8. Line Loss Methodology

8.1 Background

The term "line losses" refers to the power losses that occur when electricity is transmitted over power lines. PURPA established that, to the extent practicable: "the costs or savings resulting from variations in line losses from those that would have existed in the absence of purchases from a qualifying facility, if the purchasing electric utility generated an equivalent amount of energy itself or purchased an equivalent amount of electric energy or capacity" (18 CFR 292.304(e)(4)) should be incorporated into avoided cost payments.

D.82-12-120, D.84-03-092, and D.87-12-066 established the methodology for line losses for QF payments. Different line loss adjustment factors were established for different usage periods, such as peak, mid-peak, off-peak. Line loss factors greater than one indicate that QF production causes a reduction in utility system line losses, while line loss factors less than one indicate that QF production causes an increase in utility system line losses. For QFs connected to the grid at the transmission level, average transmission loss factors (TLFs) were set at 1.023 for Edison, 1.025 for SDG&E, and 1.000 for PG&E. For QFs connected at the primary distribution level, distribution loss factors (DLFs) were set at 1.026 for Edison, 1.06 for SDG&E, and 1.000 for PG&E.⁷

These loss factors were established on an interim basis, with the expectation that more definitive studies would lead to a more accurate line loss methodology. As the Commission stated, "[o]ur decision reflects the inconclusiveness of the record on line losses and our struggle to develop an appropriate interim solution until the line losses studies required of all three utilities are completed, reviewed, and approved." (D.84-03-092, p. 37.) The expected review and approval of these studies has never occurred, and all but one of the loss factors have been in place since.

Seeking to revise both the TLFs and the DLFs, SDG&E filed Application 98-06-045 and proposed to replace the existing TLF values with generation meter multipliers (GMMs).⁸ GMMs were developed and are used by the ISO to determine the impact on system line losses caused by generation from a particular generator. GMMs are calculated for each generator bus and each intertie⁹ every hour. The GMM's are first forecasted and published seven days in advance. An update "hour-ahead" GMM is also published. The hour-ahead GMM is also known as the ex post GMM. (See Workshop Report, Appendix C: ISO Presentation on GMMs, p. 4.) The ISO and PX use GMMs for system balancing and settlement purposes.

The Commission rejected SDG&E's GMM proposal, noting that:

"SDG&E has not demonstrated that these factors no longer reflect avoided line losses on its system, or that the generator meter multipliers of the Independent System Operator (ISO) are more appropriate to use for short-run avoided cost calculations." (D.99-03-021, p. 1.)

For the same application, SDG&E performed a new study of distribution-level QF line losses. Consequently, the Commission approved SDG&E's request to switch the DLF value to 1.00. Functioning differently from the old DLF, the new DLF of 1.00 is multiplied by the TLF in order to obtain the over-all line loss adjustment for distribution level QFs. To avoid constraining future regulatory activity, the decision also noted:

". . . nothing in this decision precludes any party from bringing up methodological proposals related to line losses, including those considered in this proceeding, in the PU Code Section 390 proceeding opened pursuant to D.99-02-085." (Ibid., p. 19.)

As directed in the rulemaking, Energy Division convened workshops and issued a workshop report addressing issues pertaining to line losses. Prior to the workshop, parties filed comments, addressing the topics set forth in the Scoping Memo. The workshop focused on developing an understanding of the existing treatment for line losses, proposed alternatives, and criteria to be used in choosing a methodology.

One of the goals of the workshop was to understand how the ISO calculates GMMs. An ISO representative presented the ISO methodology and answered questions from workshop participants. Each GMM is equal to one minus the scaled marginal loss factor. The scaled marginal loss factor is equal to the full marginal loss factor multiplied by a scaling factor. To obtain the full marginal loss factor, the ISO models an increment of power from a generator, and calculates the increase (or decrease) in system line losses that would occur if this increment of power were spread over the entire ISO grid proportionately to where the existing load is. The scaling factor (with a typical value of about 0.55) is the ratio of the system losses divided by the sum of the products, for each generator, of its full marginal loss factor times its generation level. Workshop participants discussed the validity of modeling generation as being spread throughout the grid, with no bias toward local consumption, as well as the validity of scaling of marginal loss factors. SDG&E's representative presented a February 2000 study of the effect on system line losses from the four SDG&E transmission level QFs.

Although the workshop furthered understanding of the GMM methodology, it did not produce a consensus for the treatment of line losses. The workshop report reflected this lack of consensus, cited areas that required further investigation, and made recommendations.

8.2 Parties' Positions

ORA, SCE, SDG&E, and PG&E favor use of GMMs to replace the current TLFs. As alternatives, SDG&E proposes adoption of the TLFs obtained from its recent line losses study or a TLF value of 1.00. PG&E also does not object to keeping its current TLF value of 1.00. These parties claim the following advantages for the GMM methodology:

1. GMMs have been developed and are calculated by the ISO, a neutral, knowledgeable party;

2. GMMs are specific to individual QFs, and consequently more accurate than any single number applied to all QFs;

3. GMMs vary by hour, and thus more accurately reflect the impact on line losses;

4. GMMs have been developed expressly to calculate the impact on system line losses due to power inputs from a given generator;

5. GMMs are being used by the market for purposes of calculating line losses; and

6. GMMs are readily available, and practical.

IEP, EWC, FPL and Caithness favor maintaining the status quo, citing the lack of a conclusive challenge to the existing methodology and pointing out weaknesses in all of the proposed alternatives. IEP claims that no party has successfully impugned the validity of the existing TLFs. IEP also argues that the proposed GMM method violates Commission Rule 74.3.¹⁰

Caithness objects to the use of GMMs, arguing that GMMs do not account for long-term resource decisions made in the 1980s that were responsible for determining the utilities' avoided costs today. Caithness also raises technical objections to the new SDG&E study, which calculates TLF values of approximately 1.005, significantly lower than the values currently in place. Caithness also argues that the Commission must consider the plight of remotely located alternative resources such as wind, solar, and geothermal who would likely be hit hard financially by the adoption of GMMs.¹¹ Caithness suggests that this result would be counter to California legislative policy, which is to encourage alternative generation.

CCC raises three main objections to the use of GMMs. First, CCC objects to how the ISO model spreads the incremental generation over the entire grid without giving preference to close-at-hand load, which they maintain would be a more realistic assumption. Second, CCC maintains that as a result of the ISO model's spreading the incremental generation over the grid, certain remotely located generators serving local load will be treated inaccurately and unfairly. Third, CCC argues that by forming GMMs from scaled marginal loss factors, instead of from full, unscaled marginal loss factors, the GMMs dilute the effect that a given generator has on the system line losses.

CCC developed a two-part proposal--one for QFs in general, and the other for remote QFs serving local loads. CCC's direct testimony derives a general loss factor of GMM_qf + d * (GMM_qf - GMM_sys) where: d is the inverse of the scaling factor that the ISO now uses for calculating GMMs, GMM_qf is the GMM value for the individual QF, and GMM_sys is the system average GMM. For remotely located generators serving local load, CCC derives a loss factor of d - GMM_qf * (d - 1).

Although the workshop report concluded that there was a need for more information regarding DLFs, parties declined to elaborate in their testimony and briefs. SCE proposes that the product of its Wholesale Distribution Access Tariff (WDAT)¹² and the appropriate GMM be the DLF. SDG&E proposes no change to its DLF of 1.0, which equals its WDAT. PG&E uses a DLF of 1.00 for its QF payments and proposes no changes, but uses different multipliers in its Wholesale Distribution Tariff.¹³ Other parties have been largely silent regarding DLFs, although Caithness believes that WDAT-based DLFs should be stand-alone numbers, and should not be multiplied by any other factors (such as GMMs). (Opening Brief, p. 19.)

8.3 Discussion

We begin our discussion by reviewing whether the existing methodology for addressing line losses for transmission level QFs is acceptable. The evidence indicates that it is not:

1. No party presented an explanation for the discrepancy between PG&E's TLF (1.000) and the TLFs in place for SDG&E (1.025) and for SCE (1.023);

2. The recent SDG&E TLF study suggests that the existing TLFs in place for SDG&E are much too high, leading to significant ratepayer losses;

3. D.99-03-021 explains that SDG&E's and SCE's current TLFs were based on a study that "assumed that all of the marginal line losses would be avoided by the operation of the QFs" (p.8), a difficult assumption to justify; and

4. Existing TLFs treat QF line losses in the aggregate, leading to a less fair and efficient outcome.

We conclude that replacing the existing TLFs with a simple factor of 1.000, unless there is a better methodology available, would be preferred to the existing factors. With the advantages noted above, GMMs appear to provide a superior methodology. First we examine the various arguments against GMMs more fully.

Caithness claims that the GMM does not address the long-term perspective. In order to perform the analysis proposed by Caithness, the Commission would need to speculate as to the resource procurement choices that would have been made in the 1980s, were it not for the QFs. This approach is unnecessary, as the application of line loss factors is for purposes of paying SRAC payments which clearly calls for a short-run perspective. Although we desire to promote renewable resource development, which often occurs in remote locations, there is no requirement under PURPA, or under California law, that alternative resource QFs receive special treatment for line losses.

CCC argues that the way in which the ISO model spreads incremental load over the entire grid without giving extra weight to nearby load is unrealistic. This criticism has merit. However, all models that allocate the incremental, or marginal, impact among various agents require approximating assumptions. A hypothetical raised at hearing demonstrates the problem. In the hypothetical, two generators are remotely located, and serve local load that is unable to consume all of the power from these generators. CCC witness Beach conceded that there a number of valid ways to allocate the system line losses impact in this example. (RT 862:7- 865:17.) There does not appear to be a unique, correct solution. The GMM methodology is one of the reasonable ways to allocate system losses.

A remotely located generator serving local load presents equity concerns regarding application of the GMM methodology. However, during the proceeding, no remote QF solely serving local load was identified. As discussed below, we are not convinced that the alternative approach proposed by CCC, which calls for a different formula to be applied to remote QFs serving local loads, is correct. Furthermore, the CCC proposal raises significant implementation difficulties.

Regarding scaling of marginal loss factors, it has not been demonstrated that "scaled" GMMs are wrong, or that "un-scaling" the GMMs is the right approach. In the ISO's Report to the Federal Energy Regulatory Commission: Studies Conducted Pursuant to the October 30, 1997 Order (December 1, 1999, p. 2), it states that scaling is necessary to avoid overpayments for line losses.¹⁴ Scaling is an integral part of the GMM methodology.

We will not adopt the model proposed by CCC. CCC's proposed differential line loss treatment for remote QFs and for QFs close to the load center appears tailor-made to maximize QF SRAC payments. Furthermore, the model contains numerous assumptions with which we are not comfortable. Some of these assumptions are:

1. Output from a given QF is treated separately from other generators (CCC Ex. 17, p. A-1);

2. The marginal loss rate for the QF is assumed to be constant (Id.);

3. The Total Avoided Costs equation incorporates the very GMM-based energy payments CCC is attempting to replace (Id., pp. A-3, B-1); and

4. CCC uses the same expression for marginal losses (ML_qf) for remote QFs as was developed for the standard QF scenario (Id., p. B-1.).

SDG&E is currently contesting the GMM scaling of marginal loss factors before FERC. Despite the limitations it finds with the current GMM methodology, SDG&E supports the FERC-adopted GMM methodology as the best choice to account for line losses for QF payment purposes. We expect that the GMM methodology may be revisited and refined from time to time by the FERC, and we welcome this process. Proposals to modify the GMM methodology itself should be directed to FERC.

We accept that the GMM is the best method available for measuring the impact on system line losses from an individual generator, but this is not exactly what PURPA calls for. PURPA calls for an adjustment to SRAC payments that will reflect the impact on system line losses as compared to the impact that would have occurred had the utility procured its power elsewhere.

For the case where the SRAC is PX-based, the treatment of line losses is simple. The PX procures power, ascribes line losses to each generator using GMMs, and passes these costs along to buyers in the market. Because each generator bidding into the PX market adjusts its bid to account for the GMM the PX will apply to the sale, the PX market price will reflect this collective bidding behavior. The resulting PX price reflects GMMs of all generators, thus, the clearing price reflects the system average GMM. That is, the PX clearing price reflects the cost of production as well as the cost of line losses. The PX then pays the generator the PX price times the generator's GMM. This is exactly the cost the PX avoids by purchasing from that generator. The line loss effect is captured entirely by the GMM when the SRAC is PX-based.

Unlike the PX price, the administratively determined SRAC, reflects only the cost of production. The simple GMM, when applied to the current administratively determined SRAC, fails to compare the individual QF's line losses to the line losses that would have occurred had the utility procured its power elsewhere. Under PURPA, the impact on system line losses due to generation by the individual QF must be directly compared to the system average GMM, which represents the impact on system line losses due to all of the other generation. This principle was demonstrated during cross-examination of SCE witness Mayfield.

Q: . . . You state, The generator's hourly GMM will be higher relative to the average GMM when the energy it delivers to is [sic] ISO grid decreases average transmission losses and lower than the average when the energy it delivers increases transmission losses. Now, as I understand it, this would mean that when a QFs GMM is higher than the ISO average GMM, the QF is providing line loss savings to the utility; is that right?

A: Yes.

Q: And under PURPA, the QF should be compensated for those savings, correct?

A: That's my understanding. (RT 1008:15-27.)

Therefore, if a QF has a GMM of 0.99 when the system average GMM is 0.98, the QF should receive a one percent credit for the line losses that its production helps the utility avoid. In other words, its TLF should be approximately 1.01, the QF's GMM divided by the system average GMM. This is the same proposal made by Energy Division's Workshop Facilitator James Loewen during the Line Losses Workshop and described in the Workshop Report. (P. 25.) In equation form, the new TLF equals GMM_QF / GMM_SYS. For simplicity of implementation, the simple average of all GMMs can be used to calculate GMM_SYS. Since actual ("ex post") GMMs are already listed on the ISO web site, implementing this approach will be simple and will not require any change in ISO procedures.

We will adopt GMMs as the TLFs once the Commission has made the required findings under Section 390(c) and QFs are paid a PX-based energy price. Until that time, effective with the first posting following this decision, we adopt a TLF equal to GMM_QF/GMM_SYS.¹⁵ QFs who have elected to switch to a PX-based SRAC, pursuant to D.99-11-025, should have their GMM applied to account for line losses, effective immediately.

Regarding DLFs, should we choose to rely on the utilities' WDAT factors, we face two concerns:

1. Disparity among the utilities' WDAT factors for distribution level generators;¹⁶ and

2. Lack of clarity as to whether the WDAT should be multiplied by the TLF to arrive at the correct total loss adjustment factor.¹⁷

The record provides no information as to why the factors vary so significantly between utilities or whether non-QF generators connected at the distribution level are compensated based on the GMM multiplied by the WDAT, or only on the WDAT.

Currently, the total loss factor for distribution-level QFs on PG&E's system is 1.000; PG&E's TLF is also 1.000. On SDG&E's system, the DLF is currently 1.000, and it is multiplied by the TLF to establish the total loss factor for payments to distribution-level QFs. SDG&E's DLF is the only DLF that has been updated based on a recent study and equals the WDAT. (See D.99-03-021.) SCE proposes to multiply its WDAT by the TLF to arrive at the DLF. We adopt the WDAT of SDG&E and SCE as the DLF, to be multiplied by the TLF, to arrive at the total loss factor for distribution-level QFs. This change should be effective the first posting after the effective date of this decision. Because we cannot explain the difference in the WDAT of PG&E, we retain the existing DLF of 1.000 for PG&E, to be multiplied by the TLF, to arrive at a total loss factor.