Analysis of U.S. Residential Air Leakage Database

The air leakage of a building envelope can be determined from fan pressurization measurements with a blower door. More than 70,000 air leakage measurements have been compiled into a database. In addition to air leakage, the database includes other important characteristics of the dwellings tested, such as floor area, year built, and location. There are also data for some houses on the presence of heating ducts, and floor/basement construction type. The purpose of this work is to identify house characteristics that can be used to predict air leakage. We found that the distribution of leakage normalized with floor area of the house is roughly lognormal. Year built and floor area are the two most significant factors to consider when predicting air leakage: older and smaller houses tend to have higher normalized leakage areas compared to newer and larger ones. Results from multiple linear regression of normalized leakage with respect to these two factors are presented for three types of houses: low-income, energy-efficient, and conventional. We demonstrate a method of using the regression model in conjunction with housing characteristics published by the US Census Bureau to derive a distribution that describes the air leakage of the single-family detached housing stock. Comparison of our estimates with published datasets of air exchange rates suggests that the regression model generates accurate estimates of air leakage distribution.