Occupants’ behaviour in office building: stochastic models for window opening.

The interactions between building occupants and control systems have a high influence on energy consumption and on internal environmental quality. In the perspective of a future of “nearly-zero” energy buildings, it is crucial to analyse the energy-related interactions deeply to predict a realistic energy use during the design stage. Since the reaction to thermal, acoustic or visual stimulus is not the same for every human being, monitoring the behaviour inside buildings is an essential step to assert differences in energy consumption related to different interactions. Based on measurements in eight offices in Prague, seven models concerning occupants’ window opening behaviour in office buildings were proposed. The dataset was gathered through a monitoring campaign lasting 11 months, from February 2012 until January 2013. Indoor and outdoor environmental conditions were collected in eight south-west exposed rooms, mechanically ventilated and heated through radiators. Two rooms were used by two people while the other six were single offices. Seven models for opening and closing windows were inferred using multiple logistic regression.  The model’s outputs were probabilities of a window opening and window closing event within the next five minutes. Implementing all the models into dynamic energy simulation software would lead to different simulated behaviour patterns and consequently different demands and uses of energy, representing a more reliable variability of energy consumption due to building occupants’.