Optimizing edge confinement and stability via adaptive ELM control using RMPs *

We report an adaptive real-time approach to control edge-localized modes with resonant magnetic perturbations (RMP) in a way that optimizes both the pedestal stability and confinement by exploiting the hysteresis in RMP ELM suppression [1]. Such adaptive control is essential to maximize the plasma performance while also maintaining an ELM-stable edge. We recover up to 60% of the original confinement degradation and 45% of the fusion gain factor (G=H₉₈β_N /q₉₅²) by using our method. The adaptive ELM control uses a Dα-based ELM detector and adjusts the RMP accordingly. It iteratively increases and decreases the RMP amplitude during ELMy and ELM free phase, respectively, until it converges to a stable operating point that optimizes both ELM-free and confinement. Here, we find that RMP-induced ion-scale turbulence during ELM free phase widens the ion pedestal and improves pedestal stability [2], allowing a higher pedestal pressure and amplified field penetration. Nonlinear MHD simulations based on JOREK [3] and TM1 [4] show that collisional transport due to magnetic island is insufficient to explain the ion-pedestal behavior, supporting the idea of its broadening by edge turbulence. In general, the adaptive control is not trivial because the threshold characteristics of the bifurcation in and out of ELM suppression can lead to control system oscillations. Our approach has overcome these limitations by outcomes of turbulence, which lowers the RMP threshold for recovering ELM free operation and weakens the control oscillation. Such a favorable effect is difficult to be harnessed as turbulence quickly disappears with ELMs. However, the adaptive method fully exploits it by re-increasing RMP immediately after the reduction of turbulence and before the ELMs return, which is key to control convergence.[1]F. Laggner et al.,NF60,076004 [2]T. Osborne et al., NF55,063018 [3]G. Huysmans et al.,PPCF51,124012 [4]Q. Yu et al.,NF51,073030