Combining optimization and simulation for microgrid sizing

The sizing of a microgrid is an important process to ensure safe, efficient and economical operations. It must consider a variety of technologies and the whole life of the project. Hence degradation and reinvestment must be accounted for. However, the control algorithm also influences the sizing results, especially if the microgrid is grid-tied, which offers several revenue streams and presents additional constraints such as time-varying connection limits. All this leads to a sizing problem that is large and non-convex. This paper presents a flexible approach to approximate the solution of this problem. The latter is decomposed in three nested sub-problems: the control and simulation of a microgrid for a given configuration and initial state; the evaluation of a configuration and control policy on a sample of representative days; the optimization of the configuration. Case studies highlight the importance of accounting for the control policy and device evolution along their lifetime.