Abstract Details
Abstracts
Author: Rohan Ramasamy
Requested Type: Consider for Invited
Submitted: 2026-03-09 09:04:34
Co-authors: Felix Wilms, Julian Johannes Freigang, Mohammad Miah, Orso Meneghini, and Sergei Makarov
Contact Info:
Proxima Fusion
Floessergasse 2
Munich, 81369
Germany
Abstract Text:
Predictive modeling of plasma profiles in reactor-relevant stellarator scenarios remains challenging due to the stiffness of nonlinear turbulence and sensitivity to modeling assumptions, including for pellet-driven fueling dynamics. We investigate profile prediction for a high-field stellarator targeting Q>1 using pellet-fuelled ECRH scenarios. The modeling approach couples nonlinear gyrokinetic and neoclassical fluxes computed with GENE and SFINCS to a 1D transport solver. Building on the success of the PORTALS framework applied to tokamaks [1], a Gaussian process surrogate model is used for the first time to accelerate convergence toward self-consistent profiles in stellarators. The simulations identify a clear window of operation exceeding the Q~1 requirement. The predicted performance depends strongly on scenario assumptions, particularly edge boundary conditions and the level of electromagnetic turbulence. We discuss the uncertainties in the present scenario development, including the impact of MHD in the pellet deposition and boundary condition constraints enforced by the divertor and edge transport, which must be resolved to achieve predictive reactor scenario modeling.
Characterization: 5.0
Comments: