Evaluation of power outages in Connecticut during hypothetical future Hurricane Sandy scenarios
Monday, 14 December 2015
Poster Hall (Moscone South)
Reliable electric power is a staple of our modern society.The purpose of this work was to evaluate the occurrence of power outages under more intense, future Hurricane Sandy simulations in Connecticut. In addition, we also evaluated how many crews would be necessary to restore power in 7 days, and how different vegetation scenarios might contribute to a decrease in outages. We trained five pairwise models on each current Sandy runs (2012) as training using the random forest model (each validated using 10-fold cross-validation), and used each future Sandy run as an independent test. We predict that a future Sandy would have 2.5x as many outages as current Sandy, which would require 3.23x as many crews as current Sandy to restore power in 7 days. We also found that increased vegetation management might decrease outages, which has implications for both fair-weather and storm days of all types (i.e. blizzards, thunderstorms, ice storms). Although we have only evaluated outages for electric distribution networks, there are many other types (water supply, wastewater, telecommunications) that would likely benefit from an analysis of this type. In addition, given that we have the weather simulations already processed within our 2-km weather simulation domain, we would like to expand our vulnerability analyses to surrounding utilities in New Jersey, New York, Rhode Island, Massachusetts and New Hampshire to facilitate regional coordination among electric distribution networks.