Estimating Demand Response Load Impacts: Evaluation of BaselineLoad Models for Non-Residential Buildings in California

摘要: Both Federal and California state policymakers areincreasingly interested in developing more standardized consistentapproaches to estimate verify the load impacts of demand responseprograms dynamic pricing tariffs. This study describes a statisticalanalysis performance different models used calculate thebaseline electric for commercial buildings participating ademand-response (DR) program, with emphasis onthe importance weathereffects. During DR event, variety adjustments may be made tobuilding operation, goal reducing building peak electricload. In order determine actual reduction, an estimateof what would have been on day event without any DRactions is needed. baseline profile (BLP) key accuratelyassessing from event-based programs alsoimpact payment settlements certain types programs. We testedseven sample 33 located California.These can loosely categorized into two groups: (1) averagingmethods, which use some linear combination hourly values fromprevious days predict (2) explicit weathermodels, formula based local temperature predictthe load. The were tested both andmore » morningadjustments, data adjust theestimated BLP up or down.Key findings this are: - accuracyof model currently by utilities loadreductions several (i.e., usage highest 3 out of10 previous days) could improved substantially if morning adjustmentfactor applied weather-sensitive institutionalbuildings. Applying adjustment factor significantly reducesthe bias improves accuracy all examined oursample buildings. For low variability, BLPmodels perform reasonably well accuracy. customer accounts withhighly variable loads, we found that no produced satisfactoryresults, although averaging methods best (but notbias). These customers are difficult characterize withstandard rely historic loads weather data.Implications these results program administrators andpolicymakersare: Most apply similar tocommercial industrial sector customers. our whencombined other recent studies (Quantum 2004 2006, Buege et al.,2006) suggests should flexibility andmultiple options suggesting most appropriate method forspecific customers.« less