Including occupant behavior in building simulation: Comparison of a deterministic vs. a stochastic approach

Data capture and analysis are transforming entire industries, enabling novel solutions developed from a numeric evaluation of real-world phenomena. This generally relies on gathering data on physical conditions and users to create accurate, predictive models and provide customized solutions. Increasingly, data-driven approaches are also becoming a part of architectural design, with the goal of creating user-centric and sustainable buildings. However, while simulation software can accurately model deterministic physical effects, it is still difficult to incorporate stochastic effects related to human factors. This paper analyses one aspect of occupant behavior – window operation – to give designers an intuition of the impact of occupant behavior and associated modelling approaches on building performance. To this end, behavioral patterns observed in a previous field study were incorporated into a dynamic energy simulation and compared to a deterministically modelled baseline. While the stochastic models appear to better capture the dynamic and probabilistic nature of occupants’ actions, the present study highlights the extent to which the assumption with regard to occupant behavior can influence the simulation-assisted performance based design process. The paper also makes suggestions as to how to interpret such simulation results in a way that quantifies the intrinsic uncertainty in stochastic models. We argue that increased data capture and analysis of building inhabitants could lead to a better understanding of their behavior, thereby affecting the decision-making process in favor of a more sustainable and responsive architecture.