Reducing the computational effort of stochastic multi-period DC optimal power flow with storage

Abstract

Due to the increase of intermittent renewable energy sources, it is becoming more important to consider renewable generation forecast error when solving the optimal power flow (OPF) problem. The stochastic OPF, which uses multiple forecast scenarios, generally leads to a lower cost compared to the standard deterministic OPF, which uses a single forecast. However, the stochastic OPF is computationally more demanding than the deterministic OPF. Both cost savings and computation times further increase when storage units or ramp-constrained generators are included, as they require solving a multi-period OPF problem. Our contribution is a hybrid method approaching the cost performance of the stochastic OPF while maintaining a computational burden close to the deterministic OPF. The method combines elements from both the stochastic and the deterministic OPF, and relies on Benders Cuts to interface them. Using a receding horizon approach over one year, we find that, based on eleven test cases, one version of our hybrid method leads to at least 70% of the cost improvement of the stochastic OPF, while the computation time increase is at most 40% of the stochastic OPF time increase. Furthermore, the computational advantage of our method increases with the system size. Two different versions of the method allow favoring of either the computational improvement or the cost improvement. We also identify directions for further improvement. Finally, our method can be used for more general problems in which one wishes to combine two models with different levels of complexity.