Distributed model predictive control for building demand-side management

Demand-side management is widely acknowledged as an important source of flexibility and then an essential element to balance supply and demand more effectively. A fundamental challenge is to enable buildings to participate in demand-side services without violating indoor comfort. In this paper, we present a hierarchical Model Predictive Control (MPC) approach to demand-side management in buildings, which encompasses heating/cooling systems, onsite generation and storage technologies, and enables the building to participate in balancing programs. In the upper level, a trajectory planner calculates the best possible reachable power profile to track as reference. In the lower level, a tracking problem is solved, which also minimises the energy use without violating the indoor comfort. The tracking problem is formulated so that an active-set method can be applied and the MPC problem can be solved in a distributed manner. Numerical results with real data from one university building show the promising performance and computational tractability of the proposed approach, which can enable practical implementations on building platforms.