Bulletin of the American Physical Society
75th Annual Meeting of the Division of Fluid Dynamics
Volume 67, Number 19
Sunday–Tuesday, November 20–22, 2022; Indiana Convention Center, Indianapolis, Indiana.
Session J22: Turbulence: Modeling |
Hide Abstracts |
Chair: Diego Donzis, Texas A&M University Room: 208 |
Sunday, November 20, 2022 4:35PM - 4:48PM |
J22.00001: Selected Eddy Simulation (SES): A non-DNS approach with Navier-Stokes dynamics at all scales Shilpa Sajeev, Diego A Donzis Direct Numerical Simulations (DNS) resolve all dynamically relevant modes of turbulent flows and have, thus, become indispensable to fundamental turbulence research despite its high computational cost at high Reynolds numbers. To reach realistic conditions, though, one still needs less expensive (and less accurate) methods like Large Eddy Simulation that resolves only the large scales and models the small scales. This presents a challenge in flows where small-scale physics is dominant, like reacting species. |
Sunday, November 20, 2022 4:48PM - 5:01PM Author not Attending |
J22.00002: Lagrangian Model for Passive Scalar Gradients in Turbulence Xiaolong Zhang, Maurizio Carbone, Andrew D Bragg The equation for the fluid velocity gradient along a Lagrangian trajectory can be derived from the Navier-Stokes equation, but it involves two terms that cannot be determined from the velocity gradient along the chosen trajectory, namely the pressure Hessian and the viscous term. Building on a number of previous works, a recent model has been derived that handles these terms using a multiscale, recent deformation of Gaussian fields (MRDGF) closure (Johnson & Meneveau, Phys. Rev. Fluids, 2017). This model has been shown to describe DNS data very accurately and it works for arbitrary Reynolds numbers, unlike previous models that were restricted to relatively low Reynolds numbers. Here we extend the model to the case of a passive scalar gradient, using the MRDGF closure to handle the scalar dissipation term, and also consider how the model can capture the effect of the scalar Schmidt number. Predictions from the model are compared with those from DNS, and extensions of the model to stably stratified turbulent flows will be discussed. |
Sunday, November 20, 2022 5:01PM - 5:14PM |
J22.00003: Constraining Subglacial Plume Integral Model Parameters Using Stereoscopic PIV and planar LIF Experiments Chris Lai, Muhammad Ahmad Mustafa Subglacial plume modeling is an important part in the simulations of melting glaciers, ice sheet dynamics, and ocean circulations in the poles. Because of the large disparity in spatio-temporal scales, full 3D modeling of subglacial plumes within an ocean circulation model is prohibitively costly. Instead, the plumes are incorporated into circulation models using the integral approach that is 1D and predicts the mean properties of a subglacial plume as it rises along a water-ice interface. With proper coupling between a plume integral model and a circulation model, realistic simulations of the freshwater flux from melted ice into ocean can be achieved. In this talk, we focus on testing and validating the integral model formulations and parameters of a single subglacial plume rising along a vertical ice face. We achieved this by analyzing detailed fluid velocity and passive tracer measurements of a laboratory subglacial plume. Specifically, we did stereoscopic particle image velocimetry (PIV) and planar laser-induced fluorescence (LIF) experiments at different plume cross-sections along the plume trajectory, and this allowed us to quantify the turbulent transport of momentum, mass, and heat close to the water-ice face. We also did matched experiments on a non-melting, solid wall to provide a baseline for comparison. |
Sunday, November 20, 2022 5:14PM - 5:27PM |
J22.00004: Surface Body-Force Model for Turbulent Flow over Rough Walls Giles J Brereton, Junlin Yuan, Mostafa Jouybari DNS can be used with immersed boundary methods at low Reynolds numbers to resolve almost exactly the flow over surfaces of any specified roughness, and determine roughness modeling quantities like the equivalent sandgrain roughness $k_s$. What is needed to implement this information in RANS closures for general engineering applications is a technique for reproducing the effect of roughness when computed on a smooth-wall finite-volume mesh, using computed/modeled information only in the wall-adjacent cell, without outer-flow assumptions or restrictions. In this talk, we present a surface body-force model in which the effect of roughness is modeled as a source term in the $x$-momentum equation in the wall-adjacent cell, as the quadratic drag model $f_x = - \alpha_t \vert u_c \vert u_c$, together with the \textit{near-wall} function $u_c^+ = 1/\sqrt{\alpha_t\, \Delta y}$. Calibrations against reference pipe-flow data show that the roughness `wavenumber' $\alpha_t$ is a linear function of $k_s$ for small roughness, and depends on the cube of wall-cell height $\Delta y$. In preliminary computations within a $k$-$\epsilon$ closure, this roughness model appears to describe quite accurately the wall shear shear stress in flat-plate and adverse pressure-gradient boundary layers, provided the wall-cell height $\Delta y$ is sufficiently small. |
Sunday, November 20, 2022 5:27PM - 5:40PM |
J22.00005: Continuing Investigations of Nonlocality in Rayleigh-Taylor Instability Using the Macroscopic Forcing Method Dana Lynn Lavacot, Jessie Liu, Brandon E Morgan, Ali Mani In our past work, we used the Macroscopic Forcing Method (MFM) to assess the importance of higher-order spatio-temporal moments of the eddy diffusivity kernel for 2D Rayleigh-Taylor Instability (RTI) and concluded that nonlocality should not be neglected when constructing models for RTI. In the present work, we continue our investigation of nonlocality in RTI in two areas: 1) we apply a method for accelerated statistical convergence of MFM results; 2) we examine the importance of higher-order moments of eddy diffusivity in 3D RTI. Accelerated statistical convergence is achieved by reformulating the MFM equations such that propagation of statistical errors from one moment equation to another moment equation is substantially suppressed. New MFM receiver equations are presented, and a desirable level of statistical convergence is achieved in nearly one tenth of the number of simulations needed in the original method. This accelerated method allows us to obtain statistically-well-converged MFM measurements for 3D RTI with fewer high-fidelity simulations than in the original method, when MFM for 3D RTI would have otherwise been quite costly. We present MFM measurements of eddy diffusivity moments for 3D RTI using this accelerated method. |
Sunday, November 20, 2022 5:40PM - 5:53PM |
J22.00006: A systematic procedure for determining coefficients of RANS models Jessie Liu, Ali Mani In complex turbulent flows where direct numerical simulations may be computationally cost-prohibitive, Reynolds-averaged Navier-Stokes (RANS) modeling remains an attractive and widely used alternative. Often, RANS models contain many coefficients that are determined arbitrarily. We focus on a Reynolds stress transport modeling framework targeting prediction of wall-bounded turbulent flows and separated flows. We develop a systematic procedure for determining model coefficients starting from simple canonical cases with particular attention to data from recent literature and progressively building in complexity. These cases and are chosen to incrementally activate only a few coefficients at a time with emphasis on isolating anisotropy using recent data from literature. Both a priori and a posteriori testing is used to highlight model deficiencies. |
Sunday, November 20, 2022 5:53PM - 6:06PM |
J22.00007: Resolvent analysis of the two-dimensional Kolmogorov flow Anagha Madhusudanan, Rich R Kerswell We use the resolvent analysis framework to study the two-dimensional (2-D) Kolmogorov flow, i.e. the 2-D Navier Stokes equations forced by a sinusoidal body force. In the resolvent analysis, the non-linear terms of the Navier-Stokes equations are considered to be an unknown forcing to the linear equations. The aim here is to construct a reduced-order model of the 2-D Kolmogorov flow. We therefore ask the question of what the best basis in the temporal direction is, a Fourier basis or a Wavelet basis, that would give the optimum reduced-order model of the flow? |
Sunday, November 20, 2022 6:06PM - 6:19PM |
J22.00008: Evaluation of Variable Parameter Partially Averaged Navier-Stokes Method with Non-Linear Eddy Viscosity Closure Sawan S Sinha, SAGAR SAROHA, Ajey V More We have recently augmented the partially averaged Navier-Stokes (PANS) methodology employing higher order eddy viscosity closures (both quadratic and cubic closures). Our evaluations of the augmented PANS methodology in flows past bluff bodies with heat transfer has yielded significant improvements in prediction of both hydrodynamic and heat-transfer statistics. However, so far, all these studies have used the sequential two stage PANS approach, wherein, an initial Reynolds-averaged Navier-Stokes (RANS) simulation is required to compute the appropriate value of filter control parameters. Besides the requirement of this preceding RANS simulation, the entire computational grid is simulated with static, pre-chosen value of the filter control parameters. Alternatively, the dynamic version of PANS attempts to use variable control parameters which are adaptive to the local physics of a flow field. However, so far, this version of PANS has been tested with conventional linear eddy viscosity closure only. In this work, we attempt to combine the advantages of dynamic PANS with quadratic and cubic eddy viscosity closures and extensively evaluate the performance of the resulting methodology. A detailed analysis will be presented on the interaction of the variable filter parameter and the evolution of non-linear turbulent stress field in the highly separated and three-dimensional flow past a sphere. |
Sunday, November 20, 2022 6:19PM - 6:32PM |
J22.00009: Modeling compressed turbulent plasma with rapid viscosity variations Sébastien Thevenin, Nicolas Valade, Benoît-Joseph Gréa, Gilles Kluth, Olivier Soulard We propose two-equations models in order to capture the dynamics of a turbulent plasma undergoing compression and experiencing large viscosity variations. The models account for possible relaminarization phases and rapid viscosity changes through closures dependent on the turbulent Reynolds and on the viscosity Froude numbers. These closures are determined from a data-driven approach using eddy-damped quasi normal markovian simulations. The best model is able to mimic the various self-similar regimes identified in Viciconte et al. and to recover the rapid transition limits identified by Coleman and Mansour. |
Sunday, November 20, 2022 6:32PM - 6:45PM |
J22.00010: Representation of sub-grid scale turbulence with Periodic Box Homogeneous Direct Numerical Simulation Githin Tom Zachariah, Harry E Van den Akker Large Eddy Simulations (LESs) use Sub-Grid Scale (SGS) models to account for the effects of the unresolved scales of turbulence. The complex processes that occur in the small scales make the development of SGS models challenging. In this study, we propose the use of a Periodic Box (PB) Direct Numerical Simulation (DNS), simulating Homogeneous Isotropic Turbulence (HIT), to represent the unresolved scales of a LES. As a proof of concept, the local characteristic strain rate in the LES is used as a representation of the local turbulence. This is matched with the characteristic strain rate at the large scales of the PB DNS, thereby coupling them. |
Sunday, November 20, 2022 6:45PM - 6:58PM |
J22.00011: The Myth of URANS Daniel M israel In the 1990s, RANS modeling practitioner using finer grids and higher-order models began to observe cases for which simulations never reached a steady state. Instead, they observed the emergence of structures which were, at least qualitatively, similar to the turbulent structures observed in experiments (e.g. Johansson et. al, Int. J. Num. Meth. Fld. 1993; Durbin, AIAA J. 1995). These studies raise numerous questions. Fundementally, it is not clear why some RANS simulations result in unsteady solutions, and other do not. And the results of these studies have a mixed record of quantitiatve accuracy when compared to experiments. Spalart (Int. J. Heat Fld. Flow, 2000) present systematic critiques of what they refer to as the unsteady RANS (URANS) approach, notably, the lack of a theoretical justification. The current work addresses this lacuna with a theoretical model for both the partition between the resolved and modeled contributions to the kinetic energy, as well as for the bulk parameters of the turbulence. The principle predictions of the theory are that, in the absence of forcing, URANS tends to relax to a steady solution, and, depending on how the URANS is initialized, strongly nonphysical evolution can occur. |
Sunday, November 20, 2022 6:58PM - 7:11PM |
J22.00012: Composition of resolvents enhanced by random sweeping for large-scale structures in turbulent channel flows Guowei He, Ting Wu A composition of sweeping-enhanced resolvents, referred to as the R2-model, is proposed to predict the space-time statistics of large-scale structures in turbulent channel flows. This model incorporates two key mechanisms: (i) eddy damping is introduced to represent random-sweeping decorrelation caused by nonlinear forcing, leading to a sweeping-enhanced resolvent R; and (ii) the sweeping-enhanced resolvent R2 is composited into its iterations R2 to yield nonvanishing Taylor time microscales. The resulting R2-model can correctly predict the frequency spectra and two-point cross-spectra of large-scale structures. This model is numerically compared with eddy-viscosity-enhanced resolvent models. The latter is designed to represent energy transfer instead for time decorrelation and thus under-predicts the characteristic decay timescales. The R2-model correctly yields the characteristic decay timescales over the channel. |
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