A study of the behavior and convergence of different numerical treatments of ExB shear in gyrokinetic simulations was carried out. Historically, GYRO has treated ExB shear directly in real space. On the other hand, if the equations are solved spectrally with periodic boundary conditions, a time-dependent wavenumber must be retained. However, there is significant complexity and cost required in doing so, and for that reason all existing GK codes (to our knowledge) employ a simple “array shift” scheme with no time-dependence of the wavenumber. Curiously, no detailed study of the accuracy of this approximate method has been reported. To this end, we implemented the complete wavenumber time-dependence in CGYRO. We indeed observe improved agreement with the direct method in GYRO when the proper wavenumber time-dependence is retained. The error in the pure “array shift” scheme appears to be greatest at weak ExB shear, and we have observed cases with a spurious increase in transport, rather than a decrease, at low shear.
Gary Staebler attended the ITPA Transport & Confinement Meeting in Ahmedabad, India, and presented a talk entitled “Development of a Simplified Theoretical Model for H-Mode Energy Confinement.” The work proposes using neural network fits to the EPED pedestal model and the TGLF core transport model to enable highly efficient coupled core-pedestal modeling, including predictions of global energy confinement. These could then be compared to the developing ITPA confinement database, to assess accuracy and develop improvements to the model. In addition, Phil Snyder gave a remote presentation at the ITPA Pedestal Meeting, summarizing recent results from the Pedestal Structure Working Group. These results include comparisons of the EPED model to more than 700 cases on 6 tokamaks (including new comparisons to the COMPASS tokamak in the Czech Republic), finding good agreement (average error ~18-20%). For this dataset, the level of agreement with EPED is significantly better than with empirical scaling laws developed by ITPA [J. Cordey Nucl. Fusion 43 (2003) 670], despite EPED being a theoretical model with no fitting parameters while the empirical scaling laws are fitted with 9 or 10 parameters.
The extended magnetohydrodynamics code M3D-C1 has been installed and tested on DIII-D's new computational cluster Iris. M3D-C1 is used in support numerous DIII-D experiments, particularly for the analysis of plasma response to externally-applied, three-dimensional magnetic perturbations. The new installation will allow for local running of the code in support of DOE projects. Furthermore, a new Python utility, autoC1, has been developed that allows for automated running of M3D-C1 in the linear mode with routine parameters and templates. This script allows for straightforward analysis of new kinetic EFITs. Through it, the user can preprocess kinetic EFIT files, recompute an equilibrium on the M3D-C1 grid and compare to the EFIT, adapt the finite element mesh to the equilibrium, and perform various calculations on the adapted grid, including linear stability and linear, time-independent plasma response to 3D magnetic perturbations. Anyone interested in running M3D-C1 at GA, including the autoC1 script, should contact Brendan Lyons .
Professor Howard Wilson of the University of York, UK, and his graduate student Amelia Lunniss, have been visiting GA for the past 4 weeks, with funding from EUROFusion, to work with Phil Snyder on development of a new version of the ELITE code. ELITE was originally designed for the efficient study of intermediate to high toroidal mode number (n~5-40) peeling-ballooning modes which constrain the H-mode pedestal and drive edge localized modes (ELMs). Recent additions to the formalism and code now enable ELITE to also study low mode number (n<5) edge instabilities. The new version of ELITE will enable study of conditions such as Quiescent H-mode (QH), including the recently discovered wide pedestal QH mode, where low-n modes are expected to be the limiting edge instability. In addition, a new code diagnostic has been developed which enables separately studying the various stabilizing and destabilizing terms, including field line bending, ballooning, kink and peeling terms. An interesting observation is that the kink term, driven by current gradients, is important in the pedestal even at very high n, up to wavenumbers comparable to the inverse ion gyroradius. This has important implications for gyrokinetic studies in the pedestal region, which may also need to include this term.
Previous Month Highlights | Next Month Highlights
These highlights are reports of research work in progress and are accordingly subject to change or modification