You are on page 1of 5

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO.

2, APRIL 2003

449

A Nationwide Survey of Recorded Information Used


for Calculating Distribution Reliability Indices
Cheryl A. Warren, Senior Member, IEEE, Dan J. Pearson, Member, IEEE, and Michael T. Sheehan, Member, IEEE

AbstractReliability is one of the most discussed topics in the


power industry today. Although many questions are being asked
about the reliability of utility systems across the nation, there are
currently no standards for recording or measuring interruption
minutes, number of customers affected, major events, whether to
include transmission and or distribution as well as many other issues. The methods used for calculating the indices and the definitions for the information have a major role in calculating reliability measures as defined in a recent IEEE standard. This paper
will provide the results of a nationwide survey on the recording or
measuring of data used to calculate reliability indices and associated practices. It will point out some of the pitfalls of using differing
methodologies in the recording of data behind the measures. These
differences often cause discrepancies when comparing reliability
indices between utilities.
Index TermsPower distribution, reliability.

I. INTRODUCTION

SURVEY of U.S. utilities was taken to determine the calculation of interruption information that is then used to
calculate reliability indices. From this survey, significant insight
was gained about practices used to record and measure the data
used to calculate reliability indices. This paper will outline, explain the purpose of, and show the response to the survey. The
information received in the main topic areas of the survey will
be discussed with further information regarding the issues. The
process for calculating the information used in interruption measures can vary greatly. The next section will examine what similarities exist between utilities and discuss ways to standardize
comparison between utilities that use like processes for calculating the information behind the reliability measures. The final
section will provide a conclusion on how to gain further consistency among utilities and drive consistency between utilities.

and reproducible, then a survey could be used to benchmark the


various distribution delivery strategies. Plus, there is a significant number of comparisons and benchmarking studies of utilities today. In order to do an apples to apples comparison, it is
critical that reliability reporting practices be the same.
III. SURVEY
A survey, comprised of 18 questions within six main topic
areas, was sent to 161 U.S. utilities in 1998, for the purpose of
gathering information on how interruption information was used
to calculate reliability. The main topic areas are listed below.
1) Regulatory body reporting requirements.
2) Step restoration.
3) What is included in reliability calculations (i.e., transmission/generation, reclosers, fuses, services, meters,
planned interruptions, etc.).
4) Major event inclusion and exclusions.
5) Is there a computerized interruption reporting system and
is there reporting of information to customers?
6) Is this information important and needed?
The survey results provided a wealth of information. Of the
161 utilities surveyed, 78 responded to almost all questions.
Company size seemed to have little bearing on the practices
used.
There seemed to be an issue on whether the type of protective
device that cleared the interruption was included in the interruption information used to calculate the reliability indices. The deviation in reporting practices was not clear. However, more utilities were reporting on the major protective devices only [i.e.,
there were more utilities reporting on just distribution circuit
breaker (CB) interruptions than those reporting on all device interruptions (e.g., fuses as well as CBs)].

II. RELIABILITY OBJECTIVES

IV. MAIN TOPIC AREAS

One of the primary objectives in developing reliability indices


is to benchmark the effects of various system design alternatives
on distribution performance. Fuse save versus fuse blow, covered conductors (tree wire), underground versus overhead and
various types of overhead construction are all design approaches
that will affect reliability. If the measurement system is accurate
Manuscript received July 5, 2001.
C. A. Warren is with National Grid USA, Albany, NY 12204 USA (e-mail:
cheryl.warren@us.ngrid.com).
D. J. Pearson is with Pacific Gas and Electric Company, San Francisco, CA
94177 USA (e-mail: DJP1@PGE.com).
M. T. Sheehan is with Puget Sound Energy, Seattle, WA 98009 USA (e-mail:
msheeh@puget.com).
Digital Object Identifier 10.1109/TPWRD.2002.803693

A. Regulatory Body Reporting


Question 1Are you required to report either indices or interruption information to the regulators? Y or N.
Answer 1Of the 73 utilities responding, 47 (64%) are required to report this information to the regulators.
Question 2Does the regulatory body define the process of
interruption reporting? Y or N.
Answer 2Of the 66 utilities responding, 32 (48%) have specific reporting requirements from the regulatory body.
Discussion: There seems to be a significant influence by the
regulatory bodies on interruption reporting requirements. An increasing number of utilities are being required to report reliability information to regulatory bodies. This increasing trend

0885-8977/03$17.00 2003 IEEE

450

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO. 2, APRIL 2003

seems to be following industry restructuring with regulatory


bodies watching the data to ensure that reliability does not deteriorate.
B. Restoration Process
Question 1Does your utility account for step restoration in
calculating indices? Y or N.
Answer 1Of the 76 utilities responding, 47 (62%) used step
restoration in calculating the information used in the reliability
measures reported with computerized calculations.
Question 2If the answer to 1 was N, do you use a factor to
generate a number that meets your expectations/estimation of
step restoration? Y or N.
Answer 2Of the 30 utilities that responded No to Question
1, none used any factor to estimate how step restoration would
increase the reliability measures.
Discussion: About half of the utilities that participated in the
survey are accounting for step restoration when calculating reliability indices. Step restoration mimics the actions taken by
the utility during interruption restoration, and therefore, mimics
the actual customer minutes experienced by the customer. When
step restoration is used, customer minutes are tracked accurately
as customers are restored. Depending on the methodology used,
the start and end times of an interruption can increase or decrease reliability indices significantly. By not accounting for
step restoration, it understates the values of SAIDI and makes
the comparison with other utilities more difficult. Also, when
responding to a customer request related to an interruption, logging the details of step restoration provides a good source of
information. However, it can be a large discrepancy, and brings
into question the comparisons between utilities. Inconsistencies
here are a main area of contention in utility comparisons.
C. Reliability Questions
Question 1Does your utility include in calculating reliability indices down to the following levels? Check all that
are appropriate: GenerationG, TransmissionT, Distribution SubstationDS, Circuit BreakerCB, Recloser/InterrupterRI, SectionalizerS, FuseF, Transformer
OnlyTO, ServiceS, MeterM.
Answer 1The following graph (Fig. 1) represents how 69
utilities responded with the -axis notations linking to the notations in Question 1 of item C.
Question 2Do you calculate MAIFI? Y or N
Answer 2Of the 76 utilities responding, 28 (18%) calculate
MAIFI.
Question 3If the answer to 2 was No, do you include MAIFI
in the calculation of SAIFI.
Answer 3Of the 50 utilities responding, 9 (18%) include
MAIFI in the calculation of SAIFI.
Question 4Which devices do you include in MAIFI that
generate momentaries? Check all that are appropriate.
Answer 4The following graph (Fig. 2) represents how 44
utilities responded with the -axis notations linking to the notations in Question 1 of item c).
Question 5Should transmission indices be generated separately from the distribution companies indices defined in IEEE
Std.-1366-1998? Y or N.

Fig. 1. Number of utilities including various interruption area/devices in the


calculation of reliability measures.

Fig. 2. Number of utilities including various interruption area/devices in the


calculation of MAIFI.

Answer 5Of the 71 utilities responding, 59 (83%) said that


transmission indices should be generated separately from the
distribution indices.
Question 6Do you feel that scheduled interruptions should
be calculated and indices reported separately rather than including with nonscheduled interruptions and indices? Y or N.
Answer 6Of the 76 utilities responding, 58 (76%) said that
scheduled interruptions indices should be generated separately
from the distribution indices.
Discussion: Of the utilities responding, 100% include distribution substations, and CB interruptions; while greater than
70% include recloser/interrupter, sectionalizers, and fuses; and
finally less than 50% include transformer only, services, and
meters in the data used to calculate the reliability measures denoted in [1]. One of the utilities calculates SAIDI and SAIFI
using only the distribution substation and CB level versus data
including distribution substation, CB, recloser/interrupter, sectionalizer, and fuse. The results showed that SAIDI and SAIFI
was greater with the inclusion of recloser/interrupter, sectionalizer, and fuse interruption data by 201% for SAIDI, and 166%
for SAIFI.
A smaller number of utilities (18%) include MAIFI data in
the SAIFI calculations that could significantly skew the data.
Once again, one utility ran SAIFI without MAIFI and SAIFI
with MAIFI included. The results showed that the SAIFI value
increased by 305% with MAIFI included.
Another key topic was the inclusion of scheduled/planned interruptions. The results showed that a high percentage of utilities
felt that planned interruptions should be excluded when calculating reliability indices. The belief was that scheduled interruptions do not provide pertinent information on the overall system
performance. However, all respondents felt that information on

WARREN et al.: NATIONWIDE SURVEY OF RECORDED INFORMATION

scheduled interruptions was valuable since it provided further


information on customer interruptions that could potentially improve business decision making.
Also, it was felt that under the current industry restructuring
that generation and transmission interruption data should have
their own reliability indices generated and that these should
not be included in the calculation of distribution reliability indices. Again, one utility compared the inclusion of generation
and transmission interruptions versus only distribution interruptions. The results showed that including the transmission and
generation interruptions SAIDI increased by 131% and SAIFI
by 120%.
Once again, segregating the data and indices between transmission and distribution provides a view on how each system
is performing as opposed to combining the data, which tends to
mask the performance of these different systems. The systems
are different since typically transmission is a network design
while the distribution system is typically constructed with a radial design.
There are some distribution networks used throughout the
utilities; however, the magnitude is small and designed such that
one interruption on the system does not impact customers. Only
a small percentage of utilities have distribution networks and
they are generally in the highly urbanized downtown city areas
(New York, Chicago, San Francisco, Seattle, etc.). Therefore,
inclusion of data from the network system in the reliability statistics is generally not done. It would only improve the statistics
since there are more customers that do not see an interruption.
D. Major Event Inclusion and Exclusions
Question 1Does your utility have a major event definition?
Y or N.
Answer 1Of the 76 utilities responding, 53 (70%) said that
they have a major event definition.
Question 2What is your definition of a major event for exclusion from reliability indices calculation?
Answer 2Of the 78 utilities responding, there were some
inconsistencies between all utilities. However, the most commonality among all the utilities was: 10% of the total customers
or an interruption due to extraordinary circumstances (such as
forest fires, hurricanes, major floods, or natural disasters) well
beyond the control of the company.
Question 3Is the major event definition the same as the
regulatory agency you are governed under? Y or N.
Answer 3Of the 49 utilities responding, 26 (53%) said that
their major event definition is the same as that used by the governing regulatory agency.
Question 4If the answer to 3 was N, then what is the regulatory agency major event definition?
Answer 4Of the 78 utilities responding, there were inconsistencies between all utilities. However, the most noticeable
commonality between all the utilities was the same as question
3 in the area of Major Event Inclusion and Exclusion.
Discussion: There is a wide variance in how utilities use the
major event criteria for excluding abnormal/extraordinary data
from the calculation of reliability indices. The variance between
utilities is one that surfaced based on service territory anomalies

451

(i.e., earthquakes in California, tornadoes in the Midwest, hurricanes in Florida). Where regulatory bodies are involved, the
criteria becomes more stringent and tends to match the 10% or
more of the customers impacted and extraordinary events that
are outside of the utilities control. In most of these cases, regulators allow these events to be excluded from the data used to
calculate reliability statistics.
It is generally felt that the utilities do not design their systems for these events due to the magnitude of cost that would
be imposed on the ratepayers. The events noted were: earthquakes, floods, forest fires, hurricanes, ice storms, tornadoes,
tsunamis, off distribution system events (transmission and generation), and other natural events. It seems that the regulatory
bodies are tending to drive more consistency in this area.
E. Computerized Interruption Reporting and Reporting to
Customers
Question 1Does your utility have a computerized interruption program that the interruption data is input into, and generates the reliability indices information? Y or N.
Answer 1Of the 78 utilities responding, 73 (94%) said that
they have a computerized interruption program that the interruption data is input into, and generates the reliability indices
information.
Question 2Can you provide historical interruption information to a customer upon request? Y or N.
Answer 2Of the 75 utilities responding, 73 (97%) said that
they can provide historical interruption information to a customer upon request. However, it is clear that not all interruptions can be communicated to customers since not all interruptions are stored in the utilities interruption databases noted in
Section IV-C, which discusses at what level interruptions are
reported in the interruption database.
Discussion: A high percentage of utilities use a computerized model to generate the reliability indices. The cause codes
used to classify interruptions by utilities are a major source
of uncertainty. Tree interruptions, lighting, bird or animal, and
car pole are some of the typical types used, but unknown and
catch-all categories usually make any analysis of the information more subjective than a true root cause analysis. The variability in the data source and quality of data best classifies reliability analysis as qualitative reliability based upon field experience and accuracy of the targeted improvement. Reliability
programs typically rely upon this qualitative approach to measure program effectiveness. The use of a computerized system
may allow consistency and better comparisons of data between
years, but the lack of quality input data limits the usefulness of
the analysis.
There also seems to be a direct link between the use of a
computerized system and the inclusion of step restoration (discussed in Section IV-B). Step restoration is a process that is
better handled by a computerized system to calculate SAIDI.
There were 47 utilities accounting for step restoration while 73
of the utilities have computerized systems. It is thought the 26
other utilities rely on a computerized system using a spreadsheet-based system to keep track of the data. Also, knowing 47
utilities have a computerized system and 73 utilities can provide

452

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO. 2, APRIL 2003

data to customers gives further evidence that the 26 other utilities are likely using a spreadsheet-based system versus a more
intelligent system that mimics the switching completed in the
restoration process. Generally, the more sophisticated the computerized system is, the more likely the data are consistent and
closer to actual system performance.

TABLE I
SIMILARITIES AMONG RESPONDING UTILITIES

F. Is This Appendix Needed?


Question 1Do you feel that this appendix is needed? Y or
N.
Answer 1Of the 70 utilities responding, 67 (96%) said that
this appendix was needed.
Question 2Would you like to see the results of this questionnaire? Y or N.
Answer 2Of the 75 utilities responding, 73 (97%) said they
would like to see the results of this questionnaire.
Discussion: There was a high interest in this analysis by the
utilities (50% response rate). Also, 97% of the utilities wanted
to see the tabulation of the response information. The fact that
industry restructuring is driving more regulatory involvement
and oversight into the distribution system arena seems to be the
major contributor to this conclusion.
V. SIMILARITIES BETWEEN UTILITIES
GAIN CONSISTENCY

AND

WAYS

TO

After reviewing the input from the 78 utilities, there seemed to


be a similarity between utilities in the following areas (Table 1).
Based upon this information, it seems that when comparisons
are done between utilities that they should only be done with
utilities that have like practices in the way the data are calculated
for calculating the reliability indices in [1].
There was a large discrepancy in Major Event definitions.
However, the regulatory agencies with utility input are driving
the industry toward greater consistency. Before the reliability
data can be benchmarked, it is important to establish a consistent definition of Major Event, step restoration, and transmission and distribution indices. The use of a similar major event
criterion should be implemented to gain consistency for comparing reliability indices between utilities. There is a consistent
major event criteria that is emerging but may not be acceptable
to all utilities, which is: 10% or more of the system customers
affected, and exclusions for natural events or off system disturbances outside of the control of the utilities system.
The questionnaire analysis was based upon SAIDI, SAIFI,
and CAIDI only. The California Regulatory Agency has driven
a measure related to the maintance of the system that is based
upon a sustained interruption/mile of line. Another approach includes using the customer value-of-service or value-of-services
tools. There is much more discussion and consensus needed at
the sustained interruption, value-of-service, and sustained interruption/mile of line measures.
Some utilities include data within their reliability measures
that other utilities do not. Momentaries, planned interruptions,
transmission, and generation interruptions are included in a minority of utility information. There is clearly a need for further
consistency of data when comparing utilities reliability indices
as defined in [1].

VI. CONCLUSION
Regulatory agencies are driving for consistency between
utilities. These agencies are also focusing on ensuring that
industry restructuring is accomplished without deterioration of
reliability. The IEEE needs to incorporate a process behind [1]
to ensure that consistent data calculation processes are being
used with [1] to calculate the reliability indices. For reliability
index comparisons between utilities, IEEE should consider
defining utility groupings for comparison purpose that are
based upon how the data are collected and calculated by the
utility.
IEEE Definition Areas Needed:
consistent definition of a major event;
generate reliability indices for the transmission/generation
system separately from the distribution system;
consistent definition of what is to be included in the data
used to calculate the reliability indices (i.e., exclude planned
interruptions, momentaries, etc.);
define what characteristics of a utilitys interruption data
calculations should be reviewed prior to comparing the reliability indices with those of another utility. If the characteristics
do not match, then the comparison is not applicable;
group and compare only like utilities that calculate the data
used in [1] reliability indices calculations;
the definitions already in use in [1] should be included to
ensure consistency;
new measures such as number of sustained interruptions per
mile or value of service should be defined by [1] to gain further
consistency.
REFERENCES
[1] Trial-Use Guide for Electric Power Distribution Reliability Indices,
IEEE Std. 1366-1998.

Cheryl A. Warren (S85M87SM99) received the B.Sc. and M.Sc. degrees


in electrical engineering from Union College, Schenectady, NY, in 1987 and
1990, respectively.
Currently, she is with National Grid USA, Albany, NY. She was a Principal of
the Firm at Navigant Consulting, Inc., Albany, NY. She leads distribution reliability activities and, with her colleagues, has developed asset management-based
tools. She has been with Central Hudson Gas and Electric Company, Poughkeepsie, NY. Power Technologies, Inc./Stone and Webster, Schenectady, NY.
Her main concentration of expertise has been with power quality, GIS/OMS
and enterprise-wide IT systems, and reliability. She has authored and co-authored sixteen technical papers.
Mrs. Warren chairs the IEEE Working Group on System Design that wrote the
Full-Use Guide on Electric Power Distribution Reliability Indices 13662001.

WARREN et al.: NATIONWIDE SURVEY OF RECORDED INFORMATION

Dan J. Pearson (M85) was born in Portland, OR, in January 1958. He received
the B.S.E.E. degree from Oregon State University, Corvallis.
He has been with Portland General Electric and Pacific Gas and Electric,
Portland, OR. His utility specialties are transmission and distribution reliability
and capacity.

453

Michael T. Sheehan (S71M74) received the B.S.E.E. degree from Illinois


Institute of Technology, Chicago, in 1974.
He has been with M.I.T. Center for Space Research, Commonwealth Edison,
Virginia Electric and Power Company, Richmond, VA; Department of Energy,
Washington D.C.; and Puget Sound Energy, Seattle, WA. His special fields of
interest include distributed generation, distribution reliability, and distribution
planning.
Mr. Sheehan is the past chairman of Distribution System Testing And Research, Chairman of the Western Energy Institute Distributed Generation Group,
and Chairman of the IEEE Working Group on Voltage Sags.

You might also like