Sizing of energy storage systems considering uncertainty on demand and generation

Abstract

Over recent years there has been a general consensus about the necessary changes towards modernizing the existing power grid to meet environmental and socio-economic objectives. The adoption of low carbon technologies is a milestone in this process. On the other hand, the massive and uncoordinated connection of distributed generators (e.g. solar or wind) is making the operation of electrical distribution networks more challenging, e.g. causing energy balancing problems or voltage violations. Energy storage systems represent a possible means to cope with these issues. In this paper, we consider the problem of sizing the energy storage systems installed in a low voltage network with the aim of preventing voltage violations along the feeders. Since the problem is solved at the planning stage, when future realizations of demand and generation are unknown, we adopt a two-stage stochastic formulation where daily demand and generation profiles are modelled as random processes. The cost function to be minimized takes into account installation and operation costs related to storage use. By taking a scenario-based approach, the two-stage problem is approximated via a multi-scenario optimal power flow. To reduce the computational burden of the latter problem, a heuristic strategy consisting of solving separately a sizing problem for each scenario, and then combining the solutions of the single problems through a worst-case criterion, is proposed. The multi-scenario approach and the heuristic strategy are compared in terms of both computation time and quality of the solution using real data from an Italian low voltage network with photovoltaic generation

Optimal sizing of energy storage systems under uncertain demand and generation

“…On average, the relaxed version of the single-scenario problem (21) is solved in about 2 min. As long as it can be solved, the computation time of the SDP relaxation of the multi-scenario problem (20) grows almost quadratically with the cardinality of the set D [35]. For instance, it takes about three hours for a set of 10 scenarios.…”

“…Finally, numerical results are illustrated in Section 5, and conclusions are drawn in Section 7. A preliminary version of this work appeared in the conference paper [35], where simplified models of the network and of the storage were considered, and an upper bound to the optimal cost of the multi-scenario problem was derived by finding a suitable feasible solution. Moreover, compared to [35], the case study worked out in this paper involves a more realistic test network of larger size.…”