Ten years ago, the Department of Energy put out a call for innovators to change the world of nuclear energy.
What DOE hoped to accomplish with the then-new Energy Innovation Hubs concept was “translational research” – research and development on an accelerated timeline that could solve the problems facing the nuclear industry, not only extending the life of the current reactor fleet, but also paving the way for more efficient next-generation reactors.
Those solutions would then go straight to industry as quickly as possible. DOE was willing to put $125 million toward a “hub” for at least five years to see that happen.
The Oak Ridge National Laboratory-based Consortium for the Advanced Simulation of Light Water Reactors – a national collaboration of top scientists and engineers from government, academia and industry who had the privilege of making up DOE’s first Energy Innovation Hub – showed enough success that DOE renewed its funding for a second five-year period.
The consortium wrapped in June, having solved some of the biggest nuclear reactor challenges, and is handing industry a comprehensive software suite with the tools and support to use it immediately and on an ongoing basis.
“We’ve come through on the bet,” said former CASL director Doug Kothe.
Industry buy-in was a critical element of the hub’s success, said original CASL member John Turner. The Tennessee Valley Authority, Westinghouse and the Electric Power Research Institute were partners from the start.
“That was a key part of the project, having industry involved early on, but we didn’t have to convince them how valuable this was,” Turner said. “They were coming to the table saying, ‘We need help here.’ Industry was recognizing the gaps themselves, and they respected our expertise and were motivated to collaborate with us.”
Other partners included Idaho, Los Alamos and Sandia national laboratories; Massachusetts Institute of Technology; the University of Michigan and North Carolina State University.
“It was an ambitious undertaking,” said Dave Kropaczek, a chief scientist with CASL who became its director in 2018 and was an early member of the industry council. “It had a huge scope. I was skeptical but curious.”
With hundreds of scientists and engineers at the top of their field working together, the consortium set a goal to develop broad capabilities to:
- Accurately predict and reduce instances of undesirable boiling conditions, thereby increasing fuel performance and core power. An example was departure from nucleate boiling, the point at which a steam blanket forms on the fuel rod surface, insulating it and reducing heat transfer rapidly.
- Predict and manage “crud,” which are deposits that form on fuel rods that can shorten their efficiency and lifespan, increasing the cost of power.
- Predict fuel pellet and cladding integrity during normal operation and postulated accident scenarios, giving power plant operators greater flexibility in when and how much power is produced.
- Predict how neutrons interact with large reactor components to provide a guide for which materials are likely to degrade on what timeline, as well as to help reactor owners decide when to replace parts for improved performance.
How would they do that? By developing an exceedingly accurate virtual reactor.
‘The modern era of simulation’
The field of modeling and simulation wasn’t new to the industry; it had been part of nuclear engineering for decades.
The challenge, though, was bridging the gap between its current capabilities and its possibilities, Turner said.
“At the time, industry had become more followers than leaders in simulation; they were used to lower-fidelity, lower-confidence simulations,” Kothe said. “We opened their eyes to the possibilities and brought them into the modern era of simulation. We demystified the simulation technology. It wasn’t a black box; they were part of the development, and they could roll up their sleeves and go in there and see that it’s not a bunch of smoke and mirrors. They saw that this tool was for everyone and that the staffers involved were talented and committed and listened to them.”
The stakes were high. Nuclear produces roughly 20 percent of the total U.S. power supply but more than half its carbon-free electricity. While the country’s demand for power is expected to increase by at least 25 percent by 2030, the average age of the U.S. nuclear fleet is close to 40. As of last year, 17 reactors at 16 sites were in various stages of decommissioning, yet only one new reactor has gone online in the U.S. this century. Extending the life and efficiency of these older, existing reactors meant buying time and power until the next generation of reactors is developed and put into service.
Gil Weigand, then CASL’s startup manager, “pushed us very hard to release a Version 1 software package after only one year,” Turner said. “If we looked back, we’d probably be pretty underwhelmed with what that was. But it was still a big achievement to rise to Gil’s challenge and release a software package after only a year.”
Four years after that first release when CASL’s Virtual Environment for Reactor Applications, or VERA, accurately simulated the 2016 startup of TVA’s Watts Bar Unit 2 – the only reactor to go online in the U.S. in the 21st century – it became obvious that the project would have a permanent impact on the industry.
“Early in CASL, everyone involved established a strong vision for the program with aggressive challenge problems that drove development,” said Jess Gehin, who was initially a focus area leader and became CASL’s second director. “Hard decisions were made on research directions that resulted in delivery of game-changing capabilities that showed that modern modeling and simulation capabilities can deliver significant predictive and application improvements over the engineering tools in use at the time.”
VERA’s major milestone
VERA is a suite of software codes based on reactor physics, thermal hydraulics, chemistry and fuel performance that allows insight into every part of a reactor – down to individual fuel pellets.
A typical pressurized water reactor contains 193 fuel assemblies, nearly 51,000 fuel rods and 18 million fuel pellets. VERA can simultaneously simulate all processes in a reactor core: the heating and changing phases of coolant as it flows, the fission of fuel and changes to fuel as it is depleted. It can look at individual components to accurately predict the power cycle and fuel performance.
“We developed a virtual simulation technology that could pretty convincingly tackle the problem, that could simulate the physical phenomena and give us engineering insight as to why these problems were happening and how to ameliorate them,” Kothe said. “We put the entire reactor into a computer.”
Its value became apparent when it produced a near-perfect blind prediction of the six-month startup of TVA’s Watts Bar 2 reactor, which went online in 2016.
“Until the virtual reactor showed that it could match reactor data, everybody bought into the potential, but the industry had difficulty envisioning how it could be applied to solve real-life problems,” Kothe said.
ORNL’s Andrew Godfrey, who served as deputy focus area leader for advanced modeling applications, took the lead on that simulation, which has expanded to accurately simulate more than 200 fuel cycles, representing two-thirds of the U.S. operating reactor fleet.
“It was a major milestone and was better than what the industry was using,” Turner said, “and that’s what we promised to do in the original proposal.”
VERA has also accurately modeled next-generation reactor types, including the Westinghouse AP1000 and the NuScale Small Modular Reactor, and it can simulate a reactor even after shutdown, predicting the behavior of the fuel inside over its entire lifetime. CASL team members are now leveraging the knowledge gained from VERA toward simulating other types of reactors, including molten salt reactors.
Taking VERA to industry
Over the course of a decade, leadership and researchers changed within CASL. Kothe, who left at the halfway mark to direct the Exascale Computing Project and was replaced by Gehin, estimated that 250 people were involved with the consortium in any given year, with 750 or more of the “best and brightest minds” involved throughout the lifetime of the project.
“At the start of CASL, I underestimated the challenge of integrating teams from broadly different disciplines to achieve a focused outcome,” said Gehin, now chief scientist of the Nuclear Science and Technology Directorate at Idaho National Laboratory. “We could not have delivered on the development of VERA without the contributions of every talented person who worked in CASL to create a modern light water reactor modeling and simulation environment. I’m most proud of the development of this exceptional team of researchers that delivered on the vision set forward for CASL.”
The VERA software went through several iterations as CASL worked out a way to perform software development across several different labs, solving issues of code ownership and copyrights.
But the end goal – to deliver VERA to industry – remained unchanged. Now, as the program winds down, the consortium, led by Kropaczek, is doing that.
“Our industry council of advisers said, ‘We want software that is used and useful,'” Kropaczek said. “It’s not enough to just be a research code that only a few people can run. It has to be highly intuitive.”
That also meant removing barriers. Last year, CASL worked to ensure VERA is in Nuclear Quality Assurance-1 compliance, the gold-standard rating for the nuclear industry. This involved adapting procedures, increasing documentation, writing detailed manuals and training people to control the software.
Then there was the barrier of access to high-performance computing resources. In February 2019, in preparation for CASL’s upcoming transition to the VERA Users’ Group, the consortium trained 35 people from the nuclear industry and the Nuclear Regulatory Commission to use the software suite. Users can access the software suite through a new DOE Office of Nuclear Energy high-performance computer at Idaho National Laboratory, called Sawtooth, that provides cloud supercomputing focused on nuclear energy simulations.
“This is not a code you can run on your laptop,” Kropaczek said.
VERA has now been commercially licensed, which Kropaczek said demonstrates that people recognize its worth. In the coming months, he expects more than a dozen companies to procure VERA licenses. In addition, 400 test and evaluation licenses have been issued to individuals around the world.
“This wasn’t just a researcher program,” he said. “It produced something tangible that we see as having value for specific applications of interest to the nuclear industry.”
In the end, CASL’s legacy is twofold. The VERA software suite, with its VERA Users’ Group providing support and improvements, will likely be used for decades and become a springboard for future nuclear modeling and simulation programs.
“We showed that high-performance computing, putting the best models and algorithms into the computer to emulate the reactor, really does give you very high confidence results as to what’s going on,” Kothe said. “Because of CASL, you can design reactors in the computer with high enough confidence that when you go to build the reactor, you’re really confirming the design.”
But, he added, CASL also demonstrated that putting time and money toward a multipronged issue will attract people who can solve it – a model for research that has been replicated with the Energy Innovation Hubs and beyond.
“The bet that CASL delivered on was that if you have an important national problem and enough stable funding to go after that problem for multiple years, with multiple institutions, then you’re going to attract the best and the brightest to solve that problem,” Kothe said.
CASL, in terms of size and impact, was a first for nuclear energy, Turner said.
“We were the first hub out of the gate, and a lot of people were watching, so we had a lot to prove,” he said. “We were able to give a good return on taxpayer money.”
CASL was supported by DOE’s Office of Nuclear Energy.