Embedded LES of a turbulent thermal boundary layer over ice roughness

The numerical prediction of ice accretion via icing codes relies on the proper estimation of the heat transfer coefficient on rough ice geometries. To understand the heat transfer physics at play, direct numerical simulations (DNS) on rough surfaces can be performed, although this results in a very expensive option if geometries obtained from different icing conditions want to be analyzed. Large eddy simulation (LES) presents itself as a less expensive option to perform such studies, giving insight into the physics of turbulence, as well as opening the possibility for calibration of roughness models. The present study verifies and validates a setup to perform embedded LES (ELES) of a zero pressure gradient incompressible flow over a flat plate with ice roughness heated to a constant wall temperature. An experimental database is used, which provides the geometries of rough plates obtained from unwrapped scans of ice shapes generated on a NACA0012 airfoil. The low-speed flow over the flat plate presents a $R{e}_L=3.85\cdot 10^5$ and $Pr=0.729$. The verification is carried out by analyzing the effects of the mesh resolution and the domain span on wall properties such as the skin friction coefficient and Stanton number. Additionally, an analysis of turbulence-related flow statistics is performed to guarantee the proper development of turbulence. The validation shows good agreement between ELES results and experimental data, especially for the Stanton number distributions, showcasing that this setup can be used for the study of heat transfer on ice roughness.