1. Home
  2. Energy
  3. Storage
  4. DOE eyes AI, machine learning to accelerate long-duration energy storage research – EQ Mag Pro
DOE eyes AI, machine learning to accelerate long-duration energy storage research – EQ Mag Pro

DOE eyes AI, machine learning to accelerate long-duration energy storage research – EQ Mag Pro

0
0

A proposed federal research program to accelerate research into the durability and performance of long-duration energy storage is a critical step to meeting the Biden administration’s decarbonization goals, speakers said Thursday at a Department of Energy (DOE) panel.

DOE officials said long-duration energy storage technology must be commercially ready, at scale, by 2030, in order to increase the share of renewables on the grid and meet the administrations 100% clean electricity by 2035 goal.

Most storage assets will be in use for at least 15-20 years, which means even installers and operators will not be able to validate their real-world performance over their full lifetime ahead of the 2030 timeline.

DOE has proposed a Rapid Operational Validation Initiative (ROVI) to accelerate testing and “validate technology faster than time runs on the calendar,” said Eric Hsieh, director for grid systems and components at the U.S. Department of Energy.

Dive Insight:

The Biden administration has put significant weight behind broad deployment of long-duration storage, defined as a system with at least 10 hours of continuous operation. In July, DOE announced a moonshot goal to reduce the cost of utility-scale, long-duration storage by 90% within a decade, backed by federal research, large-scale demonstrations and domestic manufacturing incentives.

Speaking Thursday at DOE’s long-duration storage summit, Deputy Energy Secretary David Turk said bringing long-duration storage to the grid wouldn’t just make it possible to rely on more renewable energy, but also “increase resilience and lower energy burdens” for vulnerable communities.

Ali Zaidi, deputy White House national climate advisor, compared the storage effort to the DOE’s SunShot Initiative, launched in 2011 to drop the cost of solar energy by 75%. Zaidi said that goal was achieved ahead of schedule, despite the fact that “some brilliant technical folks were skeptical we could make it there at all.” The storage ambitions, he said, would be equally achievable and will “put up a big, bright beacon out there for our most talented folks to step up and take that challenge.”

Although there have been technical breakthroughs on long-duration technologies — notably Form Energy’s July announcement of a 100-hour iron-air battery — experts have cautioned about the limited window to test batteries in the real world. Speaking Thursday, Craig Horne, managing director of energy storage for California-based Wellhead Electric, said there is a “limited amount of data” on both performance and production, which can hinder deployment. Electricity providers, he said, need confidence “in how these systems operate, no matter how the use case may evolve.”

ROVI, the proposed initiative from DOE’s national labs, seeks to close that information gap by using machine learning and artificial intelligence to model performance of different long-duration storage technologies, including predicting how the technology will lose performance or hold up physically over time. The initiative would rely on industry data and digital twins of the storage systems to model the long-term performance.

Brentan Alexander, chief commercial officer and chief science officer of insurance firm New Energy Risk, said that data can help “lower the soft cost” of installation by giving electricity providers more confidence in the reliability of technology. That, he said, is especially important given the long lifespan energy storage and the unpredictable weather conditions assets may be forced to endure.

“For the insurance industry, this is kind of a win-win,” Alexander said. “Our entire goal is to make these technologies possible,” he added, which means “narrowing uncertainty performances over time.”

Source: utilitydive

Anand Gupta Editor - EQ Int'l Media Network