Modelling and optimization of runaway electrons in tokamaks

Licentiatavhandling, 2025

During tokamak start-up and disruptions, strong electric fields can arise which are sufficient to cause electron runaway, whereby electrons are accelerated continuously. In future large-current tokamaks, such as ITER and SPARC, significant runaway electron generation is expected. Should the runaway electron beam come in contact with the tokamak wall, its energy can be almost instantly deposited into the wall. After a disruption, a considerable fraction of the plasma current could be carried by relativistic electrons, which could seriously damage the device. Start-up runaway electrons also risk causing damage to the tokamak, but can also impede the plasma initiation. Electron runaway is one of the major unsolved challenges in the development of fusion as a viable source of energy. This thesis focuses on accurate modelling of tokamak start-up and disruptions as well as the optimization of disruption mitigation, centering especially around runaway electrons.

The simulation tool Stream has been developed for studying runaway electrons during the burn-through and ramp-up phases of tokamak start-up. Stream uses a 0D plasma model, where the densities, currents, temperatures and electric field are evolved self-consistently. The runaway electron evolution is governed by Dreicer and avalanche generation, as well as particle transport. Using Stream, it was found that Dreicer generation plays a crucial role for start-up runaway dynamics, and can even dominate the runaway generation.

Fluid and kinetic modelling of the runaway seed generation during tokamak disruptions have been compared. It was found that the two models can give significantly different predictions of the runaway evolution. The largest difference found concerned the hot-tail generation, as the neglect of radial transport in the fluid model caused a significant overestimation of the runaway generation rate. Kinetic modelling of the seed generation was thus found to be preferable, despite the increased computational cost.

Disruption optimizations for both ITER and SPARC were performed, focused on minimizing heat loads, electromechanical forces and the runaway current. More specifically, the injected densities of deuterium and noble gases during massive material injection were optimized. For ITER, simultaneous minimization of all three objectives was found to be possible only in pure deuterium plasmas. During activated operation, low runaway currents always correlated with large transported heat losses. For SPARC, successful mitigation was found to be feasible in deuterium-tritium plasmas as well.