End-effect mitigation in multi-period stochastic programming of energy storage operations

Energy storage units offer vital balancing power for energy systems with increasing amount of variable renewable energy sources. The operation of such systems can be optimized by stochastic programming, which anticipates the uncertainty related to the variable energy generation. However, these optimization problems can only be formulated for optimization horizons of a limited length (e.g., 24 or 48 h), due to the rapidly increasing problem size. In this work, we propose the resulting end-effect to be mitigated by valuation of the terminal stored energy level based on an electricity price forecast. We present results on a hybrid energy system, consisting of photovoltaic power generation and an energy storage unit, which trades electricity in the day-ahead market.