Improved scaling of the scrape-off layer particle flux width by the Bayes theorem on EAST

TL;DR

This paper introduces a Bayesian MAP estimation method to more accurately determine the scrape-off layer particle flux width on EAST, leading to improved scalings and insights into plasma behavior relevant for fusion devices like ITER.

Contribution

The study develops a Bayesian MAP approach for estimating SOL flux width, improving accuracy and updating scalings for different plasma conditions on EAST.

Findings

MAP method improves fit accuracy by ~30% over traditional methods.

Revised scalings show dependence on connection length and differences between helium and deuterium.

Extrapolated flux widths are ~6 mm for ITER L-mode and ~13 mm for H-mode.

Abstract

The scaling of scrape-off layer (SOL) power width ({\lambda}q) is essential for advancing the understanding of particle and heat transport in the SOL. Due to the sparse layout of divertor Langmuir probes (Div-LPs) and probe erosion during long-pulse, high-performance operations on EAST, estimating SOL particle flux width ({\lambda}js, used to approximate {\lambda}q) from the ion saturation current density profile (js) often incurs substantial uncertainty. This study presents a maximum a posteriori (MAP) estimation method based on Bayes' theorem, achieving approximately 30% improvement in fitting accuracy over traditional ordinary least squares. Using this method and the FreeGS equilibrium code, we updated databases from Liu et al., Nucl. Fusion 64 (2024). Revised {\lambda}js scalings for L-mode and H-mode in deuterium and helium plasmas demonstrate better regression quality and slightly altered regression results. Unified L-mode and H-mode scalings in deuterium and helium are: {\lambda}_js^L = 0.11 L_c^1.06 n_e^0.35 Z^0.32 P_SOL^0.25 p^(-0.26) and {\lambda}_js^H = 0.11 L_c^1.28 n_e^0.56 Z^0.36 P_SOL^0.30, where L_c is the average SOL connection length, n_e the line-averaged electron density, Z the charge number, PSOL the power crossing the last closed flux surface, and p the core-averaged plasma pressure. Key findings include: (i) {\lambda}js strongly depends on SOL connection length, indicating a machine size dependence absent in the Eich scaling, and (ii) helium {\lambda}js is slightly larger than deuterium {\lambda}js. Extrapolated scalings suggest {\lambda}q ~ 6 mm for ITER L-mode (Ip = 12 MA) and ~13 mm for H-mode (Ip = 15 MA).