Type-I edge-localized modes (ELMs) have been mitigated at the JET tokamak using a static external n=1 perturbation field generated by four error field correction coils located far from the plasma. During the application of the n=1 field the ELM frequency increased by a factor of 4 and the amplitude of the D(alpha) signal decreased. The energy loss per ELM normalized to the total stored energy, DeltaW/W, dropped to values below 2%. Transport analyses shows no or only a moderate (up to 20%) degradation of energy confinement time during the ELM mitigation phase.

Abstract. Disruption mitigation is mandatory for ITER in order to reduce forces, to mitigate heat loads during the thermal quench (TQ) and to avoid runaway electrons. A fast disruption mitigation valve (DMV) has been installed at JET to study mitigation by massive gas injection (MGI). Different gas species and amounts have been investigated with respect to timescales and mitigation efficiency. We discuss the mitigation of halo currents as well as sideways forces during vertical displacement events, the mitigation of heat loads by increased energy dissipation through radiation, the heat loads which could arise by asymmetric radiation and the suppression of runaway electrons.

Disruptions are a major operational concern for next generation tokamaks, including ITER. They may generate excessive heat loads on plasma facing components, large electromagnetic forces in the machine structures and several MA of multi-MeV runaway electrons. A more complete understanding of the runaway generation processes and methods to suppress them is necessary to ensure safe and reliable operation of future tokamaks. Runaway electrons were studied at JET-ILW showing that their generation dependencies (accelerating electric field, avalanche critical field, toroidal field, MHD fluctuations) are in agreement with current theories.

This paper describes progress achieved since 2007 in understanding disruptions in tokamaks, when the effect of plasma current sharing with the wall was introduced into theory. As a result, the toroidal asymmetry of the plasma current measurements during vertical disruption event (VDE) on the Joint European Torus was explained. A new kind of plasma equilibria and mode coupling was introduced into theory, which can explain the duration of the external kink 1/1 mode during VDE. The paper presents first results of numerical simulations using a free boundary plasma model, relevant to disruptions.

Abstract:Disruptions are a critical issue for ITER because of the high thermal and magnetic energies that are released on short time scales, which results in extreme forces and heat loads. The choice of material of the plasma facing components (PFCs) can have significant impact on the loads that arise during a disruption. With the ITER-like wall (ILW) in JET made of beryllium in the main chamber and tungsten in the divertor, the main finding is a low fraction of radiation. This has dropped significantly with the ILW from 50-100% of the total energy being dissipated in the plasma with CFC to less than 50% on average and down to just 10% for VDEs. All other changes in disruption properties and loads are consequences of this low radiation: long current quenches, high vessel forces caused by halo currents and toroidal current asymmetries as well as severe heat loads. Temperatures close to the melting limit have been locally observed on upper first wall structures during deliberate VDE and even at plasma currents as low as 1.5 MA and thermal energy of about 1.5 MJ only. A high radiation fraction can be regained by massive injection of a mixture of 10%Ar with 90%D 2 . This accelerates the current quench and by this reducing halo and sideways impact. The temperature of PFCs stays below 400 ∘ C. MGI is now a mandatory tool to mitigate disruptions in closed-loop operation for currents at and above 2.5 MA in JET.

A key feature of disruptions during vertical displacement events, discovered in JET in 1996, is the toroidal variation in the measured plasma current Ip, i.e. the plasma current asymmetries, lasting for almost the entire current quench. The unique magnetic diagnostics at JET (full set of poloidal coils and saddle loops recorded either from two toroidally opposite or from four toroidally orthogonal locations) allow for a comprehensive analysis of asymmetrical disruptions with a large scale database. This paper presents an analysis of 4854 disruptions over an 18 year period that includes both the JET carbon (C) wall and the ITER-like (IL) wall (a mixed beryllium/tungsten first wall). In spite of the Ip quench time significantly increasing for the IL-wall compared to C-wall disruptions, the observed toroidal asymmetry time integral (∼ sideways force impulse), did not increase for IL-wall disruptions. The Ip asymmetry has a dominantly n = 1 structure. Its motion in the toroidal direction has a sporadic behaviour, in general. The distributions of the number of rotation periods are found to be very similar for both C- and IL-wall disruptions, and multi-turn rotation was sometimes observed. The Ip asymmetry amplitude has no degradation with rotation frequency for either the C- or IL-wall disruption. Therefore dynamic amplification remains a potentially serious issue for ITER due to possible mechanical resonance of the machine components with the rotating asymmetry.

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.

Recent experiments on JET have shown that type-I edge localized modes (ELMs) can be controlled by applying static low n = 1 external magnetic perturbation fields (EMPFs) produced by four external error field correction coils (EFCC) mounted far away from the plasma between the transformer limbs. When an n = 1 EMPF with an amplitude of a few Gauss at the plasma edge (ρ > 0.95) is applied during the stationary phase of a type-I ELMy H-mode plasma, the ELM frequency rises from ~30 Hz up to ~120 Hz and follows the applied perturbation field strength. The energy loss per ELM normalised to the total stored energy, ΔW ELM /W p , decreased from 7 % to below the resolution limit of the diamagnetic measurement (~ 2%). Transport analysis using the TRANSP code shows no or a modest reduction of the thermal energy confinement time because of the density pump-out, but when normalised to the IPB98(y,2) scaling the confinement shows almost no reduction. Stability analysis of mitigated ELMs shows that the operational point moves from intermediate n peeling-ballooning (wide mode) boundary to low-n peeling (narrow mode) boundary with n = 1 perturbation fields. The first results of ELM mitigation with the n = 2 EMPFs on JET demonstrate that the frequency of ELM can be increased by a factor of 3.5, only limited by the available EFCC coil current. During the application of the n = 1, 2 EMPFs, a reduction in the ELM size (ΔW ELM) and ELM peak heat fluxes on the divertor target by roughly the same factor as the increase of the ELM frequency has been observed. The reduction in heat flux is mainly due to the drop of particle flux rather than the change of the electron temperature. Similar plasma braking effect has been observed with n = 1 and n = 2 external fields when a same EFCC coil current was applied. Compensation of the density pump-out effect has been achieved by means of gas fuelling in low triangularity plasmas. An optimised fuelling rate to compensate the density pump-out effect has been identified. Active ELM control by externally applied fields offers an attractive method for next-generation tokamaks, e.g. ITER.