Supervisory Model-based Control using Mixed Integer Optimization for stationary hybrid fuel cell systems

Abstract This paper presents an application-oriented comparison of optimization approaches for Model-based Control of a stationary hybrid energy system. The energy system consists of a hybrid storage system with a battery and a hydrogen storage including an electrolyzer and a fuel cell system. The hybrid storage aims at improving the supply security of the intermittent renewable energy sources wind and photovoltaic in a grid-connected context. We show the potential of Mixed Integer Linear Programming (MILP) with a linear cost function in the optimization of a supervisory Model-based Control. The optimization takes into account minimum power for electrolyzer and fuel cell. Furthermore, the non-linear partial load behavior of the components is extracted from experimental data and in good fit modeled via piecewise affine functions. By utilizing this MILP in the optimization we increase the hybrid discharging efficiency by 7 % compared to Linear Programming.