Researcher aims to take guesswork out of utilities' storm preparations

Johns Hopkins University assistant professor Seth Guikema is developing a model to predict power outages caused in major storms.
Johns Hopkins University assistant professor Seth Guikema is developing a model to predict power outages caused in major storms. (Kim Hairston, Baltimore Sun)

If a hurricane threatens to barrel up the Chesapeake Bay in the next month, many Marylanders will watch the forecast with dread. Memories of spending days without power after Hurricane Irene and the June derecho are fresh and painful.

Utilities such as Baltimore Gas and Electric Co. watch the forecast, too, readying forces to assess and repair damage to the grid. To many angry customers, though, utilities' preparations appeared to be lacking in recent storms.

But what if there was a way to accurately pinpoint how many power outages a storm might cause? Could it give utilities a clearer idea of how many crews to muster and where to deploy them? It's a solution one Johns Hopkins University researcher is exploring, and even talking to BGE officials about.

Seth Guikema, a Hopkins professor trained in risk assessment and management, is developing a model that can translate weather forecast data and U.S. Census records into an estimate of power outages. He previously built such a model for a Gulf Coast utility and now is crafting one that can be applied anywhere, ahead of any type of storm.

"We have this model we can run for any hurricane that's going to impact us from Texas up through Maine," Guikema said.

The idea is the researchers can work with any utility or emergency response agency along that stretch of coast to help them better prepare for storm-related outages, he said.

Guikema first worked on such a model with colleagues while on the faculty at Texas A&M University in 2005. They formed a partnership with what Guikema would only describe as a "large investor-owned utility in the central Gulf Coast region," compiling utility data on the numbers and locations of customers to feed the model.

The model takes 120 inputs in all, including soil moisture, drought information and topography. Then, when storms approach, the scientists can plug in forecasts of wind gusts to predict how many outages could result, and where.

Their research on that model was published in a scientific journal in 2009.

But now, Guikema is working on a model that can be applied more broadly. It pulls in census data on population rather than utility-provided data on the grid, and then also factors in wind speeds and rainfall. But it isn't as detailed as the original version.

That makes getting accurate predictions out of it a work in progress, Guikema said. Hurricane Irene, which caused 756,000 power outages in BGE's territory, was the model's first significant test, and it predicted many states within 10 percent of the actual outages.

In some areas, like New York City, variables like buried power lines made it less accurate. In others, like Rhode Island, detailed data on outages after the fact are difficult to obtain, making it hard to evaluate accuracy, he said.

This year, the researchers have been collecting more data on storms like Hurricane Isaac to better inform the model. The main challenge with that storm was translating its heavy rainfall into outages, something Guikema plans to work on improving. But if more utilities would provide more detailed data, that could help even more.

Guikema said he has had some discussions with BGE officials, but no plans have been made.

BGE officials would consider using the research to improve its preparations, spokesman Rob Gould said.

"We're aware of it," Gould said of the research. "If there's value there, we certainly would have interest."

BGE was heavily criticized for its response to the June derecho, which packed hurricane-force winds but did not provide the usual two- to five-day window to prepare that a hurricane offers. The storm caused 762,000 power outages in BGE's service area and more than 1 million across the state.

The model would be less useful for sudden storms like that, Guikema said, but it could be applied to snowstorms or other types of weather hazards in addition to hurricanes.

Paula Carmody, the state people's counsel who advocates for consumers, questioned the benefit to utility customers that could be gained.

"It's intriguing; I just don't know at this stage what it would do with regard to scenarios we have right now," Carmody said, referring to the growing number and intensity of recent storms.

But that issue is something Guikema hopes to tackle, too. The research has received a $400,000 grant from the U.S. Department of Energy to also explore the potential impacts of more frequent and more intense hurricanes.



Major Maryland power outages

June Derecho: More than 1 million homes and businesses across the state, including 762,000 Baltimore Gas and Electric Co. customers, lost power, some for as long as eight days into early July.

Hurricane Irene: The August 2011 storm caused 850,000 outages across the state, and 756,000 outages in BGE's territory. The longest extended for eight days.

Hurricane Isabel: About 790,000 BGE customers lost power for up to eight days after the September 2003 storm.

Hurricane Floyd: Nearly 500,000 BGE customers were without power for up to eight days after the September 1999 storm, prompting pledges from the utility and others to invest in improvements to prevent future extended outages.

Source: Sun archives