4. Peak Definition

By D.05-09-043 in A.05-06-004 et al., we addressed the utilities proposed energy efficiency portfolio plans for the 2006-2008 program cycle. As discussed in that decision, part of the controversy over whether these portfolios were likely to meet the Commission-adopted peak demand (kilowatt or kW) reduction goals stemmed from differences in opinion over what definition of peak demand should be used when estimating portfolio savings. For a variety of reasons, including limitations to load shape data for specific measures, the utilities used a mix of metrics in the E3 calculators to prepare their portfolio plan estimates of peak kW demand reductions.

We discussed at some length in D.05-09-043 the need to develop the record further in the 2006 Update with respect to the appropriate definition of peak kW for energy efficiency planning and evaluation purposes.²⁰ In particular, one of our tasks today is to adopt a definition that will be used for the ex post evaluation of program and portfolio accomplishments during the 2006-2008 program cycle. This evaluation will be used to assess (1) whether the utilities have met the savings goals established by D.04-09-060 for this program cycle and (2) whether the minimum performance threshold is met under the risk/reward incentive mechanism being developed in R.06-04-010. The Commission has directed that the minimum performance threshold be tied to the Commission-adopted savings kWh, peak kW and therm savings goals, in a manner to be developed in R.06-04-010.

As discussed during the workshops, there are various options for defining peak demand reductions for these purposes. Currently there are four measurements of peak kW reduction used in the E3 Calculator. The first measurement is based on the definition of peak used in the Database for Energy Efficiency Resources (DEER), a database developed jointly by the CEC and this Commission and funded by ratepayers. The second is a "load factor" based definition, whereby annual energy reductions are multiplied by a fixed conversion factor.²¹ The third measurement is based on prior TOU studies of kW reductions, which conform to utility TOU period definitions that vary by utility. The fourth measurement is coincident peak kW reductions based on hourly load shapes/end-use data, where that data is available.²² As discussed in the E3 report, the mix of peak metrics is driven by the available load data information and represents a "best effort" to estimate the summer on-peak impacts of the various energy efficiency measures.

In addition, workshop participants discussed a kW metric consistent with Resource Adequacy counting rules for demand response (DR) resources. The various peak metrics are described further in Table 1, together with a summary of the associated data requirements and the pros and cons of each approach.

4.1. Workshop Consensus and Non-Consensus

After lengthy discussion in the March workshops, the participants reached general consensus that the DEER definition of peak kW should be used for the verification of goal achievements and performance basis calculations/thresholds during the 2006-2008 program cycle. DEER defines peak kW as the average grid level impact for a measure between 2 p.m. and 5 p.m. during the three consecutive weekday period containing the weekday with the hottest temperature of the year. DEER identifies these three contiguous peak kW days for each of the 16 California climate zones, based on the weather data sets developed for the California Title 24 Building Energy Efficiency Standards.²³

There was also discussion during the workshop on the related issue of how the utilities should report program and portfolio accomplishments during the 2006-2008 program cycle as they are rebalancing their portfolios and reporting program and portfolio achievements prior to ex post verification of load impacts. For those measures that are included in the DEER database, the general consensus was that the 2005 DEER Update estimates of annual kWh and peak kW reductions should be used until additional EM&V study results are available that would update those DEER values.

However, participants recognized that the DEER values do not cover all program measures, and that the utilities use peak kW and kWh savings values that are not explicitly linked to DEER measures for custom applications where the specific mix of measures is not known at the outset. For this purpose, participants agreed that the utilities should continue to use their best estimates of kW (and kWh) impacts, which would also be subject to ex post verification using the DEER definition of peak kW discussed above.

Nonetheless, workshop participants articulated the need for Joint Staff and the program advisory/peer review groups to have additional information on these non-DEER values, in order to determine if further investigation is needed for the reported kW reductions as implementation proceeds. Accordingly, at the suggestion of the ALJ, workshop participants agreed on an action item to obtain such information.

Workshop participants did not reach consensus on a peak definition for use beyond the current program cycle (2006-2008). They did, however, identify two potential peak definitions for further consideration. The first was to continue the use of the current DEER peak kW definition (or potentially a variant that would shift the block of hours based on further research and updated system peak data). The second was to use a metric for coincident peak, possibly utilizing 12 monthly single hours that represent the highest loads on the system.

Workshop participants did agree that resolution of the definitional issue over the long-term is likely to depend upon the availability of load shape information. They discussed a framework for considering what energy efficiency load shape data will be needed in the future, in order to produce peak kW metrics for energy efficiency that address a variety of purposes in Commission proceedings. These included: energy efficiency goal attainment, resource adequacy, critical peak pricing, long-term resource planning, among others. Workshop participants noted that the development of hourly energy efficiency load data, or a subset of those hours during the summer peak, would provide the granularity of information needed to develop any of the peak kW metrics that might be needed.

Rather than defining the peak for all possible Commission purposes at this time, workshop participants reached consensus that effort should be focused on assuring that load shape research and EM&V efforts produce additional hourly load data in time for the 2009-2011 program cycle. They developed an action plan for this purpose, which is discussed further in Section 8 below.

Finally, workshop participants reached consensus that a critical peak metric should not be developed at this time for energy efficiency. They concluded that such a metric is not necessary for non-dispatchable energy efficiency, and that developing such a metric is likely to exceed the accuracy of available load data.

4.2. Other Comments

Post-workshop comments on the issues discussed above echo general support for the workshop consensus recommendations. TURN additionally recommends that the Commission require the utilities to provide the data source and basis for the non-DEER energy and demand estimates, as well as establish a due date for the provision of all supplemental non-DEER information.

In its post-workshop comments, PG&E raises some concerns with respect to using the DEER definition of peak kW, at least for its own system. PG&E recommends that the peak kW definition be extended to 6 p.m. for the ex post studies of its 2006-2008 portfolio savings impacts, in order to reflect the fact that its system can peak after 6 p.m. More specifically, PG&E concludes based on further research that a summer peak period of 4 p.m. to 6 p.m. would better reflect current system level consumption patterns.

PG&E also notes that within DEER, the peak definition appears to differ both among weather-sensitive measures (i.e., for non-residential, school and residential), as well as between weather-sensitive and non-weather sensitive measures. PG&E recommends that the Commission clearly indicate which definition it is adopting, should it adopt a DEER definition.

More generally, PG&E urges the Commission to initiate the Phase 3 proceeding as soon as practicable to resolve the longer term issue of a common yardstick for defining and valuing the contribution of resource options to reducing system peak loads.

4.3. Final Report Recommendations

The 2006 Update consultants support the consensus recommendations reached during the workshop on the peak definition issues.

4.4. Discussion

We find the workshop consensus for the definition of energy efficiency peak kW reductions to be reasonable for use during the current 2006-2008 program cycle. It is a pragmatic approach to addressing load impact data limitations at this time, while taking advantage of a database that we have determined "should be the source of all assumptions that are used to estimate load impacts, to the extent possible."²⁴ Accordingly, until further notice of the Commission, we will use the 2005 DEER Update definition of peak kW for the purpose of verifying energy efficiency program and portfolio performance. In addition, until further notice, the utility program administrators are required to apply this definition to energy efficiency uses during the 2006-2008 program cycle, including any necessary portfolio rebalancing.

With respect to PG&E's concerns over the consistency of peak (kW) definitions used in the DEER database, we note PG&E quotes a sentence from the on-line DEER report that refers to the definition of peak as "the average demand savings between noon and 6:00 p.m. during the months from May through October." The ALJ has confirmed with the 2006 Update consultants that this is an isolated error in the DEER documentation, due to an oversight in updating the documentation from earlier versions (and definitions of peak), and will be corrected. The 2005 DEER Update peak demand definition for all weather sensitive DEER measures (residential and non-residential) is exactly as presented during the 2006 Update workshops in the Final Report (Attachment 3). There is no underlying inconsistency in the use of that definition within the most current version of DEER, for estimating the peak kW demand impacts for weather-sensitive measures.

PG&E also suggests that the sources of data used to develop the load impact values for non-weather sensitive measures (e.g., refrigerators, compact florescent bulbs) may not be consistent in terms of the DEER definition of peak kW used for weather-sensitive measures, and recommends that these inconsistencies be acknowledged by the Commission.²⁵ However, PG&E provides no documentation that, upon investigation of the actual data sources (provided in the DEER report sections), there are significant differences in the definition of the afternoon periods used to derive these values.

Moreover, DEER values for non-weather sensitive measures are all state-wide averages, to which the weather-sensitive definition cannot be directly applied. PG&E does not suggest that the use of state-wide average values is inappropriate for measures with load impacts that do not vary by weather/temperature or climate zone. Rather, PG&E points to two examples where the statewide values are developed from different data sources, again implying that this is an inconsistency to be reconciled-or at least noted by the Commission.²⁶ However, nowhere in its comments does PG&E evaluate whether the resulting DEER values (based on the source data it references) are inconsistent with the impacts of these measures on summer weekday loads (between 2 to 5 p.m.), based on a comparison with hourly impact data, or its own TOU load shape information (H-factors).

In sum, PG&E alleges inconsistencies with respect to DEER that do not appear to be based upon relevant analysis or documentation. Similarly, PG&E proposes an expanded peak period without documentation of the additional research to which it refers, and without the opportunity for interested parties to review the underlying data or to address the potential ramifications of this proposal. If PG&E believes that these issues warrant further evaluation, it may present them in the context of future DEER updates.

As discussed in the workshops, an appropriate long-term definition for energy efficiency peak kW impacts will need to be considered in the context of available load shape data for individual energy efficiency measures. We discuss an action plan for moving forward with the requisite load shape data collection and evaluation in Section 8 below.

The workshop discussion and comments also raise the issue of what estimates of peak kW the utilities should use for rebalancing their portfolios and reporting program accomplishments during the program cycle. The consensus recommendations on this issue are that the utilities should: (1) use DEER values for peak kW and kWh savings for those measures that are included in the DEER database and (2) continue to use their best estimates of those values for measures that are not currently included in DEER, or for programs with measure categories rather than specific measures, such as customized rebate programs. These recommendations are fully consistent with our policy rules for energy efficiency, and we will adopt them.²⁷

However, we further clarify that the utilities are required to update their ex ante estimates of kW and kWh savings for customized rebate programs as they proceed with implementation, based on site specific installations for these programs, just as they are required to do for the incremental measure costs.²⁸ In doing so, they must utilize DEER savings values for the installed measures, if that data is available in DEER. Until further notice, the utilities should present these updates of ex ante estimates to Joint Staff and the utilities' program advisory/peer review groups every six months, i.e., by June 15 and by December 15 of each year.

We also agree with workshop participants that there needs to be an ongoing exchange of information concerning the peak kW load reduction factors (ratio of kW to kWh savings) that the utilities use for portfolio rebalancing and reporting. We direct the utilities to provide the information necessary for Joint Staff and other program advisory/peer review group members to review the methodology and/or baseline load shape (measure or end use) estimates they are using to estimate peak kW reduction load factors. This information should indicate clearly where DEER and non-DEER values of kWh and peak kW impacts are used, and for the latter, present other sources of load factor data, such as the CEC load factors, as a basis for comparison.

In addition, as recommended by TURN in its comments, the utilities should include the data source and basis for the non-DEER energy and demand estimates. We direct the utilities to provide this information within 15 days from the effective date of this decision, and on an ongoing basis thereafter, as requested by Joint Staff or the utilities' program advisory or peer review groups during this program cycle. The utilities should post this information on a website and notify the following of its availability: (1) the 2006 Update service list in this proceeding and (2) the services lists in A.05-06-004 et al. and R.06-04-010. In addition, the utilities should jointly schedule a statewide meeting (or series of meetings) with their program advisory and peer review groups to present and discuss this information as soon as practicable.

The consensus recommendations recognize that, irrespective of the source (e.g., DEER), all ex ante estimates of energy efficiency load impacts are subject to ex post verification and true-up per our direction in D.05-04-051 and the adopted EM&V protocols in R.06-04-010 and its predecessor rulemaking, R.01-08-028. While D.05-04-051 allows for some exceptions to this requirement for specific measures, today's decision does not modify the Commission's general policy that load impact estimates are subject to ex post true-up in evaluating energy efficiency portfolio achievements.²⁹ Rather, today's decision provides an important clarification to this true-up process by defining the peak kW metric that will be verified in ex post studies for the 2006-2008 program cycle, namely, the DEER definition of peak demand.

²⁰ See D.04-09-060, pp. 101-114.

²¹ As discussed during the workshops, there are three different types of load factors that can be used as the basis for this conversion factor, e.g., a coincident load factor, a non-coincident load factor, or a load factor based on the historical relationship observed between kW and kWh savings from ratepayer-funded program activities.

²² "Coincident peak" generally refers to demand reductions from energy efficiency that occur at the time of the system peak, however that peak period is defined.

²³ See Final Report, Attachment 3: Definition of Demand (kW) Impacts Used in the 2005 DEER Update. As indicated in that Attachment, DEER also defines a "secondary" peak demand period for educational facilities and other buildings that tend to operate at greatly reduced use during the peak demand period defined above. For this purpose, DEER uses the next highest peak during a period in which the facility is operated in full use mode.

²⁴ D.05-04-051, mimeo., p. 25; Attachment 3, Rule IV.11.

²⁵ PG&E Comments, March 27, 2006, footnote 1, p. 3.

²⁶ Id.

²⁷ See D.05-04-041, Attachment 3, Rule IV.11, which states (in part): "To the extent possible, the assumptions that are used to estimate load impacts (e.g., kWh, kW and therm savings per unit, program net-to-gross ratios, incremental measure costs and useful lives) in the calculation of the TRC and PAC tests shall be taken from the Database for energy Efficiency Resources (DEER)."

²⁸ See Administrative Law Judge's Ruling on EM&V Protocols, R.01-08-028, September 2, 2005, p. 20.

²⁹ See D.05-04-051, mimeo., pp. 44-45.