Is This The Key To Commercial Nuclear Fusion?

Authored by Haley Zaremba via OilPrice.com,

While the concept of harnessing nuclear fusion, the process which powers the sun, for use here on Earth has long been the “Holy Grail” of energy science, it has long been too good to be true. Nuclear fusion has the potential to solve some of the world’s greatest crises: it’s several times more powerful than nuclear fission, it can be produced using only hydrogen or lithium, which means that it creates none of the hazardous radioactive waste that makes nuclear fission so infamously problematic, and it produces absolutely zero carbon emissions. 

But there is one problem. Despite the many, many teams of scientists that have tried, no one has yet figured out how to make nuclear fusion reproducible on a commercial scale. This has been so difficult to achieve because it’s no easy feat to artificially recreate the extreme conditions like those found in the core of the sun, where nuclear fusion occurs naturally. As explained by the United States Department of Energy, “fusion reactions are being studied by scientists, but are difficult to sustain for long periods of time because of the tremendous amount of pressure and temperature needed to join the nuclei together.”

Over the past year, however, it has become easier and easier to believe that commercialized nuclear fusion will not remain out of reach for much longer.

First, the International Thermonuclear Experimental Reactor (ITER), a multinational project based in Southern France, announced in July that their team is now a mere 6.5 years away from achieving “First Plasma.”Now, just this week, researchers have announced another breakthrough bringing us closer to making commercial fusion a reality.

The secret? Artificial Intelligence. An article from the supercomputing facility Oak Ridge National Laboratory reported on Wednesday that “a team of researchers has leveraged supercomputer-powered AI in an effort to address one of the key problems with scaling up fusion energy.”

That key problem is the tricky issue of managing plasma. 

Currently, in the majority of nuclear fusion research, plasma is created and maintained inside of a device called a tokamak (defined by Oak Ridge National Laboratory as “magnetic, donut-shaped fusion devices that hold fusion reactions in place so the plasma doesn’t lose its heat or interact with the surrounding materials”) but even the most cutting-edge versions of these devices still have a lot of limitations. “Instabilities in this process (‘disruptions’) allow plasma to escape, reach the walls of the tokamak, stop the reaction and potentially cause irreparable damage to the reactor itself. The problem is also scaling up: the larger the fusion reactor, the lower the surface area, increasing the risk of severe damage from a disruption.” This means that even the most promising tokamak projects, like the massive one underway at ITER, are not out of the R&D woods yet--far from it.

One team of researchers, however, says that they have found a solution for ITER’s challenges through in implementation of AI and supercomputing. The United States Department of Energy and the Princeton Plasma Physics Laboratory’s Bill Tang led a team of researchers to look into the issue, resulting in a study published in the scientific journal Nature, entitled “Predicting disruptive instabilities in controlled fusion plasmas through deep learning.” In this study, Tang says that he and his team “aim to accurately predict the potential for disruptive events before they occur, as well as understand the reasons why they happen in the first place.”

Right now, the supercomputing used in tokamak fusion experiments is simply not fast enough to detect these disruptions, which occur all but instantaneously. For Tang’s team, as summarized by Oak Ridge National Laboratory, the goal was to find a way to “meet the 95 percent correct disruption prediction threshold required by the under-construction ITER Tokamak, which will be the larger fusion reactor in the world.” The team has achieved extremely promising results with the Nvidia P100 GPUs at the Tokyo Institute of Technology’s TSUBAME 3.0 supercomputer, as well as with the world’s most powerful supercomputer Summit. This has allowed Tang’s team to move forward from detection and start broaching the all-important and even more difficult task of preventing disruptions. 

“With powerful predictive capabilities, we can move from disruption prediction to control, which is the holy grail in fusion,” Tang was quoted by Oak Ridge National Laboratory.

“It’s just like in medicine - the earlier you can diagnose a problem, the better chance you have of solving it.” 

With new breakthroughs announced in the field of fusion all the time, and such promising research currently underway, nuclear fusion feels not just like a fantastical possibility, but an inevitable eventuality.