A Stochastic Occupancy Modeling Approach to Enhance the Energy Efficiency of Residential Heating and Cooling through Occupancy Sensing Technology

Occupancy schedules are used in Building Performance Simulation (BPS) to act as proxies for human presence. However, they were not previously used to explore the potentials and limitations of human sensing systems. In this paper, we develop a simulation-based approach to support advances in occupancy sensing to specifically examine the impact of sensing errors, such as false positives, of human presence detection systems, by using occupancy schedules to quantify residential building heating and cooling energy use. The aim is to examine varying effects of human detection system configurations on thermal energy consumption in false sensing scenarios, and to introduce occupancy schedules as a means to inform processes of such sensing systems. To extrapolate stochastic transition matrices and generate reliable probabilistically driven occupancy schedules, a Markov-Chain analysis of the 2018 American Time Use Survey (ATUS) is used to develop presence schedules. We then evaluate the impact of false positives in binary occupancy modelling scenarios using Honeybee as a front-end interface in Rhino/Grasshopper, and EnergyPlus as a backend engine. Overall, the aim of this work is to recommend guidelines for various system configurations in which the use of low-cost sensing is justified for heating and cooling regulation.