We present an ultrafast neural network (NN) model, QLKNN, which predicts core tokamak transport heat and particle fluxes. QLKNN is a surrogate model based on a database of 300 million flux calculations of the quasilinear gyrokinetic transport model QuaLiKiz. The database covers a wide range of realistic tokamak core parameters. Physical features such as the existence of a critical gradient for the onset of turbulent transport were integrated into the neural network training methodology. We have coupled QLKNN to the tokamak modelling framework JINTRAC and rapid control-oriented tokamak transport solver RAPTOR. The coupled frameworks are demonstrated and validated through application to three JET shots covering a representative spread of H-mode operating space, predicting turbulent transport of energy and particles in the plasma core. JINTRAC-QLKNN and RAPTOR-QLKNN are able to accurately reproduce JINTRAC-QuaLiKiz T i,e and n e profiles, but 3 to 5 orders of magnitude faster. Simulations which take hours are reduced down to only a few tens of seconds. The discrepancy in the final source-driven predicted profiles between QLKNN and QuaLiKiz is on the order 1%-15%. Also the dynamic behaviour was well captured by QLKNN, with differences of only 4%-10% compared to JINTRAC-QuaLiKiz observed at mid-radius, for a study of density buildup following the L-H transition. Deployment of neural network surrogate models in multi-physics integrated tokamak modelling is a promising route towards enabling accurate and fast tokamak scenario optimization, Uncertainty Quantification, and control applications.
JET is used as a test bed for ITER, to investigate beryllium migration which connects the lifetime of first-wall components under erosion with tokamak safety, in relation to long-term fuel retention. The (i) limiter and the (ii) divertor configurations have been studied in JET-ILW (JET with a Be first wall and W divertor), and compared with those for the former JET-C (JET with carbon-based plasma-facing components (PFCs)). (i) For the limiter configuration, the Be gross erosion at the contact point was determined in situ by spectroscopy as between 4% (E in = 35 eV) and more than 100%, caused by Be self-sputtering (E in = 200 eV). Chemically assisted physical sputtering via BeD release has been identified to contribute to the effective Be sputtering yield, i.e. at E in = 75 eV, erosion was enhanced by about 1/3 with respect to the bare physical sputtering case. An effective gross yield of 10% is on average representative for limiter plasma conditions, whereas a factor of 2 difference between the gross erosion and net erosion, determined by post-mortem analysis, was found. The primary impurity source in the limiter configuration in JET-ILW is only 25% higher (in weight) than that for the JET-C case. The main fraction of eroded Be stays within the main chamber. (ii) For the divertor configuration, neutral Be and BeD from physically and chemically assisted physical sputtering by charge exchange neutrals and residual ion flux at the recessed wall enter the plasma, ionize and are transported by scrape-off layer flows towards the inner divertor where significant net deposition takes place. The amount of Be eroded at the first wall (21 g) and the Be amount deposited in the inner divertor (28 g) are in fair agreement, though the balancing is as yet incomplete due to the limited analysis of PFCs. The primary impurity source in the JET-ILW is a factor of 5.3 less in comparison with that for JET-C, resulting in lower divertor material deposition, by more than one order of magnitude. Within the divertor, Be performs far fewer re-erosion and transport steps than C due to an energetic threshold for Be sputtering, and inhibits as a result of this the transport to the divertor floor and the pump duct entrance. The target plates in the JET-ILW inner divertor represent at the strike line a permanent net erosion zone, in contrast to the net deposition zone in JET-C with thick carbon deposits on the CFC (carbon-fibre composite) plates. The Be migration identified is consistent with the observed low long-term fuel retention and dust production with the JET-ILW.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.