Disruption Research for SPARC

SPARC Disruption Research will deliver first-principle and machine learning-based solutions for disruption prevention strategies ensuring that SPARC will accomplish its mission goals. We present critical research advancements addressing disruption prediction, avoidance, and mitigation strategies on SPARC. First, we will discuss progress made in developing a potential candidate algorithm for the Disruption Mitigation System (DMS) trigger based on deep learning. The algorithm leverages a recently developed complex architecture, showing state-of-the-art performances on several existing tokamaks, with domain adaptation capabilities across devices. Several Massive Gas Injection valves will be installed as SPARC DMS, and a study of their effectiveness for thermal mitigation is required. Results on fueling efficiency investigations, prediction of required quantities, as well as their layout and performance optimization via 3D MHD modeling will be presented. Finally, initial results on the development of the Runaway Electron Mitigation Coil (REMC) research plan will be given: scenario dependent analyses will be presented to evaluate optimal threshold voltages for the REMC switch and added coil resistances to optimize current and effectively suppress runaway electron formation in SPARC.