Annual patterns and predictors of electric energy usage in occupied manufactured housing established through the use of spline, response surface, and stepwise regression techniques

Abstract

During 1991 electrical usage and physical characteristics data were collected on 166 randomly selected mobile homes in Central Oklahoma. Individual spline, or segmented linear regressions on each home provided a unique pattern of usage which includes occupant behavior and measures of building quality for each home. Sixteen mobile homes had no significant pattern of usage; 150 mobile homes showed a significant pattern of electrical usage. Ninety-eight of those had coefficients of determination above 0.70. Comparisons of predicted and actual usage of air-conditioning, heating, and base load of each mobile home are shown. Data of area, perimeter, value, and age from public records with heating and cooling efficiency, heating and cooling knot temperatures, and base load estimates produced by the spline regressions, such as were used in backward, forward, and step wise regressions. Using aggregate data, the best predictors of annual energy usage in mobile homes in three different usage groups were the crossproduct of a variables defined as heating or cooling efficiency multiplied by a temperature differential value which acknowledges all mobile homes are operated with unique balance, or knot, temperatures and base load. The aggregate models did not need any other variables such as perimeter, value, area or age of the homes to achieve an r-square of 0.95 in a no intercept model. The intercept models achieved a r-square values averaging 0.87. Physical characteristic variables and statistical estimates of knot temperatures and base load were determined to be components of heating and coaling efficiency and base load. There was a very wide range of usage indicative of a wide range of occupant behavior. None of the mobile homes heating or cooling over a normalized year was 1.5 standard deviation less than the mean. Those mobile homes which used more than 1.5 standard deviation above the mean were identified as the extreme energy usage frontier users. Those mobile homes were larger, more expensive, had poorer efficiency values and much higher base loads, but they used energy at a significantly different rate per square foot of area than the other mobile homes.