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 T29: Geophysical Fluid Dynamics: General |
Hide Abstracts |
Chair: Pedram Hassanzadeh, Rice Room: North 229 A |
Tuesday, November 23, 2021 12:40PM - 12:53PM |
T29.00001: Deep learning-enhanced ensemble-based data assimilation for high-dimensional nonlinear dynamical systems Ashesh K Chattopadhyay, Ebrahim Nabizadeh, Eviatar Bach, Pedram Hassanzadeh Recent advances in data-driven modeling with deep learning have seen a surge of interest in predicting the spatio-temporal evolution of high-dimensional chaotic dynamical systems and turbulent flow. In applications to such prediction tasks, models (data-driven or otherwise) are often constrained via sparse and noisy observations available from the system, which allows one to obtain improved initial conditions for prediction via data assimilation. However, data assimilation algorithms require a large ensemble of predicted trajectories to accurately compute the background covariance matrix from which these improved initial conditions are derived, thus, becoming computationally intractable for high-dimensional systems. This work presents a hybrid data-driven and numerical modeling framework that leverages deep learning to generate a large ensemble of predicted trajectories to accurately compute the background covariance matrix that enables efficient and accurate data assimilation for predicting the spatio-temporal evolution of high-dimensional geophysical turbulence. |
Tuesday, November 23, 2021 12:53PM - 1:06PM |
T29.00002: On the structures of the ice-covered flows in a bend Berkay Koyuncu, Trung B Le Hydrodynamics of ice-covered rivers is a classical problem of turbulent flows. Due to the differences in the roughness of the ice and the bed layer, two sets of velocity and length scales exist in this flow. The classical theory suggests that these velocity and length scales warrant a double log-law profile where the logarithmic layers exist both near the bed and the ice layers. In this study, we examine the theoretical bounds for these logarithmic layers as 0.26 Reτ1/2 ≤ z+ ≤ 0.15 Reτ with the Reτ → 200,000 in a river reach of the Red River in Fargo, North Dakota. Field surveys are conducted the Acoustic Doppler Current Profiler (ADCP - Sontek M9) in fixed-vessel deployment mode under ice-covered and open surface condition. In the bed layer, our measured data under open surface and ice-covered conditions show that the separation from the upper bound of the logarithmic region reaches z+ ≈ 104. In the ice layer, the extension of the logarithmic region is only limited up to z+ ≈ 3 χ 103. Therefore, the logarithmic layer of the ice layer does not follow the theoretical bounds. An alternative formulation for the velocity profile will be discussed and compared with the double log-law hypothesis. |
Tuesday, November 23, 2021 1:06PM - 1:19PM |
T29.00003: Large Eddy Simulation of ice-covered flows in a river bend Trung Le, Berkay Koyuncu Ice coverage is a critical component of riverine system in cold regions. In this study, three-dimensional flow structures of a river bend are analyzed under both ice-covered and open surface conditions. Field measurements are carried out to reconstruct river bathymetry as well as two-dimensional flow structures in a river bend, which locates in a river reach in Fargo, North Dakota (United States). The Digital Elevation Model is reconstructed from the bathymetry data and the LiDAR data. An in-house Large Eddy Simulation (LES) is used to simulate the three-dimensional flow structures in the bend using the measured discharge from the on-site United States Geological Survey station. Our field measurements and LES results show two distinctively different flow modes corresponding to the open-surface and ice-covered conditions. Under open-surface condition, the secondary flow includes a well-defined circulation. Under ice-covered condition, multiple cells are observed under the ice coverage. We observe a critical role of local bank characteristics on regulating the distribution of high velocity core (HVC) by the presence of dead zones. These zones redirect HVC toward the thalweg and excite Turbulent Kinetic Energy (TKE) to form in the vicinity of the associated shear layer. The narrowing effect of dead zone induce jet-like structures to form at the bend apex and increases TKE locally throughout the water column. Our results suggest a hidden role of ice in regulating flows in meandering rivers. |
Tuesday, November 23, 2021 1:19PM - 1:32PM |
T29.00004: A field and numerical study exploring uncertainty in high-fidelity modeling of riverine flows Kevin Flora, Ali Khosronejad
|
Tuesday, November 23, 2021 1:32PM - 1:45PM |
T29.00005: Realistic Wind Data Generation for Small Unmanned Air Systems in Urban Environment using Convolutional Autoencoders Rohit Kameshwara Sampath Sai Vuppala, Kursat Kara In urban areas, where obstacles are large and abundant, computational fluid dynamics (CFD) would be the best choice for simulating and analyzing complex scenarios for safe wind aware navigation of Small Unmanned Systems. However, owing to the large computation time required for CFD simulation, it is unfeasible for real-time predictions or close to real-time predictions for creating pilot awareness especially for avoiding gusts. In this study, we present a preliminary work on using a data-driven non-linear surrogate model based on deep learning to efficiently generate realistic wind data for urban environments. Using high-fidelity CFD data from Large Eddy Simulations (LES) and Convolutional Auto-Encoders (CAE) for non-linear surrogate modeling, we attempt to generate realistic wind data for urban environment. The non-linear surrogate model is used to extract underlying non-linear modes from the high-resolution data snapshots and LSTM network is trained on these specific modes. Modal predictions for future time-steps are then obtained using this trained LSTM network similar to time-series prediction, without the need of computationally expensive CFD simulations. These modes could then be decoded back into the physical space to obtain the relevant wind field data predictions. Since no prior information about the underlying governing equations are utilized for the predictions, the method is a completely non-intrusive in approach and could be easily extended for other applications with minimal changes. |
Tuesday, November 23, 2021 1:45PM - 1:58PM |
T29.00006: Data-driven modeling of non-local mixing phenomena in geophysical flows Jinlong Wu, Zhengyu Huang, Zhaoyi Shen, Tapio Schneider, Andrew Stuart Geophysical flows often feature mixing phenomena with a wide range of eddy sizes due to the effects of forcing and dissipation on large scales. In addition, many of them tend to exhibit mixing and no-mixing regions in statistically steady states, with eddies propagating finite distances between them. Therefore, a local closure model is sometimes not enough to accurately describe the mixing phenomena in geophysical flows, motivating us to explore non-local models that better account for mixing in geophysical flows. In this work, we propose an approach to construct neural-network-based model of non-local mixing that builds upon data-driven kernels. We test this approach by studying a barotropic flow driven by linear relaxation toward an unstable zonal jet. The results show that our approach achieves better extrapolation capability when training and testing on flows with different relaxation time and the reference unstable zonal jet. The approach also demonstrates the potential of constructing data-driven models of non-local mixing phenomena that can be generalized to different types of geophysical flows. |
Tuesday, November 23, 2021 1:58PM - 2:11PM |
T29.00007: AMR-Wind: Adaptive mesh-refinement for atmospheric-boundary-layer wind energy simulations Michael J Brazell, Shreyas Ananthan, Ganesh Vijayakumar, Lawrence Cheung, Michael A Sprague The ExaWind project is targeting high-fidelity simulations of wind farms in realistic atmospheric conditions. To help achieve this goal the ExaWind project has recently added the flow solver AMR-Wind to its software stack. AMR-Wind is an incompressible flow solver built using the AMReX library. The AMReX library enables performance portability (i.e. same code base runs on CPUs and GPUs), scalable linear solvers, and mesh adaption capabilities. In this talk we will give an overview of AMR-Wind, describe the discretization, and show our progress towards verifying and validating AMR-Wind. Validation studies will include relevant atmospheric boundary layers such as N02 (Pedersen 2014) and GABLS. In addition we will show weak and strong scaling performance studies performed on ORNL's Summit supercomputer. |
Tuesday, November 23, 2021 2:11PM - 2:24PM |
T29.00008: Simulating general flows in curvilinear domains with Dedalus Keaton Burns, Geoffrey Vasil, Daniel Lecoanet, Jeff S Oishi, Benjamin P Brown Dedalus is an open-source Python framework for solving general partial differential equations at scale with modern spectral methods. Here we will describe recent additions to the code for simulating general fluid flows in curvilinear domains. These additions are based on new methods for discretizing arbitrary tensorial equations in disks, cylinders, spherical shells, and balls. These methods enable Dedalus to directly solve both incompressible and compressible hydrodynamical models with full spectral accuracy in these domains, as no reductions via scalar decompositions are necessary. We will discuss the new capabilities of the code, detail our new interface for vector-invariant model specification, and look at a range of geophysical and astrophysical applications. We will also describe ongoing development to support more complex geometries, nonlocal boundary conditions, and data-driven subgrid closures. |
Tuesday, November 23, 2021 2:24PM - 2:37PM |
T29.00009: Turbulent structure of subcritical and supercritical gravity and turbidity currents Jorge Salinas, S Balachandar, Mariano I Cantero We study the turbulent structure of sub- and supercritical gravity currents through high-fidelity simulations that employ up to a billion grid points. Regarding supercritical currents, three families of hairpin vortices appear in the near-bed and interface layers, with their generation mechanisms dictating their shape and orientation. The interaction of near-bed and lower-interface hairpins explains the weak inviscid lid-like behavior of the streamwise velocity maximum, while the upper-interface hairpins are responsible for ambient fluid entrainment. On the other hand, only one family of hairpins can be seen in subcritical currents. An intermediate layer of counter-gradient transport of momentum caps the near-bed layer populated by hairpins. We also present energy consistent depth-averaged mass, momentum, and energy conservation equations. We compute shape factors for currents in the sub- and super-critical regimes. Finally, we discuss the energy restrictions on ambient fluid entrainment and the contributions from turbulent kinetic energy production, buoyancy, and non-self-similarity. |
Tuesday, November 23, 2021 2:37PM - 2:50PM |
T29.00010: Regime transition in subcritical and supercritical currents Jorge Salinas, S Balachandar, Mariano I Cantero Turbidity currents are sediment-laden flows that travel along sloping surfaces, typically the submarine bottom. They are driven by the density difference between the current and the deep layer of quiescent ambient fluid above them, which is the source of streamwise momentum and thus the source of the turbulence that keeps the sediment in suspension. On the other hand, the stable vertical gradient of density from the bottom to the ambient fluid above damps turbulence. The balance between these mechanisms generates two distinct flow regimes, namely super- and subcritical regimes. In this work, we focus on the transition between these two regimes, using a change in the bed slope. For this, we use highly resolved direct numerical simulations (up to 1.8 billion grid points) of spatially developing turbidity currents on a bed of varying slope. We observe that the flow reaches a quasi self-similar regime before the slope-break and far downstream from it, in a different flow regime. Between these two states, a transition zone appears. In particular, the transition from a subcritical to supercritical regime is characterized by very active turbulent mixing at the interface, with flow properties akin to a turbulent wall jet, before the flow readjusts itself and reaches a supercritical self similar state. |
Tuesday, November 23, 2021 2:50PM - 3:03PM |
T29.00011: Deep learning for surrogate modelling of 2D mantle convection Siddhant Agarwal, Nicola Tosi, Pan Kessel, Doris Breuer, Grégoire Montavon Exploring the high-dimensional parameter space governing 2D or 3D mantle convection simulations of terrestrial planets is computationally challenging. Hence, surrogates are helpful. Using 10,500 simulations of Mars’ thermal evolution carried out in a 2D cylindrical-shell geometry, we demonstrated that feedforward neural networks (FNN) can take five key parameters (initial temperature, radial distribution of radiogenic elements, reference viscosity, pressure- and temperature-dependence of the viscosity) plus time as an additional variable, and predict the 1D horizontally-averaged temperature profile at any time during 4.5 billion years of evolution (Agarwal et al. 2020). We now extend this work to predict the entire 2D temperature field which contains more information than the 1D profile such as the structure of plumes and downwellings. First, we compress the temperature fields by a factor of ~140 using a convolutional autoencoder. Then, we compare the use of FNN and long-short term memory networks (LSTM) for predicting this compressed state. While FNN predictions are slightly more accurate, LSTMs ultimately capture the flow dynamics significantly better. The entire spatio-temporal evolution of the temperature field can thus be predicted for a wide range of parameters. |
Tuesday, November 23, 2021 3:03PM - 3:16PM |
T29.00012: Experimental study on the flow behavior of a finite dense fluid release at upstream from cubic building face for different Richardson numbers Romana Akhter, Nigel B Kaye The dispersion of heavy polluted gas has become of great interest as an ever-increasing number of individuals live in big cities. Hence, it is important to have a better understanding over heavy gas dispersion in urban areas. To this end, this study illustrates the characteristics behavior of the heavy/dense gas dispersion at the wake of a cubic building for Richardson number (Ri) ranging from 5 to 40. A series of small-scale experiments of the dispersion was carried out in a water flume following the same geometry of the Thorney Island Phase II Trials 26-29. The tracer gas was released instantaneously at different locations upstream from the building face. Light-Induced Fluorescence (LIF) technique was used to visualize the flow in the wake and Acoustic Doppler velocimetry (ADV) was used to measure velocity profile. This study measured the time taken for the heavy gas to reach the building wake, the cloud height formed by the release on the leeward face of the building, and the time taken to flush the gas out of the building wake for the different Ri and release distance. It is observed that a building block significantly impacts the behavior of the dense gas movement at its wake. The results showed that the maximum concentration difference at the wake is low with the high velocity of the ambient fluid and high with the lower velocity. More phenomena of the flow were analyzed and will be discussed in this study. |
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