Modeling of DIII-D discharge #148712 finds a clear correlation between ELM suppression and stochasticity in the plasma edge. In this discharge, even-parity n=3 fields from the I-coils were applied, with the I-coil current flipped at 5 Hz. It was found that ELM suppression was significantly more robust in the phases in which the current in the upper coil at 30 degrees (IU30) was positive. Two-fluid M3D-C1 modeling of the plasma response in 24 time slices during the discharge (12 in each phase), combined with field-line tracing analysis with MAFOT, revealed significantly more stochasticity in the phases where the IU30 current was positive. This difference was especially pronounced early in the discharge, where the plasma density was lower and ELM suppression was more easily achieved. Experimentally, the difference in behavior between phases must be due to error fields; however, since the modeling did not include error fields, the difference in the modeled plasma response can only be due to differences in the reconstructed axisymmetric equilibrium. In particular, large changes in the edge plasma rotation are observed between phases early in the discharge, and are likely responsible for the difference in plasma response. It is not yet fully understood what role the change in the non-axisymmetric plasma response plays in the suppression of ELMs. This work will be presented at the 2013 APS-DPP meeting by NUF student Matthew Brown.

A generalized kinetic treatment was used to calculate the friction force from the velocity slip between the ions and neutral fluid in a supersonic MGI gas jet penetrating an ITER plasma in the current quench phase. This friction force can make the jet stop sooner than our previous published hydrodynamic (ideal) penetration calculation. In the weakly ionized core of a supersonic MGI gas jet, ions formed upstream initially move along with the neutral jet but experience a retarding Lorentz force and an even greater Hall electric field force. A velocity slip U develops downstream between the ions and neutral fluid, giving rise to the friction force on the jet. The model assumes the neutrals are stationary with zero temperature but the ion distribution function is described by a drifting Maxwellian with a temperature that is equal to the neutral particle kinetic energy in the lab frame. It was found that the ratio of the charge exchange (CX) induced friction to the friction arising from elastic scattering collisions is 0.0031 U (m/s), which for a neon jet is a factor ~ 2. The derived formula for the CX friction force was found to be identical in form to a recently published result of Fruchtman derived from the Boltzmann collision integral.

In collaboration with Nathan Howard (UCSD), who completed a one-week visit with the GA theory group, progress was made in better understanding the L-mode near-edge “transport shortfall” predicted by GYRO and TGLF, via modeling of coordinated Ip-scaling experiments in Alcator C-mod and DIII-D. Initial TGLF analysis of the DIII-D discharges, finds that consistent with previous studies, a clear shortfall persists beyond rho=0.7, and that the strength of shortfall in the ion channel increases as Ip decreases, while the electron channel exhibits a strong shortfall for all values of Ip. However, an extensive set of GYRO simulations of C-mod discharges finds no shortfall in the ion channel at any current or radii, and a shortfall in electron channel which increases with Ip, opposite the trend observed in DIII-D. Future work will focus on gyrokinetic modeling of the DIII-D discharges in order to better understand these discrepancies, as well as multi scale ITG-ETG simulations of the C-mod plasmas to identify whether ETG turbulence can provide the missing electron thermal transport at high Ip. This work was performed as part of the research activities of the CSPM SciDAC. Additional progress will be reported at the 2013 APS-DPP meeting.

Val Izzo was Session Leader in the DIII-D experiment proposed (see Highlight from April 05 2013 at Theory Weekly Highlights for April 2013) to test predictions from NIMROD calculations that the phase of the 1/1 mode during the MGI thermal quench determines the heat flux distribution. While a clear effect of external field phase was observed (as predicted), it was also found that applying the I-coil fields earlier in time prior to the shutdown noticeably changed the trend vs. phase on some diagnostics—particularly measurements far from the gas injection location. One possible explanation is a slowing of the edge rotation by the applied fields, producing a change in the toroidal impurity spreading. However, this effect must be separated from the effect of allowing the n=1 fields more time to penetrate before initiating the MGI shutdown. Further analysis will be required to make a more detailed assessment of the level of agreement between the experimental results and specific NIMROD predictions.

Dr. Gaimin Lu from the Southwestern Institute of Physics (Chengdu, China) has recently joined the GA Theory Group as a visitor for one year. She is here on a postdoctoral fellowship from the Chinese government. Dr. Lu is hosted by Ron Waltz and will mainly be working with Chris Holland (UCSD) on GYRO/ NEO/TGLF/TGYRO simulations and analysis of DIII-D transport experiments.

A GYRO linear stability analysis of a sample ITER discharge shows Alfvén eigenmode (AE)-driven transport of fusion alpha particles is robustly confined to the plasma core. Using TGLF-predicted thermal species profiles and a self-consistent classical slowing-down alpha distribution assuming no AE transport, AEs are unstable only in the central core. The n=20 modes, which are among the most strongly driven, are unstable only within 0.20⇐r/r0 ⇐ 0.55, where r0 is the radius at the top of the pedestal. The peak drive, near r/r0 = 0.35, has an alpha density gradient at about twice the local instability threshold. Under the assumption of stiff AE-induced alpha transport, the measured level of instability is insufficient to drive an alpha transport avalanche to the pedestal top, meaning no alphas are lost from the machine due to AE transport. When a Maxwellian velocity dependence is used, the peak growth rate (occurring at r/r0 = 0.45) is nearly double but the boundaries of the unstable region are unchanged.