Validation of MARS-F Modeling of 3D RMP Plasma Response and Testing of Reduced Surrogate Models

MARS-F, a simulation code which models 3D magneto-hydrodynamics in tokamak plasmas, is applied to resonant magnetic perturbation (RMP) discharges in DIII-D. Understanding these 3D RMP plasma responses in simulation is important for controlling the edge localized mode (ELM). Experimental validation datasets for plasma response are available in DIII-D, such as Thomson scattering (TS) and charge exchange recombination (CER) which both measure the displacement of the plasma boundary and the internal displacement at certain toroidal locations. This work focuses on validating the MARS-F computed plasma boundary displacement based on the linear resistive fluid model, complementary to most of the previous studies in DIII-D where magnetic probe measurements were used for comparison. Furthermore, the ability to validate against vertical and horizontal plasma boundary displacements allows systematic investigation of "mixing" of these two geometric components – an important aspect in terms of the geometric sensitivity of both measured and modeled results. A range of experiments, scanning both the plasma and RMP coil parameters, are simulated. The results (i) confirm the predictive capability of the MARS-F model for 3D response in tokamak plasmas, and (ii) help expand the available perturbed 3D equilibrium database constructed using MARS-F. This database has also been analyzed utilizing various model order reduction techniques, including singular value decomposition (SVD) combined with neutral networks (NNs). In particular, it is found that the first 5 SVD eigenstates are capable of representing ~95% of the database with a relative error of <10%.