Reynolds-constrained Large-eddy Simulation : Sensitivity to Constraint and Subgrid-scale Models

Rui Wang,Zuoli Xiao

DOI: https://doi.org/10.1007/978-3-030-27607-2_10

2018-01-01

Abstract:Computational fluid dynamics (CFD) technique is playing increasingly important roles in the field of industrial engineering. Reynolds-averaged Navier-Stokes (RANS) simulation technique has been widely used in commercial software because of low requirement for computing grids, but its application to unsteady and separated flows remains questionable and unsatisfactory due to the lack of generality of turbulence models. Largeeddy simulation (LES) can predict three-dimensional (3D) and unsteady flow fields, however, it still requires fairly fine grids when used to solve the wall-bounded turbulent flows of engineering interest. Such a situation has considerably stimulated the development of hybrid RANS/LES methods, among which is the commonly employed detached-eddy simulations (DES) technique [1]. Yet with the inherent advantages, hybrid RANS/LES methods have problematic drawbacks. For example, DES encounters the well-known log-layer mismatch (LLM) defect in predicting a simple turbulent channel flow [2]. In order for the LES to be feasible to numerically predict wall-bounded turbulent flows, Chen et al. [3] proposed a Reynolds-constrained large-eddy simulation (RCLES) method, which opened up a new way of modelling subgrid-scale (SGS) effects for LES of wall-bounded turbulent flows.