Information about http://www.netl.doe.gov/moderngrid/docs/Cost_of_Power_Interruptions_to_Electricity_Consumers_in_the_.pdf
…
Tags: berkeley national laboratory, electricity consumers, energy efficiency and renewable energy, energy program, energy technologies division, environmental energy technologies, equal opportunity employer, ernest orlando lawrence, ernest orlando lawrence berkeley national laboratory, eto, forthcoming issue, hamachi, lawrence berkeley national laboratory, lbl, lbnl, legal responsibility, power interruptions, secretary of energy, trademark manufacturer, u s department,Pages: 33
Language: english
Created: Wed Feb 15 16:51:10 2006
Display cached document
Page 1
Page 2
Page 3
Page 4
Page 5
Page 6
Page 7
Page 8
Page 9
Page 10
Page 11
Page 12
Page 13
Page 14
Page 15
Page 16
Page 17
Page 18
Page 19
Page 20
Page 21
Page 22
Page 23
Page 24
Page 25
Page 26
Page 27
Page 28
Page 29
Page 30
Page 31
Page 32
Page 33
LBNL-58164
ERNEST ORLANDO LAWRENCE
BERKELEY NATIONAL LABORATORY
Cost of Power Interruptions to Electricity
Consumers in the United States (U.S.)
Kristina Hamachi LaCommare and Joseph H. Eto
Environmental Energy
Technologies Division
February 2006
http://eetd.lbl.gov/ea/EMS/EMS_pubs.html
To be published in a forthcoming issue of Energy: The International Journal.
This work described in this paper was funded by the Assistant Secretary of Energy
Efficiency and Renewable Energy, Distributed Energy Program of the U.S. Department
of Energy under Contract No. DE-AC02-05CH11231.
Disclaimer
This document was prepared as an account of work sponsored by the United
States Government. While this document is believed to contain correct
information, neither the United States Government nor any agency thereof, nor
The Regents of the University of California, nor any of their employees, makes
any warranty, express or implied, or assumes any legal responsibility for the
accuracy, completeness, or usefulness of any information, apparatus, product,
or process disclosed, or represents that its use would not infringe privately
owned rights. Reference herein to any specific commercial product, process, or
service by its trade name, trademark, manufacturer, or otherwise, does not
necessarily constitute or imply its endorsement, recommendation, or favoring
by the United States Government or any agency thereof, or The Regents of the
University of California. The views and opinions of authors expressed herein
do not necessarily state or reflect those of the United States Government or any
agency thereof, or The Regents of the University of California.
Ernest Orlando Lawrence Berkeley National Laboratory is an equal
opportunity employer.
1
Cost of Power Interruptions to
Electricity Consumers in the United States (U.S.)
Kristina Hamachi LaCommare* and Joseph H. Eto
Ernest Orlando Lawrence Berkeley National Laboratory
One Cyclotron Road, MS 90R4000
Berkeley CA 94720-8136
Received: 7 April 2005
______________________________________________________________________________
Abstract
The massive electric power blackout in the northeastern U.S. and Canada on August 14-15, 2003
catalyzed discussions about modernizing the U.S. electricity grid. Industry sources suggested that
investments of $50 to $100 billion would be needed. This work seeks to better understand an
important piece of information that has been missing from these discussions: what do power
interruptions and fluctuations in power quality (power-quality events) cost electricity consumers?
We developed a bottom-up approach for assessing the cost to U.S. electricity consumers of
power interruptions and power-quality events (referred to collectively as "reliability events").
The approach can be used to help assess the potential benefits of investments in improving the
reliability of the grid. We developed a new estimate based on publicly available information, and
assessed how uncertainties in these data affect this estimate using sensitivity analysis.
______________________________________________________________________________
*
Corresponding author, Fax: (510) 486-7976, email address: KSHamachi@lbl.gov
2
1. Introduction
The highly publicized power outage in the Northeastern U.S. and Canada on August 14-15, 2003
produced criticism that the U.S. power system was "antiquated", comparable to that of a "Third-
World nation", and in need of modernization [1, 2]. Qualitative critiques of the power system
have been published [3, 4], but these assessments have largely neglected to quantify an important
piece of information: how much do power interruptions and fluctuations in power quality
(power-quality events) actually cost electricity consumers? Accurately estimating this cost will
help assess the potential benefits of investments in improving the reliability of the grid. This
work develops a bottom-up approach for estimating this cost in order to better understand the
significance of uncertainties inherited in these estimates.
There have been several efforts to assess the cost of power interruptions and power quality. In
1992, Keane and Woo [5] applied customer outage costs to derive an optimal reserve margin for
generation that minimizes the sum of customer outage and generation capacity costs. The
resulting optimal reserve margin is compared with and determined to be less than the one
implied from the commonly used "1-day-in-10-years loss-of-load" criterion. Because very few
data were available on outage costs, rules of thumb were widely used to derive the estimates in
the study.
Shortly after, the Electric Power Research Institute (EPRI) reported the first ever power-quality-
cost estimate of $26 billion per year for the U.S.1 [6]. However, power quality is only a subset of
reliability events that affect customers and does not include losses of power for periods longer
3
than a few seconds. In addition, the estimate does not address the residential and commercial
sectors. Finally, the estimate is based on the assumption derived anecdotally from professional
judgment that 1.5 to three cents of spending per dollar sales went to power-quality equipment.
Later, EPRI extrapolated from this figure and began reporting power-interruption costs of $50
billion per year [7].
During the same period, a U.S. Department of Energy (DOE) study published cost-of-reliability
estimates ranging from $150 to $400 billion per year, based on an extrapolation from a value-of-
service study of the Duke Power service territory to the entire U.S. [8]. Sources of potential error
and bias in this estimate include the extrapolation from a single utility service territory to the
entire U.S. and the focus only on costs to the industrial sector [9].
In 2001, Eto et al. [10] performed a scoping study of electricity reliability trends in the U.S.,
which began by updating a comprehensive list of literature on electricity reliability costs. The
report also described three distinct end-use approaches for tracking trends in reliability needs and
presented an insurance industry perspective on electricity reliability as an example of a financial
tool for addressing customers' reliability needs.
Finally and most recently, EPRI estimated the national cost of power interruptions including
power-quality events at $119 billion per year [11]. The estimate is based on a limited national
survey of selected customer groups asking these customers to estimate costs for several distinct
power-outage scenarios. Key assumptions of this study, which are reviewed in LaCommare and
1
Billion refers to 109 U.S. dollars.
4
Eto [12], are that the costs experienced by the non-surveyed customer groups (excluding
residential customers) were 25 to 50 percent of the costs experienced by the surveyed population.
Costs to residential customers were not estimated. The outage cost scenarios all asked customers
to estimate costs they would experience on summer weekday afternoons, which is the time when
customers typically reported the highest costs from outages.
These studies have helped improve our understanding of the sources of uncertainty associated
with estimating customer outage costs, which is essential for valuing the overall cost of power
interruptions to U.S. electricity customers. However, little has been done to analyze
systematically the significance of these uncertainties.
A number of recent analyses, which seek to improve energy demand forecasts, are suggestive of
an approach that may facilitate and improve our ability to understand the cost of power
interruptions and power-quality events using a bottom-up approach. In particular, research using
end-use approaches has improved our understanding of how energy is used. For example, Turiel
et al. [13] estimates the energy intensity of major end uses in the commercial sector using an
end-use approach, work that is important to understanding how each end use contributes to the
overall demand for electricity in the commercial sector. Bottom-up analyses such as these, allow
for more accurate estimates of energy consumption patterns and help improve the modeling
accuracy of energy forecasting models. Chateau and Lapillonne [14] stress the importance of
end-use models and approaches as an important tool for forecasting energy demand in long term
energy planning. Our work builds upon the analytical rigor and strengths inherent in these
5
bottom-up approaches for energy analysis by adapting it to improve our understanding of the
cost of outages.
In the work described in this article, we developed a bottom-up approach for estimating the cost
of power interruptions to U.S. electricity consumers. We illustrated the use of the approach by
drawing on existing data from a variety of sources to develop a new ("base case") estimate of the
total economic cost to U.S. electricity consumers (not including power quality events.) We
examined uncertainties and gaps in the information used to develop our estimate as one
application of our end-use approach in order to define the degree of variability that might be
expected in future estimates. Based on the identified uncertainties, we prioritized future data-
collection activities whose results can be used to refine estimates of outage costs.
2. A Bottom-Up Approach for Estimating the Economic Cost of Power Interruptions and
Power-Quality Events
A key aspect of our work is the development of a bottom-up approach for estimating the
economic costs of power interruptions and power-quality events to U.S. electricity consumers.
The framework relies on a simple mathematical expression that determines the economic cost of
power interruptions and power-quality events as follows:
m n p
Cost of Power Interruptions and Power Quality = i =1 j =1 k =1
N i , j × Fi , j ,k × C i , j ,k × Vi , j ,k (1)
where,
N = number of electricity customers in customer class i for region j
6
F = the frequency of reliability events of type k experienced annually by customer class i
in region j
C = the cost of reliability event type k per customer in customer class i for region j (2002-
Consumer-Price-Index-weighted dollars/event)
V = the vulnerability of customers to reliability event type k in customer class i for region
j (a fraction between 0 and 1)
m = the number of customers in each customer class
n = the number of regions
p = the number of types of reliability event
i,j,k = indices for customer class, region, and type of reliability event, respectively
The simplicity of this formula belies the complexities involved in estimating each of the four
parameters in equation (1). Indeed, the data used to estimate the parameters are often difficult to
define clearly. The following subsections outline current limitations on availability of the data
needed to estimate each parameter, while section 3 describes the process we used to collect these
data for our study.
2.1 Customer Data
The number of customers considered in estimating the cost of power interruptions or power-
quality events will have a significant impact on the magnitude of the estimate. Substantial
uncertainty can result from differences in how customers are defined. In the U.S., customers may
be defined in any of the following ways: (1) as a single electricity account with one (or more)
meter(s), such as a single-family detached residence; (2) as a single site/facility with multiple
7
accounts, each possibly consisting of multiple meters, such as an apartment building; or (3) as
multiple premises under common ownership, each with one or more accounts/meters, such as a
chain of retail establishments. It is important to note that customers do not equal people and that
one electricity customer often refers to one household, residence, or commercial or industrial
facility or business.
2.2 Reliability Event Data
Although many U.S. utilities maintain detailed records of customer outage experience, these data
are usually reported in summarized form as two reliability-event indices [15]: the System
Average Interruption Duration Index (SAIDI) and System Average Interruption Frequency Index
(SAIFI). SAIDI and SAIFI describe the duration and frequency, respectively, of sustained
interruptions experienced by customers of a utility in one year [16, 17]. According to the
Institute of Electrical and Electronics Engineers (IEEE), a "sustained interruption" is any
interruption that lasts at least five minutes and is not classified as "momentary". Consistent with
IEEE's definition of a sustained interruption, a momentary outage is defined as any event lasting
less than five minutes.
SAIDI represents the average duration of customers' power interruptions, while SAIFI represents
the average number of interruptions per customer per year for a specified electricity supply
system. We used SAIDI and SAIFI data to quantify the duration and frequency of sustained
electricity interruption events for a typical year. We also used the Momentary Average
8
Interruption Frequency Index (MAIFI) index, which is a useful measure for assessing the
frequency of momentary interruptions; however, these data are not as commonly collected.
Unfortunately, SAIDI, SAIFI, and MAIFI data do not distinguish among the types of customers
experiencing reliability events. Because of the way electricity distribution systems are designed
and operated, larger commercial and industrial customers in some areas may experience fewer
and shorter power interruptions than smaller commercial and residential customers [18].
Power quality, or the degree to which voltage and current deviate from a perfect 60-cycle
sinuosoid, has emerged as an important issue in recent years. Despite the growing importance of
power-quality events as a category of reliability events, there is even less information available
on these events than on power outages. There has been only one comprehensive study of power
quality, and the data collected for it are now more than a decade old [19]. Currently, there are no
ongoing public-domain data-collection efforts for power quality [17]. Hence, although the
bottom-up approach we designed allows for consideration of power-quality events, the lack of
sufficient power-quality data prevented us from considering these events in our base-case
estimate.
2.3 Cost per Outage Data
Estimating customer outage costs involve several sources of uncertainty, two of which pertain to
the accuracy of the cost measurement as well as how representative the estimate is of the area it
covers. Typically, outage-cost estimates are based on what customers say they will experience
9
under different hypothetical outage circumstances in surveys [9]. The key source of uncertainty
in these estimates is the degree to which the costs customers report for hypothetical
circumstances correspond with the actual costs they would experience [20]. No studies have
attempted to validate the results obtained from these surveys, yet this is a significant source of
uncertainty in the cost estimates that have been prepared to date. Another important source of
uncertainty inherent to customer survey data is small sample size; this can lead to significant
errors if the sample is insufficient to appropriately represent the population. A common
manifestation of this uncertainty arises from extrapolation to a wide geographic area from a
smaller subset that is not representative of the wider geography.
Assessing actual costs is complicated by the differing impacts of reliability events on various
classes of customers residential, commercial, and industrial. Costs experienced by non-
residential customers or firms are, in principle, simpler to estimate than the difficult-to-quantify
"hardship" costs experienced by residential customers. Basic accounting categories for
commercial customers, such as labor and materials costs and revenue losses, are straightforward
(though not necessarily easy to use) to apply in determining the cost of work interruptions caused
by power losses. Residential-sector costs, however, include elements such as the cost of
consumable goods (e.g. flashlights and candles) and inconvenience costs (e.g. resetting clocks,
changing plans, and coping with inconvenience, fear, anxiety, etc.).
A subtle issue that is gaining recognition is that business losses are not always directly
proportional to the duration of a reliability event [21]. In such cases, the key factor is the length
of business or production downtime caused by an outage of any length. For example, a partial
10
loss of voltage or a voltage sag can cause the same amount of downtime as a complete loss of
power, if, for example, machines need to be rebooted or production processes need to be
restarted. This issue poses a major challenge in estimating the economic cost of power
interruptions and power-quality events.
2.4 Vulnerability
The economic cost and perceived risk of reliability events has led many customers to invest in a
wide variety of technologies and measures to reduce their vulnerability to these events. Back-up
or stand-by generators are probably the most well known customer investments; smaller
investments like strip-surge protectors should be considered in this same category, however. In
between are a host of energy-storage technologies, such as batteries and flywheels, which can
reduce a customer's vulnerability to power interruption and power-quality events.
Comprehensive, national data on these investments are lacking. Hence, in the illustrative
example presented in the next section, we did not explicitly consider the potential of these
measures to reduce customers' vulnerability to reliability events. That is, we implicitly set this
factor to be equal to 1.0.
3. Deriving a New Estimate of the Economic Cost of Power Interruptions to U.S.
Electricity Consumers
Having established our framework, we applied it to develop a new estimate of the cost of power
interruptions, based on a review of the best data available in the public domain (the "base case").
11
As described earlier, even the best data available have important limitations and our new
estimate ultimately represents a mix of nationally aggregated and regional data components.
3.1 Customers by Class and Region
We used residential, commercial, and industrial electricity demand sectors to describe customer
classes as shown in Table 1. Customer population data were taken directly from the Energy
Information Administration (EIA) Electric Sales and Revenue publication [22]. The data are
reported for the year 2001 by state and demand sector. We partitioned the U.S. into regions so
that we could represent variations in outage costs as a function of differences in the population.
The regions correspond to U.S. Census Divisions as mapped by EIA, with a slight variation in
the Pacific; we extracted California, where most of the Pacific region population resides, and
treated it as a separate region.
3.2 Duration and Frequency of Reliability Events
Despite the existence of well-defined indices for a majority of reliability events, reliability event
data are not collected systematically or consistently. The National Regulatory Research Institute
reports that only 23 of 40 surveyed states require annual reporting of reliability statistics [23]. As
a result, deriving reliability event data on a national scale is a challenge.
We conducted an Internet search to gather publicly available reliability-event data. To address
biases introduced by extreme outlying values, we used "trimmed means": we removed the
12
highest and lowest five percent of observations and then calculated the means of the reduced
observations [24]. Table 2 shows the trimmed means and standard deviations for SAIDI, SAIFI,
and MAIFI used for our base case calculation. Trimming the data sets had a noticeable effect on
the resulting average duration of outages in the U.S.2 Because our Internet search did not
produce adequate data for all regions, national reliability event data were used in the base case
with regional considerations presented in our sensitivity analysis.
Several other studies have also examined national SAIDI, SAIFI, and MAIFI statistics (see Table
3). Our calculated trimmed means of 106 minutes, 1.2 sustained interruptions, and 4.3
momentary interruptions for SAIDI, SAIFI, and MAIFI, respectively, are very similar to the
estimates reported from these studies.Our data thus appear to be reasonable inputs for calculating
the cost of power interruptions to U.S. electricity customers.
Since we did not find similarly comprehensive data on the frequency of power-quality events, we
elected to exclude the costs of these events in our base-case estimate.
3.3 Cost of Reliability Events
Limited information is available on the cost of reliability events. The cost data is taken from a
recent analysis of the cost of power interruptions to customers [9]. The study is a meta-analysis
2
When we removed the outliers, the SAIDI average decreased by 13 percent, from 122 to 106 minutes, but the SAIFI and
MAIFI means changed very little. The standard deviation for both SAIDI and SAIFI were roughly halved when the outlying 10
percent of the data points were removed; the MAIFI standard deviations were reduced by more than 10 percent f. This suggests
that trimming the highest and lowest five percent of data points helped to improve significantly the robustness of our means.
13
of 24 independent customer surveys conducted by eight electric utility companies in the U.S.
during the past 13 years.
The study developed analytic expressions, called "customer damage functions", using a
regression approach that expresses customer outage costs as a function of customer class, region,
event duration, and other descriptive variables based on survey responses from more than 2,000
large commercial and industrial (C&I), 5,200 small and medium C&I, and 11,000 residential
customers. The resulting cost per outage per customer data we derived using the damage
functions were normalized and reported in year-2002 Consumer-Price-Index (CPI)-weighted
dollars. The independent descriptive variables used to derive customer outage costs were
obtained from the most recently available EIA Commercial Building Energy Consumption
Survey (CBECS) and Manufacturing Energy Consumption Survey (MECS) data.
To account for outage cost differences by time of day, week, and season, we weighted the costs
for morning, night, weekend, weekday, and summer and winter time periods. The outage cost
associated with each time period was then weighted to derive the cost per outage per customer
throughout the year.
Table 4 shows the costs per outage per customer used to calculate the total cost of power
interruptions to U.S. electricity customers in our base case analysis. The table includes costs for
momentary and sustained interruptions as well as costs for the typically reported one-hour outage
length as a basis of comparison to external estimates. Again, the U.S. outage cost is used in our
base case estimate with regional consideration in the sensitivity analysis.
14
3.4 Base-Case Estimate of the Economic Cost of Power Interruptions
We find that, based on publicly available data and subject to the limitations discussed in this
section, the economic cost of power interruptions to U.S. electricity consumers is $79 billion
annually.
Figure 1 shows how these costs are distributed among the three customer classes. The
commercial sector accounts for more than 72 percent ($57 billion) of the total outage cost. This
large share is due to the high outage cost for this sector (shown in Table 4) coupled with the
large population of commercial customers. The industrial sector represents nearly 26 percent of
the total cost, and the residential sector accounts for less than 2 percent of the total.
Although per outage per customer costs were significantly higher for the industrial sector than
for the commercial and residential sectors, the industrial sector's smaller population-- 1.6 million
customers, compared to 14.9 million commercial customers--led to a lower total estimated cost
than for the commercial sector. Commercial sector outage costs were roughly one order of
magnitude lower than those for the industrial sector. The 114.3 million residential customers
represent by far the largest population of customers in the three sectors, but their outage costs are
two to three orders of magnitude lower than for the commercial sector. Therefore, despite its
large customer population, the residential sector's contribution to the total is minor.
15
Figure 2 shows that momentary outages accounted for two-thirds of the overall annual U.S.
outage cost; the 4.3 momentary outages that were derived from a trimmed mean of collected
Internet data cost the U.S. $52 billion each year; the 1.2 sustained outages (totaling 106 minutes
together) also representing a reduced mean of the collected data account for the remaining $26
billion of the $79 billion total.
Figure 3 shows our base-case estimate by region and customer class. The South Atlantic region
bears the largest burden, $14.7 billion, or nearly 19 percent, of the total outage cost for the
nation. The Pacific-region cost, with only $2.8 billion (or 3.6 percent of the total) has the lowest
share of national outage costs.
Figure 3 also compares our regional estimates to the population of commercial and industrial
customers in each region to determine how well correlated the regional estimates are to
population density.3 The differences in the total costs by region appear to correspond well to
these regional population differences. In the South Atlantic region, the higher total cost is clearly
linked to the large population. The lowest total costs are seen in the Pacific and New England
regions where the customer populations are the smallest.
These findings point to the underlying parameters (i.e., customer definitions and populations,
outage characteristics, and outage costs) that likely have the greatest influence on our estimate.
Next, we explore the sensitivity of each of these parameters to understand better the effects of
their inherent uncertainties on the total cost of power interruptions.
3
The residential population is not included because it is such a large share of the total population, yet has little overall impact on
the total cost of power interruptions in the U.S.
16
4. Using Sensitivity Analysis to Explore Uncertainty in Power Interruption Cost
Estimates
To understand the impacts of uncertainty on results from our bottom-up approach, we performed
a sensitivity analysis in which we varied the key parameters to quantify their impact on our
results. This sensitivity analysis expresses the degree of uncertainty associated with the input
parameters as well as the variability in the resulting cost estimate.
Figure 4 shows the resulting total cost of power interruption estimates for each of the following
sensitivity analyses, in which we varied:
1. The duration and frequency of reliability events and the cost per outage per customer by
region, using the limited region-specific data we collected;
2. The duration and frequency of reliability events by one standard deviation greater and less
than the values used in our base-case estimate;
3. The value of outages based on two assumptions of their timing--whether they occur on a
summer weekday afternoon or summer weekend night; and
4. The duration and frequency of reliability events experienced by commercial and industrial
sectors to reflect the fact that they tend to be disproportionately shorter in duration and of
decreased frequency relative to events experienced by the residential sector.
We found that the annual cost of power interruptions:
17
ˇ Could be as low as $22 billion or as high as $135 billion when we consider a reasonable
range in the annual duration and frequency of power interruptions, which accounts for gaps
in the data for certain regions and possible year-to-year variations in reliability;
ˇ Could be as high as $119 billion if, as many studies (incorrectly) assume, all reliability
events take place during summer weekday afternoons when power usage and costs are high;
ˇ Could be as low as $23 billion if the typically fewer and shorter interruptions experienced by
larger commercial and industrial customers are accounted for, which is the result of the
design of many utility distribution systems; and
ˇ Is relatively insensitive to regional variation in reliability event and outage cost data--the
$90 billion estimate accounting for this variation is only 14 percent higher than the base-case
estimate that used U.S.-averaged values.
Each sensitivity exercise helped shed light on key uncertainties associated with calculating the
cost of power interruptions. The modest impact of regional differences suggests that climate and
the distribution of customers by class across the country has a limited impact on the overall cost
of power interruptions in the U.S. However, we detected significant differences in the other three
sensitivity studies. Changing the base-case SAIDI, SAIFI, and MAIFI values by just one
standard deviation, changes the overall estimate by one order of magnitude, from $22 billion to
$135 billion annually. This illustrates the large effect that outage duration and frequency, which
can realistically be expected to vary within this range, have on the overall cost. Assumptions
about the timing of events have almost as large an impact, mainly because commercial and
industrial customers' outage costs depend on whether events occur during their operating hours.
This demonstrates the importance of using time-representative outage costs when calculating
18
total costs. Finally, the total cost estimate is dramatically lowered when we account for higher
levels of electric reliability typically enjoyed by industrial sector customers relative to
commercial and residential customers. This variable, too, has a significant influence on overall
cost-of-power-interruptions estimates and should be accounted for in future analyses.
5. Conclusion
Our analysis establishes a bottom-up approach for systematically comparing and analyzing
estimates of the economic cost of power interruptions and power quality to U.S. electricity
customers. Using this approach, our base-case estimate of the annual cost for power interruptions
to U.S. electricity consumers is $79 billion. Our analysis of the uncertainty in this estimate
suggests that the costs could be as high as $135 billion or as low as $22 billion based on the
particular sensitivity assumptions we employed. Thus, we present these figures not as the final
word on the cost of power interruptions, but rather to illustrate the range of variability in
estimates, which is directly related to the assumptions on which the estimates are based and the
uncertainty inherent to these assumptions. In view of the large range of plausible estimates and
the enormous costs associated with the private and public decisions that may be based on them,
we encourage policy makers, regulators, and industry to work jointly to undertake activities
needed to improve the information available on reliability events and their costs. The cost of
such improved data collection would be modest relative to the potential costs of reliability events
in the U.S.
19
Using our bottom-up approach, we find that the majority of outage costs are borne by the
commercial and industrial sectors, not the residential sector. Although there are important
variations in the composition of customer types within each region of the country, the total cost
of reliability events by region tend to correlate roughly with the number of commercial and
industrial customers in each region. In addition, costs tend to be driven by the frequency rather
than the duration of reliability events. Momentary power interruptions, which are more frequent,
have a larger impact on the total cost of interruptions than sustained interruptions, which are less
frequent.
Priorities for future improvements in outage cost data collection fall into two broad categories:
better information is needed on customers' reliability experiences and on the cost of power
interruptions to individual customers. Specific improvements include: undertaking a coordinated,
nationwide collection of updated information on the cost of reliability events to customers;
validating the outage costs obtained from customer surveys with actual market data (e.g.,
customer ownership of backup generation), establishing consistent definitions and tracking
protocols for the frequency, duration, timing, and number and type of customers affected by
reliability events, including power-quality events, by customer class; and collecting information
on efforts by customers to reduce their vulnerability to reliability events through investments in
technology and other measures.
Acknowledgements
We acknowledge the guidance and direction provided by Imre Gyuk and Sam Baldwin, U.S.
DOE. The work described in this paper was funded by the Office of Electric Transmission and
20
Distribution, Energy Storage Program and by the Assistant Secretary for Energy Efficiency and
Renewable Energy, Office of Planning, Budget, and Analysis of the U.S. Department of Energy
under Contract No. DE-AC03-76SF00098.
References
[1] Fialka J. Power industry sets campaign to upgrade grid. Wall Street Journal, New York, A3.
25 August 2003.
[2] Schieffer B. Face the Nation with Bob Schieffer. Washington D.C. 17 August 2003.
[3] Hirst E., Kirby B. Operating reserves and bulk-power reliability. Energy 1998; 23 (11): 949-
959.
[4] Billinton R., Allan R. Basic power system reliability concepts. Reliability Engineering and
System Safety 1990; 27, 20.
[5] Keane DM, Woo C-K. Using customer outage costs to plan generation reliability. Energy
1992; 17 (9): 823-827.
[6] Electric Power Research Institute. Estimating the cost of power quality. IEEE Spectrum,
1993; 30 (6): 40-41.
[7] Douglas J. Research Analyst to EPRI. Personal communication to J. Eto. Lawrence Berkeley
National Laboratory, Berkeley, California. 2000.
[8] Swaminathan S, Sen RK. Review of power quality applications of energy storage systems.
report no. SAND98-1513. Albuquerque, New Mexico and Livermore, California. Sandia
National Laboratories. 1998.
[9] Lawton, L., M. Sullivan, K. Van Liere, A. Katz. Characteristics and trends in a national study
of consumer outage costs. Population Research Systems, LLC. San Francisco, California. 2003.
21
[10] Eto J, Koomey JG, Lehman B, Martin N, Mills E, Webber C, Worrell E. Scoping study on
trends in the economic value of electricity reliability to the U.S. economy. Report no. LBNL-
47911. Berkeley, California. Lawrence Berkeley National Laboratory. 2001.
[11] Primen. The cost of power disturbances to industrial and digital economy companies. report
no. TR-1006274 (Available through EPRI). Madison, Wisconsin. Primen. 2001.
[12] LaCommare KH, Eto JH. Understanding the cost of power interruptions to U.S. electricity
customers. report no. LBNL-55718. Berkeley, California. Lawrence Berkeley National
Laboratory. 2004.
[13] Turiel I, Craig P, Levine M, McMahon J. Estimation of energy intensity by end-use for
commercial buildings. Energy 1987; 12 (6): 435-446.
[14] Chateau B, Lapillonne B. Accounting and end-use models. Energy 1990; 15 (3, 4): 261-278.
[15] Kueck, J, Kirby B, Overholt P, Markel L. Measurement practices for reliability and power
quality, a toolkit of reliability measurement practices. report no. ORNL/TM-2004/91. Oak Ridge,
Tennessee. Oak Ridge National Laboratory. 2004.
[16] IEEE. IEEE Standard 1159-1995, recommended practice for monitoring power quality.
ISBN no. 1-55937-549-3. New York, New York. IEEE. 1995.
[17] Electric Power Research Institute. Distribution reliability indices tracking within the United
States. report no. TR-1008459. Palo Alto, California. EPRI. 2003.
[18] IEEE. Draft recommended practice for utility interface of photovoltaic (PV) systems. report
no. P929. New York, New York. IEEE. 1999.
[19] Electrotek Inc. An assessment of distribution system power quality: volumes 1-3. report no.
TR-106294-V1-3. Palo Alto, California. EPRI. 1996.
22
[20] Woo C-K, Pupp RL. Costs of service disruptions to electricity consumers. Energy 1992; 17
(2): 109-126.
[21] Eto, J, Divan D, Brumsickle W. Pilot evaluation of electricity-reliability and power-quality
monitoring in California's Silicon Valley with the I-GridŽ system. report no. LBNL-52740.
Berkeley, California. Lawrence Berkeley National Laboratory. 2004.
[22] Energy Information Administration. Electric sales and revenue 2001. Washington, D.C.
2001.
[23] National Regulatory Research Institute. The state public service commission reliability
survey. 2000.
[24] Mosteller F, Tukey JW. Data analysis and regression: a second course in statistics. Reading,
Massachusetts: Addison-Wesley Publishing Company, 1977.
23
Figure and Table Captions
Figure 1. Base-Case Estimate of the Cost of Power Interruptions by Customer Class1
Figure 2. Base-Case Estimate of the Cost of Power Interruptions by Type of Interruption1
Figure 3. Base-Case Estimate of the Cost of Power Interruptions by Region and Customer Class
with C&I Population1
Figure 4. Summary of U.S. Cost of Power Interruption Sensitivity Cases1
Table 1. Number of Customers by Region and Sector in 2001
Table 2. Summary of Trimmed Mean Reliability Event Data
Table 3. Summary of U.S. Reliability Event Estimates by External Studies
Table 4. Customer Outage Costs for the U.S. in 2001
24
Residential
2%
Industrial $1.5 Billion
26%
U.S. Total:
$79 Billion
$20.4 Billion
$56.8 Billion
Commercial
72%
1
Costs shown in U.S. 2002 CPI-weighted dollars
Figure 1. Base-Case Estimate of the Cost of Power Interruptions by Customer Class1
25
U.S. Total:
Sustained $79 Billion
Interruptions
33%
$26.3 Billion
$52.3 Billion
Momentary
Interruptions
67%
1
Costs shown in U.S. 2002 CPI-weighted dollars
Figure 2. Base-Case Estimate of the Cost of Power Interruptions by Type of Interruption1
26
$16 $14.7 B 3.5
U.S. Total = $79 Billion Residential
$14 3.0
Industrial
$12 Commercial
C&I Population (Millions)
$9.7 B $10.4 B $10.6 B 2.5
C&I Population
$10
Billions-$
$8.1 B 2.0
$8
$6.7 B $6.7 B
$5.5 B 1.5
$6
$3.6 B 1.0
$4
$2.8 B
$2 0.5
$0 0.0
New Middle East West South East South West Mountain Pacific California
England Atlantic North North Atlantic Central South
Central Central Central
1
Costs shown in U.S. 2002 CPI-weighted dollars
Figure 3. Base-Case Estimate of the Cost of Power Interruptions by Region and Customer
Class with C&I Population1
27
$160 Momentary Interruptions
$135 B Sustained Interruptions
$140
$119 B
$120
$100 $90 B
Billions-$
$79 B
$80
$60
$40 $22 B $26 B $23 B
$20
$0
LBNL Regional Data SAIDI, SAIDI, All Outages All Outages C&I
Base-Case SAIFI, SAIFI, Valued Using Valued Using Customers
Estimate M AIFI M AIFI Costs for a Costs for a Experience
Increased by Decreased by Summer Summer Fewer,
1 Standard 1 Standard Weekday Weekend Shorter
Deviation Deviation Afternoon Night Outages
1
Costs shown in U.S. 2002 CPI-weighted dollars
Figure 4. Summary of U.S. Cost of Power Interruption Sensitivity Cases1
28
Table 1. Number of Customers by Region and Sector in 2001
RESIDENTIAL COMMERCIAL INDUSTRIAL TOTAL
U.S. ESTIMATE 114,317,707 14,939,895 1,582,573 130,840,175
BY REGION
New England 5,822,935 714,049 62,677 6,599,661
Middle Atlantic 15,045,495 2,127,033 103,713 17,276,241
East North Central 18,705,754 2,110,172 158,780 20,974,706
West North Central 8,287,837 1,139,609 170,937 9,598,383
South Atlantic 22,473,797 2,842,220 270,840 25,586,857
East South Central 7,356,975 1,135,507 78,545 8,571,027
West South Central 12,883,403 1,722,873 292,035 14,898,311
Mountain 7,368,280 1,001,310 212,842 8,582,432
Pacific 3,922,426 494,778 66,699 4,483,903
California 11,841,144 1,559,258 154,261 13,554,663
Source: Energy Information Administration. 2003. Electric Sales and Revenue 2001. EIA/DOE. January.
Washington D.C.
29
Table 2. Summary of Trimmed Mean Reliability Event Data
SAIDI SAIFI MAIFI
Trimmed Mean 106 min. 1.2 4.3
(Standard Deviation) (54 min.) (0.5) (3.6)
30
Table 3. Summary of U.S. Reliability Event Estimates by External Studies
SAIFI SAIDI MAIFI
EPRI Reporta 1.1 107
IEEE 1995 Surveyb 1.3 120 5.5
EEI Annual Reportc
1998 1.2 118 5.4
1999 1.4 101 11.6
a
Source: EPRI 2003 Distribution Reliability Indices Tracking Within the U.S. TR-1008459. May. Palo Alto CA.
b
Source: IEEE 1995 Survey (http://resourceinsight.com/work/naruc_pbr_97.pdf)
c
Source: Power Sources Manufacturer's Association (http://www.psma.com/HTML/newsletter/Q2_2001/page11.html)
31
Table 4. Customer Outage Costs for the U.S. in 2001a
Residential Commercial Industrial
Momentary Interruption $2.18 $605 $1,893
Sustained Interruption $2.99 $1,067 $4,227
One-hour Interruption $2.70 $886 $3,253
a
Costs shown in U.S. 2002 CPI-weighted dollars