KSTAR as a Testbed for Reactor-Grade Density Regulation: Coordinated Gas and Pellet Fueling While Avoiding Density Limits

E. Schuster, H. Al Khawaldeh, V. Graber, S. T. Paruchuri, T. Rafiq, A. Pajares, J. W. Juhn, S.-H. Park

10th Asia-Pacific Conference on Plasma Physics

Busan, South Korea, October 11-16, 2026

Abstract

Precise regulation of plasma density is a fundamental requirement for achieving and sustaining active burn control in future reactor-grade tokamaks. To optimize fusion power output and maintain stable burning plasmas, these future machines will rely heavily on the simultaneous use of gas puffing and pellet injection. However, integrating these disparate fueling mechanisms into a unified feedback control scheme remains a significant challenge due to inherent actuation time delays and uncertain fueling efficiencies. Existing superconducting devices must serve as proving grounds to develop and validate these critical control strategies. Consequently, KSTAR is being leveraged as a key experimental testbed to demonstrate reactor-grade density regulation. Because the expected actuation challenges require advanced techniques beyond traditional PID, Model Reference Adaptive Control (MRAC) has been employed as the feedback control technique in the experiments carried out on KSTAR. This approach utilizes the recently implemented NERF-mode algorithm to faithfully emulate reactor-constrained actuation. By actively coordinating both gas and pellet actuators, the MRAC controller manages actuator constraints while regulating plasma density around the precise targets required for stable operation. Furthermore, to ensure safe and sustained operation, the baseline controller was augmented with reference governor and safe reinforcement learning algorithms. This advanced integration enables robust plasma regulation alongside strict density-limit avoidance in KSTAR. The performance of this comprehensive control system has been thoroughly examined in simulations, demonstrating robust target tracking despite actuation delays and plasma variability, and confirming its effectiveness in the KSTAR experimental campaigns.

Supported by the U.S. DOE under Awards DE-SC0010537, DE-FC02-04ER54698 and the KSTAR Experimental Collaboration and Fusion Plasma Research (EN2601-17, EN2602-13) through the Korea Institute of Fusion Energy (KFE).