Cooling towers stand against a blue sky at the Watts Bar Nuclear Plant, in Tennessee. (Photo by the Tennessee Valley Authority, via Flickr)
The shutdown was triggered automatically and no serious issues were found, but the process cost more than 100,000 hours of labor and $21 million, not to mention lost productivity with the plant offline. A team of GW researchers is now hoping to shed new light on vibrations in nuclear cores and lessen the impact of future earthquakes.
“They didn’t know how the [nuclear cores] reacted because there were no tools to predict the behavior,” said Philippe M. Bardet, a professor in the Department of Mechanical and Aerospace Engineering.
Dr. Bardet and his colleagues Elias Balaras, in mechanical and aerospace engineering, and Majid Manzari, in civil and environmental engineering, are angling to change that. Last month the U.S. Department of Energy awarded the researchers more $860,000 over three years to devise a model for simulating the impact of vibrations inside a nuclear reactor.
It’s a tool they hope could be used to help assess damage before a costly cool-down and inspection, and to help engineers design next-generation reactors.
The research, utilizing GW’s earthquake simulator and one of the world’s fastest supercomputers, will be done in collaboration with scientists at Argonne National Laboratory, in Illinois, and the French Alternative Energies and Atomic Energy Commission.
The core of a nuclear reactor contains tens of thousands of fuel rods, each about the diameter of a pencil and 13 feet long. Inside the rods uranium heats up in a chain reaction. That heat is used to turn water into steam, which drives a turbine and generates electricity.
The researchers will be modeling the most common type of reactor in the United States, called a pressurized water reactor. In a PWR, steam is made from water outside the reactor core, while water inside is used as a coolant.
“The [water] flow in there, it’s chaotic,” says Dr. Balaras, one of the GW researchers. “It’s what we call turbulent flow.”
The force of the water gives a slight push to the tall, flexible and tightly-spaced fuel rods in a dance he likened, in some sense, to a flag flapping in the breeze.
The issue is: What would happen if the rods suddenly pushed back? A vibration—from a broken water pipe, an earthquake or any other shock—could send the rods swinging against the flow of water, changing the dynamics of the core. That could get dangerous if, for example, a fuel rod gets bent.
Workers inspect fuel assemblies—the bunches of long fuel rods containing uranium pellets—that power a nuclear reactor. (Nuclear Regulatory Commission file photo)
The interaction, and at what point it becomes dangerous, is not very well understood, the researchers said. But it also hasn’t been an issue. Nuclear power plants are so conservatively designed, said Dr. Bardet, that they’re made to withstand vibrations from both an earthquake and a broken pipe simultaneously, which he said has never happened. Still, “it’s worth knowing.”
“And in light of what happened in Fukushima [Japan],” he said, “we also now have way more advanced numerical tools, so the idea is to do high-fidelity simulations of what’s going on in the reactor.”
In order to do that the researchers are building a small-scale, non-nuclear core: a bundle of 36 fuel rod stunt doubles inside a transparent-walled case. To simulate the high-velocity water flow, the “core” will be connected to a 100-horsepower pump pushing in nearly 2,000 gallons of water per minute.
High-speed cameras will capture the flow of the water, with the help of floating tracers that fluoresce in laser light, as well as the movement of the fuel rods.
And the whole structure will face temblors of various strengths atop GW’s earthquake simulator at the Virginia Science and Technology Campus.
They’ll also be using one of the world’s fastest supercomputers, through the collaboration with Argonne National Laboratory, to build a parallel computer model that will be able to account for known characteristics the team couldn’t replicate in the lab, like the actual structural details of a uranium fuel rod.
“These two approaches are complementary,” said Dr. Balaras. “What he cannot measure, we can compute. And what we cannot compute, he can measure. And we can use those two things in a synergetic way to understand the physics of the problem.”
Their hope, then, is to translate the complex supercomputer models—which might take decades to run on a regular computer—into less taxing formats for use by engineers designing new power plants.
—By Danny Freedman
