Understanding the core confinement in DIII-D super-H experiments by transport modeling

The DIII-D super-H (SH) scenario, which is characterized by a significantly higher pedestal pressure compared to standard high confinement mode (H mode) plasmas, typically exhibits two phases in its temporal dynamics. The early hot ion (SH-HI) phase has higher core ion temperatures and normalized confinement factor (H (98(y,2)) similar to 2) than the later 'standard' SH phase, which has similar pedestal pressure characteristics to the SH-HI phase but a lower confinement factor (H (98(y,2)) similar to 1.2) as well as lower pedestal T (i)/T (e) ratio. However, beyond the pedestal differences, it is also observed that in the core plasma T (i) is more peaked and has a significantly larger normalized gradient scale length a/L (Ti) in the SH-HI phase than in the SH phase. This paper identifies the physics responsible for the different core profiles via gyrokinetic and gyrofluid modeling. It is found that the ion temperature gradient (ITG) mode dominates the core transport for both phases. Absent flow shear effects, the ITG critical gradient (a/L (Ti,crit)) is shown to be far smaller in the SH-HI phase than the SH phase. The lower a/L (Ti,crit) in the SH-HI phase is shown to be mainly induced by the hollow carbon (impurity) density profile, which is strongly destabilizing relative to the nearly flat carbon density profile in the SH phase. Differences in the T (i)/T (e) ratio between these phases are found to have a minor impact. However, the significantly stronger flow shearing in the SH-HI phase relative to the SH phase enables the achievement of higher core a/L (Ti) values and is therefore mainly responsible for the higher core T (i) values observed in the early SH-HI phase. Predictive transport modeling shows that the confinement in the lower-rotation SH phase could be elevated significantly if a peaked impurity density profile can be achieved, and potential applications to the performance improvement of future reactors are discussed.