Numerical simulation techniques

Development of numerical simulation techniques for multiparticle systems

The rapid development of increased computer speed and memory allows multiparticle systems to be directly described by the dynamics of individual particles. At present, it is possible to handle up to 1010 individual particles. For the description of a real system which typically consists of 1022 particles, a subset of 1012 real particles is modelled by one numerical superparticle. The superparticle system which is generated describes the physical application with great precision and with acceptable fluctuations.

In a plasma consisting of ions and electrons, the dynamics of the system is mainly governed by electromagnetic forces. The electric and magnetic fields influence the positions and dynamics of the charged particles. In consequence, charge and current densities are changed and the fields are modified. The model thus remains consistent. Depending on the degree of approximation, a differentiation is made between hydrodynamic, multi-fluid, hybrid, and full particle techniques. We are developing codes of the last two methods.

Hybrid-Code
The electrons are described as a fluid whereas the ions are modelled by individual particles. Thus kinetic effects on the ion scale are included; characteristic effects on the electron scale are neglected. The simulation code is 3-dimensional in velocity space and also in configuration space. Out latest version is the A.I.K.E.F.-Code. The abbreviation stands for Adaptive Ion Kinetic Electron Fluid.

A.I.K.E.F.

This simulation tool is one of the first adaptive hybrid codes. In regions characterized by smaller spatial physical scales an adaptive mesh refinement is applied and the fluid equations are solved on a refined spatial grid. Simultaneously, the number and weighting of the numerical superparticles is adapted to keep the numerical noise level low.

Full particle code
Ions and electrons are described by single particles. Thus kinetic effects on the ion scale as well as on the electron scale are included. In comparison with the hybrid code, the numerical requirement is considerably greater, as the electron time scale must be resolved at a much higher level. The code is constructed for equidistant CARTESIAN grids in three spatial dimensions. The first example illustrates the code applied to beam-plasma interaction on an ion thruster.

References

Motschmann, U., Hybrid-Simulationen in der Plasmaphysik, Institut für Kosmosforschung, Berlin, IKF-2/88, 1988.

Bagdonat, T., U. Motschmann, 3D Hybrid simulation code using curvilinear coordinates, J. Comp. Physics, 183,470-485, 2002.

Müller, Joachim; Simon, Sven; Motschmann, Uve; Schüle, Josef; Glassmeier, Karl-Heinz; and Pringle, Gavin J., A.I.K.E.F.: Adaptive hybrid model for space plasma simulations, Computer Physics Communications, Volume 182, Issue 4, p. 946-966. 2011. Link