Nonlinear Model Predictive Control of a Heat Storage Tank

Sebastian Albrecht

Friday, November 24, 2017, 15:00

Room 02-014, Georges-Koehler-Allee 103, Freiburg 79110, Germany

Water has a high specific heat capacity; therefore it is appropriate for storing heat. Heat storage tanks can be installed in houses to provide the domestic hot water. Energy prices oscillate over the day, so it would be convenient to heat the storage tank in times when the prices are low. The goal is to minimize the energy costs while fulfilling the family’s demand of warm water. In our case the heat storage tank is modeled by a differential algebraic equation (DAE) of index 1. As forecasts of energy prices and warm water demand can easily be incorporated into an Optimal Control Problem (OCP), Nonlinear Model Predictive Control (NMPC) is a suitable choice to achieve the desired goals. On the one hand, due to effects like free convection this system has very fast dynamics and high sampling rates are needed. On the other hand, for good performance of the NMPC scheme a very long prediction horizon is crucial, which makes the design of a real time feasible controller challenging. We present a NMPC scheme based on the Real Time Iteration scheme to meet the desired goals.