HBT-EP program: MHD dynamics and active control through 3D fields and currents

The HBT-EP research program aims to: (i) develop new machine learning and real-time data reduction methods for active control, (ii) explore the connections between edge transport physics, non-symmetric magnetic fields, and rotation, and (iii) study the role of large-scale passively-driven magnetic perturbations on disruption dynamics, relativistic electron loss, and electromagnetic loads. A convolution neural network (CNN) was developed to predict the amplitude and phase of an n=1 MHD mode using solely optical measurements from the upgraded HBT-EP high-speed videography system[1]. A real-time application of this algorithm using an FPGA has achieved a latency of 17.6 μs, on par with the magnetic sensor GPU-based control system [2]. A two-color multi-energy EUV/SXR tangential array has been used to observe the suppression of sawtooth MHD activity correlated with the coupling of an internal 2/1 mode and an external 3/1 kink mode consistent with observations of flux pumping as has been seen in hybrid plasmas in DIII-D and JET [3]. A Runaway Electron Mitigation Coil (REMC) driven by the disruption induced loop voltage to provide a “fail‐safe” prevention of high‐energy runaway electrons has been installed inside the HBT-EP vacuum vessel to carry out the first experimental test of the REMC concept.

[1] Y. Wei, et al., PPCF 65 (2023) 074002

[2] Y. Wei, et al., RSI accepted for publication: https://arxiv.org/html/2312.00128v2

[3] Boting Li, et al., Nuclear Fusion 64 (2024) 046020.