Phase 2 of 2009 General Rate Case Marginal Cost And Sales Forecast Proposals

Transcription

1 Application No.: Exhibit No.: Witnesses: A SCE-02 Cyrus Sorooshian Steve Verdon Paul Nelson Carl Silsbee (U 338-E) Phase 2 of 2009 General Rate Case Marginal Cost And Sales Forecast Proposals Before the Public Utilities Commission of the State of California Rosemead, California March 4, 2008

2 SCE-02 Marginal Cost And Sales Forecast Proposals Table Of Contents Section Page Witness I. MARGINAL COSTS...1 P. Nelson A. Introduction And Summary...1 B. Methodology Overview Marginal Cost Principles Marginal Cost Scope And Application Cost Drivers...6 a) Electricity Usage Cost Driver...6 b) Design Demand Cost Driver...7 c) Customer Cost Driver Time-Of-Use (TOU) Issues...11 a) Generation Marginal Costs...11 b) Delivery-Related Marginal Costs Annual Cost Of Capital Investments (RECC vs. NCO)...12 C. Marginal Cost Methodology Electricity Usage Marginal Costs...18 C. Silsbee a) Generation Capacity Marginal Cost...19 b) Energy Marginal Cost...22 c) Loss Of Load Expectation Delivery-Related Marginal Costs...27 P. Nelson a) NERA Regression Methodology Customer Marginal Costs Street Lighting and Outdoor Lighting Marginal Cost i-

3 SCE-02 Marginal Cost And Sales Forecast Proposals Table Of Contents (cont d) Section Page Witness II. SALES AND CUSTOMER FORECAST...39 S. Verdon 1. Billing Determinants And Present Rate Revenue...40 Appendix A Glossary... Appendix B Circuit Analysis For Determination Of Effective Demand... Appendix C Marginal Energy Cost Analysis... Appendix D SCE Costing Period Study... Appendix E NCO Marginal Cost Methodology... P. Nelson C. Sorooshian C. Silsbee P. Nelson P. Nelson -ii-

4 SCE-02 Marginal Cost And Sales Forecast Proposals List Of Figures Figure Page Figure I-1 Illustration of the RECC Methodology...14 Figure I-2 Annual Payment for $100 Capital Investment 10% Discount Rate and 3% Annual Inflation...15 Figure I-3 Derivation of Capacity Value CT Proxy with Energy Rent Adjustment...20 iii

5 SCE-02 Marginal Cost And Sales Forecast Proposals List Of Tables Table Page Table I-1 Electricity Usage -Related Marginal Costs (2009$, at generation level)...3 Table I-2 Delivery-Related Marginal Costs (2009$, at applicable voltage level)...3 Table I-3 Marginal Customer Costs (In $/Customer, 2009$)...4 Table I-4 Generation Marginal Costs, Average (2009$)...18 Table I-5 CT Proxy Cost (2009$)...19 Table I-6 Generation Capacity Marginal Costs, Average (2009$)...22 Table I-7 Energy Marginal Costs, Average (2009 /kwh)...24 Table I-8 Relative LOLE Factors (Sum = 1)...27 Table I-9 Delivery-Related Marginal Costs (2009$)...28 Table I-10 Customer Marginal Cost Components For GS-1 Customers (In $/Customer- Year, 2009$)...30 Table I-11 Customer Marginal Costs (In $/Customer-Year, 2009)...31 Table I-12 Monthly Street Light Facility Marginal Costs (2009$)...35 Table I-13 Monthly Street Light Facility Marginal Costs (2009$) Continued...36 Table I-14 Monthly Street Light Facility Marginal Costs (2009$) Continued...37 Table I-15 Monthly Street Light Facility Marginal Costs (2009$) Continued...38 Table II-16 Forecast of Grid Sales and Customers For Years 2009 Through iv

6 1 2 I. MARGINAL COSTS A. Introduction And Summary For over twenty years, the Commission has relied on marginal cost principles for assigning revenue requirements to customers (by rate group), and as guidance for setting the level of individual rate components. 1 This chapter presents SCE s marginal costs for providing regulated utility services to our customers. 2 The starting point for calculating marginal costs is the identification of cost drivers, that is, those fundamental aspects of customer electricity requirements that directly cause SCE to incur costs. Next, marginal costs are calculated for small changes in each cost driver, by dividing the change in total cost by the change in the cost driver. Finally, these marginal costs are attributed to measurable aspects of customer requirements such as energy consumption, peak demand, and customer type. This allows the rate components most associated with these measurable customer requirements, specifically energy charges, demand charges and monthly customer charges, to be set based on the corresponding marginal cost components. Marginal costs are used to calculate marginal cost revenues that is, the revenues that SCE would collect if all of its customers were charged rates that equal marginal costs. Marginal cost revenues are then used to allocate the authorized revenue requirements to rate groups, a process called revenue allocation. Finally, marginal costs are considered when designing rates (for each rate group) to recover the allocated revenue requirements. 1 Revenue requirements are the costs of providing utility services that the Commission has determined are appropriate to recover through customer rates. Rate groups are categories into which customers are grouped, such as residential service or small general service. Rates are the regulated (tariffed) prices charged to customers in each rate group for utility services. These rates typically consist of multiple components, such as energy charges, demand charges (where metering permits) and a monthly customer charge. 2 Regulated utility services refer to electricity supply (production or procurement of power for customers), electricity delivery (transmission, subtransmission and distribution) and customer services (interconnection to the delivery system and managing SCE s relationship with customers, including handling customer communications, measuring usage, maintaining records, and billing.) -1-

7 In the testimony which follows, SCE presents marginal costs for three cost drivers: electricity usage, design demand, and number of customers. The cost of procuring electricity to meet changes in customer electricity usage varies hourly. In addition, SCE and other retailers are required to procure dependable generation resources with sufficient capacity to meet a 115% to 117% resource adequacy requirement in order to provide an adequate level of reliability to customers. Marginal generation costs (energy and capacity) are associated with the electricity usage cost driver. Marginal generation costs are aggregated and presented in time-of-use periods which group together hours with similar marginal generation costs. SCE s electric delivery system consists of a network of high-voltage (transmission and subtransmission) and low-voltage (distribution) facilities which connect generation resources to individual customers. The delivery system is designed and constructed to meet the expected peak demand placed on it, so design demand is an associated cost driver. Design demand is a localized cost driver, since portions of SCE s delivery system peak at different times depending on the area, the mix of customers in the area, and the location of the generating resources serving customer loads in the area. In addition to design demand, some of the costs which customers impose on the delivery system are fixed based on the customers location, but do not vary with customer electricity usage. That is, a portion of the delivery system represents grid infrastructure which, like streets and roads, is extended to those who live in an area regardless of actual usage. 3 Finally, the number of customers is a cost driver, reflecting the marginal costs of customer interconnection to the delivery system and various customer services. Since the marginal costs of customer interconnection and customer services vary by type of customer, there is an individual marginal cost for each customer category. SCE s customer marginal costs, as are generation capacity 3 In general, the underground conduit running through a residential or commercial development and the poles running through a right-of-way between adjacent rows of homes are not any bigger or more costly depending on how much electricity they carry. Grid infrastructure is not a new concept. The original marginal cost methodology adopted by the Commission identified a minimum distribution system component of marginal customer costs which identified the costs associated with a hypothetical redesigned distribution system capable of delivering only a minimum amount of electricity but capable of providing service to all customers. Measurement was difficult, and this approach was eventually abandoned. -2-

8 and delivery marginal costs, are calculated based on the real economic carrying charge (RECC) methodology. 4 However, in recognition that the Commission adopted an alternative new customer only (NCO) method in SCE s 1995 GRC, calculations based on the NCO methodology are also presented. SCE s recommended marginal costs are summarized in Tables I-1, I-2 and I-3. Table I-1 Electricity Usage -Related Marginal Costs (2009$, at generation level) Electricity Usage Marginal Cost ( Average) Marginal Energy Cost ( /kwh) Marginal Capacity Cost ($/kw) Summer On-Peak Summer Mid-Peak Summer Off-Peak Winter Mid-Peak Winter Off- Peak Annual Based upon the time periods from the TOU-8 tariff Table I-2 Delivery-Related Marginal Costs (2009$, at applicable voltage level) $/kw Delivery-Related Marginal Costs Marginal Subtransmission Cost (Non-ISO) - 66KVA 29 Marginal Distribution Cost 12KVA Section B of this chapter describes the principles and methodological approaches that guided the development of SCE s marginal costs. Finally, Section C presents SCE s marginal cost study and the derivation of individual marginal cost components. A glossary of terms is provided in Appendix A. Additional information supporting SCE s marginal cost study is presented in Appendices B through D. The results from applying the NCO methodology are presented in Appendix E. 4 This methodology is also called the rental value method or the economic deferral method. -3-

9 Table I-3 Marginal Customer Costs (In $/Customer, 2009$) Customer Costs (2009$) $/Customer/Year Domestic GS TC GS-2 Non-TOU 1, TOU Option 2, TOU-GS-3 ( kw) 4, TOU-8 Secondary 5, Primary 2, Sub-trans 21, PA PA-2 1, TOU-PA-5 2, AG-TOU 2, Metered Street Lights Unmetered Street Lights* Per Customer LS-1 + Per Lamp 9.34/lamp LS-2 + Per Lamp 10.09/lamp OL + Per Lamp 9.46/lamp DWL + Per Lamp 4.31/lamp *Unmetered Street Light Customer marginal cost is a per customer cost plus a per lamp cost. -4-

10 B. Methodology Overview 1. Marginal Cost Principles The Commission s reliance on marginal cost principles for revenue allocation and rate design is well founded on economic principles. Marginal cost reflects the change in cost incurred to serve a small increment or decrement in demand for utility services. Setting utility rates equal to marginal costs provides an economically correct price signal, which encourages customers to use electricity efficiently and to make appropriate choices when purchasing electricity-consuming equipment and appliances. When utility rates are not set equal to marginal cost, some users of utility services receive a subsidy and others incur a penalty, resulting in socially inefficient electricity consumption. Moreover, industry restructuring has led to growing interest in customer-site distributed generation and demand response, and increased awareness of distribution competition among utilities, municipalities and other public entities. In this environment, inefficient pricing can lead to uneconomic bypass of utility facilities, resulting in unnecessary investment in duplicative facilities and higher rates for remaining utility service customers. The Commission deviates from setting rates equal to marginal costs in order to establish overall utility rates that recover a utility s authorized revenue requirements. In practice, the Commission has frequently assigned authorized revenue requirements in proportion to marginal cost revenues by the equal percent of marginal cost (EPMC) method. 2. Marginal Cost Scope And Application SCE s marginal cost study reflects the full chain of services required to provide electricity to customers, although SCE s role in the provision of such services remains somewhat unclear. At present, state law allows some customers to directly access markets for electricity supply instead of procuring such service from SCE. This includes community choice aggregation (CCA) customers and existing direct access (DA) customers. Currently, DA is suspended for new entry. Existing DA customers are permitted to obtain some metering and billing services from their electricity supply provider instead of SCE. For the purpose of this testimony, SCE is assumed to obtain electricity supply either from wholesale market purchases or from its own generating facilities. It is also assumed -5-

11 that SCE will continue to provide metering and billing services. Thus, the marginal costs included in this study reflect the costs of serving a bundled service customer. As a result of industry restructuring, SCE s higher voltage transmission facilities were transferred to Federal Energy Regulatory Commission (FERC) jurisdiction and placed under the operational control of the California Independent System Operator (ISO). FERC-jurisdictional (ISO controlled) assets and activities have not been included in the marginal cost study. Marginal costs associated with the FERC-jurisdictional facilities and activities are excluded from marginal cost revenues and the revenue allocation process, since FERC is responsible for determining revenue requirements and rates associated with these facilities and activities. This marginal cost study is intended to represent conditions expected to occur during In particular, electricity supply marginal costs are based on a three-year forecast (expressed in constant 2009 dollars). Thus, there is no need to true-up SCE s marginal costs in annual rate design proceedings. 3. Cost Drivers The cost drivers that SCE has identified and used to determine its marginal costs are described below. a) Electricity Usage Cost Driver The cost associated with a change in customer electricity usage includes energyrelated and capacity-related components. Since SCE buys and sells power in the electricity market in which its service area is located, the market clearing price of this power is an appropriate measure of energy-related marginal generation costs. As described further in Section I.C.1., energy related marginal generation costs are forecast through a combination of market-based estimates (broker quotes) and production simulation model forecasts of market clearing prices. Capacity-related marginal generating costs are measured by annualizing the expected costs of a utility-build combustion turbine (CT) as a proxy. Because CTs operate during periods of high market prices, and are able to earn energy rents (operating profits in excess of variable operating costs) that recover a portion of their fixed costs, these -6-

12 energy rents are deducted from the annualized CT proxy costs to determine capacity-related marginal costs. Energy-related marginal costs are aggregated into time-of-use (TOU) periods. Capacity-related marginal costs are assigned to TOU periods using a loss-of-load expectation 5 (LOLE) measure, also derived from production simulation modeling. b) Design Demand Cost Driver Design demand is the amount of delivery capacity that T&D planners determine to be necessary when planning to serve the additional demand of a customer or group of customers. For a large customer, planners may investigate the customer s electrical equipment, and the expected utilization of this equipment (i.e., customer site diversity of use) in order to size the customer s final line transformer and upstream facilities. For smaller customers, planning standards have been developed to identify expected peak demand. Smaller customers frequently share a final line transformer, and design demand takes into consideration diversity of appliance use within the customer s premise, and diversity between customers served from the same transformer. Design demand is expected to have the greatest impact on the capacity of transformers in the delivery system. Power is typically delivered to the transmission system from regional generators or regional interties at 220 kv or higher voltages. This power typically goes through three stages of transformation: from 220 kv to 66 kv (subtransmission voltage), from 66 kv to 12 kv (primary voltage), and from 12 kv to between 120 and 480 volts at the customer premise (secondary voltage). When there is an increase in design demand, additional transformer capacity must be added at each of these steps to accommodate the increase. Additional substation facilities are required as a result of increases in transformer capacity. An increase in design demand also results in an increase in the number of distribution circuits 6 serving a local area. 5 Also called loss-of-load probability, or LOLP. 6 Distribution circuits are lines connecting customers in an area to a nearby substation. -7-

13 The design demand cost driver does not readily relate to the kinds of usage data normally recorded for SCE s customers. Design demand is related to a customer s expected maximum usage at the time of service installation, but it is difficult to track this over time. In older neighborhoods, for example, transformer capacity, and distribution circuit routings may have been reconfigured over time to keep up with increasing demand. Keeping track of the contribution of an individual customer to the delivery capacity built to serve an area would be highly subjective. In addition, the time in which maximum usage occurs varies by climate zone and by the mix of customers in an area. Thus, system peak demand (a measure appropriate for the capacity component of marginal generation costs) and design demand are not necessarily coincident. In order to relate design demand to measurable customer attributes, SCE developed a measure of peak load diversity, which we call effective demand. Effective demand is expressed as a factor (effective demand factor or EDF), which is the ratio of a customer s contribution to the peak load on a transmission or distribution circuit to the customer s annual noncoincident peak demand. EDFs vary by type of customer and by the voltage level of the circuit. Unlike rate group coincident demand, which is measured for customers within a particular rate group, effective demand takes intergroup diversity into account. This is important because the impact of a particular customer on delivery capacity in an area may vary depending on the characteristics of nearby customers. For example, a medium-sized business connecting to a distribution circuit primarily serving other business customers would cause planners to consider the customer s entire maximum load in circuit design. However, the same business connecting to a distribution circuit in a residential area would not have as great an impact, since residential customers tend to peak later in the day than business customers. SCE has over 4,000 distribution circuits, each of which typically provides service to customers in a variety of rate groups. Distribution circuit EDFs are calculated as follows. First, the number of customers by rate group is tabulated for each circuit, and is used to develop a profile of the number of customers by rate group on a typical distribution circuit. These profiles are calculated for each type of customer, using an average of the circuits weighted by the number of customers of the particular type. For example, the typical TOU-8 (large customer) distribution circuit has fewer -8-

14 residential and small business customers, since the design demand of the large customer leaves less capacity available for others. 7 Next, a Monte Carlo simulation method is used to randomly populate each typical circuit type with customers from SCE s load research samples. This step is performed for each circuit type. Next, individual customers on each simulated circuit are selected, and the contribution of the customer to the circuit peak is determined. For example, if the Monte Carlo simulation is for a typical TOU-8 customer distribution circuit, the effect of one of the TOU-8 customer s load on the circuit is calculated. Finally, the second and third steps are repeated a sufficient number of times to produce statistically valid results. A similar approach is used to determine EDFs for subtransmission (e.g., 66 kv) circuits. Due to the greater geographic area typically served by these higher voltage circuits, a single typical customer profile is used for all customer types. EDFs vary from around 20-30% (residential and small agricultural customers) to 60-80% (medium and large business customers). In general, higher load factor customers have higher EDFs, since their peak demands are more coincident with circuit peaks. Also, larger customers tend to have greater EDFs, since their load has proportionately greater influence on circuit peaks than smaller customers. The load research study performed to compute EDFs (by customer group and voltage level) is described in greater detail in Appendix B of this exhibit. Since EDFs relate individual customer peak demand to the customer s contribution to delivery system demand, the marginal cost revenues associated with a rate group s design demand are calculated by multiplying that rate group s annual noncoincident peak demand times the EDF for that rate group, times the marginal cost per unit of design demand. Currently, SCE has approximately 300 standby customers who self-generate to meet a portion of their electrical requirements, and rely on SCE for generation service when their generation facility is unavailable. The EDFs that SCE has calculated should be applied to both regular and standby customers, since SCE will need to reserve sufficient delivery system capability to service standby loads without adversely affecting other customers. It is sometimes alleged that utilities may not need to reserve such delivery system capability because the likelihood of multiple standby generators 7 Distribution circuits are typically sized to handle about 400 amperes of current flow. At 12 kv this is adequate to serve about 9,000 kw of demand. -9-

15 going off line at the same time is limited. This is based upon the assumption that if there are a large number of standby customers on the same circuit, then the probability that all customers generation is off-line is very low. However, if there are only a few standby customers on the same circuit, the planner must account for the potential that they may be offline at the same time. Thus, capacity must be reserved for their use. For Phase 2 of its 2006 GRC, 8 SCE performed a customer study which concluded that standby loads are widely dispersed throughout SCE s service area, so there is little opportunity to take advantage of outage diversity in the delivery system. c) Customer Cost Driver Finally, the number of customers is a cost driver, since each customer requires an interconnection with the delivery system (called a service drop for smaller customers) and a meter to measure consumption. 9 In addition, SCE incurs marginal costs in managing its relationship with customers, including handling customer communications, measuring usage, maintaining records, and billing. The cost of interconnecting a customer to the delivery system varies by type of customer, reflecting differences in size, service voltage, metering requirements, and other factors. The change in cost associated with serving a small increment or decrement in the number of customers is identified through typical customer cost studies. These studies are performed for customers in each rate group. Where there are cost differences within a rate group, such as between single-family and multifamily residential dwellings, more than one typical customer cost study is performed. The typical customer cost studies identify facilities directly associated with the customer interconnection, such as the meter, service drop, protection equipment, and final line transformer. Final line transformers are associated with the customer cost driver because the cost per kw varies for customers in different rate groups. 8 SCE s 2006 GRC Phase 2 proceeding (A ), Appendix F. 9 Technically, this description refers to a service account. Some customers, such as a firm owning a chain of retail stores or a large facility with several points of service at a single site, have more than one service account. For the vast majority of our customers, the terms customer and service account are synonymous, so we use the term customer in this testimony. In rare instances where customer usage is highly predictable, SCE provides unmetered service. -10-

16 Customer service costs include handling customer communications, measuring usage, maintaining records, and billing. We identify the specific activities and assets directly attributable to providing the particular services and then calculate the associated marginal costs. These marginal costs are calculated by customer type and size. 4. Time-Of-Use (TOU) Issues a) Generation Marginal Costs Generation marginal costs vary hourly, primarily because different units are on the margin each hour based on the level of customer demand and because the costs associated with maintaining sufficient capacity to meet reliability targets are properly associated with those hours where the gap between customer demand and generation resource availability is narrow and reliability is stressed. In this marginal cost study, generation marginal costs are averaged by TOU period, corresponding to the pricing periods in SCE s current TOU rate schedules. These TOU periods vary seasonally (summer and winter) and daily (on-peak, mid-peak, and off-peak) and are intended to group together hours with similar marginal cost characteristics. SCE periodically reviews the appropriateness of its TOU periods and recommends changes as part of its marginal cost studies when appropriate. An analysis of SCE s current TOU periods, and various alternative periods including a narrower or later in the day summer on-peak period is provided in Appendix D. The results of this study show that narrowing the existing summer on-peak period would not significantly improve the accuracy of SCE s current TOU periods. Thus, SCE does not recommend any changes to its existing TOU periods. In addition, SCE cautions that the factors which appear to trigger stresses in the generation markets which lead to periods of low reliability or higher prices are not necessarily related to periods of highest loads. 10 With the design of California s wholesale electricity markets still a work in progress, it is better to move slowly in considering changes to the TOU periods. 10 A 2002 conference paper presented by a PG&E employee observed that 29% of the California ISO s summer season Stage 1 emergency hours in occurred outside of the summer peak period, for instance. Robert Levin, Does California Avoided Cost Methodology Undervalue Energy Efficiency, presented at the 2002 Rutgers Economics Conference. -11-

17 b) Delivery-Related Marginal Costs Ideally, delivery-related marginal costs would be time differentiated to address differences in the pattern of electricity consumption by individual customers within each group. For instance, if peak demands on subtransmission and distribution facilities were consistently experienced only on summer days throughout SCE s service area, it would improve pricing accuracy to recover delivery-related marginal costs based on customers summer daytime peak demands. This would allow a customer that uses electricity predominantly in the winter or at night to pay proportionately less, in recognition that if that customer s peak load were to increase or decrease, there would be no impact on delivery-system capacity requirements. However, an analysis of customer loads at the distribution circuit level showed that there is substantial variation in the hours when individual circuits peak. 11 While a majority of circuits experience loads during summer daytime hours that are at or near their annual peak loadings, a sizable minority (about one-third) of the circuits experience peak loads during winter and nighttime periods that are equal to or near their annual peaks. Not surprisingly, these latter circuits tend to be in coastal and mountain climate zones where there is less air conditioning load. This suggests that for a substantial number of customers, an increase in winter or nighttime usage would contribute to delivery system peak demand. Time differentiating the distribution portion of rates to recover design demandrelated marginal costs based on summer peak period usage would reduce pricing accuracy for these customers. For this reason, it is not appropriate to time differentiate delivery-related marginal costs. 5. Annual Cost Of Capital Investments (RECC vs. NCO) When computing marginal costs, SCE converts capital investments into annual costs using a real economic carrying charge (RECC). This approach is sometimes called the economic deferral or rental value method. Under this approach, which is illustrated in Figure I-1, the present 11 A distribution circuit circuit study was presented in SCE s 2006 GRC Phase 2, Appendix D. SCE does not anticipate any significant changes in circuit peak behavior since the study was performed, and the conclusions from the prior study remain valid. -12-

18 worth of the annual revenue requirements 12 for an asset and its subsequent replacements are computed, and then compared to the present worth of an equivalent asset and its replacements installed one year later. The only difference between these two scenarios is that SCE loses the opportunity to use the asset in the first year of the second scenario. Thus, the difference in present worth between the two scenarios measures the economic (opportunity) cost of using the asset during this year. The resulting annual charge, when escalated at the rate of inflation over time and then discounted, yields the original cost (in terms of revenue requirement) of the investment. As shown in Figure I-2, the net present value (NPV) of the two payment streams are the same, but the RECC results in the same real payment over time. This conclusion is important because in real terms the charge for an asset is the same over time and assuming electricity customers value the service they receive the charge should be the same regardless of the age of the equipment. Therefore, the proper charge can be calculated for both existing and new customers by applying the RECC to the current cost of the equipment. This RECC approach is documented in work prepared by the National Economic Research Associates for an Electric Utility Rate Design Study, which was funded by various parties including the National Association of Regulatory Utility Commissioners. 13 The RECC approach has been used to annualize T&D and generation capacity costs by SCE, PG&E, and SDG&E, but has been controversial in determining customer marginal cost. 12 The revenue requirement includes depreciation, return on investment, income taxes, property taxes A&G, insurance, and salvage costs. 13 NERA #15 Topic 1.3, A Framework for Marginal Cost-Based Time Differentiated Pricing in the United States, February 1977, pages 90-94, and Appendix C. See also NERA #23 Topic 4, How to Quantify Marginal Costs. -13-

19 Figure I-1 Illustration of the RECC Methodology Annual Revenue Requirement Asset in Year 0 Asset in Year Year -14-

20 Figure I-2 Annual Payment for $100 Capital Investment 10% Discount Rate and 3% Annual Inflation $30 $25 RECC Payment $20 $15 $10 Annual Revenue Requirement $5 $ Year In SCE s 1995 GRC, SCE also proposed that annual marginal costs for capital investments be calculated using the rental value method. With respect to marginal customer costs, but not for marginal transmission or distribution costs, TURN advocated use of the so-called new customer only (NCO) method instead of the rental value method. 14 In that proceeding, the Commission observed 14 The NCO method omits the cost of existing customer interconnection facilities from marginal costs, because facilities such as service drops and meters are largely sunk investments that would be stranded were a customer to cease service. (See D , pp ). Although TURN supported NCO for customer interconnection facilities, it did not oppose SCE s proposal to annualize transmission and distribution investments using the RECC method (p. 66). -15-

21 that in previous utility cases it had alternated between the rental value and NCO methods, 15 but that... the record in this case supports the use of this approach (the NCO method) as an improvement over the rental method. 16 The RECC method (rental value) is an appropriate measure of marginal costs. The NCO method includes only the cost of new customer interconnection, spreading these costs across both existing and new customers. By ignoring the economic value of existing interconnection facilities, the NCO method systematically understates marginal costs. Simply because an asset is already installed and thus sunk, does not mean the asset loses its economic value. As long as the interconnection has value to the customer, there is a price at which the customer is willing to buy and the utility is willing to sell interconnection service. Since it ignores this economic value, NCO is not a valid marginal cost methodology. Considering the results of the RECC and NCO methods in the context of competitive markets makes the economic issues clearer. As an example, in recent years, the Commission has considered making provision of meters (a portion of the cost of customer interconnection) a competitive service. In a fully competitive market for meter ownership, SCE would be a price taker in competition with other entities who also offer meters. SCE s hypothetical competitors would not offer meters without a reasonable expectation of fully recovering the associated investment cost from the customers to whom meters are provided. Thus, the market clearing price for meters would reflect the full cost of a new (and newly installed) meter. In this environment, SCE would not need to discount its charges for existing meters below what a competitor would charge for a new meter. SCE would be able to fully recover the market cost of meters from existing customers, regardless of the fact that SCE s meters are a sunk investment. 17 The NCO method, which ignores existing meters in computing marginal costs, ignores this market reality. 15 D , p D , p The concept of sunk investment is sometimes confused with stranded investment. If SCE spends $1000 on a meter, then sunk investment (ignoring potential salvage value) is $1000. If new meters cost $1000, then SCE s entire (Continued) -16-

22 In addition, it is suggested that customer connection investments are different than other investments because the facilities are dedicated to a customer or that the final line transformer cannot be cheaply moved to another location so its opportunity costs are negligible. 18 Both of these claims are false. Residential and small commercial customers generally share the final line transformer so any capacity reduction by one customer can be used by another customer. If the existing transformer needs to be removed due to upgrade or relocation, the transformers can be re-used if the transformer still has a sufficient useful remaining life, it is not likely to be retired but will be refurbished and placed back in inventory. 19 In addition to its lack of theoretical foundation, the NCO method can create nonsensical results. This sensitivity is unrelated to any possible measure of market conditions, and thus served to demonstrate the flawed nature of this method. For example, assume that a customer class is expected to grow by 10 newly added customers and decline by 15 departing customers for a net reduction of 5 customers. The NCO method would yield either a zero or negative marginal cost, depending on how the method is applied. Yet the utility still incurs the new costs to install equipment for the 10 customers that were added. Changes in the growth forecast can yield declining costs one year and increasing costs the next, yet the underlying cost structure remain unchanged. The NCO method is flawed because any change to the customer growth forecast will result in a change in unit costs ($/customer) even though the cost of equipment and labor remain constant. C. Marginal Cost Methodology This section describes the calculation of electricity usage marginal costs, design demand marginal costs, grid infrastructure marginal costs, and customer marginal costs. In addition, the marginal cost of streetlight facilities (streetlight poles, luminaries, and lamps) is calculated. Continued from the previous page investment is economic, and there is no stranded investment. If the price of new meters drops to $800, then only $800 of SCE s $1000 sunk cost is economic and the remaining $200 is potentially stranded. 18 TURN s testimony in SCE s 2006 Phase 2 GRC, page GRC Phase 1, SCE-11, Volume 3, page 43, lines

23 Electricity Usage Marginal Costs The Commission has a long-standing policy of developing marginal generation costs using the deferral value 20 of a combustion turbine ( CT ) proxy for estimating the avoided cost of capacity, and a system marginal energy cost for estimating the avoided cost of energy. The onset of deregulation warranted combining capacity and energy costs into a single market price, since the market structure originally adopted in California anticipated that generators would recover their fixed investment costs from market prices during periods of scarcity when the market prices rose above the variable operating costs of generators participating in the market. However, the Commission s implementation of resource adequacy requirements has returned the utilities and the market to a structure that requires the clear separation of capacity and energy costs as the basis for generation marginal costs. The separation of energy and capacity values into two pricing streams makes it possible to evaluate the relative reliability contribution of different resources, and is more flexible across a range of capacity factors and time of use periods. The marginal cost analysis presented here is intended to represent conditions expected to occur during 2009 through The results of SCE s analysis are summarized in Table I-4. Marginal Energy ( /kwh) Marginal Capacity ($/kw-yr) Table I-4 Generation Marginal Costs, Average (2009$) Annual On- Peak Summer Mid- Peak Off- Peak Mid- Peak Winter Off- Peak Based upon the time periods in the TOU-8 tariff 20 Also referred to as the real economic carrying charge (RECC) methodology. -18-

24 a) Generation Capacity Marginal Cost The generation capacity marginal cost is based on the deferral value of a CT proxy, net of any energy rents obtained from the market prices. Energy rents are the operating profits that a proxy CT is able to earn when market prices are above the CT s variable operating costs (principally fuel and variable O&M). Because these energy rents reduce the amount of the CT s fixed costs that need to be recovered in capacity markets, energy rents are also known as energy-related capital costs (ERCC). Figure I-3 illustrates this calculation. The CT proxy is the estimated cost (in $/kw) for a new SCE-owned combustion turbine in the Southern California region, including all permitting, financing, development costs and inflation during the construction period. The annualized cost ($/kw-yr) is then calculated using the RECC methodology, plus costs for fixed O&M and property taxes. SCE has estimated the CT proxy cost based on a 2009 commercial operating date as shown in Table I-5 below: Table I-5 CT Proxy Cost (2009$) 1. Combustion Turbine Installed Cost ('09 COD) $/kw Real Economic Carrying Charge Rate 11.02% 3. Annualized CT Installed Cost (1 * 2) $/kw-yr Fixed O&M $/kw-yr Property Tax $/kw-yr Full CT Proxy Cost ('09) ( ) BOY $/kw-yr Full CT Proxy Cost Mid-Year Payment $/kw-yr

25 Figure I-3 Derivation of Capacity Value CT Proxy with Energy Rent Adjustment Fixed CT Cost Market Price Contributes to CT s Fixed Cost Value of Energy Rents Full CT Variable Costs Proxy Capacity Value Hours Due to the separation of capacity and energy prices, the CT Proxy Cost must be reduced for any energy rents obtained in the market price to avoid double counting the capacity value. Energy rents occur when the market prices exceed the variable operating costs for fuel and O&M. For example, if the marginal energy price forecast from the model is $90 per MWh, but the variable operating cost of a CT proxy is $60 per MWh for that same hour, the CT would realize a $30 per MWh contribution to its fixed costs. Failure to adjust the CT proxy for energy rents would double count the capacity value obtained in the market price. To alleviate this issue, the value of energy rents (or ERCC) is subtracted from the full CT proxy. The energy rent revenue for each year is divided by the name plate capacity which resulted in an average of $15.8/KW-year in energy rents which is subtracted from the CT Proxy. As shown in Table I-5B the Generation Capacity Marginal Cost is $90.9/KW-year (2009$). -20-

26 Table I-5B Generation Capacity Marginal Cost (2009$) 1. Full CT Proxy Cost Mid-Year Payment $/kw-yr Less Energy Rents $/KW-yr (15.8) 3. Incremental Capacity Cost $/kw-yr % General Plant Loader $/kw-yr Generation Capacity Marginal Cost $/kw-yr The marginal capacity cost calculated above is an annualized number and is not differentiated by time-of-use periods. SCE allocates the marginal capacity cost using relative LOLE values to indicate time-differentiated values based on peak period usage. 21 This is a theoretically valid approach to assigning reliability costs to time periods. LOLE is closely related to Expected Unserved Energy 22 (EUE), which identifies the potential amount of generation-related outages (in MWh of unserved energy) which would occur in a time period considering uncertainty in customer loads, resource availability, and other market conditions. If available generation increases by one MW, then LOLE is equal to the change in EUE which occurs as a result. 23 Thus, LOLE measures the improvement in reliability which occurs in a time period as a result of an increase in available generation or a decrease in customer load. The capacity value allocation results are shown in the following Table: 21 This approach is a standard utility practice and has been used in prior SCE GRC proceedings. 22 Also called energy not served, or ENS. 23 For example, if the likelihood of rolling blackouts due to a generation resource shortage is 10% in a particular hour (the LOLE) and the utility adds 100 MWH of additional generation resources, then the expected amount of unserved energy (the EUE) would go down by 10 MW (10% times 100 MW times one hour). Mathematically, LOLE is the first derivative of EUE with respect to a change in available resources. -21-

27 Table I-6 Generation Capacity Marginal Costs, Average (2009$) Annual Summer Winter On-Peak Mid-Peak Off-Peak Mid-Peak Off-Peak Based on the time periods in the TOU-8 tariff. Includes general plant b) Energy Marginal Cost The marginal energy cost forecast was based on a combination of market-derived forward prices and prices from a fundamentals-based production cost simulation. A blended approach of market forwards and fundamentals is a simple and practical method used to account for the latest market view of power prices in the near term, and to account for the declining liquidity of the market view by incorporating a fundamental view in the long term. The market-based price forecast is based on the average price quote from three power brokerage firms as of November The fundamentalsbased price forecast is based on a Market Analytics production simulation (described in further detail below). Over the GRC forecast horizon, SCE blended the market forwards and fundamental views by assigning a declining weight factor to the market forwards over time. SCE bases its fundamentals-based forecast of marginal energy costs on its best estimates 2006 Long Term Procurement Plan (LTPP) filing, 25 with updated assumptions from late 2007 for gas prices, loads and resources to better reflect current forecast conditions. In addition to these changes, the 2006 LTPP was also updated for recently announced resource additions and retirements, 24 SCE anticipates that similar information would be available to a party seeking to purchase price forecasts from a brokerage firm. 25 Rulemaking R , filed on December 11,

28 load forecasts throughout the Western Electricity Coordinating Council (WECC) region, and generic resource additions due to changes in the load forecast. Major assumptions in the LTPP include: 26 SCE meeting the 20% Renewable Portfolio Standard by 2010 per the Energy Action Plan; SONGS steam generator replacement in timeframe; Significant increases in cost effective energy efficiency and demand response programs. The LTPP and the simulated generation marginal cost analysis were developed using the Global Energy Decisions (GED) Market Analytics production simulation tool, which utilizes ProSym as the underlying engine. 27 Market Analytics divides the WECC into several transmission areas ( transareas ) based on regional differentiation in the transmission grid. Each transarea contains the hourly loads of each load-serving entity; the available thermal, hydro and renewable supply resources; and their operating characteristics, including fuel costs. These transareas are joined by paths, which reflect aggregated transmission capability between regions. Multi-year simulations, in which resources are dispatched according to least-cost economics to meet the load, are performed on an hourly basis. Energy transfer occurs between transareas, where possible, to find the optimal solution that fits all of the user-defined constraints (operating limitations, reserve requirements, etc.). The outputs of the simulations are typically the operating costs of generating units, energy-not-served, and a forecast of market clearing prices for each region. Costs included in the energy price are those for incremental fuel, variable O&M, emissions costs, 28 startup costs, congestion charges, and no-load fuel. Costs excluded from the energy 26 SCE s best estimates LTPP was adopted in D , but with a number of changes, including revisions to use a recent California Energy Commision load forecast, a higher target for energy efficiency, and a greater rate of retirement for older power plants. SCE has not yet revised its production costs models to reflect these assumptions. 27 Global Energy Decisions, formerly known as Henwood Energy Services, Inc., is the developer of the EnerPrise software package. Market Analytics, a module within EnerPrise, is the updated version of MarketSym. Both Market Analytics and MarketSym use ProSym as the core simulation engine. 28 Only traded (valued) emissions costs (i.e, NO x and SO 2 ) are included. -23-

29 price are capital costs, fixed O&M, and property taxes these are explicitly included in the CT proxy capacity price. Additionally, certain ancillary services costs are captured in the CT proxy capacity value estimate. The marginal energy prices are computed by the Market Analytics model for years in hourly increments. The prices are then sorted and averaged by time-of-use periods corresponding to the pricing periods in SCE s TOU-8 rate schedule. This forecast is based on three-year average 29 gas price of $7.49 per MMBtu for firm delivery into the SoCalGas system. The results of SCE s energy marginal cost analysis are shown in the following Table: Annual Table I-7 Energy Marginal Costs, Average (2009 /kwh) Summer Mid- On-Peak Peak Off- Peak Winter Mid- Off-Peak Peak Based upon the time periods in the TOU-8 tariff The key drivers of the energy marginal price forecast are the gas and load forecasts. The gas price assumption is a blend of market forwards 30 and a fundamentals forecast from three vendors similar to the blended power price approach described earlier. The WECC load forecasts are primarily that which came with the Market Analytics database, except for California-based system loads, which were adjusted to reflect SCE s internal estimate for annual peak and energy requirements. Additional details on gas and load assumptions are provided in Appendix C ; 2009 dollars. 30 NYMEX natural gas futures (Henry Hub plus SoCal basis swaps) plus intrastate transport charges. -24-