Evolution in Wholesale Market Design in ERCOT

1
Evolution in Wholesale Market Designin ERCOT

• Eric S. Schubert, Ph.D.
• Market Oversight Division
• Public Utility Commission of Texas
• December 13, 2004

2
ERCOT Wholesale Market Design Has Been An Ongoing
Process

• Number of problems identified over time
• Proposed remedies (2002-2004)
• Process has been ongoing for 3 years
• Stakeholders have reviewed a range of remedies
• Result has been a deeper understanding of issues
• Options for PUCT to consider in 2005
• Stay with zonal with partial fixes
• Nodal market design
• Hybrid market design

3
Major ProblemsExperienced in ERCOT Zonal Model

• Local congestion is systematic and frequent
  (2001-2002)
• Inefficient siting dispatch of resources (2002)
• Creation of new CSCs and load zones is
  problematic and creates commercial uncertainties
  (2003)
• Portfolio dispatch hinders real-time operations
  (2003-2004)
• Decentralized unit commitment may be inefficient
  (2004)

4
Potential Remedies for Five Problems in ERCOT
Wholesale Market

• Direct assignment of congestion rents on all
  transmission lines
• Resource-specific offer curves
• Centralized unit commitment in day-ahead market
• Ensure optimal resource deployment
• Assign resulting costs appropriately

5
Wholesale Market DesignAn Intricate Piece of
Machinery

• A wholesale electric market is a complex marriage
  of economics and engineering
• Engineering Balancing the power system and
  having right amount of energy in the right places
  every few seconds under transmission network
  limitations
• Economics Coordinating REPs, PGCs, and other
  market participants using price signals in a
  deregulated market
• Unintended negative consequences when the two are
  not wedded properly

6
Engineering Considerations

• No storage of electricity
• Limited transportation
• Risk of blackout or cascade failure
• Generation schedules not convex
• Start-up time
• Minimum loading of unit
• Minimum run time
• Ramp rate
• Unexpected forced outages

7
Economic Considerations

• Real-time market
• Very limited demand response
• Limited supply response
• Day-ahead market
• Unit commitment process
• Local reliability issues
• Long-term market Entry and exit of resources
• Entry of large-scale generation often occurs with
  a lag
• Exit of uneconomic resources can happen quickly
• New entrants have arrived in waves (e.g., wind,
  CC)
• Newer technologies are smaller and quicker to
  install
• Ensure the best mix of resources to achieve
  desired operational flexibility (e.g., location,
  startup, ramping)

8
Key Elements For a Good Market Design

• Sufficient system engineering elements modeled in
  economic dispatch
• Incentive compatibility Pay the way you play
• Most efficient use of resources
• Minimizing opportunities for gaming and uplift
• Flexibility to adapt to changing conditions

9
Some Possible Alternatives to the Current Zonal
Model (1)
10
Some Possible Alternatives to the Current Zonal
Model (2)

• ERCOT Zonal Model
• Simultaneous Market Clearing (SMC)
• Nodal When You Need It
• SMC Nodal When You Need It
• Nodal Light
• Transitional Alternative Pricing and Settlement
  (TAPAS)
• ADAM
• EH-DAM
• IDAM (Nodal design seen in Northeast Markets)

11
ERCOT Zonal ModelStarting Point

• Original zonal market design was based on
  stakeholder process (Sep. 1999 to Oct. 2000)
• Commission approved an improved version of
  stakeholders proposal (June 2001)
• Commission set triggers for direct assignment of
  congestion rents
• Interzonal and intrazonal congestion
• Model assumed random and infrequent congestion
• Threshold of 20 million over twelve months

12
ERCOT Zonal ModelDirect Assignment of
Congestion Rents

• Zonal design paid market participants for
  scheduled energy, not just delivered energy
• Incentives to cause congestion (DEC game)
• Schedule unit to create congestion and receive
  OOME Down payment
• Offer another unit to clear congestion and
  receive OOME Up payment
• Direct assignment eliminates the DEC game
• Charge congestion rents on all delivered output
• Eliminates OOME Down payments

13
ERCOT Zonal Model Uplift Triggers Reached

• Market Opens July 31, 2001
• Interzonal Congestion August 2001
• Threshold reached in a few weeks
• Direct assignment started on Feb. 15, 2002
• Intrazonal Congestion March 2002
• ERCOT develops partial solution
• Reduced OOME Down payments
• No direct assignment on local lines
• Not in compliance with Commission order

14
Nodal When You Need ItMODs Remedy For Uplift

• PUCT Order 23220 on direct assignment
• Implement direct assignment in a zonal market
  design
• If not feasible, then consider move to nodal
  market design
• MOD proposes Nodal When You Need It hybrid model

15
Nodal When You Need ItPortfolio Offer Curves
and Nodal Pricing

• Uses zonal two-step approach
• Uses bid premium as simplified bid curve
• Produces simplified nodal results for resources,
  zonal pricing for loads
• Complication Hedging congestion rents

16
Nodal When You Need ItDeadlock at ERCOT

• MOD presents hybrid to ERCOT (Spring 2002)
• Stakeholders debate and reject hybrid proposal
  (Summer 2002)
• Both sides agree to bring issue to Commission

17
Nodal When You Need It Commission Looks for A
Long-Term Solution

• Issue comes to Commission (Sep. 2002)
• Staff files comments on other problems
• Poor generation siting signals (McCamey)
• Resistance to adjusting CSCs and zones
• Commission decides to look for a comprehensive
  design solution (2003)

18
Resource-Specific Offer Curves Without Direct
Assignment on All Lines

• Addresses engineering problems associated with
  portfolio dispatch
• Gives ERCOT operator better tools to manage
  real-time dispatch
• Creates potential inconsistencies between
  dispatch (resource-specific) and settlement
  (zonal)

19
Adding Resource-Specific Offer Curves
Simultaneous Market Clearing (1)

• ERCOT staff and stakeholders discuss need to
  improve real-time dispatch in zonal framework /
  software (Nov. 2002- May 2003)
• ERCOT operations staff and stakeholders reviewed
  alphabet soup of options
• Simultaneous Market Clearing (SMC) addressed more
  issues than other alternatives
• SMC solves zonal and local congestion in one
  step, not two

20
Adding Resource-Specific Offer Curves
Simultaneous Market Clearing (2)

• Commissioner Perlman asks WMS to consider SMC
  (March 2003)
• No regrets list
• Long-term alternative solution
• Transition mechanism to Texas Nodal
• WMS provided Commission with recommendations
  (April 2003)
• Release 3 folded some local congestion
  management into balancing energy market (2003-4)
  but created unintended consequences

21
Adding Resource-Specific Offer Curves CAISOs
TAPAS (2004)

• Transitional Alternative Pricing And Settlement
• CAISO considering move to nodal market design
• Californias DWR signed long-term Sellers
  choice contracts at height of energy crisis
• TAPAS to institute many features similar to Texas
  Nodal without nodal pricing to avoid problems
  these contracts would cause
• Dr. Larry Ruff and CAISOs Market Surveillance
  Committee have highlighted inconsistencies of
  dispatch and settlement under TAPAS
• Decision pending on TAPAS and nodal design

22
Resource-Specific Offer Curves and Nodal Pricing
of Resources Texas Nodal Rulemaking (2002
2003)

• Rulemaking (Sep. 2002 Aug. 2003)
• Workshops (November 2002 January 2003)
• Goal to provide framework that included key
  market elements
• Commission approves Texas Nodal
• Direct assignment of congestion rents
• Resource-specific offer curves
• Centralized day-ahead energy market
• Loads settle on zonal prices

23
Texas Nodal Team (TNT) Process (2003-2005)

• Rule says stakeholders develop details
• TNT concept groups write white papers
• TNT stakeholders review protocol language
• PUCT staff has been active in attending meetings
• Market experts review market design to identify
  gaps / gaming opportunities

24
Questions?