Robustness of energy consumption and comfort in high-performance residential building with respect to occupant behavior

Highlights: • Monte Carlo simulations are carried out with validated dwelling models. • Monitored and simulated data display great variation of energy and comfort. • Occupant variables have large impacts on a dwelling's energy performance. • Thermal mass does not seem to affect the energy robustness of a dwelling. • Large housing stocks are substantially less sensitive than single dwellings. -- Abstract: Building energy simulations rely on assumptions that can affect their reliability. Occupant behavior in particular is highly uncertain, especially for residential buildings. The great variability of occupants' actions has a great impact on the energy performance of a building and can explain the failures of building simulations to accurately forecast the energy demand of a building. This study quantifies the impacts of humans on the performance of a residential building by simulating dwellings using multiple realistic occupant behavior profiles. The dwelling models were validated with monitored data. Aspects of occupant behavior covered in this study are occupancy, hot water and electricity consumption, heating set point temperature and openings of windows. The individual impact of all these aspects on energy demand and thermal comfort are analysed. Results show great variability of energy consumption and thermal comfort for a given dwelling when different occupants are living in it, with coefficient of variation of approximately 50%. Large housing stocks are less sensitive to occupant behavior than individual dwellings, but their consumption levels remain difficult to predict when using a deterministic approach to represent occupant behavior.