Characterization of reduced-order turbulence models in the L-mode pedestal-forming region in JET

TL;DR

This study uses gyrokinetic simulations to analyze turbulence modes in JET tokamak discharges near the L-H transition, revealing mode behaviors, sensitivities, and model accuracies in the pedestal region.

Contribution

It provides a detailed linear instability characterization of JET discharges with assessments of model fidelity and the impact of various assumptions on turbulence predictions.

Findings

Trapped-electron modes dominate at low densities and inner radii.

Ion-temperature-gradient modes become dominant at higher densities.

TGLF-SAT2 model agrees well with GENE simulations up to certain radii.

Abstract

Linear instability characterization of seven JET discharges just prior to the L-H transition is performed at $\rho_{\text{tor}} \in [0.85,0.9,0.95]$ with the gyrokinetic GENE code. The discharges cover both the low- and high-density branches of the L-H transition at two different triangularities. Sensitivities to driving gradients, normalized electron collisonality $\nu_e^*$, hydrogen isotope mass, magnetic geometry and finite-$\beta$ effects are all characterized. At $\rho_{\text{tor}}=0.85$ and $0.9$, trapped-electron modes (TEMs) propagating in both the electron- or ion-drift direction are observed at the lowest densities. At higher density ion-temperature-gradient (ITG) modes are dominant, some of which exhibit trapped-ion drive and unconventional ballooning structures. At $\rho_{\text{tor}}=0.95$, the low-density cases are similar to inner radii, while at higher densities subdominant modes are destabilized by higher collisionalities. The electron collisonality $\nu_e^*$ is scanned around the experimental values at all three radii and for the seven discharges studied. The experimental collisionality range corresponds to a region of minimum linear drive between ITG-TEM mode branches at lower collisionalities and resistive mode branches at higher collisionalities. Moreover, the quasilinear particle flux is directed inward only in the collisionality domain where the linear drive is minimized at $\rho_{\text{tor}}=0.85$ for all densities and $0.9$ only for the highest densities. Model fidelity reduction is performed on the GENE simulations to evaluate the impact of various assumptions and simplifications made by the state-of-the-art quasilinear models QuaLiKiz and TGLF. QuaLiKiz is found to be inadequate beyond $\rho_{\text{tor}}=0.85$, while TGLF-SAT2 agrees well with linear spectra and the quasilinear heat fluxes from GENE up to and including $\rho_{\text{tor}}=0.9$.