How can stochastic occupant-based archetypes enhance urban building energy modeling for better decision-making?

Urban building energy modeling (UBEM) is widely used to support energy planning at neighborhood and city scales. However, most existing building archetypes rely on deterministic occupant-related schedules that do not adequately capture real occupancy variability, which can reduce the accuracy of energy demand predictions. This study examines the effect of incorporating stochastic occupant-related schedules into UBEM using a stochastic occupant-centric archetype framework applied at the campus scale. The stochastic schedule generation methodology, previously developed by the authors, is applied to generate occupant-related schedules from measured datasets. Three scenarios are evaluated: (1) a standard deterministic approach, (2) a stochastic approach based on historical campus electricity data, and (3) a stochastic approach derived from the Building Genome Project dataset. A validation framework is implemented to assess the influence of occupant-related schedules on electricity demand predictions. The results demonstrate that incorporating stochastic schedules derived from measured data significantly improves UBEM predictive accuracy. Scenario 2 achieved the best performance, showing an average improvement of 32.26% compared to the standard deterministic approach.