Research in disruption prediction in tokamaks focuses on developing ways to predict harmful instabilities so they can be avoided.
This research begins with analysis of the chains of events that lead to disruptions tokamaks, and forecasting the onset of such events. This effort includes using diagnostic data and computer codes to reproduce the equilibrium state of the plasma, and to calculate its stability. Physics models or machine learning algorithms of the various processes are then implemented in real-time for disruption prediction.
Realizing this goal will require implementation and analysis of real-time diagnostic capabilities on various international tokamaks for disruption prediction and avoidance. This effort will bring significant new capabilities allowing real-time measurements of key plasma parameters such rotation profile, magnetic field pitch angle and internal magnetic perturbation profiles, electron temperature and electron temperature fluctuation profiles, evolution and decomposition of rotating magnetohydrodynamic modes, and energetic particle-driven mode onset and evolution. These measurements will allow the physics models to assess how close the plasma state is from being disrupted.
Finally, creation and implementation of control algorithms will steer the plasma away from possible disruption to preferred, sustained operational states.