The energy system is one of the most critical services. Natural disasters such as tropical cyclones, floods, and tsunamis can disrupt the system and cause power outages. Simultaneously, climate change can cause more frequent and higher-intensity hazards, which can result in longer-lasting outages. As the duration of the outage increases, it causes economic and societal losses and disrupts dependent critical infrastructure such as water and communications systems. Although many hazards can cause significant damage to energy systems, tropical cyclones are particularly dangerous. This paper aims to estimate power outages caused by tropical cyclones due to future climate scenarios up to six days before landfall for every six hours. The model builds upon McRoberts et al., 2018 static reduced network model. This 2-step outage prediction model first includes a binary classification that indicates whether an outage will occur at the census track level. In the second step, given an outage, it predicts the number of outages at each location. For further use of the model, our proposed model reduces the original set of predictor variables and uses only publicly available and accessible information, including hurricane-related characteristics, socio-demographic and environmental variables. This work estimates the fraction of customers without power in each census tract for each tropical cyclone considering changes in intensity and frequency caused by future climate change scenarios.

Reference McRoberts, D. B., Quiring, S. M., & Guikema, S. D. (2018). Improving Hurricane Power Outage Prediction Models Through the Inclusion of Local Environmental Factors. Risk Analysis, 38(12), 2722–2737. https://doi.org/10.1111/risa.12728