 1.425 1.114 1.5875 1.20 1.75 2.10 0.975 2.17 2.1125
4-18
(x = (2 mi)(.1 f/mi/yr) + (3 mi)(.l f/mi/yr) + (1 mi)(.l f/mi/yr)
eq A
+ (3 mi)(.25 f/mi/yr) + (l-.9)(2 mi)(.25 f/mi/yr)
= 1.425 f/yr
Service to load A can be restored by switching operations (.5 hr) for failures on the
1 mi and 3 mi sections of primary main and on the 1 mi and 2 mi laterals. Hence,
(u; )A = (-2 f/yr)(3 hrs/f) + (.3 f/yr)(.5 hr/f) + (.1 f/yr)(.5 hr/f)
= 1.5875 hr/yr
^eq^A (Ueq^A/^eqU
= 1.114
SAIFI = 1.31
SAIDI =1.78
CAIDI =1.36
ASAI = .999797
ALIFI = 1.28
ALIDI = 1.85
Comparing Table 4-5 with Table 4-7 it is seen that imperfect fusing causes failures
on all laterals to contribute to load point unreliability. For example, the two
mile lateral section adds (l-.9)(.5) or .05 to load point A's failure rate in Table
4-5. Load point A was unaffected by this lateral in Table 4-2. Table 4-6 depicts
a summary of the four preceding failure-modes-and-effects analyses.
4-19
Table 4-6
CASE
la 2b 3C 4d
Load Point A
Load Point B
Load Point C
System Indices
4-20
Two-Component Parallel System
The configuration shown in Figure 4-9 consists of two parallel components. If only
one of the components is actually required to carry the total load, then the parallel
structure is said to be "redundant." If both components must be available, then in
a reliability sense, this is a series configuration regardless of the actual physi
cal connections.
redundant configurations,
U = U, • u0 (4-16)
eq 1 2
Aeq = A1 + A2 - A1 A2 (4-17)
A1 (^1 ^2^
Xeq (4-18)
1 + Alrl + J2r2
r 1 r2
req (4-19)
rl + r2
0.000428 A1 = A2 = 0-999572
U = 0.183184 x 10'6
e9 yi-on
Xeq = (0.5) (0.5) ^ = 0.0004281 f/yr
r = = 3.75 hrs/failure
eq 2
28 KV
4 KV
28 KV
Figure 4-11. One of the Two Parallel Supply Paths of Figure 4-10.
Each supply path consists of two circuit breakers, a transformer, and a section of
line. Component characteristics are summarized in Table 4-7. ATI breaker failures
are assumed to be short-circuit failures; stuck-breaker modes are ignored. It is
assumed that the 28-kV bus is completely reliable.
4-22
Table 4-7
Based on component failure statistics tabulated in Table 4-7, the equivalent failure
rate for one supply path is
= 0.5432 f/yr
= 17.72 hrs/failure
U A •r = 9.62 hr/yr
eq eq eq
Now, using Eq. 4-19 and Eq. 4-20, an equivalent for two parallel supply paths can be
derived:
= 0.0019 f/yr
r 17.71
eq = 8.86 hrs
2
4-23
The 4-kV bus is in series with the parallel configuration. Thus,
= 4.46 hrs
= 0.0557 hrs/yr
It can clearly be seen that the reliability of the 4-kV bus completely dominates the
load point indices because of the series element aspect. This eliminates much of
the benefit provided by redundancy. In many systems, recognition and awareness of
series element dominance can save considerable computational effort.
Scheduled-Outage Considerations
In the case of a series system, a scheduled outage will result in loss of customer
service. With proper coordination, however, customers can be notified prior to sus
pension of service. The scheduled outages can be directly combined with the forced-
outage indices to obtain a set of total outage indices at the load point. The two
situations, though both resulting in loss of service, are not the same. One is a
random event, and the other is a scheduled event. This should be clearly understood
before combining the two to produce an overall set of statistics.
This difference becomes more important when considering a parallel redundant con
figuration such as that shown in Figure 4-9. It is assumed that a scheduled outage
would not take place if one component is already on a scheduled or a forced outage.
4-24
A forced outage could, however, occur on one component while the other component is
on either a scheduled or a forced outage.
Consider the system shown in Figure 4-9 where, in addition to the A-j, A^, r-j, and
component statistics, the components have maintenance or scheduled outage rates
a!|', x'2 and average maintenance outage times of r!j and r^. The rate at which a
forced outage will overlap a scheduled outage is A^ where
The average duration of the event in which a forced outage overlaps a scheduled out
age is given by
AiA2ri . rl r2
rm x‘]A2rl + >l2Alr2 rl + r2
(4-22)
A2Alr2 r2 rl
AlA2rl + X2Alr2 r2 + rl
am = 2a"Ar" (4-23)
r» = .r x. (4-24)
m r" + r
Consider the system shown in Figure 4-10 and add the additional scheduled outage data
shown in Table 4-8.
4-25
Table 4-8
Assuming that no simultaneous maintenance is done, the scheduled outage rate for one
side is 9.0 occurrences/year. The average scheduled outage time is given by
= 7.42 hrs/occurrence
= 17.71 x 7.42
5.23 hrs
m 17.71 + 7.42
These indices do represent random outages at the load point and can therefore be
combined with the indices obtained earlier for a forced outage overlapping another
forced outage to obtain overall load point indices. Therefore,
= 0.01249 + 0.00828
= 0.02077 f/yr
= 4.77 hrs/occurrence
4-26
U1 = 0.099 hrs/yr
The contribution to the overall load point failure rate due to a forced outage over
lapping a scheduled outage is quite significant in this system, particularly when
it is noted that the primary contributor to the load point failure rate due entirely
to a forced outage event is the 0.0113 f/yr for the 4-kV bus. This is a single con
tingency event, i.e., a series element in an otherwise redundant configuration. The
contribution due to a forced outage overlapping a scheduled outage could be reduced
by coordinating the maintenance activity and reducing the number of times one side
is down for scheduled maintenance.
and
8.2
A" = 2 4.0 x 0.5432 x = 0.00407 f/yr
m 8760
11 = 17.71 x 8.2
5.61 hrs
m 17.71 + 8.2
Scheduled outages are normally used to perform preventive maintenance. If the fre
quency of this activity is too low, the associated component failure rate will in
crease, resulting in an overall decrease in load point reliability.
Standard 346-73 (4J subdivides the outdoor environment in which a distribution com
ponent resides into the three weather classifications of normal, adverse, and major
storm disaster. There are obviously an infinite number of possible classifications
rather than just the three discrete levels selected. The difficulties in using a
large number arise in both system modeling and in data collection and validation.
The major storm disaster category is a special situation which is reserved for major
system difficulties. The random behavior pattern associated with the two-state
weather model involving normal and adverse weather is shown in Figure 4-12 for a
single component. Normal-weather failure rate is designated A and adverse-weather
failure rate is designated A1.
4-27
ADVERSE WEATHER
FAILURE
RATE
NORMAL WEATHER
-------------------------- ►
► TIME
There are many weather situations which could be considered to be adverse. The key
factor is that adverse weather is deemed to be "weather conditions which cause an
abnormally high rate of forced outages for exposed components during the period
such conditions persist." These situations are then grouped into a single classi
fication. The duration of the normal weather periods and the adverse weather periods
is considered to be a random phenomenon and can be described by an average value.
Let
FAILURE
RATE
■►TIME
4-28
The overall average failure rate xaw is usually the only statistic available-and,
in the case of a series system, is all that is required. It describes the annual
failure rate of the component or system. If two components are placed in parallel,
a load point prediction based upon average component failure rates can be quite in
correct if the adverse weather failure rate is significant. This is due to a phe
nomenon known as "failure bunching." Both components exist in the same environment
and therefore, for the relative short duration of the storm, both exist at the
stress level associated with A', the adverse weather forced outage failure rate.
The relative magnitudes of A' and A can be seen from a simple example.
Assume that
N = 200 hrs
S = 1.5 hrs
A'
N + S
If the fraction of total failures in adverse weather is F and this value is known,
then
N + S
A = A. 0 - F)
av
+ S
A' = A. [F)
av
If F = 0.5, then
201.5
A1 = 0.5 0.5 = 33.5833 f/yr of adverse weather
1.5
It can be clearly seen that during the short duration of a storm, the adverse
weather failure rate of each component is substantially higher than the overall
average annual value. Failure bunching does not imply dependency between two com
ponents. The only dependency is in terms of the common environment within which
they both reside. Individual component failures are assumed to be independent.
4-29
Series Configurations. An equivalent adverse-weather and normal-weather failure
rate can be calculated for several components in series to obtain an equivalent ele
ment.
n
A = 2 A, f/yr of normal weather
eq k=l K
n
A1 = 2 f/yr of adverse weather
eq k=1 k
Let
Ap = A + S
where
N
A = X1X2 rl + r2/ + N XlX2rl + XlX2r2
N + S
S 1 + xiX2r2 ]
B = 2A-j A^S + A^ A^r
N + S
and
A rlr2 B rlr2 , ,
r„ = Lrl + r2 J
A + B A + B ri + r2 ^
A = 2Ar
N + S
x + lA'
4-30
S
B = 2A A' S + Ar
N + S
B
r S
P A + B
Consider an example consisting of two identical components with the following char
acteristics.
N = 200 hrs,
r = 7.5 hrs.
7.5 200
A = 2 (0.2519) 0.2519 + (33.5833)
8760 201.5
= 0.000216
1.5
B = 2 (33.5833) fojtT 33.5833 + 0.2519
8760
= 0.002983
Ap = A + B = 0.003199 f/yr
r
7.5 0.002983
P 2 0.003199
= 5.15 hrs
4-31
If an Error Factor E is defined on the ratio of the two, then
0.003199
E = 7.47
0.000428
Error depends on the contribution to the individual component indices and the over
all system failure rate made by adverse-weather failures. Table 4-9 shows this
variation as a function of the percentage of individual component failures occurring
during adverse weather.
Table 4-9
Percentage of Component
Failures During Adverse
Weather Failures/yr. Error Factor
The A and B factors are also required if it becomes necessary to use the two-
component formula to obtain an equivalent component to combine with a third compo
nent. This approach, however, does not completely recognize the failure bunching
in the three-component system, and it is more accurate to use the equations developed
for a three-component parallel redundant system given in (11).
4-32
For the line
The normal and adverse weather failure rates for each component and for each series
segment are shown in Table 4-10.
Table 4-10
X = A + B
P
xp = 0.00403 f/yr
17.72 ,
r ''oo4~fcr (1,5) = 10-12 hrs/failure
P 2
The 4-kV bus statistics must now be included in the analysis as a series element.
The load point statistics now become
4-33
r = (.00403)(10.12) + (0.0113)(4) = 5_6, hrs/fa(,ure
eq .01533
a; = 2 (9.0 x 0.2856 x
= 0.004354 f/yr
The overall load-point indices, including permanent forced outages resulting from
normal and adverse weather and scheduled outage activity, are
(.01533)(5.61) + (.004354)(5.23)
r = 5.525 hrs/failure
eq .01968
The load point indices are still dominated by the series effect of the 4-kV bus. As
the percentage of component failures in adverse weather increases, the normal weather
failure rate decreases and, therefore, so does the overlapping effect of forced and
scheduled outage events, provided that no scheduled activity is done in adverse
weather. In the limit, if all failures occurred during adverse weather, then a
forced outage could not occur while maintenance activity is being conducted because
the normal-weather failure rate is effectively zero.
4-34
THE EPRI RP 1356-1 PREDICTIVE RELIABILITY ASSESSMENT MODEL (PRAM)
The approach used to describe PRAM is intended to emphasize the fundamental concept
of continuity of the electrical path between a source and load. Continuity between
a load and an electrical source is provided by an uninterrupted electrical path
between them. For a radial electrical system, failure of any component in the con
tinuous path between source and load (called the backpath) causes an interruption
to the load.
Each PSM (IPPS, EPPS, and EIPS, respectively) brings results a step closer to reality
by attempting to provide for an additional mode of protection system operation. How
ever, each model also requires more effort in collecting outage data.
4-35
Implicit Perfect Protection System
Consider, for example, the load X in the feeder configuration of Figure 4-14(a).
Because failure or improper operation of existing protection equipment is ignored,
the only components on which the integrity of the backpath depends are the shaded
(a) (b)
4-36
lines indicated in Figure 4-14(b). Outages on either of these two lines will break
the continuity of service to load X.
Temporary Restoration
Modeling the explicit location of automatic and manual protection devices introduces
a new twist to the PSM. Interruption of a load is no longer restricted to the
failure of components in the backpath. As a result, the service interruption ex
perienced at a load point as a result of component failure may not necessarily last
for the duration of the entire repair/replacement procedure. Temporary restoration
via an alternate source or the original source is a possible means of reducing the
downtime at a load.
Both EPPS and EIPS are based on explicit location of protection devices. Conse
quently, temporary restoration is a prime consideration in the normal restoration
process. PRAM evaluates possible temporary restoration via alternate feeds, as
well as the original source; however, restoration to the original source is the
preferred means. It may be possible to temporarily restore service to a load via
an alternate feed even if a failed device lies in the backpath.
In this case, the reliability of service to load X of Figure 4-15(a) is more real
istically modeled by including the effects of an explicitly defined protection
system. Although protection/isolation equipment is still assumed perfect, its
operation in response to system outages generally results in an increase in extent
over that considered by IPPS. Load X can be interrupted due to operation of the
breaker or fuse (which lies in the supply path to X) in response to component out
ages which do not occur in the backpath. Component outages which contribute to the
reliability of service to load X are shown by the shaded lines in Figure 4-15(b).
4-37
BREAKER
FUSE
SWITCH
4-38
Input
• bus-related data.
4-39
Table 4-11
DATA REQUIREMENTS
PSM
DATA IPPS EPPS EIPS
System Topology
Connectivity X X X
Substations X X X
Location of Protection Devices X X
Alternate Feeds X X
Bus-Related Data
Connected Load (kVA)* X X X
Number of Customers* X X X
Component Description
Nonprotection Components
Capital Cost* X X X
Failure Rate X X X
Restoration Time X X X
Protection Devices
Capital Cost* X X
Failure Rate (A-mode) X
Restoration Time (A-mode) X
Probability of Failure (q-mode) X
Manual Switching Time X X
4-40
Output
Based on the PSM's presented above, PRAM calculates indices of the reliability of
service to each load in a radial distribution system (Figure 4-17). In effect, for
each load point, PRAM reduces the distribution system to a single component with an
equivalent failure rate and restoration time. Average annual downtime, average
annual uptime and steady-state availability (all of which are really very similar
indices) are also calculated.
PRAM is also capable of calculating (Figure 4-18) the system indices presented in
this handbook. Calculations are based on averages of equivalent load-point indices
weighted by connected load or number of customers served at each bus.
In addition to load-point indices and system indices, PRAM provides other useful
outputs. Nodes are sorted by index and then listed in order of increasing relia
bility (Figure 4-19). This class of output is referred to as a worst-case sort.
A user can provide capital equipment costs as program input. PRAM responds by com
piling a summary of capital costs (Figure 4-20), by substation, for the configura
tion at hand. No attempt is made to provide any analysis of reliability worth.
4-41
last cast SMStanoM eacaKtas omit
aCLlAVUITV •V *061
7000 1.9712 5.7773 17.0? .waosr 7001 7.975? 5.7273 17.02 8742.98 .999057
not 2.9712 3.7??3 17.07 710? 2.9757 5.7723 17.02 87*2.98 .990057
7103 *.9*29 6.9377 33.11 • 7i6.B9 .99t>i20 7104 4.8428 8.6377 33.11 8728.89 .998220
7103 *.9*29 • .8377 33.11 B?i6.»9 .996ii0 7108 4.8428 8.8377 33.11 8728.89 .996220
7199 2.9712 5.7773 17.07 em.9» .99twr 7701 2.9752 5.7223 17.02 8742.98 .996057
7?0J 7.9757 5.7773 17.07 B712.99 .99S017 7203 2.9752 5.7223 17.02 6742.98 .9915057
720* 7.9757 5.7773 17.07 971.1.99 .999017 7205 2.9752 5.7223 17.02 0742.98 .99005’
7301 7.9757 5.7773 17.07 97k2.99 .999017 7307 2.9752 5.7223 17.02 87*2.96 .99605 ’
4-42
7303 7.9757 3.7773 17.0? 97*1.99 .999017 7304 7.9752 5.7223 17.02 07*2.98 .99005 7
7305 7.9757 1.7223 17.07 97*1.99 .999017 7306 2.9752 5.7223 17.02 87*2.96 ^eos*
7401 7.9757 5.7773 17.07 97U.99 .999017 7407 2.9752 5.7723 17.02 07*2.99 .99805 7
7403 2.9712 1.7223 17.07 97*i.99 .999017 7404 2.9752 5.7223 17.02 87*2.90 .998057
7405 7.9757 1.7223 17.07 97*i.99 .999017 7408 7.9752 5.7223 17.02 87*2.90 .996057
7407 7.9757 1.7223 17.0? 97*2.99 .999017 7408 7.9752 5.7223 17.02 87*2.90 .996057
•000 7.9757 1.7223 17.0? 97*2.99 .999017 8101 2.9752 5.7223 17.02 87*2.98 .990057
• 107 2.9712 1.7223 17.0? 97*2.99 .999017 8103 2.9752 5.7723 17.02 67*2.96 .996057
• 104 3.9112 7.8715 79.38 9710.92 .99M*9 8105 3.8552 7.8215 29.38 6730.82 .9900*6
• 701 2.9712 5.7773 17.07 97*2.99 .999017 8702 2.9752 5.7223 17.02 07*2.98 .990057
• 703 2.9712 3.7773 17.0? 97*2.99 .999017 8204 2.9752 5.7223 17.02 67*2.98 .990057
•703 7.9757 5.7773 17.0? 97*2.99 .999017 8208 2.9752 5.7223 17.02 87*2.98 .998057
• 707 7.975? 5.7773 17.0? 97*2.99 .999017 8208 2.9752 5.7223 17.02 87*2.90 .998057
• 301 7.9757 5.7773 17.0? 97*2.99 .999017 8302 4.8426 8.8377 33.11 0728.89 .9902.:O
• 305 4.847a 8.8377 33.11 9726.99 .996220 8308 4.8*28 8.8377 33.11 6728.89 .9902^0
• 307 4.4*78 8.8577 33.11 9726.99 .996220 8308 4.6*28 8.8377 33.11 8720.89 .990220
• 309 4.8478 8.8377 33.11 ' 72- 89 .996220 8*01 2.9752 5.7223 17.02 67*2.90 .99005 7
.
<fAILUAES/TEAR)
.
(HOUXS/YEAR)
HOAST-CASI SUNHAAT
153*0 .1231 OIOS .5526 0200 .5126 15302 .4326 0100 .3826 0206 .3201
usor .3651 10105 .3526 20106 .3501 OIOS .3401
5. C. F. DeSieno and L. L. Stine. "A Probability Method for Determining the Relia
bility of Electric Power Systems." IEEE Transactions on Power Apparatus and
Systems, Vol. 83, February 1964, pp. 174-179.
10. J. Endrenyi. Reliability Modeling in Electric Power Systems. New York, New
York: John Wiley and Sons, 1978.
4-46
Appendix A
GLOSSARY
The following definitions are reproduced from IEEE Standard 346-1973, which contains
terms for reporting and analyzing outages of electrical transmission and distribu
tion facilities and interruptions to customer services.
adverse weather - designates weather conditions which cause an abnormally high rate
of forced outages for exposed components during the periods such conditions
persist, but do not qualify as major storm disasters. Adverse weather condi
tions can be defined for a particular system by selecting the proper values
and combinations of conditions reported by the Weather Bureau: thunderstorms,
tornadoes, wind velocities, precipitation, temperature, etc.
adverse weather permanent forced outage rate - for a particular type of component,
the mean number of outages per unit of adverse weather exposure time per com
ponent.
exposure time - the time during which a component is performing its intended func
tion and is subject to outage.
forced outage - an outage that results from emergency conditions directly associated
with a component requiring that it be taken out of service immediately, either
automatically or as soon as switching operations can be performed, or an outage
caused by improper operation of equipment or human error.
major storm disaster - Designa-tes weather which exceeds design limits of plant and
which satisfies all of the following: (1) extensive mechanical damage to
plant; (2) more than a specified percentage of customers out of service; (3)
service restoration times longer than a specified time.
NOTE: It is suggested that the specified percentage of customers out of
service and restoration times be 10 percent and 24 hours. Percentage of cus
tomers out of service may be related to a company operating area rather than
to an entire company. Examples of major storm disasters are hurricanes and
major ice storms.
A-l
momentary interruption - an interruption of duration limited to the period required
to restore service by automatic or supervisory-control 1ed switching operations
or by manual switching at locations where an operator is immediately avail
able.
NOTE: Such switching operations must be completed in a specified time not to
exceed 5 minutes.
normal weather - includes all weather not designated as adverse or major storm
disaster.
normal weather permanent forced outage rate - for a particular type of component,
the mean number of outages per unit of normal weather exposure time per com
ponent.
outage duration - the period from the initiation of an outage until the affected
component or its replacement once again becomes available to perform its
intended function.
NOTE: Outage durations may be defined for specific types of outages; for
example, permanent forced outage duration, transient forced outage duration,
and scheduled outage duration.
outage rate - for a particular classification of outage and type of component, the
mean number of outages per unit exposure time per component.
NOTE: Outage rates may be defined for specific weather conditions and type
of outages. For example, permanent forced outage rates may be separated into
adverse weather permanent forced outage rate and normal weather permanent
forced outage rate.
permanent forced outage duration - the period from the initiation of the outage
until the component is replaced or repaired.
scheduled outage duration - the period from the initiation of the outage until
construction, preventive maintenance, or repair work is completed.
A-2
sustained interruption - a sustained interruption is any interruption not classified
as a momentary interruption.
switching time - the period from the time a switching operation is required due to
a forced outage until that switching operation is performed.
NOTE: For example, switching operations include reclosing a circuit breaker
after a trip-out, opening or closing a sectionalizing switch or circuit
breaker, or replacing a fuse link.
transient forced outage duration - the period from the initiation of the outage
until the component is restored to service by switching or fuse replacement.
NOTE: Transient forced outage duration is really switching time.
A-3
Appendix B
The calculation of system indices can be generalized into formulae which are easily
programmed on a digital computer. Those formulae are summarized below.
SAIFI = (B-l)
m ki
2 2 c T
i=l j=1 J J
SAIDI (B-2)
C
ki
m
2 2 Cij Tj
CAIDI = i=l j=l (B-3)
m
ci
i =l
m ki
8760C - 2 2
ASAI = ij 1J
i=l j=l (B-4)
8760C
m ki
1
= 1 -
8760C 2 :
i=l J
m
2 L,1
ALIFI i=l (B-5)
L
m ki
2 2 1 i j Tij
ALIDI = i=l j=l (B-6)
B-l
m = number of interruptions in subdivision (feeder, station, operating
district) for a given time period
k.j = number of restoration steps associated with i^ interruption
Note:
ki
:i = 2 cii
j=l J
. th
= total connected load (kVA) interrupted by i interruption
Note:
ki
4 - £ bo
th
1-. = connected load (kVA) restored during j restoration step
'J
J
Tij 2 S’k
k=l
2 ^eq^b ^b
beB ecl D D
SAIFI (B-7)
2 C
b
beB
2 (U eq^b
beB
SAIDI (B-8)
2
beB
B-2
2 (U1 )k C,
eq;b b
beB
CAIDI = (B-9)
2 ^eq^b Cb
beB
[8760 - <ueq>b]
2
beB
ASAI = (B-10)
8760 Cb
beB
2
beB
ALIFI = (B-ll)
2 L
b
beB
2
beB
(u;q)b
ALIDI (B-12)
2
beB
B-3
Appendix C
INTRODUCTION
• Outage Data
t Publications List
• Abstracts
The first section lists all selected publications. These publications are ordered
alphabetically by primary author. The abstracts section contains short summaries
of the most significant articles found in the publications list.
Publications List
C-l
Allan, R. N., and M. F. De Oliveira. Reliability Modeling and Evaluation of
Transmission and Distribution Systems. Proceedings IEE, Vol. 124, No. 6, June
1 977.
C-2
Capra, R. L., M. W. Gangel, and S. V. Lyon. Underground Distribution System Design
for Reliability. IEEE Transactions, PAS-88, June 1969, pp. 834-842.
C-3
Endrenyi, J., L. E. Payne, and L. Wang. A Computer Program for the Reliability
Evaluation of Power Transmission Systems. CEA Transactions, Vol. 13, Pt. 3, 1974,
Paper No. 74-SP-l51.
Good, R. D. Radio Control Cuts Interruption Time. Electrical World, January 1972,
pp. 38-40.
C-4
Heising, C. R., R. J. Ringlee, and H. 0. Simmons, Jr. A Look at Substation
Reliabi1ity. American Power Conference, April 1969.
Neagle, N. M., and D. R. Nelson. The Use of Probability in the Design and Operation
of Secondary Network Systems. AIEE Transactions, PAS-74, August 1955, pp. 567-575.
C-5
Patton, A. D. Substation Reliability Estimation is a Useful Design Tool. Electric
Light and Power, October 1969.
Ramamoorti, M., and Balgopal. Block Diagram Approach to Power System Reliability.
IEEE Transactions, PAS-89, May/June 1979, pp. 802-811.
Van Miegroet, P., and G. Dienne. Degradation Level: Method of Representing the
Reliability of Small Electric Circuit Arrangements. CIRED-73, International
Conference on Electricity Distribution.
C-6
Abstracts
C-7
Billinton, R., and M. S. Grover. Quantitative Evaluation of Permanent Outages in
Distribution Systems. IEEE Transactions, PAS-94, May/June 1975, pp. 733-741.
This paper conducts a basic investigation of the techniques proposed in
Power System Reliability - I - Measure of Reliability and Methods of
Calculation, by Gaver, Montmeat, and Patton, and Transmission System
Reliability Evaluation Using Markov Processes, by Billinton and
Bollinger, and proposes additional equations to remove some of the
inconsistencies which previously existed. Two sets of equations form
ulated in this paper consider the occurrence and non-occurrence of
repair during adverse weather periods. The fluctuating weather con
siderations are extended to three components in parallel. The minimal
cut set method is used to evaluate the reliability of a simple series
parallel configuration.
C-8
D'Agostino, R., R. L. Adamson, R. F. Mann, M. A. Lozano, and D. L. Hopkins.
Service Availability as an Approach to Selecting an Optimum Underground Distribution
System Design, Part I - Analysis of Service Continuity. IEEE Paper No. 68 CP 669-
PWR.
This paper utilizes simple series system reliability concepts to examine
nine basic feeder designs. The applications are practical and require a
minimum of mathematical manipulation. Forced and planned outages are
examined. The effects of component failure rates on the calculated
customer indices are examined together with a brief study of the restora
tion activities which form the customer outage duration.
C-9
Heising, C. R., R. J. Ringlee, and H. 0. Simons, Jr. A Look At Substation
Reliability ■ American Power Conference, April 1969.
This paper provides a relatively simple application of basic reliability
concepts to substation assessment. Industry statistics of substation
component failures and malfunctions are used to make predictions of the
frequency and duration of station caused circuit interruptions. A
reliability and availability analysis of a breaker and one half scheme
as opposed to a single bus with tie breaker scheme is used as an example.
The breaker and one half scheme is shown to be the more reliable arrange
ment using the data provided. The paper introduces the concept of com
ponent failure modes and effect analysis in substation analysis.
C-10
Mallard, S. A., and V. C. Thomas. A Method for Calculating Transmission System
Reliability. IEEE Transactions, PAS-87, March 1968, pp. 824-833.
This paper presents by application to a relatively simple configuration,
a proposed method for analyzing the reliability of a transmission system.
The approach considers generation and transmission system elements,
weather conditions, load cycle, generation dispatch, interconnections
and the effect of scheduled outages. The basic objective is to obtain
indices which describe the reliability of station supply to the low-
voltage buses. Forced and scheduled outage data for the elements used in
the configuration example are provided. The mathematical concepts
utilized in the paper are based upon the approach presented in Power
System Reliability - I - Measure of Reliability and Methods of Calculation,
by Gaver, Montmeat, and Patton, and Power System Reliability - II -
Applications and a Computer Program, by Montmeat, Patton, Zemkoski,
and Gumming. Additional equations and the use of contingency curves
are presented in order to extend the application to non series-parallel
configurations.
C-ll
provided in a more condensed form in Reliability Studies Guide Circuit
Design, by Schwobel, Shula, Gilligan, and Wood. These papers illustrate
the development of reliability profiles for each customer in the new
underground system. Reliability evaluation is used to examine alternate
equipment configurations and to add an important dimension to the evalua
tion of alternate proposals. The paper uses a minimum of mathematical
concepts and illustrates practical application.
OUTAGE DATA
Publications List
EEI Transmission and Distribution Committee. Guide for Reliability Measurement and
Data Collection, October, 1971.
Fifth Annual Report. URD Equipment and Materials Reliability in the Northwest.
Undergrounding, January/February 1973, p. 12.
Fourth Annual Report. URD Equipment and Materials Reliability in the Northwest.
Undergrounding, March/April 1972, p. 16.
C-l 2
Heising, C. R. Reliability of Electric Power Transmission and Distribution
Equipment. Twenty-Eighth Annual Technical Conference Transactions of the
American Society for Quality Control, Boston, Massachusetts, May 1974, pp. 314-319.
IEEE Committee Report. Definitions of Customer and Load Reliability Indices for
Evaluating Electric Power System Performance. IEEE Paper A 75 588-4.
IEEE Committee Report. Proposed Definitions of Terms for Reporting and Analyzing
Outages of Electrical Transmission and Distribution Facilities and Interruptions.
IEEE Transactions, PAS-87, May 1968, pp. 1318-1323. —— __
NELPA Eighth Annual Report. Northwest Utilities Report URD Product Reliability.
Transmission and DistributionT June 1976, pp. 18-23. ——— —
NELPA Sixth Annual Report. URD Equipment and Materials Reliability in the Northwest.
Undergrounding, March/April 1974, p. 62.
C-l 3
Patton, A. D. Determination and Analysis of Data for Reliability Studies. IEEE
Transactions, PAS-87, No. I, January 1968, pp. 84-100.
Paulson, N. L., and W. L. Carey. Outage Analysis Spots Trouble Areas. Electrical
World, March 21, 1966, pp. 88-93.
Abstracts
Ford, D. V. The British Electricity Boards National Fault and Interruption Reporting
Scheme - Objectives, Development and Operating Experience. IEEE Transactions, PAS-
91, No. 5, September/October 1972, pp. 2179-2188.
The British Electricity Board established a National Fault and
Interruption Reporting Scheme in 1965. It had been working for six
years at the time this paper was published. The paper describes the
Scheme and presents some of the first results from it. The objectives
and scope are described in detail together with the basic elements
involved in the reporting procedure. A description of the two annual
reports produced is provided. The paper also provides annual failure
rates for selected major components over the six year reporting period.
C-l 4
It identifies composite systems and major components of transmission and
distribution equipment for which outage data are recorded. The paper
proposes a standard component list and a set of descriptions which would
enhance data pooling and thereby increase the data base. It was sug
gested that the paper could be used as a guide by the utility industry
in setting up a standard transmission and distribution equipment outage
data bank.
IEEE Committee Report. Definition of Customer and Load Reliability Indices for
Evaluating Electric Power System Performance. IEEE Paper A 75 588-4.
This report was prepared by the Working Group on Performance Records for
Optimizing System Design of the IEEE Power Systems Engineering Committee.
It contains definitions of customer and load point reliability indices
commonly used for performance assessment of electric utility systems. A
primary objective of the paper is to illustrate clearly how each of these
indices was calculated. This is accomplished by using a simple numerical
example in which each index is illustrated and calculated.
C-l 5
IEEE Standard Definitions in Power Operations Terminology, 2. Terms for Reporting
and Analyzing Outages of Electrical Transmission and Distribution Facilities and
Interruptions to Customer Service. IEEE Standard 346-1973.
This IEEE Standard arose from the work done by the Joint Working Group on
Performance Records for Optimizing System Design of the Subcommittees on
Application of Probability Methods and System Planning of the IEEE Power
System Engineering Committee. The first publication of this information
was as Proposed Definitions of Terms for Reporting and Analyzing Outages
of Electrical Transmission and Distribution Facilities and Interruptions,
by IEEE Committee Report. This material plus the subsequent discussion
led to the present standard. Their basic purpose is to foster uniformity
and standardization of language among engineers engaged in reporting,
analyzing and predicting outages of transmission and distribution facili
ties and interruptions to customers. The definitions do not provide
specific instructions for reporting outages and interruptions. The
standardization of basic terms should, however, enhance the exchange of
data between different utilities and data collection agencies.
Publications List
C-l 6
Arceri, J. A., and E. T. Parascos. Reliability Engineering and Underground
Equipment Failure, Cost and Manufacturer's Analysis. IEEE Conference Record 76
CHI 119-7-PWR, pp. 579-782.
Berrie, T. W. How to Work Out What Quality of Electricity Supply We Can Afford.
Electrical Review, Vol. 202, No. 3, January 20, 1978.
IEEE Committee. Report on Reliability Survey of Industrial Plants - Part II: Cost
of Power Outages, Plant Restart Time, Critical Service Loss Duration Time, and Type
of Loads Lost Versus Time of Power Outages. IEEE Transactions on Industry Applica
tions, Vol. 1A-10, March/April 1974, pp. 236-241.
C-l 7
Shula, W. E., and J. H. Easley. Cost and Reliability Evaluation of Four Underground
Primary Distribution Feeder Plans. IEEE Conference Record No. 74 CH.
Shipley, R. B., A. D. Patton, and J. S. Denison. Power Reliability Cost vs. Worth.
IEEE Transactions, PAS-91, No. 5, September/October 1972, pp. 2204-2212.
Skof, L. V. Customer Interruption Costs Vary Widely. Electrical World, July 15,
1977, pp. 64-65.
Abstracts
C-18
IEEE Committee. Report on Reliability Survey of Industrial Plants - Part II: Cost
of Power Outages, Riant Restart Time, Critical Service Loss Duration Time, and Ty'p¥
of Loads Lost Versus Time of Power Outages. IEEE Transactions on Industry Applica-
tions, Vol. 1A-10, March/April 1974, pp. 236-241.
This report was compiled by the Reliability Subcommittee of the IEEE
Industrial and Commercial Power System Reliability Committee. The re
port was prepared following a survey of industrial plants in the United
States and Canada. Data from 30 companies covering 69 plants were used.
C-l 9
Appendix D
SUPPLEMENTARY EXAMPLES
Examples contained in this appendix are intended to provide additional insight into
the application of predictive methods to different configurations. To simplify the
discussion and to highlight the affects of system structure on reliability, the
following assumptions are used.
A simple looped system with manual switching is shown in Figure D-l. Failure rates
for corresponding components are tabulated in Table D-l. Note the inclusion of
failure statistics for the subtransmission supply. These numbers can be viewed as
equivalent load-point indices for the subtransmission system. In fact, they may
be calculated by applying the principles of Section 4 to the analysis of subtrans
mission reliability. Thus, distribution reliability indices can be adjusted to
include the effects of failures in the bulk supply system.
D-l
(2) make repairs on A. Given an outage on B, restoration involves three steps:
(1) Restore customers on section A by opening the switch between A and B and closing
the feeder breaker, (2) restore section C by switching, and (3) make repairs on B.
These effects are summarized in Table D-2.
Because the feeder breaker operates to clear outages on all sections, all load
points have the same interruption rate:
= 2.01
= 3.52 hrs/yr
Calculations of the remaining load-point indices (by feeder section) are summarized
in Table D-3. Corresponding system indices are tabulated below.
SAIFI = 2.01
SAIDI = 3.03
CAIDI =1.50
ASAI = .998370
Each substation transformer normally serves two feeders and operates at 50% of rat
ing. Each is also supplied from a different subtransmission circuit. In instances
of transformer failure or loss of supply, the entire load of four feeders is served
through the one operating transformer.
D-2
Figure D-l. Looped system
Table D-l
D-3
Table D-2
Downtime Array
*Restoration steps:
Step 1 - Open switch nearer breaker and close breaker (.5 hr)
Supply .10 2.0 0.2 2.0 0.2 2.0 0.2 2.0 0.2 2.0 0.2 2.0 0.2
Transformer .01 12.0 0.12 12.0 0.12 12.0 0.12 12.0 0.12 12.0 0.12 12.0 0.12
Section A .45 4.0 1.80 1.5 0.675 1.5 0.675 0.5 0.225 0.5 0.225 0.5 0.225
Section B .10 0.5 0.05 4.5 0.45 1.5 0.15 0.5 0.05 0.5 0.05 0.5 0.05
Section C .15 0.5 0.75 0.5 0.075 3.5 0.525 0.5 0.075 0.5 0.075 0.5 0.075
Section D .60 0.5 0.30 0.5 0.30 0.5 0.30 4.0 2.4 1.5 0.90 1 .5 0.90
Section E .25 0.5 0.125 0.5 0.125 0.5 0.125 0.5 0.125 4.5 1 .125 1 .5 0.375
Section F .35 0.5 0.175 0.5 0.175 0.5 0.175 0.5 0.175 0.5 0.175 4.5 1 .575
Total Effect 2.01 1.75 3.52 1.05 2.12 1.13 2.27 1 .68 3.37 1.43 2.87 1 .75 3.52
The automatic sectionalizing scheme is designed so that the feeder breaker opens
for faults on Section A and the feeder recloser operates for outages on Section B.
Corresponding restoration procedures are summarized in Table D-4. For example,
given a fault on Section A, the feeder breaker will lock open. Due to the loss of
potential, the feeder recloser will also open to permit closing the tie switch.
Calculations of the load point indices (by feeder section) are summarized in Table
D-6. Note that in instances of loss of supply or transformer failure, restoration
time is much less than actual repair/replacement time. Note also that Section A
is without service while the repair crew switches Section B back into service and
then makes repairs on Section A. Outages on Section B do not affect service to
customers on Section A.
Figure D-3 shows a simple spot-network arrangement. Basically, each secondary bus
has two redundant feeds; single contingencies do not result in interruption of ser
vice.
.46
Similarly,
.53
D-6
Figure D-2. Looped system with automatic
sectionalizing and manual ties.
Table D-4
Section B repair
D-7
Table D-5
manual
failure rate restoration time switching time
(f/yr) (hr/f) (hrs)
Table D-6
Section A Section B
X r Xr X r Xr
Cause f/yr hrs/f hrs/yr f/yr hrs/f hrs/yr
D-8
AM AM
i
30MI- -.50 Ml .10 Ml
Table D-7
D-9
Analysis of the series (in a reliability sense) service of each station is
summarized in Table D-8.
Table D-8
Supply A Supply B
X r Xr X r Xr
f/yr hrs/f hrs/yr f/yr hrs/f hrs/yr
.0024 f/yr
(r ) (4.78)(3.72)
eq'service to spot network
v eq'servi
(4.78 + 3.72)
2.09 hrs/f
(u: (.00025)(2.09)
eq'service to spot network
.0005 hrs/yr
On the average, the spot networks are virtually free of interruptions due to
failures on the primary feeders. However, during transformer outages the proba
bility of failure is much higher because both spot networks are reduced to a
single supply. This affect is commonly reduced by adding a disconnecting switch
on the primary side of the transformers.
D-l 0
ANALYSIS OF A SUBSTATION WITH THROWOVER SWITCHING
Substation configurations such as the one illustrated in Figure D-4 are used to
speed service restoration in cases of subtransmission or substation failures.
Although this arrangement is commonly set up for automatic throwover operation,
this analysis is based on manual switching operations.
Failure and switching characteristics are shown in Table D-9. Calculations are
summarized in Table D-10. Note that the chances of losing both supplies is very
smal1:
CONCLUDING REMARKS
D-ll
SUPPLY A SUPPLY B
Table D-9
tie breaker - _
1.0
D-l 2
Table D-10
Feeder A Feeder B
X r X r Xr
f/yr hrs/f hrs/yr f/yr hrs/f hrs/yr
D-13