Premium
Extensions of the 3‐Dimensional Plasma Transport Code E3D
Author(s) -
Runov A.,
Kasilov S.,
Schneider R.,
Reiter D.
Publication year - 2004
Publication title -
contributions to plasma physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.531
H-Index - 47
eISSN - 1521-3986
pISSN - 0863-1042
DOI - 10.1002/ctpp.200410002
Subject(s) - coordinate system , physics , magnetic field , plasma , tokamak , realization (probability) , monte carlo method , metric (unit) , statistical physics , diffusion , computational physics , geometry , mathematics , quantum mechanics , economics , statistics , operations management
One important aspect of modern fusion research is plasma edge physics. Fluid transport codes extending beyond the standard 2‐D code packages like B2‐Eirene or UEDGE are under development. A 3‐dimensional plasma fluid code, E3D, based upon the Multiple Coordinate System Approach and a Monte Carlo integration procedure has been developed for general magnetic configurations including ergodic regions. These local magnetic coordinates lead to a full metric tensor which accurately accounts for all transport terms in the equations. Here, we discuss new computational aspects of the realization of the algorithm. The main limitation to the Monte Carlo code efficiency comes from the restriction on the parallel jump of advancing test particles which must be small compared to the gradient length of the diffusion coefficient. In our problems, the parallel diffusion coefficient depends on both plasma and magnetic field parameters. Usually, the second dependence is much more critical. In order to allow long parallel jumps, this dependence can be eliminated in two steps: first, the longitudinal coordinate x 3 of local magnetic coordinates is modified in such a way that in the new coordinate system the metric determinant and contra‐variant components of the magnetic field scale along the magnetic field with powers of the magnetic field module (like in Boozer flux coordinates). Second, specific weights of the test particles are introduced. As a result of increased parallel jump length, the efficiency of the code is about two orders of magnitude better. (© 2004 WILEY‐VCH Verlag GmbH & Co. KGaA, Weinheim)