Data-driven planning for renewable distributed generation integration

Abstract

As significant amounts of renewable distributed generation (RDG) are installed in the power grid, it becomes increasingly important to plan RDG integration to maximize the utilization of renewable energy and mitigate unintended consequences, such as phase unbalance. One of the biggest challenges in RDG integration planning is the lack of sufficient information to characterize uncertainty (e.g., load and renewable output). In this paper, we propose a two-stage data-driven distributionally robust optimization model (O-DDSP) for the optimal placement of RDG resources, with both load and generation uncertainties described by a data-driven ambiguity set that both enables more flexibility than stochastic optimization (SO) and allows less conservative solutions than robust optimization (RO). The objective is to minimize the total cost of RDG installation plus the total operational cost on the planning horizon. Furthermore, we introduce a tight approximation of O-DDSP based on principal component analysis (leading to a model called P-DDSP), which reduces the original problem size by keeping the most valuable data in the ambiguity set. The performance of O-DDSP and P-DDSP is compared with SO and RO on the IEEE 33-bus radial network with a real data set, where we show that P-DDSP significantly speeds up the solution procedure, especially when the problem size increases. Indeed, as compared to SO and RO, which become computationally impractical for solving problems with large sample sizes, our proposed P-DDSP can use large samples to increase solution accuracy without increasing the solution time. Finally, extensive numerical experiments demonstrate that optimal RDG planning decisions lead to significant savings as well as increased renewable penetration.