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 T23: Quantitative Flow Visualization II: PIV, PTV, PLIF |
Hide Abstracts |
Chair: Huang Chen, University of Nevada, Las Vegas Room: 251 A |
Monday, November 25, 2024 4:45PM - 4:58PM |
T23.00001: Measuring high-resolution 3D density field in buoyant plumes using Optical-flow based Tomographic Background-Oriented Schlieren (TBOS) with sinogram interpolation Javed Mohd, Samiksha Dhakal, Debopam Das Tomographic Background-Oriented Schlieren (TBOS) is a three-dimensional density measurement technique in transparent media where the density gradients are related to the refractive index through the Gladstone-Dale relation. The line-integrated density gradient, based on the apparent deflection of a background pattern, from eight projection angles, is used to reconstruct the instantaneous 3D density field. Traditionally, background pattern deflections are calculated using a window-based cross-correlation approach, which, however, introduces spatial averaging and limits resolution. We employ an optical flow technique, specifically the Horn and Schunk implementation, to obtain the pattern deflection at each pixel, thus resulting in a highly dense displacement field. |
Monday, November 25, 2024 4:58PM - 5:11PM |
T23.00002: Temporal Super Resolution X-ray Particle Velocimetry for Vitrifying Flow Alaa M Ali, Simo A Makiharju Vitrifying fluids exhibit creeping flow, but are challenging to analyze due to significant temperature and viscosity gradients, and phase change which may occur in part of the fluid. Vitrification is of interest due to the potential for enabling indefinite cryopreservation of biological tissue and cells. The fluid is typically an aqueous solution with high concentrations of cryoprotective agents that penetrate biological tissue and reduce the critical cooling and warming rates necessary to avoid ice formation. During the rapid cryogenic cooling process isochoric vitrification in thick-walled metal confinement can be monitored via x-ray photon counting for imaging. We have experimentally demonstrated the feasibility of ice and cavity free isochoric vitrification in Ali et al. (2024), and now are utilizing X-ray particle velocimetry (XPV) to study previously optically inaccessible vitrifying flows. However, as rapid cooling rates are necessary to inhibit ice formation, large number of radiographs equiangularly acquired over 180 or 360 degrees, for parallel and cone beam geometries, respectively that are traditionally required for XPV are not feasible. We, therefore, propose a new temporal super resolution approach to XPV for studying vitrification. The particle tracking is achieved using a small number of radiographs and a region based convolutional neural network (R-CNN) to identify and locate the 2D projection coordinates of the tracer particles, then solve a system of linear equations that translate the projection coordinates to world coordinates. The experimental data is utilized to quantitatively characterize vitrifying flows in an unprecedented detail and is subsequently compared to computational predictions by Rabin (2021). |
Monday, November 25, 2024 5:11PM - 5:24PM |
T23.00003: Leaf motion for flow characterization: generalizing visual anemometry Roni H Goldshmid, John O. Dabiri, John E Sader The Beaufort scale exemplifies qualitative assessment of fluid flow through observations of its interaction with surrounding objects. However, translating these dynamics into quantitative measurements remains a challenge due to structure-specific calibration requirements. This is particularly true for complex structures like trees, where the dynamic response is highly sensitive to individual components (branch architecture, leaf density, water content). Consequently, current methods require reference anemometers for flow characterization from observed structural motion. This study presents a physics-based approach to model the fluid-structure interaction between the impinging wind and the observed leaf motion. We reveal a relationship governing leaf motion across a diverse dataset encompassing twelve tree species, exhibiting variations in age, size, and health. The identified relationship enables flow prediction with a single characteristic length scale, obviating species-specific anemometer calibration within the validated measurement range. |
Monday, November 25, 2024 5:24PM - 5:37PM |
T23.00004: Optimization of hyperparameters and padding for a lightweight velocimetry network Kamila Zdybal, Claudio Mucignat, Ivan Lunati Over the last decade, deep convolutional neural networks (CNNs) have gained popularity for applications to optical flow estimation and fluid velocimetry. Recently, we have proposed a novel CNN lightweight image matching architecture (LIMA), which is specifically designed for velocimetry and proved more accurate than standard methods, as well as computationally cheaper to train than other CNNs. To further optimize the performance, however, two well-known issues of CNNs need to be addressed: (i) the neural networks are highly sensitive to hyperparameters, which makes training prone to get trapped in sub-optimal local minima; (ii) padding of the convolutional kernels at the image boundary may generate spurious artifacts that propagate into the interior of the reconstructed image. Here, we first investigate the effects of various hyperparameters (e.g., the learning rate decay, network depth, and the gradient descent optimizer) and shed light onto their effects on the training speed and the accuracy of the reconstruction. Then we compare various optimization techniques such as random grid search, Bayesian optimization, reinforcement learning, and genetic algorithms, and provide insights into their robustness. Finally, we devise a method tailored to particle image velocimetry to mitigate the boundary effects introduced by kernel padding. |
Monday, November 25, 2024 5:37PM - 5:50PM |
T23.00005: Experimental Evaluation of the Maximum Likelihood Estimator Filter for Lagrangian Particle Tracking in Grid-Generated Turbulence Adhip Gupta, Griffin M Kearney, Kasey M Laurent Accurate measurements of tracer particle positions are essential for identifying key features in turbulent flows during Lagrangian particle tracking. Despite robust tracking systems, systematic errors from various sources are inevitable. These errors can be reduced using filtering techniques. This work evaluates the performance of the Maximum Likelihood Estimator (MLE) filter developed by Kearney et al. (Experiments in Fluids, 2024, 65:24) using data from water tunnel experiments with grid-generated turbulence at a Reynolds number (ReM) on the order of 10^3. The MLE filter is compared to spline filters by Gesemann (arXiv:1510.09034, 2015) and Gaussian filters by Mordant et al. (Physica D, 193(1):245–251, 2004). The MLE filter minimizes measurement errors and noise by incorporating stochastic process physics, assuming a Gaussian distribution for acceleration differences. Our objective is to experimentally determine the probability density function (PDF) of these acceleration differences as a function of ReM and separation time. This will inform the iterative process and refine our assumption of the acceleration distribution. Understanding these differences will improve particle tracking accuracy and enhance the reliability of turbulence statistics, ultimately leading to better insights and advancements in fluid dynamics research. |
Monday, November 25, 2024 5:50PM - 6:03PM |
T23.00006: Large-scale 3D Lagrangian particle tracking using soap bubbles Mano Grunwald, Lorenn Le Turnier, Claudia E Brunner Field measurements of atmospheric turbulence are challenging to conduct, not only because flow conditions are constantly changing in space and time, but also because of the high Reynolds numbers and thus the large range of scales present in the turbulence. Among the most challenging techniques is Lagrangian particle tracking, which is used to investigate turbulent mixing and dispersion behaviour. The ability to conduct highly-resolved Lagrangian measurements in the atmosphere is of interest to a wide range of applications including wind turbine flows, the spreading of pollutants and urban fluid dynamics. |
Monday, November 25, 2024 6:03PM - 6:16PM |
T23.00007: Stereo Tomography of Turbulent Air Jets Tolga Gurcan, Shaurya Aarav, Jason W Fleischer We experimentally measure the three-dimensional structure and dynamics of turbulent jets of air. Jets at variable speeds (different Reynolds numbers) are made with a standard hair dryer, and variations in air density are observed using background-oriented schlieren (BOS). Two views are obtained using binocular prisms, and 3D reconstruction of air density gradients is made by cross-correlating and triangulating the stereo measurements. Using rectangular and circular apertures, we study both planar and round free turbulent jets. |
Monday, November 25, 2024 6:16PM - 6:29PM |
T23.00008: Active Learning for Human-In-The-Loop Fluid Dynamic Measurements Julian Humml, Thomas Rösgen, Morteza Gharib Experimental aerodynamic measurements are crucial for the development of cars, planes, and various applications, including renewable energy and HVAC systems. However, the extensive setup time required for these experiments often limits their practical use. Our research introduces a novel approach that eliminates setup time, allowing a focus on the measurements while ensuring output quality. The operator only needs to outline the domain of interest, after which our algorithms ensure that measurement time is optimally utilized to sample the underlying flow field accurately. |
Monday, November 25, 2024 6:29PM - 6:42PM |
T23.00009: A Stereo-PIV system installed inside a High Reynolds Number Test Model for measuring wall shear stresses from the velocity gradients in the viscous sublayer Chintan Panigrahi, Spencer J Zimmerman, Joseph Katz Owing to the technical limitations, very little experimental data is available for the wall shear stresses generated by the high Reynolds number flows around bodies with varying surface pressure distributions. Cases involving 3D flow separation are particularly challenging due to the large variation in the magnitude and direction of wall stresses. This presentation introduces a high-resolution stereo PIV system that is installed inside a model to measure the wall shear stresses from the velocity gradients in the viscous sublayer. This system is designed to be positioned inside a 3 m long, inclined, 6:1 prolate spheroid that will be tested at Reynolds numbers varying between 1.5 x 106 to 5 x 107. The sample area is 10 x 8 mm. The flow is seeded locally by slowly injecting 2 μm particles from a series of 100 μm diameter micro-injectors located upstream of the sample area. The light source is an external laser whose beam is transmitted to the sample area via an optical fiber, that illuminates the near wall sample plane through a small window installed on the surface. Images are recorded by a pair of 4000 x 3000 pixels CMOS cameras whose data are multiplexed optically and transmitted to an external data acquisition system. These cameras view the sample area via a curved window matched with the body geometry using a transparent polymer that has the same refractive index as water. Data analysis involves ensembled correlation using an interrogation window size of 10 x 10 μm. Sample data and associated uncertainties will be discussed. |
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