Cross-device evaluation of vertical instability forecasting metrics using DECAF

Reliable avoidance of plasma vertical instability will be a crucial capability of any future tokamak power plant. In addition to scenario design and continuous vertical position feedback control, this will involve real-time control of proximity to vertical stability boundaries to avoid entering regions of operation space in which the vertical position cannot be maintained. While various metrics have been proposed for this purpose, a comparison of their predictive capability has not been performed. This work presents the results of such a study, performed in a cross-device manner using the Disruption Event Characterization And Forecasting (DECAF) code. Included in this study is a new metric based on the vertical force gradient evaluated across the poloidal cross section of the plasma, which is found to have the potential to enable the avoidance of 82% of vertical displacement events (VDEs), evaluated on a set of 400 shots from the most recent KSTAR campaign. In the interest of exploring reactor-relevance, reference is made to the relative real-time compatibility of the metrics, along with the diagnostic requirements of each. In addition to evaluation on historical data from KSTAR, MAST-U, and NSTX, results are shown for simulated data of ITER disruptions. As a demonstration of the power of such real-time-compatible vertical stability criteria, findings are presented that exemplify the utility of these metrics to better understanding the physical mechanisms underlying the causal relationship between 3D MHD phenomena, like edge-localized modes and internal reconnection events, and VDEs.

Supported by US DOE Grants DE-SC0020415, DE-SC0021311, and DE-SC0018623