In one of the coldest places on earth, this utility keeps the power on

Weather predictions helps prevent outages in New Brunswick

By | 3 minute read | November 9, 2018

Over three days in January 2017, freezing rain and ice pellets enveloped New Brunswick, Canada. Rows of jagged icicles grew on cars, signs, and buildings. Broken tree branches shattered as they hit the frozen earth. Layers of ice also smothered power lines, poles, and other infrastructure until they fell—causing widespread outages.

“New Brunswick Power experienced the most devastating weather event in our 100-year history,” said Tony O’Hara, CTO and VP Engineering at Énergie NB Power. “The severe ice storm knocked out power to 133,000 of our 400,000 customers.”

The extensive damage required NB Power to install 600 new utility poles, 150 new transformers, and 52 kilometers of new distribution lines at a steep cost of $30 million. The slippery conditions delayed repairs and made it incredibly dangerous for crews to restore power. But time was of the essence.

“Even a few days without power can cause huge economic and human impacts across the province,” said O’Hara. “Some types of industrial processes are very sensitive even to brief outages. And crucially, homes that depend on electrical heating are vulnerable to prolonged periods without power.”

As the primary electrical utility for New Brunswick, NB Power has to ensure its electrical supply keeps the province’s households warm and safe through winters that are among the world’s coldest. In 2015, temperatures plummeted to -30°C (-22°F)—in April. People can suffer from frostbite in 10 minutes if exposed to that temperature, and insulated interior pipes in homes can freeze and burst.

Utilities need to make decisions before a storm hits

With such unforgiving weather, repair crews need to quickly mobilize and get in position ahead of time to restore power faster and safer.

“If we believe bad weather is imminent, it’s essential to determine where we think our repair crews and resources will be needed most,” said O’Hara. “For an extensive restoration effort, we can be spending in excess of a million dollars per day on mobilizing people and equipment.”

To get ahead, NB Power needed to look ahead

NB Power had previously relied on manual analysis or its knowledge of past weather events to predict storm damage to its network. That was both difficult to execute well and relied on a handful of employees with years of experience. Its legacy systems were also built to manage outages after they’d occurred. It was challenging to make confident and fast decisions about where to station repair crews. NB realized that to get ahead of storms, it needed to look ahead.

“We decided to develop a more automated, objective approach to storm damage forecasting and outage prediction,” sad O’Hara. To help, NB Power turned to The Weather Company, an IBM Business.

AI and weather data: a match made in heaven

As more than 70 percent of all outages are weather-related, using a more accurate outage prediction model is key, especially, O’Hara said, as “it’s clear the severity of events is increasing.” Having detailed outage prediction models and data helps NB Power monitor weather forecasts three days in advance and then prepare and mobilize much earlier. NB can also quickly adjust its maintenance plans to avoid doing complex work when extreme weather is expected.

Similarly, if NB Power knows an ice storm is on the way, it can review its vegetation management schedules and identify and trim any overhanging trees before the weather hits. This improved operational efficiency also means utilities can be more sustainable, which is good for both their wallets and the planet. But to take advantage, they need to harness an enormous amount of data. That’s where AI comes in. By capturing The Weather Company’s data in great detail and analyzing it for patterns, AI can glean insights NB Power’s experts might miss. According to O’Hara, NB has trained its AI with data from 32 storms.

The initial “study session” is just the beginning, as AI can learn and be more precise over time the more data it ingests. But AI is already helping optimize resources across the province. “We’ve already experienced some significant storms this winter, and our decision-makers have been backing up their experience and instincts with hard data,” said O’Hara. On two occasions, outage prediction models gave NB the confidence to mobilize additional repair crews—and those extra resources made a real difference.

“We were recently able to reconnect 90 percent of our storm-hit customers within just 24 hours, which is a very positive result during the middle of the winter,” said O’Hara.