Model predictive control strategies for buildings with mixed-mode cooling

The paper presents model predictive control (MPC) strategies for buildings with mixed-mode cooling (window opening position, fan assist, and night cooling schedule) and demonstrates their potential performance bounds in terms of energy savings within thermal comfort constraints, in comparison with standard heuristic rules used in current practice. The study also identifies optimal control sequences coordinated with shading, for the control of solar gains. A transient, multi-zone building energy prediction model, with a coupled thermal and airflow network, is developed in MATLAB, and it is used within an offline MPC framework with Particle Swarm Optimization (embedded in GenOpt) as an optimizer. Simulations are performed for a period of six consecutive summer days with mixed-mode cooling strategies decided by the predictive controller, based on weather forecast and cooling load anticipation over a 24 h planning horizon. The results show that MPC can significantly reduce the cooling requirements compared to baseline night setback control while maintaining the operative temperature during the occupied period within acceptable limits. On the contrary, rule-based control strategies for the window opening position, based on simple heuristics for the outdoor conditions, create an increased risk of overcooling with lower thermal comfort acceptability.