Characterization of the core poloidal flow at ASDEX Upgrade

In tokamaks, the toroidal rotation (utor) is essentially a free parameter that is usually dominated by the external momentum input from the neutral beam sources used to heat the plasma. The poloidal rotation (upol), on the other hand, is strongly damped due to the motion of the particles from the regions with higher magnetic field to lower magnetic field (and vice versa) and the associated transfer from poloidal to toroidal rotation (magnetic pumping effect). Despite many previous investigations, the nature of the core upol remains an open question: studies at the DIII-D tokamak show that at low collisionalities, upol is significantly more in the ion diamagnetic direction than predicted from neoclassical theory. At higher collisionalities, however, a good agreement between experiment and theory was found at the TCV and DIII-D tokamaks, which is qualitatively consistent with the edge results from both ALCATOR-C and ASDEX Upgrade (AUG).
In order to study the core impurity upol at AUG the core charge exchange recombination spectroscopy systems have been upgraded with new toroidal optical heads providing 70 additional lines-of-sight from the low-field side (LFS) to the high-field side (HFS) pedestal top. From theory it is known that any plasma flow can be expressed through a component parallel to the magnetic field and a rigid body rotation such that the evaluation of the measurement of the core toroidal flow at two points of the same flux surface (LFS/HFS) enables an indirect determination of the core upol with an accuracy better than 1 km=s for certain plasma parameters and regimes.
At AUG, the core upol at mid-radius is found to be more ion diamagnetic directed than the neoclassical prediction for both L- and H-mode discharges. At the plasma edge, upol is found to be electron diamagnetic directed and in good agreement with neoclassical codes, consistent with previous investigations at AUG. In order to characterize the core upol further at AUG, different parameter dependencies were investigated in the framework of the database. In the H-mode data there is an increase in the difference between the experimentally measured and theoretically predicted upol values at lower collisionalities concomitant with larger normalized Larmor radii. A possible explanation for this result is a balance between the neoclassical damping and the turbulent drive of upol. Within this picture, the dominant driving mechanism for upol at the lowest collisionalities of the database is turbulence. At higher collisionalities, the turbulent drive is reduced and the experimental profiles agree with the neoclassical predictions within their uncertainties.