Large Eddy Simulation of Turbulent Flow Past a Bluff Body using OpenFOAM A Thesis Presented By David Joseph Hensel To The Department of Mechanical and Industrial Engineering in partial fulfillment of the requirements for the degree of Master of Science in Mechanical Engineering Northeastern University Boston, Massachusetts August 2014
27
Embed
Large eddy simulation of turbulent flow past a bluff body ...336868/... · The&opensource&CFD&software&package,&OpenFOAM,is&employed&to ... isotropic&part& of& the&residual&stress
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Large Eddy Simulation of Turbulent Flow Past a Bluff Body using OpenFOAM
A Thesis Presented
By
David Joseph Hensel
To
The Department of Mechanical and Industrial Engineering
in partial fulfillment of the requirements for the degree of
Master of Science
in
Mechanical Engineering
Northeastern University Boston, Massachusetts
August 2014
Large Eddy Simulation of Turbulent Flow Past a Bluff Body using OpenFOAM
Abstract Numerical simulation of a turbulent bluff-‐body flow is conducted using large eddy simulation (LES). The open source CFD software package, OpenFOAM, is employed to solve the LES filtered transport equations governing the three-‐dimensional incompressible flow in the wake of the body. This work is motivated by the importance of bluff-‐body flows, for example, in flame stabilization in industrial combustors and burners, as well as in aerodynamics applications. The focus of the study is on the proper generation of turbulence at the inlet boundary. Standard boundary conditions available in OpenFOAM are not sufficient for providing an accurate turbulent inlet condition without modification of the bluff geometry. An improved method describing the inflow boundary condition is developed based on the existing OpenFOAM mapping-‐type boundary condition. In this method, the boundary condition scales the standard deviation and mean value of the velocity field onto the prescribed values provided by the experimental data. The method is implemented in OpenFOAM and employed in LES prediction of a turbulent bluff-‐body flow, studied in the experiments of the Clean Combustion Research Group at the University of Sydney. The LES results show favorable agreements with the experimental data.
Thesis Committee Members
Prof. Reza Sheikhi
Contents
Table of Figures .......................................................................................................................................................................... 1
Figure 5: Magnitude of Instantaneous Vorticity Iso-‐surfaces (clipped by X-‐Y plane)
Figure 6: Line Integral Convolution of Streamwise Filtered Velocity in Bluff Region, X-‐Y Plane
16
Figure 7: LES filtered Streamwise Velocity Contours Predicted by the Smagorinsky Model.
a) Smagorinsky, X-‐Y plane, Instantaneous contours of ⟨𝑢⟩
b) Smagorinsky, X-‐Z plane, Instantaneous contours of ⟨𝑢⟩
c) Smagorinsky, X-‐Y plane, Time-‐Averaged contours of ⟨𝑢⟩
17
Figure 8: LES filtered Streamwise Velocity Contours Predicted by the Dynamic One Equation Model.
a) Dynamic One Equation, X-‐Y plane, Instantaneous contours of ⟨𝑢⟩
b) Dynamic One Equation, X-‐Z plane, Instantaneous contours of ⟨𝑢⟩
c) Dynamic One Equation, X-‐Y plane, Time-‐Averaged contours of ⟨𝑢⟩
18
Figure 9: Radial Profiles of the Mean and Resolved RMS Streamwise Velocity (x=0.003, 0.01, 0.02 [m]).
19
Figure 10: Radial Profiles of the Mean and Resolved RMS Streamwise Velocity (x=0.03, 0.04, 0.05 [m]).
20
Figure 11: Radial Profiles of the Mean and Resolved RMS Streamwise Velocity (x=0.06 [m]).
Figure 12: Radial Profiles of the Mean Radial Velocity (x=0.003, 0.01 [m]).
21
Figure 13: Radial Profiles of the Mean Radial Velocity (x = 0.02, 0.03, 0.04[m]).
22
Figure 14: Radial Profiles of the Mean Radial Velocity (x = 0.05, 0.06[m]).
5. Summary and Concluding Remarks
Large eddy simulation (LES) is conducted of a turbulent bluff body flow using OpenFOAM. The
results are comparable with the experimental data provided by the Clean Combustion Research
Group at the University of Sydney. The focus of this study is on implementing an improved method
in OpenFOAM to generate the turbulent inflow condition. This is handled by a mapping boundary
condition, which samples a location from within the computational domain and scales the field to a
prescribed mean and standard deviation.
LES predictions are obtained via two subgrid scale models: Smagorinsky and Localized Dynamic
One Equation Eddy Viscosity. Both models provide reasonable overall prediction of the turbulent
wake behind the bluff body. In general, the Smagorinsky model tends to under-‐predict the
turbulent fluctuations and the Localized Dynamic One Equation Eddy Viscosity model tends to
slightly over-‐predict turbulent fluctuations. The under-‐prediction of fluctuations by the
Smagorinsky model is likely due to the dissipative nature of this LES model. As the velocity
contours are generally well produced in the region closest to the inlet, but deviate from the
experimental data further into the domain, it is likely that numerical schemes and LES models have
23
a significant role in the accuracy of these simulations. Altering the inlet parameters provides some
control over the velocity profiles; however, those profiles past the recirculation zone are always
under-‐predicted in the results. The predictions obtained from the Dynamic One Equation model
are more consistent with the experimental data, for both first and second order moments. This
work provides a preliminary implementation of a modified mapped boundary condition in
OpenFOAM by including additional controls to match a prescribed inflow condition. OpenFOAM
along with the new inlet boundary condition has shown to provide LES prediction of turbulent
flows past a bluff body with reasonable accuracy. The near field of the flow is favorably predicted.
The far field however shows less accuracy mainly near the centerline and due to over-‐prediction of
the spread rate of the jet. The modified mapping boundary condition has the potential for further
developments including additional control over second order statistics, and scaling to maintain a
specified distribution as well as parameters to control the behavior of the jet downstream.
24
References
[1] D. C. Wilcox, Turbulence modeling for CFD vol. 1. La Cañada, CA: DCW Industries, Inc., 1993.
[2] W. Rodi, "Comparison of LES and RANS calculations of the flow around bluff bodies," Journal of Wind Engineering and Industrial Aerodynamics, vol. 69–71, pp. 55-‐75, 1997.
[3] S. B. Pope, Turbulent Flows. New York, NY: Cambridge University Press, 2000.
[4] The University of Sydney. (2014). Bluff-‐Body Flows and Flames. Available: http://sydney.edu.au/engineering/aeromech/thermofluids/bluff.htm
[5] T. G. Drozda, "Implementation of LES/SFMDF for prediction of non-‐premixed turbulent flames," Ph.D. Dissertation, University of Pittsburgh, 2006.
[6] T. Drozda, M. Sheikhi, C. Madnia, and P. Givi, "Developments in formulation and application of the filtered density function," Flow, Turbulence and Combustion, vol. 78, pp. 35-‐67, 2007.
[7] D. A. Lysenko, I. S. Ertesvåg, and K. E. Rian, "Modeling of turbulent separated flows using OpenFOAM," Computers & Fluids, vol. 80, pp. 408-‐422, 2013.
[8] N. M. C. Salvador, M. T. Mendonça, and W. M. da Costa Dourado, "Large Eddy Simulation of bluff body stabilized flame with turbulent premixed flame," Guidelines of Workshop on Space Engineering and Technology, June 2012.
[9] G. Tabor, M. Baba-‐Ahmadi, E. de Villiers, and H. Weller, "Construction of inlet conditions for LES of turbulent channel flow," in Proceedings of the ECCOMAS congress, Jyväskylä, Finland, 2004.
[10] J. Volavy, M. Forman, and M. Jicha, "Influence of the turbulence representation at the inlet on the downstream flow pattern in LES of backward-‐facing step," EPJ Web of Conferences, vol. 25, 2012.
[11] OpenFOAM Foundation. (2014). OpenFOAM® Release History. Available: http://www.openfoam.org/download/history.php
[13] OpenFOAM Foundation. (2014). OpenFOAM® Programmer's C++ documentation -‐ Smagorinsky Class Reference. Available: http://foam.sourceforge.net/docs/cpp/a02317.html
[14] OpenFOAM Foundation. (2014). OpenFOAM® Programmer's C++ documentation -‐ dynOneEqEddy Class Reference. Available: http://foam.sourceforge.net/docs/cpp/a00580.html
[15] G. R. Tabor and M. H. Baba-‐Ahmadi, "Inlet conditions for large eddy simulation: A review," Computers & Fluids, vol. 39, pp. 553-‐567, 2010.