Bulletin of the American Physical Society
66th Annual Meeting of the APS Division of Fluid Dynamics
Volume 58, Number 18
Sunday–Tuesday, November 24–26, 2013; Pittsburgh, Pennsylvania
Session M4: Turbulence: Modeling IV |
Hide Abstracts |
Chair: Sanjiva K. Lele, Stanford University Room: 326 |
Tuesday, November 26, 2013 8:00AM - 8:13AM |
M4.00001: A subfilter-scale stress model for large eddy simulations Amirreza Rouhi, Ugo Piomelli In most large eddy simulations, the filter width is related to the grid. This method of specification, however, causes problems in complex flows where local refinement results in grid discontinuities. Following the work of Piomelli and Geurts (\textit{Proce.\ 8th Workshop on DLES}, 2010) we propose an eddy-viscosity approach in which the filter width is based on the flow parameters only, with no explicit relationship to the grid size. This model can achieve grid-independent LES solutions, vanishing dynamically in the regions of low turbulence activity and a computational cost less than the dynamic models. The Successive Inverse Polynomial Interpolation (Geurts \& Meyers \textit{Phys.\ Fluids} \textbf{18}, 2006) was used to calculate the model parameter. Calculating implicitly the eddy-viscosity at each time-step removes the numerical instabilities found in previous studies, while maintaining the local character of the model. Results of simulations of channel flow at $Re_\tau$ up to 2,000, and forced homogeneous isotropic turbulence will be presented. [Preview Abstract] |
Tuesday, November 26, 2013 8:13AM - 8:26AM |
M4.00002: Grid-independent large-eddy simulation (LES) of turbulent flow around a circular cylinder using explicit filtering Satbir Singh, Donghyun You The explicit filtering technique has the potential to provide grid-independent and error-quantified large-eddy-simulation (LES) solutions. Bose {\it et al.} [Phys. Fluids {\bf 22}, 105103 (2010)] and Singh {\it et al.} [Phys. Fluids {\bf 24}, 085105 (2012)] recently obtained grid-independent LES solutions for turbulent channel flow using one-dimensional discrete filter functions implemented on Cartesian grids. Many complex flow configurations, however, employ arbitrary shape grids, for which it is difficult to design such discrete filter functions. In the present work, we employ an elliptic differential filter to solve explicit-filter LES equations on arbitrary shaped grids. The coefficients of the elliptic filter are determined by comparing its filtering characteristics with those of a Gausian filter. The elliptic filter is applied to a homogeneous isotropic turbulence flow field and the coefficient is adjusted until a filtered energy spectra similar to that of the Gaussian filter is obtained. The filter coefficients thus obtained are then employed to solve explicit-filter LES equations for turbulent channel flow at $Re_{\tau}=395$ and turbulent flow over a circular cylinder at $Re_D = 3900$. Grid-independent solutions are obtained for both flow configurations. [Preview Abstract] |
Tuesday, November 26, 2013 8:26AM - 8:39AM |
M4.00003: LES of three-dimensional, shear-driven turbulent wall flow at $Re_\tau\approx2000$ using a nested-LES wall-modeling approach Yifeng Tang, Rayhaneh Akhavan Accurate prediction of high Reynolds number non-equilibrium wall flows presents a major challenge for traditional LES and wall modelling approaches such as hybrid RANS/LES or analytical wall functions. In this study, we investigate the applicability of the nested-LES wall-modeling approach (Tang $\&$ Akhavan 2012) to non-equilibrium flow in a 3D turbulent channel at $Re_\tau\approx2000$. The three-dimensionality was introduced by imposing an impulsive spanwise motion of the walls in an initially 2D equilibrium turbulent channel flow and suddenly stopping the spanwise motion after the turbulence had adjusted to the wall motion. The progression of turbulence statistics in both the sheared and recovery stages was in good agreement with experiments in 3D, shear-driven boundary layers at comparable Reynolds number (Driver $\&$ Hebbar 1987). The nested-LES wall-modeling approach couples coarse-grained LES in a full-size domain ($L_x=4\pi h, L_y=2\pi h$) with nested fine-grained LES in a minimal domain ($L_x^+=3200, L_y^+=1600$), both using $64^3$ grids. At each iteration, the velocity fields in both domains are renormalized to match the $U_i$ and $u_{i,{rms}}$ to those of the minimal domain in the near-wall region and the full-size domain in the outer region, respectively. [Preview Abstract] |
Tuesday, November 26, 2013 8:39AM - 8:52AM |
M4.00004: Turbulence Shell Models for Initial and Inflow Conditions in Direct and Large-Eddy Simulations Tomasz Drozda, Jeffery White, Robert Rubinstein Initial and inflow conditions continue to present a challenge for simulations of turbulent flows via Direct and Large-Eddy Simulations (DNS and LES). The current work utilizes the output of a Sabra~[1] shell model of turbulence to synthesize a three-dimensional (3D) homogeneous, isotropic, incompressible, turbulence-in-a-box velocity field. This approach is motivated by recent work of~van~de~Water~et~al.~[2] on generation of wind tunnel turbulence with active grids. The properties of the new synthetic turbulence are assessed for several values of the Reynolds number by computing higher order statistics. DNS of the decay of homogeneous isotropic turbulence are also considered with initial conditions obtained using both the new method and Gaussian turbulence. \\[4pt] [1] L'vov, V., Podivilov, E., Pomyalov, A., Procaccia, I., and Vandembroucq, D., Improved shell model of turbulence, \emph{Phys. Rev. E.}, \textbf{58}:1811--1822, 1998.\\[0pt] [2] van~de Water, W., Cekli, H.E., and Joosten, R., Stirring turbulence with turbulence, in \emph{Bulletin of the American Physical Society}, the American Physical Society, Baltimore, MD, 2011. [Preview Abstract] |
Tuesday, November 26, 2013 8:52AM - 9:05AM |
M4.00005: Subgrid model evaluation through lockstep DNS/LES of a turbulent jet Ankit Bhagatwala, Venkat Raman, Jacqueline Chen The aim of this study is to analyze the validity of the common Smagorinsky type closure assumptions employed in LES scalar mixing and scalar dissipation rate models. This is done using a unique DNS-LES lockstep methodology, wherein a DNS is run simultaneously with several LES instances. The LES only solves for the scalar fields and obtains the velocity fields directly from the filtered DNS solution at every substep of time. The LES is also solved on the same grid as the DNS. This eliminates two primary sources of error in LES, numerical error associated with a coarser grid and modeling error arising from the modeled velocity field. The only source of error then, is from the closure assumption made for the LES model. One instance of DNS and three LES instances of a 3D turbulent slot jet at a Reynolds number of 7500 are simulated. The three LES simulations correspond to three different filter widths. Predictions of resolved and subgrid contributions of scalar second moment, scalar variance and scalar dissipation rate are compared. Implications for turbulent combustion models that heavily rely on these parameters are discussed. [Preview Abstract] |
Tuesday, November 26, 2013 9:05AM - 9:18AM |
M4.00006: Experimental study of the SGS pressure-strain-rate correlation in the convective atmospheric surface layer Khuong Nguyen, Chenning Tong The subgrid-scale (SGS) stress and flux are studied using measurement data obtained in the atmospheric surface layer during the Advection Horizontal Array Turbulence Study (AHATS) field program, which notably includes measurement of the resolvable- and subgrid-scale pressure. We analyze the terms in the transport equations of the SGS stress and SGS heat flux, conditioned on the resolvable-scale velocity, for different filter scales and atmospheric stability. The results show that the pressure destruction terms in the budgets of the SGS shear stress and the SGS heat flux play the usual role of return-to-isotropy and generally counter the trends of the conditional production for all filter scales and unstable conditions. In contrast, the pressure-strain-rate correlations in the budgets of the normal SGS stress components can be the main cause of anisotropy of the SGS stress under convective conditions, depending strongly on the resolvable-scale velocity. These effects are most significant at large filter scales and have strong implications for modeling the near-wall SGS pressure-strain-rate correlation. [Preview Abstract] |
Tuesday, November 26, 2013 9:18AM - 9:31AM |
M4.00007: Higher-order moments and their modeling approximations in turbulent channel flow Elbert Jeyapaul, Gary Coleman Third- and fourth-order moments and the terms in their budgets are evaluated using results from Direct Numerical Simulations (DNS) of turbulent channel flow at $Re_{\tau}=395$, to aid in the development of higher-order Reynolds-Averaged Navier Stokes (RANS) closure models. These models have been proposed as a means of obtaining more accurate predictions of complex flows. The DNS data is used to test the assumptions that have been made to model the turbulent diffusion, velocity-pressure gradient and dissipation terms in the higher-order transport equations. The validity of using the Gram-Charlier Probability Density Function (PDF) to extrapolate the fourth-order moments from the lower-order ones is examined, as is the Millionshchikov hypothesis of quasi-normal distributions of the fourth-order moments. The wall-correction-free velocity-pressure gradient model of Poroseva (2001) is assessed, along with the assumption for wall-bounded flows of zero-dissipation in the third- and fourth-order equations. [Preview Abstract] |
Tuesday, November 26, 2013 9:31AM - 9:44AM |
M4.00008: Stochastic model representation of the energy transfers in turbulent channel flow Vassili Kitsios, Juan A. Sillero, Julio Soria, Jorgen S. Frederiksen A stochastic model is used to represent the energy transfers in the direct numerical simulation (DNS) of turbulent channel flow. The DNS has a Fourier discretisation in the streamwise and spanwise directions, and a Chebyshev discretisation in the wall normal direction. We spectrally decompose the DNS into large and small horizontal scales, and develop a stochastic subgrid model representing the effect the removal of the small scales has on the large. The stochastic model consists of a deterministic drain dissipation acting on the resolved field and a stochastic backscatter force. Positive values of the drain operator indicate energy sent from the large to the small scales (dissipation), whilst negative values represent energy sent from the small to the large (deterministic backscatter). The variance of the stochastic force quantifies the extent to which the backscatter is random as opposed to deterministic. We are also able to produce large eddy simulations using this stochastic subgrid model that reproduces the time averaged kinetic energy spectra of the DNS within the resolved scales. Results are presented for various Reynolds numbers up to $Re_\tau=950$. [Preview Abstract] |
Tuesday, November 26, 2013 9:44AM - 9:57AM |
M4.00009: Formulation of a $k-\omega$ based DDES model Karthik Rudra Reddy, Paul Durbin DES models fall under the category of Hybrid RANS/LES models, and they employ RANS to resolve near-wall boundary layers in the flow domain and LES away from the surface. The idea of DES is applicable to any RANS model, and various versions of the method reflect this. A general DDES formulation was put forth (Spalart et al, 2006) suitable for use with any RANS model, and later adapted for use with the $k-\omega$ SST model (Gritskevich et al, 2012). The current work develops a variant based on the $k-\omega$ model. In this version, length scales enter directly in the subgrid eddy-viscosity, rather than being used in the dissipation term of the k-equation; indeed, one can approach it as replacing length-scale clipping, in standard DDES, with velocity scale clipping. The length scales were modified in order to account for the log-layer mismatch, a well-known issue with DDES. Simulation results from channel flow and flow over a backward-facing step are presented. [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. |
© 2022 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
1 Research Road, Ridge, NY 11961-2701
(631) 591-4000
Office of Public Affairs
529 14th St NW, Suite 1050, Washington, D.C. 20045-2001
(202) 662-8700