Effective Scenario Selection for Preventive Stochastic Unit Commitment during Hurricanes

In 2017, four hurricanes made U.S. landfalls, leading to millions of customer outages. Our previous work shows that weather forecast can be used to estimate the failure of transmission lines during hurricanes; these failure estimations can be effectively used in stochastic optimizations and guide preventive operation to reduce outages. However, the large number of possible contingency scenarios, caused by hurricanes, makes preventive operation extremely computationally burdensome. The problem can be practically solved with only a small number of representative scenarios. Thus, the effectiveness of preventive operation would directly depend on the scenario selection process. This paper examines two scenario selection methods, which eliminate (a) the unlikely and (b) the inconsequential scenarios. Simulation studies were carried out on IEEE 118-bus system, mapped to the Texas transmission network, using Hurricane Harvey wind data. The paper sheds light on the effective selection of an appropriate number of scenarios with acceptable computational complexity.