Towards Tractable Stochastic Unit Commitment for Preventive Operation during Hurricanes

Hurricanes cause power outages in the coastal areas of the United States every year. Previous studies show that the integration of weather forecast data within power system operation can substantially reduce such outages. However, preventive power system operation, modeled as a stochastic unit commitment, is extremely computationally burdensome, and, thus not applicable to real systems. The present paper aims to address this challenge by developing a computationally-efficient stochastic unit commitment model. The proposed algorithm proactively schedules generation units during hurricanes, taking the hurricane-induced damages into consideration. The proposed model was able to solve a large-scale 2000-bus Texas system in 7 hours, achieving an acceptable level of computational tractability. In the simulated hurricane, the proposed model was also able to avoid 80% of the power outages by only adding 5% to the dispatch cost. The results confirm the viability of stochastic unit commitment as a preventive operation tool to reduce the power outages during hurricanes.