Imaging Plasma Simulations

One of the biggest challenges facing fusion researchers is handling large plasma instabilities, which can cause excessive erosion of tokamak materials. This must be solved before we build commercial fusion power plants, as constantly shutting down to replace damaged components would be very inefficient.

One of the plasma instabilities in a tokamak is called an edge localised mode or ELM for short. A lot of research is currently being done on ELMs. They appear as large filamentary structures of plasma where the filaments violently erupt at the edge of the main plasma. This eruption often leads to hot plasma touching, and thus damaging, the material surfaces inside the tokamak chamber. We know that filamentary structures appear during ELM instabilities because we are able to observe them in fusion experiments using diagnostics. One such diagnostic is the fast camera diagnostic. The fast camera creates images of visible spectrum light emitted during a plasma discharge. It can capture images of the ELM filaments, and the data gathered can be used to analyse the filamentary structures and dynamics.

The challenge is to understand these ELM instabilities and develop a solution to mitigate or completely suppress them. Fusion physicists are attempting to understand these instabilities through experiments, theory and numerical simulations. My work concentrates on the latter, where I am using a code called JOREK [1] to run numerical simulations that test existing plasma models. Numerical simulations show which models best describe ELMs, which increases our understanding of plasma physics. Numerical simulations can also be used to make predictions for future tokamaks, most notably ITER.

In particular, I’m researching and trying to understand if altering the geometry of a tokamak exhaust system can significantly reduce the hot plasma and resulting heat loads reaching the surface materials. The new exhaust geometry upon which my research is based, is called the “Super-X”. It will be tested on the MAST-U Super-X tokamak, which is based at CCFE in the UK and is scheduled to start operation in the near future. If successful, this could make ELMs less of a concern, directly benefitting future fusion devices.

Another challenge occurs when attempting to simulate ELM instabilities.  Some of the simulations I perform take up to three million core hours. To put this into perspective, an average laptop has two to four cores, so it would take roughly 85 years to run one of these simulations on your laptop. Luckily, we run the simulations on supercomputers, but even so it takes a few months to complete a large simulation. Once an ELM simulation is complete, I can choose from many analysis techniques to make sense of the data. One such technique uses a synthetic fast camera diagnostic. Just like a real camera diagnostic takes pictures of actual plasmas, the synthetic diagnostic code uses data from the simulations to create pictures.

Just like a real camera diagnostic takes pictures of actual plasmas, the synthetic diagnostic code uses data from the simulations to create pictures.

Synthetic Photography

I use the synthetic fast camera to create images of the filaments in a simulation. The eight synthetic fast camera snapshots I’ve included show the evolution of the plasma during a simulated ELM. A filter has been applied so that we can observe the filaments. Without this filter, only the bright light from the exhaust regions would be visible.

Image a) shows a well-confined plasma before the instability occurs. Images b) to f) show the filaments starting to form and then violently erupting from the plasma edge, which degrades the plasma confinement. Afterwards, the filaments start to reduce in size - as seen in images g) to h), and the fast camera images also show that plasma has moved to the exhaust regions (the upper and lower bright regions in images d) to h)) during the ELM.

The advancements in supercomputing power have made these simulations possible, and the models used in the simulations are also becoming more advanced. Fortunately, this ELM simulation showed a reduction in the heat loads to the exhaust material surfaces, which is promising for the Super-X exhaust geometry at MAST-U. I find it exciting to contribute to one of the big challenges in fusion and hopefully my work can provide useful guidance for future research.

[1] G. Huysmans and O. Czarny, Nucl. Fusion 47, 659 (2007)