In 2018, Alphabet CEO Sundar Pichai described artificial intelligence as “one of the most important things humanity is working on.” The technology, he said, is “more profound than, I dunno, electricity or fire.”
Back in 2018, Pichai’s prediction was met with skepticism. But five years later, nearly every sector of business is grappling with the peril and the promise of the technology. Since OpenAI unveiled the game-changing ChatGPT last November, Microsoft’s partnership with the AI developer has been regarded as one of the most successful strategies in the history of the company.
But will that transformative potential extend to the energy sector?
Many have speculated about how utilities could embrace AI, from data management to weather forecasting. And, at the margins, some of this is already underway. But whether a complete transformation is imminent remains an open question: utilities have long been fundamentally slow to adopt new software, and integrating AI to its fullest potential would involve a dramatic shift in how they operate.
Jesse Morris, CEO of the Energy Web Foundation, highlighted this tension in our live episode of The Carbon Copy in April: “I think it’s important to understand how analog a lot of our energy systems are right now. We actually have a data management problem on the grid side.”
Up to half of all new generation assets in the next decade could come from distributed resources. Morris lives in California and has solar, an EV, and heat pumps integrated into his home: “My utility has no idea I have those assets,” he said. “They don’t have the ability to collect data on them — let alone run some AI-based algorithm on them.”
Although many utilities are still grappling with the fundamentals of how to integrate AI, Andrew Hoffman, the chief development officer at Leap, said that the potential is real, and that it could have implications for decarbonization.
“I see [AI] as a key enabler for us moving to a more transactive grid,” Hoffman said, “where we have technologies throughout the grid — both at the bulk system and all the way at the grid edge — that are really solving this massive optimization problem and over time pulling power plants off the grid and pulling carbon out of the grid.”
There are some indications within the industry that 2023 will be a year of “significant digital progress” for power companies. But the reality is that these distributed resources will not succeed without a strong digital layer — and AI could help strengthen that layer. So where could AI actually make a near-term impact?
One area where AI is already beginning to transform the work of utilities is in data management. The technology is already “simplifying some of the more complex decisions that need to be made,” said David Groarke, managing director of Indigo Advisory Group.
For instance, Vistra Corp. turned to AI to monitor the hundreds of variables (steam temperatures, fan speeds, etc.) contributing to the thermal efficiency of its plants. After analyzing two years of data, the models were converted into an engine, reportedly saving the utility money and abating roughly 1.6 million tons of carbon per year.
On the distributed energy side, companies like Leap that work with utility data are finding an optimization problem that AI can help solve.
“We might have 20,000 smart thermostats and 5,000 EV chargers and 50 building management systems that are part of a virtual power plant,” said Hoffman. While Leap has long used machine learning and big data to estimate curtailment ability and grid resource capability for these assets, he added, more sophisticated AI is an “accelerator.”
And there’s a need for speed. According to the International Energy Agency’s projections, achieving net zero by 2050 will require renewables to make up 33% of global energy consumption by 2030; we’re on track for just 18% at present.
When it comes to physically managing new renewable assets, Groarke said AI could be hugely useful for identifying the right location to build, for instance, a new solar plant. Once utilities onboard those assets, he added, the use of AI for monitoring and management “effectively means there’s a better marriage of supply and demand of energy” despite their inherent intermittency.
As the grid becomes increasingly complex, there was an assumption that utilities’ own optimization and computing capacity would evolve alongside it, said Leap’s Hoffman. But those in the U.S. have only done an “okay job” of managing intermittency with new technologies. Meanwhile, other markets like Germany’s have kept their grids reliable even with much higher rates of renewable production.
This could be crucial because grid reliability and resiliency is an acute concern for major utilities. And it’s another place where AI could step in.
Climate change is worsening the reliability problems for the grid, as record heat waves spike energy demand and extreme weather like the storm that swept through Texas in early 2021 knocks out supply. Both Groarke and Hoffman cited the predictive capabilities of AI to improve utilities’ ability to weather these events, which are likely to become more frequent.
“Utilities are looking at different types of weather data, and they’re using more granular demand data on the customer side, and they’re able to forecast and dispatch and have the grid operate much more efficiently,” Groarke said. “And that’s not overhyped.”
There are hundreds of use cases for AI, which could work in tandem as utilities transition to managing a more complex grid. Per Groarke, a blue-sky vision of the future would involve renewables dispersed globally, with grid systems talking to each other, dispatching automatically, and integrating many types of third-party data, from weather to customer usage.
Data centers — representing about 1% of our global electricity demand — would be automating all this usage, “and operators are just sitting there eating doughnuts.”
It’s a compelling vision, especially as the latest global warming estimates remain dire. But it’s one that is “many years, if not decades” away, Groarke said. So far, utilities have made only incremental improvements each year when it comes to integrating AI.
The digital tools exist to ease the path to decarbonization, if only traditionally slow-moving utilities are willing to make full use of them. For instance, Hoffman said, utilities could be using AI to fully take advantage of the battery capacity of new EVs being added to the grid. But instead — fearful of sacrificing reliability to the transition — they may end up replacing one-for-one today’s large, high-emissions power plants with huge battery plants, overbuilding and overinvesting in the process.
“That’s not an elegant way to solve the energy transition,” said Hoffman. “And now we have the tools at our disposal, so I think we need to put them to good use.”
This story was originally published on the Post Script Media website on May 25, 2023
