Fusion energy is the same process that powers the Sun and stars. While it could one day provide virtually “limitless” power, the intense heat presents engineering challenges. A collaboration between Commonwealth Fusion Systems (CFS), Princeton Plasma Physics Laboratory (PPPL), and Oak Ridge National Laboratory led to the discovery of an AI tool that reportedly speeds up the calculations vital to protecting internal components.
The tool is called HEAT-ML, and it rapidly finds safe havens on a fusion vessel’s interior walls that protect it from heat called “magnetic shadows.”
A New AI Approach is Speeding Up Calculations for Fusion Energy
Researchers believe the HEAT-ML could pave the way for software that accelerates the design of future fusion systems. The technology can reportedly assist in real-time decision-making during fusion operation. Researchers explain that it adjusts the plasma to prevent problems before they happen.
Explore Tomorrow's World From Your Inbox
Get the latest science, technology, and sustainability content delivered to your inbox.
I understand that by providing my email address, I agree to receive emails from Tomorrow's World Today. I understand that I may opt out of receiving such communications at any time.
“This research shows that you can take an existing code and create an AI surrogate that will speed up your ability to get useful answers, and it opens up interesting avenues in terms of control and scenario planning,” said Michael Churchill, the paper’s co-author and head of digital engineering at PPPL.
Heat from the plasma can reach temperatures hotter than the Sun’s core. This extreme heat poses significant risks to the integrity of a tokamak, a device used to contain the plasma. Associate research physicist Doménica Corona Rivera notes, “the worst thing that can happen is that you would have to stop operations” if the components are damaged. Researchers must predict where the heat will hit to avoid this.
The new AI simulates a specific section of a tokamak currently being built by CFS called SPARC. Researchers focus on a small part of the machine’s exhaust system that is subjected to the most heat. To understand the impact, researchers create 3D maps of the protected areas called “shadow masks.” The original open-source program used to perform these calculations. However, a single simulation would take as long as 30 minutes.
HEAT-ML reportedly overcomes this and reduces the calculation time to only a few milliseconds. It uses a type of AI that finds patterns by applying a series of mathematical operations. This deep neural network was trained using a database of about 1,000 SPARC simulations from the previous open-source program to learn how to calculate the shadow masks efficiently.
While HEAT-ML is tied to SPARC’s exhaust system, the team is optimistic about expanding its capabilities.
