Precision Predictions: Computer Power To Forecast Storms, Harness Renewables

Oct 15, 2014

Vermont utilities have teamed up with computer scientists and weather experts to develop more precise forecasts of severe storm events. The project was unveiled Wednesday at a meeting of emergency planners in Waterbury. 

Tom Dunn is CEO of the Vermont Electric Power Company, which operates the statewide transmission grid. He says Vermont has seen an increase in the frequency and severity of storm events. Tropical Storm Irene washed away a half mile of power line. "In 2013, the Vermont Electric Co-op and Green Mountain Power spent over $22 million in storm response," he said.

The computer software project – based on a predictive forecast model called Deep Thunder, and developed by IBM – should help provide more precise early warnings of storms so utilities can mobilize line crews in advance of outrages.

The sophisticated forecasting software should also help boost the use of renewable energy in the region. "Renewable energy production has a strong dependency on weather; likewise, energy demand also depends on weather; therefore, high resolution, high accuracy forecasting will be a key enabler of the coming transition to clean energy," said Chandu Visweswariah, director of IBM's Smarter Energy Research Institute.

Take wind for example. Two summers ago, some wind projects in Vermont had to be ramped down because the regional grid operator couldn’t integrate their electricity into the grid even at times of high demand.

"What we don't have as a grid operator is visibility or predictability as to what those [renewable] resources will be producing on a sub-hourly basis... So I think as a grid operator this capability that we're talking about through the Vermont Weather Analytics Center is particularly exciting." - Tom Dunn, Vermont Electric Power Company CEO

Dunn said some local transmission bottlenecks were partly to blame in those cases. But he said the grid is always trying to precisely balance electricity supply with demand. And in general renewable resources can be harnessed more effectively if the operator knows in close to real time when and how much electricity will be available, and where the demand needs to be filled.

“What we don’t have as a grid operator is visibility or predictability as to what those resources will be producing on a sub-hourly basis,” he said. “It’s really what we need to both understand the value they’re providing and allow them to be more successful, in terms of seeing some of that value coming back. So I think as a grid operator this capability that we’re talking about through the Vermont Weather Analytics Center is particularly exciting.”

In the meantime, probably the most immediate outcome of the work is more exact forecasts of storm damage. Jason Shafer is associate professor of atmospheric science at Lyndon State Colleg. He used some computer generated animation to illustrate aspects of his research. The computer graphics showed a thunderstorm moving across Vermont in September 2013 and the number of places that lost power.

“So we have a line of thunderstorms moving through. And we have outages that occur. And so we’re doing analysis of events like this to try to build a model that will eventually better predict that,” he said.

The goal is to produce accurate forecasts up to 48 hours in advance and down to areas as small as two square kilometers across the state.