First-principles performance prediction of burning plasmas with self-consistent kinetic profiles

Self-consistent, high-fidelity predictions of kinetic profiles are required to accurately scope future experiments and pilot plants. In this presentation, we present a framework that allows for first-principles simulations of core profiles with an accelerated convergence rate [Rodriguez-Fernandez, NF 2022]. The high computational cost of self-consistent, first-principles, multi-channel predictions (which require flux-driven, gyrokinetic frameworks) has long precluded their use to study experiments and to aid the design of new devices. Thanks to the surrogate-based optimization framework presented here, nonlinear simulations of the core of the SPARC tokamak using the CGYRO code have been possible. The predictions of kinetic profiles and performance with full nonlinear turbulence simulations are compared to empirical and TGLF modeling. While all predictive frameworks show the Primary Reference Discharge in SPARC to be above the burning plasma regime (Q>5), clear differences are seen in predicted profiles, in particular in the density peaking. Improvements to the workflow to further reduce the computational cost will also be presented, as well as the potential of the surrogate-modeling techniques to predict variations of the nominal scenario at minimal computational expense.