New York Congestion Costs Experience and Plans

1. New York Congestion CostsExperience and Plans FERC OfficesApril 14, 2005 James Mitsche JMitsche@power-gem.com PowerGEM Power Grid Engineering & Markets

2. Meeting Topics • Congestion • It’s Nature • Analysis Tools • NYISO Activities • NYISO Congestion Analysis • Congestion Measures • Congestion Analysis Approach • Historic Congestion Results • Sensitivity Studies • Congestion Planning • PowerGEM Observations PowerGEM Power Grid Engineering & Markets

3. Remember • Numbers • Average 4 Simultaneous Constraints Every Hour in NY • NY Congestion Impact: 2003 = $85 Million; 2004 = $72 Million • Observations from a Planning Perspective • Congestion Occurs at All Load Levels • Most Congestion Impact is Due To Chronic Events, Not Price Spikes • Bottleneck Relief Is Not as Cost Effective as Generally Believed • Analysis of Large Persistent Transmission Constraints is Reasonable PowerGEM Power Grid Engineering & Markets

4. Congestion • The Cost of Committing or Running Generating Units Out of Merit Order • Caused By Combination of • Insufficient Transmission, Requiring Local More Expensive Supply to Maintain Load/Generation Balance • Generator Characteristics (e.g. min run time, ramp rate, unit MW minimum) • Unit Commitment Process Commits/Decommits Units to Minimize BPC, but Increases Congestion PowerGEM Power Grid Engineering & Markets

5. Nature of Congestion Cost Congestion is: Persistent Usually Not Dramatic At All Load Levels Average $0.59 / MW PowerGEM Power Grid Engineering & Markets

6. How Often are There Simultaneous Transmission Constraints? Answer: Almost Always Average for NY 4.25, Max 13 in 2004 PowerGEM Power Grid Engineering & Markets

7. Congestion Is an Everyday Thing Bid Production Cost Impact PowerGEM Power Grid Engineering & Markets

8. Congestion Is an Every Hour Thing Bid Production Cost Impact PowerGEM Power Grid Engineering & Markets

9. What Causes Congestion At Shoulder Load ? • Unit Commitment Decisions and Restrictions • Unit Minimums • Ramp Rates • Minimum Run Times • Generation Outages • Transmission Outages • Bidding Patterns PowerGEM Power Grid Engineering & Markets

10. Congestion Terms • Bid Production Cost • The Total Cost to Serve Demand at the Bid Price (Not Clearing Price) • Shadow Price • The Incremental Change in Bid Production Cost for an Incremental Change in Constrained Facility Rating • LMP = Shadow Price x Distribution Factor • LMP Congestion Component • Difference Between Local and Reference Point LMP (excepting Losses) • Distribution Factor • Change in Constraint Flow for a Change at a Location, Relative to a Reference Point • Reference Point • Assumed Uncongested Location PowerGEM Power Grid Engineering & Markets

11. Congestion Calculation Tools • Shadow Prices Result From Optimization Simulations • Security Constrained Unit Commitment • Security Constrained Economic Dispatch • These Tools Minimize Total Bid Production Cost Subject to Energy Balance, Flow Limits, and Other Constraints • Payments are Made at the LMP Clearing Price, Not the Bid Price PowerGEM Power Grid Engineering & Markets

12. NYISO Congestion Analysis • NYISO has reported historic congestion costs since 2001 • Included in NYISO Market Advisor's annual "State of the Market" report • Electric System Planning Working Group ("ESPWG") was formed in 2003 to assist in the development of a Comprehensive Planning Process for the NYISO •  ESPWG efforts have led to a detailed methodology for the analysis of historic congestion costs •  This process is now included in the NYISO Tariff (Attachment Y) PowerGEM Power Grid Engineering & Markets

13. ESPWG Objectives To Better Understand Congestion Costs & Causes • Define congestion • Analyze causes of congestion • Provide better information for the market • Consider implications for NYISO Comprehensive Planning Process Characterize Historic Congestion • Measures & Metrics • Magnitude • Location • Zones • Constraints PowerGEM Power Grid Engineering & Markets

14. Congestion Measures • Market Impact Measures • Change in BPC • Incremental BPC Impact (Patton Method) • Payment Impact Measures • Congestion Payments • Demand and Supply Payments PowerGEM Power Grid Engineering & Markets

15. PowerGEM Power Grid Engineering & Markets

16. David Patton’s 2003 NYISO Markets Report • Defines Congestion as: • The difference between total congestion payments from loads and payments to generators and imports; plus • Congestion costs collected from physical bilaterals • Utilizes a “marginal cost” definition: • “Shadow price” x Power flowing across congested interface • Includes both Day-Ahead and Real-Time congestion components • Relative, Not Absolute Measure • Total for 2003: $688 Million PowerGEM Power Grid Engineering & Markets

17. BPC Congestion Impact and Patton Number • Both Numbers are Extreme Cases • BPC Impact Removes All Transmission Constraints (Big Change) • BPC Represents the Maximum Savings Possible • Patton Number is for a Small Incremental Change Only • Better Approach • BPC Change for Discrete Grid Improvements or Other “What If” Tests • It is Believed That Both Methods Should Converge When Looking at Discrete Situations PowerGEM Power Grid Engineering & Markets

18. Comparison of 2004 Totals By Different Metrics, PowerGEM Power Grid Engineering & Markets

19. Congestion Reporting Posting(www.NYISO.com)Total and Zonal Congestion Metrics Congestion Payments By Constraint PowerGEM Power Grid Engineering & Markets

20. Metrics Calculated Using PROBE Software • PROBE(PoRtfolio Ownership Bid Evaluation) Software • A “Study Mode” for the NYISO SCUC • Day Ahead Market Only • Driven by Same Data as the SCUC • Bids (real and virtual, load and generation), Hourly Network Models, Business Rules, TCC • Market Optimization • All Markets (Generation, Price Capped Load, Virtual Load & Generation, Imports, Exports, Wheels, Ancillary Service) • Full Security Constrained Unit Commitment and Dispatch • Includes Mitigation • Produces Same Hourly Results as SCUC (LMP’s, dispatch, flows, etc.) + Insights • 2 Modes • SCUC Viewer • Simulation • Used Since January 2002 • NYISO Market Monitoring & Performance Unit Uses Routinely • Used for Daily PJM Market Advice and Economic Planning PowerGEM Power Grid Engineering & Markets

21. PROBE Daily Security Constrained Unit Commitment NYISO Supplied Daily Simulation Tool Hourly Network Models (24) • User Controls • Ramp Rate Handling • Bids to Optimize • PAR Handling Daily Security Constrained Dispatch Generator Bids Virtual Load & Gen Bids Shadow Prices & LMP’s Imports & Export Bids Fixed Load & Price Capped Load Bids • Energy & Congestion $ for All Bid Types • Generators • Fixed Load • Price Capped Load • Virtuals • Imports & Exports Unit Characteristics (Status, Min Up & Down Time, Ramp Rates) PAR Settings Excel Spreadsheets Monitored Branches, Interfaces Limits, Contingencies Relatively Static or Derived Data Branch Flows Marginal Bids Congestion Impact on Bidders TCC Contracts Zonal Factors PowerGEM Power Grid Engineering & Markets

22. Important Assumptions • The Congestion Impact Calculation is the Difference of the “As Given” Network and a Totally Unconstrained System. Removing All Constraints is Not Truly Practical, but Quantifies the Maximum Possible Savings • Bids are Assumed to Be Mitigated and Unchanged for All Calculations • Bilateral Market Hedging is Not Included • TCC Hedging Attributed Totally to Load. No Inclusion of TCC Auction Cost • Virtual Load, Virtual Generation, and Price Capped Load Assumed to be Fixed MWHr in the Unconstrained Case ($ Payments Change) • No Adjustment for “Unusual Events” • PAR’s Fixed at SCUC Supplied Setting PowerGEM Power Grid Engineering & Markets

23. SCUC – PROBE Results Comparison Note: Sept. 22 Imports were very small (5% of Gen+import Revenue); a small SCUC/PROBE difference results in a larger than normal % difference. All dates in 2003 Conclusion: Overall Revenue and Bid Production Cost is Being Reproduced by PROBE With Acceptable Precision PowerGEM Power Grid Engineering & Markets

24. 2003 & 2004 Summary PowerGEM Power Grid Engineering & Markets

25. 2003 & 2004 Summary Not all Constraints Shown. 95% of 2003, 90% 0f 2004 PowerGEM Power Grid Engineering & Markets

26. - Number Means Congestion Decreases Load Payments 2003 Totals Negative Number Means Gen Payments Congestion Hurt Gens + Number Means Congestion Increases Supplier Production Cost + Number Means Congestion Increases Load Cost PowerGEM Power Grid Engineering & Markets

27. 2003 TotalsUnhedged Congestion Impact by Constraint(Price Increasing Impact only) PowerGEM Power Grid Engineering & Markets

28. - Number Means Congestion Decreases Load Payments 2004 Totals Negative Number Means Gen Payments Congestion Hurt Gens + Number Means Congestion Increases Supplier Production Cost + Number Means Congestion Increases Load Cost PowerGEM Power Grid Engineering & Markets

29. 2004 TotalsUnhedged Congestion Impact by Constraint PowerGEM Power Grid Engineering & Markets

30. Congestion Impact By Day PowerGEM Power Grid Engineering & Markets

31. Observations • 2004 Congestion Impact Not Dramatically Different than 2003, but is Lower • After a Congested January, Summer 2004 Congestion was Lower then 2003 • The Upstate/Imports to Zone J and K Congestion Impact Divide Continues • Individual Constraints 2003 vs 2004 • Consistent • Dunwoodie – Shore Rd 345 • Rainey – Vernon 138 kV • Up • Dunwoodie – Rainey 345 • Pleasant Valley – Leeds 345 kV • Sprainbrook – 49th Street • Down • Central East Interface PowerGEM Power Grid Engineering & Markets

32. “Release Constraints in Sequence” Test Assumes No Network Impedance or Connection Change PowerGEM Power Grid Engineering & Markets

33. Economic Planning Example Using PROBE May 2004 • Dunwoodie – Shore Rd. 345 kV was 25% of Congestion Payments • 2 Tests • 10% Increase in Dunwoodie – Shore Road Rating • 600 MW added in Zone K @ $50/MW PowerGEM Power Grid Engineering & Markets

34. Handling “Unusual Days” Approach • Use a Statistical Approach to Identify Days with an Unusual Amount of Congestion Impact • Define Congestion Impact Using the Primary Congestion Measure (Constrained – Unconstrained) Mitigated Bid Production Cost • Analyze and Characterize Why the Days Were Unusual • Report Congestion Metrics Separately for “Unusual Days” PowerGEM Power Grid Engineering & Markets

35. Unusual Days Part of the Preliminary 2003 Congestion Metrics PowerGEM Power Grid Engineering & Markets

36. Economic Planning Thoughts Using PROBE Everyone Agrees “What if” Studies are Worthwhile Study Approach Questions • Analyze All Hours or a Sample? • Take Out “Unusual Events” in the Base Calculation? • Solution to Each Problem or Multiple Problems ? • What about the “Just Submerged” Constraints ? PowerGEM Power Grid Engineering & Markets

37. Economic Planning Thoughts Using PROBE “Cost Effectiveness” • Measured by the Change in Bid Production Cost with the Upgrade (Economic Efficiency Improvement) • Whole State Net Benefit Viewpoint • 10 Year Planning Horizon • “Benefit” is Present Worth of Production Cost Improvement Net of • Cost of the Transmission Upgrade • Cost of the Upgrade Outage PowerGEM Power Grid Engineering & Markets

38. Chronic Congestion May Be Viewed As: • All Monitored Elements that contributed to the top X% of unhedged congestion payments in 2004 PowerGEM Power Grid Engineering & Markets

39. PowerGEM Observations • More Comprehensive Calculation Significantly Lowers Congestion Impacts • Cost Shifting Rather than Savings • Effect on Energy Markets • Another Constraint is Usually in Hiding • Studies to Date Have Not Shown Instances Where Transmission Development Lowers Cost Enough to Justify the Investment on a Purely Economic Basis • Reliability Upgrades Affect Congestion • Diligence is Justified • Persistent High Impact Constraints • Low Cost Transmission Investments PowerGEM Power Grid Engineering & Markets

40. PowerGEM Observations • LMP Based Rewards for Transmission Investment are Weak • LMP’s are Volatile • Congestion is Shifty and Difficult to Predict • Small Changes Can Make a Large Local Difference • Upgrades Remove the Source of Reward • Free Rider Issues • LMP’s Are Strong Indicators for Generation • Affect on Bid Production Cost is a More Dependable Reward for Transmission Investment PowerGEM Power Grid Engineering & Markets

41. PowerGEM Lessons Learned • Congestion is Persistent • LMP’s are Volatile • Shadow Prices are Even Yet More Volatile • Unit Commitment Has a Major Impact on Congestion Location and Price • Congestion Cost Measurement Depends on Application • Transmission Upgrades Shift Rather than Reduce Cost to Customers • Bid Production Cost Change is a More Stable Measure of Upgrade Merits PowerGEM Power Grid Engineering & Markets

42. Remember • Numbers • Average 4 Simultaneous Constraints Every Hour in NY • NY Congestion Impact: 2003 = $85 Million; 2004 = $72 Million • Observations from a Planning Perspective • Congestion Occurs at All Load Levels • Most Congestion Impact is Due To Chronic Events, Not Price Spikes • Bottleneck Relief Is Not as Cost Effective as Generally Believed • Analysis of Large Persistent Transmission Constraints is Prudent PowerGEM Power Grid Engineering & Markets