Studies of Edge Fluctuations of Negative Triangularity Plasma on TCV using a New Gas-Puff Imaging Diagnostic

Abstract

Successful operation of magnetic confinement fusion devices requires thorough analysis of edge plasma turbulence, which plays a key role in dealing with the exhaust heat and particle loads on the wall materials. Edge plasma turbulence differs depending on various plasma conditions including plasma shape, e.g., the triangularity (δ). Negative δ is of great interest, as it is known to exhibit a substantial increase in con-finement compared to the usual positive δ, D-shaped plasmas. Recent experiments in the Tokamak à Configuration Variable (TCV) and DIII-D tokamaks have correlated the confinement improvement to a reduction of fluctuations within the plasma core, but, until now, relatively little was known about the effect of δ on plasma edge dynamics.

This thesis explores the edge turbulence in negative δ plasmas on TCV using Gas-Puff Imaging (GPI). GPI measures the spatially-resolved edge fluctuations by imaging atomic spectral-line emission from a local neutral gas puff. In collaboration with the TCV team at EPFL, a new GPI system was installed at TCV and has been operational since 2018. The most prominent features appearing in the GPI images are blobs, which are intermittent turbulence with large amplitude having filamentary structures (elongated along the field) in the scrape-off layer (SOL). Estimation of the size, speed, and frequency of blobs allows evaluation of particle fluxes leaving the magnetically confined plasma boundary. Traditional approaches to blob analysis in the GPI images, including conditional averaging sampling and cross-correlation techniques, have the limitations of only providing averaged characteristics of blobs. These limitations were tackled in this work by implementing a machine learning method with standardized models. The models were trained with synthetic GPI images and demonstrated excellent performance in predicting blob contours for real GPI data.

Using the GPI diagnostic, together with probe measurements, first-wall interaction is found to be completely suppressed in sufficiently negative δ, for both limited and diverted L-mode plasmas in TCV. This phenomenon can be explained by blobs being ejected along the reduced connection length, which is intrinsic to negative δ plasmas. In addition, edge fluctuations for negative δ plasmas were investigated in high density plasmas, near the density limit. The suppression of first-wall interaction for sufficiently negative δ is maintained at high densities, and the density limit appears to be similar for positive and negative δ. Furthermore, the blob-tracking method was applied for a detailed blob-by-blob analysis to compare cross-field particle transport in positive and negative δ plasmas. This revealed that plasmas with smaller δ tend to have less frequent blobs, most of which have large area and low radial speed, leading to a lower cross-field particle transport. All in all, this work provides experimental evidence of reduced edge turbulence in negative δ via analysis of blobs in the GPI measurements on TCV, strengthening the prospects of negative triangularity plasmas as a potential reactor solution.