249 Diagnosing the Potential for Thunderstorm-Induced Power Outages with the Rapid Refresh Model

Monday, 7 January 2019
Hall 4 (Phoenix Convention Center - West and North Buildings)
Rachel Cucinotta, Univ. of North Carolina at Chapel Hill, Chapel Hill, NC; and M. D. Eastin

Diagnosing the Potential for Thunderstorm Induced Power Outages with the Rapid Refresh Model


Department of Geography and Earth Sciences, University of North Carolina at Charlotte,

Charlotte, North Carolina


Duke Energy Corporation

Charlotte, North Carolina


Severe thunderstorms consisting of strong winds and lightning are frequently disturbing power transmission and distribution lines, resulting in customer’s loss of power for an extended period of time and expensive repair crew costs. Pre-existing weather models leave mitigation decision based on insufficient knowledge on timing, location and severity of the storms and often result in two types of scenarios. The first insistence is a “short fuse” event, where an inadequate number of repair crews are on call for the large volume of outages and the second insistence being a “non” event, where repair crews are overstaffed for the number of outages that have transpired. An in-depth, statistical study of how environmental conditions vary during different event types has been developed for three distinct regions (Midwest, Carolinas and Florida) in an attempt to help energy forecasters to discriminate between the different event types and to better pre-position repair crews in a timely manner. Hourly output from the Rapid Refresh Model (RAP) was used as a proxy for local environmental conditions. A multitude of severe weather diagnostic parameters were then extracted from the RAP grid cells contained within each region, along with the total number of reported thunderstorm-induced power outages. Well-correlated parameters will then be used as predictors in a multiple regression model designed to identify locations most susceptible to power outages. Details concerning the data, methods, optimal parameters, and predictive models will be presented at the conference.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner