Modelling Techniques for Large-Eddy Simulation of Wall-Bounded Turbulent Flows

Abstract: Large-eddy simulation (LES) is a highly accurate turbulence modelling approach in which a wide range of spatial and temporal scales of the flow are resolved. However, LES becomes prohibitively computationally expensive when applied to wall-bounded flows at high Reynolds numbers, which are typical of many industrial applications. This is caused by the need to resolve very small, yet dynamically important flow structures found in the inner region of turbulent boundary layers (TBLs). To remove the restrictive resolution requirements, coupling LES with special models for the flow in the inner region has been proposed. The predictive accuracy of this promising approach, referred to as wall-modelled LES (WMLES), requires further analysis and validation. In this work, systematic simulation campaigns of canonical wall-bounded flows have been conducted to support the development of a complete methodology for highly accurate WMLES on unstructured grids. Two novel algebraic wall-stress models are also proposed and shown to be more robust and precise than the classical approaches of the same type. For turbulence simulations, it is often challenging to provide accurate conditions at the inflow boundaries of the domain. Here, a novel methodology is proposed for generating an inflow TBL using a precursor simulation of turbulent channel flow. A procedure for determining the parameters of the precursor based on the Reynolds number of the inflow TBL is given. The proposed method is robust and easy to implement, and its accuracy is demonstrated to be on par with other state-of-the-art approaches. To make the above investigations possible, several software packages have been developed in the course of the work on this thesis. This includes a Python package for post-processing the flow simulation results, a Python package for inflow generation methods, and a library for WMLES based on the general-purpose software for computational fluid dynamics OpenFOAM. All three codes are publicly released under an open-source licence to facilitate their use by other research groups.