Are Utilities Prepped To Deal With A Slew Of Hurricanes And Wildfires?

On June 28, 2024, Hurricane Beryl began to form, initially sending mixed signals about its path. In the early days, many believed it would strike Mexico’s eastern seaboard. However, the storm took a right turn and headed straight for Houston, Texas. To deal with this, CenterPoint Energy gathered 4,000 utility workers. The storm was so powerful it knocked out power to 2.2 million customers—about 80% of those it serves. Sadly, 73 people died, and the storm caused more than $8 billion in damages.

Extreme weather is now typical. Artificial intelligence is like a weather-predicting superhero for utilities. Power companies can calculate when harsh weather is coming by analyzing old weather data and satellite pictures. This intelligence lets them take action, like positioning crews in the right places and ensuring they have the right equipment. If the AI is right, it can prevent many headaches—for utilities and customers alike.

The pressure to maintain the grid’s integrity and ensure a consistent power supply will only increase. Consider the goal of electrification and its demands on network operators: the need for more wind and solar energy, a growing number of electric vehicles, and the emergence of power-hungry AI and data centers. Additionally, the current grid is outdated and lacks self-healing capabilities. Experts estimate that to modernize the system, we need to invest $1 trillion by 2030.

With these frequent and severe storms, how can we efficiently respond with limited resources? How can we equip field workers with the best tools to restore power? I spoke with Amazon Web Services, Consumers Energy, and Hitachi Energy experts. They made the point we have all heard for years: utilities are reluctant to take risks and get ahead of the curve when allocating capital. Nonetheless, more and more of them understand they must break free from their silos and invest in cloud services that give them a holistic view of what is happening in their territories.

“Does better planning lead to better outcomes? If you have a plan going into battle, you’ll be better armed, even though that will change. Mike Tyson said, ‘Everyone has a plan till you get punched in the face,’” says Jeff Pauska, global product product director at Hitachi. “So, at the end of the debate, a utility may not have the best information. You have a lot of siloed systems and processes, and that’s creating risk.”

Give Decision Makers More Info

We’ve seen the wildfires, hurricanes, and floods. Crisis management teams meet and send workers to climb poles and fix wires. Undoubtedly, they need boots on the ground. But the utilities also must have satellites up high and drones flying closer to where the damages are—the 21st Century of dealing with disasters. Those devices collect data and store it in the cloud, giving the decision-makers more knowledge.

Imagine power companies having a supercomputer that can analyze thousands of pictures of transmission and distribution systems. It’s like having a team of experts that can run through all sorts of scenarios and potential patterns, something that’s impossible for humans to do.

For example, AI uses wind speed, humidity, and temperature to predict wildfires and equipment breakdowns. In California, strong winds and droughts increase the risk of wildfires, so network operators can turn off power lines in high-risk places.

“To scale up digitally, you must centralize all of the data. Then, you can use machine learning”—exercises that let utilities run through thousands of potential possibilities so they can effectively respond to disasters, says Howard Geffen, general manager of energy and utilities at Amazon Web Services. “This will help them manage their assets and crews and predict where they need to be, allowing them to focus their resources accordingly.”

Consider ISA Transelca, a transmission company in the Barranquilla region of Colombia, South America. What happens if someone reports a fire that is hard to reach because of mountains and dense forests? Does the company take a wait-and-see approach, or does it use the latest technologies to predict the fire’s travel pattern and, thus, minimize devastation?

“Now we scan the globe with 28 satellites,” says Geffen. This system sends updates every 90 minutes, helping the utility track weather patterns and predict the fire’s path. It starts by using technology and AI to reach the fire before it gets too big.

Taking A Holistic View

Consumer’s Energy, which serves 2 million customers across the Lower Peninsula of Michigan, faced a severe ice storm at the end of March. The event caused widespread damage, with downed power lines as well as fallen trees and utility poles. It looked like a hurricane had swept through. For the utility’s part, it mobilized teams and resources to get the power back on. How?

It’s moving all its data to the cloud, allowing utility workers to transcend their departments and get a company-wide view of the territory. In all cases, its goal is to get the lights on within 24 hours.

Aaron Rajda, Consumers Energy’s chief data officer, explains that machine learning models now consider workforce numbers and equipment levels. By analyzing this data, the system can predict when customers will regain their power. Unlike relying on team members at the HQ to make guesses, AI is always up-to-date and accurate.

Rajda says the ice storm in March was the worst in a century. And guess what? The next day, thunderstorms and winds of 96 miles per hour hit the region, causing even more destruction. Data collection, storage, and analysis are not rigid processes—one-size-fits-all for each weather event. They are constantly adjusted, with the “goal of making it better each time.”

Consumer’s Energy high-voltage distribution system stretches across thousands of miles of Michigan. It’s tough to keep an eye on it all. But they use helicopters to take videos and drones to collect data. The company has even trained an AI model to make smart decisions and ensure things get fixed before they break. During the big ice storm, the drone pilots sent the data to mobile command centers, which directed field crews and ensured they had everything they needed.

Can utilities afford to make these investments in technology? Rajda says the worst-case scenario is to spend $5 million and learn months later that it didn’t work. Consumers Energy has a system in which it will test technologies for five weeks to validate their assumptions and see if the tools will deliver positive results. He said the cloud model has proven itself, allowing the utility to be proactive and save money.

“The point is that there’s a whole lot of data in the system, and it needs to be put in one place, managed, accessed, and optimized,” adds Geffen with Amazon Web Services. It’s in everyone’s interest to limit the devastation.

Utilities have generally entered the 21st Century. Many are making these investments, sped up by the fast pace of electrification and the spread of machine learning.