Recent trends in power system reliability and implications for evaluating future investments in resiliency
Introduction
In the U.S. and abroad, recent catastrophic weather events; existing and prospective government energy and environmental policies; and growing investments in smart grid technologies have drawn renewed attention to ensure the reliability of the electric power system [42], [6]. Over the past 15 years, the most well-publicized efforts to assess trends in electric power system reliability have focused only on a subset of all power interruption events [3], [8] —namely, the very largest events, which trigger immediate emergency reporting to federal agencies and industry regulators. Anecdotally, these events are believed to represent no more than 10% of the power interruptions experienced annually by electricity consumers. Moreover, a review of these emergency reports has identified shortcomings in relying upon these data as accurate sources for assessing trends, even for the reliability events they target [16].
Recent work has begun to address these limitations by examining trends in reliability data collected annually by electricity distribution companies [13], [14]. In principle, all power interruptions experienced by electricity customers, regardless of size, are recorded by the distribution utility. Moreover, distribution utilities have a long history of recording this information, often in response to mandates from state public utility commissions [12]. Thus, studies that rely on reliability data collected by distribution utilities can, in principle, provide a more complete basis upon which to assess trends or changes in reliability over time.
Eto et al. [13], [14] was one of the first known studies to apply econometric methods to account for utility-specific differences among electricity reliability reports. This study found that the annual average amount of time and frequency customers are without power had been increasing from 2000 to 2009. In other words, reported reliability was getting worse. However, the Eto et al. [13], [14] paper was not able to identify statistically significant factors that were correlated with these trends. The authors suggested that “future studies should examine correlations with more disaggregated measures of weather variability (e.g., lightning strikes and severe storms), other utility characteristics (e.g., the number of rural versus urban customers, the extent to which distribution lines are overhead versus underground), and utility spending on transmission and distribution maintenance and upgrades, including advanced (“smart grid”) technologies” [13], [14]. Ahvehag and Söder [2] describe a reliability model that correlate two severe weather metrics (lightning, wind speed) to distribution system failure rates (SAIFI) and restoration times (SAIDI) in Sweden. The aforementioned authors found that the “stochasticity in weather has a great impact on the variance in the reliability indices” [2]; p. 910). However, the Ahvehag and Söder [2] study does not consider other factors, which may contribute to reliability including utility spending and the presence of outage management systems—among other things.
This paper seeks to extend the Eto et al. [13], [14] and Ahvehag and Söder [2] analyses along exactly these lines. This paper attempts to identify statistically significant factors, including various aspects of “abnormal weather”, but also other utility characteristics, using up to 13 years of information on power interruptions for a large cross-section of U.S. electricity distribution utilities. These utilities, taken together, represent approximately 70% of both total U.S. electricity sales and customers. We also consider the possibility that utility operations and maintenance spending may impact reliability and that weather and reliability have a non-linear relationship. Following Hoen et al. [25]; we employ a sequential modeling approach to ensure model (1) performance; (2) parsimony; and (3) coefficient stability is achieved prior to interpretation.
In this work, we seek to answer the following questions:
Are warmer/cooler, wetter/drier, and/or windier than average years correlated with changes in the annual average number of minutes and/or frequency of power interruptions?
Are the number of customers, annual sales of electricity, share of underground lines, or the presence of outage management systems (OMS) correlated with changes in the annual average number of minutes and/or frequency of power interruptions? Is previous year T&D operations and maintenance (O&M) spending correlated with changes in the annual average number of minutes and/or frequency of power interruptions in the following year?
Are there trends in the annual average number of minutes and/or frequency of power interruptions over time, which we cannot explain by considering the above factors?
Answers to these questions have important implications for efforts to project long-term trends in reliability and the associated benefits of strategies to improve grid resiliency to severe weather—both in the U.S. and abroad.
Section snippets
Reported causes of power outages
Utilities in the U.S. publicly report a number of causes associated with increased frequency and duration of outages. This section reviews causes of reliability events as reported by a subset of the U.S. electric utilities evaluated in the broader econometric analysis. The following utility reliability reports were consulted to determine the causes of past reliability events: Florida Public Utilities Company [17]; Rocky Mountain Power [41]; Interstate Power and Light Company [27]; Jersey
Econometric analysis method
We used the following regression equation to analyze the relationship between utility-specific attributes and weather variability on the duration (SAIDI) and frequency (SAIFI) of power interruptions:
The general model specification described in Equation (3) above follows the general form used in earlier energy-related multivariate panel regressions [10], [13]. In Equation (1), annual utility reliability (measured by SAIDI or SAIFI with or without major
Model performance and selection
We developed a sequence of model specifications (each a distinct regression equation following the form outlined in Section 3) and conducted a series of robustness tests to evaluate them following procedures outlined in Hoen et al. [25]; which evaluated the impact of wind power projects on residential property values.6
Principal findings
This section describes the principal findings from our analysis. Fig. 5 through Fig. 8 show results for the SAIDI and SAIFI regressions, both with and without major events included.
Major events are causing decreases in U.S. power system reliability over time
A key finding of this analysis is that there is an increasing trend in the annual average number of minutes of power interruptions over time. The trend is larger when major events are included, which means that increases in the severity of major events over time has been the principal contributor to the observed trend. Fig. 9 and Fig. 10 show the year coefficients for all seven SAIFI and SAIDI models, respectively, both without and with major events included. Fig. 9 shows that both when major
Research implications and conclusion
The principal finding from this research—that reliability is getting worse over time due to severe-weather related increases in annual average power interruption frequency and number of minutes customers are without power—has important implications for planners, policymakers, and other industry stakeholders. At the highest level, this finding suggests that increased attention to preparation for and recovery from major events may be warranted. Utilities and regulators should consider planning
Acknowledgment
The work described in this report was funded by the Office of Electricity Delivery and Energy Reliability, National Electricity Delivery Division of the U.S. Department of Energy (DOE) under Contract No. DEAC02-05CH11231.
References (44)
- et al.
Fixed effects, random effects or hausman–taylor? A pretest estimator
Econ Lett
(2003) - et al.
Estimating global impacts from climate change
Glob Environ Change Part A
(2004) Energy sector vulnerability to climate change: a review
Energy
(2012)2011 service quality report to the public utility commission of Texas
(2012)- et al.
“A reliability model for distribution systems incorporating seasonal variations in severe weather”
IEEE Trans Power Deliv
(2011) Challenges in reliability, security, efficiency, and resilience of energy infrastructure: toward smart self-healing electric power grid
- et al.
Instrumental variables and the search for identification: from supply and demand to natural experiments
J Econ Perspect
(2001) - et al.
Ready or not, here comes the smart grid!
Energy
(2011) - Keywords: Smart grid; Energy modeling; Distributed energy; Energy policy, Cameron, A. C., and Trivedi, P. (2009)....
Weather-related power outages and electric system resiliency
Form EIA-861 data files
The impact of power market reforms on electricity price-cost margins and cross-subsidy levels: a cross country panel data analysis
Energy Policy
Cointegration, exogeneity, and policy analysis: an overview
Tracking the reliability of the U.S. Electric power system: an assessment of publicly available information reported to state public utility commissions
An examination of temporal trends in electricity reliability based on reports from U.S. Electric utilities
Distribution-level electricity reliability: temporal trends using statistical analysis
Energy Policy
Transmission and distribution system line miles
Understanding bulk power reliability: the importance of good data and a critical review of existing sources
Reliability, wood pole inspections, storm hardening plan, and storm preparedness initiatives
Investigating causal relations by econometric models and cross-spectral methods
Econometrica
Econometric analysis
Specification tests in econometrics
Econometrica
Cited by (22)
The quest to quantify the value of lost load: A critical review of the economics of power outages
2022, Electricity JournalThere is no economic case for new coal plants in India
2021, World Development PerspectivesThe impact of variable renewable energy resources on power system reliability
2021, Energy PolicyCitation Excerpt :We conclude in Section 6 by providing a summary of the main findings, discussing the policy implications of this work, and potential future work from this research. Much of the prior published work related to electric grid system reliability has focused on uncovering time-trends in bulk power system (BPS) interruptions, which have implications for public policy and investment decisions surrounding the revitalization of the U.S. electrical grid (Eto and LaCommare 2008; Hines et al., 2009; Larsen et al. 2015, 2016).5 Past findings suggest most adverse system interruptions, when and if they do occur, occur at the distribution level (Hines et al., 2009; Eto et al., 2012).
Severe weather, utility spending, and the long-term reliability of the U.S. power system
2020, EnergyCitation Excerpt :γ measures the change in reliability over time. The data considered in this analysis largely follows earlier research by Larsen et al. [16]. However, key improvements involved adding: (1) three additional years of data (now through 2015); (2) more robust measures of extreme weather; and (3) additional details on annual utility capital and operations and maintenance spending.
Improving the estimated cost of sustained power interruptions to electricity customers
2018, EnergyCitation Excerpt :In our 2006 study, we were not able to consider the impact of customer-initiated purchases of equipment designed to protect against the impacts of power outages. Recent work has highlighted the importance of installing protective devices to shelter against possible interruptions, especially critical infrastructure [27]. In the current study, we considered the impacts of customer-installed backup generators designed to protect against the effects of electricity interruptions.