Quantifying Reliability and Service Quality
To ensure proper incentives are in place to maintain traditional levels of reliability and service quality, regulators and utilities are using different approaches, including performance-based, prescriptive and customized programs, to meet customer needs.
Regardless of the approach, any attempt to regulate or compare service quality requires quantifying reliability and service in a consistent manner. This article examines common methods of calculating and reporting performance, as well as some important developments in trying to quantify reliability and service quality.
The Trend Toward Performance-Based Rates
The concern for deteriorating reliability levels has been a major focus for many regulators. This concern applies primarily to the distribution company, or wires company. Based on a 1999 survey conducted by Edison Electric Institute (EEI), 92% of all outages experienced by customers resulted from distribution causes. In order to assure adequate levels of reliability, regulators are rethinking the cost of the service/rate of return (COS/ROR)-based rate-making mechanism traditionally used by regulatory bodies. COS/ROR regulation permits utilities to charge rates that allow them to recover reasonable operating expenses and to earn a fair return on investment.
This regulatory trend is toward incentive-based regulations, which are broadly known as performance-based rates (PBR). In mid- to late-1990, regulatory efforts in the United Kingdom and Australia resulted in performance-based incentives/penalties for distribution companies. This trend moved through North America with the implementation of various forms of penalty/incentive criteria adopted by Public Utility Commissions (PUCs) in more than 11 states. Today, more than 27 U.S. states have some type of reliability standards — an increase of 800% since 1996 — which may give way to more performance-based contracts.
These types of policies link an electric utility's performance indicators with the rate structure and ROR allowed for a given utility. PBRs are essentially where a distribution company is penalized or rewarded based on sustained interruptions as quantified by reliability indices.
In its most general form, the PBR scheme is a contract that rewards utilities for good performance and penalizes them for poor performance. The jury may still be out as to whether PBR is a valid regulatory approach for improving or maintaining the overall quality of supply of a distribution system. Nevertheless, regulatory agencies and other policy makers worldwide are implementing such schemes in increasing numbers. Figure 1 illustrates the reliability improvement (average customer minutes lost) achieved in Italy since the implementation of rate incentives in 1998-1999. “Customer minutes lost” represents the total interruption duration that is characterized as the responsibility of the distribution company (regulated portion of the customer minutes lost). The “customer minutes lost total” also includes interruptions caused by “third parties” and “acts of God” that are not characterized as the responsibility of the distribution company.
Common Indices and Calculation Methods
With this new trend, regulators and other policy makers are searching for ways to benchmark individual state utilities against other states as well as against the utility industry in general. Some utilities also pay bonuses to managers and other employees based in part on indices. Some commercial and industrial customers factor such indices into their decision-making process for locating a facility within the utility's service territory. All of these applications for reliability measures require that indices are calculated in a consistent manner.
IEEE Standard 1366-2003 provides definitions for the most important indices used to characterize reliability. A survey conducted by the Working Group on System Design showed that the indices most commonly used for tracking reliability are SAIFI, SAIDI, CAIDI and ASAI. The standard also proposes a new method of identifying “major events” for purposes of reliability calculations. This is very important because major events, such as ice storms, hurricanes and other system problems that affect very large numbers of customers, can have a disproportionate impact on the overall reliability indices and may distort the reliability performance that is actually attributable to the design and operation of the distribution system.
Identifying Major Events
Major events also can cause significantly increased variability in the reported reliability levels. Figure 2 illustrates 10 years of reliability data for utilities that report data with and without major events included. Obviously, the effect is important. Unfortunately, there are many different ways to determine whether events should be categorized as major events, and this can result in significant differences in reported indices. The IEEE 1366-2003 standard recommends a statistical approach for categorizing major events that result in much more uniformity in reporting reliability indices if it was adopted. It is based on the assumption that daily reliability-performance indices exhibit a log-normal distribution when plotted over a whole year. The average (alpha) and the standard deviation (beta) for the distribution of daily SAIDI values can be used to determine the events that should be considered major events. An approach of using events that are more than 2.5 beta from the average value is recommended.
| Definition & Data Classification | Service Territory |
|---|---|
| • Major Events • Interruption • Planned/Unplanned • Distribution/Transmission • Animal Activity |
• Geography • Weather Pattern • Vegetation Pattern • Vehicle Access Pattern |
| Data Collection Process | System Design |
| • Outage Notification • Outage Reporting • Step Restoration Process • Customer-to-Network Connectivity |
• Urban/Rural/Downtown • Load Characteristics • Underground/Overground • Voltage Level • Protection Scheme |
Other Factors Affecting Reliability Performance
Even if the reliability indices are calculated in a consistent manner, it is not possible to compare reliability performance across different systems. There are several factors that affect reliability, and it would not be reasonable to expect all utilities to be able to achieve the same levels. Therefore, future efforts should focus on characterizing the effects of different factors and coming up with an approach for normalizing the calculated reliability levels for factors that are outside the control of the utility. The table illustrates some of the factors that must be considered.
A Probabilistic Approach for Characterizing Reliability Performance
Because reliability levels vary from site to site around the system and vary from year-to-year to a variety of factors, it is reasonable to try and represent the expected performance using probabilistic methods rather than with simple indices. The probabilistic characterization can help in understanding the uncertainty and the variability inherent in reliability indices.
Variability represents heterogeneity or diversity that is inherent in the service-quality data (or, for that matter, in any real-world data). Some feeders will perform better than the others due to differences in topology, weather and existing system conditions. Also, some years will be more severe than others in terms of storms, lightning flashes, tornados and a slew of other effects. Therefore, variability is present across the system data (spatial variation), as well as over a period of time (temporal variation). An average does not reflect variability in the data.
Uncertainty arises due to the limited amount of information (called sample in statistical jargon) that is used to characterize the entire dataset (called population). Uncertainty is unavoidable in service-quality data. No utility has resources to monitor all the feeders in its service territory. System performance should be predicted from the limited information available. Also, monitored data at a feeder will be available only for a limited period of time. In any event, an average cannot account for uncertainty.
A probabilistic risk assessment approach enables consumers to see the full range of variability and uncertainty as opposed to presenting service-quality indices as simple point values. Figure 3 shows a sample two-dimensional representation of variability and uncertainty using a probabilistic framework for characterization.
Predicting Ongoing Reliability Levels
Many utilities track reliability performance on a month-to-month basis, and employee bonuses may even be tied to meeting reliability objectives. Dashboard summaries are developed that provide an ongoing view of the reliability levels. When tracking reliability levels through the year and trying to understand what the end of the year results might be, it is important to take into account the normal seasonal variations in interruption performance. This can be done in a probabilistic manner as well, and ongoing predictions can be provided for ongoing reliability performance that takes into account historical seasonal variations. Figure 4 is an example of such a prediction showing the range of possible levels that can be expected for the remainder of the year. Figure 5 shows a Web-based application for forecasting reliability. The tool was developed for a utility using the Rpad technology (www.rpad.org).
This type of analysis is preferable to a straight-line prediction without any probabilistic assessment. Better evaluations of possible measures to improve performance can be made with this type of information.
Service Quality Needed
As mentioned previously, different customers have different expectations and needs regarding the quality of service. For some customers, limiting the number and length of long duration interruptions is all that is needed for their service. Other customers may be sensitive to momentary interruptions or even instantaneous voltage sags. Many regulators are starting to require reporting of momentary interruption performance (MAIFI) along with traditional reliability indices. Many utilities also are tracking voltage sag performance (SARFI), so the information can be provided to customers that have economic impacts associated with these events. Fortunately, monitoring systems and tools for calculating expected performance in all these categories are available.
The probabilistic techniques described above can be applied to the wider range of indices that may be important as well. Figure 6 illustrates an example of an approach for calculating the overall service quality that can be expected at a particular location, taking into account traditional reliability characteristics and important power-quality characteristics. This service characterization tool is currently undergoing field trial by Duke's PQ Group and its Customer Service Group. Ultimately, Duke hopes to link the methodology with the overall Duke Power customer connectivity and reliability database to produce real-time service-level characterization at any part of the network based on historical performance data.
Future Efforts
Significant progress has been made in developing standardized approaches for characterizing reliability and even service-quality levels. Consistent methods of identifying major event days are a good starting point, and many regulators are considering the recommendations of the latest IEEE 1366 standard. The next areas of focus will involve development of methods to normalize reliability levels for critical factors that are different from system to system. These normalization procedures, combined with probabilistic methods for characterizing the uncertainty and variability of the information, will allow more direct comparison of system performance around the country and around the world. These same concepts than can be applied to the wider range of service-quality characteristics that may be important for many customers. Finally, as customer requirements are better defined, service quality can be tailored to the needs of individual customers and rates can be designed for these customer specific services.
Arshad Mansoor, vice president of engineering of EPRI PEAC, provides technical leadership to a variety of engineering projects, including power system analysis; power distribution system reliability and quality evaluation; industrial and commercial power-quality analysis; distributed generation and renewable resource application; and assessment of performance-based rate structure for electric utilities. Mansoor has conducted more than 50 training courses, published numerous IEEE papers and presented in many industry forums in these areas both in the United States and overseas. He received a BSEE degree from the Bangladesh University of Engineering and Technology, and an MSEE degree and Ph.D. from the University of Texas in Austin.
Mark McGranaghan, vice president of consulting services for EPRI PEAC, coordinates a wide range of services offered to electric utilities and critical industrial facilities throughout the world. These services include research projects, seminars, monitoring services, power system analysis, testing, failure analysis and designing solutions to improve system performance. His main areas of focus have been power-quality monitoring systems, development of power-quality programs, analysis of distribution systems, customer services and analysis software applications. McGranaghan received his BSEE and MSEE degrees from the University of Toledo and his MBA degree from the University of Pittsburgh.
Karen Forsten manages operations and technical transfer services for EPRI PEAC Corp.'s power-quality and reliability programs. She works with clients in various technical programs and projects, including Security, Quality, Reliability and Availability (SQRA), CEIDS and strategic technical business development. Additionally, she manages and coordinates all technical seminars, training, conferences and workshops for EPRI PEAC. She is an active member of the IEEE, has published papers in journals and conference proceedings, and has served as chairperson and speaker in various conference forums. Forsten received her BEE degree from Auburn University in Alabama.
For more information about EPRI Quality and Reliability programs and activities, or EPRI PEAC's PQ Knowledge-Based Services Program 97, contact Karen Forsten at 865-218-8052 or go to www.mypq.net.
Want to use this article? Click here for options!
© 2008 Penton Media Inc.
