14. SDG&E Service Reliability Incentives

14.1. Summary

This section of the decision addresses several tried and several new indices with incentives to improve or enhance the electric system reliability for SDG&E's customers. It is also the most jargon-intense section of the decision. We adopt updated targets and incentives for all existing incentives, agree to delete one existing incentive, and adopt two new indices with incentives.

SDG&E presently has three reliability incentives, which it proposes to continue, based on a System Average Interruption Duration Index (SAIDI), System Average Interruption Frequency Index (SAIFI), and Momentary Average Interruption Frequency Index (MAIFI) which have specific quantitative performance targets and reward/penalty mechanisms, last adopted in D.05-03-023.39 SDG&E argues the SAIDI and SAIFI results "put SDG&E in the top quartile of utilities using similar performance indicators." (Sempra Opening Brief, p. 418.) The two new incentives are the System Average Interruption Duration Index Exceeding Threshold (SAIDET) and the Estimated Restoration Time (ERT), which we discuss below.

As already noted, we believe incentives can improve the quality (or safety) of service beyond the base line level funded in the rate case. We will therefore consider the parties' recommendations and adopt a set of reliability incentives with the belief that we expect to see significant improvements in exchange for the offered rewards.

We note that CCUE agrees with SDG&E and that we should not have dead bands (CCUE Opening Brief pp. 17-19) believing the dead band "dampen" the effect of an incentive. As already noted, we disagree: dead bands eliminate the random difference in results immediately around the target and therefore we adopt dead bands for the reliability incentives consistent with the customer service incentives.

The following table summarizes the proposed and adopted incentives:

SDG&E Reliability Performance Incentives40

14.2. Excludable Major Events

SDG&E proposes to modify the method for outage event exclusion, when calculating the rates or incurrence of outages, starting in 2008 "to provide a more standardized approach to address events like storms, fires, etc."

The current definition of an excludable major event is:

Each utility will exclude from calculation of its reliability indices major events that meet either of the two following criteria: (a) the event is caused by earthquake, fire or storms of sufficient intensity to give rise to a state of emergency being declared by the government, or (b) any other disaster not in (a) that affects more than 15% of the system facilities or 10% of the utility's customers, whichever is less for each event. (D.96-09-045, Appendix A.)

SDG&E proposes to exclude events as defined by IEEE 1366-2003, which is a statistical method. Applying this new definition SDG&E believes "will help minimize variations in SAIDI and SAIFI values caused by factors beyond SDG&E's control." (Sempra Opening Brief, p. 420.)

DRA recommends that the Commission continue to use the present definition for excludable major events adopted in D.96-09-045, to preserve consistency in comparing SDG&E's future performance with its past performance.

CCUE argues that there is no need to change the definition of excludable events (CCUE Opening Brief, pp. 14-16) arguing persuasively that the IEEE-1366-2003 method is a complex calculation unlike the current Commission exclusion based on actual events.

We agree with DRA and CCUE: we believe we are better-served with a lifetime consistency of results (including rewards and penalties) and we are less concerned about comparison to other non-jurisdictional utilities. We will, therefore, continue with our existing definition for excludable major events for incentive purposes. SDG&E has not persuaded us that the current method is wrong, only that IEEE has a new standard. SDG&E (or others) may always present a translation of historical results in order to draw a comparison between SDG&E and others who may comply with IEEE 1366-2003, but we have no control over the accuracy or inevitable interpretive application of IEEE 1366-2003 by non-jurisdictional utilities. We may only hope to be consistent in our application of our own standard over time.

14.3. System Average Interruption Duration Index

SDG&E proposes performance targets for SAIDI based on a new methodology. The proposed performance target for 2008 uses a five-year historical average (including the proposed IEEE 1366-2003 and government event exclusions) plus a predicted increase in cable failures (Ex. SDG&E-4). The 77.0 minutes performance target is the rounded total of the annual average performance for the five most recent complete years of SDG&E (using the IEEE Standard 1366-2003 and the government event exclusions criteria), plus a predicted increase in SAIDI due to cable failures, less a stretch factor of 1.0 minute. This stretch factor of one minute is based upon the SAIDI stretch factor adopted in D.05-03-023. (Sempra Opening Brief, p. 425.)

DRA argues that SDG&E inappropriately added 9.45 minutes to the last-adopted SAIDI target of 68 minutes. While SDG&E states that the predicted increase is due to cable failures, DRA believes SDG&E did not provide any statistical basis for its methodology for predicting SAIDI increases resulting from cable failures, or provide any rationale for its incremental SAIDI per failure formula. The additional minutes appear to DRA to be an overestimation. DRA recommends a SAIDI target of 68 minutes based on the five-year average from 2002 - 2006, equal to 68.9 minutes, less a one-minute stretch factor.41 DRA also opposed adopting IEEE 1366-2003. (DRA Opening Brief, p. 535.) We agree we should continue with the established measures for excludable events.

CCUE also objects to SDG&E's cable failure rates included in the SAIDI calculation (CCUE Opening Brief, pp. 7-10) and states that failure rates have been consistent with customer growth so failures per customer have remained constant. CCUE also points out that DRA's estimate of a target does not allow for continued improvement and recommends annual targets with a 5% annual improvement:42

2008 - 68 minutes

2009 - 65 minutes

2010 - 61 minutes

2011 - 58 minutes

As explained above, we reject SDG&E's proposal to revise the definition of excludable events. Furthermore SDG&E has not justified its prediction of an increase in the duration of cable failures on a per customer basis. Therefore, we will adopt the 2008 DRA/CCUE target of 68 minutes and CCUE's 5% annual improvements requirement, but with a dead band. Consistent with our approach for annual targets and our preference for including dead bands, we will adopt the SDG&E balanced $250,000 incentive per increment of change with a maximum reward or penalty at $2,000,000.

14.4. System Average Interruption Duration Index Exceeding Threshold

SDG&E argues, based on customer studies, that it found customers are less likely to be satisfied when they experience more than two outages annually or if the outages last for more than two to three hours, and that they want to know a timeframe when their electric power will be restored. This matters because SDG&E has found for example in 2004, only 10% of SDG&E's customers experienced approximately 60% of the SAIDI minutes. (Sempra Opening Brief, p. 418.)

SDG&E has therefore developed and proposed a new "customer focused" performance indicator System Average Interruption Duration Index Exceeding Threshold (one hundred and fifty minutes or greater) (SAIDET), which is "SAIDI exceeding a threshold." Thus, SAIDET represents the SAIDI minutes experienced by customers for outage durations beyond an annual interruption minute threshold. SDG&E recommends implementation of SAIDET for three reasons:

1) SAIDET is a "customer focused" index rather than a system wide average index;

2) SAIDET will focus company resources on customers who are experiencing more frequent or longer outages, and

3) SAIDET is more likely to improve customer satisfaction.

SDG&E argues SAIDET, in conjunction with the continued use of SAIDI and SAIFI, will allow it to work on the overall goal of improved reliability, yet allow SDG&E to focus on customers who experience a disproportionate share of the outages.

DRA argues that "in terms of reliability improvements, there is no essential difference between SAIDI at the 150 minute threshold [i.e., SAIDET] and SAIDI. The two indicators are mathematically very similar and statistically very highly correlated." (Ex. DRA-24, p. 24-21.) DRA recommends that the Commission not adopt a SAIDET incentive.

CCUE supports the adoption of SAIDET (CCUE Opening Brief, p. 20.) because it focuses on those customers who experience the most outages. CCUE also suggests an initial incentive amount of $100,000 per increment for the reward or penalty, compared to either SDG&E's $250,000 and DRA's unbalanced incremental proposals.

We agree that the measures appear to be very close. We therefore propose to adopt both measures but essentially divide the potential reward/penalty between the two. SDG&E previously had a maximum reward of SAIDI of $3,750,000. It now proposes a SAIDI reward of $2,000,000 and $2,500,000 for SAIDET, for a total of $4,500,000 for two very similar mechanisms. We note with concern that SDG&E assigns the largest maximum reward to the new and untested SAIDET. We will therefore allocate a maximum total reward/penalty of $3,750,000 with a $2,000,000 limit to the existing SAIDI measure and $1,750,000 to the new SAIDET measure. SDG&E offered no persuasive argument to raise the combined total of the incentive and therefore we can gain experience with the new measure without increasing the cost to the ratepayers.

14.5. System Average Interruption Frequency Index

The SAIFI is the annual system average number of sustained outages a customer experiences, distinct from the separate measure of momentary interruptions (MAIFI). SDG&E proposes a SAIFI performance target of 0.65 outages per year; a 0.15 outage live band;43 no dead band, and a reward/penalty calculated at 0.01 outage increments at a rate of $250,000 per increment for a maximum allowable reward/penalty of $3,750,000 starting in 2008. SDG&E's performance target of 0.65 outages per year is the rounded total of the annual average performance for the five most recent complete years of SDG&E performance (using the proposed IEEE Standard 1366-2003) and the government event exclusions, plus a predicted increase in SAIFI due to cable failures, less a stretch factor of 0.0110 outages.44 (Sempra Opening Brief, p. 431.)

DRA states that the current SAIFI target is 0.67 outages, which was based on a five-year average (1999 - 2003) less a stretch factor of 0.01 outages, as adopted in D.05-03-023. DRA objects to SDG&E's use of the IEEE Standard 1366-2003, and argues SDG&E has not justified the increased cable failure factor. Consistent with its positions on other incentives, DRA proposes unbalanced reward/penalties and continued use of the dead bands. DRA calculates a target of 0.61 outages, based on the five-year (2002 - 2006) average. DRA argues it used the formula the Commission used in the prior proceeding to determine the SAIFI target. (DRA Opening Brief, p. 536.)

CCUE supports adoption of a SAIFI and proposes a 2008 target of 0.56 outages with annual improvements of 0.03 outages (thus, 2009 would be 0.56 - 0.03 = 0.53, etc.). CCUE would set a higher target incentive for SAIFI (and SAIDI) and would set the incremental SAIFI rate at $450,000, which is higher than SDG&E's proposed $250,000. (CCUE Opening Brief, p. 13.) Even if CCUE is right, that inflation and customer growth affect the incentive, CCUE has not shown that it would be reasonable to adopt a rate greater than SDG&E's proposal, and the company is prepared to accept the lower amount as a reasonable incentive. Parties may revisit this in greater depth in subsequent proceedings.

We will adopt a balanced reward/penalty, using our prior formula (not including the IEEE Standard 1366-2003) and we will adopt the DRA target of 0.61 outages with a 0.02 dead band because it reflects the more recent data and the dead band is consistent with our general intent. We will also adopt CCUE's annual improvement factor of 0.03 outages because we do not want static targets that reward subsequent years for prior year's results. We believe consistency in the mechanism is critical to our long-term evaluation of this mechanism.

14.6. Momentary Average Interruption Frequency Index

Although SDG&E believes that MAIFI has proven to be a useful performance indicator, SDGE proposes to discontinue MAIFI for several reasons. First, SDG&E believes that MAIFI will begin to increase as a result of future improvements in SAIDI, SAIDET and SAIFI. As Supervisory Control and Data Acquisition (SCADA) equipment proliferates, SDG&E believes its distribution system customers may see an increased number of shorter duration outages with a commensurate decrease in longer outages. Second, the vast majority of MAIFI events are linked to the overhead electrical system. With the gradual transition to an underground system, SDG&E believes there may be a subsequent decline in MAIFI events in the long term.

CCUE concur with ending MAIFI, and its elimination is consistent with DRA's antipathy to incentives generally proposing rewards smaller than penalties.

We find that SDG&E essentially expects that short duration interruptions may increase while concurrently there will be a notable decrease in the number of longer outages - thus, an incentive reducing momentary interruptions may conflict with the goal of reducing the durations of outages generally. SDG&E is proposing incentives geared towards a system where outages are shorter in total duration, but not necessarily fewer in number. The next new incentive mechanism (ERT) is clearly focused on quick fixes for outages, at least within the estimated restoration time.

We will not adopt a MAIFI in this proceeding because: a continuation of MAIFI may create a perverse incentive; there is a new incentive designed with similar goals as the MAIFI did; and we are reluctant to require an incentive where the company no longer endorses the mechanism.

14.7. Estimated Restoration Time

SDG&E proposes a second new reliability indicator, Estimated Restoration Time (ERT), with the target of providing 50% of affected customers with an estimated time of service restoration that is within one hour of the actual restoration time. DRA is concerned about the adequacy of the data to set a benchmark. (Ex. DRA-24, p. 24-22.) The difference in the proposed incentive is significant: SDG&E seeks a maximum of $2,000,000 and DRA recommends one-eighth of that, or $250,000. This is due to DRA's proposal for all incentives to be weighted by its equalization factor. (DRA Opening Brief, p. 538.)

We agree with DRA that we should be conservative with new mechanisms. With the new SAIDET, above, we limited its maximums for reward/penalty and preserved the SAIDI. We could consider either reducing the incentive incremental amount - SDG&E proposes $200,000 per 1% change in accuracy - or the overall range of the incentive. If we reduce the incremental reward value, we risk under-rewarding or penalizing SDG&E: the company describes its proposed incremental incentive amounts as sufficient or necessary to induce changed behavior.

CCUE opposes the adoption of ERT (CCUE Opening Brief, pp. 24-26) and the value of knowing when service will be restored when customers are mainly concerned about actual restoration. We think there is some value to a reliable estimate of restoration. CUEE also warns of the potential for SDG&E to game the incentive. We are concerned, therefore, that this mechanism rewards SDG&E for restoring service within the time it forecasts. Thus, by under-promising (exaggerating how long a repair may take) SDG&E directly affects the likelihood of beating an exaggerated estimate. We will ask DRA to examine in the next proceeding the care and precision with which SDG&E forecasts service restorations.

While we agree with the need for an adequate incremental incentive amount, the option of reducing the over-all range of the incentive mechanism would allow an appropriate reward (if we accept the $200,000/1% as appropriate) but protect ratepayers from a large exposure should the ERT improvements prove to be easily obtainable. As noted already, we expect adequate service as a part of the adopted revenue requirement; incentives must result in significant verifiable improvements. We will, therefore, adopt a limit of $1,000,000 for the penalty/reward, so that the eligible range is limited to a 5% change around a dead band.

Consistent with our intent to have an annual improvement to raise the bar, we will adjust the ERT by 1% annually, which is the incentive reward/penalty increment. We expect parties to address an appropriate stretch factor in the next GRC as a part of evaluating the continuation of ERT. These incentive mechanisms are discretionary and we should be conservative in exposing ratepayers to additional costs.

We will adopt the 3% dead band proposed by DRA for two reasons: we are still convinced that dead bands eliminate unnecessary rewards or penalties due to changes in the indicators caused by chance rather than corporate actions; and because there is no history to this incentive and its measurement that would provide complete confidence in the target. We also adopt a 1% annual improvement factor so the 2009 target, for example, is 51% (50% in 2008 + 1%). We believe that a beginning target of only 50% accuracy is a very low objective and raising the target by the 1% dead band is a reasonable adjustment that should be achievable by the company.

39 SAIDI is the system average sustained (five minutes or greater) outage duration a customer experiences annually; SAIFI is the system average number of sustained outages a customer experiences annually (frequency); and MAIFI is the average number of momentary (less than five minutes) outages a customer experiences annually.

40 Ex. DRA-24, Table 24-6.

41 We can use 2006 operational data for SAIDI, as proposed by DRA, because it is not subject to any adjustments comparable to the adjustments necessary for 2006 financial data to be comparable with the presentation of prior years' financial data.

42 The annual Improvement factor is similar to a "stretch factor," i.e., a requirement to improve. This is consistent with SDG&E's incentives adopted in D.05-03-023, although the incentives here have a small annual increase which makes all of the incentives, including the safety incentives, slightly harder to earn a reward annually. The record showed that SDG&E and SoCalGas were successful at earning rewards under the terms of D.05-03-023, where there were not annual improvements to modestly improve the targets in successive years but only an initial "stretch" to the base target.

43 A live band represents the range where a penalty or reward can be earned.

44 Stretch factor of 0.01 outages is based upon the SAIFI stretch factor applied in D.05-03-023.

Previous PageTop Of PageNext PageGo To First Page