While GYRO and TGLF simulations of core transport and turbulence intensity for typical DIII-D L-mode discharges appear to be in good agreement with experiment, it has become apparent that there is a large shortfall in the simulated transport and turbulence in the edge region compared to the observed levels in the experiments. This shortfall was first pointed out by Holland and White in 2009. The source of this discrepancy is now being investigated. In electromagnetic global GYRO simulations of the near edge region (0.7 < r/a < 0.95) for discharges #128913 at 1500 msec, and 101391 at 2790 msec, the shortfall results mainly because the high near-edge electron-ion collisionality stabilizes the trapped electron modes (TEMs) so that the diffusivity decreases as T^7/2/n toward the cold edge. Artificially zeroing collisions so that the diffusivity decreases as T^3/2 robustly enhances the transport in both channels. By including higher k, electron temperature gradient (ETG) modes, the experimentally observed electron transport levels are recovered but the shortfall in the ion channel remains. The edge transport is highly local; artificially inducing huge turbulence levels at r/a > 0.9 does not reduce the shortfall at r/a=0.85. Additional driving mechanisms not in the standard GYRO formulation, such as ionization instability from interaction with edge recycled neutrals, and electron current drift gradient drive, were investigated but without success. Further investigation of the discrepancy is ongoing.

NIMROD modeling of L-mode plasmas with an applied RMP fields is being carried out to compare experimental probe measurements of edge profiles with the calculated plasma response. Initial simulations of Ohmic only cases with applied n=1 fields show little screening and only a slight phase shift in the 3/1 island due to low edge rotation. Probe data from DIII-D experiments indicates screening only in cases with higher edge rotation due to co-NBI. Further NIMROD simulations will investigate the higher rotation plasma and compare screening results to probe profiles. In the same set of experiments, n=3 RMPs were also applied, and the plasma response in these cases will be calculated with NIMROD following the n=1 simulations.

Several members of the Theory group actively participated in providing input to the DIII-D Program Advisory Committee sessions.

GA hosted a meeting of the Fusion Simulation Program (FSP), held at General Atomics. There were over 75 attendees, from six national labs, two companies, and nine universities. At this meeting the objectives of the FSP were discussed, along with computer and plasma science aspects of the program. The immediate goal is to complete a compelling plan for the FSP based on reports written by Integrated Science Activity subgroups. Dave Hill gave a plenary talk on experimental planning in major fusion facilities and how FSP can be involved.

Recent upgrades to the GYRO code have, for the first time, enabled definitive identification of subdominant global eigenmodes in the gyrokinetic model. Previously, the linear eigenvalue solver, GKEIGEN, could provide the complete spectrum of unstable eigenmodes only for relatively small local cases. The processor limit (usually 32 cores) in GYRO linear runs constrained the problem size through speed and memory limitations. A new GKEIGEN parallelization scheme now allows an effectively unlimited processor count, opening the possibility of large global eigenvalue calculations requiring 20,000 or more cores. Preliminary tests show almost perfect weak scaling as both the processor count and total problem size scale upward. This work was motivated by efforts to track global reverse shear Alfvén eigenmodes (RSAEs) in DIII-D which are often dominated by a toroidal Alfven eigenmode and/or an energetic particle mode with larger growth rate. Now the RSAE and any other subdominant, unstable eigenmodes, including drift waves excited by thermal species, can be tracked together in a straightforward way.