High-performance computing boosts uranium research

Consider the way air flows over an airplane wing, how slight variations in design allow the plane to take off or keep it on the ground.

The first airplane wings looked very different from what they do today, as engineers have come to understand the fluid dynamics of air flow and computer simulations have helped reveal how even a slight design change can optimize the performance of a dynamic system.

In the same way, use of supercomputers at ORNL to simulate fluid dynamics in gas centrifuges is revitalizing uranium science research in the United States.

Gas centrifuges are used to create low-enriched uranium for nuclear power but can also create highly enriched uranium for nuclear weapons. (Enrichment refers to how much of a quantity of uranium is the isotope uranium-235.)

The creation of physics-based gas centrifuge simulations to better understand centrifuge design for uranium enrichment is the primary purpose of a program called the Advance Simulation Initiatives for Nonproliferation Applications, or ASINA. These simulations will enable scientists and engineers to better inform policymakers regarding nonproliferation missions.

After Oak Ridge built the world’s first plutonium production reactor — the Graphite Reactor — during World War II’s Manhattan Project, the U.S. nuclear weapons program grew in capability and expertise for four decades. But, as priorities shifted following the end of the Cold War, that scientific advantage diminished. The scientists and engineers behind ASINA intend to get it back.

“Today, there are key gaps in our technical understanding of the performance and proliferation capabilities of certain centrifuge designs,” said Jared Johnson, Uranium Science and Technology program manager at ORNL. “ASINA and related efforts will close that gap and give the nation the ability to reliably analyze, assess and predict centrifuge performance and operations.”

Behind ASINA is a multidisciplinary team of engineers and scientists who understand computational science, fluid dynamics, multiphysics coupling, materials science and isotope enrichment. It takes these different areas of expertise to create a computer simulation capable of handling the various computational scenarios needed for gas centrifuge research — a mix of expertise readily available at ORNL.

“ORNL has historic strength in uranium enrichment, having the centrifuge program here for decades,” said Matt Bement, deputy principal investigator for ASINA. “With the HPC and code-development experience, Oak Ridge is the most logical place to do this.”

Using computers to simulate gas centrifuges is not new; code for modeling these centrifuges has been around since the 1970s. Today’s software engineers are dusting off the old code and boosting it with a much-needed upgrade to leverage the speed of high-performance computing and the intelligence of machine-learning algorithms.

“ASINA gives us an opportunity to explore things that we can't do or can't see experimentally,” said Franklin Curtis, ORNL Flow, Thermal, and Data Science Group leader, referring to the expensive and time-consuming process of building physical centrifuge plants and running individual experiments. “Having a better simulation capability gives us this new insight into how these very simple-sounding but complicated processes work.”

Results can be delivered within a few seconds or hours to collaborators anywhere in the U.S., and these new insights could help researchers and analysts better understand an adversarial nation’s ability to enrich uranium beyond the 3 to 5 percent needed for nuclear energy. 

2D and 3D models are among ASINA’s key outputs. Using computational fluid dynamics, the models answer questions about internal flow in a gas centrifuge. Simulations show different gas speeds, turbulence and pressure changes, as well as the resulting effects on internal gas dynamics. Researchers can run thousands of simulations at once, each with a tiny difference, to see how outcomes change.

2D modeling has long been part of the gas centrifuge research program and remains a significant part of today’s research, albeit with greater compute power. Small 2D calculations can be run on a desktop computer; however, more challenging tasks, such as quantifying uncertainty in predictions with multiple inputs, need significantly more power. Having the ability to run thousands of calculations on supercomputers can mean producing results promptly enough to effectively influence decision-making.

3D modeling is another challenge altogether, but the benefits of 3D simulation will be a gamechanger for those working to understand how objects actually move along multiple planes. Greg Davidson, senior researcher for the HPC Methods for Nuclear Applications Group, believes 3D will put high-fidelity modeling at the fingertips of centrifuge engineers. The challenge of 3D is that researchers need very powerful computers to run the calculations efficiently.

“When you add that extra dimension, the computer you need gets a lot bigger, and how you write your code has to be very thoughtful,” said aerospace research scientist Ryan Glasby. 3D calculations will be at least 1,000 times harder than 2D calculations, he said, and will require leveraging HPC to meet the demands of additional computational power.

In addition to ASINA’s implications for global nuclear nonproliferation, the program will improve domestic uranium capabilities, such as carbon-free nuclear power and medical isotopes. Engineers are diving into the future of gas centrifuge design mechanics and building the ability to identify weaknesses in materials or system design.

Gas centrifuges are costly. Simulating gas centrifuge behavior allows engineers to test designs before construction by conducting trials in a mock environment, accounting for risk and tolerance in the manufacturing process.

The ASINA program is one of two efforts ORNL is leading for the National Nuclear Security Administration’s Nonproliferation Stewardship Program. The other, the Uranium Science and Technology Center, is building unique laboratories and test beds that will allow the research community to advance the state of the art in uranium sciences. Together, ASINA and USTC will also focus on building the next generation of U.S. nonproliferation experts.

Flow, Thermal, and Data Science Group Leader Curtis and his colleagues are proud to build on the country’s uranium science legacy to revive and advance the field. They are working to take centrifuge technology, a consistent part of the nuclear science industry since the 1950s, and use it to advance uranium’s place in the peaceful use nuclear energy for economic and medical progress.

“We're setting the stage to allow the United States to become independent of isotopes from other countries and to be able to continue on the legacy that the people here at the lab in the past have done,” Curtis said.