Bulletin of the American Physical Society
2005 58th Annual Meeting of the Division of Fluid Dynamics
Sunday–Tuesday, November 20–22, 2005; Chicago, IL
Session AP: Turbulence Modeling: LES I |
Hide Abstracts |
Chair: Massimo Germano, Politecnico di Torino Room: Hilton Chicago Stevens 1 |
Sunday, November 20, 2005 8:00AM - 8:13AM |
AP.00001: Additive filters for LES Massimo Germano In the framework of a research devoted to analyze the algebraic properties of the filtering operators and the possible constraints that they impose on the filtered terms, some recent results concerning the additive filters are presented. These filters are produced by summing two or more generic filters and in a sense they are dual of the product filters that have been usefully employed in the dynamic modeling procedures. In this contribution we present particular additive RANS/LES and RANS/DNS filters and their properties are discussed in detail. As regards the possible applications it is interesting to remark that the associated subgrid stresses are derived by mixing filters, and not by mixing models. As such they seem appropriate to the development of controlled hybrid RANS/LES and RANS/DNS filtering procedures that gradually match different computational zones. [Preview Abstract] |
Sunday, November 20, 2005 8:13AM - 8:26AM |
AP.00002: Constrained Optimization for Dynamic Optimal Finite-Volume LES Paulo Zandonade, Robert D. Moser Optimal large eddy simulation models have been shown to produce accurate simulations, but until now they have been based on statistical data obtained from direct simulation. Here, the need for such DNS data is eliminated. A dynamic optimal finite-volume LES (FVLES) model is developed based on the constrained optimization of the mean-square error. Similar to previous optimal models, the present model is a stochastic estimate approximating the ideal LES, but with theoretically derived and dynamically determined correlations replacing the DNS data. In addition, constraints are imposed to control the large-scale dynamical behavior and numerical properties of the model in the presence of the uncertainties introduced by the dynamic procedure. The quadratic part of the model is constrained to be second-order accurate, while the coefficients for the linear part are constrained to be numerically stable. Results from the constrained dynamic FVLES models are compared to dynamic Smagorinsky results for forced isotropic turbulence. While the results are quite promising, the constraints applied to the linear part are \emph{ad-hoc} and must be further justified and analyzed. [Preview Abstract] |
Sunday, November 20, 2005 8:26AM - 8:39AM |
AP.00003: Properties of the Optimal LES Kernel for wall-bounded turbulence Amitabh Bhattacharya, Robert Moser Das (2004) succesfully performed an Optimal LES (OLES) of turbulent channel flow ($Re_{\tau}=590$), based on correlations obtained from DNS. A very coarse Fourier representation was used in all directions (including the wall normal direction), along with the filtered boundary formulation of Das \& Moser (2001). The Optimal subgrid model is purely dissipative, suggesting that simplier dissipative models such as Smagorinsky might be useful, and so a simulation employing the Dynamic Smagorinsky model was also performed, yielding poor results. We investigate the reason Dynamic Smagorinsky behaves poorly compared to the OLES model, with the goal of devising simple models that do not require DNS data, and that perform as well as OLES. We study the critical properties of the OLES model (e.g. dissipation spectra, kinetic energy spectra \& anisotropy of the OLES kernel), in comparison with the properties of the Dynamic Smagorinsky model. The desirable properties of subgrid models for LES of wall bounded flows are thus determined. [Preview Abstract] |
Sunday, November 20, 2005 8:39AM - 8:52AM |
AP.00004: Progress in Dynamic SGS Modeling for the SCALES Methodology -- Part I. Daniel E. Goldstein, Oleg V. Vasilyev, Nicholas K.-R \mbox{Kevlahan}, Giuliano \mbox{De Stefano} This is the first of two talks, which describe ongoing Dynamic SGS model development for the Stochastic Coherent Adaptive Large Eddy Simulation (SCALES) methodology. The SCALES methodology has the potential for significant improvement over regular grid LES methodologies with its ability to resolve and dynamically track the most energetic coherent structures in a turbulent flow through dynamic grid adaptation based on wavelet threshold filtering. The effect of unresolved SGS modes is modeled using a new dynamic eddy viscosity model based on modified Germano's dynamic procedure redefined in terms of two wavelet thresholding filters. As in LES, the modified dynamic procedure, when applied locally, results in model coefficient that can be locally positive or negative, which allows for local backscatter of energy to resolved scales. In practice, it has been found that locally negative values of model coefficient cause numerical instabilities. To deal with this difficulty, two different approaches are explored. In Part I, homogeneous direction averaging is considered, while in Part II Lagrangian pathline averaging is explored. The details for both approaches are given and the results of three-dimensional SCALES simulations of isotropic turbulence using both models are presented. [Preview Abstract] |
Sunday, November 20, 2005 8:52AM - 9:05AM |
AP.00005: Progress in Dynamic SGS Modeling for the SCALES Methodology -- Part II. Giuliano De Stefano, Daniel E. Goldstein, Oleg V. Vasilyev, Nicholas K.-R. Kevlahan This is the second of two talks, which describe ongoing Dynamic SGS model development for the Stochastic Coherent Adaptive Large Eddy Simulation (SCALES) methodology. The SCALES methodology has the potential for significant improvement over regular grid LES methodologies with its ability to resolve and dynamically track the most energetic coherent structures in a turbulent flow through dynamic grid adaptation based on wavelet threshold filtering. The effect of unresolved SGS modes is modeled using a new dynamic eddy viscosity model based on modified Germano's dynamic procedure redefined in terms of two wavelet thresholding filters. As in LES, the modified dynamic procedure, when applied locally, results in model coefficient that can be locally positive or negative, which allows for local backscatter of energy to resolved scales. In practice, it has been found that locally negative values of model coefficient cause numerical instabilities. To deal with this difficulty, two different approaches are explored. In Part I, homogeneous direction averaging is considered, while in Part II Lagrangian pathline averaging is explored. The details for both approaches are given and the results of three-dimensional SCALES simulations of isotropic turbulence using both models are presented. [Preview Abstract] |
Sunday, November 20, 2005 9:05AM - 9:18AM |
AP.00006: Near-wall formulation for LES, Part 1: RANS/LES coupling Gorazd Medic, Jeremy Templeton, Georgi Kalitzin A novel near-wall treatment for LES is presented. A RANS eddy-viscosity corrected dynamically using the resolved turbulent stress is imposed near the wall. This formulation is derived by comparing the averaged LES equations for channel flow to the RANS equations. Another new element is the procedure for computing the RANS eddy-viscosity, which is obtained from the non-dimensional RANS equation for channel flow using the averaged velocity profile from the LES. Results obtained using the proposed formulation are compared to LES and DNS for channel flow at $Re_{\tau} = 395$. In contrast to simulations that impose RANS eddy-viscosity in the near-wall region (similar to DES), the correct mean velocity profile is recovered and fluctuations are retained near the wall. Energy spectra and flow structures compare well to DNS results. [Preview Abstract] |
Sunday, November 20, 2005 9:18AM - 9:31AM |
AP.00007: Near-wall formulation for LES, Part 2: High Reynolds number flows Georgi Kalitzin, Jeremy Templeton, Gorazd Medic The near-wall eddy-viscosity formulation described in Part 1 is applied to flows at high Reynolds numbers. For high Reynolds numbers, computational cost limits LES to coarse grids where the near-wall region is not resolved. In place of no-slip boundary conditions, wall stress boundary conditions are applied. Coarse LES of channel flow has been performed for flows with Reynolds numbers up to $Re_{\tau} = 1,000,000$. Several computational grids have been used to assess the grid dependency of the method. In contrast to simulations with only the wall stress boundary conditions, the proposed near-wall treatment predicts well the mean velocity profile over a range of Reynolds numbers. Unlike RANS simulations, this method retains velocity fluctuations and coherent structures. Stress balances show that resolved stress is significantly larger than the modeled stress. [Preview Abstract] |
Sunday, November 20, 2005 9:31AM - 9:44AM |
AP.00008: Progress in the Multi-scale, Multi-domain Approach to Wall Modelling in LES M.U. Haliloglu, R. Akhavan The Multi-Scale, Multi-Domain (MSMD) method is an approach to wall-modelling based on the solution of the near-wall region in a minimal flow unit at fine resolution, repeating this unit periodically or quasi-periodically, and coupling this solution to a full domain solution in the core at coarse resolution. In this talk, we review the details of the MSMD approach and the progress made to date on the topic. We will discuss issues such as the minimum required size for the near- wall unit, effect of interface boundary conditions and scaling of the computational cost with Reynolds number. It will be shown that the near-wall unit needs to have minimum dimensions of $\sim\!5000$ wall units in the streamwise direction, $\sim\!2000$ wall units in the spanwise direction, and $\sim\!250$ wall units in the wall-normal direction in order to accommodate the near- wall hairpin vortex packets. The interface boundary conditions need to be specified such that they break the periodicity on the core side of the interface. With these conditions, simulations can be performed with a resolution of $32\times32\times17$ in the near-wall layer and $32\times64\times33$ in the core for up to $\mathrm{Re}_\tau\approx10000$. For higher $\mathrm{Re}_\tau$, larger near-wall units are required to provide some overlap between the grid spacing in the core and the overall size of the near-wall region. The results obtained in a turbulent channel flow based on application of the above concepts will be discussed. [Preview Abstract] |
Sunday, November 20, 2005 9:44AM - 9:57AM |
AP.00009: A dynamic globalization subgrid-scale model for large eddy simulation of complex turbulent flows Haecheon Choi, Noma Park, Jinseok Kim, Jungil Lee In the present study, a dynamic subgrid-scale model is proposed for large eddy simulation of turbulent flows in complex geometry. The eddy viscosity model recently proposed by Vreman [Phys. Fluids, 16, 3670 (2004)], which guarantees theoretically zero SGS dissipation for various laminar shear flows, is considered as a base model. A priori tests with the original Vreman model show that it predicts the correct profile of subgrid-scale dissipation in turbulent channel flow but the optimal model coefficient is far from universal. Dynamic procedures of determining the model coefficient are proposed based on the ‘global equilibrium’ between the subgrid-scale dissipation and the viscous dissipation. An important feature of the proposed procedures is that the model coefficient determined is globally constant in space but varies only in time. A posteriori tests of the proposed dynamic model are conducted through large eddy simulations of forced isotropic turbulence, turbulent channel flow, flow over a sphere, and flow over a three-dimensional model vehicle. The proposed dynamic model produces excellent performance for all flows considered. The proposed model is quite robust and it can be readily applied to complex flows without homogeneous direction. [Preview Abstract] |
Sunday, November 20, 2005 9:57AM - 10:10AM |
AP.00010: Evaluation of a multifractal model for the LES of a filtered passive-scalar field Gregory Burton A model for the flux of a filtered passive scalar in large eddy simulation is developed from the multifractal structure of the scalar-dissipation field at inertial-range scales in high Reynolds-number turbulence. The model is based on a scale-invariant multifractal cascade governing the spatial structure the subgrid-scalar dissipation field $\chi^{sgs}({\bf x},t)$, and an additive cascade for the progressively isotropic decorrelation of the subgrid-scalar gradient orientations from those of the smallest resolved scale in the simulation. This approach permits the determination of the subgrid-scalar concentrations, $\zeta^{sgs}$, permitting the direct calculation of the filtered nonlinear term, $\overline{u_j \,\zeta}$ in the filtered passive-scalar transport equation. Results are then presented from {\it a posteriori} evaluations of the model in LES of homogeneous, isotropic turbulence with a passive-scalar field that is forced by a mean-scalar gradient. [Preview Abstract] |
Sunday, November 20, 2005 10:10AM - 10:23AM |
AP.00011: Relations between the local structure of turbulence and the Smagorinsky coefficient Marcelo Chamecki, Charles Meneveau, Marc B. Parlange The objective of this study is to uncover systematic dependencies of the Smagorinsky coefficient upon dimensionless parameters that characterize the local flow conditions in turbulence. The local flow can be characterized using invariants of the velocity gradient tensor that are related to some fundamental dynamic/ kinematic characteristics of turbulent flows, such as dissipation, enstrophy, vortex stretching, self-amplification of strain rate, etc. Such invariants are often used to classify local flow topology and they are expected to play an important role in scale-interactions of relevance to subgrid-scale modeling for Large Eddy Simulation. In this a-priori study we use the HATS field data obtained from 14 3D-sonic anemometers arranged in two horizontal arrays in the atmospheric surface layer. For comparison, DNS data of isotropic turbulence is used as well. We make use of conditional averages to investigate how the SGS dissipation and the Smagorinsky coefficient depend on local properties of the resolved field. Besides the invariants of the resolved velocity gradient tensor, for the HATS data the local gradient Richardson number is also used. We briefly discuss how this information can be used to improve dynamic SGS models. [Preview Abstract] |
Sunday, November 20, 2005 10:23AM - 10:36AM |
AP.00012: Time-Dependent Boundary Conditions for Large Eddy Simulation Traian Iliescu, Jeff Borggaard, Alexey Miroshnikov We present a new boundary treatment for Large Eddy Simulation of flows with time-dependent boundary conditions. The approach uses approximate deconvolution to approximate the velocity near the boundary. With these approximations and the exact value of the velocity at the boundaries, the method computes the approximation for the filtered velocity at the boundaries at the new time-step. We illustrate the new method in the numerical simulation of three-dimensional channel flows with time-dependent boundary conditions. [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