Paper Conference

Proceedings of BSO Conference 2020: Fifth Conference of IBPSA-England


Occupant Behaviour: A Data Driven Markov Model for Occupancy Presence in Residential Buildings

Benjamin Halls, Steven Firth

Abstract: The presence and movement of occupants in a building can have a significant influence on energy demand. This is due to the requirement that occupants need to be present to interact with buildings systems such as lighting, heating, and appliances. Currently, building simulation tools use standard and repeatable occupancy schedules to simulate occupants and predict energy demand. This may not represent the stochastic nature of occupants in a building environment which can contribute to the performance gap. A better understanding of when occupants are present in a building can aid in improving energy demand predictions. This study investigates occupancy presence in domestic buildings with monitored data from 20 houses. The study aims to develop a better understanding of occupancy presence and to develop a stochastic model to integrate into building simulation tools. The results suggest occupancy presence varies noticeably across different homes, time of week, and number of occupants in a household. The average difference between the predicted and observed means is between 14-17.5% for whole house and room level predictions. The stochastic approach can introduce a level of uncertainty into energy demand predictions, showing the variation with occupant behaviour. Further work will investigate the application of occupant behaviour models in building design.
Pages: 1 - 8