A stochastic model of integrating occupant behaviour into energy simulation with respect to actual energy consumption in high-rise apartment buildings

• Energy use was 10–30% deviated from averages in apartments built in the 1970s–1980s.
• 17–20 °C set temperatures and 3–8 operating hours result in 25% deviation in heating.
• 25% deviation in electricity use is derived from diverse ranges of appliances.
• Sets of parameter in 50–90% probability reach 10–30% deviation in actual energy use.

Apartment buildings have evolved to be self-sufficient for occupants. Thus, energy use is individually controlled in apartment units, which can be considered as independent thermal zones within buildings. However, this has been disregarded in conventional energy modelling which is mainly applicable for reducing energy demands of buildings with standardised conditions, rather than reflecting actual consumption. This approach has been questioned due to the high levels of uncertainty formed with real buildings. In this study, a model considering occupant random behaviour consuming heating and electricity is developed to reflect variations in actual energy consumption in apartments. Moreover, the effects of various parameters of occupant behaviour in relation to the model were examined. In total 96 apartment blocks in Seoul were used as samples. Gaussian Process Classification was applied to modify occupant random behaviours corresponding to the probability of energy consumption. As a result, it has been found that occupants’ general heating controls (25% deviation) are between three and eight hours, with 17–20 °C set temperatures. Moreover, the operating hours of electric appliances and lighting are also approximated with the probabilities. This methodology could reduce uncertainties in building simulations, and provide a broader application in buildings with similar development stages.