Theory Weekly Highlights for March 2009
March 27, 2009
The new unified F90/95 version of EFIT for different operating platforms (see highlight from February 20 2009 at Theory Weekly Highlights for February 2009) has now been fully developed and tested and the MPI option and high spatial grid 257 and 513 options are now available. A single version of EFIT and its Makefile can now be used for both HP-based UNIX and Linux-based workstations. Benchmarking calculations against the public EFIT version have also been completed and results from this new version agree closely with those from the public version. Work is underway to convert all EFIT arrays to be dynamically allocated so that a single EFIT version can also be employed for different spatial grid sizes and tokamak devices.
March 20, 2009
Transport simulations using the Trapped Gyro-Landau-Fluid (TGLF) transport model have successfully reproduced changes in the temperature profiles in DIII-D hybrid discharges resulting from the extension of the hybrid scenario to ITER-relevant conditions. In particular, a large increase in temperature was observed during ELM suppression experiments with resonant magnetic perturbations as a result of the decrease in density and TGLF modeling reproduced these changes including the relatively large rise in the ion temperature. The TGLF transport model also matched the measured increase in the electron temperature with strong ECH in hybrid plasmas; however, in this case the predicted drop in ion temperature during ECH was smaller than the experimental observation. In these hybrid plasmas, calculations indicate that about half of the electron thermal transport is due to the electron temperature-gradient (ETG) mode.
March 13, 2009
In a comparison between NIMROD simulations with fixed-amplitude and ramped-up RMP fields very similar steady state solutions were obtained but differences in transient behavior produced some effects lasting several reconnection times. In two simulations, the RMP fields were held fixed, one with no rotation and one with toroidal plasma rotation. In two more recent, but otherwise identical simulations, the RMP fields were ramped linearly with a ramp time of 0.1 ms, comparable to a reconnection time, and then held fixed for several more reconnection times as a steady state solution is achieved. With no rotation the plasma responds violently to the superimposed vacuum fields in the fixed amplitude case and a small residual n=1 oscillation continues as a steady state is approached. In contrast, the case with the RMP field ramp obtains identical final amplitudes for the n=3 and n=1 field components, but does not exhibit the n=1 oscillation. The two simulations with rotation also achieve very similar steady state solutions; both rotating cases exhibit a large amplitude n=1 oscillation which is independent of the RMP ramp and is associated with the plasma rotation frequency. Hysteresis is possible in a rotating case but is not expected in the highly resistive and viscous regime of the simulations. In the future, hysteresis effects will be explored in a higher rotation or lower resistivity regime.
March 06, 2009
A version of the TGLF confinement model based on the GCNMP global Newton solver has been installed and is operational in the ONETWO transport code. Several improvements to GCNMP were made to handle the extremely computationally intensive TGLF model. These include a Jacobian-free solution method, and finite-difference form of the transport equations in terms of fluxes rather than diffusivities. Also, a “compact derivative” scheme that uses all grid points to simultaneously define higher-order derivatives across the entire computational grid was implemented to provide a smoother representation of derivatives and this results in smoother predicted temperature profiles. Benchmarking calculations for two DIII-D discharges and an FDF configuration show good agreement against the XPTOR transport code and the experimentally measured profiles to within 10 to 15%. Some of this difference may be due to the difference in the background neoclassical confinement models used in these two codes.
Disclaimer
These highlights are reports of research work in progress and are accordingly subject to change or modification