Reproducing LHD Deuterium Plasma Using Leading Edge Supercomputers
In achieving nuclear fusion power generation, it is necessary to confine the energy stored in a plasma for a long period of time and to maintain the extreme high temperature condition of higher than 100 million degrees. The phenomenon occurring in a plasma that most greatly affects energy confinement is “turbulence.” In high temperature plasma there sometimes occurs the generation of large and small flows and eddies, and those become irregularly disturbed, leading the plasma to a turbulent state. Turbulence is a familiar phenomenon that occurs in the seas and in the atmosphere. However, when turbulence occurs in a plasma, the temperature falls due to the fact that the plasma becomes mixed. In order to weaken this turbulence as much as possible and achieve plasma of greater performance it is necessary to clarify both the conditions in which the turbulence was generated and the details of its properties.
At the National Institute for Fusion Science we are preparing experiments for the Large Helical Device (LHD) that will utilize deuterium in order to improve plasma performance. In deuterium plasma, which bears a mass twice that of light hydrogen which has been used to date, an important issue is clarifying what types of changes will occur in turbulence and in energy confinement. Further, the turbulence phenomenon in plasma is extremely complicated. In understanding those properties and in predicting energy confinement, in addition to experiments, large-scale simulations that utilize the full capabilities of leading edge supercomputers will perform an extremely important role. Here, we will introduce turbulence simulation research in deuterium plasma.
Plasma turbulence simulation built upon numerous ions and electrons requires immense computational complexity. Flow phenomena in water and in the atmosphere can be investigated through solving equations numerically that express movement in three-dimensional space. However, in order to compute the movement of turbulence in a plasma confined by a magnetic field, because of the unique complexities and diversity of plasma we need an immense number of coordinates for expressing movement in a five-dimensional space (a mathematical space in which two components for a particle’s velocity are added to three spatial coordinates for the position) and perform calculations. Making the plasma turbulence simulation all the more difficult is that we must treat the movement of the incredibly fast-moving electrons (several tens of thousands of kilometers per second) and the movement of the comparatively slow ion particles (several hundreds of kilometers per second). Because of the growing speed difference between the heavier ion particles and the electrons, a simulation in a larger scale and in a longer time duration is necessary. Thus, to date, turbulence simulations using light hydrogen plasmas have been conducted in the LHD. Deuterium plasma simulations were extremely problematic. This time, using NIFS’ leading edge supercomputer “Plasma Simulator,” which was newly enhanced in June 2015, even larger scale simulations have become possible. We achieved for the first time in the world a deuterium plasma turbulence simulation in the LHD. Further, we compared turbulence born of “trapped particles” that move back and forth in the magnetic field by means of light hydrogen plasma and deuterium plasma, and we predicted that turbulence is suppressed and energy confinement is improved by deuterium plasma. In addition, we investigated that mechanism and clarified that the flow called the “zonal flow” which has a comparatively large structure is strong in deuterium plasma, and that that flow effectively grinds large eddies and waves, suppresses turbulence, and improves energy confinement.
These results brought by the high performance of the “Plasma Simulator” will inform research in high performance plasma in the upcoming LHD deuterium experiment.