Predictability of time-resolved cross-envelope pressures driving natural infiltration in low-rise residential buildings

Natural infiltration in residential buildings has two major drivers: indoor-outdoor temperature differences (stack effect) and wind effect. While residential infiltration models are long established, their validity has not been evaluated with measurements, and they have rarely been deployed to explain time-resolved indoor-outdoor exchange. Pressure differentials (ΔP) across building envelopes are an intermediate step in modeling; if they cannot be well predicted from the driving forces, then neither can infiltration. We report nearly 16,000 h of environmental and ΔP data, in nine homes, at one-minute resolution that reflects the transient nature of air exchange. Under conditions of low wind (less than 0.25 m/s) and heating (outdoor temperature below indoor), stack pressure is predicted exceptionally well. Biases between observed and predicted values average 0.11 Pa or less across all sites. Biases increase by about a factor of two under cooling conditions, but observations under these conditions were of insufficient length to diagnose the causes. Wind influence on pressure, and hence on infiltration, is not well predicted even with practical, site-based measurements. Airport and site wind speeds, and site wind and envelope pressure, are correlated only modestly, even accounting for wind direction. Simple terrain and shielding classifications cannot reproduce intersite variation. Infiltration models overestimate the influence of wind on pressure even when the most extreme shielding and terrain classes are used in scaling airport data. In addition to evaluating infiltration drivers, this study establishes the difference between time-resolved, cross-envelope pressure differentials at separate points in a single zone (Δ−ΔP) as a building diagnostic.