Some observations on the computational sensitivity of rotating cavity flows

Across the open literature, there is no clear consensus on what the most suitable modeling fidelity is for rotating cavity flows. Although it is a widely held opinion that unsteady Reynolds-averaged-Navier–Stokes (URANS) approaches are unsuitable, many authors have succeeded in getting reasonable heat transfer results with them. There is also a lack of research into the validity of hybrid URANS/large eddy simulation (LES) type approaches such as detached eddy simulation (DES). This paper addresses these research challenges with a systematic investigation of a rotating cavity with axial throughflow at Grashof numbers of 3.03×109 and 3.03×1011⁠. The disk near-wall layers remained laminar at both conditions, meaning that a turbulence model should not be active in these regions. The disk heat transfer was observed to affect the near-disk aerodynamics, which in turn affect the disk heat transfer: this feedback loop implies that conjugate heat transfer computations of rotating cavities may be worth investigating. On the shroud, additional eddy viscosity in URANS and DES was found to interfere with the formation of heat transfer enhancing streaks, whilst these were always captured by LES. DES exhibited a concerning behavior at the higher Grashof number. Locally generated eddy viscosity from the shroud was injected into the bulk fluid by the radial inflow. This contaminated the entire cavity with nonphysical modeled turbulence. As the radial inflow is a characteristic feature of rotating cavity flows, these results show that caution is necessary when applying hybrid URANS/LES approaches to this type of flow