2010 State of the Market Report for PJM: Technical Reference for PJM Markets

Transcription

1 200 State of the Market Report for PJM: Technical Reference for PJM Markets The Independent Market Monitor for PJM March 0, 20 Monitoring Analytics 20

2 This page intentionally left blank 2 Monitoring Analytics 20

3 Introduction 200 State of the Market Report for PJM: Technical Reference for PJM Markets This Technical Reference document includes technical details related to PJM markets, which are essential to a full and detailed understanding of the operation of PJM markets but which are not necessary for the general audience of the state of the market reports. The MMU will continue to supplement this Technical Reference and publish periodic updates. Monitoring Analytics

4 Capacity Market Background PJM and its members have long relied on capacity obligations as one of the methods to ensure reliability. Under the Reliability Assurance Agreement (RAA) governing the Capacity Market operated by the PJM regional transmission organization (RTO), each load serving entity (LSE) must own or purchase capacity resources greater than, or equal to, its capacity obligation. On June, 2007, the Reliability Pricing Model (RPM) Capacity Market design was implemented in PJM, replacing the Capacity Credit Market (CCM) Capacity Market design. This appendix explains certain key features of the RPM design in more detail. Demand VRR Curves Under RPM, PJM establishes variable resource requirement (VRR) curves for the PJM RTO and for each constrained locational deliverability area (LDA). The VRR curve is a demand curve based on three price quantity points. The demand curve quantities are based on negative and positive adjustments to the reliability requirement. The demand curve prices are based on multipliers applied to the net cost of new entry (CONE). Net CONE is CONE minus the energy and ancillary service revenue offset (E&AS). 2 The PJM reliability requirement, measured as unforced capacity, is the RTO peak load forecast multiplied by the RTO forecast pool requirement (FPR) less the sum of any unforced capacity (UCAP) obligations served by fixed resource requirement (FRR) entities. The FPR is calculated as ( + Installed Reserve Margin) times ( Pool Wide Average EFORd), where the Installed Reserve Margin (IRM) is the level of installed capacity needed to maintain an acceptable level of reliability. The PJM reliability requirement represents the target level of reserves required to meet PJM reliability standards. This section relies upon the cited PJM manuals where additional detail may be found. 2 See Manual 8: PJM Capacity Market, Revision 8 (Effective January, 200), p. 6 < > (.27 MB). 4 Monitoring Analytics 20

5 Load Obligations 200 State of the Market Report for PJM: Technical Reference for PJM Markets Participation by LSEs in the RPM for load served in PJM control zones is mandatory, except for those LSEs that have elected the FRR alternative. 3 Under RPM, each LSE that serves load in a PJM zone during the delivery year is responsible for paying a locational reliability charge equal to its daily unforced capacity obligation in the zone multiplied by the final zonal capacity price. LSEs may choose to hedge their locational reliability charge obligations by directly offering resources in the Base Residual Auction (BRA) and Second Incremental Auction or by designating self supplied resources (resources directly owned or resources contracted for through unit specific bilateral purchases) as self scheduled to cover their obligation in the Base Residual Auction. Base UCAP Obligations A base RTO UCAP obligation is determined after the clearing of the BRA and is posted with the BRA results. The base RTO UCAP obligation is equal to the sum of the UCAP obligation satisfied through the BRA plus the forecast RTO interruptible load for reliability (ILR) obligation, for delivery years prior to 202/203, or plus the RTO Short Term Resource Procurement Target for the delivery years 202/203 and forward. Base zonal UCAP obligations are defined for each zone as an allocation of the RTO UCAP obligation based on zonal, peak load forecasts and zonal ILR obligations, for delivery years prior to 202/203, or the zonal Short Term Resource Procurement Target for the delivery years 202/203 and forward. The zonal UCAP obligation is equal to the zonal, weather normalized summer peak for the summer four years prior to the delivery year multiplied by the base zonal RPM scaling factor and the FPR plus the forecast zonal ILR obligation, for delivery years prior to 202/203, or plus the zonal Short Term Resource Procurement Target for the delivery years 202/203 and forward. Final UCAP Obligation Prior to the 2009/200 delivery year, the final RTO UCAP obligation is determined after the clearing of the Second Incremental Auction (IA) and is posted with the second IA results. 4 For the 2009/200 through 20/202 delivery years, the final RTO UCAP obligations are determined after the clearing of the Third IA. Effective with the 202/203 delivery year, the final RTO UCAP obligations are determined after the clearing of the final incremental auction. Prior to the 202/203 delivery year, the final RTO UCAP obligation is equal to the sum of the UCAP obligation satisfied through the 3 See Reliability Assurance Agreement among Load Serving Entities in the PJM Region, Substitute Original Sheet No. 40 (Effective June, 2007), Schedule See Manual 8: PJM Capacity Market, Revision 8 (Effective January, 200), p. 86 < > (.27 MB). Monitoring Analytics

6 BRA and the second IA plus the forecast RTO ILR obligation. Effective with the 202/203 delivery year, the final RTO UCAP obligation is equal to the total MW cleared in PJM Buy Bids in RPM Auctions, including cleared MW in the BRA, less the total MW cleared in PJM Sell Offers in RPM Auctions for the given delivery year. Prior to the 2009/200 delivery year, the final zonal UCAP obligation is equal to the base zonal UCAP obligation plus the RTO UCAP obligation satisfied in the second IA multiplied by the zone s percentage allocation of the obligation satisfied in the second IA. For the 2009/200 through 20/202 delivery years, the final zonal UCAP obligation is equal to the zonal allocation of the RTO UCAP obligation satisfied in the BRA and second IA plus the zonal forecast ILR obligation. The allocation of the RTO UCAP obligation satisfied in the BRA and second IA to zones is on a pro rata basis based on the final zonal peak load forecasts. For the 202/203 delivery year and beyond, the final zonal UCAP obligation is equal to the zonal allocation of the final RTO UCAP obligation. The allocation of the final RTO UCAP obligation to zones is on a pro rata basis based on the final RTO and zonal peak load forecasts for the delivery year. LSE Daily UCAP Obligation Obligation peak load is the peak load value on which LSEs UCAP obligations are based. The obligation peak load allocation for a zone is constant and effective for the entire delivery year. The daily UCAP obligation of an LSE in a zone/area equals the LSE s obligation peak load in the zone/area multiplied by the final zonal RPM scaling factor and the FPR. Capacity Resources Capacity resources may consist of generation resources, load management resources and qualifying transmission upgrades, all of which must meet specific criteria. 5 Generation resources may be located within or outside of PJM, but they must be committed to serving load within PJM and must pass tests regarding the capability of generation to serve load and to deliver energy. Generation Resources Generation resources may consist of existing generation, planned generation, and bilateral contracts for unit specific capacity resources. Existing generation located within or outside PJM is eligible to be offered into RPM Auctions or traded bilaterally if it meets defined requirements. 6 Planned generation that is participating in PJM s Regional 5 See Manual 8: PJM Capacity Market, Revision 8 (Effective January, 200), p. 29 < > (.27 MB). 6 See Manual 8: PJM Capacity Market, Revision 8 (January, 200), p. 22 < (.27 MB). 6 Monitoring Analytics 20

7 Transmission Expansion Planning (RTEP) Process is eligible to be offered into RPM Auctions if it meets defined requirements. Load Management Resources Load management is the ability to reduce metered load upon request. 7 A load management resource is eligible to be offered as a demand resource (DR) or, prior to the 202/203 delivery year, interruptible load for reliability (ILR). DR is a load resource that is offered into an RPM Auction as capacity and receives the relevant LDA or RTO resource clearing price. ILR is a load resource that is not offered into the RPM Auction, but receives the final zonal ILR price determined after the close of the second incremental auction. DR and ILR resources must meet defined requirements. Energy Efficiency Resources Existing or planned Energy Efficiency (EE) resources may be offered in an RPM auction starting with the 202/203 delivery year and receive the relevant LDA or RTO resource clearing price. An EE resource is a project, including installation of more efficient devices or equipment or implementation of more efficient processes or systems, exceeding then current building codes, appliance standards, or other relevant standards, designed to achieve a continuous (during peak periods) reduction in electric energy consumption that is not reflected in the peak load forecast prepared for the delivery year for which the Energy Efficiency Resource is proposed, and that is fully implemented at all times during such delivery year, without any requirement of notice, dispatch, or operator intervention. 8 Qualified Transmission Upgrades A qualifying transmission upgrade may be offered into the BRA to increase import capability into a transmission constrained LDA. Such transmission upgrades must meet the identified requirements. 9 7 See Manual 8: PJM Capacity Market, Revision 8 (Effective January, 200), p. 28 < (.27 MB). 8 See Reliability Assurance Agreement among Load Serving Entities in the PJM Region, First Revised Sheet No. 35C (Effective March 27, 2009), Schedule 6, section M. 9 See Manual 8: PJM Capacity Market, Revision 8 (Effective January, 200), p. 38 < (.27 MB). Monitoring Analytics

8 Obligations of Generation Capacity Resources The sale of a generating unit as a capacity resource within PJM entails obligations for the generation owner. The first four of these requirements, listed below, are essential to the definition of a capacity resource and contribute directly to system reliability. Energy Recall Right. PJM rules specify that when a generation owner sells capacity resources from a unit, the seller is contractually obligated to allow PJM to recall the energy generated by that unit if the energy is sold outside of PJM. This right enables PJM to recall energy exports from capacity resources when it invokes emergency procedures. The recall right establishes a link between capacity and actual delivery of energy when it is needed. Thus, PJM can call upon energy from all capacity resources to serve load. When PJM invokes the recall right, the energy supplier is paid the PJM Real Time Energy Market price. Day Ahead Energy Market Offer Requirement. Market sellers owning or controlling the output of a generation capacity resource that was committed in an FRR Capacity Plan, self supplied, offered and cleared in any RPM auction, or designated as replacement capacity, and that is not unavailable due to an outage are required to offer into PJM s Day Ahead Energy Market. 0 When LSEs purchase capacity, they ensure that resources are available to provide energy on a daily basis, not just in emergencies. Since day ahead offers are financially binding, PJM capacity resource owners must provide the offered energy at the offered price if the offer is accepted in the Day Ahead Energy Market. This energy can be provided by the specific unit offered, by a bilateral energy purchase, or by an energy purchase from the Real Time Energy Market. Deliverability. To qualify as a PJM capacity resource, energy from the generating unit must be deliverable to load in PJM. Capacity resources must be deliverable, consistent with a loss of load expectation as specified by the reliability principles and standards, to the total system load, including portion(s) of the system that may have a capacity deficiency. In addition, for external capacity resources used to meet an accounted for obligation within PJM, capacity and energy must be delivered to the metered, PJM boundaries through firm transmission service. 0 See OA Schedule,.0.A (d). Deliverable per PJM. Reliability Assurance Agreement among Load Serving Entities in the PJM Region, Original Sheet No. 50 (Effective June, 2007), Schedule 0. 8 Monitoring Analytics 20

9 Generator Outage Reporting Requirement. Owners of PJM capacity resources are required to submit historical outage data to PJM pursuant to Schedule 2 of the RAA. 2 CETO/CETL Since the ability to import energy and capacity into LDAs may be limited by the existing transmission capability, PJM conducts a load deliverability analysis for each LDA. 3, 4 The first step in this process is to determine the transmission import requirement into an LDA, called the capacity emergency transfer objective (CETO). This value, expressed in MW, is the transmission import capability required for each LDA to meet the area reliability criterion of loss of load expectation due to insufficient import capability alone, of one occurrence in 25 years when the LDA is experiencing a localized capacity emergency. The second step is to determine the transmission import limit for an LDA, called the capacity emergency transfer limit (CETL), which is also expressed in MW. The CETL is the ability of the transmission system to deliver energy into the LDA when it is experiencing the localized capacity emergency used in the CETO calculation. If CETL is less than CETO, capacity related transmission constraints may result in locational price differences in the RPM. 5 This will also trigger the planning of transmission upgrades under the RTEP Process. Prior to the 202/203 delivery year, only an LDA with CETL less than.05 times CETO was modeled as a constrained LDA in RPM. Effective with the 202/203 delivery year, an LDA with CETL less than.5 times CETO is modeled as a constrained LDA in RPM. Starting with the 202/203 delivery year, regardless of the CETO/CETL results, separate VRR curves will be established for any LDA with a locational price adder in one or more of the three immediately preceding BRAs, any LDA that PJM determines in a preliminary analysis is 2 See Reliability Assurance Agreement among Load Serving Entities in the PJM Region, Original Sheet No. 53 (Effective June, 2007), Schedule. 3 See Manual 4B: PJM Region Transmission Planning Process, Attachment C: PJM Deliverability Testing Methods, Revision 4 (Effective February, 200), p. 45 < (887.5 KB). PJM Manual 4B indicates that all electrically cohesive load areas are tested. 4 See Manual 20: PJM Resource Adequacy Analysis, Revision 3 (Effective June, 2007), p. 32 < ( KB). 5 See Manual 8: PJM Capacity Market, Revision 8 (Effective January, 200), p. 0, < (.27 MB). Monitoring Analytics

10 likely to have a locational price adder based on historic offer price levels, and EMAAC, SWMAAC, and MAAC LDAs. Generator Performance: NERC OMC Outage Cause Codes Table 0 includes a list of the North American Electric Reliability Council (NERC) GADS cause codes that PJM deems outside management control (OMC). PJM does not automatically include cause codes as outside management control for the purposes of calculating unforced capacity, with the exception of code 9250 under certain conditions. 0 Monitoring Analytics 20

11 Table 0 NERC GADS cause codes that PJM deems outside management control 6 (OMC) Cause Code Reason for Outage 3600 Switchyard transformers and associated cooling systems - external 36 Switchyard circuit breakers - external 362 Switchyard system protection devices - external 369 Other switchyard equipment - external 370 Transmission line (connected to powerhouse switchyard to st Substation) 3720 Transmission equipment at the st substation (see code 9300 if applicable) 3730 Transmission equipment beyond the st substation (see code 9300 if applicable) 9000 Flood 900 Fire, not related to a specific component 9020 Lightning 9025 Geomagnetic disturbance 9030 Earthquake 9035 Hurricane 9036 Storms (ice, snow, etc) 9040 Other catastrophe Lack of fuel (water from rivers or lakes, coal mines, gas lines, etc) where the operator is not in 930 control of contracts, supply lines, or delivery of fuels 935 Lack of water (hydro) Labor strikes company-wide problems or strikes outside the company's jurisdiction such as 950 manufacturers (delaying repairs) or transportation (fuel supply) problems 9200 High ash content 9205 Poor quality natural gas fuel low heat content 920 Low grindability 9220 High sulfur content 9230 High vanadium content 9240 High sodium content 9250 Low Btu coal 9260 Low BTU oil 9270 Wet coal 9280 Frozen coal 9290 Other fuel quality problems 9300 Transmission system problems other than catastrophes (do not include switchyard problems in this category; see codes 3600 to 3629, 3720 to 3730) 9320 Other miscellaneous external problems 9500 Regulatory (nuclear) proceedings and hearings - regulatory agency initiated 9502 Regulatory (nuclear) proceedings and hearings - intervener initiated 9504 Regulatory (environmental) proceedings and hearings - regulatory agency initiated 9506 Regulatory (environmental) proceedings and hearings - intervenor initiated Plant modifications strictly for compliance with new or changed regulatory requirements (scrubbers, cooling towers, etc.) Miscellaneous regulatory (this code is primarily intended for use with event contribution code 2 to indicate that a regulatory-related factor contributed to the primary cause of the event) 6 See NERC. Generator Availability Data System Data Reporting Instructions, Appendix K < > (49 KB). Monitoring Analytics 20

12 Financial Transmission and Auction Revenue Rights Section 3 provides examples of topics related to Financial Transmission Rights (FTRs) and Auction Revenue Rights (ARRs): The sources of total congestion revenue and the determination of FTR target allocations and congestion receipts; The procedure for prorating ARRs when transmission capability limits the number of ARRs that can be allocated; and The establishment of ARR target allocations and credits through the Annual FTR Auction. FTR Target Allocations and Congestion Revenue Table 0 shows an example of the sources of total congestion revenue and the determination of FTR target allocations and congestion receipts. 2 Monitoring Analytics 20

13 Table 0 Congestion revenue, FTR target allocations and FTR congestion credits: Illustration Day-Ahead Congestion Revenue Pricing Node Day-Ahead Congestion Price Day- Ahead Load Load Congestion Payments Generation Congestion Credits Transmission Congestion Charges Day-Ahead Generation A $0 0 $0 00 $,000 ($,000) B $5 50 $750 0 $0 $750 C $20 50 $, $2,000 ($,000) D $25 50 $,250 0 $0 $,250 E $30 50 $,500 0 $0 $,500 Total 200 $4, $3,000 $,500 Balancing Congestion Revenue Real-Time Congestion Price Load Congestion Payments Generation Congestion Credits Transmission Congestion Charges Pricing Node Load Deviation Generation Deviation A $8 0 $0 0 $0 $0 B $8 0 $0 0 $0 $0 C $25 3 $75 5 $25 ($50) D $20 (5) ($00) 0 $0 ($00) E $40 7 $280 0 $0 $280 Total 5 $255 5 $25 $30 Transmission congestion charges accounting Balancing transmission congestion charges $30 +Day-ahead transmission congestion charges $,500 =Total transmission congestion charges $,630 FTR Target Allocations Positive FTR Target Allocations Negative FTR Target Allocations Path Day-Ahead Path Price FTR MW FTR Target Allocations A-C $0 50 $500 $500 $0 A-D $5 50 $750 $750 $0 D-B ($0) 25 ($250) $0 ($250) B-E $5 50 $750 $750 $0 Total 75 $,750 $2,000 ($250) Congestion accounting Transmission congestion charges $,630 +Negative FTR target allocations $250 =Total congestion charges $,880 Positive FTR target allocations $2,000 -FTR congestion credits $,880 =Congestion credit deficiency $20 FTR payout ratio 0.94 Monitoring Analytics

14 ARR Prorating Procedure Table 0 2 shows an example of the prorating procedure for ARRs. If line A B has a 00 MW rating, but ARR requests from two customers together would impose 75 MW of flow on it, the service request would exceed its capability by 75 MW. The first customer s ARR request (ARR #) is for a total of 300 MW with a 0.50 impact on the constrained line. It would thus impose 50 MW of flow on the line. The second customer s request (ARR #2) is for a total of 00 MW with a 0.25 impact and would impose an additional 25 MW on the constrained line. Table 0 2 ARR allocation prorating procedure: Illustration Line A-B Rating = 00 MW ARR # Path Per MW Effect on Line A-B Requested ARRs Resulting Line A-B Flow Prorated ARRs Prorated Line A-B Flow C-D E-F Total Equation 0 Calculation of prorated ARRs Individual prorated MW = (Line capability) i (Individual requested MW / Total requested MW) i ( / per MW effect on line). The equation would then be solved for each request as follows: ARR # prorated MW award = (00 MW) i (300 MW / 400 MW) i ( / 0.50) = 50 MW; and ARR #2 prorated MW award = (00 MW) i (00 MW / 400 MW) i ( / 0.25) = 00 MW. Together the prorated, awarded ARRs would impose a flow equal to line A B s capability (50 MW i 0.50) + (00 MW i 0.25) = 00 MW. ARR Credits Table 0 3 shows an example of how ARR target allocations are established, how FTR auction revenue is generated and how ARR credits are determined. The purchasers of FTRs pay and the holders of ARRs are paid based on cleared nodal prices from the Annual FTR Auction. If total revenue from the auction is greater than the sum of the ARR target allocations, then the surplus is used to offset any FTR congestion credit deficiencies occurring in the hourly Day Ahead Energy Market. For example, the FTR auction revenue is only $75 for the ARR on line A D while the ARR target allocation is $50. The surplus FTR auction revenue from the other ARR paths is enough to cover the $75 deficiency and fulfill the ARR target allocation of $50. 4 Monitoring Analytics 20

15 Table 0 3 ARR credits: Illustration Self-Scheduled ARRs 200 State of the Market Report for PJM: Technical Reference for PJM Markets Path Annual FTR Auction Path Price ARR MW ARR Target Allocation FTR MW FTR Auction Revenue ARR Credits A-C $0 0 $00 0 $00 $00 A-D $5 0 $50 5 $75 $50 B-D $0 0 $0 20 $200 $0 B-E $5 0 $50 5 $75 $50 Total 30 $ $450 $400 ARR payout ratio = ARR credits / ARR target allocations = $400 / $400 = 00% Surplus ARR revenue = FTR auction revenue - ARR credits = $450 - $400 = $50 Table 0 4 shows an example of two ARR customers, one of which self schedules ARRs and one of which retains ARRs. During an Annual ARR Allocation, both ARR customers # and #2 are allocated 0 MW ARRs on line A B. ARR customer # self schedules 0 MW ARRs on line A B as FTRs during the subsequent Annual FTR Auction while ARR customer #2 retains 0 MW ARRs on line A B. Based on cleared nodal prices from the Annual FTR Auction, ARRs on line A B are valued at $0 per MW. Customer #2 will receive $00 in ARR credits. Customer # converts all of the 0 MW ARRs on line A B to FTRs during the Annual FTR Auction and, as a result, this customer needs to pay $00 to purchase the associated self scheduled FTRs although this cost will be fully offset by the same amount of ARR credits. Based on the difference in LMPs, FTRs on line A B are valued at $5 per MW. Customer # will receive $50 in FTR credits. In summary, Customer # receives a net $50 in FTR credits as a result of self scheduling the 0 MW of allocated ARRs on line A B as FTRs, while Customer #2 receives $00 in ARR credits as a result of retaining the 0 MW ARRs on line A B. Table 0 4 Self Scheduled ARR credits: Illustration Customer # Path ARR MW Annual FTR Auction Path Price ARR Credits Converted to FTRs? Cost of Conversion to FTRs Day-Ahead FTR Path Price FTR Credits Total Credits A-B 0 MW $0 $00 Yes $00 $5 $50 $50 2 A-B 0 MW $0 $00 No $0 $5 $0 $00 Total credits = ARR credits - Cost of conversion to FTRs + FTR credits Monitoring Analytics

16 Calculating Locational Marginal Price In order to understand the relevance of various measures of locational marginal price (LMP), it is important to understand how average LMPs are calculated across time and across buses. This appendix explains how PJM calculates average LMP and loadweighted, average LMP for the system, for a zone and, by extension, for any aggregation of buses, for an hour, for a day and for a year. 7 This appendix also explains how the Market Monitoring Unit (MMU) calculates average LMP for states, consistent with the PJM method for other aggregates. Real-Time Hourly Integrated LMP and Real-Time Hourly Integrated Load In PJM a real time LMP is calculated at every bus for every five minute interval. The system real time, five minute, average LMP is the load weighted, average LMP for that five minute interval, calculated using the five minute LMP at each load bus and the corresponding five minute load at each load bus in the system. The sum of the product of the five minute LMP and the five minute load at each bus, divided by the sum of the five minute loads across the buses equals the system load weighted, average LMP for that five minute interval. In PJM, the real time hourly LMP at a bus is equal to the simple average of each hour s 2 five minute interval LMPs at that bus. This is termed the hourly integrated LMP at the bus. The hourly load at a bus is also calculated as the simple average of each hour s 2 five minute interval loads at that bus. This is termed the hourly integrated load at the bus. The hourly values for LMP and load are the basis of PJM s settlement calculations. Day-Ahead Hourly LMP and Day-Ahead Hourly Load The day ahead LMP is calculated at every bus for every hour from the day ahead dispatch required to meet estimated nodal loads derived from the distribution factors plus nodal load from decrement bids (DECs) and price sensitive load and nodal supply from generation offers and increment offers (INCs). The result is a full set of day ahead nodal LMPs and cleared, nodal loads. This measure of nodal, day ahead load is used in system load weighted, average LMP calculations. This is termed nodal, total day ahead load here. Zonal, day ahead hourly aggregate load is assigned to buses in the relevant zone using zonal distribution factors. Day ahead zonal distribution factors are calculated from historical real time, bus level load distributions that were in effect at 8 AM seven days prior. The use of load data 7 The unweighted, average LMP is also referred to as the simple average LMP. 6 Monitoring Analytics 20

17 from a period seven days prior to the DA price calculations provides a week day match, but the lack of adjustment for other factors that affect bus specific loads, including temperature, introduces a potentially significant inaccuracy in the load data used to clear the day ahead market. This would be an issue to the extent that weather or other factors changes the relative size of nodal loads. Zonal, day ahead, load weighted LMP is calculated from nodal day ahead LMP using zonal distribution factors as the load weights. This measure of load weights excludes bus specific loads, such as DECs, that clear in the day ahead market. The exclusion of bus specific loads from the calculation of day ahead load weighted LMP means that the zonal day ahead load weighted prices reported by PJM do not reflect the load weighted price paid by all load in a zone, but instead reflect only the price paid by the load that settles at the day ahead hourly zonal price. Factor distributed load, used in the calculation of state load weighted average LMP, is calculated by multiplying day ahead zonal hourly load (fixed plus price sensitive load only) by day ahead distribution factors. The factor distributed load calculation provides bus specific load weights, derived directly from the day ahead zonal distribution factors, which are used to calculate day ahead load and load weighted average LMP for states with load buses in multiple zones or parts of zones. This methodology is used because it results in weighted LMPs that are consistent with how zonal factor weighted prices are determined by PJM. This means that where the zone buses are the same as state buses, the result will be the same. For example, the state of Maryland contains buses from the AP, BGE, DPL and Pepco zones, but the areas encompassed by these aggregates, with the exception of BGE, extend beyond the borders of the state. AP, for example, extends past the western portion of Maryland into Pennsylvania, Ohio, West Virginia and Virginia. To provide Maryland specific results for load and LMP, a Maryland aggregate is calculated using only those AP, BGE, DPL and Pepco load buses that are physically within the geographic boundaries of the state of Maryland. Load-Weighted, Average LMP Real Time The system real time, load weighted, average LMP for an hour is equal to the sum of the product of the hourly integrated bus LMPs for each load bus and the hourly integrated load for each load bus, for the hour, divided by the sum of the hourly integrated bus loads for the hour. The zonal real time, load weighted, average LMP for an hour is equal to the sum of the product of the hourly integrated bus LMPs for each load bus in a zone and the hourly integrated load for each load bus in that zone, divided by the sum of the real time hourly integrated loads for each load bus in that same zone. Monitoring Analytics

18 The real time, load weighted, average LMP for an hour for a state is equal to the sum of the product of the hourly integrated bus LMPs for each load bus in a state and the hourly integrated load for each load bus in that state, divided by the sum of the realtime hourly integrated loads for each load bus in that state. The system real time, load weighted, average LMP for a day is equal to the product of the hourly integrated LMPs for each load bus and the hourly integrated load for each load bus, for each hour, summed over every hour of the day, divided by the sum of the hourly integrated bus loads for the system for the day. The zonal real time, load weighted, average LMP for a day is equal to the product of each of the hourly integrated LMPs for each load bus in a zone and the hourly integrated load for each load bus in that zone, for each hour, summed over every hour of the day, divided by the sum of the hourly integrated bus loads at each load bus in that zone for the day. The real time, load weighted, average LMP for a day for a state is equal to the product of each of the hourly integrated LMPs for each load bus in a state and the hourly integrated load for each load bus in that state, for each hour, summed over every hour of the day, divided by the sum of the hourly integrated bus loads at each load bus in that state for the day. The system real time, load weighted, average LMP for a year is equal to the product of the hourly integrated LMPs and hourly integrated load for each load bus, summed across every hour of the year, divided by the sum of the hourly integrated bus loads at each load bus in the system for each hour in the year. The zonal real time load weighted, average LMP for a year is equal to the product of each of the hourly integrated bus LMPs and hourly integrated load for each load bus in a zone, summed across every hour of the year, divided by the sum of the hourly integrated bus loads at each load bus in that zone for each hour in the year. The real time load weighted, average LMP for a year for a state is equal to the product of each of the hourly integrated bus LMPs and hourly integrated load for each load bus in a state, summed across every hour of the year, divided by the sum of the hourly integrated bus loads at each load bus in that state for each hour in the year. Day Ahead The system day ahead, load weighted, average LMP for an hour is equal to the sum of the product of the hourly LMP at each load bus and the corresponding nodal, total dayahead hourly load at each load bus in the system, divided by the sum of the nodal, total day ahead hourly loads across the buses. 8 Monitoring Analytics 20

19 The zonal day ahead, load weighted, average LMP for an hour is equal to the sum of the product of the hourly bus LMPs for each load bus in a zone and the hourly estimated load distribution factors for each load bus in that zone. The zonal day ahead, loadweighted, average LMP does not use the full nodal, total day ahead hourly loads used in the other calculations of day ahead average LMP. The day ahead, load weighted, average LMP for an hour for a state is equal to the sum of the product of the hourly bus LMPs for each load bus in a state and the hourly factor distributed load, from each contributing zone, for each load bus in that state. The state specific day ahead, load weighted, average LMP does not use the full nodal, total dayahead hourly loads used in the other calculations of day ahead average LMP. The system day ahead, load weighted, average LMP for a day is equal to the product of the hourly day ahead LMPs for each load bus and the nodal, total hourly day ahead load for each load bus, for each hour, summed over every hour of the day, divided by the sum of the nodal, total hourly day ahead loads for the system for the day. The zonal day ahead, load weighted, average LMP for a day is equal to the product of each of the hourly day ahead LMPs for each load bus in a zone and the hourly estimated load distribution factors for each load bus in that zone and the hourly day ahead load for the zone, summed over every hour of the day, and divided by the corresponding estimated total zonal load for the day. The zonal day ahead, load weighted, average LMP does not use the full nodal, total day ahead hourly loads used in the other calculations of day ahead average LMP. The day ahead, load weighted, average LMP for a day for a state is equal to the product of each of the hourly day ahead LMPs for each load bus in a state and the hourly factor distributed load, from each contributing zone, for each load bus in that state, summed over every hour of the day, and divided by the corresponding estimated total hourly factor distributed load for the day. The zonal day ahead, load weighted, average LMP does not use the full nodal, total day ahead hourly loads used in the other calculations of day ahead average LMP. The system day ahead, load weighted, average LMP for a year is equal to the product of the hourly LMPs and nodal, total hourly load for each load bus, summed across every hour of the year, divided by the sum of the nodal, total hourly bus loads at each load bus in the system for each hour in the year. The zonal day ahead, load weighted, average LMP for a year is equal to the product of each of the hourly LMPs for each load bus in a zone and the hourly estimated load distribution factors for each load bus in that zone and the hourly day ahead load for the zone, summed over every hour of the year, and divided by the total estimated zonal load for the year. The zonal day ahead, load weighted, average LMP does not use the Monitoring Analytics

20 full nodal, total day ahead hourly loads used in the other calculations of day ahead average LMP. The day ahead, load weighted, average LMP for a year for a state is equal to the product of each of the hourly LMPs for each load bus in a zone and the hourly factor distributed load, from each contributing zone, for each load bus in that state, summed over every hour of the year, and divided by the corresponding estimated total hourly factor distributed load for the year. The zonal day ahead, load weighted, average LMP does not use the full nodal, total day ahead hourly loads used in the other calculations of dayahead average LMP. Equation 0 LMP calculations 20 Monitoring Analytics 20

21 Load Definitions 200 State of the Market Report for PJM: Technical Reference for PJM Markets PJM measures load in a number of ways. The Market Monitoring Unit (MMU) makes use of two basic measures of load in its analysis of the PJM market: peak load and accounting load. In the 200 State of the Market Report for PJM, both measures of load are used, as appropriate for the specific analysis. The measures of load and their applications changed after PJM s June, 2007, implementation of marginal losses. Peak Load PJM uses emtr data for both peak loads and as the basis for accounting loads. emtr data is supplied by PJM electricity distribution companies (EDCs) and generators and is based on the metered MWh values of tie lines and the metered values of generation MWh. For PJM Western Region and Southern Region EDCs (ComEd, AEP, DAY, DLCO, AP and Dominion), emtr load values implicitly include local, EHV (extra high voltage) and non EHV losses. emtr load values for PJM Mid Atlantic Region EDCs implicitly include local and non EHV losses plus an explicit allocation of metered Mid Atlantic Region EHV losses. PJM uses this emtr load data to measure peak loads. This measure of load provides the total amount of generation output and net energy imports required to meet the peak demand on the system. It is not strictly a measure of load, but rather a measure of the output and imports necessary to meet load. Accounting Load PJM uses emtr load data, excluding losses, as accounting load in the settlement process. Prior to June, 2007, accounting load for all EDCs was equal to emtr load and thus included losses. Since the implementation of marginal losses on June, 2007, accounting load without losses is calculated by subtracting State Estimator losses from emtr load and allocating the net amount to load buses based on State Estimator loads. Since June, 2007, accounting load without losses has represented the actual retail customer load and is referred to here as accounting load. Accounting load is used in the 200 State of the Market Report for PJM to measure daily, monthly and annual load. Accounting load is also used in the 200 State of the Market Report for PJM to weight LMP in load weighted LMP calculations. Prior to June, 2007, accounting load included losses and after June accounting load excludes losses. Prior to June, 2007, LMP did not include losses. After June, 2007, LMP includes losses. Marginal Losses On June, 2007, PJM revised its methodology for determining transmission losses from average losses to nodal, marginal losses. Marginal loss pricing is based on the incremental losses that result from an increase in output. Marginal loss pricing is designed to permit more efficient system dispatch and decreased total production cost. Monitoring Analytics

22 Under the new methodology, PJM s locational marginal price (LMP) at a bus i is comprised of three distinct components: system marginal price (SMP), marginal losses component of LMP at bus i (Li) and the congestion component of LMP at bus i (CLMPi). Equation 0 shows the components of LMP at bus i. Equation 0 LMP components LMPi = SMP + Li + CLMPi SMP is calculated at the distributed load reference bus, where the loss and CLMP contribution to LMP are zero. The LMP at bus i is comprised of losses and congestion effects, either positive or negative, that are determined by the bus s location on the system relative to the SMP at the load weighted reference bus. Total, Average and Marginal Losses Total transmission losses are equal the product of the square of the current flowing across the line (I) and the resistance of the line (R). The materials constituting the conductors and other elements of the transmission system exhibit a characteristic impedance to the flow of power. Total transmission losses over a line can also be expressed as the product of the resistance of the line (R) times the square of the power consumed by the load (P), divided by the square of the voltage (V). 8 While this relationship differs somewhat in an alternating current (AC) as compared to a direct current (DC) system, the magnitude of losses can be approximated by the equation: Equation 0 2 Total transmission losses Total Losses = I R= ( P R)/ V Defining a 2 = R/ V and substituting into Equation 0 2 results in: Equation 0 3 Total transmission losses Total Losses = 2 a P Average transmission losses per MW from a given power flow P across a transmission element are: Equation 0 4 Average transmission losses = a P P = a P. Average Losses ( 2 / ) 8 Equation 0 2 incorporates the substitution of the relationship I=P/V, derived from Ohm s Law, for the variable I. 22 Monitoring Analytics 20

23 Marginal transmission losses are the incremental losses resulting from an increase in power flow P across the transmission element and are equal to the first derivative of total losses with respect to power flow P: Equation 0 5 Marginal losses d = a P = 2 a P. dp 2 Marginal Losses ( ) For a given power flow P, the marginal losses for an increase in P are, therefore, equal to twice the average losses for the associated total flow P. Effect of Marginal Losses on LMP The following equations illustrate the effect of marginal losses on least cost dispatch. In this simple example, the least cost dispatch problem involves meeting system load and the losses associated with serving that load. Equation 0 6 defines the total cost of generation (CT), which is a function of generator output (P) of units i through N. Equation 0 6 Total cost of generation C N = [ C ( P) ] T i i i= Equation 0 7 is the power balance constraint, where total injections ( total withdrawals ( P load Equation 0 7 Power Balance Constraint N N Pload + Ploss Pi = Pi i= i= N i P ) must equal ) plus total losses ( P loss ), where losses are a function of ( P ). i Together, Equation 0 6 and Equation 0 7 form a system of equations which can be represented by a Lagrangian (ζ ), as defined in Equation 0 8. Equation 0 8 System N N N ζ( P) = C ( P) + λi ( P + P ( P ) P) i i i i load loss i i i= i i i= i N i Optimizing Equation 0 8 for Pi n results in Equation 0 9 and Equation 0 7: Monitoring Analytics

24 Equation 0 9 Lambda dc =λi dp dp i loss ( ) dp i Equation 0 7Power Balance Constraint (from above) N N Pload + Ploss Pi = Pi i= i= Note, that Equation 0 9 shows that the optimal dispatch of each generator i must account for losses associated with using that unit to meet load. This measure of losses is the marginal loss penalty factor (Pff ) for incremental power from generator i to serve system load: Equation 0 0 Penalty factor Pfi =. Ploss Pi The incremental cost of using output from generator i to meet load includes incremental losses. 9 Ploss The term is called the loss factor and represents the change in system losses for a Pi change in output from generator i to meet load. If an increase in power from generator i results in an incremental increase in losses, then the loss factor is positive: Ploss 0< <, P i and the resultant penalty factor at busi would be greater than one: 9 Note, as presented here, the marginal effect is on total losses, not losses at any particular load bus. 24 Monitoring Analytics 20

25 Pf i = >. P loss Pi Conversely, if an increase in power results in a decrease in losses, then the loss factor is negative: Ploss < < 0, P i and the resultant penalty factor at bus i would be less than one: Pf i = <. P loss Pi The unit offer curve of a generator is multiplied by the respective penalty factor for serving the load (Equation 0 0). To the system operator, seeking to minimize the costs of serving a given level of load, the existence of losses modifies the relative costs of output from the unit relative to the case where losses are not accounted for. If the relevant penalty factor is greater than one, system losses would be made greater by increasing the output of that generator to serve load, and the unit offer curve, from the system operator perspective, would be shifted upward relative to the case where losses were not accounted for. Similarly, if the penalty factor associated with generator i delivering power to load is less than one, system losses associated with serving system load would be reduced by increasing the output of generator i, and the unit offer curve would shift downward relative to the case where losses are not accounted for. These marginal loss related adjustments in relative costs will affect the optimal dispatch, and the resulting LMPs, for any given level of load relative to the case where marginal losses are not accounted for. LMPs at specific load buses will reflect the fact that marginal generators must produce more (or less) energy due to losses to serve that bus than is needed to serve the load weighted reference bus. The LMP at any bus is a function of the SMP, losses and congestion. Relative to the system marginal price (SMP) at the load weighted reference bus, the loss factor can be either positive or negative. Loss Revenue Surplus As demonstrated in Equation 0 5, revenues resulting from marginal losses are approximately twice those collected from average losses. As demonstrated in Equation 0 2, losses are equal to the square of the power, P. As such, two loads of equal size at the same location, served simultaneously, result in losses four times greater than the losses incurred in serving either of them separately. By utilizing the penalty factor in the dispatch, losses are paid based on marginal losses rather than based on average losses. Monitoring Analytics

26 Other than the effect on the optimal dispatch point, LMP at the marginal generator bus, and therefore the payment to the generator, is not affected. By paying for losses based on marginal instead of average losses at the load bus, a revenue over collection occurs. Using the example of two loads, of equal size at the same location, being served simultaneously, the marginal losses associated with the combined effect of the loads are greater than the sum of the losses incurred by each load separately, thus resulting in an over collection. Properly accounting for marginal losses allows for an optimal, least cost solution to the system of equations that make up the market to serve load. Over collection is a direct outcome of marginal cost pricing and not a cause for concern. Prices set on this basis reflect the true incremental cost of serving load at any bus, and provide efficient incremental resource signals. Of concern under these circumstances is what is done with the over collection and how it is distributed among the market participants. These disbursements should be provided to the market participants that pay for the marginal losses in their energy charges, in this case the loads. To maintain an efficient price signal, any reallocation of the excess revenues must not interfere with the price signal at the margin. The solution to this problem generally takes the form of lump sum payments to market participants. The next issue is how to distribute the payments among the loads. To the extent that the causality of total marginal losses related costs are not generally directly attributable to specific load serving entities, the actual allocation methodology used to distribute the lump sum payments, while important from a policy perspective, is more a question of equity than market outcome efficiency. Under these circumstances, where there are common costs attributable to providing a service to a number of parties, it is general accepted practice to allocate the common costs, or benefits, to participants in proportion to their contribution to total load. This is the approach adopted by PJM. Under PJM s tariff, excess total loss related revenues are allocated to transmission users based on load plus export ratio shares: Equation 0 Excess loss revenue allocation Loss Credit = (Total Loss Surplus) i Customer total MWh delivered to load + exports Total PJM MWh delivered to load + exports. 26 Monitoring Analytics 20

27 Calculation and Use of Generator Sensitivity/Unit Participation Factors Sensitivity factors define the impact of each marginal unit on locational marginal price (LMP) at every bus on the system. The availability of sensitivity factor data permits the refinement of analyses in areas where the goal is to calculate the impact of unit characteristics or behavior on LMP. 20 These factors include the impact on LMP of the cost of fuel by type, the cost of emissions allowances by type, frequently mitigated unit adders and unit markup by unit characteristics. 2 Generator sensitivity factors, or unit participation factors (UPFs), are calculated within the least cost, security constrained optimization program. For every five minute system solution, UPFs describe the incremental amount of output that would have to be provided by each of the current set of marginal units to meet the next increment of load at a specified bus while maintaining total system energy balance. A UPF is calculated from each marginal unit to each load bus for every five minute interval. In the absence of marginal losses, the UPFs associated with the set of marginal units in any given interval, for a particular load bus, always sum to.0. UPFs can be either positive or negative. A negative UPF for a unit with respect to a specific load bus indicates that the unit would have to be backed down for the system to meet the incremental load at the load bus. Within the security constrained, least cost dispatch solution for an interval, during which the LMP at the marginal unitʹs bus equals the marginal unitʹs offer, consistent with its output level, LMP at each load bus is equal to each marginal unitʹs offer price, multiplied by its UPF, relative to that load bus. In some cases, the bus price for the marginal unit may not equal the calculated price based on the offer curve of the marginal unit. These differences are the result of the LPA marginal unit offer being overridden with its UDS LMP or ex ante dispatch rate. When overridden, the LPA marginal unit s current offer is replaced by the UDS LMP and this sets the price. The UDS LMP does not reflect the LPA marginal unit s offer curve and does not represent the offer behavior of the marginal units in the LPA whose offers are overridden. The UDS LMP is a result of the marginal units in UDS and reflects the offer curve and behavior of these units. Any difference between the price based on the offer curve and the actual bus price when no override occurs is categorized as dispatch differential. 20 The PJM Market Monitoring Unit (MMU) identified applications for sensitivity factors and began to save sensitivity factors in Before the 2006 State of the Market Report, state of the market reports had shown the impact of each marginal unit on load and on LMP based on engineering estimates whenever there were multiple marginal units. Monitoring Analytics