Predictive management for energy supply networks using photovoltaics, heat pumps, and battery by two-stage stochastic programming and rule-based control

Predictive management for energy supply networks using photovoltaics generation (PV) units, heat pump water-heating units (HPUs), and battery units is developed by uniquely combining two-stage stochastic schedule programming and rule-based control to enhance their operating performances, including operating cost reduction under low selling prices of the surplus PV output after a feed-in tariff system and uncertain input conditions. The forecast scenarios of input conditions are generated from the probability distributions at each sampling time. The schedule planning problem is formulated using two-stage stochastic mixed-integer linear programming and solved by inputting the forecast scenarios and the initial operation states of network components. In the operation control, the energy flow rates are modulated according to the actual input conditions under the obtained operation schedule. The forecast scenario generation and stochastic schedule planning are updated using a receding horizon approach. The developed management is applied to an annual operating simulation of a residential energy supply network for a housing complex, consisting of a shared PV unit, four sets of an HPU and thermal storage tank, and shared battery unit. The simulation results reveal that the decrease in the annual operating cost reduction by the developed management from the ideal management based on the previously-known PV output is just 1.57% point. Moreover, at the selling price of the surplus PV output higher than 6 yen/kWh, the developed management has an advantage in the annual operating cost over the charging- and exporting-priority managements.