Physics and computational simulations of plasma burn-through for tokamak start-up

Abstract

This thesis will discuss the fundamental process of high temperature plasma formation, consisting of the Townsend avalanche phase and the subsequent plasma burn-through phase. By means of the applied electric field, the gas is partially ionized by the avalanche process. In order for the electron temperature to increase, the remaining neutrals need to be fully ionized in the plasma burn-through phase, as radiation is the main contribution to the electron power loss. The radiated power loss can be significantly affected by impurities resulting from interaction with the plasma facing components. The parallel transport to the surrounding walls is determined by the so called connection length in the plasma. Previously, plasma burn-through was simulated with the assumptions of constant particle confinement time and impurity fraction. In the new plasma burn-through simulator, called the DYON code, the treatment of particle confinement time is improved with a transonic ambipolar model for parallel transport, by using the effective connection length determined by the magnetic field lines, and Bohm diffusion model for perpendicular transport. In addition, the dynamic evolution of impurity content is calculated in a self-consistent way, using plasma wall interaction models. The recycling of the particles at the walls is also modelled. For a specific application, the recent installation of a beryllium wall at Joint European Torus (JET) enabled to investigate the effects of plasma facing components on plasma formation and build-up of plasma current in the device. According to the JET experiments the Townsend avalanche phase was not influenced by the replacement of the wall material. However, failures during the plasma burn-through phase, that could occur with a carbon wall, are not observed with a beryllium wall. In order to obtain a deeper insight in these effects this thesis will present detailed modelling of plasma burn-through. For the first time a quantitative validation of the simulation results to experimental data is documented. The simulation results with the DYON code show consistent good agreement against JET data obtained with the carbon wall as well as the beryllium wall. According to the DYON results, the radiation barrier in a carbon wall is dominated by the carbon radiation. The radiation barrier in the beryllium wall is mainly from the deuterium radiation rather than the beryllium radiation, as far as the radiated power from other impurities (i.e. carbon, nitrogen, etc) is not significant. These issues are of crucial importance for the International Thermonuclear Experimental Reactor (ITER) where the allowable toroidal electric field for plasma formation is limited to 0.35 V/m, which is significantly lower compared to the typical loop voltage ( 1 V/m) used in the current devices. Using the validated DYON code, predictive simulations for ITER are given, showing a need for RF heating to allow reliable plasma burn-through.