This image shows the electron temperature fluctuations in a multiscale simulation of a DIII-D, ITER-like plasma where turbulent structures existing on different spatial scales are clearly visible.  These structures are responsible for the experimental levels of heat and particle transport observed in current tokamak plasmas.

Chris Holland

Nathan Howard receives INCITE leadership computing award

Paul Rivenberg  |  PSFC

A multi-institutional team consisting of Plasma Science and Fusion Center research scientist Nathan Howard, Chris Holland (University of California, San Diego) and Jeff Candy (General Atomics), has received a prestigious INCITE leadership computing award. The team will receive 100,000,000 CPU hours on Oak Ridge National Laboratory’s Titan XK7 supercomputer for a fusion project entitled, “Understanding How Multiscale Transport Determines Confinement in Burning Plasmas.”

The project will use a new gyrokinetic code for simulating plasma turbulence called CGYRO, recently developed at General Atomics, and specifically optimized for extreme computational problems presented by multiscale simulations. These simulations will study the impact of cross-scale coupling of ion and electron-scale turbulence in conditions relevant to burning plasmas, and will validate the gyrokinetic model through direct comparison with experimental measurements of heat and particle (impurity) transport, as observed in GA’s DIII-D tokamak fusion experiment.

The completion of this work will begin to shed light on the dynamics of turbulence and on the multiscale nature of heat and particle transport in tokamaks. These phenomena must be understood in order to accurately predict the performance of future fusion reactors, including ITER, an international fusion experiment currently under construction in France.

Ultimately this understanding is key to the success of fusion as a viable commercial energy source. The reason for this is that in burning plasmas they are mostly self-heated by the energetic helium ion produced by the fusion. The fusion reaction is produced by ions, yet the energetic helium heats the electrons, which in turn couple their energy to the ions, which then can fuse. This makes cross-coupling a key issue but it is a grand computational challenge because the fundamental size scale of their orbit size around the magnetic field is different by a factor 60 between electrons and ions.

Titan XK7, currently the fastest supercomputer in the U.S., is able to carry out quadrillions of calculations each second, is an invaluable resource for performing simulations that simultaneously capture the dynamics of plasma turbulence on the scale of both the ion and the electron gyroradius.  Even with the increased speed, approximately 17,500 times faster than a modern laptop, a single multi-scale simulation could take up to 20 days on Titan – hence the importance of this computer-time grant.

This competitive award is sponsored by the U.S. Department of Energy, and is open to academia, government labs, and industry. It provides supercomputer time for projects that address “grand challenges” in science and engineering. Howard and his colleagues were the only plasma physics proposal funded in magnetic fusion energy this year.


Topics: Magnetic fusion energy, Plasma theory & simulation, Plasma turbulence