Optimal predictive control of phase change material-based energy storage in buildings via mixed-integer convex programming

Abstract

This paper describes a mixed-integer convex programming-based control strategy to optimize the operation of phase change material-based energy storage integrated in building supply ducts. To improve the numerical feasibility, the original nonlinear control problem is pre-conditioned and transformed to a mixed-integer convex program through convexification of the cooling system control model, mixed-integer reformulation of the phase change material dynamics and discretization of the supply airflow. The overall control framework was leveraged toward development of two model predictive control strategies to optimally charge/discharge the phase change material storage, through supply air temperature reset, and the building passive thermal mass, via scheduling of the zone air temperature setpoint. These strategies were tested and compared to two baseline control strategies using a simulation case study over three summer days. Test results show that using the phase change material energy storage alone, energy cost savings of 2.9% and peak demand reduction of 46.7% could be achieved, compared to a conventional fixed-supply air temperature and zone air temperature night setup control strategy; when both the active (phase change material) and passive (building thermal mass) storage capacities are utilized, the savings potentials could increase to 8.4% for the energy cost and 65% for the demand charge. • Integration of PCM energy storage in supply ducts is effective in reducing demand. • Predictive control of PCM storage offers significant utility cost savings potential. • Charging/discharging of PCM thermal storage results in efficiency losses. • Combining active and pass storage can provide the maximum load shifting potential.