Balancing Ratio Determination Issue Patrick Bruno Sr. Engineer, Capacity Market Operations Markets Implementation Committee April 4, 2018
Issue Recap 1. The current rules set the default Market Seller Offer Cap ( MSOC ) for Capacity Performance ( CP ) Resources equal to Net CONE times the average historical Balancing Ratios experienced during Performance Assessment Intervals in the three calendar years that immediately precede the Base Residual Auction ( BRA ) for the Delivery Year Average historical Balancing Ratio becomes indeterminable when no Performance Assessment Intervals have occurred during the prior three calendar years If determinable, may not be in time to inform the unit-specific MSOC submission deadline 120 days prior to the BRA (mid-january) 2. The CP Non-Performance Charge Rate currently uses an assumed 30 Performance Assessment Hours for the Delivery Year 30 hour assumption should be reviewed; No emergency actions triggering Performance Assessments Hours/Intervals have occurred since CP implementation 2
Key Work Activities 1. Provide education on the calculation of the MSOC and Balancing Ratio 2. Provide education on the determination of Non-Performance Charge Rates 3. Develop and discuss alternative Balancing Ratio calculation methodologies for use in the determination of the default MSOC 4. Develop and discuss alternative methods to determine the Non- Performance Charge Rate 3
Timeline Education & Interests Design Components & Solutions Packages & MIC Endorsement MRC/MC and FERC Filing (tariff changes) Feb-Mar MIC Mar-May MIC Jun-Jul MIC Jul-Aug MRC (Sept MC) File endorsed changes with FERC by early October 2018 4
MSOC Balancing Ratio Solution Option A
Solution Description To estimate an expected future average Balancing Ratio for use in the default MSOC Take the average Balancing Ratios during the three Delivery Years that immediately precede the BRA using: a) actual Balancing Ratios calculated during RTO PAIs of the Delivery Year, and b) for any Delivery Year with less than H clock hours of PAIs, estimated Balancing Ratios calculated during the peak load hours of the RTO that do not overlap a PAI H represents expected number of hours of PAIs in the DY (currently 30) CP Default MSOC = Net CONE x estimated Balancing Ratio 6
Example of Delivery Year with less than H Clock Hours (30) of PAIs Hour Peak Hourly Avg Date HE PAIs Count Hour Bal Ratio 1 Jul 18 14 8 Y 93.4% 2 Jul 18 15 12 Y 93.7% 3 Jul 18 16 12 Y 95.2% 4 Jul 18 17 12 Y 95.1% 5 Jul 18 18 4 Y 90.8% 6 Aug 2 15 12 Y 89.5% 7 Aug 2 16 12 Y 90.9% 8 Jan 11 7 4-83.4% 9 Jan 11 8 12 Y 84.2% 10 Jan 11 17 6 Y 84.3% 11 Jan 11 18 12-76.7% 12 Jan 11 19 12-78.5% 13 Jul 18 13 - Y 93.1% 14 Jul 19 16 - Y 92.8% 15 Jul 19 17 - Y 92.5% 16-30 (a) 12 hourly average Balancing Ratios from actual PAIs (118 in total) (b) 18 hourly estimated Balancing Ratios during RTO peak hours that do not overlap a PAI Balancing Ratio for the DY equals average of both (a) and (b) 7
Solution Pros 1. Straight-forward solution that augments the existing methodology by providing reasonable proxy hours and Balancing Ratios to use when no, or relatively few, actual PAIs occur Peak load hours used as reasonable proxies due to correlation of high load hours and PAI triggers 2. Resultant Balancing Ratios appear on par with the values calculated from actual data during historical RTO emergency actions 3. Determinable in time to inform the unit-specific offer cap submission deadline for documentation 120 days prior to the BRA (mid-january) 8
Comparison of Balancing Ratios under Existing and Proposed Methodologies Delivery Year Existing Proposed Prior 3 DYs 2018/2019 85.0% 88.3% 11/12, 12/13, 13/14 2019/2020 81.0% 85.3% 12/13, 13/14, 14/15 2020/2021 78.5% 83.8% 13/14, 14/15, 15/16 2021/2022 78.5% * 86.8% 14/15, 15/16, 16/17 Balancing Ratios during historical RTO emergency actions from 2011-14 Summer (16 hours): Avg = 93.5% Min = 87.7% Max = 95.1% Winter (26 hours): Avg = 78.3% Min = 71.5% Max = 84.9% 9
Assumed Performance Assessment Hours H in the Non-Performance Charge Rate
Non-Performance Charge Rate Non-Perf. Charge Rate* = Net CONE x 365 days / H (30 hours) Where: Net CONE is the Net Cost of New Entry (stated in $/MW-Day, ICAP terms) for the relevant Delivery Year and LDA in which the resource is modeled H or 30 hours is the current estimated number of Performance Assessment Hours that may occur in a Delivery Year Non-Performance Charge Rate is expressed in $/MWh to be multiplied by a unit s Performance Shortfall to calculate the assessed penalty charges * Charge Rate does not reflect the filed change with 5-minute Settlements, which further divides the rate by the number of Real-Time Settlement Intervals in an hour 11
Historical RTO Performance Assessment Hours Hours 35 30 25 20 15 10 5 0 2011/2012 2012/2013 2013/2014 2014/2015 2015/2016 2016/2017 2017/2018 * Delivery Year CP Transition Delivery Years Note: Hours shown prior to 2016/2017 reflect Emergency Actions that would have triggered a Performance Assessment Hour under the CP rules 12
GE MARS Preliminary Study to Estimate H GE MARS is a planning software tool capable of calculating standard reliability indices for a given power system (e.g. daily and hourly LOLE) The tool also allows for review of emergency operating procedures, by calculating the expected number of days per year at a specified margin e.g. A margin set at the typical Primary Reserve requirement might be used to estimate the number of Primary Reserve Warnings The tool uses a sequential Monte Carlo simulation to calculate the probability of events, and requires a fair number of inputs and assumptions to run 13
GE MARS Study Assumptions 1. Same generator supply used in IRM Study Operating histories randomly generated with each Monte Carlo replication for all units (reflects unit-specific forced outages rates) Total Available Capacity determined for each hour 2. Solved peak load from IRM Study at reserve requirement Monthly load shape using forecasted monthly peak loads; daily and hourly loads determined from an historical typical load shape Hourly load levels varied in MARS simulations based on 7 load uncertainty levels, each with an associated probability 3. Specified Margin based on dispatch of Pre-Emergency DR Estimated DR (8200 MW) Operating Reserves/Regulation (3400 MW) 14
GE MARS Study Results (1,000 replications run at each load level) PAHs per year 22 20 21/22 Target IRM 18 16 14 12 10 8 6 4 2 0 13.8% 15.8% 17.8% 19.8% 21.8% 23.8% 25.8% 27.8% 29.8% IRM 15
GE MARS Study Results / Observations H significantly varies at different assumed reserve levels for the future DY IRM of 15.8%: ~ 15 Hours IRM of 21.8%: ~ 2 Hours Virtually no Performance Assessment Hours occurred in winter months of the preliminary analysis; almost all risk and emergency hours in summer months Balancing Ratios calculated during the triggered Performance Assessment Hours of the program around 95 to 96 percent on average 16
GE MARS Study Conclusions H in the Non-Performance Charge Rate should reflect the expected PAHs at the target IRM Consistent with using Net CONE in the numerator, as both represent the longterm market at equilibrium Consistent with CP design that aims to discourage non-performing resources from taking on capacity obligations due to penalties offsetting capacity revenues, especially when new entry is needed Recommend using an H between 15 and 30 hours in denominator of the Non- Performance Charge Rate 15 hours seen at target IRM in GE MARS Study for just summer months 30 hours seen historically (i.e. 13/14 DY, even with high reserve margin) 17
Appendix - Prior Education
CP Default Market Seller Offer Cap CP default MSOC = Net CONE x Balancing Ratio (B ) Where: Net CONE is the Net Cost of New Entry (stated in $/MW-Day, ICAP terms) for the relevant Delivery Year and zone in which the resource is located Balancing Ratio (B ) is the historical average of the Balancing Ratios experienced during Performance Assessment Intervals/Hours in the three most recent calendar years preceding the Base Residual Auction for such Delivery Year Represents the expected Balancing Ratio across all Performance Assessment Intervals/Hours for a Delivery Year CP default MSOC is expressed in $/MW-Day 19
Historical RTO Performance Assessment Hours Hours 35 30 25 20 15 10 5 0 2011/2012 2012/2013 2013/2014 2014/2015 2015/2016 2016/2017 2017/2018 * Delivery Year CP Transition Delivery Years Note: Hours shown prior to 2016/2017 reflect Emergency Actions that would have triggered a Performance Assessment Hour under the CP rules 20
Historical MSOC Balancing Ratios Average Balancing Ratios calculated for use in the default MSOC by Delivery Year Delivery Year MSOC Balancing Ratio 2018/2019 85.0% 2019/2020 81.0% 2020/2021 78.5% 2021/2022 78.5% * * 2021/2022 Balancing Ratio in the MSOC set to the same value as prior Delivery Year due to absence of Performance Assessment Hours in prior three calendar years (2015-2017), as approved by FERC A list of the underlying Performance Assessment Hours and corresponding Balancing Ratios used to determine the above averages are included in Appendix 2 of PJM s response to FERC on April 10, 2015 in Docket No. ER15-623-001 21
Balancing Ratio in Performance Assessment Intervals The calculated Balancing Ratio for a Performance Assessment Interval represents the percentage share of total generation capacity commitments needed to support the load and reserves on the system within the Emergency Action Area during the interval i.e. (Load + Reserves) / Generation Capacity Commitments The Balancing Ratio is used to set the Expected Performance level of Generation Capacity Performance Resources within the Emergency Action Area during the Performance Assessment Interval Expected Performance = Capacity Commitment (UCAP) x Balancing Ratio 22
Non-Performance Charge Rate Non-Perf. Charge Rate* = Net CONE x 365 days / 30 hours Where: Net CONE is the Net Cost of New Entry (stated in $/MW-Day, ICAP terms) for the relevant Delivery Year and LDA in which the resource is modeled 30 hours is the estimated number of Performance Assessment Hours that may occur in a Delivery Year Based on Emergency Action hours seen during 2013/2014 Non-Performance Charge Rate is expressed in $/MWh to be multiplied by a unit s Performance Shortfall to calculate the assessed penalty charges * Charge Rate does not reflect the filed change with 5-minute Settlements, which further divides the rate by the number of Real-Time Settlement Intervals in an hour 23
Non-Performance Charge Rates LDA 18/19 Non-Performance Charge Rate ($/MWh) 19/20 Non-Performance Charge Rate ($/MWh) 20/21 Non-Performance Charge Rate ($/MWh) RTO $3,424.80 $3,401.17 $3,329.31 MAAC $3,095.44 $2,977.55 $2,868.54 EMAAC $3,245.22 $3,223.07 $3,217.35 SWMAAC $2,770.72 $2,612.79 $2,300.60 PSEG $3,395.35 $3,446.56 $3,488.06 PS-NORTH $3,395.35 $3,446.56 $3,488.06 DPL-SOUTH $2,943.36 $2,980.31 $2,897.73 PEPCO $2,856.98 $2,775.37 $2,574.50 ATSI $3,096.05 $3,000.64 $2,968.21 ATSI-CLEVELAND $3,096.05 $3,000.64 $2,968.21 COMED $3,649.39 $3,732.33 $3,748.21 BGE $2,684.33 $2,450.29 $2,026.74 PPL $3,244.97 $3,156.12 $3,038.16 DAYTON $3,104.21 DEOK $3,210.14 24
Annual Stop-Loss of Non-Performance Charges Stop-Loss = Net CONE x 365 days x 1.5 x Committed MW Where: Net CONE is the Net Cost of New Entry (stated in $/MW-Day, ICAP terms) for the relevant Delivery Year and modeled LDA in which the resource resides Committed MW is the resource s capacity commitment in UCAP Based on the maximum clearing price allowed by the VRR curve at Net CONE times 1.5 At 30 assumed Performance Assessment Hours in the Non- Performance Charge Rate, a resource will hit the stop-loss after 45 hours of zero Actual Performance 25
CP Default MSOC Rationale
CP Default MSOC The default MSOC reflects the amount that a competitive resource with low net going forward costs (Low ACR Resource) would accept in the capacity market A Low ACR Resource is one whose net avoidable costs are less than its total expected Bonus Performance payments as an energy-only resource Represents the lost opportunity costs incurred by taking on a capacity obligation and foregoing some expected Bonus Performance payments The Balancing Ratio (B ) is a component of the default MSOC calculation to reflect the percentage share of expected Bonus Performance payments that are foregone by taking on a capacity obligation A resource will receive Bonus Payments for its production that exceeds the Balancing Ratio share of its capacity obligation during Performance Assessment Intervals/Hours regardless of it having a capacity obligation Note: A resource with high net going forward costs that exceed expected Bonus Performance payments can go through the resource-specific MSOC process for a higher CP offer cap 27
CP Default MSOC Example Capacity Resource Energy-Only Nameplate (MW) 100 100 Capacity Obligation (UCAP MW) 100 0 Net CONE ($/MW-day) $250 $250 Balancing Ratio (B') 0.9 0.9 Actual Performance (A') 100 100 Expected Performance (MW) 90 - Bonus Performance (MW) 10 100 Bonus Rate ($/MWh) $3,042 $3,042 Bonus Performance Hours 30 30 Annual Bonus Performance ($/year) $912,500 $9,125,000 Foregone Bonus Performance ($/year) $8,212,500 - Lost Opportunity Cost ($/MW-day) $225 - Default MSOC of Net CONE x B' ($/MW-day) $225-28
CP Competitive Offer p = PPR x H x B + max{0, (ACR PPR x H x A )} Where: p: Offer price in RPM on a UCAP basis ($/MW-year) PPR: Non-Performance Charge Rate ($/MWh) Assumed to be equivalent to the Bonus Performance Rate H: Expected number of Performance Assessment Hours in the year (hours/year) B : Expected value of balancing ratio across all Performance Assessment Hours in year ACR: Net ACR (net going forward costs) for a resource ($/MW-year) A : Expected value of availability across all Performance Assessment Hours in year Note: The full overview and explanation of the Capacity Performance Offer Cap Logic can be found in Appendix 1 of PJM s April 10, 2015 response to FERC in Docket No. ER15-623-001 29
CP Competitive Offer for Low ACR Resource Low ACR Resource is one whose net avoidable costs are less than its total expected Bonus Performance payments as an energy-only resource Second term of competitive offer drops to zero PPR substituted with Non-Performance Charge Rate p ($/MW-year) = PPR x H x B + max{0, (ACR PPR x H x A )} p ($/MW-year) = (Net CONE x 365 / H) x H x B p ($/MW-year) = Net CONE x 365 x B p ($/MW-day) = Net CONE x B CP default MSOC 30
CP Competitive Offer for High ACR Resource High ACR Resource is one whose net avoidable costs are greater than its total expected Bonus Performance payments as an energy-only resource Second term of competitive offer remains greater than zero PPR substituted with Non-Performance Charge Rate Competitive offer dependent on unit-specific ACR and expected resource performance compared to B, requiring a unit-specific review of its MSOC An appropriate unit-specific risk premium may also be included in the unit-specific review p ($/MW-year) p ($/MW-year) p ($/MW-year) p ($/MW-day) = PPR x H x B + (ACR - PPR x H x A ) = ACR + PPR x H x (B - A ) = ACR + (Net CONE x 365 / H) x H x (B - A ) = ACR + Net CONE x (B - A ) 31