Vertical Displacement Event Forecasting to Enable Disruption Avoidance on KSTAR with DECAF

Development of robust disruption avoidance techniques is a key goal of current tokamak devices in order to enable reliable and efficient operation of future tokamak-based power plants. Due to their relevance to disruptions, the ability to forecast vertical displacement events (VDEs) will be an important component of any of these systems. On KSTAR, analysis of data from previous shots using DECAF has yielded a real-time-capable method of detecting VDEs that achieves 100% accuracy on the full shot-years of data to which it has been applied with an average identification time before disruption of 15 ms [1]. A study of the accuracies of common vertical stability metrics, including the vertical instability growth rate and a vertical force balance, for the purpose of forecasting VDEs has resulted in the development of an approach that advances the VDE warning time and can facilitate avoidance of the ideally vertically uncontrollable regime altogether. Progress on employment of this model in the real-time plasma control system is reported. This study and the performance of a resulting 'time to disruption' estimator indicate this approach is an attractive strategy for implementation on future tokamak reactors as a means of minimizing the risk of VDE-induced disruptions.