Subfilter-scale Stress Modelling for Large Eddy Simulations


Autoria(s): Rouhi, Amirreza
Contribuinte(s)

Queen's University (Kingston, Ont.). Theses (Queen's University (Kingston, Ont.))

Piomelli, Ugo

Mechanical and Materials Engineering

Data(s)

29/11/2016

29/11/2016

Resumo

A subfilter-scale (SFS) stress model is developed for large-eddy simulations (LES) and is tested on various benchmark problems in both wall-resolved and wall-modelled LES. The basic ingredients of the proposed model are the model length-scale, and the model parameter. The model length-scale is defined as a fraction of the integral scale of the flow, decoupled from the grid. The portion of the resolved scales (LES resolution) appears as a user-defined model parameter, an advantage that the user decides the LES resolution. The model parameter is determined based on a measure of LES resolution, the SFS activity. The user decides a value for the SFS activity (based on the affordable computational budget and expected accuracy), and the model parameter is calculated dynamically. Depending on how the SFS activity is enforced, two SFS models are proposed. In one approach the user assigns the global (volume averaged) contribution of SFS to the transport (global model), while in the second model (local model), SFS activity is decided locally (locally averaged). The models are tested on isotropic turbulence, channel flow, backward-facing step and separating boundary layer. In wall-resolved LES, both global and local models perform quite accurately. Due to their near-wall behaviour, they result in accurate prediction of the flow on coarse grids. The backward-facing step also highlights the advantage of decoupling the model length-scale from the mesh. Despite the sharply refined grid near the step, the proposed SFS models yield a smooth, while physically consistent filter-width distribution, which minimizes errors when grid discontinuity is present. Finally the model application is extended to wall-modelled LES and is tested on channel flow and separating boundary layer. Given the coarse resolution used in wall-modelled LES, near the wall most of the eddies become SFS and SFS activity is required to be locally increased. The results are in very good agreement with the data for the channel. Errors in the prediction of separation and reattachment are observed in the separated flow, that are somewhat improved with some modifications to the wall-layer model.

Doctor of Philosophy

Identificador

http://hdl.handle.net/1974/15254

Idioma(s)

en

Relação

Canadian theses

Direitos

CC0 1.0 Universal

Queen's University's Thesis/Dissertation Non-Exclusive License for Deposit to QSpace and Library and Archives Canada

ProQuest PhD and Master's Theses International Dissemination Agreement

Intellectual Property Guidelines at Queen's University

Copying and Preserving Your Thesis

This publication is made available by the authority of the copyright owner solely for the purpose of private study and research and may not be copied or reproduced except as permitted by the copyright laws without written authority from the copyright owner.

http://creativecommons.org/publicdomain/zero/1.0/

Palavras-Chave #Turbulence #Computational Fluid Dynamics #Large Eddy Simulation #Subgrid-scale Modelling
Tipo

Thesis