Using IBM’s The Weather Company, Vector uses AI and data analytics to predict and preempt downed poles and lines.
As storms and hurricanes have increased in number and strength over the last 30 years, as the impact of climate change is felt, everyone has felt the effects. This includes drastic increases in the length and breadth of power failures. According to IBM, 70% of the disruptions worldwide are due to weather-related causes such as wind, rain and ice.
But what if there was a way utilities could know where the power failures would hit before they happened? What if utilities could predict and preempt downed poles and lines that can leave thousands without power for days at a time?
Thanks to IBM’s The Weather Company, this is now a reality for New Zealand-based Vector Energy. Its partnership with IBM has led to a reduced number of power disruptions thanks to artificial intelligence and data-driven analytics, IBM said.
“All utilities are impacted by bad weather, all of them care and all are trying to figure out how to handle this problem better, of how to prevent outages either by trimming the right branches at the right time or restoring power based on mobilizing crews faster to the right places,” said IBM Manager Robbie Berglund.
“Many utilities have used a traditional approach based on experience. What they realize is the old way of trying to solve problems when you’re in a region that gets a lot of severe weather is inefficient and not as effective. To continue the old ways of trying to simply look at the weather and decide is too much for the human brain to handle. That’s left a lot of room for computation and analytics to help improve and augment what they already do,” he added.
SEE: Artificial intelligence: A business leader’s guide (free PDF) (TechRepublic Premium)
Berglund explained that IBM has been working with Vector for about a year and a half and it took that long to make the tools fully operational. The IBM system helps primarily with two significant things: preventing disruptions with vegetation management and anticipating where line failures will be so crews can be deployed quickly and efficiently.
Such prediction has been around longer than vegetation management, according to Berglund, and IBM has worked on it with dozens of utilities across the world. IBM says this has delivered immediate cost savings because utilities can now avoid over-mobilizing in places that are not seriously impacted by severe weather.
Image: IBM
With prediction, utilities can also avoid a costly practice called “mutual aid,” which is when they borrow power from one another in the event that one is suffering from significant power issues. By knowing beforehand which areas will probably suffer the greatest number of disruptions, they can optimize their predictions and potentially avoid dipping in to mutual aid entirely.
“When you put a truck with four people in it on the road and multiply that by many trucks, if they’re not driven to or assigned to where the outages happen quickly, it gets expensive. If you can combine your decision making with data-driven analytics you will reduce the costs of your mobilization and you should restore the power faster. That’s what Vector is after,” Berglund said.
Vegetation management is something IBM has recently started working on. Using aerial footage from drones, airplanes or satellites, the company’s AI system is able to tell which lines are in danger of being disturbed by tree branches or other natural impediments that may bring them down in the event of a storm.
“Before the storms ever get there, we tell them, ‘here’s where you should trim in advance’ so that 12 months out, we can say where they should be trimming trees and vegetation and where the worst spots are,” Berglund added.
“That may sound easy, but when you have a utility that has 10, 20, 30, 40 or 100 thousand miles of power line, you’ve got to do more than guess and it’s really expensive to drive people around and try to assess where the worst areas are. So you can do that with AI and remote sensing with satellite imaging or aerial imaging and tell a utility where their worst areas are,” he said.
IBM’s AI system analyzes the historical performance of a utility’s lines and poles to see how storms of certain intensities will affect them. The company is now working with many utilities on these specific tasks, including Canada’s Hydro One.
Traditionally, utilities like Vector would simply rely solely on their employees’ expertise and decide what to do based on their past experiences. But Berglund said nothing IBM provides can replace the kind of institutional knowledge that longtime utility workers have.
“What I always tell utilities is, keep your approach of the guy down the hall who’s been there 30 years and he knows this storm is going to be a doozie. Keep that experience but augment it with a data-driven approach that has no emotion in it,” Berglund said.
“It is black and white modeling that performs pretty well. The two together is the best approach.”
Vector and IBM say they are seeing tangible results from their partnership. According to Berglund, the number and severity of power failures has decreased since the partnership began last year.
Though some utilities are choosing to do all of this on their own without the help of a company like IBM, many may find it is difficult and beyond their comprehension.
“Utilities usually find out they’re really good at being a utility but not as good at handling large, complex data sets or analytics and especially not artificial intelligence,” Berglund said. “They know that to do better, they need to use data and analytics. When somebody buys in, what I’ve seen so far is that they absolutely are changing the process.”
For most utilities, he said, the process starts about three days before a storm hits. About a week before a storm hits, Vector will get a notification with basic facts about the storm like whether there will be snow or how strong the winds will be. Based on this information, two days before a storm hits is when utilities decide how big their crews will need to be to handle downed poles and lines and what areas will potentially be hit.
IBM plans to work with Vector for years on this project and hopes to expand its work with other utilities. Berglund said almost every utility could benefit from a tool like IBM’s.
“We don’t see any utility that doesn’t need a data-driven analytic approach. Most of the utilities recognize that as well,” he said. “I really believe it’s just a matter of time before most or all utilities are leveraging data and analytics or AI to address these two problems.”
Tech News You Can Use Newsletter
We deliver the top business tech news stories about the companies, the people, and the products revolutionizing the planet.
Delivered Daily
Sign up today
Also see
IT leader’s guide to deep learning (TechRepublic)
Telemedicine, AI, and deep learning are revolutionizing healthcare (TechRepublic download)
Managing AI and ML in the enterprise 2019: Tech leaders expect more difficulty than previous IT projects (TechRepublic Premium)
What is AI? Everything you need to know about Artificial Intelligence (ZDNet)
6 ways to delete yourself from the internet (CNET)
Artificial Intelligence: More must-read coverage (TechRepublic on Flipboard)
How IBM uses Fortran and POWER9 in GRAF, the first “high-resolution, hourly forecast model” (TechRepublic)
Source: TechRepublic
