Abstract Details
Abstracts
Author: Adam Stanier
Requested Type: Consider for Invited
Submitted: 2022-03-03 15:28:06
Co-authors: L. Chacon
Contact Info:
Los Alamos National Laboratory
P.O. Box 1663
Los Alamos, NM 87545
USA
Abstract Text:
Hybrid kinetic-fluid models have long been used to model magnetized plasma experiments. Full-orbit versions can be complementary to the gyrokinetic (GK) model, due to their extended validity in regimes where GK ordering parameters may not be small. This can include strong gradient regions such as the tokamak pedestal, or fast occurring phenomena such as reconnection in a sawtooth crash or a tokamak disruption.
We will describe a novel implicit, electromagnetic particle-based scheme for this model [1]. It uniquely conserves mass and energy for general curvilinear meshes [2], as well as momentum for a subset of adaptive (packed) tensor-product meshes. The use of implicit time-stepping, along with exact conservation, lends favorable stability properties to the scheme. The basic algorithm is extended to treat multi-scale problems by using adaptive sub-stepping and orbit averaging of the full-orbit ions, to integrate their orbits accurately and reduce noise. Physics-based preconditioning, using a low-order fluid model to accelerate the high-order kinetic scheme, yields a significant performance gain when taking large timesteps. We will demonstrate the utility of the scheme for several numerical examples, including an m=1 kink mode in helical geometry.
Finally, we will comment on the mitigation of a previously unidentified cancellation issue [3] that arises from the hybrid discretization (with ions treated as particles and electrons on a mesh), and is common to all hybrid-kinetic schemes of this type.
[1] Stanier, A., Chacón, L., & Chen, G. (2019). J. Comp. Phys., 376, 597-616.
[2] Stanier, A., & Chacon, L. (2021). arXiv preprint arXiv:2110.03886.
[3] Stanier, A., Chacon, L., & Le, A. (2020). J. Comp. Phys., 420, 109705.
Comments:
Topic is Computer Simulation of Plasmas