Bulletin of the American Physical Society
76th Annual Meeting of the Division of Fluid Dynamics
Sunday–Tuesday, November 19–21, 2023; Washington, DC
Session X43: Turbulence: Modeling III 
Hide Abstracts 
Chair: Scott Salesky, University of Oklahoma Room: 207B 
Tuesday, November 21, 2023 8:00AM  8:13AM 
X43.00001: A hidden mechanism of dynamic LES models Xiaohan Hu, Keshav Vedula, George I Park The dynamic model is one of the most successful inventions in subgridscale (SGS) modeling as it alleviates many drawbacks of the static coefficient SGS stress models. The model coefficient is dynamically calculated through the minimization of the Germanoidentity error (GIE). However, the driving mechanism behind the dynamic model's success is still not well understood. In the present work, we show that the essence of the dynamic procedure is contained in special lowdimensional subspaces of the resolved velocity field. Specifically, we find that minimization of the GIE along only the principal direction(s) of the resolved strainrate tensor, in lieu of its nine components in its original formulation, produces equally comparable results as the original dynamic model when examined in canonical turbulent channel flow, a threedimensional turbulent boundary layer, and a separating flow over periodic hills. On the other hand, when only the nonprincipal components are considered in the GIE, the model performs as bad as no SGS model case. This suggests that not all components of the Germano identity are equally important for the success of the dynamic model, and that there might be dynamically more important directions for modeling the subgrid dynamics. A potential extension of this idea to a tensorial coefficient Smagorinsky model will also be discussed. 
Tuesday, November 21, 2023 8:13AM  8:26AM 
X43.00002: Error analysis of SES, a mixeddynamics model to capture all turbulent scales Shilpa Sajeev, Diego A Donzis Direct numerical simulations of high Reynolds number turbulent flows are computationally expensive owing to the large number of modes that need to be resolved. Selected Eddy Simulation (SES) is a mixeddynamics approach consisting of solving a subset of modes and modeling the rest using simple dynamics. Since only a fraction of the modes are solved at every step, this can reduce the computational cost associated with high Reynolds number simulations. We have previously applied SES to study isotropic turbulence and shown that one can capture the dynamics of turbulence with just 10% of modes. This reduced computational cost comes at the cost of accuracy. 
Tuesday, November 21, 2023 8:26AM  8:39AM 
X43.00003: Modeling the subgrid scale scalar variance: a priori tests and application to supersaturation in cloud turbulence Scott T Salesky, Kendra Gillis, Jesse C Anderson, Ian Hellman, Will Cantrell, Raymond A Shaw The subgrid scale (SGS) scalar variance represents the “unmixedness” of the unresolved small scales in large eddy simulation, and is critical to model for a variety of applications, including turbulent mixing, turbulent reacting flows, and cloud microphysical processes. In the context of cloud turbulence, Lagrangian microphysics models often require information about the SGS supersaturation variance; thus the fidelity of the SGS model plays a critical role for numerically simulated cloud droplet growth. Using data collected turbulent RayleighBenard convection in the Michigan Technological University Pi chamber (aspect ratio Γ = 2) for Rayleigh numbers Ra ~ 10^{8}10^{9}, we perform a priori tests of the SGS supersaturation variance. Data from a spatial array of ten thermistors is spatially filtered and used to calculate the true SGS variance, a gradient model, and a scale similarity model for three dimensionless filter widths. While the gradient model exhibits low correlations (ρ ~ 0.2), the similarity model is highly correlated (ρ ~ 0.8) with the true SGS variance and exhibits good local performance in terms of joint probability density functions. Implications for large eddy simulations of cloud turbulence will be discussed. 
Tuesday, November 21, 2023 8:39AM  8:52AM 
X43.00004: Numerical investigation of wall modeling for LES using convolutional neural network Golsa Tabe Jamaat, Yuji Hattori Wall modeling in large eddy simulation (LES) is crucial as the computational cost of LES rises significantly for high Reynolds number wallbounded flows which can make the LES impossible for such flows. Therefore, it is important to develop a wall model with reasonable accuracy and computational cost. In the recent years, with the increase in the computational resources and the proven ability of datadriven approaches in making predictions, they have become a popular tool for different applications in the fluid mechanics including turbulence modeling in LES. In the present study, the convolutional neural network (CNN) is used as a tool to develop a datadriven nonlocal wall stress model for the LES of turbulent channel flow. First, a hyperparametric study is performed and the model performance is checked in the a priori test. Finally, the model is embedded in an actual LES to investigate how well the model performs in the simulation. In the a posteriori test, initially, the model is tested for the same condition as used for the training. Then, the generalizability of the model is checked by using the model under various conditions different from those used for the training data. The results show that CNN is successful in establishing a wall model with simple input choices and has reasonable accuracy in predicting the wall shear stress and flow field. 
Tuesday, November 21, 2023 8:52AM  9:05AM 
X43.00005: Lagrangian Large Eddy Simulations via PhysicsInformed Machine Learning Yifeng Tian, Michael Woodward, Mikhail Stepanov, Chris L Fryer, Criston M Hyett, Daniel Livescu, Michael Chertkov In this work, we present a novel approach for developing particlebased Lagrangian turbulence models using Large Eddy Simulation (LES) heuristics within the framework of Physicsinformed Machine Learning. We generalize the evolutionary equations of Lagrangian particles moving in weakly compressible turbulence with extended, physicsinformed parameterization and functional freedom, by combining physicsbased parameters and physicsinspired Neural Networks to describe the evolution of turbulence within the resolved range of scales. The subgrid scale contributions are modeled separately with physical constraints to account for the effects from unresolved scales. We build the resulting model under the Differentiable Programming framework to facilitate efficient training and then train the model on a set of coarsegrained Lagrangian data extracted from fullyresolved Direct Numerical Simulations. We experiment with loss functions of different types, including trajectory, field, and statisticsbased ones to embed physics into the learning. Through extensive analysis and validation, we demonstrate that our Lagrangian LES model successfully reproduces both Eulerian and unique Lagrangian turbulence structures and statistics across a wide range of turbulent Mach numbers. 
Tuesday, November 21, 2023 9:05AM  9:18AM 
X43.00006: Statistical modeling of Burgers turbulence with a superposition of characteristic functionals Gabriel B Apolinário, Michael Wilczek We study an ensemble of random fields, each with statistics described by a general characteristic functional. The typical length scale of these fields is a random variable and is used to model intermittency. This ensemble decomposition approach [Wilczek, New J. Phys. 18, 125009 (2016)] allows for an analytically tractable approximation to turbulent statistics, and is applied to the Burgers equation. By choosing smooth correlation functions and an appropriate distribution for the typical length scale, we build a description of Burgers turbulence at an infinite Reynolds number, in which the velocity statistics are Gaussian, but increment statistics are bifractal. Skewness, a hallmark of turbulent fields, is obtained by truncating a Taylor expanded cumulant generating functional, and this is shown to be a useful approximation. 
Tuesday, November 21, 2023 9:18AM  9:31AM 
X43.00007: Turbulent flow prediction: Lagrangian Particle TrackingDeep Learning (LPTDL) based models Reza Hassanian, Ásdís Helgadóttir, Clara M Velte, Morris Riedel Turbulent flow is a common occurrence in various natural and artificial processes. Predicting and modeling turbulent flow poses a significant challenge, as traditional numerical methods are limited by computational costs and constraints. Furthermore, conducting experimental studies on turbulent flow is constrained by factors like scale and expenses. 
Tuesday, November 21, 2023 9:31AM  9:44AM 
X43.00008: Numerical MultiFractal Cascade of Atmospheric Turbulence Arturo Rodriguez, Vicente Corral, Piyush Kumar, Vinod Kumar The modeling of turbulence cascade has been executed using a multitude of methods, among which we utilized the multifractal representation for a more precise portrayal of turbulence. Typically, energy dissipation characteristics are dictated by specific partial differential equations such as the NavierStokes Equations. However, in climate modeling, the Kolmogorov turbulence cascading approximation often leads to an isotropic representation. In recent years, a shift from the Kolmogorov assumptions has been proposed by Meneveau et al., advocating for multifractal models that accommodate a novel anisotropic representation. Our research is geared towards using Direct Numerical Simulations (DNS) from the JHU Turbulence Database and Large Eddy Simulations (LES) that we created via OpenFOAM. This is to ascertain the accuracy of these simulations in mirroring the experimental procedures of Meneveau, employing numerical simulations that adhere to the same rigorous mathematical paradigms. We hope that the modeling of turbulence cascading using higher fidelity data will yield advancements in the field, and generate quicker, superior remote sensing metrics. We have developed computer code to scrutinize DNS and LES data, delving into the multifractal nature of energy dissipation. We employed the boxcounting method to discern the multifractal dimension spectrum of the DNS and LES data in all directions. This aligns with Meneveau's work and facilitates a more accurate representation of turbulencecascading effects within the atmosphere. 
Tuesday, November 21, 2023 9:44AM  9:57AM Author not Attending 
X43.00009: Abstract Withdrawn

Tuesday, November 21, 2023 9:57AM  10:10AM 
X43.00010: A length scale for nonlocal multiscale gradient interactions in isotropic turbulence Miguel P Encinar Threedimensional turbulent flows enhance velocity gradients via strong nonlinear interactions of the rateofstrain tensor with the vorticity vector, and with itself. For statistically isotropic flows, their total contributions to gradient production are related to each other by conservation of mass, and so are the total enstrophy and total dissipation. However, locally they do not obey this relation and have different (often extreme) values, and for this reason both production mechanisms have been subject to numerous studies, often decomposed in multiscale interactions. In general lines, their dynamics and contributions to the cascade processes and turbulent kinetic dissipation are different, which posses a difficulty for turbulence modelling. In this paper, we explore the consequence of the 'Betchov' relations locally, and show that they implicitly define a length scale. This length scale is found to be about three times the size of the turbulent structures and their interactions. It is also found that while the nonlocality of the dissipation and enstrophy at a given scale comes mostly from larger scales that do not cancel, the nonlocal production of strain and vorticity comes from multiscale interactions. An important consequence of this work is that isotropic cascade models need not distinguish between vortex stretching and strain selfamplification, but can instead consider both entities part of a more complex transfer mechanism, provided that their detailed pointvalue is not required and a local average of reasonable size is sufficient. 
Follow Us 
Engage
Become an APS Member 
My APS
Renew Membership 
Information for 
About APSThe American Physical Society (APS) is a nonprofit 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 207403844
(301) 2093200
Editorial Office
100 Motor Pkwy, Suite 110, Hauppauge, NY 11788
(631) 5914000
Office of Public Affairs
529 14th St NW, Suite 1050, Washington, D.C. 200452001
(202) 6628700