Data-driven simulation for energy and local comfort optimization: Case study of a laboratory

Abstract

Recent control strategies for heating, ventilation and air conditioning (HVAC) in buildings consider ambient conditions, spatial temperature distribution and occupancy density to maximize occupant comfort at minimum energy. Simulators capturing the energy and thermal behaviour of buildings with HVAC systems provide a convenient platform for developing and testing such control strategies. Commonly used simulators are limited to zonal temperature simulation and are unable to model spatial distribution of temperature within the zones. Thus, these simulators cannot be used for control development that uses finer spatial temperature measurements. This paper proposes a simulation framework that emulates the HVAC behavior and energy consumption as well as temperature distribution in actual buildings. Data-driven techniques are used to facilitate model development and calibration. Since the simulator generates spatial thermal distribution, it is used to devise and test air conditioner (AC) set-point control policy that optimizes local comfort as opposed to the usual zonal comfort. The control thus designed is applied to optimize the AC setpoint schedules in a live academic setting and is shown to induce more energy savings than the control based on zonal comfort optimization.