Application of a long-term MPC formulation to hybrid GEOTABS buildings

Optimal controllers such as Model Predictive Control (MPC) are desired in hybrid GEOTABS buildings due to their operation complexity and large thermal inertia. However, the time constant of ground dynamics of the geothermal borefield is too large to be captured by the typical prediction horizons of MPC. This paper develops and applies a long-term MPC formulation that accounts for these long-term dynamics to a typical hybrid GEOTABS building emulator model, which also serves as controller model. The long-term MPC formulation uses prescribed predictions of the building heating and cooling needs and static energy balance equations to formulate a shadow cost for the long-term horizon. A 5 years building simulation demonstrates that there is potential beyond a standard short-term MPC, achieving energy savings up to 20% if these long-term dynamics are incorporated. However, the use of fixed long-term predictions can detriment the potential energy savings of the methodology, and can result in an energy use increase of the same magnitude. Future work will develop variable long-term predictions that account for the building indoor conditions.