1. Probabilistic Approach to Resource Adequacy Resource Adequacy Forum Technical Committee Meeting Portland, OR January 23, 2009

3. Current Method Monte Carlo simulation with various random variables Monte Carlo simulation with various random variables Record curtailment events Record curtailment events Define a curtailment “threshold” to count only “significant” or bad events Define a curtailment “threshold” to count only “significant” or bad events Calculate LOLP as the number of simulations with a bad event divided by the total number of simulations Calculate LOLP as the number of simulations with a bad event divided by the total number of simulations Adequacy for winter energy and summer/winter capacity have different thresholds and are assessed separately Adequacy for winter energy and summer/winter capacity have different thresholds and are assessed separately

4. Current Random Variables Streamflow conditions (1929-98) Streamflow conditions (1929-98) Loads (based on regionally weighted daily average temperatures, 1929-2005) Loads (based on regionally weighted daily average temperatures, 1929-2005) Thermal resource performance Thermal resource performance

5. Current Correlations Streamflows and temperatures dependent Streamflows and temperatures dependent Loads and temperature are dependent Loads and temperature are dependent Streamflows and loads are “lock stepped” Streamflows and loads are “lock stepped” Thermal performance is independent Thermal performance is independent

6. Proposed Method Monte Carlo simulation with various random variables Monte Carlo simulation with various random variables Define a “contingency” resource for each month of the year Define a “contingency” resource for each month of the year Record curtailment events across all months of the year Record curtailment events across all months of the year Calculate a single LOLP as the number of simulations with curtailments divided by the total number of simulations Calculate a single LOLP as the number of simulations with curtailments divided by the total number of simulations

7. Proposed Random Variables Streamflow conditions Streamflow conditions Loads (based on temperature) Loads (based on temperature) Thermal resource performance Thermal resource performance Wind generation Wind generation

8. Proposed Correlations Streamflows and temperatures are independent Streamflows and temperatures are independent Loads and temperature are dependent Loads and temperature are dependent Wind generation is temperature dependent Wind generation is temperature dependent Thermal performance is independent Thermal performance is independent

9. Problems with LOLP Does not address magnitude of problems (doesn’t offer any guide toward solutions) Does not address magnitude of problems (doesn’t offer any guide toward solutions) Could have two scenarios with the same LOLP but with different worst cases Could have two scenarios with the same LOLP but with different worst cases May not be an issue if there is a strong correlation between LOLP and magnitude May not be an issue if there is a strong correlation between LOLP and magnitude Does not address multiple events within a single simulation Does not address multiple events within a single simulation May not be an issue if the same solution resolves each event May not be an issue if the same solution resolves each event

10. Other Adequacy Metrics LOLE: loss of load expectation (%) LOLE: loss of load expectation (%) Number of hours with curtailment divided by the total number of hours simulated Number of hours with curtailment divided by the total number of hours simulated Can be more intuitive, i.e. 99.5% reliable Can be more intuitive, i.e. 99.5% reliable Does not address magnitude Does not address magnitude EUE: expected unserved energy (MW-hr) EUE: expected unserved energy (MW-hr) Average amount of unserved energy per year Average amount of unserved energy per year Lacks specific information about severe events Lacks specific information about severe events

11. Other Adequacy Metrics Frequency of occurrence – how often is an event likely to happen? Frequency of occurrence – how often is an event likely to happen? A variation of LOLE A variation of LOLE Magnitude of occurrence – how bad is an event likely to be? Magnitude of occurrence – how bad is an event likely to be? Average loss of energy in MW-hrs Average loss of energy in MW-hrs Duration of occurrence – how long can I expect an event to last? Duration of occurrence – how long can I expect an event to last? Average duration per event (hours) Average duration per event (hours) These measures do not capture information for severe events These measures do not capture information for severe events

12. Possible Modifications to the Current Method Replace LOLP with an alternative adequacy metric Replace LOLP with an alternative adequacy metric Use LOLP in conjunction with an alternative adequacy metric Use LOLP in conjunction with an alternative adequacy metric Use LOLP in conjunction with the magnitude of the most severe event (or an average of the worst 10% of events) Use LOLP in conjunction with the magnitude of the most severe event (or an average of the worst 10% of events)

13. Redefining Thresholds How is the “adequate” threshold for LOLP (or other metric) assessed? How is the “adequate” threshold for LOLP (or other metric) assessed? What is the region’s tolerance for “unwanted” events? What is the region’s tolerance for “unwanted” events? What does it really mean to customers? What does it really mean to customers? Should we continue to “translate” the adequacy metric into deterministic measures (annual L/R balance and surplus peak reserve margins)? Should we continue to “translate” the adequacy metric into deterministic measures (annual L/R balance and surplus peak reserve margins)?

14. Next Steps Finish the benchmarking process Finish the benchmarking process More extensive review of curtailment events and causes More extensive review of curtailment events and causes Contract with McCoy/PSRI to review methodology Contract with McCoy/PSRI to review methodology Work with other NERC sub-regions to be as consistent as possible Work with other NERC sub-regions to be as consistent as possible Propose revised approach for 2010 assessment Propose revised approach for 2010 assessment