Yueqiang Liu gave an invited presentation at the 61st Annual Meeting of the APS Division of Plasma Physics on “Toroidal modeling of post-disruption runaway beam by MHD instabilities in DIII-D”, describing simulations of a recent DIII-D experiment, where fast-growing magnetohydrodynamic (MHD) instabilities were observed when the plasma edge safety factor dropped to around 2. This instability then suppresses the runaway beam.
Klaus Hallatschek from the Max-Planck-Institute in Garching visited GA for two weeks to work with Emily Belli and Jeff Candy on new mathematical approaches for computations involving the linearized Landau collision operator. In particular, Klaus has developed a pseudo-spectral approximation method for the full linearized operator at arbitrary wavelength. This new operator has been implemented in a development version of CGYRO and preparation of a publication is underway.
For an upcoming DIII-D hydrogen isotope campaign, the Neutral Beam Injection (NBI) ion profiles were predicted with the TGLF-EP+Alpha 1D critical gradient model combined with TRANSP simulations. This is the most recent example of the predict-first paradigm being pursued at DIII-D. In addition to the usual all deuterium operation, the upcoming experiments will use a hydrogen beam into a deuterium plasma and a hydrogen beam into a hydrogen plasma. The reduced collisional slowing-down time of hydrogen (compared to deuterium) reduces beam-ion pressure and drive of transporting Alfvén eigenmodes (AEs), but hydrogen beam velocities can be super-Alfvénic and lead to greater AE drive. In the case of a hydrogen beam into a deuterium plasma, these competing effects roughly cancel. In this case, TGLF-EP+Alpha finds only modest change from the deuterium-beam reference, with the core beam-ion depletion (from classical) modified up or down a few percent depending on the discharge time. For the case of a hydrogen beam into a hydrogen plasma, the beam Alfvénic Mach number is closer to the reference case so the reduced beam pressure wins out. Here, the beam-ion core pressure deficit versus classical is reduced by about 15% across tested beam power levels and time slices. Ongoing streamlining of the TGLF-EP+Alpha transport model will enable more such predict-first beam-ion studies.
An interface to the GA TokSearch [Sammuli FED 2018] data mining tool has been developed in OMFIT. TokSearch's massively parallel data ingestion and processing capabilities, combined with high performance computing hardware can provide data throughput orders of magnitude faster than single-threaded methods. The new OMFIT interface streamlines the use of this new technology, and makes it transparently available to OMFIT physics modules. It is expected that the efficient utilization of experimental data archives will be critical in supporting new and upcoming large-scale machine learning studies. Public access to DIII-D users will become available as a new dedicated TokSearch server comes online.
Disclaimer
These highlights are reports of research work in progress and are accordingly subject to change or modification