Dual Transport and Equilibrium Scenario Optimization for NSTX-U using COTSIM

B. Leard, T. Rafiq, E. Schuster

66th Division of Plasma Physics (DPP) Annual Meeting of the American Physical Society (APS)

Atlanta, GA, USA, October 7-11, 2024

Abstract

Determining the correct combination of actuator trajectories to reach a desired plasma state is pivotal to successfully operating a tokamak. One method to shape these trajectories is algorithmic feedforward optimization, which couples a model of the plasma dynamics such as the Control Oriented Transport SIMulator (COTSIM) to an optimizer. To accelerate the scenario optimization, the modeling capabilities of COTSIM have been enhanced with neural-network surrogate models for NUBEAM and MMM, which model the NBI sources and heat/momentum diffusivities, respectively. Additionally, fixed-boundary equilibrium solvers have been integrated into COTSIM to enable self-consistent equilibrium + transport simulations. These COTSIM capabilities have enabled a scheme that optimizes the evolutions of the plasma shape, current, and density along with NBI powers to reach and sustain a target plasma scenario. Additionally, several nonlinear constraints have been incorporated into the optimization scheme to avoid MHD instabilities. Finally, the evolutions of the optimized plasma shape and internal properties can be fed into the free-boundary equilibrium solver in COTSIM to determine the needed poloidal-field coil currents. This optimization scheme has been used to design scenarios for NSTX-U.

*Supported by the US DOE under DE-SC0021385