Analysis of Air Leakage Measurements from Residential Diagnostics Database

The LBNL Residential Diagnostics Database (ResDB) contains blower door measurements and other diagnostic test results for US homes. Most of the data have been contributed by weatherization assistance programs and residential energy efficiency programs. We analyzed the air leakage measurements, using normalized leakage as the metric, of 134,000 single-family detached homes. Almost all 50 US states are represented. We performed regression analyses to examine the relationship between normalized leakage and various house characteristics. We identified parameters that are useful as explanatory variables, including floor area, height, vintage, and climate zone. Foundation type and whether ducts are located outside or inside the conditioned space also were found to be useful parameters for predicting normalized leakage. We developed a regression model that explains approximately 68% of the observed variability across US homes. A more spatially refined model for 4,500 California homes in ResDB explains 76% of the observed variability. Comparison of the air leakage measurements before and after retrofit shows a reduction in normalized leakage, i.e. an improvement in airtightness, of 20% to 30%. Homes that are rated for energy efficiency have normalized leakage values that are, on average, 30% lower than non-rated homes. The resulted regression model can be used to predict air leakage values for individual homes, and distributions for groups of homes, based on their characteristics. For the US housing stock, the regression model predicts normalized leakage to range between 0.22 and 1.95, with a median of 0.67.