Princeton Plasma Physics Laboratory (PPPL) scientists are creating simulations to advance magnetic mirror technology, a potential path to safe, clean nuclear fusion power that would use relatively small, inexpensive and quickly assembled devices — but was previously considered unworkable.
“The magnetic mirror is an exciting opportunity that may have very big applicability to fusion science overall,” says Manaure Francisquez, associate research scientist at PPPL. He leads a collaboration with groups that are building experimental magnetic mirror devices with support from the Department of Energy Advanced Research Projects Agency-Energy (ARPA-E).
Francisquez has been modeling plasmas for nuclear fusion applications such as tokamak reactors for years. For magnetic mirror research, he and his colleagues received an ASCR Leadership Computing Challenge award of 93,000 node-hours on the Perlmutter supercomputer at the National Energy Research Scientific Computing Center (NERSC) to develop algorithms and simulations that will help guide the other groups’ experiments.
In a magnetic mirror, a plasma of deuterium-tritium nuclear fusion fuel is held in a horizontal core formed by two powerful, ring-shaped magnetic coils that face each other to form a cylindrical volume between them. In such a device, the magnetic field is strongest at the coil centers. That sets up conditions for plasma ions to reflect between the coils, ping-ponging inside the core and preventing most of the ions from escaping out the ends. The plasma could then be heated over 100 million degrees, triggering enough fusion to generate power. On their way to that goal, many research groups around the world have worked to improve magnetic mirror containment.
Hopes for mirror technology were high in the 1970s and 1980s, but after losses of major funding and frustrations with fuel containment, most researchers abandoned the approach. Over the past decade, containment has improved and new mirror-related technologies have arisen, sparking renewed interest. Now the main challenges are plasma instabilities that threaten to damage mirror devices or prevent plasmas from getting hot enough to produce fusion.
“These plasmas are turbulent, and they can draw energy from a lot of things,” Francisquez says. “They can draw energy from gradients in the pressure, for example, and they can use that energy to move around the machine in uncontrolled ways.”
‘You need the plasma to remain confined.’
A common instability happens when fuel bursts out the side of the core, a massive plasma ejection that melts the machine’s inner wall. Researchers working on other types of nuclear-fusion facilities, such as tokamaks, have developed techniques to control these events. But magnetic mirror devices face an additional challenge called the interchange instability, or when the plasma volume wobbles like a rolling pin with off-center handles.
As this instability progresses, the plasma can also make damaging contact with the inner wall. The instability also can let hot particles and heat escape from the plasma core. “You need the plasma to remain confined,” Francisquez says. “It needs to be hot enough and dense enough for long enough to produce enough fusion energy, but this instability can degrade the magnetic confinement, and it can allow more particles and energy to leak out.”
His group works with two ARPA-E-funded teams who are incorporating new approaches with the potential to quell the interchange instability. At the University of Maryland, College Park, the Centrifugal Mirror Fusion Experiment (CMFX) is building a device that will force the core to spin around its axis at supersonic speeds. At the University of Wisconsin–Madison, researchers are working toward a device that uses high-temperature superconducting (HTS) magnets. This project, called the Wisconsin HTS Axisymmetric Mirror (WHAM), is a partnership among the UW–Madison, MIT and Commonwealth Fusion Systems in Devens, Massachusetts, and Realta Fusion Inc., a Madison-based spinoff of the WHAM experiment that is funding the ongoing research.
“For these devices to have a very useful and prominent role in the fusion ecosystem,” Francisquez says, “we’re going to have to have some robust and reliable computational tools that model them.”
He envisions a two-fold project. First, the group will develop algorithms and models that will work not just for WHAM and CMFX but also for future devices. “When these machines turn on, we can use their experiments to validate the models, and that will give us confidence that what we’re predicting for the future is truthful,” he says.
Second, they aim to reduce the many uncertainties inherent in the experiments. For example, although experimenters can roughly predict the core temperature and density of a plasma core, they can’t know how these parameters will vary with the radius or change over time. “They’ve given us the parameters that they anticipate — some macro description of the plasma — and it is up to us and our models to figure out, okay, what are the micro-instabilities? What are the micro-details of the plasma? Is it going to be stable? And hopefully we can help them.”
To accomplish these goals, the group will apply an open-source plasma-modeling code named Gkeyll, currently hosted at PPPL, to the problem on the Perlmutter Cray EX petascale supercomputer at NERSC.
“Perlmutter is particularly convenient for our tools,” Francisquez says. His group is familiar with its architecture and writes its code for the type of NVIDIA graphics processing units in the machine. “We have a long history of using NERSC. It’s just an incredible platform to do computational sciences. It’s a large machine, but it’s also easy to use.”
First, he and his co-workers will develop a detailed model of the interchange instability. “Let’s model this scenario,” he says. “How much is it moving away from the axis? What’s the rate at which you’re losing energy, and how does that compare to the duration of the experiment?” Then they’ll simulate the experimentalists’ proposed approaches to tame the instability and see what works.
The team is running and analyzing their simulations. Francisquez looks forward to seeing the new computational tools benefit the wider field of nuclear fusion power.
“One of the things that attracted me computationally about this project is that I thought we could do this modeling more realistically than we can for tokamaks. But also we can do it with the same tools that we use for tokamaks. And as we’re developing these tools for the mirror, we’re going to develop them in a way that also benefits the tokamaks.”