Sherwood 2015

Abstract Details

files Add files

ITG turbulence in coupled XGC1-XGCa multiscale simulations

Author: Salomon Janhunen
Requested Type: Poster Only
Submitted: 2015-01-20 13:54:25

Co-authors: R.Hager, C-S.Chang, S-H.Ku, J.Hesthaven, J.Choi, F.Zhang, M.Prashar

Contact Info:
Princeton Plasma Physics Lab
100 Stellarator road
Princeton, NEW JERSEY   08543
United States

Abstract Text:
We have developed a novel multiscale coupling algorithm for the acceleration of gyrokinetic full-f particle simulations in diverted geometry using the codes XGC1 and its axisymmetric version XGCa. In order to establish a basis for further development of our approach, we studied how ITG turbulence induced transport is modified by coupled XGC1/XGCa operation as compared to pure XGC1 simulations.
Both XGC1 and XGCa calculate particle motion in a 5-dimensional (5D) phase space, but while XGC1 is equipped with a 3D Poisson solver for turbulence simulations, XGCa has an axisymmetric 2D Poisson solver and is used for the simulation of the neoclassical evolution of the plasma background. Acceleration is achieved through relaxed constraints on numerical requirements in XGCa, such as mesh resolution, step size and total number of markers.
Coupled simulations have been performed in the weak coupling regime, where the turbulent component of the electrostatic potential (|n|>0) from XGC1 is used as a scattering operator while evolving the neoclassical equilibrium self-consistently with XGCa. Our results demonstrate strict coherence requirements between potential and ion density for driving anomalous transport. Even in cases where exactly the same initial kinetic state is used, the transport level is observed to decrease to roughly half of the level of the fully self-consistent simulation with XGC1. This behavior is observed both in the linear growth phase and in fully saturated turbulence states.
In addition to the phenomenon discussed above, we introduce in-memory techniques used in the coupling between the fine-scale and coarse models, applicable for massively parallel simulations of long term evolution of kinetic plasma equilibria in the presence of turbulent and neo-classical transport processes.

Comments:

March 16-18, 2015
The Courant Institute, New York University