To: Toby Tyler cc: DAWG B Team From: Jim Price, DRA Date: 7/9/96 Re: Reply to PG&E re Load ProfilingMany thanks for your thoughtful response to DRA's position paper on load profiling. I hope further discussion of this issue can result in a solution acceptable to all parties, and was glad to see your acknowledgements that (1) if we are to proceed with load profiling as a means of enhancing consumer choice, stratifying load profiles within customer classes represents a desirable goal, and (2) where customer load patterns differ in a meaningful and identifiable way, we ought to work toward developing profiles that capture these differences.
However, your comments left me wondering about a couple of points. First, how do you propose to make direct access (presumably via aggregation) available to small customers (residential, small commercial, and agricultural) during the period of time before hourly meters become cost effective at low usage levels, and before meters can be installed in large numbers? It is simply not realistic to move ahead without taking the needs of these customers into account, and this is the point of implementing load profiling. At the end of your comments, you suggest that if a load profiling option is offered to customers, it ought to be offered on a rate schedule basis. An issue raised in Phase 2 (Rate Design) of both the recent PG&E and Edison GRCs is whether single- vs. multiple-family dwellings, and large vs. small residential customers have different load characteristics. In the Edison GRC, a different customer charge was adopted for multiple-family dwellings because this difference could be established; in PG&E's GRC, TURN is asking for information to be developed to address this cost-of-service issue. Although PG&E did not have data available to address this issue for Test Year 1996, cost-of-service differences are easily identifiable for other groups within the residential (as well as other) classes -- that's why PG&E uses stratified statistical sampling. It is unrealistic to assume that aggregators' customer populations will have the same mix as customers remaining under UDC service, and since differences in cost-of-service have already been identified between different strata, we must be prepared to reflect those differences in ISO settlements. Thus, offering load profiling only on a rate schedule basis does not account for the needs of small customers.
Second, I sense some possible confusion as to the timing needs and the cost-effectiveness of load profiling. Perhaps using the term "statistical sampling" would be clearer. In past uses of load research data (i.e., in rate cases), noticable time requirements have indeed been present, to collect a year of data and then analyze it and present it during construction of a utility's rate case proposal, which is then litigated before the Commission. For ISO settlements, though, the use of the data would be real time: forecasting would need to be used only for submission of load schedules to the PX and ISO, and data collected in real time would be used for final settlements once monthly meter reads occur. That is, any needed enhancements to the load research sample do not need to be in place for settlements before the time when the data is used. This is indeed putting the load research data to a different purpose than its original one, but that is a common characteristic of many aspects of the electric industry restructuring. As for cost-effectiveness, are you arguing that it is more expensive to install meters for a sample of customers than for the entire customer population to which competitive options will be offered? If so, please explain your reasoning. I believe that enhancing the load research sample so that it will be acceptable for settlements of small customers' loads, among aggregators and the UDC, is both feasible and cost-effective. Allowing the load research sample to be enhanced prospectively, as needed, could also offer a way to allow aggregators to achieve RTP rate response for their customers: procedures could be developed for segmenting load research strata for customers for whom particular load management devices are installed, without relying on engineering estimates of the devices' effectiveness. Of course, for load research enhancements of this type, it seems appropriate for the beneficiaries to pay the added cost.
Similarly, if you believe the use of load profiling will somehow require the UDC to develop new tariff schedules, you should explain your reasoning. We can then develop solutions that meet the Commission's stated goal of offering competitive options to all types of customers. I believe the development of new tariff schedules will not be a requirement, but rather a marketing decision of the UDC. SDG&E referred at the July 2 meeting to a perspective on obligation to serve, by which the obligation of the UDC is to provide distribution services to all customers, to connect them to the competitive marketplace, and by which it is the PX that has an obligation to provide generation services. This context seems to match the fact that the UDC's primary function is no longer to market generation. I infer that the UDC would still have a role in providing brokerage services (e.g., it would be submitting a load forecast to the PX and ISO), but it would be up to the UDC as to how aggressively it markets its brokerage services.
As for your final comment that "if limited to the right set of customers, load profiling should not have to contribute to any additional cost shifting", as well as your comment in response to UCAN that "net metering could introduce unintended cost shifting", I'm left puzzled. The load at an upstream point will be the sum of downstream loads plus line losses, which can be calculated. Net metering doesn't produce cost-shifting -- it defines a boundary for billing via load profiling, which reduces its margin of uncertainty. Net metering, as I understand it to work, would take the directly metered hourly loads as a given, and divide the residual among customers who have "agreed" to be served via load profiling. Some of your concerns do support my own inclination to calculate net metering at the Distribution Planning Area level (instead of at the substation level or lower levels): this would take the "looped" design of the distribution system into account, avoid any illusion that placement at the final line transformer level has metered the individual customer, and would achieve the best cost-effectiveness by reducing the number of primary distribution-voltage meters that are required.
As stated in my earlier paper, the task before us should not be to debate the merits of load profiling, because it is necessary for at least some period of time to meet the stated goal of competitive options for all customers. Instead, our task should be to identify the details needed to make load profiling work. I'm eager to hear your thoughts on these implementation issues.