Abstract. Disruption generate large thermal and mechanical stresses on the tokamak components. For a future reactor disruptions have a significant impact on the design since all loading conditions must be analyzed in accordance with stricter design criteria (due to safety or difficult maintenance). Therefore the uncertainties affecting the predicted stresses must be reduced as much as possible with a more comprehensive set of measurements and analyses in this generation of experimental machines, and avoidance/predictive methods must be developed further. The study of disruptions on ASDEX Upgrade is focused on these subjects, namely on: (1) understanding the physical mechanisms leading to this phenomenon and learning to avoid it or to predict its occurrence (with neural networks, for example) and to mitigate its effects; (2) analyzing the effects of disruptions on the machine to determine the functional dependence of the thermal and mechanical loads upon the discharge parameters. This allows to dimension or reinforce the machine components to withstand these loads and to extrapolate them to tokamaks still in the design phase; (3) learning to mitigate the consequence of disruptions.
IAEA 2003