Local Market Power Mitigation Enhancements Draft Final Proposal and - PowerPoint PPT Presentation

Local Market Power Mitigation Enhancements Draft Final Proposal and Retrospective Analysis May 13, 2011 Cynthia Hinman, Sr. Market Design and Policy Specialist Lin Xu, Sr. Market Development Engineer

Agenda for today’s meeting Estimated Topic Presenter Time 10:00 – 10:10 Introduction Chris Kirsten 10:10 – 10:40 Review Draft Final Proposal Cynthia Hinman 10:40 – 12:00 Retrospective Analysis Discussion Lin Xu 12:00 – 1:00 Lunch 1:00 – 2:30 Inline Competitive Path Jeff McDonald Assessment 2:30 – 2:45 Next Steps Chris Kirsten Page 2

ISO Policy Initiative Stakeholder Process POLICY AND PLAN DEVELOPMENT Issue Straw Draft Final Board Paper Proposal Proposal Stakeholder Input We are here

The LMPM proposal seeks to: • Meet FERC requirement to use bid-in demand • Incorporate elements of: – Convergence bidding – Proxy demand resource • Improve accuracy of mitigation in real-time market • Incorporate inline competitive path designation Page 4

LMP Decomposition process Note: Virtual bids are not mitigated. Perform all constraints (AC) run Dynamic CPA Decompose LMP for each location < 0 no market power concerns Evaluate non- Default Energy competitive Bid component >0 indicates market power; mitigate to either: Competitive LMP Run IFM with (if it is lower than the mitigated bids unmitigated bid) Page 5

The methodology for the decomposition method is: For location i : EC + LMP i LC + LMP i CC + LMP i LMP i = LMP i NC Where: EC = the energy component LC = the loss component CC = the competitive constraints congestion component NC = the non-competitive constraints congestion component Page 6

The congestion component contains two elements. Congestion Component For location i : EC + LMP i LC + LMP i CC + LMP i LMP i = LMP i NC Where: EC = the energy component LC = the loss component CC = the competitive constraints congestion component NC = the non-competitive constraints congestion component Page 7

Each element of the LMP congestion component is calculated in the following way: • Competitive constraint component = the sum of the shift factor times the shadow price for all competitive constraints • Non-competitive constraint component = the sum of the shift factor times the shadow price for all non-competitive constraints Page 8

The competitive LMP is analogous to the LMP produced in the current CC run. Competitive For location i : LMP EC + LMP i LC + LMP i CC + LMP i LMP i = LMP i NC Where: EC = the energy component LC = the loss component CC = the competitive constraints congestion component NC = the non-competitive constraints congestion component Page 9

RMR Condition 1 & Condition 2 Mitigation • Concern - Use of bid-in demand and virtual bids could cause over or under commitment of RMR resources • Proposed solution – – Condition 1 units - market bids will be utilized in the AC run and RMR proxy bids will be used in place of DEBs. The same LMP decomposition will be used to determine when RMR proxy bids replace market bids. – Condition 2 units – ISO operators will manually dispatch these resources if needed and RMR proxy bids will be utilized in the market. For 2011, there is only one RMR contracted resource. Page 10

Questions: Cynthia Hinman chinman@caiso.com 916-608-7060 Submit comments to: LMPM@caiso.com

A Retrospective Analysis of LMPM enhancements Lin Xu, PhD Sr. Market Development Engineer

An analysis road map LMPM CPA DMM Current LMPM Current LMPM Current CPA Dynamic CPA (RSI) CAISO CAISO perform study New LMPM New LMPM Current CPA Dynamic CPA (RSI) Pending DMM to provide RSI We are here Page 13

Analysis • Apply new LMPM method on actual historical data – All-constraint run – The same seasonal CPA as today • Study period – 57 days from day-ahead market in February and March 2011 • Mitigation reference bus choice – Midway or Vincent 500KV bus vs load distributed slack bus • Mitigation threshold – LMP i NC > thres > 0 Page 14

Binding non-competitive constraints • Total of 175 hours – Exactly one non-competitive constraint for every hour Constraint Type Congested Hours SDGE_PCT_UF_IMP_BG Flowgate 109 SLIC 1417897_IV_CB_7022_OUT_NG Nomogram 15 36957_MCSN TP1_230_36961_MOCCASIN_230_BR_1 _1 Flowgate 13 32228_PLACER _115_32238_BELL PGE_115_BR_1 _1 Flowgate 10 SLIC 1446790 EGL_SLV_FLTN SOL-1 Nomogram 9 SLIC 1368530_SDGE_IV_CB_7022 Nomogram 6 SSONGS_BG Flowgate 6 SLIC 1434491_Moorpark_Pardee_NG Nomogram 5 22716_SANLUSRY_230_24131_S.ONOFRE_230_BR_3 _1 Flowgate 2 Page 15

Performance of current LMPM Non-competitive constraint Current mitigation binding 175 hours 644 hours (New mitigation hours) 12 hours 163 hours 481 hours False positive hours For example, mitigate PG&E unit Note: These are not false when congestion positives for the CPA. happen in SDGE These LMPM false positives might be caused by modeling differences between the CC run and the AC run. Page 16

New LMPM vs Current LMPM within these 175 hours with binding non-competitive constraints in AC run severe: >$200 above DEB false negative moderate: <=$200 above DEB subset of the economic withholding units severe economic withholding 1.5 units moderate economic withholding 5.8 units below competitive 0.1 units false positive Total 35.6 units below DEB 0.8 units 11.8 units self schedule 11.6 units overlapping identification unavailable 0.8 units 4.8 units New LMPM flagged Current LMPM flagged 35.6 units 1.6 units Page 17

Mitigation reference bus • Mitigation reference bus – Only used in the LMP decomposition in the LMPM process – No change to market optimization • Choices – Option MV: Midway 500KV bus if path 26 power flow from North to South, Vincent 500 KV bus otherwise – Option DS: load distributed slack bus • In the study period, for every hour and every unit, NC than option DS – Option MV always has a higher LMP i – Option MV always flags more units than option DS • On average, option DS has been impacted by market power by $1.42/MWh compared with option MV – Inflated price in a local area will be aggregated in the load distributed slack bus LMP according to load distribution factor • Option MV is better than option DS for LMPM purpose Page 18

Mitigation threshold impact Mitigated units vs mitigation threshold 40.0 35.0 Average numer of flagged units 30.0 25.0 unavailable self schedule 20.0 below DEB 15.0 below competitive moderate economic withholding 10.0 severe economic withholding 5.0 0.0 0 1 2 3 4 5 6 7 8 9 10 NC > thres LMP i Mitigation threshold ($/MWh) Page 19

LMP decomposition vs individual shift factor test • With zero mitigation threshold LMP i NC > 0 – Produce different results only when multiple looped non- competitive constraints binding simultaneously, produce exactly the same results otherwise – Produced exactly the same results in the study period NC > thres • With positive mitigation threshold LMP i – Produce different results • Advantages of LMP decomposition over shift factor test – Loop flow effect – Less concern of over-mitigation – Provide competitive LMP protection floor – Can accommodate significance test with positive threshold Do we want to mitigate if the non-competitive LMP component is trivial? Page 20

Summary • Compared new LMPM method and current LMPM method based on actual day-ahead market data • Demonstrated that the new LMPM is able to flag potential market power more accurately than the current LMPM • Proved the Midway or Vincent 500KV bus option is a better choice of mitigation reference bus than load distributed slack bus • Analyzed sensitivity with respect to positive mitigation threshold • New LMPM is fully compatible with dynamic CPA Page 21

Dynamic Competitive Path Assessment Jeffrey McDonald, Ph.D. Manager, Analysis and Mitigation

Summary of proposal for dynamic path assessment. • Dynamic assessment performed before market runs – After all-constraint run in IFM – After all-constraint run in HASP (hourly inter-tie market) – After ancillary service run before RTD (5-minute market) • Use three pivotal supplier test to determine competitiveness for each potentially binding constraint. – All potentially binding constraints will be tested each market run. – Designations will be “Competitive” unless test is failed. • Assessment includes current market and grid conditions. – Resource and transmission availability. – Test what is likely to bind, not what has historically bound. Slide 23

Specific elements of proposal • Assessment run – Day-ahead all constraint run (hourly) – Hour-ahead scheduling process all constraint run (hourly) – Real-time ancillary service procurement run (each 15-min) • Test for competitiveness – Test only binding constraints for binding interval – Pivotal supplier test to use three supplier residual supply index: RSI(3) < 1  non-competitive – Default designation is competitive – Will account for • Resource ramp capability including A/S procurement • Tolling contracts • Current resource and transmission derates • Convergence bids (cleared on counterflow side) Page 24