5. Undervaluation from TOU Averaging

In calculating resource benefits, the E3 calculator maps 8760 hours of avoided cost values to available load impact data for energy efficiency measures. TOU-based load shapes only present information at the aggregate TOU period level.³⁰ PG&E and SCE use five TOU periods to represent the year, and SDG&E uses six. In contrast, hourly load shapes have 8760 values. For measures with hourly load data, the mapping of avoided costs is one-to-one and the resource benefit calculations are performed on an hourly basis. For measures that depend on TOU-based load shapes, the hourly avoided costs are averaged over each TOU period.

Parties agree that this averaging process could undervalue measures that produce relatively more load reduction during the highest cost hours. To estimate this potential undervaluation, E3 presents in the Draft Report a comparison between the avoided costs obtained when hourly load shape versus TOU-averaged load shape data is used for office cooling, office lighting, residential air-conditioning (a/c) and residential refrigeration energy efficiency measures. The comparison was developed from hourly load shape data available for these measures within PG&E's service territory, in two different climate zones. E3 presented the comparison for four different TOU period definitions (i.e., May-Oct. noon to 6 p.m., June to Sept. noon to 6 p.m., May-Oct. 2-5 p.m., and July-Sept. 2-5 p.m.).

This analysis shows that residential a/c is the most undervalued end-use when using TOU period averaging, ranging from 5.7% to 12.5% undervaluation in climate zone 13 (Fresno, Bakersfield), depending on the TOU definition. The narrower summer on peak period definition (July-Sept. 2-5 p.m.) produces the lower end of the range. For climate zone 3 (San Francisco Bay Area), the analysis showed a similar but smaller effect, i.e., an undervaluation of residential air conditioning that ranged from 5.0% to 8.8%, depending on the TOU period definition.

In order to further develop their recommendations, workshop participants requested to see the magnitude of the undervaluation problem for all of PG&E's end uses that have hourly load shapes, as well as the undervaluation inherent in the use of TOU averages for residential and small commercial air conditioning hardware measures based on utility-specific DEER hourly load shapes. E3 presented this information in its Final Report, which is reproduced in summary form in Attachment 2. The magnitude of the undervaluation is expressed as a ratio of the average avoided costs calculated using hourly loads and avoided costs, divided by the average avoided costs using TOU averaging. The ratio represents a multiplier factor that would be applied to the TOU average value to correct for the undervaluation.³¹ We refer to this multiplier factor as a "correction factor" in our discussion below.

For residential a/c upgrades, the 2006 Update consultants developed correction factors for representative units with SEER 14, SEER 15 and SEER 16 ratings, and presented factors for each SEER rating, by utility and climate zone.³² These results were then weighted by the estimated percentage of installations for each SEER rating to produce an average residential a/c correction factor by utility and climate zone. To present a single correction factor for each utility, the climate zone-specific correction factors were then weighted by the expected distribution of units across each zone.

A similar process was used to derive the commercial sector correction factors utilizing DEER data. More specifically, the 2006 Update consultants produced correction factors for the most representative unit upgrade for the commercial sector, and presented the results for two building types (small office and retail) that were selected to bound the results across all sub-sectors.³³

In calculating the utility average correction factors, the 2006 Update consultants assumed an equal distribution of installed units across climate zones, i.e., weighted each climate zone-specific factor by 1.00. In doing so, the consultants stated that the weighting should be based on the expected number of unit installations in each climate zone relative to the total expected number of installations, but that they did not have the requisite information to perform these calculations for the Final Report. By ruling dated April 3, 2006, the assigned ALJ requested that the 2006 Update consultants prepare a supplement to the Final Report that would present correction factors weighted in the manner recommended, and to consult with the utilities in obtaining the requisite information. The ALJ directed that the supplement include a description and source(s) of data used to develop the weights. Parties were given an opportunity to comment on the reasonableness of the new weights used to recalculate the correction factors.

The analysis showed that the correction factor differs for each utility, reflecting differences in impact shapes and differences in TOU period definitions, as well as other factors. It is noteworthy that SDG&E's correction factor is significantly higher than the correction factor for SCE. This is largely due to SDG&E having a summer peak period that contains many more hours than SCE's. The more hours in the summer peak TOU period, the more averaging that occurs and the larger the averaging undervaluation.

5.1. Workshop Consensus and Non-consensus

There was consensus at the workshop that residential a/c measures that use TOU shapes should receive an adder to correct for the TOU undervaluation. There was no consensus as to the level of that correction, and whether the correction should vary by climate zone and/or utility. In addition, there was no consensus as to whether other customer sectors and measures that use TOU shapes should also receive a correction adder.

5.2. Other Comments

As summarized below, post-workshop comments clarified parties' positions on a correction factor to address TOU averaging.

Based on its review of the DEER correction factors, SDG&E recommends a 24% upward adjustment for residential a/c and a range of 10% to 20% for commercial a/c measures to account for TOU averaging. SDG&E would apply the lower end of this range to businesses that operate primarily on weekdays (e.g., office buildings), and the higher end of the range to businesses operating 7 days a week (e.g., retail).

For residential a/c measures, PG&E recommends using utility and climate-zone specific factors to adjust the TOU-averaged avoided costs. PG&E argues that the correction factors for its service territory should be based on PG&E's hourly load shape data, rather than the DEER load shape data. In PG&E's view, this is reasonable because its own hourly load shape data is based on metered data, rather than building simulation data. PG&E does not support adopting a correction factor for small commercial a/c at this time. PG&E contends that there is currently insufficient data to determine the appropriate TOU correction factor for this sector, and that applying one to this sector could be difficult because of definitional differences among the utilities. The correction factors that PG&E would use to adjust the avoided costs for its residential a/c measures range from 1.12 (zone 4) to 1.20 (zones 2 and 16).³⁴

Initially, TURN and DRA recommended using climate-zone specific correction factors developed from the DEER hourly load shapes to calculate the avoided costs associated with residential and small commercial a/c units. However, noting that the correction factors in the Final Report were based on an equal distribution of measures across climate zones, TURN and DRA recommended that the distribution instead be based on the expected number of energy efficiency measures to be installed in each climate zone. In its written comments, SCE supports this approach for residential a/c correction factors, but is silent on the issue of whether correction factors should be adopted for commercial a/c measures, and if so, how.

Based on the additional information provided in the Supplement, TURN and DRA modified their recommendations. Instead of using climate-zone specific correction factors, they recommend that utility territory-wide factors be used for residential a/c energy efficiency valuation, using the weights presented in the Supplement. These are: 1.171 for PG&E, 1.202 for SCE and 1.276 for SDG&E.³⁵

For commercial a/c installations, TURN and DRA recommend that the utilities apply the averaged correction factors weighted by climate-zone to retail building types. However, TURN and DRA would not use the weights presented in the Supplement by SDG&E for its program. In their view, the assumption implicit in that factor is unreasonable, namely, that all non-residential small packaged a/c unit efficiency improvements in SDG&E's service territory will be in retail buildings. They suggest that SDG&E revise this assumption to be more realistic. While not adverse to correction factors for office buildings, TURN and DRA believes that the need for correction factors for this sub-sector is less obvious.

No measures other than residential or commercial a/c equipment upgrades were proposed for the TOU-averaging correction adjustment or evaluated for such an adjustment in the Final Report and Supplement.

5.3. Final Report Recommendation

For residential a/c measures using TOU-based load shapes, the 2006 Update consultants recommend applying the territory-wide weighted average correction factor.

With respect to commercial a/c measures, the Final Report points out that the E3 calculator does not currently differentiate avoided costs by commercial sub-sector (i.e., office building or retail), and similarly, the data entered into the calculator does not make this distinction. As a result, additional data input as well as modification to the E3 calculator would be required to apply different correction factors at the sub-sector level.

In view of the above, and to be conservative on the level of adjustment, the 2006 Update consultants suggest that the territory-wide average correction factors for the office building sub-sector be applied to all commercial a/c installations that utilize TOU-based load shapes.

5.4. Discussion

We recognize, as do the parties, that the averaging that occurs through the use of TOU-based load shapes will both undervalue avoided costs during some of the highest load (peak) hours and at the same time overvalue avoided costs during some of the lowest load (off-peak) hours. Therefore, there are inaccuracies produced by this averaging process that could be addressed through the use of correction factors for each measure and end-use, in varying degrees. Nonetheless, the record in this proceeding supports the workshop consensus that TOU-averaging significantly undervalues measures that produce relatively more load reduction during the highest cost hours, such as residential and small commercial a/c equipment upgrades. In addition, the Final Report and Supplement provides a reasonable basis for adopting correction factors to adjust the avoided cost valuation of these particular measures. As discussed in Section 8, improvements to load shape data over the next 18 months should diminish the need to make such adjustments during the next program cycle.

We agree with TURN and others that the correction factors should be based on the DEER data. As TURN notes in its comments, PG&E's load shape data has not be subjected to the same level of public vetting and data quality control as have the DEER hourly impact shapes. Moreover, the workshop discussion clearly revealed that PG&E itself has elected not to use its own hourly load shapes at any point heretofore in the 2006-2008 energy efficiency planning and design process in any significant manner. Instead, as the 2006 Update consultants documented during that workshop, PG&E used TOU blocks or shapes in its application for approval of its 2006-2008 energy efficiency programs and budgets that are in closer agreement with DEER hourly shapes converted to TOU shapes than the TOU shapes created from PG&E's own hourly load shapes.³⁶

As that discussion also revealed, the PG&E hourly load shapes are building end use shapes rather than measure impact shapes, and many of them are from relatively old (late 1980's and early 1990's) data collection exercises using relatively small sample sizes. While it is true that DEER load shapes are based on simulations, the documentation of DEER indicates that those simulations utilize field data that is more recent, more extensive, and more representative of climate and vintage variations than the PG&E hourly load shapes.³⁷ G&E has actively participated in every DEER update, including the most recent 2005 updating process. If PG&E truly believed that its building end use shapes were more characteristic of the hourly load impacts of energy efficiency measures than the DEER measure impact shapes, it would have (1) argued this issue during the 2005 DEER Update process and/or (2) proposed to use its building end use shapes during the 2006-2008 planning process in a significant manner for further consideration by the Commission. PG&E did neither.

In fact, an examination of PG&E's average hourly load shapes for office and retail indoor lighting reveal that these shapes show approximately 18-20% and 21-24% respectively, of all electric lighting power being consumed during the summer off-peak period, which does not seem to be representative of current office and retail building operation. PG&E's hourly load profiles for office and retail cooling indicate a large use of retail a/c during the night in mild climates compared to hot climates, with almost equal cooling electric use during the summer on-peak and off-peak periods for these building types. Again, these patterns do not appear representative of the load impacts associated with a/c energy efficiency measures, which may be why PG&E did not use these load shapes in developing its 2006-2008 program plans.³⁸

In view of the above, we find PG&E's position in this proceeding on the issue of what hourly load shape data to use as the basis of the correction factors to be unpersuasive. We do not adopt it. Instead, we will adopt correction factors based on the DEER data presented in this proceeding, as described below.

5.4.1. Correction Factors for Residential A/C Units

We adopt the correction factors recommended by TURN and DRA for the installation of efficient residential a/c units. As noted in their joint comments, the data reflect that the climate-zone specific weighting results in higher correction factors than those where an equal distribution of measures across zones is assumed. This is logical and reasonable because the small packaged a/c unit energy efficiency savings are likely to be higher in the hotter climate zones, where efficiency improvements result in higher energy savings because of a/c usage patterns. Moreover, we agree with their observation that the additional precision gained from individual climate-zone correction factors does not justify the complexity and possible confusion resulting from having eight or nine different climate-zone correction factors. Accordingly, the following utility territory-wide correction factors will be applied to the avoided cost valuation using TOU shapes for residential small, packaged a/c unit energy savings: PG&E: 1.171; SCE: 1.202 and SDG&E: 1.276.³⁹

These correction factors should be applied to all residential a/c unit installations. The correction factors should be applied to the total avoided cost valuation for the installations, excluding transmission and distribution avoided costs. If the utilities do not currently identify residential a/c unit installations in the E3 calculator and the associated peak savings (or in other formats where projected savings are presented), they will need to develop a consistent and joint approach for doing so. This may entail estimating the fraction/percentage of installations for cooling end-use measures that represent the a/c unit hardware upgrades, and applying the correction factor to that fraction, or some other approach that is reasonable, consistent across utilities and practicable. We discuss in Section 11 the process for reviewing these and other updates to the E3 calculator and inputs in response to today's direction.

5.4.2. Correction Factors for Commercial A/C Units

The analysis presented in the Final Report and Supplement is based on data associated with two building types (small office and retail) within the commercial sector, which were selected to present a reasonable range of the potential undervaluation associated with TOU-averaging for small packaged commercial a/c units across all building types. That data from Attachment 2 is summarized below:

The weighting options result in very little differences to the conversion factors calculated for PG&E and SCE. As TURN and DRA point out, SDG&E has projected that all of its installations of small a/c commercial units will occur in the retail sector. Therefore, the climate-zone specific weighting for SDG&E will produce conversion factors of zero for commercial sub-sectors other than retail.

Irrespective of which weighting approach is used, the DEER data indicates that the potential undervaluation associated with TOU-averaging in the commercial sector ranges from approximately 5% to 12% for PG&E, from 6% to 14% for SCE and from 10% to 19% for SDG&E, based on the building types used to bracket this analysis. Except for PG&E's general contention of insufficient data and TURN/DRA's concern over SDG&E's estimation of where the a/c units will be installed, no party raises objections to the methodology or data employed by the 2006 Update consultants to develop these ranges for the purpose of calculating a correction factor for commercial a/c units. In particular, no party asserts that bounding the range of correction factors utilizing data from the retail and office sub-sectors is unreasonable.⁴⁰ Given the inherent uncertainty in predicting the potential impact of TOU-averaging, we believe that the 2006 Update consultants have utilized the best available data and reasonable methodology for developing the potential correction factors for this sector.

However, as pointed out by the 2006 Update consultants, there are implementation challenges associated with adopting sub-sector (i.e., building type) specific correction factors based on this data. In particular, it appears that over 98% of PG&E's commercial a/c measures are currently entered into the E3 calculators with a generic "commercial" label. Applying sector specific correction factors to SDG&E and SCE's E3 calculator inputs would be more straightforward, given the manner in which that data is currently disaggregated. Still, adopting sector-specific correction factors would require additional data transfer for all the utilities.

We therefore question the value of approaching the commercial a/c correction factor on a sector-specific basis, in view of these additional implementation complexities. Moreover, the record indicates that such an approach is not likely to produce significantly improved accuracy. In particular, the expected distribution of small packaged commercial a/c units across sub-sectors appears to be less certain for commercial applications, than for residential applications, especially in light of TURN and DRA's comments concerning SDG&E's projections for its service territory.

In view of the above, we believe it is reasonable to adopt correction factors for all commercial sector installations of packaged small a/c units based on a simple average of the low (office) and high (retail) end of the range presented in the Final Report. Accordingly, we adopt the following utility territory-wide correction factors: PG&E-1.085; SCE-1.105 and SDG&E-1.145.

These correction factors should be applied to all packaged small commercial a/c unit installations in the commercial sector. The correction factors should be applied to the total avoided cost valuation for the installations, excluding transmission and distribution avoided costs. As discussed above, if the utilities do not currently identify these types of installations in the E3 calculator and the associated peak savings (or in other formats where projected savings are presented), they will need to develop a consistent and joint approach for doing so. This may entail estimating the fraction/percentage of installations for cooling end-use measures that represent the small packaged commercial a/c unit hardware upgrades, and applying the correction factor to that fraction, or some other approach that is reasonable, consistent across utilities and practicable.

³⁰ More specifically, for energy, the TOU-based load shape indicates the share of the load shape's total annual energy consumption that occurs in each of the five or six TOU periods (e.g: Summer On-Peak = 20%, Summer Partial Peak = 25%, summer Off Peak = 5%, Winter Partial Peak = 35%, Winter Off-Peak = 15%, and the sum of the shares for all periods sums to 100%). For demand, the TOU values indicate the relative magnitude of the peak demand that occurs in each TOU period. (e.g., Summer Partial Peak = 90%).

³¹ For example, if the ratio is 1.15 and the TOU avoided cost is $100, one would multiply $100 by 1.15 (or increase $100 by 15%) to correct the TOU-averaged avoided cost to the hourly equivalent ($115).

³² "SEER" stands for Seasonal Energy Efficiency Ratio, which is calculated by dividing the amount of cooling supplied by an air conditioner or heat pump (btus per hour) by the power (watts) used by the cooling equipment under a specific set of seasonal conditions. The higher the SEER rating, the more efficient the unit.

³³ The level sub-sector disaggregation varies for each utility, however, in addition to retail and small office, the commercial sub-sectors include categories such as lodging, health care, colleges/universities, warehouses, K-12 schools, restaurants, grocery, assembly, and other commercial.

³⁴ See Attachment 2, Table 1 from the 3/21/06 Final Report.

³⁵ See Attachment 2, Table 2 from the 4/11/06 Supplement.

³⁶ At the March workshop, the 2006 Update consultants presented materials contained within a workbook and a PowerPoint document that presented this comparison. The workbook used in making that presentation is contained in a ZIP archive located at http://www.doe2.com/download/AvoidedCost/Compare2006AvoidedCostCalcs_2006-03-10.zip with the document that explains its contents, use and the data and methods used to create the workbook found in the PDF document located at: http://www.doe2.com/download/AvoidedCost/Compare2006AvoidedCostCalcs-Description_2006-03-10.pdf and the PowerPoint presentation can be found at http://www.doe2.com/download/AvoidedCost/AComparisonOfMeasureAvoidedCostCalculationsv4.ppt.

³⁷ See DEER documentation at www.energy.ca.gov/deer.

³⁸ To review these load shape data, open the PG&EComViewer.xls to the "Viewer" tab and select the desired climate zone, building type and "COOL" end use. The archive containing the IOU load shape data, as used in the E3 calculators, can be found at http://www.doe2.com/download/AvoidedCost/PGE-SCE-SDGE_LoadshapeViewers.zip.

³⁹ See Attachment 2, Table 2 from 4/11/06 Supplement.

⁴⁰ In fact, selecting these two building types to bound the results is consistent with PG&E's sub-sector data as well. See Tables 6 and 7 of the Final Report, on p. 11.