Traditional programs based on feature engineering are under performing on a steadily increasing number of tasks compared with Artificial Neural Networks (ANNs), in particular for image analysis. Image analysis is widely used in Fluid Mechanics when performing Particle Image Velocimetry (PIV) and Particle Tracking Velocimetry (PTV), and therefore it is natural to test the ability of ANNs to perform such tasks. We report for the first time the use of Convolutional Neural Networks (CNNs) and Fully Connected Neural Networks (FCNNs) for performing end-to-end PIV. Realistic synthetic images are used for training the networks and several synthetic test cases are used to assess the quality of each network predictions and compare them with state-of-the-art PIV software. In addition, we present tests on real-world data that prove that ANNs can be used not only with synthetic images but also with more noisy, imperfect images obtained in a real experimental setup. While the ANNs we present have slightly higher Root Mean Square (RMS) error than state-ofthe-art cross-correlation methods, they perform better near edges and allow for higher spatial resolution than such methods. In addition, it is likely that one could with further work develop ANNs which perform better that the proof-of-concept we offer.
Simultaneous Particle Image Velocimetry (PIV) measurements of stratified turbulent air/water flow in a horizontal pipe have been performed using water droplets as tracers in the gas-phase. The use of water droplets as tracers ensures that the water surface tension remains unaffected and thus allows small scale interfacial structures, such as capillary waves to occur naturally. Experiments have been conducted in a 31 m long, 100 mm diameter PVC pipe using air (density 1.20 kg/m 3 and viscosity 18.4 µPa•s) and water (density 996 kg/m 3 and viscosity 1.0 mPa•s) as test fluids. For the purpose of validation of the experimental setup and the suggested seeding technique, single-phase measurements of both air and water were compared to each other and to DNS results provided by "Wu X. and Moin P., 2008, A direct numerical simulation study on the mean velocity characteristics in turbulent pipe flow, J. Fluid Mechanics, Vol. 608.", showing very good agreement. The two-phase measurements are presented in terms of mean-and rms-profiles. These measurements offer a qualitative demonstration of the behavior of the interfacial turbulence and its correlation with the various interfacial flow patterns. The observations made in this paper are in agreement with the conclusions drawn from the DNS study of "Lakehal D., Fulgosi M., Banerjee S. and De Angelis, Direct numerical simulation of turbulence in a sheared air water flow with a deformable interface, 2003, J. Fluid Mechanics, Vol. 482.". The present results may eventually provide a better explanation to many important phenomena related to the physical characteristics of stratified two-phase flow such as scalar mixing between phases, and to challenges related to its modeling.
The Lagrangian paths, horizontal Lagrangian drift velocity, $U_{L}$, and the Lagrangian excess period, $T_{L}-T_{0}$, where $T_{L}$ is the Lagrangian period and $T_{0}$ the Eulerian linear period, are obtained by particle tracking velocimetry (PTV) in non-breaking periodic laboratory waves at a finite water depth of $h=0.2~\text{m}$, wave height of $H=0.49h$ and wavenumber of $k=0.785/h$. Both $U_{L}$ and $T_{L}-T_{0}$ are functions of the average vertical position of the paths, $\bar{Y}$, where $-1<\bar{Y}/h<0$. The functional relationships $U_{L}(\bar{Y})$ and $T_{L}-T_{0}=f(\bar{Y})$ are very similar. Comparisons to calculations by the inviscid strongly nonlinear Fenton method and the second-order theory show that the streaming velocities in the boundary layers below the wave surface and above the fluid bottom contribute to a strongly enhanced forward drift velocity and excess period. The experimental drift velocity shear becomes more than twice that obtained by the Fenton method, which again is approximately twice that of the second-order theory close to the surface. There is no mass flux of the periodic experimental waves and no pressure gradient. The results from a total number of 80 000 experimental particle paths in the different phases and vertical positions of the waves show a strong collapse. The particle paths are closed at the two vertical positions where $U_{L}=0$.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.