Citation

Shi S, Li H, Ding X, Gao X. Build. Environ. 2020; 170: e106610.

Copyright

(Copyright © 2020, Elsevier Publishing)

DOI

10.1016/j.buildenv.2019.106610

PMID

unavailable

Abstract

Window opening behavior is an important factor influencing building performance, including building energy consumption, indoor thermal environment and indoor air quality. Traditionally utilized linear logistic regression model, which describes the correlations between the window opening probability and the environmental driven factors, fails to include the variety among different residences. This study utilized multilevel logistic regression model to analyze the effect of household features on the residential probability of window opening. Here the environmental factors were set as lower-level predictors while the household features were set as higher-level predictors. The studied household features include the area, the located floor, the home renovation condition and the special type of residents (i.e., smoker and old people). The analysis was conducted on a measured dataset monitored in 10 residential apartments in Nanjing. According to the results, 84% of the difference in the probability of window opening was attributable to the residence-level variety, which cannot be ignored. Although the higher-level household features show no direct effect on the window opening probability, all of them significantly affect the correlations between the probability of window opening and the lower-level environmental predictors. Together with future field measurements of representative households, this research will be a crucial basis to build up the multilevel structure of predictors for the residential window opening behaviors, which will further make an improvement to represent the residence-level variety in building performance simulations.

Language: en

Keywords

Environmental factors; Household features; Multilevel logistic regression; Natural ventilation; Residence; Window opening behavior