Klaus Hallatschek from IPP (Garching) visited GA for two weeks to work with Emily Belli and Jeff Candy on edge physics. In particular, studies of GAM oscillation and damping in the fluid/collisional limit were made. The extreme values of parameters used for these studies led to refinements in the field-particle operator in CGYRO as well as the energy discretization approach. This work is now being prepared for a short publication on tje fluid limit of zonal flows and GAM oscillations.
A series of integrated simulations of China Fusion Engineering Test Reactor (CFETR) was performed in collaboration with University of Science and Technology of China (USTC) and Institute of Plasma Physics, Chinese Academy of Sciences (ASIPP) in Hefei, using the OMFIT framework. The CFETR workflows were improved to allow more comprehensive self-consistent core-pedestal-equilibrium coupled model that includes particle, heating, and current-drive sources from the ONETWO transport code, profile evolution using the TGYRO transport manager with the TGLF turbulent and NEO neoclassical transport models, the EFIT free-boundary equilibrium solver, and the EPED pedestal width and height model. Three fully non-inductive CFETR Phase I and Phase II scenarios have been developed, which show that the CFETR Phase II scenarios can reach the goal of fusion power greater than 1000 MW with a bootstrap current fraction of greater than 85% at a q95 of 4.5. A new larger size CFETR design reduces the heating and current-drive requirements and lower divertor heat flux and neutral wall loading. Preliminary results indicate that the lower normalized beta of ~ 1.9 phase I configuration is stable to the ideal n=1,2 MHD modes without a conducting wall, whereas that higher normalized beta > 3.2 Phase II configurations required a close conducting wall for stability. Future work will include self-consistent impurity and alpha-particle transport, pellet fueling, and a more self-consistent rotation source for turbulent suppression.
The automated M3D-C1 capability within OMFIT has been exploited to perform 3D plasma response analysis during the run day of an experiment. A sequence from ELM suppression with 3D magnetic perturbations, at high rotation, to ELMing at lower rotation was produced on DIII-D. This sequence was analyzed with OMFIT workflows. A series of kinetic EFITs was created and input into the automated M3D-C1 analysis module. This allowed linear plasma response results to be obtained just hours after the experiment had been performed. These preliminary results indicated that the transition of the plasma from ELM-suppressed to ELM-free to ELMing correlated with a consistent decrease and inward shift of the peak resonant magnetic perturbation in the pedestal. This work demonstrates a valuable new tool for run-day interpretation and guidance of experiments.
In a collaboration with ASIPP-Hefei, the parallel Graphical Processing Unit (GPU)-optimized version of EFIT, P-EFIT, has been extended to include Motional-Stark-Effect (MSE) and full kinetic equilibrium reconstructions. These require P-EFIT to extend its reconstruction capacity to allow for more detailed diagnostics and physics constraints, more comprehensive and flexible current and pressure profile representations such as tension spline, higher spatial resolution, and convenient data interface. P-EFIT parallel algorithms have been improved and redesigned to meet these requirements. These include the development of parallel algorithms to efficiently distribute computational threads, making optimum use of GPU cores and high-speed shared memory while minimizing the expensive communication and synchronization costs. Reconstructions of DIII-D discharges using experimental kinetic profiles, MSE and external magnetic data show that P-EFIT can accurately reproduce the EFIT reconstruction algorithms at a fraction of the EFIT computational time. The results suggest that P-EFIT can potentially offer real-time full kinetic equilibrium reconstructions with high spatial resolution allowing more accurate and better plasma control and other applications.
A second code-camp for developers of the OMFIT data analysis and modeling framework was hosted at General Atomics this week (Feb. 27-Mar. 3). In addition to numerous bug fixes and improvements to the framework itself, significant progress was made on a broad range of its physics modules. For instance, the TORAY module developed by M. Brookman is capable of setting up and running parallel TORAY simulations that are set up based on time-dependent profile reconstructions generated via the OMFITprofiles module. An Ion Cyclotron Emission (ICE) module was developed by K. Thome to analyze the 1TB/shot data collected by the DIII-D ICE diagnostic with an efficient algorithm to minimize memory usage. C. Rea continued the development of the DB_ANALYSIS module to leverage machine learning algorithms for investigating the correlation of time-dependent signals on experimental databases. B. Lyons updated the M3D-C1 module that he developed to enable post-processing of M3D-C1 simulations on local as well as remote workstations. The TRIP3D-GPU module has been further improved by G. Trevisan to support the execution and visualization of simulation ensembles, as well as to convert internally to OMFIT the original TRIP3D-GPU file format to NetCDF for reduced file size and improved performance. J. McClenaghan worked on integrating the Sauter model into the GA Systems Code module of OMFIT, and improving the module's neutron peaking calculation. The original TIMCON module by D. Eldon has been re-written by T. Wilks and N. Logan to use scaling laws to preview the range of obtainable power, torques, beta and rotation, as well as the effect of changing the beam voltages. The OMFIT TIMCON module will be used to support the work of DIII-D beam operators. The next OMFIT code-camp is scheduled to take place at GA the week of Aug. 21.
These highlights are reports of research work in progress and are accordingly subject to change or modification