Duke Energy customers frequently experience electrical power outages caused by thunderstorms that produce strong surface winds and subsequently damage power distribution infrastructure. To better anticipate such outages, forecasters would benefit from outage-focused guidance regarding storm strength and timing so proper outage-mitigation can be initiated prior to each event. This study identified meteorological parameters that could best predict total daytime and nighttime power outages across each Duke Energy service area using generalized linear models (GLMs). A total of 392 event dates were stratified with regard to five service areas, two seasons, six convective modes, and three dominant severe weather types. Daytime and nighttime GLMs were attempted for each stratification with more than 10 event dates and the resulting GLMs were deemed "operational" if predictors remained statistically significant (p-values < 0.05) for ten trials using a randomly-selected portion of the training dataset. In total, 58 operational GLMs were developed using 24 unique parameters as predictors. The most common predictors were low-level vertical velocity and composite parameter used to predict large hail. The resulting operational GLMs can be implemented within an ensemble framework to provide Duke Energy with a total outage estimate for each service area.