Berlin 2014 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

P: Fachverband Plasmaphysik

P 6: Theory and Modelling

P 6.4: Talk

Monday, March 17, 2014, 17:15–17:30, SPA HS202

Weakly adaptive particle weighting algorithms in Particle in Cell Codes. — •Denis Eremin, Ralf Peter Brinkmann, and Thomas Mussenbrock — Ruhr-Universität Bochum, Universitätsstrasse 150, Bochum, Deutschland

Particle-in-cell simulations of low-temperature plasmas are often made in 2D cylindrical (r,z) geometry, following typical symmetry of the experimental facilities. In such a geometry one usually disretizes electrostatic field (electromagnetic fields) on a grid, which is either uniform in r squared, or uniform in r. The second method of discretization yields better resolution of the field(s) near the axis, but requires an adaptive weighting algorithm for the particles. Assuming constant number of superparticles in each cell at the beginning of the simulation (which is warranted by the statistical nature of the PIC codes), initial superparticle weights must be proportional to their radial position. During the simulation, superparticle with large weights coming from the radial periphery close to the axis tend to replace the particles with small weights initially located at the axis, which leads to dramatic increase of the statistical noise there. The superparticles with small weights, in turn, tend to accumulate close to the periphery of the discharge, which is undesired due to the memory limitations. Once in a while, superparticles with large weights must therefore be split in the course of their motion toward the axis, and superparticles with small weights must be merged as they tend toward the radial periphery of a discharge. In this contribution the authors discuss how to construct such an algorithm which would distirb the physics of the discharge only minimally.