Brookhaven National Laboratory will lead a three-year project funded by a $14.2 million investment to develop an artificial intelligence platform for electric grid management. The project will pool resources from institutions including Stony Brook University and companies such as IBM Research, alongside utilities LIPA and Con Edison, to rapidly model grid scenarios. Researchers aim to simulate billions of possibilities within 24 hours, a process currently taking months, to address challenges from fluctuating loads to integrating new power sources.

“The grid is an incredibly complex organism,” says BNL director John Hill. “It’s a big system that keeps us all alive, so you want to be able to protect that system.”

GridFM Model to Simulate Billion-Scenario Electric Grid Analysis

The $14.2 million project, spearheaded by Brookhaven National Laboratory, aims to dramatically accelerate electric grid analysis through the development of GridFM, an artificial intelligence grid foundation model. This new system will be capable of simulating billions of scenarios within a 24-hour period, a process that currently requires months of computation time for grid operators. Long Island Power Authority and Con Edison are directly collaborating with Brookhaven and IBM Research to provide crucial operational data and real-world system experience for the model’s development.

Hendrik Hamann, chief artificial intelligence scientist for innovation, science and security at Brookhaven Lab, explained that the project seeks to reframe the perception of artificial intelligence as a grid burden, instead positioning it as a potential advantage. The open-source development, facilitated through a partnership with the Linux Foundation Energy, will allow for broad accessibility and collaborative refinement of the GridFM model.

Researchers anticipate the system will be instrumental in quickly deploying new resources and adapting to fluctuations in electrical load, enhancing grid reliability and efficiency. During peak demand events, such as heat waves, current grid management strategies often involve load shedding to conserve energy; however, preliminary simulations suggest that GridFM-powered AI systems could reduce the need for these disruptive measures by more than 50 percent.

Artificial intelligence is often seen as a growing burden on the electric grid, but through the Genesis Mission, we have an opportunity to turn that challenge into an advantage.
Hendrik Hamann, chief artificial intelligence scientist for innovation, science and security at Brookhaven Lab

$14.2M DOE Project Integrates AI for Rapid Grid Adaptation

This capability represents a leap beyond current modeling techniques, which researchers say require months to achieve comparable analyses. The system’s development focuses on rapidly assessing the impact of new demands on the grid, optimizing the deployment of renewable energy sources, and maintaining overall reliability. Beyond immediate crisis response, GridFM is intended to support long-term grid planning, enabling operators to proactively address challenges and integrate new technologies. By providing utilities with enhanced artificial intelligence tools, the Department of Energy hopes to improve grid operation, reliability, and cost-effectiveness, ultimately supporting economic growth and affordable power delivery.

During a heat wave, you do load shedding to help conserve the energy so that you can keep the lights on or the cooling on in some areas, and [using artificial intelligence systems] they are able to reduce that [need to shut off customers] by more than 50%.

Source: https://www.newsday.com/long-island/politics/brookhaven-national-lab-ai-electric-grid-lipa-yt6cybrv

More like this