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 Time Topic Presenter 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 Assessment Jeff McDonald 2:30 – 2:45 Next Steps Chris Kirsten

Page 2

ISO Policy Initiative Stakeholder Process

POLICY AND PLAN DEVELOPMENT

Issue Paper

Board

Stakeholder Input

We are here

Straw Proposal Draft Final Proposal

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

Perform all constraints (AC) run Decompose LMP for each location Evaluate non- competitive component < 0 no market power concerns >0 indicates market power; mitigate to either: Default Energy Bid

Competitive LMP

(if it is lower than the unmitigated bid)

Run IFM with mitigated bids

LMP Decomposition process

Page 5

Note: Virtual bids are not mitigated. Dynamic CPA

The methodology for the decomposition method is:

For location i: LMPi = LMPi

EC + LMPi LC + LMPi CC + LMPi 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

For location i: LMPi = LMPi

EC + LMPi LC + LMPi CC + LMPi NC

Where: EC = the energy component LC = the loss component CC = the competitive constraints congestion component NC = the non-competitive constraints congestion component

The congestion component contains two elements.

Page 7

Congestion Component

Each element of the LMP congestion component is calculated in the following way:

Page 8

• 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

The competitive LMP is analogous to the LMP produced in the current CC run.

For location i: LMPi = LMPi

EC + LMPi LC + LMPi CC + LMPi 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

Competitive LMP

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.

Page 10

For 2011, there is only one RMR contracted resource.

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

A Retrospective Analysis

• f LMPM enhancements

Lin Xu, PhD

• Sr. Market Development Engineer

Current LMPM Current CPA New LMPM Current CPA Current LMPM Dynamic CPA (RSI) New LMPM Dynamic CPA (RSI) DMM CAISO CAISO perform study Pending DMM to provide RSI We are here

An analysis road map

LMPM CPA

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

– LMPi

NC > thres > 0

Page 14

Binding non-competitive constraints

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

• Total of 175 hours

– Exactly one non-competitive constraint for every hour

Page 15

Page 16

163 hours 12 hours 481 hours Current mitigation 644 hours Non-competitive constraint binding 175 hours (New mitigation hours) False positive hours

Performance of current LMPM

For example, mitigate PG&E unit when congestion happen in SDGE Note: These are not false positives for the CPA. These LMPM false positives might be caused by modeling differences between the CC run and the AC run.

severe economic withholding moderate economic withholding below competitive below DEB self schedule unavailable

5.8 units 0.1 units 11.8 units 11.6 units 4.8 units

New LMPM vs Current LMPM within these 175 hours with binding non-competitive constraints in AC run

Total 35.6 units

New LMPM flagged 35.6 units Current LMPM flagged 1.6 units

false positive 0.8 units

• verlapping identification

0.8 units false negative subset of the economic withholding units 1.5 units severe: >$200 above DEB moderate: <=$200 above DEB

Page 17

Mitigation reference bus

Page 18

• 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,

– Option MV always has a higher LMPi

NC than option DS

– 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 19

0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 1 2 3 4 5 6 7 8 9 10 Average numer of flagged units Mitigation threshold ($/MWh)

Mitigated units vs mitigation threshold

unavailable self schedule below DEB below competitive moderate economic withholding severe economic withholding

Mitigation threshold impact

LMPi

NC > thres

LMP decomposition vs individual shift factor test

• With zero mitigation threshold LMPi

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

• With positive mitigation threshold LMPi

NC > thres

– 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

Page 20

Do we want to mitigate if the non-competitive LMP component is trivial?

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

Specific elements of proposal (cont.)

• Default designations (failure of assessment run)

– If uncompetitive in past seven days, then default is uncompetitive.

• Position on bid mitigation

– Day-ahead bid mitigation straight forward – HASP: All resources that fail LMP decomposition trigger mitigated for HASP, short-term unit commitment, and ancillary service runs. – RTPD: Full reevaluation for RTD. Mitigation applies to balance of hour.

Page 25

Timing and execution of dynamic CPA (real time)

• Highlighted blue hour is actual operation or trade hour.
• Mitigation must cover intertie dispatch (HASP), short-

term unit commitment, and internal 5-minute dispatch.\

• Several opportunities for assessment and mitigation.

Slide 26

1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 RTD 1 2 3 4 1 2 3 4 1 2 3 4 RTPD

* *

HASP 1:00 HE 11 HE 12 HE 13 10:00 11:00 12:00 RTPD at 11:37 and RTD at 11:52, 11:57, and 12:02 CPA and LMPM performed in RTPD enables RTD to use mitigated bids for RTD 12:00 - 12:15. RTPD is re-run every 15 minutes effective for the next three 5minute RTD intervals HASP at 10:47 and STUC at 11:07 CPA and LMPM performed in HASP enables STUC to commit units based on mitigated bids for hours 1:00pm - 4:00pm. RTPD at 11:22 Performs unit commitment using mitigated bids from previous the HASP run for hours 12:00-1:00pm.

The pivotal supplier test for competitiveness – residual supply index.

• Use the residual supply index (RSI) to determine if there

is competitive (effective) supply of counter-flow.

• Remove effective supply of up to three largest net

suppliers for each binding constraint.

• If RSI ≥ 1 then constraint is competitive, otherwise it is

uncompetitive.

• Pass tested competitive / uncompetitive designations to

LMPM process for potential mitigation.

Slide 27

Calculating the RSI

• The residual supply index for a constraint is the ratio of the

effective supply of counter-flow without selected suppliers to the demand for counter-flow.

• Effective supply:

– Only remove effective supply of net suppliers. – Accounts for current availability and effectiveness. – Accounts for tolling agreements. – Includes virtual bids cleared.

• Demand for counter-flow:

– Measured using dispatch of effective resources from pre- market run.

• Formulas for calculating the RSI will be updated in the next

release of DMM paper.

Slide 28

RSI calculations for 2010 competitive paths and untested non-competitive paths (IFM and real time)

• Day-ahead

– Few instances of non-candidate and competitive paths with an RSI<1  More effective counter flow available in day-ahead

• Real time

– More frequently congested competitive paths have RSI > 1 in IFM but show frequent RSI < 1 in real time. – Non-candidate paths in real time tend to be either competitive or non-competitive nearly 100% of the congested hours.

Slide 29

20 40 60 80 100 120

MARBLE_BG IID-SCE_BG 24074_LA FRESA_230_24065_HINSON … 30060_MIDWAY _500_24156_VINCENT … 32218_DRUM _115_32244_BRNSWKT2_115_B… SYLMAR-AC_BG 22708_SANLUSRY_69.0_22712_SA NLUSRY_138_XF_3 24601_VICTOR _230_24085_LUGO _230_BR_1 … 32232_HIGGINS _115_32238_BELL PGE_115_BR_1 _1 LOSBANOSNORTH_BG 22100_CALAVRTP_138_22760_SH ADOWR _138_BR_1 _1 24807_MIRAGE _115_24819_CONCHO … 33047_CC JCT1 _115_33045_FIBRJCT1_115_BR_… 22192_DOUBLTTP_138_22300_FRI ARS _138_BR_1 _1

Hours of Congestion in IFM

RSI1 < 1 RSI2 < 1 RSI3 < 1 RSI3 > 1 * Excludes paths with < 10 hours of IFM congestion Non-candidate paths

Slide 30

50 100 150 200 250 300

33912_SPRNG GJ_115_33914_MI- WUK _115_BR_1 _1 SCE_PCT_IMP_BG HUMBOLDT_BG 33203_MISSON _115_33204_POTRERO … SDGEIMP_BG 33200_LARKIN _115_33204_POTRERO … 33206_BAYSHOR1_115_33208_MA RTIN C_115_BR_1 _1 IVALLYBANK_XFBG SDGE_CFEIMP_BG 33205_HNTRS PT_115_33208_MARTIN … 33207_BAYSHOR2_115_33208_MA RTIN C_115_BR_2 _1 33205_HNTRS PT_115_33208_MARTIN …

Hours of Congestion in IFM

RSI1 < 1 RSI2 < 1 RSI3 < 1 RSI3 > 1 Competitive paths

Frequency of RSI < 1 for IFM

10 20 30 40 50 60

24074_LA … 31464_COTWDPGE_115_31463_… 24807_MIRAGE … SOUTHLUGO_RV_BG 30060_MIDWAY … 32232_HIGGINS … LOSBANOSNORTH_BG 32290_OLIVH J1_115_32214_RIO … 24601_VICTOR … NSONGS_BG 33047_CC JCT1 … 32218_DRUM … 36266_SNTA … 32212_E.NICOLS_115_32214_RI… 30261_BELDENTP_230_30300_T… NOB_BG 30487_ELECTRA … SSONGS_BG 34724_KRN OL … 34778_FELLOWS … 22192_DOUBLTTP_138_22300_F… 24156_VINCENT … 24156_VINCENT …

Hours of Congestion in RTD

RSI1 < 1 RSI2 < 1 RSI3 < 1 RSI3 > 1

50 100 150 200

33912_SPRNG GJ_115_33914_MI- WUK _115_BR_1 _1 SCE_PCT_IMP_BG SDGEIMP_BG IVALLYBANK_XFBG SDGE_CFEIMP_BG 33203_MISSON _115_33204_POTRERO _115_BR_1 _1 31000_HUMBOLDT_115_31001_HMBL T TM_ 1.0_XF_1 HUMBOLDT_BG 33200_LARKIN _115_33204_POTRERO _115_BR_2 _1 31000_HUMBOLDT_115_31452_TRINI TY _115_BR_1 _1 33206_BAYSHOR1_115_33208_MARTI N C_115_BR_1 _1 33205_HNTRS PT_115_33208_MARTIN … 33205_HNTRS PT_115_33208_MARTIN …

Hours of Congestion in RTD

RSI1 < 1 RSI2 < 1 RSI3 < 1 RSI3 > 1 Non-candidate paths Competitive Paths

Frequency of RSI < 1 for Real Time

Next Steps

May 23 Comments to LMPM@caiso.com June 29, 30 Board of Governors

Page 32