Up-to-Congestion Transactions in PJM

Transcription

1 Up-to-Congestion Transactions in PJM Adam Keech

2 Creation of Bid Type June 1, 2000 UTCs first became available in PJM coincident with the implementation of the DA Market Originally used to allow participants to specify the amount explicit congestion they were willing to pay Intention was that if they did not clear in DA they would take their chances with the transaction in real-time Used as a financial hedge for a physical transaction Required a transmission service reservation for DA scheduling 2

3 Marginal Losses June 1, 2007 PJM implemented marginal losses Due to the transmission service reservation required for UTCs, they receive a share of the loss overcollection 3

4 3/1/2008 Increased Spread but Reduced Points Max spread for up-to congestion transactions increased to +/- $50/MWh but limited valid source/sinks Result of a tradeoff during the Reserve Market Working Group process Parties interested wanted larger price spreads than the +/- $25/MWh that were currently available Increased bids were impacting PJM s software due to infeasible transactions and general volume 4

5 Loss Over-collection Issues September 17, 2010 Transmission service reservation requirement for UTCs is removed Bidding activity for some UTCs is geared towards taking no financial position, just to receive a portion of the loss allocation More info ( Removal of transmission service reservation eliminated the marginal loss over-collection allocation to UTCs Made product purely virtual 5

6 6

7 7

8 UTC vs. Inc / Dec Gross Profit Per Day Comparison May 1, 2012 April 30, 2013 UTC (Gross) Inc/Dec (Gross) Delta (Inc/Dec UTC) Total Cleared (MWh) 367,504, ,297, ,206,623.9 Average Daily Profit ($/day) $364, $188, $175, Avg Profit/Cleared MWh ($/MWh) $ $ $ Standard Deviation of Daily Gross Profit $681, $1,356, $675, % of Days Where Bid Type Has Negative Gross Profit 28.8% 51.5% 22.7% 8

9 Observations Over the past year, on-average Cleared UTC MWh are greater than cleared Inc/Dec bids by factor of 3:1 Each cleared Inc/Dec bid makes 36% more money than a cleared UTC Inc/Dec gross profits are more volatile than UTCs Per cleared MWh, Incs/Decs are a higher risk, higher reward, product compared to UTCs UTC volume appears to be higher because they do not incur OR charges 9

10 Day Ahead Market Studies PJM selected 5 recent Day Ahead Market cases and reran the cases without UTCs Goal of analysis: Determine whether or not UTC improve price convergence between DA and RT Initial intent was also to analyze revenue adequacy/inadequacy with and without UTCs To many other factors can influence this to be able to tie this to a particular bid type with any confidence 10

11 Price Convergence To measure price convergence PJM did the following Calculate the average daily price for each hub and zone for each scenario (unweighted average) Calculate the difference between each daily LMP per hub and zone Average all of the differences to determine a daily difference (unweighted average) 11

12 Price Convergence Statistics Total LMP UTCs appear to be beneficial in contributing to Day-ahead vs Real-time price convergence Date RT DA (w/ UTCs) ($/MWh) RT DA (w/o UTCs) ($/MWh) Improvement with UTCs included ($/MWh) May 2, May 4, May 22, May 23, May 27,

13 Price Convergence Statistics Congestion Component UTCs appear to improve Day-ahead vs Real-time congestion price convergence Date RT DA (w/ UTCs) ($/MWh) RT DA (w/o UTCs) ($/MWh) Improvement with UTCs included ($/MWh) May 2, May 4, May 22, May 23, May 27,

14 Virtuals: Unit Commitment & Dispatch 12/10/13 12/14/13 12/18/13 12/23/13 Base Case Units Committed Units Decommitted w/o Virtuals Units Committed w/o Virtuals Net Commitment Difference

15 UTC s: Unit Commitment & Dispatch 12/10/13 12/14/13 12/18/13 12/23/13 Base Case Units Committed Units Decommitted w/o UTCs Units Committed w/o UTCs Net Commitment Difference

16 Conclusions Incs, Decs, UTCs, etc., will all impact the day ahead commitment and dispatch This can occur to differing degrees on different days in an unpredictable manner Given that all impact scheduling and dispatch, all should receive some share of uplift costs 16

17 Current Review in PJM Currently in PJM Incs/Decs pay uplift but UTC s do not A function of how UTCs evolved over time A wholesale review of uplift is currently underway in PJM This topic will be visited as well as many others 17

18 Appendix: How UTC s Work

19 How UTCs Work UTCs are submitted in the Day Ahead Market and take a position on the price separation between two points For UTCs in the direction of congestion, the are profitable when RT congestion > DA congestion In the counter-flow direction, they are profitable when RT congestion < DA congestion 19

20 Example DA Settlement Forward Flow 100 MWh UTC cleared between points A and B Node A: DA LMP = $50 Node B DA LMP = $75 UTC (A to B) = (Sink LMP Source LMP) * Cleared MWh UTC (A to B) = ($75 - $50) * 100 MWh = $2500 (charge) Equivalent to moving 100 MWh of power from A to B and paying the congestion costs 20

21 Example Balancing Settlement 100 MWh UTC cleared between points A and B Node A: RT LMP = $45 UTC (A to B) = (Sink LMP Source LMP) * Cleared MWh UTC (A to B) = ($80 - $45) * 100 MWh = $3500 (credit) Profit = RT Position DA Position = $ $2500 = $ Node B RT LMP = $80 Transaction does not exist in realtime and relieves flow on the facility. Relief provided in real-time was worth more than the congestion imposed in DA and therefore the UTC is profitable.

22 Alternative Balancing Settlement 100 MWh UTC cleared between points A and B Node A: RT LMP = $45 UTC (A to B) = (Sink LMP Source LMP) * Cleared MWh UTC (A to B) = ($60 - $45) * 100 MWh = $1500 (credit) Loss= RT Position DA Position = $ $2500 = -$ Node B RT LMP = $60 Transaction does not exist in realtime and relieves flow on the facility. Relief provided in real-time was worth less than the congestion imposed in DA and therefore the UTC is NOT profitable.

23 Key Observations UTCs profit when they contribute to convergence of Day-ahead and Real-time congestion When they increase congestion in DA drawing it closer to real-time, they make money When they create congestion in DA that does exist in RT, they lose money This principles applies for both prevailing and counter-flow congestion Therefore, the economic incentive for UTCs is to provide price convergence 23

24 Bid Fees Up-To-Congestion Transaction Applicable Fees o Schedule 9-3 Market Support Service o Schedule 9-MMU o Schedule 9 PJM Settlement o Schedule 9-PJM Settlement = Market Support Offset These are Identical to Increment Offers and Decrement Bids 24