Optimal Sizing of Battery Energy Storage Systems Driven by Clustered Load Profiles

The optimal exploitation and management of electrical energy passes through the possibility to store energy. Battery energy storage systems offer new, important potentialities to optimally manage electrical energy and to exploit renewables in a microgrid. Sizing the battery energy storage systems is however a critical point, as their total costs are still expensive. Researchers have tackled this aspect in deterministic and probabilistic frameworks due to the many sources of randomness that influence the battery sizing. Probabilistic sizing is, in principle, computationally intensive, as it requires several scenarios to be run to individuate the size ranges left to the decision maker. This paper provides a methodology for sizing battery energy storage systems, considering the randomness of load profiles in distribution networks, where scenarios are considered based on clustered load profiles which make the sizing procedure less computationally intensive. Particularly, load profiles are clustered through a dedicated methodology based on k-means algorithm, aiming at reproducing groups of days that share similar features. The clustered load profiles are used as inputs of the sizing methodology. Numerical experiments based on actual data show the effectiveness of the proposal.