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.

The JET exploitation plan foresees D-T operations in 2020 (DTE2). With respect to the first D-T campaign in 1997 (DTE1), when JET was equipped with a carbon wall, the experiments will be conducted in presence of a berylliumtungsten ITER-like wall (ILW) and will benefit from an extended and improved set of diagnostics and higher additional heating power (32 MW NBI + 8 MW ICRH). Among the challenges presented by operations with the new wall, there are a general deterioration of the pedestal confinement, the risk of heavy impurity accumulation in the core, which, if not controlled, can cause the radiative collapse of the discharge, and the requirement to protect the divertor from excessive heat loads, which may damage it permanently. Therefore, an intense activity of scenario development has been undertaken at JET during the last three years to overcome these difficulties and prepare the plasmas needed to demonstrate stationary high fusion performance and

The JET 2019-2020 scientific and technological programme exploited the results of years of concerted scientific and engineering work, including the ITER-like wall (ILW: Be wall and W divertor) installed in 2010, improved diagnostic capabilities now fully available, a major Neutral Beam Injection (NBI) upgrade providing record power in 2019-2020, and tested the technical & procedural preparation for safe operation with tritium. Research along three complementary axes yielded a wealth of new results. Firstly, the JET plasma programme delivered scenarios suitable for high fusion power and alpha particle physics in the coming D-T campaign (DTE2), with record sustained neutron rates, as well as plasmas for clarifying the impact of isotope mass on plasma core, edge and plasma-wall interactions, and for ITER pre-fusion power operation. The efficacy of the newly installed Shattered Pellet Injector for mitigating disruption forces and runaway electrons was demonstrated. Secondly, research on the consequences of long-term exposure to JET-ILW plasma was completed, with emphasis on wall damage and fuel retention, and with analyses of wall materials and dust particles that will help validate assumptions and codes for design & operation of ITER and DEMO. Thirdly, the nuclear technology programme aiming to deliver maximum technological return from operations in D, T and D-T benefited from the highest D-D neutron yield in years, securing results for validating radiation transport and activation codes, and nuclear data for ITER.

Abstract. L to H transition studies at JET have revealed an n=0 m=1 magnetic oscillation starting immediately at the L to H transition (called M-mode for brevity). While the oscillation is present a weak ELM-less H-mode regime is obtained, with a clear increase of density and a weak T e pedestal, with medium confinement, between high (H-mode) and low (L-mode). In ICRH heated plasmas or low density NBI plasmas the mode and the pedestal pressure can remain steady for the duration of the heating phase, of order 10 s or more. The axisymmetric magnetic oscillation has period ~ 1-2 ms, and poloidal mode number m=1: it looks like a pedestal localised up/down oscillation, although it is clearly a natural oscillation of the plasma, not driven by the position control system. Electron Cyclotron Emission, interferometry, reflectometry and fast Li beam measurements locate the mode in the pedestal region. D α , fast infrared camera and Langmuir probe measurements show that the mode modulates heat and particle flux to the target. The mode frequency appears to scale with the poloidal Alfvén velocity, and not with sound speed (i.e., it is not a Geodesic Acoustic Mode). A heuristic model is proposed for the frequency scaling of the mode. We discuss the relationship between Mmode and other related observations near the L-H transition.