Modification to Additional Load Carrying Capability Calculation Formula

1. Agenda Item 4.0 Reliability Committee Meeting May 27, 2010 Modification to Additional Load Carrying Capability Calculation Formula Peter Wong Manager, Resource Adequacy

2. Background • The Westinghouse Capacity Model is a computer program that uses probabilistic mathematics to calculate the resource adequacy of a power system. • New England has been using this program for establishing its regional capacity requirements since the 1970s. • One of the assumption of this program was that the daily peak loads used for calculating system adequacy is normally distributed. • In 1993, a study was conducted to investigate whether the daily peak load distribution actually exhibits a normal distribution. A report titled “Skewness of Daily Load Distribution” was published by NEPLAN in August 1993.

3. Background (cont.) • One of the conclusions of that report was: “The third moment of daily peak load distribution should be incorporated into NEPOOL reliability calculations to represent the skewness of these distributions. This parameter can be forecast using the daily peak load distribution produced by NEPOOL’s short-run forecasting model.” • Between 1994 to 1996 the Westinghouse Capacity Model was modified to accept daily peak load distribution with a skewness consistent with the analytical results. • One of the issues raised by the LSR, TSA, Zones Stakeholder Process relates to using the GE MARS program for calculating Installed Capacity Requirements.

4. Background (cont.) • The Westinghouse Capacity Model program and the GE MARS program should produce similar LOLE results when using the same input assumptions on a single area basis. • One of the most important outputs of the Westinghouse Capacity Model for ICR calculations is the additional load carrying capability (ALCC). • ALCC reflects the ability of the system to meet additional load under capacity surplus conditions. • The program determines the percentage point each weekly peak load distribution can increase so that the system is exactly at 0.1 days/yr LOLE • In response to the stakeholder’s desire to use GE MARS to calculate ICR, ISO developed an algorithm to calculate ALCC using MARS because the program does not have an ALCC feature.

5. Background (cont.) • Following completion of the ALCC algorithm and integration into GE-MARS results from the modified MARS program were benchmarked against Westinghouse. • Initial benchmarking results indicated a small discrepancy between results generated by the two models • Upon investigation it was determined that the logic used by the Westinghouse Capacity Model to identify the ALCC did not fully implement the skewness modification. • After modifying this feature and changing the peak load distribution in the ALCC calculations to reflect the skewness, the ICR results between the Westinghouse Capacity Model and the GE MARS match exactly.

6. Comparison of ICR Values with and Without Skewness in ALCC Calculation

7. Comparison of ICR using Scaled Skewness in ALCC CalculationsScenarios with proxy units and 2012/13 Loads

8. Observations • Skewness has minimal impact on ALCC when the installed resource base is close to the one day in ten requirement. • The impact of not assuming skewness in ALCC calculations is that ICR values are understated under conditions of capacity surplus and overstated during conditions of shortage. • LRA is not impacted by this but MCL is. MCL would increase or decrease by the same amount that ICR changes. MCL = ICR – LRA. Since LRA stays the same value, MCL would increase when ICR increases.

9. Recommendation ISO is recommending that the calculation of the ALCC be modified to fully implement the skewness of the daily peak load distributions as previously intended. No Market Rule changes are required. The revised ALCC calculation methodology will be used to calculate ICR prospectively.

10. Questions