Understanding the occurrence of electron temperature profile collapses and their use as disruption predictors across tokamak databases using DECAF

Disruption-free operation in tokamak devices requires both the development of stable operational scenarios and effective disruption avoidance techniques. The DECAF code [1] advances this goal by developing a physics-based understanding of disruption triggers and event chains across a wide range of conditions and devices. The present work focuses on understanding the physics of electron temperature profile collapses preceding disruption events across multiple tokamak databases using DECAF. In particular, we have shown that electron temperature collapse (TEC) events can be used to forecast early disruption triggers in KSTAR, with warning times of up to ~1 second [2]. It has been found that the main conditions leading to TEC events include, among others, MHD precursors in high-beta plasmas that locally reconnect and seed neoclassical tearing modes (NTMs), field line stochasticity following mode locking, and high-Z impurity accumulation that asymmetrically cools down the plasma producing hollow temperature profiles. Systematic identification of various electron temperature collapse types—also including sawtooth relaxations and ELMs—has been performed to understand how they initiate disruption chains and under what conditions they occur. To support cross-device analysis, we have developed a hybrid ML/physics-based generalized algorithm to detect and characterize electron temperature collapses by mapping a reconstructed ‘crash profile’ to a set of self-adapting basis functions that capture key features that link each collapse to its underlying physical cause. The findings presented in this work were tested during initial experiments in the most recent KSTAR campaign, where a performance-degrading TEC event was successfully avoided by reducing ECH power just before a TEC onset, leading to stable operation with ~40% higher stored energy than the reference shot.

References

[1] S.A. Sabbagh, et al., Phys. Plasmas 30, 032506 (2023); https://doi.org/10.1063/5.0133825

[2] G. Bustos-Ramirez et al, ‘Exploring connections between electron temperature profile evolution and disruptions in KSTAR through DECAF analysis’ iFPC 2024, Seoul, South Korea.