Investigating the impact of stochasticity in microgrid energy management

Abstract

Various stochastic programming methods have been used to account for penetration of uncertain renewable energy generation in microgrids. However, these stochastic methods may be unnecessary. Energy storage combined with rescheduling based on a rolling time horizon gives a microgrid powerful tools to adapt to any unexpected events. Add to that the natural tendency to over-engineer new systems and one begins to wonder how much value can be gained by stochastic optimization over deterministic methods. We investigated this question by looking at an existing residential microgrid in Hoover, AL. We compare various stochastic approaches for scheduling with deterministic approaches and show that there is little value of using stochastic programming. Instead, we find that considering longer time horizons is a better use of computational resources.