Parameters determining the radial propagation of type-I edge localized modes (ELMs) in ASDEX Upgrade

The radial propagation of type-I edge localized modes (ELMs) in ASDEX Upgrade has been measured using a number of techniques. The most reliable technique uses a filament probe to measure the time difference between two separated probes and yields a mean radial velocity obtained of ~1.5?km?s?1. The radial velocity calculated using a time of flight technique suffers from uncertainties in the start time and yields radial velocities which are 3?4 times lower than those based on the filament probe. The filament probe data show that the filaments leave the last closed flux surface (LCFS) over a period of up to 400??s. The velocity derived from the floating potential at two poloidally separated probes suffers from the neglect of temperature differences, local turbulence effects and the smoothing used. The mean radial velocity calculated is ~1100?m?s?1 for a smoothing of 0.5??s (the data acquisition time) decreasing to ~850?m?s?1 for a 5??s smoothing. In spite of the differences in the size of the radial velocity all the methods suggest that Vr is independent on distance from the LCFS for 20 < ?RLCFS < 100?mm and on ELM size. The ion saturation current e-folding length scales with q95 and inversely with the square root of the temperature pedestal. Once these dependences are accounted for the evolution of the e-folding length with ?WELM/Wped is consistent with Vr being independent of ELM size.