Pedestal profile predictions with quantified uncertainty

The confinement in H-mode plasmas is strongly influenced by the pedestal structure. However, predictions of pedestal profiles remain limited by gaps in our understanding of pedestal transport. While high-fidelity nonlinear simulations are computationally expensive and impractical for routine use, existing MHD-based reduced models rely on transport constraints that limit their applicability in many relevant plasma scenarios (e.g. ELM-free regimes). Moreover, they lack predictive capabilities regarding necessary heating power and/or the pedestal density.

This work addresses these limitations by developing fast, validated transport models for the pedestal. The models are based on a quasilinear mixing-length approach, utilizing linear gyrokinetic simulations performed with GENE. A Bayesian uncertainty quantification framework is employed to calibrate and validate the models against experimental pedestal profiles from the DIII-D tokamak, incorporating experimental uncertainties via Gaussian process regression. The forward propagation of uncertainties is implemented using the integrated modeling framework ASTRA. Initial validation results are presented for two transport models: one based on electron temperature gradient (ETG) modes [1] and another targeting electromagnetic modes.

[1] Hatch et al., Nucl. Fusion, 2024