Disruption Prediction and Avoidance in High Beta Long Pulse KSTAR Plasmas ¿ Real Time Expansion

Disruption prediction and avoidance in tokamaks to enable long-pulse plasma operation are the highest Tier 1 DOE Fusion Energy Sciences (FES) strategic priority elements. These critical issues are the present-day “grand challenge” topics for high performance tokamak applied plasma science. The research in this award directly addresses these highest priority DOE FES elements. This award is a coordinated experimental and physics analysis research program that leverages the unique capabilities of an advanced, international tokamak facility. It focusses on the critical issue of maintaining long pulse, high performance tokamak plasma operation with very low disruptivity on the Korea Superconducting Tokamak Advanced Research (KSTAR) facility located at the National Fusion Research Institute (NFRI) in South Korea. The research program continues the development of our disruption event characterization and forecasting (DECAF) code capability based on multiple tokamak database analysis, with quantitative performance evaluation of disruption prediction including validated physics modeling, focusing on KSTAR long-pulse, high beta plasmas. It further includes the expansion of this capability to real-time. This will be the first-ever implementation of DECAF disruption forecasting models in real-time, made possible by significant new diagnostic and hardware upgrades, and will produce initial studies of real-time disruption avoidance. The physics analysis in the proposed work will fully exercise our present analysis capability with a focus on advanced high beta, long pulse, high / fully non-inductive plasma experiments with self-consistent current profiles. These conditions are expected to present unique challenges for disruption avoidance. Physics-based DECAF analysis capability will expand in scope to include physics-based density limits, transport considerations such as loss of power balance, and other key disruption-inducing events. Machine learning (ML) techniques, already initiated in present DECAF work, will be expanded as a powerful numerical tool to support the physics analysis in innovative ways beyond the usual application of ML for disruption prediction. The proposed effort will substantially expand diagnostic and sensor capabilities including: new real-time motional Stark effect capability with internal measurement of magnetic field fluctuations; new real-time plasma rotation measurement; new real-time electron temperature profile and temperature fluctuation measurement; and energetic particle mode measurement aimed to support disruption prediction and plasma profile control for disruption avoidance. Plasma control elements in the proposed work will include plasma shape, stored energy (beta), toroidal rotation, and global magnetohydrodynamic mode control. Quantitative performance metrics including the observed plasma disruptivity, predictions by full DECAF analysis, and predictions by the real-time DECAF implementation will be compared and iteratively improved.