• A probabilistic framework is developed for estimating town-level outage return thresholds using Generalized Extreme Value (GEV) analysis. • Spatial variability in outage risk is modeled using a Generalized Linear Model (GLM), revealing weather, land cover, and topographic factors as key drivers of power outage vulnerability. • The study helps power utilities identify highly vulnerable towns, explain the factors affecting increased vulnerability, and prioritize resilience investments using data-driven insights. Weather-related power outages are becoming increasingly frequent and severe, posing significant challenges for grid reliability and public safety. Such outages lead to significant disruption and economic losses, especially in regions with diverse environmental and geographic characteristics. This study aims to quantify town-level outage risks within Eversource Energy’s Connecticut service territory using return period thresholds and identify key environmental drivers of outage vulnerability. We collected outage data for 294 rain wind storms that affected the region from 2005 to 2023 and fitted Generalized Extreme Value (GEV) distributions using L-moments. Thresholds for the number of affected customers corresponding to multiple return periods were estimated and modeled using a Generalized Linear Model (GLM) based on geographic and environmental variables. Our analysis reveals differences in power system vulnerability across towns, with higher risk linked to factors such as dense canopy cover, high elevation, and stronger winds. By identifying the most vulnerable towns, the findings contribute to effective emergency preparedness and resilience planning. The data-driven approach followed in this study can also be adapted for outage risk assessment in other areas.
Munabia et al. (Sat,) studied this question.