Numerical study on neoclassical tearing mode stabilization via minimum growth rate seeking method

Abstract

Neoclassical tearing mode (NTM) is one of critical instabilities which need to be stabilized to achieve high fusion performance. It degrades plasma confinement and sometimes can cause plasma disruptions. Injecting a localized electron cyclotron current drive (ECCD) has been proven experimentally as an effective method to stabilize NTM by replacing the missing bootstrap current in the island. As the efficiency of this method strongly depends on the alignment between the island and the beam deposition, real-time feedback control is essential to achieve robust suppression of NTM. In this thesis, a concept of minimum growth rate seeking control is proposed and the feedback control of the growth rate of the magnetic island is shown to be faster and more efficient than that of the island width. As the minimum growth rate seeking method is a non-model-based control method, it has advantages of no requirement of precise real-time equilibrium reconstruction and EC ray-tracing calculation or NTM plasma system identification. An integrated numerical model is setup for time-dependent simulations of NTM evolution where plasma equilibrium, transport, and heating and CD by EC are coupled with the solver of simplified Rutherford equation in a self-consistent way. It is applied to KSTAR plasmas for reproducing the evolution of the island width in a NTM stabilization experiment. To evaluate the performance of the growth rate control concept, predictive feedback control simulations are performed based on two types of minimum seeking controller in KSTAR

finite difference method (FDM) and sinusoidal perturbation. Realistic control input and output parameters are used as the format of measured island growth rates from diagnostics and the poloidal launcher angle, respectively. The results are compared with the minimum island width seeking method. It is revealed that the proposed control concept is less limited in minimum seeking and more robust in reducing misalignment in shorter time scales.