Bulletin of the American Physical Society
70th Annual Meeting of the APS Division of Fluid Dynamics
Volume 62, Number 14
Sunday–Tuesday, November 19–21, 2017; Denver, Colorado
Session L31: Computational Fluid Dynamics: Large Eddy SimulationCFD
|
Hide Abstracts |
Chair: J. M. McDonough, University of Kentucky Room: 108 |
Monday, November 20, 2017 4:05PM - 4:18PM |
L31.00001: Energy Cascade Analysis: from Subscale Eddies to Mean Flow Mohamad Ibrahim Cheikh, Louis Wonnell, James Chen Understanding the energy transfer between eddies and mean flow can provide insights into the energy cascade process. Much work has been done to investigate the energy cascade at the level of the smallest eddies using different numerical techniques derived from the Navier-Stokes equations. These methodologies, however, prove to be computationally inefficient when producing energy spectra for a wide range of length scales. In this regard, Morphing Continuum Theory (MCT) resolves the length-scales issues by assuming the fluid continuum to be composed of inner structures that play the role of subscale eddies. The current study show- cases the capabilities of MCT in capturing the dynamics of energy cascade at the level of subscale eddies, through a supersonic turbulent flow of Mach 2.93 over an 8◦ compression ramp. Analysis of the results using statisti- cal averaging procedure shows the existence of a statistical coupling of the internal and translational kinetic energy fluctuations with the corresponding rotational kinetic energy of the subscale eddies, indicating a multiscale transfer of energy. The results show that MCT gives a new characteri- zation of the energy cascade within compressible turbulence without the use of excessive computational resources. [Preview Abstract] |
Monday, November 20, 2017 4:18PM - 4:31PM |
L31.00002: Very-high-Reynolds-number vortex dynamics via Coherent-vorticity-Preserving (CvP) Large-eddy simulations Jean-Baptiste Chapelier, Bono Wasistho, Carlo Scalo A new approach to Large-Eddy Simulation (LES) is introduced, where subgrid-scale (SGS) dissipation is applied proportionally to the degree of local spectral broadening, hence mitigated in regions dominated by large-scale vortical motion. The proposed CvP-LES methodology is based on the evaluation of the ratio of the test-filtered to resolved (or grid-filtered) enstrophy: $\sigma = \widehat{\overline{\xi}}/\overline{\xi}$. Values of $\sigma \simeq 1$ indicate low sub-test-filter turbulent activity, justifying local deactivation of any subgrid-scale model. Values of $\sigma < 1$ span conditions ranging from incipient spectral broadening $\sigma \la 1$, to equilibrium turbulence $\sigma = \sigma_{eq} < 1$, where $\sigma_{eq}$ is solely as a function of the test-to-grid filter-width ratio $\widehat{\overline{\Delta}}/\overline{\Delta}$, derived assuming a Kolmogorov's spectrum. Eddy viscosity is fully restored for $\sigma \le \sigma_{eq}$. The proposed approach removes unnecessary SGS dissipation, can be applied to any eddy-viscosity model, is algorithmically simple and computationally inexpensive. A CvP-LES of a pair of unstable helical vortices, representative of rotor-blade wake dynamics, show the ability of the method to sort the coherent motion from the small-scale dynamics [Preview Abstract] |
Monday, November 20, 2017 4:31PM - 4:44PM |
L31.00003: Autonomic Closure for Turbulent Flows Using Approximate Bayesian Computation Olga Doronina, Jason Christopher, Peter Hamlington, Werner Dahm Autonomic closure is a new technique for achieving fully adaptive and physically accurate closure of coarse-grained turbulent flow governing equations, such as those solved in large eddy simulations (LES). Although autonomic closure has been shown in recent a priori tests to more accurately represent unclosed terms than do dynamic versions of traditional LES models, the computational cost of the approach makes it challenging to implement for simulations of practical turbulent flows at realistically high Reynolds numbers. The optimization step used in the approach introduces large matrices that must be inverted and is highly memory intensive. In order to reduce memory requirements, here we propose to use approximate Bayesian computation (ABC) in place of the optimization step, thereby yielding a computationally-efficient implementation of autonomic closure that trades memory-intensive for processor-intensive computations. The latter challenge can be overcome as co-processors such as general purpose graphical processing units become increasingly available on current generation petascale and exascale supercomputers. In this work, we outline the formulation of ABC-enabled autonomic closure and present initial results demonstrating the accuracy and computational cost of the approach. [Preview Abstract] |
Monday, November 20, 2017 4:44PM - 4:57PM |
L31.00004: Sub-grid scale models for discontinuous Galerkin methods based on the Mori-Zwanzig formalism Eric Parish, Karthk Duraisamy The optimal prediction framework of Chorin et. al., which is a reformulation of the Mori-Zwanzig (M-Z) formalism of non-equilibrium statistical mechanics, provides a framework for the development of mathematically-derived closure models. The M-Z formalism provides a methodology to reformulate a high-dimensional Markovian dynamical system as a lower-dimensional, non-Markovian (non-local) system. In this lower-dimensional system, the effects of the unresolved scales on the resolved scales are non-local and appear as a convolution integral. The non-Markovian system is an exact statement of the original dynamics and is used as a starting point for model development. In this work, we investigate the development of M-Z-based closures model within the context of the Variational Multiscale Method (VMS). The method relies on a decomposition of the solution space into two orthogonal subspaces. The impact of the unresolved subspace on the resolved subspace is shown to be non-local in time and is modeled through the M-Z-formalism. The models are applied to hierarchical discontinuous Galerkin discretizations. Commonalities between the M-Z closures and conventional flux schemes are explored. [Preview Abstract] |
Monday, November 20, 2017 4:57PM - 5:10PM |
L31.00005: Structure-Preserving Variational Multiscale Modeling of Turbulent Incompressible Flow with Subgrid Vortices John Evans, Christopher Coley, Ryan Aronson, Corey Nelson In this talk, a large eddy simulation methodology for turbulent incompressible flow will be presented which combines the best features of divergence-conforming discretizations and the residual-based variational multiscale approach to large eddy simulation. In this method, the resolved motion is represented using a divergence-conforming discretization, that is, a discretization that preserves the incompressibility constraint in a pointwise manner, and the unresolved fluid motion is explicitly modeled by subgrid vortices that lie within individual grid cells. The evolution of the subgrid vortices is governed by dynamical model equations driven by the residual of the resolved motion. Consequently, the subgrid vortices appropriately vanish for laminar flow and fully resolved turbulent flow. As the resolved velocity field and subgrid vortices are both divergence-free, the methodology conserves mass in a pointwise sense and admits discrete balance laws for energy, enstrophy, and helicity. Numerical results demonstrate the methodology yields improved results versus state-of-the-art eddy viscosity models in the context of transitional, wall-bounded, and rotational flow when a divergence-conforming B-spline discretization is utilized to represent the resolved motion. [Preview Abstract] |
Monday, November 20, 2017 5:10PM - 5:23PM |
L31.00006: Isolating Numerical Error Effects in LES Using DNS-Derived Sub-Grid Closures Ayaboe Edoh, Ann Karagozian The prospect of employing an explicitly-defined filter in Large-Eddy Simulations (LES) provides the opportunity to reduce the interaction of numerical/modeling errors and offers the chance to carry out grid-converged assessments\footnote{Radakrishnan and Bellan, \textbf{J. Fluid Mech.}, 697, 399-435 (2012)}, important for model development. By utilizing a quasi \emph{a priori} evaluation method -- wherein the LES is assisted by closures derived from a fully-resolved computation\footnote{De Stefano and Vasilyev, \textbf{Theoret. Comput. Fluid Dyn.}, 18, 27-41 (2004)} -- it then becomes possible to understand the combined impacts of filter construction (e.g., filter width, spectral sharpness) and discretization choice on the solution accuracy. The present work looks at calculations of the compressible LES Navier-Stokes system and considers discrete filtering formulations in conjunction with high-order finite differencing schemes. Accuracy of the overall method construction is compared to a consistently-filtered exact solution, and lessons are extended to \emph{a posteriori} (i.e., non-assisted) evaluations\footnote{Edoh \emph{et al.}, AIAA Paper 2017-3952 (2017)}. [Preview Abstract] |
Monday, November 20, 2017 5:23PM - 5:36PM |
L31.00007: Effects of numerical dissipation and unphysical excursions on scalar-mixing estimates in large-eddy simulations Nek Sharan, Georgios Matheou, Paul Dimotakis Artificial numerical dissipation decreases dispersive oscillations and can play a key role in mitigating unphysical scalar excursions in large eddy simulations (LES). Its influence on scalar mixing can be assessed through the resolved-scale scalar, $\bar{Z}$, its probability density function (PDF), variance, spectra, and the budget of the horizontally averaged equation for $\bar{Z}^{2}$. LES of incompressible temporally evolving shear flow enabled us to study the influence of numerical dissipation on unphysical scalar excursions and mixing estimates. Flows with different mixing behavior, with both marching and non-marching scalar PDFs, are studied. Scalar fields for each flow are compared for different grid resolutions and numerical scalar-convection term schemes. As expected, increasing numerical dissipation enhances scalar mixing in the development stage of shear flow characterized by organized large-scale pairings with a non-marching PDF, but has little influence in the self-similar stage of flows with marching PDFs. Flow parameters and regimes sensitive to numerical dissipation help identify approaches to mitigate unphysical excursions while minimizing dissipation. [Preview Abstract] |
Monday, November 20, 2017 5:36PM - 5:49PM |
L31.00008: Grid-adaptation for chaotic multi-scale simulations as a verification-driven inverse problem Johan Larsson The grid-spacing has a direct effect on both the numerical and the modeling errors in coarse-grained simulations of multi-scale problems (e.g., large eddy simulations of turbulence). We make the argument that it is impossible to estimate where errors are introduced in such simulations with absolute certainty, and thus that the problem of finding an ``optimal'' adapted grid must be framed as a verification-driven problem. After posing this new problem, one possible general approach to solving it is proposed. This is then tested and demonstrated on a modified Kuramoto-Sivashinsky equation. [Preview Abstract] |
Monday, November 20, 2017 5:49PM - 6:02PM |
L31.00009: Filter-induced bifurcations of Navier--Stokes solutions J. M. McDonough, Ali Naghi Ziaei, Neda Sheikhrezazadeh Nikou We study 3-D lid-driven cavity (LDC) flows at both laminar and turbulent Reynolds numbers ($Re$), employing LES as the solution technique, and demonstrate that application of filters (both explicit and implicit) can lead to computed solutions exhibiting features that are not even qualitatively correct, when compared with DNS and experiment, as a result of bifurcations associated with changing filter parameter values. We carry this out for two specific values of $Re$: $1000$ for laminar (steady) flows and $10000$ for turbulent flows, using a wide range of numerical grid spacings (both uniform and nonuniform) to permit assessment of accuracy and aliasing effects. No geometric smoothing is employed in corners of the impulsively-started, driven lid, leading to non-classical solutions to the equations of motion, and the need for mollification (filtering). Power-spectral analysis of time series is used to identify qualitative behaviors, and it is found that on coarse grids any of steady, periodic, quasiperiodic and chaotic states can be achieved by changing filter-parameter values such as the Smagorinsky constant and the Shuman filter parameter. [Preview Abstract] |
Monday, November 20, 2017 6:02PM - 6:15PM |
L31.00010: A Sensitivity Study of the Navier-Stokes-$\alpha$ Model Sean Breckling, Monika Neda We present a sensitivity study of the of the Navier Stokes-$\alpha$ model (NS$\alpha$) with respect to perturbations of the differential filter length $\alpha$. Parameter-sensitivity is evaluated using the sensitivity equations method. Once formulated, the sensitivity equations are discretized and computed alongside the NS$\alpha$ model using the same finite elements in space, and Crank-Nicolson in time. We provide a complete stability analysis of the scheme, along with the sensitivity results of several benchmark problems in both 2D and 3D. We further demonstrate a practical technique to determine the reliability of the NS$\alpha$ model in problem-specific settings. Lastly, we investigate the sensitivity and reliability of important functionals of the velocity and pressure solutions. [Preview Abstract] |
Monday, November 20, 2017 6:15PM - 6:28PM |
L31.00011: LES of stratified turbulent wake with temperature and salinity dependent density stratification Reetesh Ranjan, Suresh Menon Large eddy simulation (LES) based investigation of doubly-diffusive stratified turbulent wake flow is performed in this study. Such flows are observed in underwater naval applications, where the density stratification usually depends upon two scalar fields, namely, temperature and salinity through a linear/nonlinear equation of state. A disparity in the range of scales associated with these scalars and differential diffusion effects leads to exorbitant computational cost, thus making LES a viable alternative. In this study, LES is performed by employing the dynamic one-equation based eddy viscosity model, which has been extended for stratified flows. The extended LES formulation is first assessed for its predictive capabilities by comparing with the reference direct numerical simulation results. Afterward, the effects of doubly-diffusive turbulence on the temporal evolution of the wake structure are examined by comparing two cases with stable/unstable stratification of temperature/salinity fields, while still maintaining an overall stable density stratification, where the density is determined by a linear equation of state. The results from these cases are compared with reference results employing density stratification dependent upon only one scalar field. [Preview Abstract] |
Monday, November 20, 2017 6:28PM - 6:41PM |
L31.00012: Effect of inflow condition on near-field prediction of Large Eddy Simulations of isothermal and non-isothermal turbulent jets Sasan Salkhordeh, Mark Kimber In order to develop an experimentally validated computational model, turbulent round jets have been studied extensively under both isothermal and non-isothermal conditions using Large Eddy Simulation (LES) methodology. Capturing the near-field physics of a turbulent jet has been a challenge when utilizing LES. To address this concern, the effect of inlet flow profile and turbulent fluctuations on the evolution of both type of jets has been analyzed in detail by performing separate large eddy simulations of the flow in the nozzle upstream of the jet inlet to accurately determine the inlet turbulent spectra. From the precursor simulations, the accurate turbulence fluctuations at the jet nozzle can be sampled and then implement to the inlet boundary of the main jet simulation. Properly specifying the turbulent fluctuations at the jet inlet was found to play a vital role in order to accurately predict key characteristics throughout the computational domain. For isothermal jets, the experimental measurements of Hussein et al. (Journal of Fluid Mechanics. 1994 Jan;258:31-75) has been simulated computationally using LES. The experimental measurement of Mi et al. (Journal of Fluid Mechanics. 2001 Apr;432:91-125) has been chosen for performing LES for a non-isothermal jet at the same Reynolds number and identical temperature difference. The LES results show good agreement for first and higher order statistics of velocities and temperatures in both near field and far-field data. [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. |
© 2025 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