Research Highlight | Nuclear Technologies and National Security

Supercomputers deliver faster nuclear safety simulations
Supercomputers deliver faster nuclear safety simulations image
A Nek5000 HPC large-eddy simulation shows a turbulent jet of gas breaking apart a layered mixture of air and helium — mimicking the kind of dangerous gas buildup that can occur inside a nuclear reactor containment building after an accident like Fukushima (Image by Argonne National Laboratory).

When people think about nuclear safety, they may picture thick concrete walls and backup systems. But safety also depends on something less visible: knowing, in detail, how fluids and gases move during normal nuclear reactor operations and in rare accidents.

Researchers at the U.S. Department of Energy’s Argonne National Laboratory are using high-performance computing to make those predictions more accurate with advanced fluid dynamics simulations. Their focus is on turbulent flow, a hard problem for many older computer models. Turbulence is the chaotic, swirling motion you see in smoke, boiling water or a fast river. In a reactor or its containment building, turbulence can affect how heat is transferred and how gases mix.

The Argonne team uses two related simulation tools, Nek5000 and NekRS. Both are open-source computational fluid dynamics (CFD) codes that model how fluids and gases flow and transfer heat. Nek5000 runs mainly on central processing units (CPUs). NekRS is a newer version built to run efficiently on graphics processing units (GPUs). GPUs can handle streams of calculations at once, which can cut the time it takes to run large simulations. 

Originally designed for general CFD applications, Nek5000 and NekRS have been adapted to address nuclear-specific challenges, such as predicting hydrogen-related problems in reactor containment structures—a key issue highlighted by the Fukushima accident.

A major test for this work was started by a larger team through an international ​“blind benchmark” PANDA proposed through the Nuclear Energy Agency (NEA) of the Organization for Economic Co-operation and Development (OECD). In a blind benchmark, research teams get the same tank shape, materials and conditions, then predict what the flow will do before they see the experimental results. It is like taking a test without being able to check the answers. Argonne’s team demonstrated the accuracy and reliability of their tools albeit for shorter experiment runtimes.

That success helped start a collaboration with the Office of Nuclear Regulatory Research of the U.S. Nuclear Regulatory Commission (NRC). Their staff was interested in similar kinds of flows and containment-like geometries, where older, less accurate tools can struggle. Over time, NRC staff worked with Argonne to learn the approach and produce the simulations themselves. 

“Regulators need tools they can trust, especially for rare scenarios that are hard to test,” said Aleksandr Obabko, an Argonne computational scientist.

A key next step is moving more of these cases from Nek5000 to NekRS and optimizing them for the Argonne Leadership Computing Facility. This could help run the same kinds of safety studies faster, while keeping the physics details that regulators and designers need. The team is now preparing to run simulations on Aurora, one of the world’s most advanced supercomputers.

“By leveraging high-performance computing, we’re not just solving today’s problems—we’re laying the groundwork for the future of nuclear energy,” saidDillon Shaver, Argonne nuclear engineer. 

The implications of this work extend beyond safety. By improving the accuracy of simulations, Argonne’s tools can reduce the need for costly physical experiments and accelerate the regulatory approval process for new reactor designs. 

Looking ahead, the team plans to investigate how to integrate artificial intelligence and machine learning into their workflow. These technologies could further accelerate simulations and enhance predictive capabilities, making it easier for industry and regulators to adopt advanced nuclear technologies.

This research is part of the U.S. Department of Energy’s Nuclear Energy Advanced Modeling and Simulation (NEAMS) program, which aims to modernize the nuclear industry through innovative computational tools. With its focus on safety, efficiency, and collaboration, Argonne’s work is a testament to the power of science and technology to address critical challenges in energy and beyond.