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.

We study the effect of a diverted magnetic geometry on edge plasma turbulence, focusing on the three-dimensional structure and dynamics of filaments, also called blobs, in simulations of the WEST tokamak, featuring a primary and secondary X-point. For this purpose, in addition to classical analysis techniques, we apply here a novel fully 3D Blob Recognition And Tracking (BRAT) algorithm, allowing for the first time to resolve the three-dimensional structure and dynamics of the blobs in a turbulent 3D plasma featuring a realistic magnetic geometry. The results are tested against existing theoretical scalings of blob velocity [Myra et al, Physics of Plasmas 2006]. The complementary analysis of the 3D structure of the filaments shows how they disconnect from the divertor plate in the vicinity of the X-points, leading to a transition from a sheath-connected regime to the ideal-interchange one. Furthermore, the numerical results show non-negligible effects of the turbulent background plasma: approximately half of the detected filaments are involved in mutual interactions, eventually resulting in negative radial velocities, and a fraction of the filaments is generated by turbulence directly below the X-point.

Collisionality is one of the key parameters in determining turbulent transport in the plasma edge, regulating phenomena such as "shoulder formation", separation of scale lengths in the scrape-off layer, turbulence damping and zonal flow dynamics. Understanding its role is therefore of primary importance for future reactors like ITER. Obtaining reliable predictions and a better characterization of plasma flow properties when varying collisionality remains, however, a critical challenge for the simulations. This paper focuses on the impact of varying collisionality in a nonisothermal three-dimensional fluid model of the plasma edge. A high field side limited configuration encompassing open and closed magnetic field lines with parameters typical of a medium-sized tokamak is considered. The present model can consistently account for the variations of collisionality and its impact on both the parallel resistivity η and the ion and electron parallel thermal conductivities χ e,i . Details on mean flow and turbulence properties are given. Changing collisionality leads to significant changes in the flow properties both on the mean and fluctuating quantities. In particular, lowering collisionality decreases the size of coherent structures, the fluctuation levels of turbulence, and steepens the density and temperature equilibrium profiles around the separatrix leading to a global reduction of the turbulent transport. The scrape-off layer (SOL) width is observed to increase with collisionality, eventually resulting in the disappearance of the scale lengths separation between near and far SOL, consistently with previous experimental observations. At low collisionality, where the presence of narrow feature is well-established, a contribution of heat conduction increases up to compete with heat convection.