Abstract Details
Abstracts
Author: Eric Held
Requested Type: Poster
Submitted: 2026-03-20 09:12:08
Co-authors: Jacob King , Robert Hager, Fatima Ebrahimi, Cody Balos, David Gardner, Carol Woodward, Natalie Beams and the CETOP team
Contact Info:
Fiat Lux, LLC
132 S 300 W
Logan, Utah 84321
US
Abstract Text:
Accurate predictions of neoclassical transport in the pedestal edge provide insight into high-confinement operation in tokamaks. Transport coefficients inferred from such predictions can modify the growth rates of linearly unstable edge modes, which are of interest to the SciDAC Center for Edge of Tokamak OPtimization (CETOP). In order to quantify such effects, NIMROD CDK predictions of neoclassical transport were benchmarked against XGC for CETOP relevant equilibria. Although initial results were encouraging, the block preconditioning strategy used in NIMROD CDK led to unreasonably large GMRES iteration counts when including the collisional plasma from the bottom of the pedestal out to the open flux region. In collaboration with the GPU/Time-Integration CETOP group, we have implemented an enhanced time discretization scheme that separately inverts the linearized Coulomb collision and advection (streaming/bouncing) operators in a Douglas-Rachford (DR) split (J. Douglas and H. H. Rachford, Transactions of the American Mathematical Society, 82(2):421–439, 1956). For the collisional part, small systems that only couple the 2D velocity space degrees of freedom are solved at each of the quadrature points in NIMROD’s 2D FE spatial domain. The result is fed into a perfectly preconditioned solve for the spatial advection piece effectively removing the need for GMRES iterations. In addition to presenting further details of the implementation of the DR scheme, we quantitatively compare NIMROD-CDK and XGC predictions for a variety of CETOP-relevant equilibria.
*Work supported by DOE SciDAC program, Center for Edge of Tokamak OPtimization (CETOP), under a subcontract to Fiat Lux LLC, Award Number S-AGR-25223-P-250012934.
Characterization: 1.0
Comments: