Scrape-Off Layer Turbulence in Tokamaks Simulated with a Continuum Gyrokinetic Code

TL;DR

This paper introduces a new continuum gyrokinetic code, Gkeyll, capable of simulating scrape-off layer turbulence in tokamaks with energy-conserving schemes, demonstrating initial results in various geometries and boundary conditions.

Contribution

The paper presents a novel Maxwellian-weighted discontinuous Galerkin method for gyrokinetics that conserves energy and reduces computational costs significantly.

Findings

Successfully simulated turbulence on open field lines in LAPD device

Demonstrated energy conservation with Maxwellian-weighted DG scheme

Achieved reduced computational cost compared to traditional methods

Abstract

We are developing a new continuum gyrokinetic code, Gkeyll, for use in edge plasma simulations, and here present initial simulations of turbulence on open field lines with model sheath boundary conditions. The code implements an energy conserving discontinuous Galerkin scheme, applicable to a general class of Hamiltonian equations. Several applications to test problems have been done, including a calculation of the parallel heat-flux on divertor plates resulting from an ELM crash in JET, for a 1x/1v SOL scenario explored previously, where the ELM is modeled as a time-dependent intense upstream source. Here we present initial simulations of turbulence on open field lines in the LAPD linear plasma device. We have also done simulations in a helical open-field-line geometry. While various simplifications have been made at present, this still includes some of the key physics of SOL turbulence, such as bad-curvature drive for instabilities and rapid parallel losses with sheath boundary conditions. This is useful for demonstrating the overall feasibility of this approach and for initial physics studies of SOL turbulence. We developed a novel version of DG that uses Maxwellian-weighted basis functions while still preserving exact particle and energy conservation. The Maxwellian-weighted DG method achieves the same error with 4 times less computational cost in 1v, or 16 times lower cost in the 2 velocity dimensions of gyrokinetics (assuming memory bandwidth is the limiting factor).