Modeling of the equilibrium of a tokamak plasma

The simulation and the control of a plasma discharge in a tokamak require an efficient and accurate solving of the equilibrium because this equilibrium needs to be calculated again every microsecond to simulate discharges that can last up to 1000 seconds. The purpose of this thesis is to propose numerical methods in order to calculate these equilibrium with acceptable computer time and memory size. Chapter 1 deals with hydrodynamics equation and sets up the problem. Chapter 2 gives a method to take into account the boundary conditions. Chapter 3 is dedicated to the optimization of the inversion of the system matrix. This matrix being quasi-symmetric, the Woodbury method combined with Cholesky method has been used. This direct method has been compared with 2 iterative methods: GMRES (generalized minimal residual) and BCG (bi-conjugate gradient). The 2 last chapters study the control of the plasma equilibrium, this work is presented in the formalism of the optimized control of distributed systems and leads to non-linear equations of state and quadratic functionals that are solved numerically by a quadratic sequential method. This method is based on the replacement of the initial problem with a series of control problems involving linear equations of state. (A.C.)