Nuclear power provides roughly 20 percent of the United States’ electricity and about half of its low-carbon electricity. Whether you think nuclear power is a good or a bad thing, the fact is that the existing nuclear power fleet contributes a significant amount of energy to the US grid, and all that capacity is rapidly approaching its sunset years.
Most nuclear power plants were built in the 1970s and ’80s and were planned around a 40-year lifespan. Only one new nuclear reactor has been put into commission since the ’90s. This presents a problem: the US is facing a fast-approaching loss of a significant source of zero-carbon electricity, which it can either replace with intermittent renewables or fossil fuels. The first option may require expensive storage to smooth out the times when wind or sun are not available, and the second is undesirable given the nature of climate change.
One group at the Oak Ridge National Laboratory is trying to help utilities and energy companies extend the lives of their aging reactors. The Consortium for Advanced Simulation of Light Water Reactors (or CASL, for short) has been building and refining a reactor modeling program called VERA (an acronym for Virtual Environment for Reactor Applications), which offers high-resolution computer modeling of nuclear reactor equipment.
A long-term project for long-term projects
CASL was formed under the Obama administration as a way to make the most out of existing nuclear reactors. The group is funded by the Department of Energy (and is currently relying on pre-approved Fiscal Year 2019 funding, so it’s not subject to the partial government shutdown).
VERA can simulate safety concerns, reactor startups, and fuel-rod behavior, among other things. Dave Kropaczek, the director of CASL, spoke to Ars last week about his group’s work and said it’s not inconceivable that utilities could extend the lifespans of their 40-year reactors to 80 or even 100 years. But that’s only possible if you can show how extending the life of a reactor impacts the materials it’s composed of, Kropaczek said.
VERA appears capable of showing this accurately. Kropaczek told Ars that the CASL team has been refining VERA simulations for years and has been able to validate the results of those simulations with real-world data from things like spent fuel rods.
The high-resolution simulations allow nuclear power plant operators to eliminate some of the extreme conservatism they had when commissioning the reactors in the 1970s. Now, instead of completely retiring a functioning reactor, Kropaczek said reactor owners are able to see which components need replacing and which are OK to keep using.
He compared this plan to buying a 100-year-old house. “People buy houses from the 1900’s and they say ‘Oh, it’s a Victorian’… it’s the same thing [with nuclear]: you replace lots of things in the house, and there [are] lots of things you don’t replace.” That is, a house that needs new electrical and central air can be fixed up, while if it needs a new foundation, it may not be worth the extra effort and should be scrapped.
Solid foundations
VERA can help determine if the foundation—here, major components like the reactor vessel—is in good condition. When many US reactors were built in the ’70s and ’80s, they were over-designed and given extremely conservative safety margins, Kropaczek told Ars. So by extending the life of a rector and maybe replacing a few parts, “you’re not giving up anything” in terms of safety, Kropaczek said. “You’re just saying, ‘Hey, I’ve sharpened my pencil to a really fine point.'”
The work that CASL has done on VERA has been supported with real-world data from private companies like reactor designer and fuel distributor Westinghouse (which went bankrupt in 2017 but was purchased in 2018 by Brookfield Business Partners), the Tennessee Valley Authority (TVA), and the Electric Power Research Institute.
Another way VERA simulation can help utilities is by modeling their energy output as the grid changes. Say you have an abundance of wind and solar on the grid; during the mid-afternoon, when the sun is shining and the wind is blowing, all those renewable resources will be producing energy. Nuclear is a challenge then, because reactors take so much time to shut down and start up—they have to be running constantly to be economical. As a result, a lot of that wind and solar might not get used, because it would keep the grid manager from maintaining a specific frequency on the grid.
Instead, perhaps a nuclear plant could turn down its power temporarily to make use of that renewable energy (an operation called load-following). That could stress the fuel—but by how much?
“Nuclear power plants need to be boring with respect to operations, with respect to the fuel. I don’t want to see any surprises—that’s the goal of a plant: nothing happens,” Kropaczek said. “Load-following is no longer steady state and boring. We need to go from 100-percent to 50-percent power and back again. So you’re worried about the fuel and stresses on the fuel and things are changing.”
However, “VERA can actually… model fuel behavior for load-follow, [reporting] what the fueling is going to do and what is the fuel operating level?… We can look at every fuel rod, every fuel pellet in the core; we can look at those stresses in the fuel.” As long as the fuel stresses are kept within reasonable bounds, more aggressive load-following could be possible.
Training utilities on a big computer
Ultimately, CASL hopes that making better use of existing plants will buy the US some time to roll out more advanced nuclear technologies, like reactors that can take advantage of spent fuel from the current fleet of Light Water Reactors. (This was the idea behind TerraPower, whose demonstration plant was recently stymied by restrictions placed on working with Chinese partners, according to reports.)
Until then, getting VERA in the hands of utilities and energy companies is important. The program has been used before by clients including the TVA, which used VERA to model the startup of the Watts-Barr Unit 2 reactor in Tennessee in 2015. “VERA tracked the startup very, very accurately,” Kropaczek said.
VERA was also used by Westinghouse to model the startup of its first AP 1000 reactor in Sanmen, China, in 2018. Although the program is only available to US nationals and people with the appropriate qualifications, Westinghouse used VERA to simulate the Chinese reactor startup from a distance.
Now, CASL hopes to open up a lot more functionality to additional US utilities. CASL is setting up VERA on a 1,000-core computing cluster at the Idaho National Laboratory this month to host a week of training for 24 staff at 12 different nuclear energy companies. The companies receive their own accounts and can use VERA to run their own simulations.
“We want the technology to be transferred to industry,” Kropaczek told Ars. “That’s our goal.”
A path to better regulation
One common criticism lobbed at nuclear policy in the United States is that the industry’s governing body, the Nuclear Regulatory Commission (NRC), slows down progress unnecessarily through years of regulatory pushback and requirements. Perhaps with a more progressive approach to regulation, keeping old nuclear power plants alive longer wouldn’t be as necessary, and we could move on to more advanced nuclear reactors.
Kropaczek doesn’t feel that way. “The regulator is doing their job,” he said. They’re constantly asking the utilities: “tell me why you can do this.”
But the director believes that CASL’s work will eventually speed up the regulatory process. “If I [as the owner of a nuclear plant] can improve my modeling and simulation to the point where it’s trustworthy, I’m supplementing data with modeling and simulation. Add the two together and it will allow the regulator to make a decision in a much more informed way.”
The point is to push the conservative 40-year-old planned limits on our nuclear fleet without compromising on real safety limits. As such, modeling with VERA can “address issues that no single utility could take on. This is big science and big technology together,” Kropaczek said.
Source: Ars Technica
