Max Planck Institute for Plasma Physics, Garching
Local Project ID:
HPC Platform used:
SuperMUC of LRZ and Hemit of HLRS
Fusion energy is a very attractive option to provide large-scale and CO2-free electricity production for centuries to come. Here, the goal is to mimic the way the sun generates its power under laboratory conditions on earth. To this aim, one confines a plasma (i.e., an ionized gas), consisting of two heavy versions of hydrogen, namely deuterium and tritium, in a doughnut-shaped magnetic cage and heats it to about 100 million degrees. In order for this to work, however, the energy confinement – which is controlled by turbulent transport – must exceed a certain level.
Thus, one of the key physics challenges on the way towards future fusion power plants is the understanding, prediction, and control of turbulence in these devices. Given the inherently nonlinear and multi-scale character of turbulence and the need to work with 5D (reduced) kinetic descriptions in lieu of 3D hydrodynamic ones, progress in this research area hinges on the latest advances in supercomputing.
In the context of the present project, one of the world-leading plasma turbulence codes, GENE (developed in the group of Prof. F. Jenko at IPP Garching), has been employed to carry out pioneering studies of actual large-scale fusion devices with an unprecedented level of realism, bringing the community one step closer to the ultimate goal of a “virtual tokamak”.
The first goal was to ensure that the experimentally inferred heat fluxes in the present-day experiments ASDEX Upgrade and JET are recovered even quantitatively, thus validating both the code and basic approach.
An additional goal was to better understand the scaling of the transport levels with system size.
According to present thinking, the magnetic confinement should improve substantially with increasing size of the device, and the design of the new flagship fusion experiment ITER, currently under construction in Southern France, has been chosen accordingly. With the help of individual GENE simulations ranging from 100,000 to many million core-hours, both of these issues could be addressed successfully, breaking new ground scientifically.
Moreover, it was shown that a newly discovered effect regarding the interaction between the fast particles originating from the fusion reactions and the background plasma can result in an unexpected substantial improvement of the confinement. Obviously, this is very good news for ITER, based on state-of-the-art computations on state-of-the-art supercomputers provided by the GCS Infrastructure.
The project was made possible through the Partnership for Advanced Computing in Europe (PRACE). As simulation platforms served GCS HPC systems Hermit of HLRS (23.3 mill core hours) and SuperMUC of LZR (26.7 mill core hours).
Principal Investigator: Frank Jenko (Max Planck Institute for Plasma Physics)
Researchers: Daniel Told, Tobias Görler, David Hatch, Stephan Brunner, Tilman Dannert
Project partner: Ecole Polytechnique Federale de Lausanne (CRPP), RZG
Prof. Dr. Frank Jenko
Max Planck Institute for Plasma Physics
Boltzmannstr. 2, D-85748 Garching/Germany