Loosely Coupled Under-Resolved LES/RANS Simulation Augmented by Sparse Near-Wall Measurement

Abstract

We investigate scenarios, where only sparse wall shear stress measurements are available, while accurate wall shear stress and velocity profiles are sought. Applying discrete adjoint-based data assimilation, with only near-wall measurements, accurate wall shear stress profiles are achieved at the expense of unrealistic velocity profiles. We therefore add and employ internal reference data generated by performing a relatively cheap hybrid simulation. We modified the dual-mesh hybrid LES/RANS framework recently proposed by Xiao and Jenny~\cite{xiao12} by loosely coupling under-resolved LES in the interior with steady RANS near the walls. The framework was developed in OpenFOAM and tested for flow over periodic hills with Re = 10595. Results show that the devised framework outperforms conventional dual-mesh hybrid LES/RANS and standalone sparse wall-data assimilated RANS models.