Supervisory model predictive controller (MPC) for residential HVAC systems: Implementation and experimentation on archetype sustainable house in Toronto

Abstract

Residential heating, ventilation and air conditioning (HVAC) systems usually employ on/off controllers to regulate the temperature. Generally, there is no supervisory controller which can use the weather forecast and electricity time of use (TOU) price information to reduce the overall operating cost of the equipment. Lack of a supervisory controller results in higher operating costs of the HVAC system. In this paper, a model predictive control (MPC) based supervisory controller is designed to shift the heating and cooling load of a house to off-peak hours. Supervisory MPC generates the temperature set-points trajectory for on/off controllers. The approach was verified through extensive experiments in an experimental archetype house. By employing the supervisory controller with variable set-points 16% cost savings were obtained when compared to the fixed zone temperature set-points at 25°C. More significant savings of about 50% were seen when MPC-based centralized controller was compared with the fixed zone temperature set-points of 24°C.