Predicting Outages

Severe weather is the leading cause of damage to the overhead electric distribution grid. With advanced planning and state-of-the-art research and technology we are predicting storm severity, path and impact on the electric distribution system.

The UConn Outage Prediction Model (OPM) forecasts a storm’s impact, which Eversource combines with meteorological data to proactively pre-stage crews and expedite power restorations. The OPM provides an up to three-day advanced picture of a storm’s anticipated impact, updated every six hours, and is a leading-edge approach in the electric industry.

Outage predictions, along with proactive tree and forest management, are providing the greatest benefits for utility customers by avoiding and shortening outages, and enhancing electric system reliability.

Project Goals and Updates

The primary scope of the proposed project is to develop the next generation of storm-based damage forecasting by investigating the impact of weather forecast variability and different statistical models. We are significantly improving the functionality and ability of the system to predict outages,  better characterizing the confidence of weather forecasts, enhancing a utility’s ability to conduct “what if” weather vulnerability studies to the overhead distribution network in extreme events, and implement the system on an operational basis facilitated by a website and fed with real-time weather forecast data.

We are accounting for differences in geographic variables (i.e. tree conditions, soil conditions, elevation) and improving the characterization of vegetation conditions (i.e. leaves on tree) results to improve model performance.

Our outage prediction research has expanded to include customers in western Massachusetts and central-coastal Connecticut, and we look forward to expanding our research to surrounding utilities to improve the region’s emergency response.

• Over 90 percent of power outages during storm events in Connecticut are tree related
• UConn has developed an Outage Prediction Model that integrates high-resolution weather predictions (winds, heavy rain, ice, snow, flooding) with vegetation characteristics (height, density, leafs on/off) and other geographic data to accurately predict damages on the electric grid and to prepare for and respond to damages.
• 3.6 million electric and gas customers across three geographic regions are benefiting from UConn’s Outage Prediction Model. Currently, UConn is working with Eversource Connecticut, Massachusetts, and New Hampshire and United Illuminating in central coastal Connecticut.
• The Outage Prediction Model performs accurately for a range of storms including tropical systems (tropical storms and hurricanes) and winter storms (excludes ice storms).
• Our weather forecast model, the Weather Research and Forecasting Model (WRF), runs on seven Haswell nodes (Intel x64; 24-core nodes with 128GB RAM each) with a total scratch storage of 620TB. This computing power is equivalent to 168 laptop computers performing calculations at the same time, and the scratch storage is large enough to store 10,540,000 hours of music!

Team Members

• Emmanouil Anagnostou, Professor of Civil and Environmental Engineering, University of Connecticut
• Diego Cerrai, Assistant Professor of Civil and Environmental Engineering, University of Connecticut
• Xinxuan Zhang, PhD Candidate, Department of Civil and Environmental Engineering, University of Connecticut
• Peter Watson, PhD Candidate, Department of Civil and Environmental Engineering, University of Connecticut
• Aaron Spaulding, PhD Candidate, Department of Electrical and Computer Engineering, University of Connecticut
• Sita Nyame, PhD Candidate, Department of Civil and Environmental Engineering, University of Connecticut
• Shah Saki, PhD Candidate, Department of Civil and Environmental Engineering, University of Connecticut
• William Taylor, PhD Candidate, Department of Civil and Environmental Engineering, University of Connecticut
• Guannan Liang, PhD Candidate, Department of Computer Science and Engineering, University of Connecticut