Bulletin of the American Physical Society
74th Annual Meeting of the APS Division of Fluid Dynamics
Volume 66, Number 17
Sunday–Tuesday, November 21–23, 2021; Phoenix Convention Center, Phoenix, Arizona
Session E11: Turbulence: Modeling & Simulations II: RANS |
Hide Abstracts |
Chair: Wayne Strasser, Liberty University Room: North 125 AB |
Sunday, November 21, 2021 2:45PM - 2:58PM |
E11.00001: Data-driven computational model selection via recommender systems Xinyi Huang, Robert F Kunz, Xiang Yang RANS is a widely used computational tool in fluid engineering. The accuracy of a RANS calculation depends on the RANS model and the numerics of the code. As a result, experience plays an important role in selecting a RANS model for reliable results. This work aims to distill such human experience in a recommender system. A recommender system is a subclass of information filtering systems that seeks to predict the "rating" a user would give to an item. It has been well developed in commercial applications as recommending content for social media. Here, we utilize it to address the challenge of RANS model selection. We generate a simulation dataset for commonly used RANS models for 9 flows with diverse flow characteristics. Human experience is exploited in rating the performance of each model in quantities of interest (QoIs). We feed the recommender system with this rating data and train. The recommender system then gives estimates of the unknown performance of a specific model in a specific flow of interest. Detailed analysis shows that the accuracy of the prediction relies on the consistency of human experience and the richness of flow feature input. |
Sunday, November 21, 2021 2:58PM - 3:11PM |
E11.00002: Data-driven learning of Reynolds stress tensor using nonlocal models Huansheng Chen, Yue Yu, Justin Jaworski, Nathaniel Trask, Marta D'Elia The Reynolds-averaged Navier-Stokes (RANS) equations are well-known for their efficiency in simulating turbulent flows but require a model for the nonlinear Reynolds stress tensor to close the system of equations. While closures are traditionally obtained via introduction of transport models involving local differential operators, recent work has suggested that nonlocal interactions are important to correctly treat turbulent statistics. To this end, a machine learning framework is pursued to discover nonlocal closures from data. This work presents a data-driven approach to learn a nonlocal model for the Reynolds stress tensor from high-fidelity direct numerical simulation datasets. In this approach, the Reynolds stress tensor is modeled as an integral operator acting over a finite range of lengthscales, and therefore is capable of capturing important aspects of the small-scale anisotropic behavior in turbulent flows. To obtain a nonlocal stress tensor that efficiently and accurately captures turbulent behavior, we propose an optimization-based operator regression approach where the residual of the RANS equations is minimized to calibrate the RANS closure to direct numerical simulation (DNS) data. Different nonlocal kernels are investigated and the resulting accuracy of optimal Reynolds stress closures are demonstrated for fully-developed internal flows. |
Sunday, November 21, 2021 3:11PM - 3:24PM |
E11.00003: Modeling an active grid generated turbulence decay using a coupled transient RANS model with a stochastic turbulence simulator. Mohd. Hanzla, Christopher Ruhl, Arindam Banerjee Kinetic energy decay in grid turbulence has been studied over the past few decades. The computational fluid dynamics approach used includes tuning closure coefficients of turbulence RANS models to match the decay exponent. We take a completely different approach to model turbulence decay in which the closure coefficients in a transient k-w SST model are not tuned. Rather, a 3D time-resolved velocity field is specified for the domain inlet using a stochastic turbulent inflow simulator (TurbSim). Power spectral density (PSD) data to be used in TurbSim is obtained from laboratory water tunnel experiments fitted with a Makita-type active grid. Statistical properties like mean velocity, Reynold’s stress, and turbulence decay for different grid operating conditions are compared to test the modularity of this approach. The methodology developed in this study is robust and can be used to model grid-generated turbulence decay. |
Sunday, November 21, 2021 3:24PM - 3:37PM |
E11.00004: The idea of Fourier-Averaged-Navier-Stokes formulations for turbulent flow modeling Benjamin Freeman, Arman Hemmati Periodic or quasi-periodic flow behaviour is commonly observed in many turbulent flows in engineering problems. The power spectra of periodic fluid motion often have distinct peaks in frequency space, suggesting that large-scale, energetic and predictable motions are present in the flow. Conventional Reynolds-Averaged-Navier-Stokes (RANS) formulations commonly fail to accurately model these inhomogenous and anisotropic structures. Eddy-viscosity models assume that all variations are isotropic and uncorrelated in space and time. Therefore, a formulation for turbulence models that is inherently capable of capturing these periodic variations is desirable. By deriving transport equations for the Fourier coefficients of a periodic flow, we are proposing a class of models that would directly capture periodic flow behaviour. The intention of this modeling paradigm is to simplify and limit the number of computational resources needed to capture large, anisotropic, regularly occurring structures, while increasing the accuracy of turbulence modeling. The newly proposed Fourier-Averaged-Navier-Stokes (FANS) formulations are tested against Direct Numerical Simulation results for 2D wake of a square cylinder, Couette flow and Hagen-Poiseuille flow with an oscillating pressure gradient. |
Sunday, November 21, 2021 3:37PM - 3:50PM |
E11.00005: On the accuracy of eddy-viscosity and eddy-conductivity models in square duct flow Davide Modesti We carry out a priori tests of linear and nonlinear eddy-viscosity models using direct numerical simulation (DNS) data of square duct flow up to friction Reynolds number Reτ = 2000. We focus on the ability of eddy-viscosity models to reproduce the anisotropic Reynolds stress tensor components aij responsible for turbulent secondary flows, namely the normal stress a22 and the secondary shear stress a23. We perform two types of tests: i) on constitutive relations and ii) on RANS models. A priori tests on constitutive relations for aij are performed using the tensor polynomial expansion of Pope, and they allow us to assess the maximum accuracy that one can achieve for different orders of the tensor polynomial, where one tensor base corresponds to the linear eddy-viscosity hypothesis and five bases return the exact representation of aij. Models performance are quantified using the mean correlation coefficient with respect to DNS data Cij, which shows that the linear eddy-viscosity hypothesis always returns very accurate values of the primary shear stress (C12>0.99), whereas two bases are sufficient to achieve good accuracy of the normal stress and secondary shear stress (C22=0.911, C23=0.743). Instead, a priori analysis carried out on popular RANS models, including k–ε and v2–f, reveals that none of them achieves ideal accuracy. The only model which approaches ideal performance is the quadratic correction of Spalart, which has an accuracy similar to models using four or more tensor bases. An equivalent approach, based on vector integrity bases, is used to assess the accuracy of eddy-conductivity models for approximating the turbulent-heat flux. |
Sunday, November 21, 2021 3:50PM - 4:03PM |
E11.00006: Characterization of nonlocal eddy diffusivity in homogeneous shear flow Young R Yi, Ali Mani Downgradient models are often used to estimate or “close” terms such as the Reynolds stresses and turbulent scalar fluxes in the Reynolds-averaged Navier-Stokes equations. These models often invoke assumptions of isotropy and locality when relating the closure terms to mean gradients. In this work, we present a quantitative characterization of the nonlocal eddy diffusivity by incorporating unimodal harmonic forcing to the passive scalar equations, which allow the mean scalar gradient to vary in space while satisfying periodic boundary conditions. We assess how the nonlocal eddy diffusivity depends on the forcing wavenumber by varying it in the k1, k2 plane (streamwise and shear directions). Lastly, we contrast the eddy diffusivity operators between homogeneous shear flow and isotropic turbulence. |
Sunday, November 21, 2021 4:03PM - 4:16PM |
E11.00007: Modeling and Simulation of Mixing Flows Filipe S Pereira, Daniel M israel, Luke van Roekel Modeling and simulation of mixing flows are rife with challenges. These complex variable-density problems can feature transient flow, onset and development of turbulence, non-equilibrium turbulence, density fluctuations, and production of turbulence kinetic energy by shear and buoyancy mechanisms, which are difficult to parameterize with one-point RANS closures. Hence, we have been working on extending the bridging partially-averaged Navier-Stokes equations (PANS) method to material-mixing problems. In this presentation, we discuss the current state of our work and present the framework to extend PANS closures to such a class of variable-density problems. The Taylor-Green Vortex and Rayleigh-Taylor benchmark flows are predicted to illustrate the potential of the model. The results show that the PANS method can accurately predict these variable-density flows at a fraction of the cost of Large-Eddy or Direct Numerical Simulations (LES and DNS). Such a cost reduction can exceed a factor of fourteen for the TGV flow. A new verification and validation technique is also discussed. |
Sunday, November 21, 2021 4:16PM - 4:29PM |
E11.00008: Comparing SST and RSM Predictions of Flow Currents and Heat Generation Within an LPDE Autoclave Reactor Eric Turman, Wayne Strasser An investigation of a Shear Stress Transport (SST) verses a differential Reynolds Stress Model (RSM) is presented. The evaluation tests both models’ ability to predict complex flows and heat generation within an industrial low-density polyethylene (LDPE) reactor. The CFD model consists of a rotating stirrer shaft and temperature dependent polymerization kinetics to simulate the production of LDPE. The geometry consists of multiple paddles, zone baffles, and stirring elements that induce mixing via local shear. Regions of unbounded heat generation throughout the model can negatively impact polymer properties. The initial goal of the study was to understand the formation of hot spots within the reactor. RSM proved to be less diffusive, reporting higher temperatures throughout the reactor. The increase in temperature affected the polymerization kinetics within the model and resulted in a 17% decrease, compared with SST, in catalyst at one of the zonal outlets. The SST model predicted concentrations of hotter fluid isolated near the centralized stirring shaft, while the RSM revealed new mixing patterns with colder currents penetrating into those previously isolated regions. In tracking each tensorial stress, RSM provides a less diffusive approach in simulating the reactor and is shown to be the more desirable model for predicting flow behavior. |
Sunday, November 21, 2021 4:29PM - 4:42PM |
E11.00009: Evaluating the Importance of Nonlocal Eddy Diffusivity for Rayleigh Taylor Instability Dana Lynn Lansigan, Ali Mani, Brandon E Morgan, Jessie Liu, Hannah Williams In this work, we investigate the importance of the nonlocality of the eddy diffusivity operator in RANS models for Rayleigh-Taylor Instability (RTI) using the macroscopic forcing method (MFM). With this method, we measure the moments of the eddy diffusivity in RTI, including the zeroth order moment, which is purely local, and higher order moments in space and time, which characterize nonlocality. We show that the nonlocal effects cannot be ignored when constructing a RANS model, as omission of the terms corresponding to higher order moments results in erroneous predictions. Finally, we present a framework for constructing a nonlocal model for RTI based on our MFM measurements. |
Sunday, November 21, 2021 4:42PM - 4:55PM |
E11.00010: Contemporary Challenge of Connecting Turbulence Models with Time and Spectral Analytical Acoustic Sources Steven A Miller Contemporary Reynolds-averaged Navier-Stokes (RANS) based turbulence models provide Reynolds stresses, turbulent kinetic energy, dissipation, and a multitude of other variables via their particular closure. Recently, my students and myself have proposed both time-domain and spectral-domain sources of noise based on recent closed-form acoustic prediction equations. These noise source terms, which have recently been validated for turbulent jet flows and homogeneous turbulence, do not directly coincide with terms of RANS based turbulence models. In this talk, I identify sources that include fine-scale mixing noise, shock wave turbulence interaction, and large-scale coherent turbulent structures. These source terms are compared with one-equation, two-equation, and Reynolds stress based RANS models. We explain how these terms are approximated via RANS results and present the challenge of modifying RANS closures to directly find source terms. |
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