Error-based grid adaptation methods for plasma edge simulations with SOLPS-ITER

Abstract

Only recently, plasma edge simulations up to the wall have been enabled with SOLPS-ITER. This requires dedicated gridding techniques to reconcile grid alignment with the magnetic field and refinement toward the wall as grid quality is primordial to ensure fast and reliable convergence. Therefore, the gridding approach for the grids up to the wall is analyzed and improved. A truncation error analysis is performed on the discrete operators of the discretization scheme in SOLPS-ITER, resulting in indicators of grid properties that are undesired. Based on these indicators, grid adaptation and grid smoothing algorithms are developed to reduce truncation errors and improve the overall grid quality. The resulting methods are applied on an AUG single-null case. Here, the impact of the new gridding strategy is examined on the divertor heat load, a typical quantity of interest for plasma edge simulations. The new gridding methods allow to mitigate spurious numerical spikes in the target heat load profiles, reduce the convergence time with a factor 30, and improve the accuracy of the heat load with a factor 3 compared to original grids with similar total number of cells.