Table of Contents
Fetching ...

Numerical methods for stellarator simulations in BOUT++

David Bold, Brendan Shanahan

Abstract

Modeling the Scape-off layer (SOL) of stellarator fusion devices is challenging due to the complicated magnetic topology, requiring numerical tools to solve transport equations for realistic geometries. Previously the flux coordinate independent (FCI) method has been successfully applied to model the SOL in simplified geometries. The current work presents some of the recent improvements for the BOUT++ modeling implemented to simulate the SOL in realistic geometries with the example of Wendelstein 7-X. The changes include improvements for the grid generation tool, the physics model as well as the BOUT++ library itself. A short outlook is given on current modeling work using the new features.

Numerical methods for stellarator simulations in BOUT++

Abstract

Modeling the Scape-off layer (SOL) of stellarator fusion devices is challenging due to the complicated magnetic topology, requiring numerical tools to solve transport equations for realistic geometries. Previously the flux coordinate independent (FCI) method has been successfully applied to model the SOL in simplified geometries. The current work presents some of the recent improvements for the BOUT++ modeling implemented to simulate the SOL in realistic geometries with the example of Wendelstein 7-X. The changes include improvements for the grid generation tool, the physics model as well as the BOUT++ library itself. A short outlook is given on current modeling work using the new features.

Paper Structure

This paper contains 17 sections, 5 equations, 18 figures.

Figures (18)

  • Figure 1: Example of a cell for the finite volume scheme showing a cell. The cell is an octagon around the point of the grid.
  • Figure 2: Convergence plot of a MMS test of the differential operator for a hollow square domain for different radial resolution $nx$.
  • Figure 3: Convergence plot of a MMS test of the differential operator for a W7-X grid and comparison to previous differential operators.
  • Figure 4: Scaling of the MMS test of the PETSc parallelized implementation of the FCI operator.
  • Figure 5: Scaling of wall clock computation time per evaluation of the right hand side of the Hermes-2 physics model. This includes several FCI parallel derivatives and other MPI communications, but does not include the time solver and disk i/o.
  • ...and 13 more figures