Earliest deadline control of an energy source and a group of heatpumps with uncertain demand

Renewable energy sources are important for reducing carbon emission. There are many ways to integrate renewable energy into existing supply-demand chains. One way is to replace natural gas by biogas. In Meppel, a new urban area is to be equipped with a hybrid energy system with a Combined Heat and Power unit (CHP) at the heart of the system for the production of heat and electricity. The heat is transferred into a district heating system which is fed by multiple heat sources. The electricity is distributed to houses, each having a heat pump for space heating and domestic hot water. Earlier we investigated centralized control options for this hybrid energy system, focusing on the optimal control of the heat pumps with the objective to balance electricity production by the CHP and demand by the heat pumps. In [2] we investigated a control method based on MILP (Mixed Integer Linear Program). Although this method properly balances electricity production and consumption and fulfill all energy demands as required, the optimization steers towards frequent shifting of the heat pumps, which is problematic for the efficiency and lifetime of the heat pumps. Therefore, in [1] we investigated possible solutions to improve steadiness or robustness of the heat pump and CHP control. We developed a control method based on earliest deadline theory and dynamic programming and we demonstrated that with this method it is possible to reach near-optimum control strategies within very short computational time with the desired robustness of the control of a large group of heat pumps and a central CHP. In our previous papers [2], [4], [1], we assumed perfect prediction in our control strategies, i.e. all input data (e.g. occupancy related thermal gains and losses, ventilation and domestic hot water consumption) for the realization of the control are identical to the predicted values. Of course this is an unrealistic assumption since weather forecasts are not perfect and occupancy related schedules may vary substantially. This paper improves previous results by considering inaccuracies in predictions and it develops a control method based on earliest deadline theory which handles these inaccuracies.