6. Load Forecasting Issues

6.1. Best Estimate vs. Current Customers
Approach

D.04-10-035 adopted a protocol whereby LSEs are required to submit load forecasts using their best estimates of future customers and their loads. The Commission rejected an alternative approach that would have required LSEs to assume that their customer base will remain fixed for the forecast period, i.e., that load migration will not occur. This issue resurfaced in the Phase 2 workshops, and TURN filed a petition for modification in which it sought to vacate the Commission's adoption of the best estimate approach and to defer final resolution of the matter to this Phase 2 decision. In support of its petition, TURN noted that Phase 2 workshop discussions addressed several alternative proposals for dealing with the effects of customer load migration. AReM and Sempra Energy Solutions filed responses in opposition to the petition.

TURN's petition refers generally to workshop discussions regarding the problem of customer load migration and alternatives for dealing with it, but it does not refer to any particular impacts on LSE load forecasts or point to any specific new facts or arguments that arose in those discussions. We therefore conclude that the petition is procedurally deficient.14 The question of which load forecast method to use has already been resolved, and we will not revisit the question here. In accordance with D.04-10-035, LSEs should prepare and submit hourly load forecasts based on the best estimates approach.

As the CAISO noted in its comments, an organized capacity market might provide LSEs with a means of addressing the impact of load migration on their RA obligations. While we deny TURN's petition for the reasons stated herein, and we are committed to going forward with the RAR program using the best estimate approach, we are willing to revisit this topic at an appropriate time in the future. In particular, if a capacity market is in place and it has been shown that the load migration problem can be readily addressed by the ability of LSEs to acquire and dispose of increments of capacity sufficiently small (and located where needed) to match such migration, then it would be reasonable to revisit this topic. We note that Sempra Global, an opponent of TURN's proposal to vacate the best estimate approach, agrees that a capacity market would readily accommodate load migration.

6.2. Coincidence Adjustment Methodology

D.04-10-035 provided that RA obligations should rest upon coincident peaks rather than the unadjusted peaks of each LSE. Two alternative approaches to the coincidence adjustment were discussed in the Phase 2 workshops: (1) use of historic coincident factors (historic approach) and (2) determination of coincident peaks directly from the hourly load forecasts submitted by the LSEs (forecast approach). The workshop report described advantages and disadvantages of both options and invited comment on which of them should be adopted. The comments revealed preferences for both options.

We adopt the historic approach. While, in theory, forecasts might be more accurate (and as CAISO observes, more in line with our decision to use the best estimate rather than the current customers approach), we have insufficient experience with these forecasts to justify making that conclusion. It may be the case that the historic approach is just as accurate, if not more so. As PG&E notes, using load forecasts based on differing forecast methods of individual LSEs could introduce "forecast noise" to the analysis. SCE and SDG&E make similar points. SWP/SWC also underscore the report's point that the historic approach simplifies the planning process and would permit the coincident adjustment factor to be identified earlier. At least until we have gained experience with the RAR program, we think this benefit outweighs any theoretical gain in accuracy that might be realized with the forecast approach.

In addition to the issues identified in the workshop report, PG&E makes two additional recommendations for coincidence adjustments. First, non-coincident load should be defined as the difference between (1) the sum of the LSEs' non-coincident peaks and (2) the CAISO control area's coincident peak, expressed as a percentage of the sum of the LSEs' non-coincident peaks. According to PG&E, this percentage is the average coincidence adjustment factor of all LSEs in the control area, and it takes advantage of the pooling effect of LSEs with diverse peaks and load shapes within the CAISO control area. Second, PG&E recommends adoption of a single adjustment factor for all LSEs. Thus, each LSE's forward procurement obligation would be its final, forecasted non-coincident load for a month, as determined by the CEC, reduced by a factor that reflects the average load diversity in the CAISO's control area in that month. As PG&E notes, averaging is more stable and easier to calculate, monitor, and apply. We adopt the PG&E approach, and grant discretion to the CEC to determine the exact method by which the PG&E approach is implemented.

6.3. Allocating Demand Side Impacts

D.04-10-035 outlined how energy efficiency (EE), demand response (DR), and distributed generation (DG) programs will affect load forecasts. While dispatchable demand side options will be considered as resources and counted as qualifying capacity, non-dispatchable DR and EE programs will be accounted for in load forecasts.

The Phase 2 workshop discussions regarding the quantification of EE and DR impacts and the allocation of those impacts to LSEs led to a working group (WG) paper which includes the following allocation recommendations:

Regarding the allocation of the EE/DR impacts, the WG recommends that for RA purposes the impacts associated with the utilities' EE/DR programs be allocated to the LSEs in fixed proportion using metrics that are transparent, equitable, and relatively simple to quantify and apply.

In principle, the EE/DR impact should be allocated to the LSEs in proportion to the funding their respective customers provide toward the utilities EE/DR programs. In order to simplify the allocation process, as a proxy for their funding contribution, the WG recommends allocating the impacts in proportion to the LSEs energy sales, as follows:

· For EE programs, the WG recommends using the percentage of total IOU retail sales (i.e., bundled plus DA) represented by incremental EE savings for each utility to determine an LSE's share of that utility's incremental EE impact. [Material omitted.]

· For DR programs, as a proxy for funding the WG recommends using the percentage of each LSE's sales to the sum of all LSEs' sales within a utility's service area to allocate that utility's DR impact. Because an LSE's funding contribution to a utility's DR programs can vary by program (at least in the case of the CPA program now administered by PG&E), the allocation percentages for DR impacts can vary by program. [Footnote omitted.]

This report provides estimated allocation percentages for the existing utility EE/DR programs by utility. The WG recommends that the utilities determine the EE/DR RA impacts and allocation percentages annually. (Phase 2 workshop report, Appendix C, pp. 1-2.)

Even though the WG paper is denoted as a draft, we understand that it is the WG's proposal. As set forth in the Phase 2 workshop report, staff believes that the WG paper generally makes sense, but that certain elements were problematic. For example, staff expressed concern that the report did not recognize the distinction between programs that are dispatchable and those that are not. Upon review of the comments filed by PG&E and SCE, we are persuaded that the staff's concerns have been addressed and do not require further discussion. Nothing in the WG paper is inconsistent with D.04-10-035's determination that dispatchable programs are to be counted as resources, and the paper supports the idea that EE impacts are to be debited from load forecasts. We accept the WG recommendations for allocation of demand side impacts and adopt them as our own.

6.4. Measurement and Evaluation

With respect to the quantification of the EE/DR impacts, the WG recommended that parties continue using their present methodologies, and review and evaluate those methodologies based on the results from measurement and evaluation (M&E) efforts currently planned for next year in R.01-08-028 (in particular the December 30, 2004 ALJ ruling) for EE, and in R.02-06-001 for DR.

The WG paper also notes a need for improved M&E efforts, and the workshop report observes that modification of existing M&E efforts for various program categories is a key linkage to resource adequacy needs that should be pursued in terms of the research design changes and funding required to accomplish these new studies in a timely manner. The workshop report goes on to recommend that the Commission "direct EE, DR, and DG [M&E] efforts to support the hourly load shape impact assessments that are necessary to the inclusion of the impacts of policy-preferred resources within RAR." PG&E and SCE are supportive of this recommendation. We adopt the staff's recommendation in principle, and ask that our staff provide us with specific recommendations for its implementation.

6.5. Responsibility To Quantify EE, DR, and DG
Effects

The Phase 2 workshop discussions surfaced the need for the three IOUs to prepare and document the hourly impacts of EE, DR and DG programs within their service areas and to provide these impacts to the CEC for use in the adjustment of LSE load forecasts. Staff notes that these impact evaluation responsibilities must be completed and documented for handoff to the CEC each spring. Staff notes further that to the extent that the Commission assigns programmatic M&E activities for EE, DR or DG to entities other than the IOUs, then these entities must also provide comparable impact products to the CEC.

Staff recommends that the IOUs and any independent evaluators be required to prepare EE, DR, or DG impacts according to the informational needs of RAR. While this recommendation is largely uncontested, SCE points to the need for funding of additional studies and SDG&E's concerns about its lack of historical data are applicable. We adopt staff's recommendation. While there is no funding proposal before us for studies, we commend to the appropriate EE and DR proceedings consideration of the data needs of the RAR program and the specification of, and funding options for, studies to develop such data.

The workshop report recommends that the Commission direct IOUs to make monthly estimates of EE, DR, and DG for all twelve months of the year despite any uncertainties of responsibility about program administration. PG&E supports this recommendation. SCE does not explicitly support the recommendation but it notes that program administration responsibility has been clarified, and that IOUs are better able to provide monthly forecasts for EE, DR, and DG programs. SCE also notes that IOUs have little control over many variables that can affect estimates, and that there can be large month-to-month variations in program impacts. Finally, SCE notes that the IOUs will have little choice but to rely on nameplate ratings for monthly estimates for DG.

SDG&E states that it does not collect monthly or hourly history nor are there studies to guide estimates. SDG&E believes that any attempts to quantify hourly or monthly impacts would be unreliable. Nonetheless, developing impacts in this manner is essential to estimating peak impacts. We will direct the IOUs to make monthly estimates of hourly EE, DR, and DG program impacts as recommended by staff. In view of the understandable data problems, the IOUs shall work with Commission staff and CEC staff to develop estimating methods appropriate for each IOU's existing measurement and evaluation data.

6.6. DG Impacts

Staff reports that the workshop participants generally agreed that DG impacts are less important than those of EE and DR. Where there are thousands of megawatts of aggregate impacts from EE and DR programs, DG programs appear to have no more than a few hundred megawatts. The participants essentially agreed that each IOU would prepare DG penetration and stereotypical electrical production patterns that would allow development of hourly impacts. These will be provided to the CEC for use in adjusting preliminary LSE load forecasts on a pro-rata basis like those of EE and DR. We agree with staff's conclusion that if DG impacts appear to become more significant in the future, then more sophisticated methods identifying impacts and attributing them to the specific customers of individual LSEs may become important.

At this time, a simple DG impact assessment methodology is acceptable for RAR forecasting. IOUs shall provide data to the CEC in accordance with the foregoing discussion.

6.7. Total Losses Methodology

D.04-10-035 directed LSEs to include all losses in their load forecasts, including distribution losses, transmission losses, and unaccounted for energy (UFE), and it directed further consideration of implementation details for this policy in Phase 2. In the Phase 2 workshops, the CAISO presented data that could form the basis for a simplified approach in which hourly distribution loss factors (DLFs) would be used with an upward adjustment of three percentage points for both transmission losses and UFE. The 3% adder would apply in all hours. The workshop participants agreed on this approach, and the comments reflected universal support for it, at least for initial implementation of the RAR program. We will adopt this straightforward approach for transmission losses and UFE, and leave possible refinements to future proceedings.

Staff is concerned that while the IOUs have DLFs available on their websites to support direct access load scheduling and settlement, these factors are intended for short-term purposes and may not be compatible with developing long-term forecasts. It is apparent that further study may identify DLFs that are more appropriate for purposes of the RAR program. For initial implementation of the program, however, the simplified approaches suggested by PG&E and SCE are adequate. CEC, in consultation with the IOUs, should develop DLFs on the basis of the best information available. The DLFs should be made available to all LSEs, and proposals to use website postings appear to be reasonable for this purpose.

14 While TURN's petition did not refer to such facts or arguments, its opening comments on the workshop report referred to workshop discussions of how the LSEs' month-ahead RAR showings might be updated to reflect customer migration. However, even if we were to accept this late-supplied information as resolution of the petition's procedural deficiency, and reconsider the best estimate approach, we are not persuaded that the ability to update forecasts a month ahead adequately addresses the LSEs' legitimate concerns regarding load migration. Additionally, it would be quite disruptive to the LSE load forecast review process that is already underway, and put timely implementation of RAR at risk. (See June 24, 2005 ruling of the ALJ directing LSEs to submit load forecast data.)

Previous PageTop Of PageNext PageGo To First Page