Optimal operation of a residential photovoltaic/fuel-cell energy system using scenario-based stochastic programming

The operational degree of freedom in a residential energy system has dramatically increased recently because an energy system tends to be composed of many devices including some energy buffers. It is challenging to construct the framework of the operational planning problems, which update the plan based on the short-span predictions. This framework has to plan the operation before uncertain demand and PV output are realized; hence the plan is based on ex-ante decisions. The aim of this paper is to apply a stochastic programming framework to the operational planning of a residential energy system considering the prediction, and is to show basic directions for the planning. The residential energy system includes a fuel cell cogeneration system with a hot water tank, the photovoltaic system with an electrical battery. The parameters of the numerical experiments are three kinds of temporal precision and the number of predicted scenarios. The operation means the timing of the fuel cell's start-stop, and the stored energy levels of the battery and the hot water tank. As a result, the expected value based on scenarios was 21% greater than the minimized value based on perfect information in terms of daily primary energy consumption. The number of predicted scenarios was reasonable around 10 at 15-min temporal precision, because the great number of input scenarios made a decision not to operate the fuel cell cogeneration system for the entire day, and needed a great deal of the computational time.