Development of a lattice boltzmann solver for analysis of separated flows over aerofoils at high reynolds numbers
File(s)
Author(s)
Mitchell, Sam
Type
Thesis
Abstract
Axial compressor blades are routinely subjected to flow separation that is detrimental to their aerodynamic performance, and can thus compromise the efficiency and stability of the compressor. The accurate prediction of flow separation is therefore of importance to the analysis of compressor performance and establishment of operational safety margins. It is however considerably challenging to numerically simulate separation, and methods of lower fidelity struggle to do so. This is particularly the case for compressor blades given their challenging geometric features such as small leading-edge radii and thin profiles. In the work contained in this thesis, the ability of the Lattice Boltzmann Method of computational fluid dynamics to simulate separated flows at high Reynolds numbers was investigated. Firstly, towards the identification of the most suitable Lattice Boltzmann model for this task, a novel linear stability analysis of the compressible finite-difference Lattice Boltzmann model was performed. This however revealed strict limits on the range of Reynolds numbers achievable with this particular model. It was therefore then decided instead to consider the entropic multiple relaxation time LB model, which has recently become known for its stability at high Reynolds numbers. A Lattice Boltzmann solver for high Reynolds number flows was then developed by implementation of this model into an existing high-performance Lattice Boltzmann framework. The new developments were then verified, where accurate results for turbulence and wall-resolved cylinder flows were achieved. It was then utilised for a novel application to an aerofoil test case at a Reynolds number of 100000. A reasonable agreement with an established wall-resolved large eddy simulation code was achieved at a low angle of attack. At higher angles of attack the results between both methods however differed substantially, highlighting the need for further developments in Lattice Boltzmann models for wall-resolved simulations of challenging geometries at high Reynolds numbers.
Version
Open Access
Date Issued
2023-03-15
Date Awarded
01/07/2023
License URL
Advisor
Stapelfeldt, Sina
Puente Rico, Ricardo
Brandstetter, Christoph
Morgans, Aimee
Sponsor
Engineering and Physical Sciences Research Council
Rolls-Royce (Firm)
Grant Number
NPIF2017
Publisher Department
Department of Mechanical Engineering
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)