Abstract
As a consequence of considerable uncertainty about occupancy, occupant behaviour, and the corresponding effect on thermal loads in buildings, it is difficult to correctly size heating, ventilation, and air-conditioning (HVAC) equipment. Mechanical engineers avoid liability of potential under-capacity and corresponding thermal discomfort by making conservative assumptions about occupants. Meanwhile, there has been a surge in research on characterizing occupants through increasingly advanced modelling approaches to support building performance simulation, but these have focused on agent-based models representing individual occupants, which may be impractical for building-level HVAC equipment sizing. This paper describes the development of a data-driven stochastic tenant model using 15 months of data from 17 independent commercial tenants. The model is implemented in EnergyPlus to examine its potential for an improved HVAC equipment-sizing procedure. The results show: the standard schedules are reasonable though conservative; oversizing equipment does not greatly improve comfort; and the tremendous importance of modelling inter-tenant diversity.
Acknowledgments
The authors have greatly benefited from discussion with project partners RWDI, Autodesk, and National Research Council Canada. The provider of the data from the anonymous office building is also acknowledged. This paper was inspired by the authors’ involvement with IEA EBC Annex 66.
ORCID
William O'Brien http://orcid.org/0000-0002-0236-5383