On the direct use of core nonlinear gyrokinetic profile predictions for the planning of burning plasma experiments

Surrogate-based optimization techniques in PORTALS [P. Rodriguez-Fernandez et al 2022 Nucl. Fusion 62 076036] greatly reduce the cost of determining plasma profiles consistent with turbulent transport as predicted by high-fidelity, nonlinear gyrokinetics. Leveraging the capability of PORTALS to scan input parameters at a much-reduced cost thanks to surrogate pre-training, the operational space of SPARC L-modes has been scoped with ion-scale, nonlinear CGYRO and NEO simulations. Profile predictions and the associated performance were evaluated for conditions that span anticipated density ranges of operation (1.0-2.5x10²⁰m^-3) and input power (5-15 MW), with a range of boundary conditions (T₉₅=0.9-2.3 keV). This first-of-a-kind database was possible only due to the use of GPU-accelerated CGYRO, coupled with PORTALS techniques. These predictions are used to help plan the first SPARC campaigns, with the goal of maximizing the probability of reaching breakeven in pre-H-mode operations. The simulations agree closely with empirical scalings of density peaking, and it is found that the ion heat flux is always greater than the electron heat flux, which could be favorable for accessing high edge temperature gradients (such as in I-mode). The high sensitivity of fusion power to edge conditions reveals that the attainment of high edge temperatures (r/a=0.95) could be important to achieve performance milestones in early SPARC campaigns.