150 likes | 288 Views
Day-Ahead Market (DAM) Clearing Process follow-up from February WMS. Matt Mereness, ERCOT Manager DAM & CRR. Outline of Discussion from WMS regarding DAM OD 2/6/13. Two action items from WMS in Feb What was the amount of load resource proration Details and requirements of DAM Optimization
E N D
Day-Ahead Market (DAM) Clearing Process follow-up from February WMS Matt Mereness, ERCOT Manager DAM & CRR
Outline of Discussion from WMS regarding DAM OD 2/6/13 Two action items from WMS in Feb • What was the amount of load resource proration • Details and requirements of DAM Optimization Other observations • A/S linked offer summary • A/S trends since Feb 6 • MIP trend since NPRR322 implemented Feb 12 Appendix • Basic example of DAM Optimization in A/S shortage • Reminder of A/S trades and self-arrangement positions
Q1- WMS Load Proration Question for HE1 OD 2/6/13 • Question from WMS of how many Load Resources and MW were prorated in DAM for HE01 on Feb 6 • Answer is that no Loads were prorated for HE01 • Some loads not fully awarded due to COP for Resources (HSL below Offer amount so only awarded up to HSL)
Q2- DAM DesignDAM design and objective function in protocols Protocols 4.5.1(4) The DAM uses a multi-hour mixed integer programming algorithm to maximize bid-based revenues minus the offer-based costs over the Operating Day, subject to security and other constraints, and ERCOT Ancillary Service procurement requirements. (a) The bid-based revenues include revenues from DAM Energy Bids and Point-to-Point (PTP) Obligation bids. (b) The offer-based costs include costs from the Startup Offer, Minimum Energy Offer, and Energy Offer Curve of any Resource that submitted a Three-Part Supply Offer, DAM Energy-Only Offers and Ancillary Service Offers.
Q2- DAM DesignDocumentation • DAM Requirements document from Go-Live posted at http://nodal.ercot.com/docs/pd/mms/index.html#req • From Appendix 1, DAM clearing mathematical function (see next slide for market clearing objective function) • Energy and A/S Prices both driven by shadow prices for next MW
Q2- DAM DesignDAM clearing process and analysis • DAM engine – Unit commitment creates potential commitment patterns/solutions, then sends to network security analysis. • Based on the results, adjusts commitment pattern • Continues to iterate until all violations are resolved and the solution meets all parameters, including the MIP threshold • As previously noted, the MIP gap/threshold represents the area wherein multiple possible solutions that meet the constraint set become equivalent and indistinguishable to the DAM engine • Once an acceptable solution is found and DAM converges, the final prices are calculated in a pricing run • All awards and prices sent to Price Validation Tool analysis and investigated and validated in parallel by DAM Operator • See appendix for basic market clearing and pricing example
Other Observations:A/S Linked Offer Summary Majority of online AS offers from generators (~75%) were not linked with Non-Spin on Feb 6 • Example is 50 MW capacity offering $10 RegUp and $8 RRS. Perhaps the QSE would also be willing to sell Non-Spin at $8 from the same capacity, but that is not reflected in the offer • No A/S substitution • More linked offers from online resources may have helped the non-spin scarcity in Hour 1
Other ObservationsA/S Offer trends since Feb 6 OffNon-Spin TREND
Other observations:MIP threshold since NPRR322 • MIP threshold is necessary because DAM is not a straight marginal dispatch problem • DAM solution is a combination of binary and continuous variables • Binary variable decisions (turning on a generation resource, awarding a fixed quantity block bid, etc.) are subject to an imperfect solutions within the MIP gap • Purpose of the DAM Make-Whole payment in the case of valid Three-Part Offers • Looks at total profit for the OD, however; doesn’t help if DAM extends the commitment by an hour when turning off sooner was more optimal, for instance • Continuous variable decisions (such as awarding an energy bid curve) are not subject to it and will always be consistent with the clearing prices • Binary variables for recent DAMs: approx 240 TPOs plus 50 fixed quantity bid/offers times the hours contained therein
Other Observations:MIP threshold • How MIP gap is calculated: • As each iteration completes, the engine runs a mini auction where all the binary variables become continuous (e.g., that a resource with a 200 MW LSL can be awarded just 0.1 MW) • This removes the lumpiness from the solution and provides a baseline to compare to the real solution • So the “optimal” solution that is used for comparison is one that is not feasible nor attainable, but it gives an indication of the current solution quality.
Other Observations:MIP Threshold Refinement since NPRR322 on Feb 12 MIP TREND
Example with Linked AS Capacity Simplified example to illustrate price effect from linking AS capacity in AS Offers:
Reminder of Mechanics of AS Trades = QSE AS Obligation - Self-Arranged AS submitted AS amount ERCOT buys in DAM for QSE = + QSE AS Responsibility Total DAM AS Offer awards Total SASM AS Offer awards Self-Arranged AS + AS sold in Trades - AS bought in Trades