Cost-Optimizing Heat-Pump and Battery Dispatch in Office Building Using Predictive EMS : A case study of Magasin X in Uppsala

Rising and volatile electricity prices puts economical pressure of commercial buildings. This study designs an Energy Management System (EMS) for Magasin X in Uppsala, equipped with solar panels, a battery, and heat pumps, to lower operational costs through the optimization of heat pump and battery operation in relation to the dynamic electricity price. With random forest regression and deep neural networks, accurate predictions of electric loads, solar production, and indoor temperature were achieved. They are inputs to optimization models: battery scheduling through mixed-integer linear programming with degradation cost and heat pump scheduling through linear programming with thermal dynamics. Results show significant cost savings by using the building’s thermal inertia to shift heat pump operation to low-price periods and optimize battery charge/discharge for spot price arbitrage, all while maintaining occupant comfort. The EMS is found to be robust with respect to uncertainties in weather forecasts, highlighting its cost saving potential for flexible building energy management.