Empirical estimates indicate that the transient divertor power load induced by type-I ELMs in the standard H-mode ITER scenario might result in intolerably short target life times. A more quantitative extrapolation, however, requires a much better physical understanding. During past years, a variety of theoretical approaches have been made to describe the underlying physical processes, and spontaneous ELMs have been investigated in many divertor tokamaks with high temporal and spatial resolution in order to improve the experimental data base. In addition, signi cant experimental e ort has been put on the exploration of recipes to avoid or at least mitigate the ELM size by external intervention. Experimental ELM mitigation studies (and related modelling) are of interest from di erent points of views. First, the active intervention gives access to new, externally controlled ELMy discharge scenarios and ELM diagnostic techniques. Second, the ELM response on the type and amplitude of the applied perturbation rovides additional hints on the ELM stability. And third, mitigated ELMs, possibly different from spontaneous ones, and their scaling are of interest 'per se', in case that mitigation is to be nally applied in ITER. In this contribution we report on ELM mitigation results in ASDEX Upgrade, where we further restrict the scope to those methods involving 3D, local non-axisymmetric perturbations. It has been demonstrated long ago that local 3D perturbations from cryogenic pellet injection can 'promptly' trigger ELMs, i.e. the onset of the ELM starts within about 100 µs after the onset of the perturbation. In addition to deuterium (D) pellet injection, techniques and methods tried on ASDEX Upgrade were localised supersonic pulsed injection of D gas and laser blow o from targets carrying C or Al micro pellets. The corresponding perturbations differ strongly both in the space and time domain and provide an instrument to probe the local stability. Varying the parameters of the external perturbations by e.g. changing the amount of injected material or the injection speed we mapped out the ELM trigger threshold with high spatial and temporal resolutions. For perturbative triggered ELM this threshold is lowest close to the pedestal top and rises strongly towards the separatrix, where even massive perturbations fail to release a prompt ELM. Only small changes of the trigger threshold were found in the time domain. For a short interval (~ 5ms) after the ELM event its value is raised modestly. But even there it is possible to trigger yet another ELM reducing dramatically the e ective ELM repetition time. The ELM intensity shrinks with reduced delay from the previous ELM. No signiffcant diference was found so far between triggered ELM and their intrinsic counterparts at similar repetition rates. All ELM events seem to originate from the plasma low field side, irrespective of the poloidal location of the imposed perturbation. Poloidal displacements of the applied perturbation - realised by D pellet injection from the high field side - may allow to conclude on different possible trigger transport mechanisms. Disentangling radial pellet flight time and perturbation transfer times to the low eld in a pellet velocity scan showed that the ELM starts at the low field side only a few µs after imposing the perturbation - primarily a small high pressure plasmoid surrounding an ablating pellet - at the high eld side. Since this is short compared to a sound transit time, other candidates, e.g. shear Alfvén waves or electron transport along eld lines, have to be considered. |
Trabalhos Científicos 
Comunicações a Reuniões Científicas