US lab adapts ChatGPT-like transformer tech to speed up nuclear reactor simulations

Researchers at the U.S. Department of Energy’s Argonne National Laboratory are using transformer-based models to speed up simulations of fluid dynamics in advanced nuclear reactors, potentially helping engineers study reactor safety and performance faster. Transformers are the neural network architecture behind many generative AI systems, but Argonne researchers are adapting the technology for a different task: modeling how fluids move and transfer heat inside nuclear power systems. The team is integrating transformer architectures into Argonne’s System Analysis Module (SAM), a simulation tool used to study advanced nuclear reactors. The goal is to improve turbulence modeling, which helps researchers predict complex fluid behavior and its effects on reactor systems. By combining transformer-based models with turbulence simulations, the researchers aim to produce results that are both faster and more accurate than conventional approaches. The work could eventually allow engineers to model entire nuclear power plants, including reactors, cooling systems and other supporting infrastructure. Transformers tackle reactor turbulence In nuclear reactors, understanding fluid movement is critical because fluids transport heat and interact with components throughout the plant. Turbulence modeling helps researchers predict these complex flows, but detailed simulations can require significant computing resources. Argonne’s transformer-based approach analyzes relationships between physical data points, including locations, velocities, and fluid flows. The model has already demonstrated high accuracy in representing resistance to fluid flow and heat transfer, two properties that are important for reliable turbulence simulations. “With AI, we can be as accurate as the complex methods and as fast as the simple methods,” said Rui Hu, principle nuclear engineer and manager of the Safety and Engineering Analysis Department in Argonne’s Nuclear Science and Engineering Division (NSED). “It is a union of accuracy and speed.” According to Argonne, AI-based models can produce simulation results almost instantaneously while maintaining the accuracy associated with more complex computational methods. That could allow researchers to run more realistic simulations without the same time and computing demands. The improved modeling could support work on reactor design, performance and safety. It could also help engineers evaluate how changes in fluid behavior affect the wider nuclear system. Digital twins enter nuclear simulations The researchers’ next step is to apply the model to simulations of entire power plants. These could include the reactor itself as well as cooling, safety, auxiliary, and other supporting systems. The team is also exploring digital twins, virtual representations of physical systems that can operate and update in real time. Argonne researchers say they are among the first to explore transformer architectures for digital twin technology in nuclear systems. Digital twins could eventually allow engineers to monitor and simulate nuclear facilities while incorporating real-time information from physical systems. Combined with faster simulation models, the technology could provide another tool for studying plant performance and identifying potential issues. Future work will focus on expanding the transformer-based models, improving their accuracy and flexibility, and integrating new AI-enabled capabilities into SAM’s simulation workflow. The project is supported by the U.S. Department of Energy’s Nuclear Energy Advanced Modeling and Simulation (NEAMS) Program. Recommended ArticlesGet the latest in engineering, tech, space & science - delivered daily to your inbox.With over a decade-long career in journalism, Neetika Walter has worked with The Economic Times, ANI, and Hindustan Times, covering politics, business, technology, and the clean energy sector. Passionate about contemporary culture, books, poetry, and storytelling, she brings depth and insight to her writing. When she isn’t chasing stories, she’s likely lost in a book or enjoying the company of her dogs.