A statistical model was developed to relate residential building shell leakage to building characteristics such as building height, floor area, floor leakage, duct leakage, and year built or the age of the house. Statistical regression techniques were used to determine which of the potential building characteristics best described the data. Seven preliminary regressions were performed to investigate the influence of each variable. The results of the eighth and last multivariable linear regression form the predictive model. The major factors that influence the tightness of a residential building are participation in an energy efficiency program (40% tighter than ordinary homes), having low-income occupants (145% leakier than ordinary) and the age of a house (1% increase in Normalized Leakage per year). This predictive model may be applied to data within the range of the data that was used to develop the model.

%I Lawrence Berkeley National Laboratory %8 01/2006 %G eng %12.3

%2 LBNL-59041 %& Chapter