Particle and Power Exhaust Studies in the EAST Fusion Tokamak and Computational Analysis Using Coupled Fluid-Kinetic Plasma Codes

This PhD aims contributing in finding a solution for the particle and power exhaust problem in tokamak operation. Because the combination of a double null magnetic configuration and neon seeding seems promising in full-carbon experiments, the main focus is on exploring this potential in metallic devices. This thesis is done in the EU-CN collaboration framework of EUROfusion and experiments are performed at the EAST tokamak, located at ASIPP in Hefei (China). This tokamak has an upper tungsten and a lower carbon divertor, allowing experiments in a partially metallic environment. On top, it can handle stable neon seeded discharges in high confinement mode. Due to the condition of the lower divertor, no coupled double null configuration was possible. Therefore, experiments with and without neon seeding, in single and disconnected double null configuration with main upper divertor were performed in 2019. The impact of the disconnected double configuration on the particle and heath fluxes reaching the divertor targets was negligible due to a too large separation between the separatrices. The neon seeding, on the other hand, seems to have a significant impact on the plasma profiles. An analysis of the experimental data shows that neon drove the discharge into a high-recycling regime. The ultimate goal, however, is to achieve detachment. Energy detachment was clearly reached, but based on the experimental data, no hard statement about the momentum detachment (which seems to be absent) and the particle detachment can be made. Therefore, SOLPS-ITER simulations are used for a further analysis of the two discharges in upper single null configuration. This numerical code solves the Braginskii equations and is a coupling between the fluid finite volume B2.5 plasma and the kinetic Monte Carlo EIRENE neutral codes. In the performed SOLPS-ITER simulations, attention is put on the introduced numerical errors. As expected, the main contribution is caused by the discretization of the B2.5 grid. The employed EIRNE grid does not have a large impact on the numerical error. Besides the B2.5 discretization error, also the bias should be considered while interpreting the simulation result for the plasma parameters. The use of a Monte Carlo treatment of the neutrals, introduces also a statistical error in the final result. By making an appropriate choice of the EIRENE input parameters and employing the earlier developed SOLPS-ITER averaging procedure, this error can be kept small. The numerical error on the neutral parameters is determined by a contribution due to the coupled simulation, and one due to the last two purely EIRENE steps. These last steps are performed with much more Monte Carlo particles to decrease the statistical error. By making an appropriate choice of the input variables, also the errors on these parameters can be kept small. By solving the original equations of the SOLPS-ITER code, reasonable agreement with the experiments is obtained at the outer midplane, but poor agreement is found at the outer divertor target. Therefore, sensitivity studies towards some not exactly known input parameters are performed. These studies, however, do not show the required changes in the plasma profiles to improve the agreement. This seems only possible by solving the extra electric current continuity equation making it possible to include drift effects in the simulation. The next points should be fulfilled to include drifts successfully in SOLPS-ITER: • Avoid too fine grid cells at the divertor targets. As a rule of thumb, use a minimum radial and poloidal cell width of 1 mm. • Start from a fully converged (and averaged) simulation without drifts. • Ensure that the anomalous conductivity is much smaller than the neoclassical one, but not too small to avoid a slow convergence. For EAST simulations cσα,0 = 5 · 10–5 is a suitable choice. • Apply the speed-up technique for convergence of Kaveeva et al. [57] in which the time derivative simulation time step size for different regions in the problem is changed, making it possible to use a time step of 10–5 s. • Increase the neutral source terms to improve the convergence speed. Depending on the simulation with or without neon seeding, some further requirements for the input parameters should be imposed. By including these drift effects, reasonable agreement between experiments and simulations is possible for all plasma profiles. An analysis of these final simulation results shows that detachment can be reached due to the added neon. The increased ion saturation current towards the outer target can be explained by the increased deuterium injection in the neon seeded discharge. This means that neon can drive EAST discharges in full detachment.

Jaar van publicatie:2021

Toegankelijkheid:Open