Advancements in Disruption Event Characterization and Forecasting (DECAF) research including first real-time multi-Event disruption avoidance demonstration*

Physics-based disruption event characterization and forecasting (DECAF**) research continues to broaden high accuracy prediction of plasma disruptions and their physical event underpinnings across entire device databases and provide forecasts with sufficiently early warning to cue successful disruption avoidance [1]. Such success in real-time (100% accuracy in dedicated experiments exhibiting MHD-induced disruptions) with sufficiently early warning (~ 1 second on KSTAR) has motivated recent experiments that produced the first real-time demonstration of disruption avoidance by enabling multiple DECAF “Events” to control actuators using “Event feedback”. These Events now examine various physical phenomena including plasma current anomalies, vertical instability, MHD mode-locking, and impurity radiative collapses. Supporting analysis used to create the VDE Event included testing the underlying model on data from thousands of plasmas in the KSTAR, MAST-U, and NSTX full device databases resulting in predictive accuracies of 100%, 100%, and 98.6% respectively. For earlier warning in real-time, a VDE forecaster Event (VDE-f) was created based on a vertical force balance model including the applied equilibrium field, 2-D plasma current, and device eddy currents [2] and was connected to plasma shape and current profile (ECCD) actuators that independently and in combination produced disruption avoidance. The target plasmas produced high transient normalized beta up to 3.9 (record levels for KSTAR with the new tungsten divertor). Disruption avoidance was demonstrated in both upper / lower single null configurations with separatrix strike points on the carbon / tungsten divertor, respectively. The LTM-f Event actuates an n = 1 rotating field prepared to avoid mode locking. New DECAF Events are studied that correlate with plasma disruptions including a generalized capability to diagnose electron temperature collapses (TEC) that provide early disruption prediction (~ 0.7s). *Supported by U.S. DOE grants DE-SC0020415, DE-SC0021311, and DE-SC0018623. **U.S. and international patents pending.

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

[2] M.T. Tobin, et al., Plasma Phys. Control. Fusion 66 (2024) 105020