Analytical Uncertainty Propagation for Multi-Period Stochastic Optimal Power Flow

Analytical uncertainty propagation and storage usage in a high RES Turkish transmission grid scenario

“…For the optimization of generation in the face of large wind power outputs and storage, we use the second-order cone problem (SOCP) presented in [4]. This optimal power flow problem is based on a DC power grid, and can handle uncertainty using decision variables modelled with Gaussian processes (GPs) and chance-constraints (CCs).…”

“…As wind energy depends on natural wind resources that are only to some extent predictable, we have to include uncertainty. We choose Gaussian processes (GPs) for modelling, as they yield sufficient representation of smoothed time series and enable our analytic reformulation of the optimization problem [4]. GPs are well-suited to model time series, both smooth and volatile [14].…”

“…One such major change can be distributed storage, a widely used strategy complementing RE integration [9]. There are plenty of optimization models already including storage (see e.g., [4]). However, they are usually evaluated on test grids and simulated data.…”

“…This paper addresses this problem by providing a real-world transmission grid of Turkey, combined with real-world wind data. On this setup we apply our previously presented second-order cone problem formulation [4] and evaluate the optimal placement of distributed storage.…”

“…The contribution of this paper is then threefold: (1) We provide the Turkish transmission grid as a real-world power grid to the research community. (2) We test our previously presented SOCP method [4], that is convex, tractable and analytically exact, on this Turkish transmission grid using real-world wind data for forecasts of wind power. (3) We evaluate a scenario of wind production that is realistic in terms of resources and demand and show that storage plays stabilises the Turkish transmission grid and the storage units are best placed close to demand.…”