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 J29: CFD: LES and Hybrid LES/RANS |
Hide Abstracts |
Chair: Scott Morris, University of Notre Dame Room: 237 |
Sunday, November 20, 2022 4:35PM - 4:48PM |
J29.00001: Development of Hybrid RANS/LES Model for Oceanic Turbulent Flows Filipe S Pereira, Daniel M israel, Luke van Roekel The ocean plays a crucial role in Earth's climate as it acts as a reservoir of heat, energy, and carbon dioxide. As such, it is imperative to develop numerical methods and parameterizations to predict global ocean circulation flows efficiently. Although direct-numerical and large-eddy simulation techniques are optimal for achieving high-fidelity ocean computations, the cost of resolving all or most turbulent scales makes these methods excessively expensive for practical applications. In contrast, Reynolds-Averaged Navier-Stokes (RANS) parameterizations can significantly reduce the cost of ocean simulations. However, the limitations of such parameterizations representing ocean physics often lead to low-fidelity computations. We extended the hybrid RANS/LES Partially-Averaged Navier-Stokes equations (PANS) to oceanic flows by developing a scale-aware closure for such a class of problems. The PANS model aims only to resolve scales that are not amenable to parameterizing. This strategy leads to the concept of accuracy-on-demand and can significantly increase the efficiency of ocean computations. This work presents the governing equations of the new PANS closure and discusses the results of the predictions of the flow problems constituting the initial validation space of the model. |
Sunday, November 20, 2022 4:48PM - 5:01PM |
J29.00002: Wavenumber Adaptive Simulation of Taylor-Green Vortex Flow Jose Padilla, Aleksandar Jemcov, Scott C Morris Simulations of Taylor-Green vortex flow were performed at Re = 3000 using a two-equation Wavenumber Adaptive Simulation (WAS) model. The model is based on the partial-averaging closure formulation for hybrid turbulence modeling. Turbulent viscosity was computed using the standard k-ω SST URANS turbulence model and scaled directly and locally by the ratio of unresolved-to-total turbulent kinetic energy (f_{k}). The scaling ratio f_{k} was computed via partial integration of the turbulent kinetic energy spectrum, dependent on local cell cut-off wavenumber. Simulations of Taylor-Green vortex flow were performed at different levels of spatial resolution to study the effect of cut-off wavenumber. Simulations were repeated using standard k-ω SST URANS turbulence model and compared to highlight improved accuracy when using the WAS model. The turbulent kinetic energy spectrum and transient behavior of total kinetic energy and dissipation rate were examined to establish the efficacy of WAS a hybrid turbulence model. |
Sunday, November 20, 2022 5:01PM - 5:14PM |
J29.00003: Application of a Scale-Resolving Hybrid model to high Reynolds number canonical and industry-relevant turbulent flows Michael D Mays, Sylvain Laizet, Sylvain Lardeau High-fidelity scale resolving methods remain prohibitively expensive to study turbulent flows of industrial relevance, necessitating the development of methods that can provide high-accuracy data at reduced computational cost. Consequently, a Scale-Resolving Hybrid method based on a time-filtering formalism is applied to different Reynolds-Averaged Navier Stokes base models, permitting a reduction of the required near-wall resolution. The closures are validated and compared on academic canonical flows including a turbulent channel flow and the flow around a finite height cylinder. In general, performance is strong with good agreement with experiment and fully resolved methods, but at lower computational cost. |
Sunday, November 20, 2022 5:14PM - 5:27PM |
J29.00004: Turbulence anisotropy characterization in an IC engine intake flow: a scale-resolved modeling study Suad Z Jakirlic, Maximilian Bopp, Sebastian Wegt, Louis Krüger, Andrea Pati, Christian Hasse The intake flow of an internal-combustion engine, the so-called 'Flowbench' configuration, is computationally investigated using a transient modeling strategy that focuses on a RANS-based, eddy-resolving formulation referred to as Instability-Sensitive Reynolds Stress Model - IS-RSM (RANS - Reynolds-averaged Navier-Stokes), Jakirlic and Maduta (2015, Int J Heat Fluid Flow 51). The model provides the solution for all components of the corresponding residual stress tensor. The turbulence-resolving capability of this subscale stress model is enabled by an additional production of the length-scale-determining quantity in accordance with the SAS-proposal (SAS - Scale-adaptive Simulation), Menter and Egorov (2010, Flow Turbulence and Combustion 85). The results obtained on relatively coarser grids are discussed along with an appropriately resolved LES (Large-eddy Simulation), performed in parallel, and the experimental reference provided by Freudenhammer et al. (2014, Experiments in Fluids 55). The comparative results assessment illustrates correctly predicted instantaneous character of the intake flow as well as its assemble-averaged pattern, highlighting the Reynolds-stress anisotropy characterization revealing differently structured flow regions within the 'Flowbench' configuration. |
Sunday, November 20, 2022 5:27PM - 5:40PM |
J29.00005: Development of a coupled hybrid RANS-LES solver for wall-bounded turbulent flows Ashutosh K Jaiswal, Anupam Dewan, Amitabh Bhattacharya Most hybrid RANS-LES turbulence models suffer from mean stress depletion at the interface of the RANS and LES zones while simulating turbulent wall bounded flows. This issue in turn can lead to wrong estimation of skin friction distribution at the wall, and also adversely affect resolved turbulent stress profiles. In this work, we formulate a new hybid coupled RANS-LES (CRL) model, in which both LES equations and RANS equations are solved simultaneously on the same computational grid. Three zones, namely RANS zone, LES zone and hybrid zone, are defined in the flow domain, based on the distance from the wall. The hybrid zone is sandwiched between the near-wall RANS zone and free stream-LES zone. Effective eddy viscosity (i.e. from resolved+modeled turbulent stress) in the RANS and LES zones is calculated from the modeled and resolved stresses respectively. In the hybrid zone, a novel interpolation scheme is used to obtain the effective eddy viscosity. The mean stress obtained from the net effective eddy viscosity is then imposed in in the mean momentum equation the RANS and hybrid zones. This method essentially allows inputs from both LES and RANS to be blended smoothly in the near-wall hybrid zone, without the need for an extra zonal grid for RANS (e.g. unlike WMLES). Issues related to mean stress depletion are also successfully avoided in this method. We implement a flow solver based on this formulation in OpenFOAM and conduct simulations for turbulent channel flow and backward facing step. Comparison of the results against existing DNS/LES data indicates that the CRL solver is able to predict more accurate mean velocity profiles and resolved turbulent stress profiles compared to DES solvers. |
Sunday, November 20, 2022 5:40PM - 5:53PM |
J29.00006: Deep learning of subfilter-scale turbulence for large eddy simulation Sean Current, Saket Guruker, Vilas Shinde, Datta Gaitonde, Srinivasan Parthasarathy High-fidelity fluid dynamics simulations, namely, Direct Numerical Simulations (DNS) and Large Eddy Simulations (LES), are computationally expensive for high Reynolds number problems. Although Reynolds Averaged Navier-Stokes (RANS) methods have shown qualified success for such flows, many aspects such as heat transfer rates are poorly reproduced, while others such as unsteadiness are outside their purview. Hybrid RANS/LES methods provide a middle ground depending on the scale and required fidelity. As the scope of problems grows larger, it is imperative that new techniques and methods are developed to extend modeling capabilities and improve the efficiency of calculation. To this end, we implement Convolutional Neural Networks (CNN) that learn the features of subfilter-scale turbulence in higher resolution simulations and augment the fidelity of lower resolution simulations, utilizing the highly parallel nature of neural networks to efficiently predict future results. Moreover, we introduce a recurrent training paradigm to stabilize predictions and encourage the consistency of future results. The case studies comprise a lid-driven cavity at Mach 0.5 and a supersonic (Mach 2.7) shock boundary layer interaction. Testing of a variety of neural network architectures ranging from shallow CNNs to popular deep ResNet and U-Net setups as well as Generative Adversarial Networks (GAN) indicate that the models are capable of accurately predicting future timesteps, but only for relatively short time durations. The results enforce the need to explicitly encode physics information into machine learning models. By incorporating physical constraints and partial differential equation information into the models, we anticipate greater stability in the training process and the delivery of an efficient and accurate method for subfilter-scale turbulence modeling. |
Sunday, November 20, 2022 5:53PM - 6:06PM |
J29.00007: Performance enhancement of the Partial-Averaged Navier-Stokes method by using fractional step algorithm Branislav Basara, Aleksandar Jemcov, Zoran Pavlovic An improved accuracy of hybrid RANS/LES models is evident in many industrial CFD applications, but the excessive computing time is still the main obstacle for more extensive use of such models. Therefore, the acceleration of hybrid RANS/LES simulations should be treated with the highest priority. In this work we deal with the Partially Averaged Navier-Stokes (PANS) formulated by Girimaji et al. (2003) and Girimaji (2006), which belongs to the bridging hybrid RANS-LES methods. This approach is being increasingly used to solve complex and high-Reynolds number industrial flows. It is designed to resolve a part of the turbulence spectrum adjusting seamlessly from RANS to DNS (Direct Numerical Simulation). In the previous reported work, we balanced speed and accuracy by using different models for the main PANS resolution parameter, namely, the unresolved to total kinetic energy ratio (fk=ku/ktot). Furthermore, an adaptive mesh refinement based on ‘fk’ criteria (refining cells above prescribed fk value) have been already reported. In this work, we employ the fractional step algorithm for coupling velocity and pressure fields and for solving the pressure correction equation. Preliminary calculations for the external car aerodynamics using meshes larger than one hundred million computational cells have shown significant speed up when compared to well-known SIMPLE like pressure algorithms. Measurements, but also previous RANS calculations, are used as a reference point to the present calculations. |
Sunday, November 20, 2022 6:06PM - 6:19PM |
J29.00008: Large-Eddy Simulation of Flow Around a Rectangular 5:1 Cylinder with Different Inflow Turbulence Conditions Yuxin Zhang, Shuyang Cao, Jinxin Cao The aerodynamics characteristics around the bluff body is a simple but fundamental problem of fluid physics. With the increase of computational power, more and more bluff aerodynamic results based on the Computational Fluid Dynamics (CFD) method have been reported. However, so far, most of the CFD simulations paid attention to the bluff aerodynamics under laminar inflow, and the results under the turbulent inflow are relatively few. In this study, the numerical simulations of the aerodynamics characteristics around a rectangular 5:1 cylinder impacted by different turbulence fields were carried out by using the Large-Eddy Simulation (LES) method. Total of nine cases with different turbulence intensities and different ratios of integral scales to depth were calculated, and the comparison of results between each case showed good agreement with that of the experiments and theories.The flow structures around the rectangular cylinder were also illustrated to better understand the turbulence effects on bluff bodies. |
Sunday, November 20, 2022 6:19PM - 6:32PM |
J29.00009: Evaluation of an LES-based Multi-Fidelity framework for wind loading predictions. Mattia Fabrizio Ciarlatani, Catherine Gorlé Finely resolved Large-Eddy Simulations (LES) can correctly predict wind loading on the surface of high-rise buildings. However, these simulations require substantial computational time. Thus, to allow LES to be routinely used for wind loading predictions, a significant computational speedup is needed. This work evaluates the use of multi-Fidelity modeling for wind loading predictions to significantly reduce the computational time required by CFD. We aim to combine data at many points from a Low-Fidelity model, namely a coarse LES, with data at a few points from a High-Fidelity model, namely a fine LES, to provide predictions with accuracy close to that of the fine LES at a fraction of the cost. To do so, we build two surrogate models over the discrepancy between the Low- and the High-Fidelity LES, and we use them to correct Low-Fidelity predictions. We also explore how the Low-Fidelity LES resolution affects the predictive ability of the two models by using two different mesh setups for the coarse LES. The results show that both frameworks allow for a reduction in the RMSE of the Low-Fidelity LES predictions with a substantial decrease in the computational cost. Future work will explore a machine learning-based Multi-Fidelity framework to further reduce the predictions RMSE. |
Sunday, November 20, 2022 6:32PM - 6:45PM |
J29.00010: Suppression of vortex shedding in the wake of a bluff body by active control Ananthu J P, Prasanth P Nair, Vinod Narayanan The study of the wake behind the bluff body has been the topic of interest for engineers for the last seven decades. A wide range of real-life applications, including aircraft, bridges, automobiles, and ships, makes vortex shedding an important topic for research. Wake control is vital in the defense application to avoid the trace. Active control of the wake prevails over other passive control methods for the drag reduction for a bluff body with a blunt trailing edge. Therefore active control has been used in the current study. Introducing a secondary flow in the trailing edge of the bluff body prevents the growth of wake bubbles. Thus, the periodic formation of vortices from the trailing edge will be limited due to the smaller wake bubble. Reduced oscillations, in turn, reduce the vorticity across the wake of the bluff body. Hybrid RANS/LES has been used for the current simulation. Improved delayed detached eddy simulation (IDDES) turbulence model is used. OpenFOAM has been used for the simulation. A parametric study is conducted with the velocity ratio of freestream velocity and the injection velocity from the base of the bluff body as the parameter. The numerical model accurately captured the flow development, wake bubble growth, and von-Karman vortex street. The vorticity across the wake of the bluff body with a blunt trailing edge is reduced by 40% to 55% when a secondary flow is introduced at the rear face of the body. Proper Orthogonal Decomposition (POD) and Dynamic Mode Decomposition (DMD) sheds light on the dominant frequency and the modes responsible for the vortex shedding. |
Sunday, November 20, 2022 6:45PM - 6:58PM |
J29.00011: Extension of the Active Model Split Hybrid Turbulence Model for Wind Energy Relevant Flows Jeremy Melvin, Marc T Henry de Frahan, Robert D Moser, Michael A Sprague In this talk, we discuss efforts to extend the Active Model Split (AMS) hybrid turbulence model to wind energy relevant flows. AMS, a recently developed Reynolds-Averaged Navier Stokes (RANS) / Large Eddy Simulation (LES) modeling framework (S. Haering et al., 2022), has many conceptual advantages over existing RANS/LES models. The main premise is the splitting of the modeled stress into two distinct terms, one in the role of RANS, to approximate the mean subgrid stress and the other in the role of LES, to provide energy transfer between resolved and unresolved scales. AMS has shown promising results in canonical turbulence validation cases when coupled with the $\overline{v}^2-f$ RANS model. However, for large scale wind farm simulations, models with less computational overhead are desired. To simulate these flows, we extend AMS by providing the mean subgrid stress through the Shear Stress Transport (SST) RANS model (Menter et al., 2003), commonly used in wind turbine simulations. We compare AMS-SST models with DES variants on high Reynolds number (Re) flows of airfoils, with comparisons to LES and/or experimental data where available. The advantages of the AMS hybrid models are discussed and areas where further AMS development is being pursued are emphasized. |
Sunday, November 20, 2022 6:58PM - 7:11PM |
J29.00012: A 3-D Kinetic-Based Discrete Dynamic System and its Surrogate Models by Machine Learning Huidan Yu, Xiaoyu Zhang, Jianhua Yin, James M McDonough, Xiaoping Du Naiver-stokes based discrete dynamical system (DDS) has been used in turbulence modeling to capture the subgrid-scale motion. We have recently derived a 3-D DDS for incompressible flows based on the kinetic-based lattice Boltzmann equation (LBE). Five bifurcation parameters, including a relaxation time from the LBE, a splitting factor to separate large-scale and sub-grid scale (SGS) motion, and three wavevector components from the Fourier space Galerkin procedure used to derive the DDS. Numerical experiments employing combinations of these bifurcation parameters have produced laminar and turbulent flow behaviors indicated by the patterns of power spectral density (PSD) of the time series. To use this DDS for generating SGS information in large-eddy simulation (LES) of pulsatile turbulence via the lattice Boltzmann method, we intend to systematically study the effects of the bifurcation parameters on capturing laminar and turbulence behaviors. To overcome the demands of computation time, we developed surrogate models for the DDS through physics-based machine learning classification techniques including Support Vector Machines and Artificial Neural Networks. The surrogate models from both machine learning methods result in prediction precision varying from 93% to 99% based on the size of test point sets. For 15,000 test points, the surrogate models reduce 98% of computation time to the DDS. We will use the surrogate models to explore phase diagrams in the 5-dimensioanl parameter space, which will be critically important for appropriately selecting bifurcation parameters for LES modeling of pulsatile turbulence. |
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