Model predictive control applied to residential self-assisted ground source heat pumps

The seasonal performance and feasibility of a "self-assisted" ground source heat pump system are examined. The system uses an electric heating element to inject heat into the ground and reduce peak energy demand by eliminating the need for auxiliary heating. Doing so allows for an undersized borehole to be used with a smaller penalty on peak power consumption. A simulation model for an energy efficient residential building in the Montreal area is implemented in the Modelica language. A model predictive control strategy with a week-long prediction horizon is applied to determine the optimal heat injection rate and timing for the electric heating element. The model uses an improved cell shifting load aggregation scheme for ground temperature predictions. On a borehole that is undersized by 16.7%, the self-assisted configuration with model predictive control reduces peak energy demand by 58% over a 5 year period, with an increase in total annual energy consumption of 2.8% when compared to an unassisted configuration