Bulletin of the American Physical Society
2006 59th Annual Meeting of the APS Division of Fluid Dynamics
Sunday–Tuesday, November 19–21, 2006; Tampa Bay, Florida
Session FM: Turbulence Simulations II |
Hide Abstracts |
Chair: Meng Wang, University of Notre Dame Room: Tampa Marriott Waterside Hotel and Marina Meeting Room 10 |
Monday, November 20, 2006 8:00AM - 8:13AM |
FM.00001: Joint Frequency-Velocity-Scalar Filtered Mass Density Function for Large Eddy Simulation of Turbulent Reacting Flows Reza Sheikhi, Peyman Givi, Stephen Pope A new methodology, termed the frequency-velocity-scalar filtered mass density function (FVS-FMDF) is developed for large eddy simulation (LES) of turbulent reacting flows. The FVS-FMDF represents the joint frequency-velocity-scalar probability density function of the subgrid scale quantities and is obtained by solving its modeled transport equation. In this equation, the effects of convection and chemical reaction appear in closed forms. The unclosed terms are modeled in a fashion similar to PDF methods [1]. The modeled FVS-FMDF transport equation is solved by a Lagrangian Monte Carlo method. The methodology is employed to simulate turbulent shear flows. The predicted results are assessed by comparisons with data generated by direct numerical simulation (DNS). The predictions are in good agreements with DNS data. The FVS-FMDF is the most comprehensive form of the FDF up to now. \newline \newline [1] Pope, S. B., Turbulent flows, Cambridge University Press, Cambridge, UK (2000). [Preview Abstract] |
Monday, November 20, 2006 8:13AM - 8:26AM |
FM.00002: Effect of discontinuous filters on LES Seongwon Kang, Ugo Piomelli, Frank Ham, Gianluca Iaccarino Simulations of turbulent flows in complex geometries often rely on unstructured grids, in which sharp interfaces between regions with widely different resolutions may occur. Here we perform LES in which interfaces between fine and coarse grids are artificially introduced. We distinguish two types of interfaces: parallel to the mean advection direction, and normal to it. In the first case, while the resolved stresses decrease as the grid is coarsened, the subgrid-scale (SGS) ones increase proportionately. If the grid interface is placed very close to a solid surface, a thicker sublayer results. When the interface is normal to the main advection we observe more complex behaviors. A sudden grid coarsening results in aliasing error and in loss of phase information. None of the SGS models tested is capable of adjusting properly to the decrease of the resolved stresses. A coarse-to-fine interface has a more benign character. Improvements based on decoupling the filter-width from the grid size and explicit filtering are proposed. [Preview Abstract] |
Monday, November 20, 2006 8:26AM - 8:39AM |
FM.00003: Dynamic modeling for LES of turbulent jet mixing in spherical coordinates without explicit test filtering O.S. Sun, L.K. Su, T.M. Burton The dynamic model for LES requires explicit test filtering of grid-resolved quantities to compute model coefficients. On a regular, Cartesian grid, the test filter at any location in the computational domain is usually implemented by taking a local average of quantities at neighboring grid points. In complex flows involving irregular or unstructured grids, or simulations performed in non-Cartesian coordinate systems, the test filtering operation is less intuitively interpreted. Another approach (Chester, Charlette, Meneveau (2001)) approximates test filtering by expanding grid-resolved quantities locally in a truncated Taylor series. In this work, we implement this test filter approximation in LES of scalar mixing in a round turbulent jet. This approach provides a more robust and consistent definition of the test filter on the spherical coordinate grid used in the simulation, and allows more accurate determination of the dynamic coefficient for the subgrid scale (SGS) models. Previous studies have shown the resolved-scale scalar field to be particularly sensitive to the numerics of the simulation, including the spatial discretization scheme, filter width, as well as the amount of `backscattering' allowed by the subgrid scalar flux model. We aim to determine how the evolution of the dynamic coefficients for the SGS stress and SGS scalar flux models influences the mixing properties of the simulated turbulent jet, at both the resolved and subgrid scales, and how closely the resulting velocity and scalar fields correlate with experimental measurements. [Preview Abstract] |
Monday, November 20, 2006 8:39AM - 8:52AM |
FM.00004: Subgrid-scale modeling effects on wall pressure fluctuations Meng Wang Accurate computations of wall pressure fluctuations underneath a turbulent boundary layer are of interest in aeroacoustic and hydroacoustic applications. In this study LES predictions of the spatio-temporal statistics of fluctuating wall pressure are systematically investigated in a plane channel at friction Reynolds number of 395 by comparison with DNS results. As expected, LES predicts well the low wavenumber/frequency ranges of the wall pressure spectra, but underpredicts the higher wavenumber/frequency spectra and hence the magnitude of wall pressure fluctuations due to a combination of filtering, subgrid-scale modeling, and numerical errors. Another significant effect of LES is the overprediction of correlation time and length scales, which is consistent with previous observations regarding velocity space-time correlations. Furthermore, subgrid-scale modeling is shown to significantly alter the convection velocities of small-scale pressure fluctuations. The underlying causes for these errors are analyzed, and means to improve LES predictions are explored. [Preview Abstract] |
Monday, November 20, 2006 8:52AM - 9:05AM |
FM.00005: A Study of Passive-Scalar Mixing by Round Turbulent Jets using Inertial Large-Eddy Simulation with Multifractal SGS Modeling Gregory Burton Large-eddy simulation of passive scalar mixing by a round incompressible turbulent jet is evaluated using the Inertial LES methodology with multifractal subgrid-scale modeling. The present work extends previous efforts in which the Inertial LES approach with multifractal modeling has been successfully applied to LES of forced and decaying homogeneous isotropic turbulence, as well as turbulent mixing of a passive scalar. The Inertial LES approach involves the direct calculation of both the inertial term $< u_i \, u_j >$ in the filtered incompressible Navier-Stokes equation, and the scalar flux term $< u_j \, \phi >$ in the filtered scalar advection-diffusion equation, rather than the use of traditional artificial viscosity closures. The approach requires models both for the subgrid velocity $u^{sgs}$ and scalar fields $\phi^{sgs}$, which are based on the multifractal structure of the enstrophy field and scalar dissipation field, respectively. The method produces high accuracy in the local spatial structure of the momentum, kinetic-energy and passive-scalar energy transfer fields between the resolved and subgrid scales, with correlations exceeding 0.99. The Inertial LES approach also has already produced accurate simulations of high-Schmidt number turbulent mixing. The presentation will focus on the evaluation of the Inertial LES approach in the significantly more complex case of scalar mixing by a turbulent round jet at both low and high Schmidt number. [Preview Abstract] |
Monday, November 20, 2006 9:05AM - 9:18AM |
FM.00006: Application of hybrid LES-RANS model to turbulent boundary layer flows over rough wall Kojiro Nozawa, Tetsuro Tamura In this study, Large Eddy Simulation (LES) of turbulent boundary flows over homogenous roughness were performed using hybrid LES- RANS model which can represent appropriately and efficiently the roughness condition on ground surface[F. Hamba, Theoret. Comput. Fluid Dynamics 16, 387--403 (2003)]. In LES of boundary layer flows over vegetation fields, leaves and plants are too thin to resolve them by the sufficient number of grid points. So, the effect of those leaves and plants on the flow must be treated with an artificial model. The turbulence closure model for plant canopy flows in this study was proposed by Hiraoka and Ohashi[Proc. of The 4th Int. Symp. on Comp. Wind Eng., 693-696, (2006)], which is formulated based on RANS($k-\epsilon$) turbulence model. The upper limit of RANS region is set at 2$h$, where $h$ is the height of the canopy, and the higher region is simulated using one-equation SGS model of LES. To reduce the mismatch of mean velocity profile between RANS and LES regions due to a steep velocity gradient at the interface, the buffer region is introduced, where LES works well by the smooth changes of the filter width according to the distance from the LES-RANS interface. The boundary layer thickness is approximately 200$h$ and the Reynolds number ($=U\delta/\nu$) is 2.0$\times$10$^7$. The mean velocity profiles and turbulence characteristics are compared to the results obtained by full scale measurements of the past studies. [Preview Abstract] |
Monday, November 20, 2006 9:18AM - 9:31AM |
FM.00007: Large Eddy Simulations of Turbulent Flow and Heat Transfer in Three-Dimensional Lid-Driven Shallow Cavities Arindom Joardar, Surya Vanka, Anthony Jacobi Large-eddy simulations (LES) of turbulent flow and heat transfer in three-dimensional lid-driven cavities of aspect ratios (AR) of 1, 2 and 4 are carried out at Reynolds numbers of 5000, 10000 and 20000. The governing equations are box-filtered implicitly by a finite volume scheme and are solved using a fractional-step method. Subgrid scale stresses are represented using Yoshizawa's SGS kinetic-energy transport equation. It was found that the three-dimensional vortical structures play an important role in the heat transfer characteristics at the confining walls. For AR=1, the highest levels of temperature fluctuations occur within the free shear layer between the primary vortex and the downstream secondary eddy, similar to that observed for the velocity field. Mean and RMS velocity profiles were found to be in good agreement with published experimental data. For higher aspect ratio cases the upstream secondary eddy (USE) was found to bifurcate at high Reynolds number as observed from the streamlines of mean velocity field at the mid-span plane. Interestingly for AR=4, the bifurcations of the USE further increased to five at Re=20000 which may be attributable to the Kelvin-Helmholtz type instability at the free shear layer between the Primary eddy (PE) and USE. Profiles of u$_{rms}$ fluctuations and v$_{rms}$ fluctuations are studied. [Preview Abstract] |
Monday, November 20, 2006 9:31AM - 9:44AM |
FM.00008: Large Eddy Simulation of the Neutrally Stratified Atmospheric Boundary Layer Tie Wei, James Brasseur A long standing problem in large eddy simulation of the neutrally stratified atmospheric boundary layer (ABL) is the excessive shear ($\phi_m=\frac{\kappa z}{u_*}\frac{\partial U}{\partial z}$) found in the semi-resolved region close to the surface. Due to the high Reynolds number of the flow Schumann-Grotzbach type `wall stress models' are commonly used for the lower wall boundary conditions. The overshoot of $\phi_m$ has been attributed to the subfilter scale model and the lower wall boundary conditions. To improve the accuracy of the simulation, some researchers have focused on modification of the subfilter scale model, while using the classical wall stress model. To determine the relationship between the subfilter scale model, the lower wall boundary conditions and the overshoot of $\phi_m$, we carried out simulations with several subfilter scale models and various modified lower wall boundary conditions. We examined the profiles of the mean resolved Reynolds stress and subfilter stress, and explored their relation to the structure of turbulent boundary layers. Our results show that small changes in the lower wall boundary conditions have a large influence on the simulation results. We aim to determine a pertinent combination of modifications to both the lower wall boundary conditions and the subfilter scale models to improve the accuracy in the near-wall region. [Preview Abstract] |
Monday, November 20, 2006 9:44AM - 9:57AM |
FM.00009: Large-Eddy Simulation (LES) of the NASA Hump Flow Using Dynamic Sub-Grid-Scale (SGS) Model Subhadeep Gan, Urmila Ghia, K.N. Ghia LES using dynamic SGS models is employed to investigate turbulent flow over the NASA Hump flow. This has a simple geometry, but, nevertheless, is rich in many complex flow phenomena such as shear layer instability, separation, reattachment, and vortex interactions. The flow is first simulated using the dynamic SGS model (Germano \textit{et al.}, 1991) with freestream Reynolds number of approximately 936,000 and standard atmospheric conditions. Next, LES with an integral-type localization two-parameter dynamic SGS model (Wang and Bergstrom, 2004) is employed. A multi-block structured 3D computational grid is used for the simulation. Mean-velocity contours, turbulent kinetic energy contours, and streamlines will be examined. Detailed comparisons will be made of mean and turbulence statistics such as the pressure coefficient, skin-friction coefficient, Reynolds stress profiles, and wall shear stress, with experimental results. The location of the reattachment behind the hump will be compared with previously published numerical simulations and experimental results. The correlation between the large-scale coherent structures and the SGS events is expected to be predicted more accurately by the integral-type localization two-parameter dynamic SGS model in comparison to eddy viscosity models. [Preview Abstract] |
Monday, November 20, 2006 9:57AM - 10:10AM |
FM.00010: A LES-Langevin model for turbulence Rostislav Dolganov, B\'ereng\`ere Dubrulle, Jean-Philippe Laval The rationale for Large Eddy Simulation is rooted in our inability to handle all degrees of freedom ($N\sim 10^{16}$ for $Re\sim 10^7$). ``Deterministic'' models based on eddy-viscosity seek to reproduce the intensification of the energy transport. However, they fail to reproduce backward energy transfer (backscatter) from small to large scale, which is an essentiel feature of the turbulence near wall or in boundary layer. To capture this backscatter, ``stochastic'' strategies have been developed. In the present talk, we shall discuss such a strategy, based on a Rapid Distorsion Theory (RDT). Specifically, we first divide the small scale contribution to the Reynolds Stress Tensor in two parts: a turbulent viscosity and the pseudo-Lamb vector, representing the nonlinear cross terms of resolved and sub-grid scales. We then estimate the dynamics of small-scale motion by the RDT applied to Navier-Stockes equation. We use this to model the cross term evolution by a Langevin equation, in which the random force is provided by sub-grid pressure terms. Our LES model is thus made of a truncated Navier-Stockes equation including the turbulent force and a generalized Langevin equation for the latter, integrated on a twice-finer grid. The backscatter is automatically included in our stochastic model of the pseudo-Lamb vector. We apply this model to the case of homogeneous isotropic turbulence and turbulent channel flow. [Preview Abstract] |
Follow Us |
Engage
Become an APS Member |
My APS
Renew Membership |
Information for |
About APSThe American Physical Society (APS) is a non-profit membership organization working to advance the knowledge of physics. |
© 2024 American Physical Society
| All rights reserved | Terms of Use
| Contact Us
Headquarters
1 Physics Ellipse, College Park, MD 20740-3844
(301) 209-3200
Editorial Office
100 Motor Pkwy, Suite 110, Hauppauge, NY 11788
(631) 591-4000
Office of Public Affairs
529 14th St NW, Suite 1050, Washington, D.C. 20045-2001
(202) 662-8700