Energy scheduling of a fuel cell based residential cogeneration system using stochastic dynamic programming

Abstract

Decarbonization of residential systems is one of the effective ways to achieve the low carbon goal. Due to the advantages of cleanness, efficiency and flexibility, hydrogen fuel cells play a key role in driving the transition to low carbon or carbon-neutral systems. This paper studies the optimal energy scheduling of an off-grid residential cogeneration system composed of a fuel cell, photovoltaic (PV) device, battery, thermal energy storage (TES) and heat pump. With the objective function being fuel cost, the economic optimization scheduling model is developed based on security constraints of devices and power balances of supply and demand. To handle the system uncertainties and achieve global optimality, the stochastic dynamic programming (SDP) algorithm is used, where the random variables are modeled as time-varying Markov chains. Simulation results under the typical winter day scenarios show that the SDP algorithm can achieve efficient scheduling while enhancing the device sustainability. Besides, comparison results with dynamic programming (DP) in scenarios with different uncertainty degrees demonstrate the optimality and robustness of SDP. Additionally, the impact of TES capacity on the system operation is also discussed, which has certain instructive significance for system capacity configuration in terms of economy and sustainability.