Towards Predictive Gyrokinetic Simulations of Edge and SOL Turbulence in Reactor-Relevant Conditions

Predicting turbulent transport across the edge and scrape-off layer (SOL) is critical for the development of magnetic confinement fusion devices, including tokamaks and stellarators. However, first-principles full-f gyrokinetic (GK) simulations remain challenging due to complex magnetic geometries and their prohibitive computational cost. The high-performance full-f GK code GENE-X was specifically designed to overcome these challenges. Recently, a major computational acceleration was achieved through the implementation of a velocity-space spectral method, reducing the computational footprint by more than an order of magnitude. This breakthrough enables advanced first-principles studies of larger devices and high-confinement regimes, such as H-mode. Leveraging this new capability, we first present spectral simulations of L-mode plasmas in TCV. The new approach is validated through a detailed comparison with previously benchmarked grid-based simulations, showing excellent agreement in outboard midplane profiles, turbulence characteristics, and power balance. In this case, a computational speedup of nearly 50× is achieved. We then simulate the pre-L–H transition phase in AUG, focusing on a hydrogen discharge. As the L–H transition is approached, a region of strongly sheared radial electric field (Er) develops near the separatrix. Force-balance analysis reveals that the Er well is dominated by turbulence-driven poloidal flows. The turbulence is characterized by structures propagating in the electron diamagnetic direction, associated with electron drift waves and trapped-electron-mode turbulence. Notably, the predicted ion heat flux across the edge—a key quantity for L–H transition predictions—shows good quantitative agreement with experimental measurements. The role of turbulence-driven poloidal flows in shaping the Er well and their implications for H-mode access are further illustrated through comparisons of favorable and unfavorable configurations in AUG and TCV, validated against experiments. Overall, these results demonstrate that the spectral approach implemented in GENE-X provides a powerful and reliable first-principles predictive tool for turbulent transport studies in reactor-relevant scenarios, including ITER.