Integrated Predictive Transport Modeling with Neural Networks and Self-Consistent L-H/H-L Transitions in COTSIM

K. Shabbir, B. Leard, T. Rafiq, E. Schuster

67th Division of Plasma Physics (DPP) Annual Meeting of the American Physical Society (APS)

Long Beach, CA, USA, November 17-21, 2025

COTSIM is a control-oriented predictive simulation code augmented with neural-network surrogate models for plasma sources and transport. A neural network version of the updated Multi-Mode Model (MMM 9.0.10), referred to as MMMnet, predicts turbulent transport with high accuracy and significantly reduced computational cost compared to the conventional MMM. The coupled transport and current-drive equations are evolved self-consistently, incorporating Ohmic heating, ECRH (via TORAYnet), and neutral beam injection (via NUBEAMnet). The magnetic equilibrium is computed in this study using a fixed-boundary solver. Edge pedestal height and width are predicted using the PEDESTAL module, which includes an empirical model for the L-H transition. H-L transitions are triggered when the separatrix power drops below the L-H threshold. The integration of neural-network surrogates and tightly coupled solvers enables fast and accurate predictions of electron and ion temperature and safety factor profiles in DIII-D discharges.

*Supported by the U.S. DOE under Awards DE-SC0010661 and DE-FC02-04ER54698.