An on-line predictor of the time to disruption has been installed on the ASDEX Upgrade tokamak. It is suitable either for avoidance of disruptions or for mitigation of those that are unavoidable. The prediction uses a neural network trained on eight plasma parameters and their time derivatives extracted from 99 disruptive discharges. The network was tested off-line over 500 discharges and was found to work reliably and to be able to predict the majority of the disruptions. The trained network was installed on-line, tested over 128 discharges and used to inject killer pellets to mitigate the disruption loads.

First results are reported on the prediction of disruptions in one tokamak, based on neural networks trained on another tokamak. The studies use data from the JET and ASDEX Upgrade devices, with a neural network trained on just 7 normalised plasma parameters. In this way, a simple single layer perceptron network trained solely on JET correctly anticipated 67% of disruptions on ASDEX Upgrade in advance of 0.01 seconds before the disruption. The converse test led to a 69% success rate in advance of 0.04 seconds before the disruption in JET. Only one overall time scaling parameter is allowed between the devices, which can be introduced from theoretical arguments. Disruption prediction performance based on such networks trained and tested on the same device shows even higher success rates (JET: 86%; ASDEX Upgrade 90%), despite the small number of inputs used and simplicity of the network. It is found that while performance for networks trained and tested on the same device can be improved with more complex networks and many adjustable weights, for cross machine testing the best approach is a simple single layer perceptron. This offers the basis of a potentially useful technique for large future devices such as ITER, which with further development, might help to reduce disruption frequency and minimise the need for a large disruption campaign to train disruption avoidance systems.

Recent experiments at ASDEX Upgrade have achieved advanced scenarios with high β N (>3) and confinement enhancement over ITER98(y, 2) scaling, H H98y2 = 1.1-1.5, in steady state. These discharges have been obtained in a modified divertor configuration for ASDEX Upgrade, allowing operation at higher triangularity, and with a changed neutral beam injection (NBI) system, for a more tangential, off-axis beam deposition. The figure of merit, β N H ITER89-P , reaches up to 7.5 for several seconds in plasmas approaching stationary conditions. These advanced tokamak discharges have low magnetic shear in the centre, with q on-axis near 1, and edge safety factor, q 95 in the range 3.3-4.5. This q-profile is sustained by the bootstrap current, NBI-driven current and fishbone activity in the core. The off-axis heating leads to a strong peaking of the density profile and impurity accumulation in the core. This can be avoided by adding some central heating from ion cyclotron resonance heating or electron cyclotron resonance heating, since the temperature profiles are stiff in this advanced scenario (no internal transport barrier). Using a combination of NBI and gas fuelling line, average densities up to 80-90% of the Greenwald density are achieved, maintaining good confinement. The best integrated results in terms of confinement, stability and ability to operate at high density are obtained in highly shaped configurations, near double null, with δ = 0.43. At the highest densities, a strong reduction of the edge localized mode activity similar to type II activity is observed, providing a steady power load on the divertor, in the range of 6 MW m −2 , despite the high input power used (>10 MW).