Predictions of Reactor Performance Utilizing Nonlinear Gyrokinetic Simulations: SPARC, ITER, and ARC-class reactors

For the first time, nonlinear gyrokinetic simulations have been used to predict the performance of next generation fusion devices, namely ITER, SPARC, and ARC class devices. Utilizing machine learning techniques, steady state plasma profiles and performance were predicted using high fidelity gyrokinetics with greatly reduced computational expense. Predictions of SPARC H-modes like the Primary Reference Discharge (PRD) and L/I-mode like plasmas were performed to attempt to assess profile shapes and overall reactor performance. These results indicate that burning plasma conditions should easily be accessible in ELM-y H-mode and that operation at moderately high density (ne(0) > 2.5e20) and input power (15MW) is likely required to reach Q>1 goals in lower confinement regimes. The first attempts at modeling ARC-like reactor conditions were also completed, demonstrating that optimization of the design point is likely needed, but that conditions with P_fus ~ 350MW with Q~15 are likely attainable. Good agreement was generally found with physics-based models such as TGLF SAT2. The ITER baseline scenario was also simulated, indicating that ITER should realize its mission goals and may have potential avenues for further optimization of its performance.