A more detailed analysis of the linear equilibrium response to an external n = 3 nonaxisymmetric perturbed field from the I-coils in DIII-D for discharge #142603 found that the predictions from MARS-F, M3D-C1, and IPEC are actually in reasonable quantitative agreement. Previously (see the highlight for February 17 2012 at Theory Weekly Highlights for February 2012) it was found that these predictions were qualitatively the same and similar to the nonlinear prediction from M3D-C1 but all different from the VMEC prediction. The codes include different physics and implement the nonaxisymmetric fields in different ways; for example, IPEC sets the fields and MARS-F and M3D-C1 specify models for the coil currents. Once the phasing of the applied fields was correctly sorted out, the agreement between MARS-F and M3D-C1 particularly was found to be within about 10% in phase and amplitude, despite the fact that the latter case was for a resistive rotating plasma but MARS-F was ideal with rotation. The IPEC results were for an ideal plasma and did not include plasma rotation and there was a residual small phase shift of several degrees and 40% larger amplitude, both consistent with the expected shielding effect from rotation. Future work will focus on identifying the reason for the qualitatively different VMEC solutions.
Criteria for RMP ELM suppression have been evaluated by analyzing a series of DIII-D discharges in which RMP were applied. Various criteria, including island overlap width, local Chirikov parameters, and pedestal width were considered, including and omitting the plasma response. Including the plasma response (using calculations with M3D-C1) improved the correlation of the island overlap width with ELM suppression. The local Chrirkov parameter at the top of the pedestal was found to correlate more strongly to ELM suppression than island overlap width, which is consistent with the hypothesis that RMP ELM suppression is caused by stochasticity at the top of the pedestal that limits the pedestal width. Among measures of the equilibrium profile, it was found that the location of the center of the pedestal (in normalized poloidal flux) correlated more strongly with ELM suppression than the pedestal width, height, or maximum gradient. The database of discharges, which consisted of approximately 100 equilibria, is representative of a relatively small range of accessible parameter space in DIII-D and will be expanded in future studies.
For the first time the TGLF quasilinear model of gyrokinetic turbulent transport and the NEO model of neoclassical transport have been used to predict the plasma profiles in a DIII-D quiescent double barrier (QDB) discharge. All of the plasma profiles were predicted simultaneously (electron density, electron and ion temperature and toroidal rotation) for a three species plasma consisting of electron, deuterium, and carbon VI. The agreement between the predicted and measured profiles is excellent. Using the GLF23 model in place of TGLF predicts a run away to extreme density and rotation for this discharge due to the transport being reduced to neoclassical levels by the rotational shear. The higher linear growth rate accuracy and continuous wave number spectrum coverage of TGLF compared to GLF23 was found to have been crucial in yielding the successful prediction.
Two NIMROD simulations have been run in which a DIII-D discharge is terminated by Ne massive gas injection (MGI) while n=1 fields are applied in an attempt to influence the phase of the unstable m/n = 1/1 mode during the thermal quench (TQ). In MGI simulations without applied fields, the 1/1 mode always orients itself with a particular phase relative the toroidal location of the gas jet source. In one simulation the applied fields are aligned with the natural phase of the mode, while in the other case the applied fields have the opposing phase. In the latter simulation, the applied fields did not successfully force the unstable 1/1 mode to take on the opposite phase, but did delay the growth of the mode and reduce its amplitude. However, a larger radiated power spike was observed in that case. Simulations with larger amplitude applied fields will be carried out. In an upcoming DIII-D experiment, the I-Coils will be used to apply a 1/1 perturbation prior to MGI; these simulations suggest that measurable differences should be observed in the timing and amplitude of the measured radiated power, even if the mode is not successfully locked to the applied fields.
Gregorio Trevisan of Consorzio RFX is visiting GA for 1 month to collaborate with Aaron Sontag (ORNL) and Lang Lao on DIII-D 3D reconstructions with the V3FIT 3D equilibrium reconstruction code.
These highlights are reports of research work in progress and are accordingly subject to change or modification