Mitigation of initial transients in gyrokinetic turbulence simulations using numerical distribution function of neoclassical equilibria
POSTER
Abstract
The presence of initial transients with geodesic acoustic mode oscillations—arising from initializing total-f/full-f gyrokinetic simulations with a Maxwellian distribution and profile gradient—can cause prolonged relaxation phases, complicating gyrokinetic turbulence studies. One method to avoid these transient oscillations is to use a canonical Maxwellian distribution, which is as an analytical steady-state solution of the Vlasov equation in the absence of radial electric fields. However, the Coulomb collisions and the development of radial electric fields quickly render the canonical Maxwellian a poor steady-state approximation. In this work, we present a strategy to mitigate the initial transients by employing a numerically computed neoclassical quasi-steady state distribution as the initial condition. This distribution is from the actual magnetic geometry and plasma profiles, enabling a "quiet start" for total-f/full-f gyrokinetic simulations. Results using the XGC code in realistic diverted tokamak geometries show marked improvements in the numerical performance and fidelity of gyrokinetic turbulence simulations.
*This research is supported by DOE FES through the SciDAC-5 Center, Computational Evaluation and Design of Actuators for Core-Edge Integration (CEDA), under award DE-AC02-09CH11466. Simulations utilized resources from the NERSC under award FES-ERCAP002368, as well as the Oak Ridge Leadership Computing Facility at ORNL, under award DE-AC05-00OR22725.
Presenters
-
S.-H. Ku
- Princeton Plasma Physics Laboratory
- Princeton Plasma Physics Laboratory (PPPL)