Presentation on theme: "Constraint Enforcement Issues in Market Optimization Jim Price Lead Engineering Specialist CAISO Market Analysis and Development Presentation to WECC Seams."— Presentation transcript:
Constraint Enforcement Issues in Market Optimization Jim Price Lead Engineering Specialist CAISO Market Analysis and Development Presentation to WECC Seams Issues Subcommittee November 15-16, 2010
Slide 2 Management of market optimization when bids are insufficient Background on uneconomic adjustment parameters Market clearing software uses optimization techniques to minimize bid costs based on submitted/mitigated bids, while respecting power balance, resource operating constraints, transmission constraints, and a set of established scheduling priorities. Each market process (day-ahead, residual unit commitment, real-time) requires coordinated way adjust self-schedules and/or relax constraints uneconomic adjustments when reasonably effective economic bids are exhausted before the optimization reaches a feasible solution. One task in market design is to tune the parameter settings used for uneconomic adjustments to achieve practical market results. Presentation highlights parameters in CAISO market that are the most comparable to issues in EIM and ECC. CAISO parameters also address additional issues: over-generation, energy shortage during fast ramping, maximum energy from use-limited resources, minimum on-line capacity, ancillary service self-provision, details of residual unit commitment, quick-start resource capacity, etc.
Slide 3 Scheduling run establishes unit commitment and adjusts schedules using priorities. Scheduling run uses priorities of uneconomic adjustment parameters for adjusting self-schedules and constraint relaxation (penalty prices). If effective economic bids are insufficient to manage constraints, market optimization respects self-schedules, in order of priority (listing is for CAISOs day-ahead market): Reliability Must Run (RMR) generation pre-dispatch; Transmission Ownership Right (TOR) self-schedules; Existing Transmission Contract (ETC) self-schedules, observing ETC priority levels provided by the responsible transmission owners; Other CAISO demand self-schedules, exports explicitly identified in a Resource Adequacy (RA) plan to be served by RA capacity, and export self- schedules explicitly sourced by non-RA capacity; Export self-schedules not explicitly sourced by non-RA capacity or identified in RA plan; Day-ahead Regulatory Must-Run and Regulatory Must-Take Generation; Other supply self-schedules.
Slide 4 Then, pricing run sets the final prices. Pricing run establishes bid prices reflecting economic bid ranges. Extends and limits dispatched bid range as determined by scheduling run. Pricing run avoids unwarranted price impacts by limiting costs of transmission constraint enforcement, using a lower penalty price in pricing run between the original limit and any relaxed limit from the scheduling run.
Slide 5 Illustration: Economic & uneconomic bids for generator Scenario: Generator bids a self-schedule plus an economic bid. A specific output is needed for RMR. The optimization uses negative bid prices for RMR pre-dispatch and remaining self-schedule.
Slide 6 Illustration: Economic & uneconomic bids for demand Bids for Demand (loads in CAISO, and exports) can also include both self-schedules and economic bids. The optimization uses positive bid prices for self-schedules for Demand.
Slide 7 Additional provisions for market clearing & pricing Penalty prices for constraint relaxation and uneconomic bid prices are inter- related and require coordination. Between RMR/ TOR/ ETC priorities and other self-schedules, transmission constraints within CAISO are relaxed Setting RMR requirements considers transmission capacity. In case of conflict, optimization honors RMR dispatch. TOR and ETC self-schedules are adjusted only through operator action Relaxation of intertie scheduling capacity exists in software, but at a penalty price beyond other priorities IFM first tries to clear market using economic bids. If non-competitive transmission constraints cannot be resolved: 1.Step 1: Schedule energy from capacity that was submitted as self-provided ancillary services. 2.Step 2: If Step 1 is not sufficient, relax transmission constraints consistent with operating practices. Steps reflect priorities, not sequential order due to relative effectiveness.
Slide 8 Testing by modifying market cases determined the optimization parameters Testing criteria reflect trade-offs including but not limited to: Protecting self-schedules and maintaining pre-specified priorities, and allowing effective bids within bid cap to be accepted, while avoiding unreasonable pricing and scheduling outcomes. Market outcomes consistent with good operating practices. Allow prices to reach reasonable levels reflecting scarcity. Shift-factor effectiveness threshold. Important: To illustrate uneconomic adjustment parameters, examples do not reflect normal conditions. In order to create infeasible transmission constraints and other constraint violations, it was necessary to make adjustments such as: Reduce transmission limits to small percentages of normal limits Create energy shortages and imbalances Increase ETC reservations to exceed intertie capacity Reduce MW and increase price of ancillary service bids
Slide 9 Example 1: Transmission impact on large areas SDG&E LAP For Hour Beginning 16:00: Total LMP = $59.86/MWh, Energy component = 102.00, Congestion = -43.01, Loss = 0.88 SCE LAP For Hour Beginning 16:00: Total LMP = $73.00/MWh, Energy component = 102.00, Congestion = -32.25, Loss = 3.26 North of SONGS corridor limited to 500 MW Parallel paths Example: The North of SONGS corridor connects SDG&E to SCE. For parameter testing, its limit was reduced by 79.5%, to 500 MW. Other transmission limitations discussed for other examples also apply.
Slide 10 When economic bids resolve constraints, price impact of severe constraints can be moderate Result of example: Due to sufficiency of economic bids, scheduling and pricing runs produce consistent prices, and LMP impacts are moderate and understandable. No reduction of self-schedules is needed to enforce this constraint.
Slide 11 Example 2: Scheduling and pricing run results when self-schedules are reduced 1: Economic bids are limited 2: Generic self- schedules are constrained 3: ETC self- schedules are constrained Final schedules for economic bids-59 MW-80 MW-12.5 MW Final generic self-schedules154 MW30 MW0 MW Final ETC self-schedules5 MW150 MW112.5 MW Intertie shadow price (scheduling)$55.36/MW$601.42/MW$5554.81/MW Scheduling run LMP$2.87/MWh-$550/MWh-$5500/MWh Pricing run LMP$2.87/MWh-$30/MWh Example: Radial intertie capacity is reduced to 100 MW. (For now, assume penalty price on transmission constraint is high, to enforce the constraint.) Case 1: All self-schedules are feasible, and economic bids are limited to enforce binding intertie constraint. (Imports are shown with positive sign.) Case 2: ETC self-schedule increases to 150 MW. Other (generic) self-schedules must be reduced, to the point where the constraint is enforced. Case 3: Reduced export bids require reduction of the ETC self-schedule, after other self-schedules are reduced to zero MW.
Slide 12 Refining the penalty price for transmission In the radial intertie example, enforcing the transmission constraint assumed the penalty price for transmission constraints to be higher than the uneconomic bid prices for self-schedules. Example 2 is a simple case, since the intertie constraint is radial. In a looped network, further definition of penalty prices is needed. In some local areas, the most effective resource for managing a constraint is only about 10% effective. Principle for penalty price at which transmission constraint relaxes in scheduling run, at start of CAISOs new market: a resource that is at least 10% effective in managing a constraint, and bids energy at the bid cap, should be accepted before relaxing the constraint. Penalty price of transmission constraints in scheduling run = initial bid cap ($500) / 10% = $5000/MW
Slide 13 Example 3: Relaxation of transmission constraint For illustration of market clearing and pricing principles: Capacity of Tesla – Ravenswood 230 kV line is reduced to less than 33% of capacity, causing constraint relaxation in several hours of test case. Most effective resource for managing constraint is Potrero (San Francisco). LMPs at Potrero (16% effectiveness) during affected hours are $721 to $865.
Slide 14 Example 4: Maintaining TOR & ETC priority Test of impact on representative ETCs constrained key lines to identified pumps and generators utilizing transmission rights. Each site has multiple lines as feeders from CAISO 230 kV grid. In event of outage, transmission capacity is determined by capacity of remaining feeder lines. More severe constraint is derate of one in-service lines capacity. Power flow then limits parallel lines to flows similar to the derate. 230 kV lines, derated to 10 MW each: 30580_ALTM MDW to 38610_DELTAPMP 30765_LOSBANOS to 38615_DS AMIGO 30970_MIDWAY to 38600_BUENAVJ1 38620_HYATT to 30300_TABLMTN ETC priority applies to individual resources, not balanced adjustment. If generation source were curtailed, market would match demand pending operator action.
Slide 15 Priorities would be honored in scheduling run If ETC priority level = $3200, only Banks (Delta) pumps would have self-schedule adjustments in test case. The following self-schedules were not adjusted by optimization: Dos Amigos: Highest LMP in scheduling run = $3185 CDWR07 aggregation: Highest LMP in scheduling run = $2537 Hyatt-Thermalito: Lowest LMP in scheduling run = $-1509 At Banks pumps, ETC priority level = $3600 would avoid self- schedule adjustments Banks: Highest LMP in scheduling run = $3531 Final uneconomic bid prices provide additional schedule protection, by relaxing transmission before adjusting TOR or ETC schedules.
Slide 16 LMPs reflect shadow price of transmission and shift factor (a.k.a. effectiveness) of resources. These LMPs result from scheduling run parameter values and the severe constraints in this test case (including ones not stated above).
Slide 17 Setting final market prices For final prices, re-dispatch costs using bid prices should set LMPs, instead of an administrative price from scheduling run: Transmission price between original and relaxed limits (plus epsilon) = bid cap = $500 in pricing run (acts as floor on final transmission shadow price) Uneconomic bids in pricing run are the bid cap and floor Penalty Price Description (subset of parameters) Scheduling Run (1 st year) Pricing Run (1 st year) RMR pre-dispatch$-6000$-30 Transmission ownership rights & ETCs5100 - 5900500 Transmission constraints (branch, corridor, nomogram, contingency) 5000500 Self-scheduled CAISO demand and exports using non-RA supply 1000500 Self-scheduled supply-550-30
Slide 18 References Market Operations Business Practice Manual (BPM) Section 6.6.5 https://bpm.caiso.com/bpm/bpm/version/000000000000109 Final comment on Uneconomic Adjustment in the MRTU Market Optimizations from Market Surveillance Committee http://www.caiso.com/2059/2059765f39fb0.pdf Draft Final Proposal on Parameter Tuning for Uneconomic Adjustments in the MRTU Market Optimizations http://www.caiso.com/1fe1/1fe1e4d012730.pdf