Advances in Multi-device Tokamak Disruption Event Characterization and Forecasting (DECAF) Research

Disruption event characterization and forecasting (DECAF**) research determines the relation of events leading to disruption providing event onset forecasts with high accuracy and sufficiently early warning to allow disruption avoidance [1]. Real-time application of DECAF was made on the KSTAR superconducting tokamak including initial connection to control actuators producing over 50 plasma shots with nearly equal disrupted / non-disrupted cases that were forecast with 100% accuracy. With highly successful objective prediction performance numbers established both in database analysis and real-time applications of DECAF, research advances to broaden the physics and technical events needed to produce such high accuracy for any plasma analyzed. A multi-device study conducted for plasma vertical instability produced real-time capable modelling with prediction accuracy of 98.6% - 100% over multiple devices. A DECAF VDE forecaster event has recently been produced based on a stabilization physics model to give 3 times earlier VDE warnings. High bandwidth T_e profile measurements are used to reconstruct real-time capable ‘crash profiles’ to computationally identify sawteeth, ELMs, and more global MHD as NTM triggers and as direct disruption precursors. Impurity radiative collapses are examined showing that trigger Events often lead to more complex Event chains before disruption. Research also advances past feedforward control by connecting DECAF Events in feedback control for disruption avoidance.

**U.S. and international patents pending.