Database of the 14 measured roughness geometries used for training the roughness function model
Database of the 14 measured roughness geometries used for training the roughness function model
Abstract
The two-part publication deals with roughness investigations on in-service high-pressure compressor (HPC) blades, both in terms of measurements and simulations. In this paper (Part II), first, stripe measurements of surface roughness coming from the suction side of the blades are conducted, using a highly accurate Alicona measuring device (see Part I for details of the measurement approach). Then, these roughness distributions are used to construct the walls of zero-pressure-gradient, fully turbulent channel flow simulations. Body-fitted unstructured grids of up to 80 M nodes are generated, on which wall-resolved large eddy simulations (LES) as well as Reynolds-averaged Navier–Stokes (RANS) simulations with the k–ω shear stress transport (SST) turbulence model are performed. The computational fluid dynamics (CFD) setup is first validated on a smooth channel reference case against LES and direct numerical simulations (DNS) data from the relevant literature. In addition, the impact of Reynolds number on several rough channel flow simulations is explored, using two different setups at Reτ = 540 and Reτ = 880. Finally, after an identification of the most important roughness parameters (given the relatively limited database at hand), a new roughness function model is proposed, which would allow the prediction of the flow over a rough surface without the need of geometrically resolving the roughness scales.