Bulletin of the American Physical Society
69th Annual Meeting of the APS Division of Fluid Dynamics
Volume 61, Number 20
Sunday–Tuesday, November 20–22, 2016; Portland, Oregon
Session G29: CFD: Large-eddy Simulation |
Hide Abstracts |
Chair: Karthik Duraisamy, University of Michigan Room: F150 |
Monday, November 21, 2016 8:00AM - 8:13AM |
G29.00001: Statistical Mechanics-based Closures for Large Eddy Simulations Eric Parish, Karthik Duraisamy, Ayoub Gouasmi The simulation of high Reynolds-number fluid flows is made challenging by the presence of an enormous range of temporal and spatial scales. The Mori-Zwanzig (MZ) formalism originates from non-equilibrium statistical mechanics and provides a formal backdrop for the construction of coarse-grained models. In this work, a class of models inspired from the Mori-Zwanzig formalism are applied to turbulent flows. The MZ-models are derived directly from the governing equations and require minimal heuristics. The resulting closures are non-Markovian and are akin to modeling the divergence of the sub-grid stress. Non-local temporal effects are captured through a finite memory approximation of the MZ memory kernel. Numerical simulations of rotating homogeneous turbulence and turbulent channel flow are presented. The MZ-based models are shown to accurately characterize the behavior of the unresolved dynamics associated energy transfer mechanisms. [Preview Abstract] |
Monday, November 21, 2016 8:13AM - 8:26AM |
G29.00002: Unphysical scalar excursions in large-eddy simulations Georgios Matheou, Paul Dimotakis The range of physically realizable values of passive scalar fields in any flow is bounded by their boundary values. The current investigation focuses on the local conservation of passive scalar concentration fields in turbulent flows and the ability of the large-eddy simulation (LES) method to observe the boundedness of passive scalar concentrations. In practice, as a result of numerical artifacts, this fundamental constraint is often violated with scalars exhibiting unphysical excursions. The present study characterizes passive-scalar excursions in LES of a turbulent shear flow and examines methods for error diagnosis. Typically, scalar-excursion errors are diagnosed as violations of global boundedness, i.e., detecting scalar-concentration values outside boundary/initial condition bounds. To quantify errors in mixed-fluid regions, a local scalar excursion error metric is defined with respect to the local non-diffusive limit. Analysis of such errors shows that unphysical scalar excursions in LES result from dispersive errors of the convection-term discretization where the subgrid-scale model (SGS) provides insufficient dissipation to produce a sufficiently smooth scalar field. Local scalar excursion errors are found not to be correlated with the local scalar-gradient magnitude. [Preview Abstract] |
Monday, November 21, 2016 8:26AM - 8:39AM |
G29.00003: Wall-resolved large-eddy simulation of flow past a circular cylinder W. Cheng, D.I. Pullin, R. Samtaney Wall-resolved large-eddy simulations (LES) about a smooth-walled circular cylinder are described over a range of Reynolds number from $Re_D = 3.9\times 10^3$ (subcritical) to above the drag crisis, $Re_D=8.5\times 10^5$ (supercritical), where $D$ is the cylinder diameter. The span-wise domain is $3\,D$ for $Re_D \le 10^5$ and $D$ otherwise. The numerical method is a fourth-order finite-difference discretization on a standard curvilinear O-grid. The stretched-vortex sub-grid scale model is used in the whole domain, including regions of large-scale separated flow. For $Re_D \le10^5$, calculations of the skin-friction coefficient versus polar angle $\theta$ along the cylinder surface and its dependence on $Re_D$ are well captured in comparison with experimental data. Proper separation behavior is observed. For high $Re_D$, a fine mesh $8192\times 1024\times 256$ is used. It is found that a blowing/suction-type perturbation of the wall-normal velocity along a span-wise strip, with angular position at $\theta =50-60^o$, is then required in order to produce flow separation in accordance with experiment at Reynolds numbers in the drag-crisis regime. Results presented will focus on the skin-friction behavior and details of flow separation. [Preview Abstract] |
Monday, November 21, 2016 8:39AM - 8:52AM |
G29.00004: Entropy-viscosity based LES of turbulent flow in a flexible pipe Zhicheng Wang, Fangfang Xie, Michael Triantafyllou, Yiannis Constantinides, George Karniadakis We present large-eddy simulations (LES) of turbulent flow in a flexible pipe conveying incompressible fluid. We are interested in quantifying the flow-structure interaction in terms of mean quantities and their variances. For the LES, we employ an Entropy Viscosity Method (EVM), implemented in a spectral element code. In previous work, we investigated laminar flow and studied the complex interaction between structural and internal flow dynamics and obtained a phase diagram of the transition between states as function of three non-dimensional quantities: the fluid-tension parameter, the dimensionless fluid velocity, and the Reynolds number. Here we extend our studies in the turbulence regime, Re from 5,000 to 50,000. The motion of the flexible pipe affects greatly the turbulence statistics of the pipe flow, with substantial differences for free (self-sustained) vibrations and prescribed (forced) vibrations. [Preview Abstract] |
Monday, November 21, 2016 8:52AM - 9:05AM |
G29.00005: A Stabilized Scale-Similarity Model for Explicitly-Filtered LES Ayaboe Edoh, Ann Karagozian, Venkateswaran Sankaran Accurate simulation of the filtered-scales in LES is affected by the competing presence of modeling and discretization errors\footnote{Ghosal, \textbf{J. Comp. Phys.}, 125, 187-206 (1996)}. In order to properly assess modeling techniques, it is imperative to minimize the influence of the numerical scheme. The current investigation considers the inclusion of resolved and un-resolved sub-filter stress ([U]RSFS) components in the governing equations\footnote{Carati \emph{et al.}, \textbf{J. Fluid. Mech.}, 441, 119-138 (2001)}, which is suggestive of a mixed-model approach. Taylor-series expansions of discrete filter stencils are used to inform proper scaling of a Scale-Similarity model representation of the RSFS term\footnote{Pruett \emph{et al.}, \textbf{Phys. of Fluids}, 13, 2578-2589 (2001)}, and accompanying stabilization is provided by tunable and scale-discriminant filter-based artificial dissipation techniques\footnote{Edoh \emph{et al.}, AIAA 2016-3794} that represent the URSFS term implicitly. Effective removal of numerical error from the LES solution is studied with respect to the 1D Burgers equation with synthetic turbulence, and extension to 3D Navier-Stokes system computations is motivated. [Preview Abstract] |
Monday, November 21, 2016 9:05AM - 9:18AM |
G29.00006: LES of a submarine model in self-propulsion Antonio Posa, Elias Balaras Wall-resolved LES computations are presented on the flow over a notional submarine geometry in self-propelled conditions at Re=1.2e+06 (based on the free-stream velocity and the length of the body). The rotational speed of the propeller was dynamically adjusted during the simulation using a proportional-integral controller, in order for the propeller thrust to balance the overall drag on the system. An immersed-boundary methodology was adopted to enforce boundary conditions over the body. Comparisons with the same submarine geometry in the towed configuration allowed us to verify that the boundary layer over the hull surface is affected only in the stern region, while that over the cylindrical mid-body is roughly in equilibrium in both towed and self-propelled conditions. The quasi-streamwise structures of the turbulent boundary layer and their development along the stern dominate the flow ingested by the propeller. The wake and junction flows from the fins break the axial symmetry of such flow. Comparisons with the towed configuration verify that the propeller affects substantially its own inflow, defined by the overlapping effects of adverse pressure gradient, due to the tapered geometry of the stern, and suction generated by the same propeller. [Preview Abstract] |
Monday, November 21, 2016 9:18AM - 9:31AM |
G29.00007: An investigation of slip wall boundary condition for wall-modeled large-eddy simulation Hyunji Jane Bae, Adrian Lozano-Duran, Sanjeeb Bose, Parviz Moin Wall models for large-eddy simulation are necessary to overcome the resolution requirements near the wall for high Reynolds number turbulent flows. In the present study, the slip wall boundary condition is examined (Bose and Moin, Phys. Fluids, 2014). The optimal slip length and its dependence on Reynolds number, grid size, subgrid scale model, etc. is investigated in turbulent channel flows up to $Re_\tau = 4200$. Two families of slip wall models are introduced. The first is derived from the Navier Stokes equations, and the second is based on error minimization in two different filter levels. These new and existing models are tested and compared to the optimal slip length and the filtered direct numerical simulation results. [Preview Abstract] |
Monday, November 21, 2016 9:31AM - 9:44AM |
G29.00008: Further reduction of near-wall resolution for wall-modeled LES Alexandre Marques, Qiqi Wang, Johan Larsson, Gregory Laskowski, Sanjeeb Bose One of the greatest challenges to the use of Large Eddy Simulations (LES) in engineering applications is the large number of grid points required near walls. To mitigate this issue, LES is often coupled with a model of the flow close to the wall, known as wall model. One feature common to most wall models is that the first few (about 3) grid points must be located below the inviscid log-layer ($y/\delta \le 0.2$), and the grid must have near isotropic resolution near the wall. Hence, wall-modeled LES may still require a large number of grid points, both in the wall-normal and span-wise directions. Because of these requirements, wall-modeled LES still is unfeasible in many applications. We present a new formulation of wall-modeled LES that is being developed to address this issue. In this formulation, LES is used to solve only for the features of the velocity field that can be adequately represented on the LES grid. The effects of the unresolved features are captured by imposing a balance of momentum integrated in the wall-normal direction. This integral momentum balance translates into a dynamic PDE defined on the walls, which is coupled to the LES equations. We discuss details of the new formulation and present results obtained in laminar and turbulent channel flows. [Preview Abstract] |
Monday, November 21, 2016 9:44AM - 9:57AM |
G29.00009: A dummy cell immersed boundary method for incompressible turbulence simulations over dirty geometries Keiji Onishi, Makoto Tsubokura A methodology to eliminate the manual work required for correcting the surface imperfections of computer-aided-design (CAD) data, will be proposed. Such a technique is indispensable for CFD analysis of industrial applications involving complex geometries. The CAD geometry is degenerated into cell-oriented values based on Cartesian grid. This enables the parallel pre-processing as well as the ability to handle 'dirty' CAD data that has gaps, overlaps, or sharp edges without necessitating any fixes. An arbitrary boundary representation is used with a dummy-cell technique based on immersed boundary (IB) method. To model the IB, a forcing term is directly imposed at arbitrary ghost cells by linear interpolation of the momentum. The mass conservation is satisfied in the approximate domain that covers fluid region except the wall including cells. Attempts to Satisfy mass conservation in the wall containing cells leads to pressure oscillations near the IB. The consequence of this approximation will be discussed through fundamental study of an LES based channel flow simulation, and high Reynolds number flow around a sphere. And, an analysis comparing our results with wind tunnel experiments of flow around a full-vehicle geometry will also be presented. [Preview Abstract] |
Monday, November 21, 2016 9:57AM - 10:10AM |
G29.00010: Wall-layer model for LES with massive separation Ahmad Fakhari, Vincenzo Armenio, Federico Roman Currently, Wall Functions (WF) work well under specific conditions, mostly exhibit drawbacks specially in flows with separation beyond curvatures. In this work, we propose a more general WF which works well in attached and detached flows, in presence and absence of Immersed Boundaries (IB). First we modified an equilibrium stress WF for boundary-fitted geometry making dynamic the computation of the k (von Karman constant) of the log-law; the model was first applied to a periodic open channel flow, and then to the flow over a 2D single hill using uniform coarse grids; the model captured separation with reasonable accuracy. Thereafter IB Method by Roman et al. (Phys. Fluids, 2009) was improved to avoid momentum loss at the interface between the fluid-solid regions. This required calibration of interfacial eddy viscosity; also a random stochastic forcing was used in wall-normal direction to increase Reynolds stresses and improve mean velocity profile. Finally, to reproduce flow separation, a simplified boundary layer equation was applied to construct velocity at near wall computational nodes. The new scheme was tested on the 2D single hill and periodic hills applying Cartesian and curvilinear grids; good agreement with references was obtained with reduction in cost and complexity. [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