NEO analysis of deuterium and tritium neoclassical transport in ITER-relevant scenarios finds that the Neutral Beam Injection (NBI) particle source has an important role in producing the final stationary values of the neoclassical pinch. While the neoclassical diffusion coefficients are similar with and without the NBI particle source, without NBI the large inward pinch of D and T in the core is significantly reduced. For both cases, the density gradient from the opposite ion species plays a dominant role in determining the pinch velocity from mid-radius to the edge. However, for the case without NBI, it is instead the temperature gradient of the opposing species which largely determines the pinch velocity near r/a=0.1. The next step is to see how the inward pinch builds up as the NBI source is turned on. In addition, analysis of tungsten transport for a high temperature case, for which the transport is likely to be neoclassically-dominated, finds that the tungsten particle flux in the pedestal is very strongly radially outward, indicating no central accumulation. This is attributed to the good temperature screening.

Runaway electron (RE) confinement was studied in the context of a shell-pellet-like rapid shutdown scenario in DIII-D using NIMROD [see May 20 highlight]. It was found that when the plasma is cooled from the center due to direct deposition of impurities in the core, the inner flux surfaces are destroyed first, and the outermost flux surfaces remain intact until the end of the thermal quench (TQ) phase when nearly all of the thermal energy has already been radiated. Drift-orbits for over 2000 randomly seeded runaway REs were calculated during a NIMROD shell-pellet simulation. So long as the vast majority of the seeded REs remain confined while the outer flux surfaces remain intact, once they are destroyed nearly all of the tracked REs escape and hit the outer divertor within a ~ 0.1ms time window. This represents a much larger loss fraction than is usually seen in MGI simulations for DIII-D, which is typically on the order of 50% to 70%. The so-called “prompt loss” of REs during the TQ is desirable since it potentially prevents the small seed population formed during the rapid temperature drop from avalanching into a larger and more menacing RE plateau. The successful separation of RE loss to the divertor from electron heat loss to the divertor seen in this simulation is especially promising as a disruption mitigation scenario.

Dylan Brennan and Chang Liu visited this week to discuss analysis and future plans for runaway electron experiments on DIII-D.

We welcome Dr. Federico Halpern, who joined the Turbulence & Transport Group at GA earlier this week. Previously Dr. Halpern was at EPFL in Lausanne where he studied turbulence in the scrape-off layer, physics of the sawteeth, and magnetic reconnection. At GA he will continue work on fluid and kinetic modeling of the plasma edge region and also provide support for our future efforts in integrated modeling.

A preliminary implementation of the neural-network versions of both TGLF (tglfnn) and EPED1 (eped1nn) has been made directly into TGYRO. This removes a loose-coupling loop previously done in OMFIT and replaces it with tight coupling via the TGYRO root-finding algorithm. An additional advantage was found since the neural networks yield outputs that are smooth functions of the inputs, the convergence properties of TGYRO in this case are remarkably good.