Book Description
A significant number of people worldwide live in rural villages with only part time electrical power or no power at all. In an effort to address this deficiency, the U.S. National Renewable Energy Laboratory (NREL) is developing hybrid renewable power systems composed of photovoltaic panels, wind turbines, battery banks, and diesel generators to deliver 24 hour power. In keeping life cycle costs of these systems to a minimum, an appropriate dispatch strategy can contribute as much to reducing cost as the proper choice of system architecture. Current dispatch strategies consider only current net load and current state of charge of batteries. The framework developed here considers future net load realizations in determining optimal dispatch strategies. The model is a Markov Decision Process solved using a Policy Iteration algorithm. The algorithm produces a policy that maps each state that the hybrid renewable power system can occupy to an optimal action. The dispatch policies consider immediate and long run ramifications of an action on life cycle cost. The results show that over different values of diesel fuel cost, battery wear cost, and loss of service penalties the optimal strategies minimize the long run average cost. Furthermore, minimal implementation costs make the optimal strategies an attractive alternative to heuristic strategies that are currently used to dispatch hybrid renewable power systems.