Open and closed loop simulation for predictive control of buildings

The paper presents the system modeling, controller design and numerical simulation results for thermal energy management of a real office building. Focus is set on an efficient and unitary approach which leads from detailed civil engineering specifications of the building elements to compact and effective models which are suitable for control. A modular semi-automated approach in used in order to derive the discrete state-space representation of the system model. This combines the key thermal dynamics of the constructions with a modular list of thermal loads and losses, defined as external heat fluxes. A balanced trade-off is thus achieved between model accuracy and complexity through a compact and effective representation of the plant dynamics. The control strategy is based on a predictive controller which evolves an optimized system input vector, in a closed loop. Paths for occupant feedback integration into a single framework, by using human-in-the-loop models via disturbance channels are also discussed.