Bulletin of the American Physical Society
77th Annual Meeting of the Division of Fluid Dynamics
Sunday–Tuesday, November 24–26, 2024; Salt Lake City, Utah
Session ZC11: Low-Order Modeling and Machine Learning in Fluid Dynamics: Other Applications II |
Hide Abstracts |
Chair: Hamid Reza Karbasian, Southern Methodist University Room: 155 A |
Tuesday, November 26, 2024 12:50PM - 1:03PM |
ZC11.00001: Adaptive Physics-Informed Learning for Downscaling Fluid Flows over Irregular Geometries Thi Nguyen Khoa Nguyen, Christophe Millet, Thibault Dairay, Raphaël Meunier, Mathilde Mougeot The computation of high-resolution flow fields, which is essential for various applications in engineering and climate sciences, is typically achieved by solving partial differential equations (PDEs). In applications such as design optimization or uncertainty quantification, solutions of these PDEs are computed for varying geometries. While physics-informed neural networks have emerged as a new surrogate, their usage for downscaling remains underexplored due to the need for repetitive and time-consuming training. In this work, we address this problem by combining an adaptive mesh learning strategy with a latent representation of irregular geometries. By adaptively refining the distribution of unsupervised training points during the training process, this strategy effectively captures critical couplings between physical fields over complex terrains. The performance is demonstrated through solving 2D stratified boundary layer-topography interaction for various Richardson numbers and mountain shapes. The numerical results show that the surrogate's downscaled fields reproduce local-to-global patterns such as trapped waves and upward propagating gravity waves. Moreover, it is shown that fine-scale fields can be predicted on new geometries using a single coarse field, such as buoyancy. |
Tuesday, November 26, 2024 1:03PM - 1:16PM |
ZC11.00002: Abstract Withdrawn
|
Tuesday, November 26, 2024 1:16PM - 1:29PM |
ZC11.00003: Dynamic Mode Decomposition of Wake Flow Structures for Supersonic Oscillating Genesis Atmospheric Entry Capsule Ashraf Kassem, Shafi Al-Salman Romeo, Bipin Tiwari, Omer San, Kursat Kara Atmospheric entry capsules experience significant oscillations and tumbling during deceleration, which can jeopardize successful parachute deployment and overall mission success. |
Tuesday, November 26, 2024 1:29PM - 1:42PM |
ZC11.00004: Optimizing progress variables for ammonia/hydrogen combustion using encoding-decoding networks Kamila Zdybal, James Sutherland, Alessandro Parente We demonstrate a strategy to optimize low-dimensional parameterizations of turbulent flames using an encoding-decoding artificial neural network architecture. A gradient descent optimizer is informed by the reconstruction quality of important quantities of interest (QoIs) that enter the optimization as the decoder outputs. Our focus is on the combustion of ammonia/hydrogen blends. The literature on ammonia combustion to date lacks an efficient definition of a reaction progress variable (PV) to parameterize the thermo-chemical state-space. A quality parameterization should be able to represent the thermo-chemical state variables accurately, as well as any functions of those, e.g., the source terms of the non-conserved PVs. Our approach incorporates information about the reaction source term of a PV and about important combustion products into the PV optimization. This approach naturally promotes parameterizations where a QoI is uniquely and smoothly represented over the manifold. We show that with an adequate definition of a PV, we can steer the model's accuracy towards improved representation of selected products and pollutants. The definition of a PV automatically adapts to best complement the remaining physics-based parameters, such as the mixture fraction or the enthalpy defect. |
Tuesday, November 26, 2024 1:42PM - 1:55PM |
ZC11.00005: Deep Generative Modeling for Predicting Turbulence Structure in Urban Flows Aakash Patil, Tomek M Jaroslawski, Beverley J McKeon This study introduces a deep generative modeling approach for predicting turbulence structures in urban street-canyon flows, grounded in extensive experimental data. Utilizing high-fidelity particle image velocimetry measurements from wind tunnel experiments, we develop a novel deep learning framework that combines convolutional encoder-decoder architectures with transformer models. Our approach is tailored to capture the complex spatio-temporal dynamics of urban turbulence across various canyon geometries and upstream roughness conditions. The model is trained on detailed flow measurements at the roof level of street canyons, encompassing different width-to-height ratios and flow regimes. By integrating autoregressive training strategies and exploring diffusion model techniques, we enhance the model's ability to generate realistic flow field snapshots and predict key turbulent statistics, two-point correlations, and dominant flow structures. This research demonstrates the potential of deep generative modeling in bridging experimental fluid dynamics with advanced predictive capabilities, offering new insights into urban flow phenomena and turbulence prediction. |
Tuesday, November 26, 2024 1:55PM - 2:08PM |
ZC11.00006: Flow-Informed Path-Planning for Safe Autonomous Flight in Cities Alejandro Stefan-Zavala, Julian Humml, Peter Ian James Renn, Morteza Gharib Autonomous flight in cities is a high-stakes open challenge. The labyrinth of buildings, antennas and trees interacts with wind to produce dynamic flow disturbances that can catastrophically perturb small drones. The proximity to people and property increases the standard for reliable flight control, even in these challenging conditions. |
Tuesday, November 26, 2024 2:08PM - 2:21PM |
ZC11.00007: Deep Learning Strategies for Transport Properties Prediction in Flow Condensation via Acoustic Signatures Ying Sun, Dylan Wallen, Han Hu, Christy Dunlap Flow condensation is critical to the efficient operation of power generation, refrigeration, water purification, and other important applications. Flow condensation in tubes has several advantages over surface condensation, especially in compact and high-power-density applications, where real-time, nondestructive monitoring and prediction of the regime transitions is desirable. Compared to image-based techniques, wideband acoustic sensing (e.g., acoustic emission and accelerometer) allows for higher sampling rates to capture high-frequency interface oscillations that are critical to the flow regime transitions and works well even for condensation in opaque tubes. In this paper, a machine learning framework is introduced to detect the annular to slug flow regime transition and interfacial instabilities, as well as characterizing the instantaneous vapor quality, heat flux and pressure drop of flow condensation, based on imaging, acoustics/vibration, temperature, and pressure data. Multimodal data fusion is implemented to integrate different signals at various operation conditions, sampling rates and test section dimensions. The utilization of multimodal data provides greater prediction accuracy and better feature extraction over a single data source. The model is used for the real-time, non-destructive detection of regime transitions and accurate prediction of thermofluidic performance of flow condensation, thereby improving overall system performance and reliability. |
Tuesday, November 26, 2024 2:21PM - 2:34PM |
ZC11.00008: Quantifying Uncertainty in Groundwater Vulnerability Assessment: a Bayesian Approach Invited Speaker: Nasrin Taghavi Assessing groundwater vulnerability is essential for understanding the risk of pollutants infiltrating groundwater systems after being introduced at the ground surface. Traditional methods for groundwater vulnerability assessment (GVA) often rely on deterministic or empirical approaches, lacking a robust probabilistic framework. This study introduces Bayesian inference as a comprehensive method for GVA, focusing on nitrate concentrations as a proxy for contamination risk in agricultural areas. The proposed model defines a linear relationship between nitrate levels and various hydrological and geological parameters. Two Bayesian algorithms, Joint Maximum a Posteriori (JMAP) and Variational Bayesian Approximation (VBA), are applied to GVA in the Burdekin Basin, an agricultural catchment in Queensland, Australia. Different model ranking metrics are used to compare the models, revealing the Bayesian posterior to be a robust metric for model ranking. Additionally, the Bayesian framework demonstrates superior performance compared to traditional GVA methods in terms of Pearson correlation coefficients (R) between observed and predicted nitrate concentrations. This research highlights the benefits of Bayesian methods in GVA, offering improved model ranking, parameter estimation, and uncertainty quantification. |
Tuesday, November 26, 2024 2:34PM - 2:47PM |
ZC11.00009: Predicting swirling flow states in finite-length pipes using physics-informed neural networks Yuxin Zhang, Diego Rangel Monroy We investigate the capabilities of physics-informed neural networks (PINNs) in predicting the dynamics of axisymmetric, inviscid-limit swirling flow states in finite-length pipes. The inlet flow is described by the circumferential and axial velocity profiles, along with a fixed azimuthal vorticity, while the outlet flow is characterized by a zero radial velocity state. A fully-connected deep neural network is implemented to learn the solution to the unsteady stream function-circulation equations governing the dynamics of swirling flows. To evaluate the results predicted by PINNs, we also solve the problem using global analysis techniques and numerical simulations. We establish a correlation between the outlet states of the solutions obtained using these three different approaches. Our results suggest that PINNs have great potential in predicting swirling flow states. Moreover, the results provide insights into the stability of various states and the nature of flow evolution. |
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. |
© 2025 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