Model-based statistical sampling for electric utility load research. Final report
This report describes and tests the model-based statistical sampling (MBSS) approach to load research. The MBSS system is shown to give: (1) expansion procedures that achieve high efficiencies by using all relevant information known about the classes and subclasses of interest, (2) sampling plans that are efficiently stratified for the expansion procedures, and (3) realistically estimated sample sizes matched to the characteristics and objectives of each project. The report tests MBSS using two applications - one drawn from a residential class and one drawn from a general service class. In the residential application, MBSS is used to relate demand to usage, for a reduction in sample size of about 20%. The report also shows how MBSS can be used to study particular subclasses, in this case based on family size. The general service application shows the importance of analyzing the characteristics of each class and the objectives of each project. In the general service application, MBSS gives savings of about 40% when demand is related to usage. In this class the required sample sizes are substantially smaller than in the residential class. The general service application also demonstrates the potential effectiveness of multivariate MBSS techniques. In particular, when demand is related to both usage and maximum noncoincidental demand, the savings reach 70%.
- Research Organization:
- Wright (Roger L.) and Associates, Plymouth, MI (USA)
- OSTI ID:
- 5344746
- Report Number(s):
- EPRI-EA-3286; ON: DE84920167
- Country of Publication:
- United States
- Language:
- English
Similar Records
Dramatic Demand Reduction In The Desert Southwest
Orange County Photovoltaic Project & Educational COmponent