May 03, UFE ANALYSIS Old – New Model Comparison Compiled by the Load Profiling Group ERCOT Energy Analysis & Aggregation May 03, 2007
2 UFE Basics Sources of UFE include: ■ Generation Measurement Errors ■ Load - Missing/Erroneous Usage Data - Model Error - Load Profile ID Assignment Error ■ Losses -Model Error - Loss Code Assignment Error Negative UFE indicates load/losses are overestimated UFE (unaccounted for energy) is computed as follows: UFE = Generation – (Load + Losses) UFE % is computed as follows: UFE % = UFE / Generation
May 03, UFE Basics Net Generation for Settlement Interval Interval Data Energy Usage Profiled Energy Usage Non-Interval Data Non-Metered Accounts Losses: Transmission & Distribution UFE GAP > Net Generation Compared to Load Buildup
May 03, Analysis Overview Objective: Estimate UFE impact for the 48 new profiles models Analyze UFE impacts on an annual basis and across seasons, day-types and time-of-day Winter (Dec, Jan, Feb) Spring (Mar, Apr) Summer (May – Sep) Fall ( Oct, Nov) Analysis based on current profile type assignments Analysis window January 1, 2005 – July 31, 2006 Analysis limited to the most recent settlement run for trade date Create a dataset of usage-sum values Compute the sum of all meter readings with the same Profile Type, Weather Zone, read start date, and read stop date. Based on usage data available and Profile ID assignments in December 2006
May 03, Create daily profile kWh for new and old models by summing the profile intervals for each day of the analysis window Apply the daily profile kWh values to the usage-sum data to allocate the usage-sum across the days in the meter reading period Result is daily kWh based on new and old profiles Assumed usage records spanning trade days within the analysis window were completely loaded Apply new and old profiles to the daily kWh to allocate it to intervals within the day Sum the profiled interval values across all 48 profile type and weather zone combinations for each interval Result is an old and new version of the total profiled load (for the 48 combinations) for each interval Analysis Overview (continued)
May 03, Compute the difference between the old and new versions of the total profiled load for each interval Difference = Profiled Load old – Profiled Load new Add the difference above to UFE (Old UFE) in the interval from the latest settlement run New UFE = Old UFE + Difference All other settlement values thus are held constant Losses were not recalculated … they were also held constant as a simplifying assumption Proceeded with comparison analysis following the patterns of previous UFE Analysis Reports Analysis Overview (continued)
May 03, UFE on Ercot Peak Day August 23, 2005 Old Models vs New Models
May 03, UFE MWh by Month New Models vs Old Models UFE │UFE│ Abs. UFE MWh for New Models is 13% lower than Old Models
May 03, UFE Cost │UFE Cost│ UFE Cost by Month New Models vs Old Models UFE Cost for New Models is 40% lower than Old Models Abs. UFE Cost for New Models is 19% lower than Old Models
May 03, Statistical Comparisons – Study Period New Models vs Old Models
May 03, Statistical Comparisons - Monthly New Models vs Old Models
May 03, Statistical Comparisons - Monthly New Models vs Old Models
May 03, Statistical Comparisons - Monthly New Models vs Old Models
May 03, Statistical Comparisons - Monthly New Models vs Old Models
May 03, Statistical Comparisons - Monthly New Models vs Old Models
May 03, Statistical Comparisons - Monthly New Models vs Old Models
May 03, Statistical Comparisons - Monthly New Models vs Old Models
May 03, Statistical Comparisons - Monthly New Models vs Old Models
May 03, Statistical Comparisons - Monthly New Models vs Old Models
May 03, Statistical Comparisons - Monthly New Models vs Old Models
May 03, Distribution of UFE MW New Models vs Old Models
May 03, Distribution of UFE Cost New Models vs Old Models UFE Cost = ∑(MCPE × UFE_MWh)
May 03, UFE % by Day of Week New Models vs Old Models Medians, 5 th and 95 th Percentiles compiled over entire analysis period
May 03, UFE % - Average Weekday New Models vs Old Models Medians, 5 th and 95 th Percentiles compiled over entire analysis period
May 03, UFE Cost by Day of Week New Models vs Old Models Medians, 5 th and 95 th Percentiles compiled over entire analysis period
May 03, UFE Cost - Average Weekday New Models vs Old Models Medians, 5 th and 95 th Percentiles compiled over entire analysis period
May 03, UFE % vs Load New Models vs Old Models
May 03, UFE % – Average Weekday New Models vs Old Models
May 03, UFE Cost vs Load New Models vs Old Models
May 03, UFE Cost – Average Weekday New Models vs Old Models
May 03, UFE % vs Load by Season New Models vs Old Models
May 03, UFE % Average Weekday by Season New Models vs Old Models
May 03, UFE Cost vs Load by Season New Models vs Old Models
May 03, UFE Cost – Average Weekday by Season New Models vs Old Models
May 03, UFE % vs ERCOT Load New Models Medians, 5 th and 95 th Percentiles compiled over entire analysis period
May 03, UFE % vs ERCOT Load Old Models Medians, 5 th and 95 th Percentiles compiled over entire analysis period
May 03, UFE Cost vs ERCOT Load New Models Medians, 5 th and 95 th Percentiles compiled over entire analysis period
May 03, UFE Cost vs ERCOT Load Old Models Medians, 5 th and 95 th Percentiles compiled over entire analysis period