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 L37: Turbulence: Large Eddy Simulations and Direct Numerical Simulations |
Hide Abstracts |
Chair: Meng Wang, University of Notre Dame Room: 245 |
Monday, November 21, 2022 8:00AM - 8:13AM |
L37.00001: Stable a posteriori LES of forced two-dimensional turbulence using shallow artificial neural networks Aditya Sai Pranith Ayapilla, Yuji Hattori Over recent years, data-driven methods to model subgrid-scale (SGS) stresses have garnered attention and been shown to outperform traditional SGS closure frameworks. It is well known that the statistical characteristics of an SGS model are more important than the accurate representation of the SGS term to obtain a stable LES. Two of the most important statistical requirements include the accurate representation of mean dissipation and the accurate prediction of the mean SGS stresses. Previous studies on data-driven SGS modeling showed that despite the accurate prediction of the inter-scale energy transfers and/or the SGS term, instabilities would often arise in an a posteriori LES, which would often require ad-hoc post-processing, one of which is to eliminate backscatter. In the present work, using forced two-dimensional turbulence test cases, we developed an SGS model using shallow (2 hidden layers) artificial neural networks that could obtain high correlation coefficients of both the SGS stresses and the inter-scale energy transfers, using a Gaussian filter with several filter widths. More importantly, the developed SGS model also resulted in a stable a posteriori LES. |
Monday, November 21, 2022 8:13AM - 8:26AM |
L37.00002: The role of the areal parameters on turbulent flow over gaussian roughness Federica Bruno1, Stefano Leonardi2, Mauro De Marchis1 1 Department of Engineering and Architecture, University Kore Enna, Italy 2 Department of Mechanical Engineering, University of Texas at Dallas, Richardson, Texax, USA Federica Bruno, Stefano Leonardi, Mauro De Marchis Reliable prediction of friction drag and roughness function ∆U+ induced by wall roughness nowadays is an open issue, despite the extensive effort of the scientific community. Recent studies illustrate how a single feature of the roughness geometry is not able to adequately predict ∆U+. In order to find a geometrical parameter able to match the stochastic nature of random roughness and its effect on the flow, several Large Eddy Simulations at Re??= 400 over 2D rough walls were carried out. Designed throughout gaussian functions, the geometries are characterized by different roughness average heights, distributions and densities. The results confirm that the higher values of the roughness function are associate with higher roughness elements. With the aim to find a universal correlation between ∆U+ and rough surface, new geometrical quantities based on the area occupied by the roughness were investigated. The proposed parameter, called Effective Area, EA, is based on the correlation between the Effective Slope and the roughness area A. The achieved results show a good correlation between ∆U+ and EA. |
Monday, November 21, 2022 8:26AM - 8:39AM |
L37.00003: Exact decomposition of the kinetic helicity flux in homogenous turbulence Damiano Capocci, Perry L Johnson, Sean Oughton, Luca Biferale, Moritz Linkmann In homogeneous turbulence, the relative contributions of different physical mechanisms to the energy cascade can be quantified by an exact decomposition of the energy flux (P. Johnson, Phys. Rev. Lett., 124, 104501 (2020), J. Fluid Mech. 922, A3(2021)). We extend the formalism to the transfer of kinetic helicity across scales and quantify the contributions of different physical effects in the inertial range. All sub-fluxes transfer helicity from large to small scales. About 50% of the mean flux is due to the scale-local deformation of vortices into vortex sheets in a way that the vorticity deformation tensor aligns with the strain rate tensor and a strain-vorticity coupling with the deformation of resolved-scale vorticity. We derive an exact relation between these effects, asserting that the mean contribution of the former is three times larger than that of the latter. Scale non-local effects account for the remaining 50%, with approximate equipartition between scale-nonlocal versions of the two aforementioned effects and the alignment of resolved-scale vorticity strain with small-scale vorticity. |
Monday, November 21, 2022 8:39AM - 8:52AM |
L37.00004: A Data-Driven Nonlinear Eddy Viscosity Model for Sub-Grid Scale Stress Closure Samantha Friess, Basu Parmar, Aviral Prakash, John A Evans The growing popularity of machine learning in fluid mechanics research has unveiled the massive potential of Big Data in the turbulence modeling community to reduce model uncertainties. The vast amounts of previously unmanageable high-fidelity flow field simulation and experimental data are now being harnessed to systematically inform medium to lower-cost turbulence models, which often outperform current state-of-the-art approaches. Through the extraction of meaningful statistics from a nominal amount of open-source Direct Numerical Simulation data, we have constructed a data-driven framework to model the Sub-Grid Scale Stress (SGS) tensor to close the filtered Navier Stokes equations. Our proposed Nonlinear Eddy Viscosity (NLEV) model imbeds frame, Galilean, time, and dimensional invariance properties directly into the machine learning model by training over an integrity basis of invariant scalars and tensors. Unlike previous approaches, our NLEV model form has been extended to handle anisotropic filter widths. We demonstrate the robustness of our low-cost data-driven framework and show improved predictive performance over classical SGS models for Large Eddy Simulations in both a priori and a posteriori tests. |
Monday, November 21, 2022 8:52AM - 9:05AM |
L37.00005: Large Eddy Simulations of wall bounded turbulent flows with fixed filter width Sandip Ghosal, Rahul Agrawal, Ahmed Elnahhas, Perry L Johnson In large eddy simulations (LES), the geometry-dependent, large, energy-carrying scales are computed, and the subgrid scales are modeled. In wall-bounded turbulence, the filter width must approach zero for LES equations to strictly hold, which leads to prohibitively expensive grid-resolution requirements at high Reynolds numbers. Further, this also leads to additional unclosed terms as this filtering operation does not commute with the differentiation. |
Monday, November 21, 2022 9:05AM - 9:18AM |
L37.00006: Large eddy simulation of MHD turbulent Taylor-Couette flow in axial magnetic field Hiromichi Kobayashi, Takahiro Hasebe, Takayasu Fujino, Hidemasa Takana We conduct large eddy simulation of magnetohydrodynamic (MHD) turbulent Taylor-Couette (TC) flows in the axial magnetic filed. The Hartmann number is proportional to the magnetic flux density. We investigate the effect of the Hartmann number on the flow and electric current fields. As increasing the Hartmann number, turbulent fluctuations are suppressed, and the turbulent flow distributions approach the MHD laminar flow distributions. The orientations of turbulent vortices change from the azimuthal flow direction to the axial magnetic field direction. The triplet high-speed streaky structures emerge near the inner wall. |
Monday, November 21, 2022 9:18AM - 9:31AM |
L37.00007: Wall-pressure fluctuations in a high-Reynolds-number turbulent-boundary-layer flow over a forward step Di Zhou, Yi Liu, Meng Wang Large-eddy simulations are performed to investigate the turbulent boundary-layer flow over a small forward-facing step at momentum-thickness Reynolds number of 15,500 and step-height Reynolds number of 26,600, with a focus on the effect of the step on wall-pressure fluctuations. The step height is 15% of the unperturbed boundary-layer thickness. Both wall-modeled LES and wall-resolved LES are employed, and consistent results are obtained for the spatiotemporal characteristics of the fluctuating wall pressure. The LES predicts step-elevated wall-pressure frequency spectra comparable to those from earlier experimental and numerical studies at the same step-height Reynolds number and lower momentum-thickness Reynolds numbers, but underpredicts the low-frequency spectral level relative to the measurements at Virginia Tech (Awasthi et al., J. Fluid Mech. Vol. 756, 2014) that match both Reynolds numbers. Two-point and space-time correlations are predicted well compared with the experimental data. They show a significant decrease in correlation length and time scales at the high Reynolds number in the unperturbed boundary layer, and a drastic increase in correlation scales after the step due to step-induced disturbances, which decay slowly in the downstream. The fluid dynamic sources of the wall-pressure fluctuations and their downstream evolution will be discussed. |
Monday, November 21, 2022 9:31AM - 9:44AM |
L37.00008: DNS and LES of turbulent channel flows with temperature-dependent variable viscosity Kazuhiko Suga, Yusuke Kuwata Modification of turbulence by temperature-dependent fluid properties is significant in high Prandtl number flows with relatively large temperature differences. The present DNS study describes how large the turbulence and turbulent scalar fields are modified by the temperature-dependent viscosity in a water channel flow at Ret=650. The present study also has performed an LES and discusses the importance of including the correlation term between velocity and variable viscosity. A comparison between the DNS and the LES data suggests that although the effects of the velocity-variable viscosity correlation term is not totally ignorable, the LES without such a term can capture most of the essential characteristics of variable viscosity turbulence. The present study also provides a database for the velocity-variable viscosity correlation term to discuss its modelling. |
Monday, November 21, 2022 9:44AM - 9:57AM |
L37.00009: Explainable transfer learning for data-driven closure modeling of Rayleigh-Benard turbulence across parameters Yifei Guan, Ashesh K Chattopadhyay, Pedram Hassanzadeh In this work, we develop a data-driven subgrid-scale (SGS) model for large eddy simulation of turbulent thermal convection using a fully convolutional neural network (CNN). With the filtered DNS (FDNS) data, we train the CNN with the filtered state variables, i.e., vorticity, temperature, and stream function as inputs and the nonlinear SGS term (subgrid momentum flux and heat flux) as an output. A-priori analysis shows that the CNN-predicted SGS term accurately captures the inter-scale energy transfer. A-posteriori analysis indicates that the LES-CNN outperforms the physics-based models in both short-term predictions and long-term statistics. Although the CNN-based model is promising in predicting the SGS term, it lacks generalizability to different flow scenarios, e.g., various Rayleigh or Prandtl numbers. To relieve this shortcoming of CNN, here we use the transfer learning (TL) technique which utilizes a previously trained CNN from a base system and a fraction of data (<5%) from a target system. Here we extend the explainable TL framework proposed by our group [1] to guide the TL process in a Fourier-Chebyshev domain with spectral analyses. With proper re-training, a-priori and a-posteriori analyses show that the CNN with TL enhances the SGS model and allows the data-driven model to work stably and accurately in a different flow scenario. |
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