Application and improvement of a scale-resolving hybrid method for simulations of complex turbulent flows at high Reynolds number
File(s)
Author(s)
Mays, Michael
Type
Thesis
Abstract
Computational Fluid Dynamics (CFD) plays an increasingly major role in the design, validation and optimisation of industrial processes. Despite the dramatic increase in computing power, resolved simulations, using Large-Eddy or Direct Numerical simulation, of full-scale geometries in high Reynolds number flows remain prohibitively costly. Consequently low-cost Reynolds Averaged Navier-Stokes (RANS) turbulence modeling is widely used in industry to simulate turbulent flows. Unfortunately, such approaches continue to show inaccuracy in simulations of complex flows featuring separation. One method to overcome the cost-accuracy trade-off is the Hybrid RANS/LES turbulence model class which combines Large-Eddy simulation (LES) and RANS methods.
The Hybrid Temporal LES (HTLES) model from the literature was taken as the basis of a Scale-Resolving Hybrid (SRH) method to treat high Reynolds separated flows. The method was extended to the Realisable k-epsilon and Lag Elliptic Blending k-epsilon RANS models and applied to high and low y+ wall treatments. Modifications were made to the base method including the incorporation of the Durbin realisability limiter and imposing a shielding function on the dissipation enhancement function.
Validation was conducted using canonical cases featuring a range of flow dynamics consisting of decaying isotropic turbulence, channel flow, flow over periodic hills and flow over a finite-height cylinder. Across the cases the SRH model showed desirable features including a general robustness to variation in temporal resolution, dependent on the base RANS model, the ability to return LES quality results on grids of varying resolution and and accurate channel flow results using the two wall treatments. Having shown robust behaviour for the canonical cases, the model was used to predict the drag difference between two DrivAer automotive model geometries, a test case realistic to industry. The SRH method provided results comparably accurate to other scale-resolving methods, using a large time step and reduced computational cost.
The Hybrid Temporal LES (HTLES) model from the literature was taken as the basis of a Scale-Resolving Hybrid (SRH) method to treat high Reynolds separated flows. The method was extended to the Realisable k-epsilon and Lag Elliptic Blending k-epsilon RANS models and applied to high and low y+ wall treatments. Modifications were made to the base method including the incorporation of the Durbin realisability limiter and imposing a shielding function on the dissipation enhancement function.
Validation was conducted using canonical cases featuring a range of flow dynamics consisting of decaying isotropic turbulence, channel flow, flow over periodic hills and flow over a finite-height cylinder. Across the cases the SRH model showed desirable features including a general robustness to variation in temporal resolution, dependent on the base RANS model, the ability to return LES quality results on grids of varying resolution and and accurate channel flow results using the two wall treatments. Having shown robust behaviour for the canonical cases, the model was used to predict the drag difference between two DrivAer automotive model geometries, a test case realistic to industry. The SRH method provided results comparably accurate to other scale-resolving methods, using a large time step and reduced computational cost.
Version
Open Access
Date Issued
2023-12
Date Awarded
2024-11
Copyright Statement
Creative Commons Attribution NonCommercial Licence
License URL
Advisor
Laizet, Sylvain
Sponsor
Engineering and Physical Sciences Research Council
Grant Number
EPSRC ICASE 2018 (EP/S513635/1)
Publisher Department
Aeronautics
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)