The high confinement mode (H-mode) is the operational scenario foreseen for ITER, DEMO and future fusion power plants. At high densities, which are favorable in order to maximize the fusion power, a back transition from the H-mode to the low confinement mode (L-mode) is observed. In present tokamaks, this H-mode density limit (HDL) occurs at densities on the order of, but below, the Greenwald density.In gas ramp discharges at the fully tungsten covered ASDEX Upgrade tokamak (AUG), four distinct operational phases are identified in the approach towards the HDL. These phases are a stable H-mode, a degrading H-mode, the breakdown of the H-mode and an L-mode. They are reproducible, quasi-stable plasma regimes and provide a framework in which the HDL can be further analyzed. During the evolution, energy losses are increased and a fueling limit is encountered. The latter is correlated to a plateau of electron density in the scrape-off layer (SOL). The well-known extension of the good confinement at high density with high triangularity is reflected in this scheme by extending the first phase to higher densities.In this work, two mechanisms are proposed, which can explain the experimental observations. The fueling limit is most likely correlated to an outward shift of the ionization profile. The additional energy loss channel is presumably linked to a regime of increased radial filament transport in the SOL. The SOL and divertor plasmas play a key role for both mechanisms, in line with the previous hypothesis that the HDL is edge-determined.The four phases are also observed in carbon covered AUG, although the HDL density exhibits a different dependency on the heating power and plasma current. This can be attributed to a changed energy loss channel in the presented scheme.
Natural density formation as an H-mode operational limit