How does a utility deliver the right level of reliability when it serves different customers with different needs from the same point in the network? If a utility upgrades the network to accommodate its most sensitive customers, the system tends to be over-built for the surrounding customers. And, if utilities don't upgrade the network, sensitive customers are left to their own devices to compensate for grid disturbances.

Because customers' needs vary from site to site, it's difficult to equitably set utility design standards that meet the reliability needs of all customer types in a given area. Furthermore, if societal costs are to be minimized, optimizing the network design will require utilities to take a “minimum requirements” approach to ensure that power reliability meets, but doesn't exceed, customers' needs.

As deregulation progresses and retail competition drives local distribution companies (LDCs) into more of a “wires only” delivery role, the need to determine minimum design requirements will increase. Utilities can take different approaches to determine the minimum design requirements, but each approach involves applying reliability standards to ensure LDCs are meeting their service responsibilities. Approaches include benchmarking, performance-based rates, differentiated service and service guarantees.

To address the above issues, T&D World sought the insight of Tom Eyford, PacifiCorp's manager of risk planning, Cheryl Warren, National Grid's manager of T&D systems engineering, and Richard Brown, a principal consultant with KEMA. Their comments follow.

T&D World: Should utilities benchmark against one another?

Eyford: Utility benchmarking can be a challenging exercise with suspect value. Utilities have been reporting reliability statistics for decades; however, only recently have the problems associated with accurate collection and reporting of these statistics been widely appreciated. Over the last 10 years, a significant percentage of utilities have implemented outage management systems to drive automated outage reporting. These systems have undeniably increased the accuracy of such reports, but they have also highlighted a very important fact: Most utilities historically severely underreport their reliability measures, with some reporting triple or more outages than they had previously. As a result of this implementation, a dichotomy now exists between utilities with automated outage reporting and those without, making benchmarking studies including utilities in both groups nearly meaningless. In order to benchmark effectively, it is necessary to include utilities similar not only in size and makeup, but also in outage management/outage reporting techniques and experience level.

Warren: I think using the newly developed IEEE “2.5 beta methodology” is a first step to comparing apples to apples — albeit, Granny Smiths to Red Delicious. This method allows utilities to use the same process to determine when either their system design or operating capability has been exceeded. It's based on daily SAIDI. Once the SAIDI per-day threshold has been exceeded, that day is classified as a major event. The major event days are reviewed individually while the performance from the remaining days is used to calculate traditional reliability metrics (SAIFI, SAIDI and CAIDI). Using this approach will eliminate one of the two hurdles to obtaining comparable benchmarks, namely that all utilities will be reporting on the same basis. The second hurdle is to obtain information that has been collected with the same level of accuracy. In order to get truly comparable benchmarks, data must be collected in the same manner. Some other key information should be provided during the collection process, such as customer density, percent of overhead versus underground and other system measures.

T&D World: Of what value are customer surveys?

Brown: There are some countries, such as Norway and Sweden, that link utility revenue to inferred customer interruption cost based on surveys. The hope is that this approach will motivate the utility to improve reliability until the marginal cost of improvement is equal to the marginal benefit to the customer. This approach is problematic on many levels. First, customer surveys always overstate a customer's willingness to pay, especially if the customer knows that these surveys will impact reliability. Second, this thinking implies that customer-specific reliability improvements should be paid in full by the utility, but only for customers with sensitive loads.

T&D World: Should we have industry standards?

Warren: There are already many standards being employed throughout the United States to assess service quality. In some cases, penalties and/or rewards are assessed based on service quality (meeting pre-established targets). The current regulatory environment for distribution assets is mandated at the state level, giving rise to the potential for 50 different approaches to reliability standards. Even though some would argue that standards are detrimental to the industry, I think standards are necessary to ensure customers and regulators that customers are receiving at least a minimum level of reliability.

To date, most performance-based rate (PBR) systems that have been designed may not be achieving the goals that regulators were striving to achieve. Dr. Roy Billinton, Dr. Richard Christie and Dr. Richard Brown have written papers describing the impact of current service quality penalty plans, with all three reaching similar conclusions.

T&D World: How should standards be applied against the different customer types?

Warren: Inherently embedded in this question is the cost question. If standards — state or national — are mandated, how much will it cost utilities to meet the new standards? System improvements or degradations will be required to move from the current reliability level to the mandated level. Some other key questions are: How much will it cost customers and are they willing to pay? How can reliability improvements be quantified at the customer level? There is a normal distribution of customer experience rooted within the traditional SAIFI and SAIDI indices we use today. Some customers experience 15 interruptions each year while others experience one or none. In some cases, the customers who experience many interruptions are more satisfied than those with fewer interruptions; it all depends on their expectations. Customer satisfaction is a tough issue to quantify as it relates to both cost and reliability.

T&D World: Should standards be mandated or enforced at the state level?

Warren: If you ask a state regulator, they would tell you it is a state right to regulate distribution utilities. The issue, of course, is their autonomy, which in turn leads to potentially 50 different approaches. It would be more cost effective for all involved to have one uniform standard, not to mention that customers would benefit by having a common level of expectation that would transfer from state to state. One way to obtain one common standard is to set it at the federal level. We will still have to deal with the cost required to move to a new level and, in some cases, with customers seeing a degradation in service where they previously had exemplary service.

In order to achieve this end, much work on asset value versus performance would be required. Before a utility can move from its current level of performance to an arbitrary new level, certain things must be ascertained. In 1996, Peru undertook such an exercise. At the time, its systems were 30% underground and 70% overhead. They wanted to determine the cost and performance differential if, over time, it moved to a 30% overhead and 70% underground system.

T&D World: Are PBRs a viable approach?

Eyford: PBRs are definitely a favorable option. Properly implemented, PBRs create a very real economic incentive for utilities to provide the highly reliable electric service that today's customers are coming to expect. However, targets must be set responsibly; excessively stringent requirements may drive costly design and operating standards. Utility commissions must be cognizant of the trade-off between reliability and cost; in general, a sizable amount of “low-hanging fruit” for improving reliability still exists in many utilities, but significant improvement beyond these areas can be prohibitively expensive.

An interesting concept for reliability design is that of differentiated service. Urban areas cost less to serve than rural areas; because urban customers pay the same rates as rural customers, currently urban customers partially subsidize their rural counterparts. A potentially more equitable solution is to provide more reliable service to areas that cost less to serve, while still retaining an adequate minimum level of service for all customers.

T&D World: Is there a tendency to set reliability targets too low?

Brown: Reliability targets are typically set based on historical performance, such as “10% better than the average of our last five years.” Although these types of targets are generally in line with customer expectations, they are probably too highly based on customer willingness to pay. Existing reliability is definitely too low for some, but most would probably prefer to pay less and wouldn't even notice 10% or 20% more interruptions.

T&D World: Will the penalties imposed be large enough to encourage continued investment?

Warren: If designed properly, PBRs can be a viable approach. Moving from system-wide averages to customer-based numbers will provide greater insight into reliability problems. New PBR designs will be required in order to make them more effective.

There are other issues besides size of penalty to be considered when answering this question. Not only do PBRs often carry a monetary penalty, but they also carry a reputation penalty. Consistently paying penalties of any size leaves the impression of poor performance. Some states have taken a slap on the wrist level of penalty while others have assessed millions of dollars. One question remains: What do we do with the fines? They could be applied to state-mandated reliability programs, they could be refunded to customers or they could be used to offset future performance.

T&D World: Isn't there a significant financial risk to natural unforeseen factors, including weather?

Brown: Electric utilities have historically had the least volatile of all stock prices, and any potential increase in financial risk will not be received well by the investor community. That said, most performance penalties in the United States are immaterial to utility earnings. A penalty of $3 million might seem high, but not when compared against revenues of $3 billion. Much more important is the regulatory response to reliability performance, such as the ability of utilities to retain cost savings achieved under rate freezes and the maximum allowable rate of return.

Warren: Most PBR plans do not have a large enough disincentive to cause utilities to make infrastructure changes solely on that basis.

T&D World: After minimum design requirements have been established, should the cost of enhanced reliability be borne by the customer?

Warren: Getting to minimum standards will be a monumental task, both in terms of expense and time to arrive at new service levels. If a customer wishes to receive even more reliability than minimum standards provide, additional fees should be leveled. The age-old problem is how to provide one customer with more service than another on the same feeder. Maybe DG solutions can provide some assistance in this area.

Whether or not this best serves the customers depends on the customer. Residential customers, probably; industrial/commercial customers will continue to demand higher reliability without additional cost.

T&D World: Do cost-effective reliability solutions result from a reliability-based design?

Eyford: Utilities must be careful to balance reliability-based design with prudent reliability improvement. While a solid set of planning guidelines is important to ensuring long-term reliability improvement and sustainability, relying completely on design standards is very shortsighted. A comprehensive circuit-hardening program, coupled with a review of operational practices, can yield impressive results. Generally speaking, network-based reliability results in a greater overall reliability impact than does reliability on customer premises. Beyond the cost impacts, implementing strictly customer-based reliability solutions can have a negative overall impact on the network, as well as creating a situation where new customers may experience vastly different reliability than their neighbors, resulting in unpredictable planning and ultimately higher connection costs. However, in many cases, customers may want more reliable service than their differentiated area may warrant. In these cases, the customer should be responsible to pay for the increased reliability, and the most economical solution (network versus customer premise) should be studied and implemented.

T&D World: Would differentiated service best drive the reliability marketplace?

Brown: In a sense, differentiated service already exists. Customers are free to augment their reliability through backup generators, uninterruptible power supplies and a host of other options. Since most people choose not to purchase these products, one can safely assume that reliability at the meter is probably too high for most.

Differentiated service also exists in many urban areas where certain customers can opt for multiple supplies or spot networks. On a broad scale, differentiated reliability is unlikely because it is difficult to customize meter-level reliability in a cost-effective manner. If a single residential customer wants high reliability, an in-home solution makes the most sense. If an entire neighborhood wants high reliability, it can certainly approach the utility for potential solutions.

For now, high existing reliability precludes differentiated service on a broad scale. More realistic is a set of reliability guarantees that allows customers to choose lower levels of service as an option. If many customers in an area choose low rates and low reliability, the utility has a politically palatable mechanism to lower reliability in this specific area. Low-cost electric meters could be used to automatically track customer-level reliability and credit monthly bills if reliability falls below guaranteed levels.

T&D World: Will reliability guarantees really ensure enhanced network investment or performance?

Warren: It is doubtful that reliability guarantees will ensure network investment. Often, the problems experienced by customers having the most issues cannot be solved with network solutions. Instead, solutions to those problems require expensive, custom solutions for customers who are served from electrically remote portions of the network. Maybe DG will assist with resolving this type of customer issue, but they will come with a cost.

T&D World: Will reliability guarantees closely follow expected performance or will utilities just become better risk managers? If these guarantees require utilities to take on more risk, is that a role desired for LDCs?

Brown: Reliability guarantees on a broad scale will force utilities to become better risk managers. They will have to price premiums correctly, and make system modifications to both improve expected profit and to improve the predictability of profit. Certainly, the utility will be motivated to improve reliability in areas with high expected payouts. Free riders will still remain a problem, such as when everyone in a neighborhood signs up for high reliability except a few people. Why should these people get the same benefits of reliability improvement but pay lower rates?

T&D World: When network improvements require upgraded feeders, how do you deal with the free riders who aren't paying for the enhanced reliability?

Eyford: This is the age-old question. When a customer requires greater reliability, the improvements may result in similar improvements to neighboring customers who receive it for free. Utilities deal with this situation all the time; where economical, utilities currently install redundant service, improving reliability for customers on the affected circuits. Customers who happen to be in close proximity to their substation on well-protected lines also experience fewer outages than those at the end of long radial lines. The way to deal with this situation is the guarantee process. If a customer is paying for more reliable service, it has recourse with the utility to ensure that its level of service remains at the appropriate level. Those customers who are not paying for it have no such recourse, except if the level of service is below the minimum requirements.

About the Participants

Richard E. Brown is a principal consultant for KEMA in Raleigh, North Carolina. He has published more than 60 technical papers related to distribution system reliability and asset management, is author of the book Electric Power Distribution Reliability and is a registered professional engineer. He is a senior member of the IEEE, chair of its working group on Distribution Planning and Implementation, and recipient of the Walter Fee Outstanding Young Engineer award. Brown earned his BSEE, MSEE and PhD degrees from the University of Washington in Seattle and earned his MBA degree from the University of North Carolina at Chapel Hill.

Tom Eyford is manager of risk planning in the Distribution Asset Management department of PacifiCorp (Portland, Oregon). He is responsible for the long-range assessment of value and risk for all of the Distribution Business Unit CAPEX and OMAG expenditures. He co-holds U.S. patent #5,973,899, Automated Power Feeder Restoration System and Method, and has published an article on outage management implementation in Transmission & Distribution World. He holds a BSEE degree from Washington State University in Pullman, Washington, and an MBA from Utah State University in Logan, Utah.

Cheryl A. Warren is the manager of T&D Systems Engineering for National Grid USA Service Co. Inc. (Albany, New York). Her main areas of expertise are distribution reliability analysis, power quality, GIS/OMS and enterprise wide IT systems integration. Warren has authored and co-authored 23 technical papers. She chairs the IEEE Working Group on System Design that wrote the Guide on Electric Power Distribution Reliability Indices 1366-2004. She received her BSEE and MSE degrees from Union College in Schenectady, New York.