MOSITER : MOdeling and Simulation for ITER
Partners:
Institut de Mathématiques de Toulouse (Pierre Degond, Naoufel Ben Abdallah): coordinator
IRIT (Luc Giraud, Mathias Paulin)
Project information:
Funding: Université Paul Sabatier
Period: 2009-2010
Summary:
The ITER project (International Thermonuclear Experimental Research) aims at mastering the fusion reaction which are the energy source of the stars. The goal is to confine a ionized gas or plasma with a magnetic field and to heat it during a time long enough to ignate an atomic fusion reaction between deuterium and tritium. ITER belongs to the class of Tokamak devices where the shape of the confinement magnetic field lines are toroidal.
In France, the Fédération nationale de Recherche Fusion par Confinement Magnétique – ITER provides the framework for the collaborative effort between the Universities, the CNRS, the CEA, the INRIA.
The goal of the MOSITER project is to enhance the collaboration between, the Toulouse groups involved in research on ITER, namely the IMT (math.), the LAPLACE (plasma physics), the LCAR (quantum physics) and the IRIT (Computer Science). It concerns the numerical modeling of various aspects of ITER and the search for efficient algorithms allowing breakthroughs in computing efficiency compared to the state-of-the-art.
Three lines of actions have been defined:
Action 1: Beyond state-of-the-art numerical methods for the plasma and the neutrals injector:
In the current state-of-the-art, specialized codes for the various locations (the core, the edge, the divertor, etc. ) are available and a lot of effort is put in the coupling of these codes one to each other. However, the results often depend on the coupling strategies and lack robustness. In a decade or two from now, the available computer power will allow global simulations of the full device and our goal is to prepare the numerical algorithms which will be needed to face this challenge. The core of this project consists of the so-called Asymptotic-Preserving methods which allow simulations over a large range of scales with the same model and numerical method. These methods represent a breakthrough with respect to the state-of-the art. They will be developed specifically to handle the various challenges related to the simulation of the ITER plasma and of its neutral particle source.
Action 2: Pre-conditioning and parallelization of strongly anisotropic diffusion equations
Due to the very large magnetic field, the plasma conductivity is strongly anisotropic and leads to strongly anisotropic diffusion equations. Specific methods for handling this kind of problem are being developed. A key problem in the use of this method is their pre-conditioning and their ability to run on massively parallel machines. The MOSITER project will aim at finding efficient pre-conditioning and parallelization algorithms for these problems.
Action 3: Massive parallelization of particle methods on Graphic Process Units (GPU)
An accurate physical modeling of the ITER plasma requires kinetic models (due to the low collisional plasma) and their resolution through Particle-In-Cell methods. Extremely powerful algorithms for ray tracing are currently developed and adapted for massively parallel Grapics Process Units. The MOSITER project aims at extending these algorithms to PIC methods.
Description of the work performed during the project
Related publications: