Power System Operation and Planning under Uncertainty with Robust Optimization

Recently, US government policy gives more incentives to renewable energies and is working towards more integration of renewable energy sources into current power system. This poses new challenges to system operators due to the intermittent nature of the renewable energy sources. This thesis presents a collection of research studies in power system operation and planning under uncertainty. The studies investigate the use of robust optimization and distributionally-robust optimization approaches to the power system planning problems under uncertainty, specifically, unit commitment problem and network reconfiguration problem.;In the first chapter, we propose an acceleration techniques for solving two-stage robust unit commitment problem with consideration of full transmission line constraints. The key idea of this research is to include only critical transmission line constraints dynamically as we solve the two-stage robust optimization models. We conduct computational studies for the modified IEEE-118 bus system and show that the proposed approach can greatly reduce the total solution time of the two-stage robust unit commitment problem. In the second chapter, we solve the distributional network reconfiguration problem with the load uncertainty using the robust optimization approach. Computational results for 5 test cases that are studied in the literature are provided. We find that the reconfiguration solution from the robust optimization approach is reliable and robust against all load scenarios considered in the uncertainty set. In the third chapter, we research the distributionally-robust optimization approach and solves the unit commitment problem using the proposed distributionally-robust optimization models. The distributionally-robust optimization models provide less conservative solution than the robust optimization models and are free of committing to a specific distribution as in stochastic optimization. We present the framework and algorithm to solve two-stage distributionally-robust optimization. We conduct computational study of the distributionally-roust optimization model using 6-bus and 118-bus system.