Mardi 19/11/2013, 10h00-11h30
Salle René GRAVIER 506 rdc, CEA Cadarache

devant le jury composé de :


Prof. Laurent VILLARD,            Ecole Polytechnique Fédérale de Lausanne (EPFL)   Rapporteur

Prof. Eric SONNENDRÜCKER,    Max Planck Institute of Plasma Physics (IPP)          Rapporteur

Prof. Peter BEYER,                   Université d’Aix-Marseille                                     Examinateur

Prof. Steve COWLEY,               UK Atomic Energy Authority (UKAEA)                    Examinateur

Prof. Nabil NASSIF,                  American University of Beirut (AUB)                      Examinateur

Prof. Yanick SARAZIN,             CEA Cadarache                                                   Examinateur

Dr. Maurizio OTTAVIANI,          CEA Cadarache                                                   Directeur de thèse


Mots-clés: field-aligned coordinates, turbulence simulations



The primary thrust of this work has been the development and implementation of a new approach to the problem of field-aligned coordinates in magnetized plasma turbulence simulations called the FCI approach, standing for Flux-Coordinate Independent. The method exploits the elongated nature of micro-instability driven turbulence which typically have perpendicular scales on the order of a few ion gyro-radii, and parallel scales on the order of the machine size. Mathematically speaking, it relies on local transformations that align a suitable coordinate to the magnetic field to allow efficient computation of the parallel derivative. However, it does not rely on flux coordinates, which permits discretizing any given field on a regular grid in the natural coordinates such as (x, y, z) in the cylindrical limit. The new method has a number of advantages over methods constructed starting from flux coordinates, allowing for more flexible coding in a variety of situations including X-point configurations. In light of these findings, a plasma simulation code FENICIA has been developed based on the FCI approach with the ability to tackle a wide class of physical models. The code has been verified on several three dimensional test models. The accuracy of the approach is tested in particular with respect to the question of spurious radial transport. In this regard, numerical radial diffusion is shown to be easily kept under control with the choice of suitable algorithms, at a minimal computational cost. Further tests on  3D models of the drift wave propagation and of the  Ion Temperature Gradient (ITG) instability in cylindrical geometry in the linear regime demonstrate again the high quality of the numerical method. Moreover, simulations in the nonlinear ITG models allows one to recover, at reduced numerical cost, the standard features of slab ITG turbulence. Finally, the method is shown to be able to deal with an X-point configuration such as one with a magnetic island with good, convergence and conservation properties.

Contact : TH099389