Dive Brief:
- Utilities are beginning to use artificial intelligence to operate the electric grid more reliably and efficiently, but there are also challenges and risks to consider, including data privacy, cybersecurity and AI’s own energy use, experts told the House Energy and Commerce Subcommittee on Energy, Climate, and Grid Security on Thursday.
- AI’s energy use cases include load and weather forecasting, predictive maintenance, grid management, enhancing the output of wind and solar resources, faster storm recovery, wildfire risk assessment and methane leak detection, witnesses told lawmakers.
- Excluding China, AI represents 4.3 GW of global power demand today and could grow almost five-fold by 2028, Sreedhar Sistu, vice president of artificial intelligence offers for Schneider Electric, told the subcommittee. Of that demand, 30-45% is estimated to be in the United States.
Dive Insight:
“AI is already helping stakeholders better forecast power consumption,” Sistu told the subcommittee. Electric utilities and microgrid managers leverage the new technology “to forecast short-term energy consumption more quickly and efficiently across the entire geography of their grids.”
But while AI can be used to operate a more efficient electric grid, Sistu also warned of its growing energy demand. Global AI power demand, excluding China, is estimated to reach 13.5 GW to 20 GW by 2028, he said.
“With a grid that is already straining to meet existing demand, it is imperative this committee consider how strategic, future investments in physical infrastructure can support the growth of AI in America, which will in turn support the future of our grid,” he said.
The Electric Power Research Institute has been involved in more than 70 projects utilizing AI, said Jeremy Renshaw, the group’s senior technical executive for AI, quantum, and innovation.
“While automated grid operation may still be more than a decade away, current models may be useful in assisting grid operators as decision support agents to provide recommended actions to maintain grid operation,” Renshaw said.
The new technology can contribute to the safety, affordability, and reliability in energy generation, delivery, and consumption, he said. But EPRI “has found it also poses new challenges, such as cybersecurity risks, ethical considerations ... and data privacy and security.”
Paul Dabbar, the CEO of Bohr Quantum Technology and former DOE under secretary for science, said AI is “beginning to remake energy operations” but warned about the potential for security threats.
“National security can be placed at risk with the new hardware and software deployment,” Dabbar said. China can “place physical back doors on their chips, and holes in AI algorithms, to allow sabotage. The security challenges of this new AI and digital infrastructure are acute.”
Cybersecurity “has become an increasingly large concern,” Renshaw said, but he noted that AI has the opportunity “to be both an offensive and defensive force-multiplier for cybersecurity applications.”
As cybersecurity standards are developed for AI, Schneider Electric’s Sistu said Congress “should work alongside us to both ensure these standards support the government’s goals.”
Protection of consumer data “must also be a critical focus of AI regulation,” Sistu said. “Industry must apply and exceed standards for data protection and, like cybersecurity standards, should work alongside government to develop meaningful data privacy and protection standards for AI applications.”
Rep. Frank Pallone, D-New Jersey, expressed some reticence at the pace of AI adoption.
“It’s clear that AI will likely be a part of our energy future,” Pallone said. Utilized alongside distributed energy resources, the technology “can help us transition to a clean energy future. ... with that being said, we must proceed with caution. Guardrails must be put in place to ensure the adoption of AI is responsible.”
AI can help power plants run efficiently
Rep. Jeff Duncan, R-S.C., chair of the subcommittee, asked witnesses how AI could be used to improve operations in oil and gas production and in the nuclear sector.
AI can be used to improve the reliability and throughput of oil and gas producers, said Edward Abbo, president and chief technology officer for C3.ai. The technology can be used to increase production at refineries or from wells; reduce energy consumption in oil and gas production; and to survey assets in the field such as pipelines.
“Essentially, AI can be used across the upstream, midstream and downstream sectors,” Abbo said.
Generator performance can be improved by installing sensors on turbines and breakers, said Dabbar. Over time power plant operators can collect sufficient data to efficiently perform predictive maintenance and “drive up availability, making more power for cheaper cost. It’s more reliable. It’s already being deployed on certain types of power plants.”
