LAUSR

Abstract Window-opening behavior profoundly affects indoor air quality, thermal comfort, and the energy consumption of HVAC systems in office buildings. This paper presents the results of a field study on manual window control conducted in five typical individual offices in the building of a research institute in the city of Xi'an, located in the cold region of China. The average duration of window-open and window-closed states was 5.3 h/day and 6.7 h/day during working hours, respectively, and thus the window-open state accounted for 44% of the total duration during the transition season. Unlike previous studies, the results show that once a window was opened when occupants first come to the office, it would not be left like that for long, until the occupants left the office at the end of day. In addition, window opening behavior is more strongly correlated with indoor PM2.5 concentration and indoor temperature. A new algorithm has been developed that applies dummy variables based on a binary logistic to provide a more accurate description of the relationship between the window state and influencing factors at different levels. The proposed model increased the prediction accuracy of the window being open by 5.9% compared to the continuous variable model. This study focuses on understanding the characteristics of window operation and establishing a more accurate model that offers suitable data and theoretical support for applications to simulate energy consumption in office building.