Conditional Demand Analysis for Estimating Residential End-Use Load Profiles

This paper reports some preliminary results from an ongoing study that uses regression methods to break down total household load into its constituent parts, each associated with a particular electricity-using end use or appliance. The data base used for this purpose consists of 15-minute integrated demand readings on a random sample of statistical control group customers from the Los Angeles Department of Water and Power TOD (time of day)-pricing experiment for the months of August 1978 (132 customers), 1979 (108 customers), and 1980 (80 customers). Twenty-four regression equations are fitted, each one aimed at explaining variation in the time-averaged load (averaged over days of the month) over customers as a function of temperature, house size, and binary indicator variables that indicate the presence or absence of each of the end uses of interest. This sort of method for extracting the individual contributions of end uses to total household load has become known as conditional demand analysis (Parti and Parti, 1981). The success of this method for isolating end-use loads statistically, without direct metering of the appliance, depends crucially on whether the ownership patterns of appliances are well mixed. For example, if (as in our sample) everyone owns at least one refrigerator, it will be impossible to isolate refrigerator load. Similarly,