Two-stage robust energy storage planning with probabilistic guarantees: A data-driven approach

Shorter-term (e.g., hourly) uncertainties, which are not explicitly accounted for in conventional power system planning practice, become imperative in the longer-term planning with deepening penetration of renewable energy resources. This paper addresses this central issue in power system planning: the challenges induced by the increasing short-term and long-term uncertainties and the pivotal opportunities from the rapidly growing flexible resources (e.g., storage devices). By leveraging the abundant operation data, we propose a data-driven power system planning framework based on robust optimization and the scenario approach. The proposed framework considers a broad range of operation conditions and provides rigorous theoretical guarantees on the future risk of planning decisions. By connecting two-stage robust optimization with the scenario approach theory, we show that the operation risk level of the robust solution can be adaptable to the risk preference set by planners. The theoretical guarantees hold true for any distribution, and the proposed approach is scalable towards real-world power systems. Furthermore, we show that the column-and-constraint generation algorithm, which is a popular algorithm to solve two-stage robust optimization problems, is capable of tightening theoretical guarantees. We substantiate this framework through a planning problem of energy storage in a power grid with significant renewable penetration. Case studies are performed on large-scale test systems (modified IEEE 118-bus system) to illustrate the theoretical bounds as well as the scalability of the proposed algorithm.