Particle Image Velocimetry Particle Research Articles

Cross-correlation-based convolutional neural network with velocity regularization for high-resolution velocimetry of particle images

Particle image velocimetry (PIV) stands as a pivotal experimental technique in fluid dynamics, enabling the visualization and analysis of fluid flows. Traditional methods for extracting velocity fields from particle images often rely on window-cross correlation PIV or, more recently, optical flow techniques rooted in intensity conservation principles. However, the former approach suffers from low resolution, whereas the latter is hampered by computational inefficiency and a high susceptibility to noise. Recent studies have demonstrated the effectiveness of convolutional neural networks (CNNs) in processing particle images to obtain high-resolution and high-accuracy velocity fields, though traditional CNN architectures are still not satisfying in accuracy. The present study introduces an enhanced network, En-FlowNetC, based on the cross correlation-based CNN FlowNetC, specifically designed to process PIV particle images and achieve high-accuracy, high-resolution velocity fields. It incorporates a velocity regularization and is trained and validated on canonical datasets. The results indicate that En-FlowNetC surpasses traditional CNN networks in accuracy and markedly outperforms the classic Horn–Schunck optical flow method in both complex and simple flow scenarios. Furthermore, this study confirms the beneficial impact of velocity regularization, when judiciously applied, on network accuracy. The proposed modifications compared to the original FlowNetC are also examined in the ablation experiments. Overall, En-FlowNetC provides an effective deep-learning solution for PIV analysis, paving the way for future endeavors aimed at achieving highly accurate and resolved velocimetry.

PIV investigation on the effect of gas distributor design for fluidization of Geldart A spherical particles

This is a hydrodynamics study on gas distribution during fluidization of the industrially relevant Geldart A type particles. The impact of 10 different distributor designs comprising perforated distributors (pitch and orifice arrangement pattern varied) and nozzle distributors (nozzle size varied) were studied experimentally on a bench-scale unit. Particle flow homogeneity in air was determined by a novel Particle Image Velocimetry (PIV) analysis of the freeboard velocity field. Over 50 PIV particle phase vector field images are presented and analysed. Bed expansion, Particle attrition with and without fines, velocity profiles with and without fines and effect of fines on bed expansion were ascertained. Best distributor design for a unit such as Fluid Catalytic Cracking (FCC) regenerator and FCC riser were found to be 0.8 cm and 0.6 cm nozzle plates distributors respectively. Confirmation with previous study revealed a correlation between homogeneous distribution of particulate flow and lower number of turbulence zones.

Augmentation of experimentally obtained flow fields by means of Physics Informed Neural Networks (PINN) demonstrated on aneurysm flow

Abstract Biofluid mechanics play an important role in the study of the mechanism of cardiovascular diseases and in the development of new implants. For the assessment of hydrodynamic parameters, experimental methods as well as in-silico approaches can be used, such as particle image velocimetry (PIV) and Deep Learning, respectively. Challenges for PIV are the optical access to the region of interest, and time consumption for measuring and post-processing analysis in particular for three dimensional flow. To overcome these limitations state-of-the-art deep learning algorithms could be utilized to augment spatially coarse resolved flow fields. In this study, we demonstrate the use of Physics Informed Neural Networks (PINN) to augment PIV measurement data. To demonstrate a combined workflow, we investigate the flow of a Newtonian fluid through a simplified aneurysm under laminar conditions. Generation of synthetic PIV particle images of a single measurement plane and the corresponding PIV vector calculations were performed as the basis for the PINN algorithm. Based on the Navier-Stokes equations the PINN reconstructs the entire 3D flow field and pressure distribution inside the aneurysm. We observed qualitative agreements between ground through data and PINN predictions. Nevertheless, there are substantial differences in the quantitative, locally resolved comparison of the flow metrics, despite the generally tendency for the PINN algorithm to correctly augment the flow field.

Marangoni flow induced in a waterbody by the impact of a raindrop

In this study, the Particle Image Velocimetry (PIV) technique is employed to investigate the transient flow that arises when a freshwater drop impinges onto the surface of a water body. Our measurements show that in addition to the impact of the drop itself, surfactant and salt concentration play a crucial role in determining the resulting flow on and near the surface. The latter contributions arise from surface tension gradients that arise from merging the drop’s water with the water on the surface, which reduces the concentrations of salt and surfactant. As the surface tension of water falls with decreasing salt concentration, the decrease in salt concentration leads to a Marangoni flow, which moves radially outward from the impact area. On the other hand, the reduction in surfactant concentration causes a reverse Marangoni flow, which flows towards the impact point as the surface tension of water increases with decreasing surfactant concentration. The overall direction and strength of the flow depend on the Weber number and the surfactant and salt Marangoni numbers. Experiments indicate that when the surfactant Marangoni number dominates, the radial flow due to the drop’s impact quickly reverses direction giving rise to a jet-like flow from the impact point in the downward direction. In contrast, when the salt Marangoni flow dominates, the radial flow persists for a longer duration, and the diameter of the region free of PIV particles on the surface is larger than for the freshwater case. The height of the water column increases with increasing Weber and surfactant Marangoni numbers but decreases with salt Marangoni number. The strength and direction of the surface flow are crucial for enhancing the mixing of dissolved gases in the near-surface layer into the water body.

Experimental study of velocity and turbulence fields in a square mixer-settler tank: Comparison of shake the box PTV and 2D2C PIV

The characterization of flows in agitated vessels is essential for a good understanding of physical and chemical processes and to validate computational fluid dynamics code. The hydrodynamics of cylindrical vessels have been well studied at various scales, mostly in the turbulent regime, typically using particle image velocimetry (PIV). Studies of square tanks or at moderate Reynolds numbers are much rarer however, and 3D Shake-The-Box (STB) Lagrangian particle tracking has never been used in this context. The objective of the present work was to compare two-dimensional, two-component PIV (2D2C-PIV) and STB particle tracking in characterizing the hydrodynamics of a stirred 20 cm sided (8 L) square tank at Reynolds numbers ranging from 1700 to 4300. The velocity fields, velocity profiles, turbulent kinetic energy fields and turbulent kinetic energy dissipation rates obtained using the two approaches were compared. The velocity fields measured with the two methods were found to be in good agreement, except for a few discrepancies in the velocity profiles of the turbine discharge at 250 rpm. The dimensionless velocity profiles were similar above 100 rpm and indicated that the flow was turbulent at the Reynolds numbers considered. The turbulent kinetic energy fields obtained with the two approaches were globally similar, although with STB, the energy tended to be underestimated, especially in the jet discharge area. The STB method was also found to have lower spatial resolution and precision in measuring turbulent kinetic energy dissipation rates, a consequence of a too large interrogation window. Recommendations are proposed to improve the precision of STB measurements.

Experimental study of flow field and sediment suspension in the gap area of tandem piles under oscillatory flows

This study experimentally investigates the local scour mechanism around tandem piles in oscillatory flows with emphasis on the flow characteristics and sediment suspension during the scouring process. The instantaneous flow field is measured by the particle image velocimetry (PIV) technology meanwhile the sediment transport is measured by an image processing method. A preprocessing technique for the same PIV particle image is employed to identify the water particle and sediment particle, thereby realizing the simultaneous measurement of flow field and sediment movement. It shows that with the temporal development of scour, the scour profiles tend to be symmetrical, and the equilibrium scour depth is expected increased with the increase of KC number. Suspended sediment is observed significant in the gap area of the piles with large magnitude of suspended sediment concentration, which contributes most to the bed scour. From the instantaneous flow field, a noticeable upward flow is occurred in the gap area, which is identified as the primary driving force lifting the sediment. From the time-averaged flow field, a vortex is formed, inducing a strong rising flow with high velocity magnitude below it, which is responsible for the suspension of sediment. It is found that there exists a linear correlation between flow velocity and sediment motion velocity.

Effect of basset history drag on the accurate determination of streaming flow with PIV measurement

In acoustofluidics, the conflation of acoustic streaming and acoustic pressure effects on suspended particles, especially for particle image velocimetry (PIV) with micron-sized particles, leads to inaccurate streaming measurements. Typical multiphase streaming models fail to account for the Basset history drag, which emerges due to a mismatch between the momenta of the particle and the local flow. The history drag is proportional to a fractional-order derivative of the mismatch. Since this force integrates over time, its absence leads to significant modeling errors while also augmenting errors from other sources (e.g., particle-fluid density mismatch and acoustic radiation forcing). We investigate the effects of particle size selection on accurate streaming flow determination from PIV microacoustofluidic system measurements. We achieve this through experimental observation ofparticle behaviors under varied ultrasonic forcing frequencies (30–500MHz) and particle sizes. We pair the empirical study with an analytical fractional-order differential model of a spherical particle under the influence of radiation forces. Rigorous comparative analyses between the model and observations are undertaken to more completely characterize the magnitude of various effects leading to spurious PIV data. We expect the results of this investigation to clearly delineate optimal PIV particle size selection for the continued study of microacoustofluidic systems.

Velocity refinement of PIV using global optical flow

In this study, we propose a method to enhance the particle image velocimetry (PIV) velocity resolution using global optical flow along with image warping. A global optical flow formula proposed by Brox et al. (High accuracy optical flow estimation based on a theory for warping. In: Proceedings of the 8th European conference on computer vision, vol 4, pp 25–36, 2004) is adopted to compensate the intensity changes of PIV image pairs, which depend on the set-up and synchronization of a laser and a camera. The proposed method is quantitatively evaluated and validated using synthetic particle image pairs generated for Rankine vortices and reference DNS-based velocity data. The proposed method outperforms the conventional PIV method in capturing small scale vortex and turbulent structures due to its enhanced spatial resolution. In addition, the proposed method shows good performance in large displacement fields and varying image intensity whereas optical flow is applicable to small displacement and susceptible to image intensity variation in general. Finally, the proposed method is applied to real PIV particle images of a multiple rectangular jet flow. The results show that the proposed method successfully works out high-resolution fluid mechanical structure and quantities while preserving the conventional PIV results.

Effect of Different Acoustic Parameters on NOx Emissions of Partially Premixed Flame

The effects of acoustic frequency (f)/0–400 Hz and amplitude (A)/0–1400 Pa on nitrogen oxides (NOx) emissions of a partially premixed flame were investigated experimentally. The mechanism of NOx emissions was analyzed by the evolution of the vortex, which was shown by particle image velocimetry (PIV). From the relationship of NOx emission index (EINOx) and acoustic parameters, it was concluded that a critical frequency (fc) from 170 Hz to 190 Hz appeared. When the frequency was less than fc, EINOx decreased linearly with an increase in amplitude. The flame length became shorter, which led to a decrease in the global residence time, and hence, a reduction in reaction time for NOx. However, a direct proportional relationship between EINOx and amplitude was not found when the frequency was larger than fc. Based on PIV particle scattering images, with an increase of the acoustic frequency, the effects of the acoustic field on the flame base became less significant, but the flame length and reaction space of NOx were gradually increased.

Simultaneous direct measurements of concentration and velocity in the Richtmyer–Meshkov instability

The Richtmyer–Meshkov instability (RMI) is experimentally investigated in a vertical shock tube using a broadband initial condition imposed on an interface between a helium–acetone mixture and argon (Atwood number $A\approx 0.7$). In the present work, a shear layer is introduced at the interface to serve as a statistically repeatable, broadband initial condition to the RMI, and the density interface is accelerated by either an $M=1.6$ or $M=2.2$ planar shock wave. The development of the ensuing mixing layer is investigated using simultaneous planar laser-induced fluorescence (PLIF) and particle image velocimetry (PIV). PLIF images are processed to reveal the light-gas mole fraction, while PIV particle image pairs yield corresponding two-component planar velocity results. Field structure and distribution are explored through probability density functions (PDFs), and a decomposition is performed on concentration and velocity results to obtain a mean flow field and define fluctuations. Simultaneous concentration and velocity field measurements allow – for the first time in this regime – experimentally determined turbulence quantities such as Reynolds stresses, turbulent mass-flux velocities and turbulent kinetic energy to be obtained. We show that by the latest times the mixing layer has passed the turbulent threshold, and there is evidence of turbulent mixing occurring sooner for the higher Mach number case. Interface measurements show nonlinear growth with a power-law fit to the thickness data, and that integral measurements of mixing layer thickness are proportional to threshold measurements. Spectral analysis demonstrates the emergence of an inertial range with a slope ${\sim}k^{-5/3}$ when considering both density and velocity effects in planar turbulent kinetic energy (TKE) measurements.

High Momentum Jet Flames at Elevated Pressure: Detailed Investigation of Flame Stabilization With Simultaneous Particle Image Velocimetry and OH-LIF

A model FLOX® combustor, featuring a single high momentum premixed jet flame, has been investigated using laser diagnostics in an optically accessible combustion chamber at a pressure of 8 bar. The model combustor was designed as a large single eccentric nozzle main burner (Ø 40 mm) together with an adjoining pilot burner and was operated with natural gas. To gain insight into the flame stabilization mechanisms with and without piloting, simultaneous particle image velocimetry (PIV) and OH laser-induced fluorescence (LIF) measurements have been performed at numerous two-dimensional (2D) sections of the flame. Additional OH-LIF measurements without PIV particles were analyzed quantitatively resulting in absolute OH concentrations and temperature fields. The flow field looks rather similar for both the unpiloted and the piloted cases, featuring a large recirculation zone next to the high momentum jet. However, flame shape and position change drastically. For the unpiloted case, the flame is lifted and widely distributed. Isolated flame kernels are found at the flame root in the vicinity of small-scale vortices. For the piloted flame, on the other hand, both pilot and main flame are attached to the burner base plate, and flame stabilization seems to take place on much smaller spatial scales with a connected flame front and no isolated flame kernels. The single-shot analysis gives rise to the assumption that for the unpiloted case, small-scale vortices act like the pilot burner flow in the opposed case and constantly impinge and ignite the high momentum jet at its root.

An optical flow algorithm based on gradient constancy assumption for PIV image processing

Particle image velocimetry (PIV) has matured as a flow measurement technique. It enables the description of the instantaneous velocity field of the flow by analyzing the particle motion obtained from digitally recorded images. Correlation based PIV evaluation technique is widely used because of its good accuracy and robustness. Although very successful, correlation PIV technique has some weakness which can be avoided by optical flow based PIV algorithms. At present, most of the optical flow methods applied to PIV are based on brightness constancy assumption. However, some factors of flow imaging technology and the nature property of the fluids make the brightness constancy assumption less appropriate in real PIV cases. In this paper, an implementation of a 2D optical flow algorithm (GCOF) based on gradient constancy assumption is introduced. The proposed GCOF assumes the edges of the illuminated PIV particles are constant during motion. It comprises two terms: a combined local-global gradient data term and a first-order divergence and vorticity smooth term. The approach can provide accurate dense motion fields. The approach are tested on synthetic images and on two experimental flows. The comparison of GCOF with other optical flow algorithms indicates the proposed method is more accurate especially in conditions of illumination variation. The comparison of GCOF with correlation PIV technique shows that the proposed GCOF has advantages on preserving small divergence and vorticity structures of the motion field and getting less outliers. As a consequence, the GCOF acquire a more accurate and better topological description of the turbulent flow.

Spatiotemporal analysis of turbulent jets enabled by 100-kHz, 100-ms burst-mode particle image velocimetry

100-kHz particle image velocimetry (PIV) is demonstrated using a double-pulsed, burst-mode laser with a burst duration up to 100 ms. This enables up to 10,000 time-sequential vector fields for capturing a temporal dynamic range spanning over three orders of magnitude in high-speed turbulent flows. Pulse doublets with inter-pulse spacing of 2 µs and repetition rate of 100 kHz are generated using a fiber-based oscillator and amplified through an all-diode-pumped, burst-mode amplifier. A physics-based model of pulse doublet amplification in the burst-mode amplifier is developed and used to accurately predict oscillator pulse width and pulse intensity inputs required to generate equal-energy pulse doublets at 532 nm for velocity measurements. The effect of PIV particle response and high-speed-detector limitations on the spatial and temporal resolution are estimated in subsonic turbulent jets. An effective spatial resolution of 266–275 µm and temporal resolution of 10 µs are estimated from the 8 × 8 pixel correlation window and inter-doublet time spacing, respectively. This spatiotemporal resolution is sufficient for quantitative assessment of integral time and length scales in highly turbulent jets with Reynolds numbers in the range 15,000–50,000. The temporal dynamic range of the burst-mode PIV measurement is 1200, limited by the 85-ms high-energy portion of the burst and 30-kHz high-frequency noise limit.

Detection of Passive Scalar Interface Directly from PIV Particle Images in Inhomogeneous Turbulent Flows

This article proposes a technique to estimate the cross-sectional scalar interface (outer boundary) in an inhomogeneous turbulent flow from a conditioned particle image velocimetry (PIV) experiment, which is suitable for medium to high Reynolds numbers. The scalar interface is estimated directly by using conditioned PIV particle images which have distinguishably high particle seeding density in the area of interest, whereas conventionally in water based experiments, scalar interface is often determined from planar laser induced fluorescence (PLIF) or equivalent dye images. By comparing quantities in the vicinity of this scalar interface, it also shows that in terms of separate turbulent and non-turbulent regions, this technique could also replace the function of PLIF images in water experiments, with slightly lower spatial resolution. At the same time, if velocity information is also required simultaneously then the cost of a separate camera-laser system can be saved. The effect of particle field inhomogeneity on the PIV accuracy can be well reduced to an insignificant level by an image local normalisation treatment. This article shows that the interfacial layer could be detected fairly accurately by enhancing the particle images by wavelet based thresholding methods. The degree of detection accuracy is quantified by synthetic particle image analyses, where a scalar interface can be artificially pre-defined. The proposed technique is tested in two water based experiments but is expected to be particularly useful in gas-phase based experiments or some combustion applications, where liquid-phase dye cannot be applied.

Maps of deformations in a cantilever beam using particle image velocimetry (PIV) and speckle patterns

PIV (particle image velocimetry) has been spreading in studies that use the movement of particles to monitor the displacement of an object or the flow of a fluid by means of velocity vectors using optical techniques and second order statistics. PIV is also known as laser speckle velocimetry when associated with speckle patterns. This technique has been used in works involving fluids, in general, building a map of velocity vectors representing the flow under analysis. This paper presents an approach by using PIV associated to speckle patterns for deformation measurements in a cantilever beam (ASTM A36 steel), one of the most common examples used in civil engineering, without the introduction of external particles. Results showed that the difference between PIV associated to speckle patterns and the analytic displacement values is increased along the beam length for a load of 1.96 N as an evidence of sensitivity of the proposed measurement method. This indicates that PIV is also capable for detecting displacement fields associated with laser speckle patterns in solid mechanics generating a map of deformation as an additional option for non-destructive tests.

Numerical Simulations of the University of Michigan Shock Boundary-Layer Interaction Experiments

Computational fluid dynamic (CFD) analyses of the University of Michigan (UM) Shock Boundary-Layer Interaction (SBLI) experiments were performed as an extension of the CFD SBLI Workshop held at the 48th AIAA Aerospace Sciences Meeting in 2010. In particular, the UM Mach 2.75 Glass Tunnel with a semi-spanning 7.75 degree wedge was analyzed in attempts to explore key physics pertinent to SBLI’s, including thermodynamic and viscous boundary conditions. A fundamental exploration pertaining to the effects of particle image velocimetry (PIV) on post-processing data is also shown. Results showed an improvement in agreement with experimental data with key contributions by adding a laminar zone upstream of the wedge (the flow is considered transitional downstream of the nozzle throat) and the necessity of mimicking PIV particle lag for comparisons. All CFD analyses utilized the OVERFLOW solver.

PIV camera response to high frequency signal: comparison of CCD and CMOS cameras using particle image simulation

We present a quantitative comparison between FlowMaster3 CCD and Phantom V9.1 CMOS cameras’ response in the scope of application to particle image velocimetry (PIV). First, the subpixel response is characterized using a specifically designed set-up. The crosstalk between adjacent pixels for the two cameras is then estimated and compared. Then, the camera response is experimentally characterized using particle image simulation. Based on a three-point Gaussian peak fitting, the bias and RMS errors between locations of simulated and real images for the two cameras are accurately calculated using a homemade program. The results show that, although the pixel response is not perfect, the optical crosstalk between adjacent pixels stays relatively low and the accuracy of the position determination of an ideal PIV particle image is much better than expected.

스프링클러 작동시 분무액적이 연기거동에 미치는 영향에 관한 연구

본 연구는 성능적인 피난안전설계에 있어서 스프링클러를 응용한 설계를 위하여 스모크로깅현상에 대한 연기층 하강기류에 대해 고찰하였다. 이를 위해 PIV(particle image velocimetry)장비를 이용하여, 국내 스프링클러 헤드별 분무액적특성을 실험적으로 고찰하고, 화원의 크기 및 스프링클러 헤드의 종류에 따른 연기층의 하강기류를 분석하였다. 그 결과, 평균입경, 입자속도, 살수분포에 관한 데이터를 얻었으며, 살수량과 입경과의 관계를 얻었다. 또한 하강기류에 관하여, 하강기류속도가 0.6 m/s이 되면 연기층이 하부층으로 이동하는 것을 확인하였으며, 100 kW 이상의 화원크기가 존재하여야 안정적인 하강기류가 발생하는 것을 확인하였다. This study contemplated the descending air current from the smoke layers related to the smoke logging phenomenon in the sprinkler applied design for effective evacuation safety design. To do this, using the PIV(Particle Image Velocimetry) device, the property of watering spray for each sprinkler head types available in Korea were experimented and the descending air current from smoke layers for each size of fire source or the type of sprinkler head was analyzed. As a result, database on the average particle diameter, particle velocity and distribution of sprinkling was obtained and the relationship between the water amount and particle diameter was obtained. Also, in relation to descending air current, the movement of smoke layer to the bottom at the descending air current velocity of 0.6m/s was observed and stable descending air current was observed in existence of fire source over 100kW in size.

덕티드 로켓의 이차 연소기 내에서 입자의 크기와 속도비가 유동 혼합에 미치는 영향

본 연구에서는 유동의 속도비와 입자의 크기가 덕티드 로켓의 2차 연소기 내의 유동 혼합 특성에 미치는 영향을 고찰하였다. 연구의 실험적 기법으로는 PIV(Particle Image Velocimetry)가 적용되고 수치적 방법으로는 LES(Large Eddy Simulation)가 사용되었다. PIV 입자로는 직경 <TEX>$5{\mu}m$</TEX>와 <TEX>$50{\mu}m$</TEX>의 폴리스틸렌이 각각 사용되었고, 속도비는 각각 5, 3, 1.5로 수행되어졌다. 고속유로를 통한 유동이 저속유로의 유동과 혼합되어 재부착점과 재순환 영역을 형성하였고, 속도비가 작을수록 재순환 영역이 커졌다. 큰 입자를 적용한 경우가 증가된 모멘텀에 의해 재순환 영역이 크게 나타났고, 유동 특성에 대한 속도비의 영향은 작은 입자를 적용한 경우에 비해 크지 않게 나타났다. In this study, the effect of velocity ratio and particle size on the flow mixing characteristics in the secondary combustor was investigated. Both PIV(Particle Image Velocimetry) technique and LES(Large Eddy Simulation) were applied. Two sizes of Polystyrene PIV seeding particle of 5 and <TEX>$50{\mu}m$</TEX>, and three velocity ratios of 5, 3, and 1.5 were considered. Results showed that the mixing of two air streams created reattachment and recirculation regions. The size of the recirculation region was decreased as the velocity ratio increased. For the larger particle cases, due to the increased momentum by the larger particles, the size of the recirculating regions were larger than that of the smaller particle cases and the effect of the velocity ratio was not as significant as in the smaller particle case.

In Situ PLIF and Particle Image Velocimetry Measurements of the Primary Entrainment Fuel Jet in a Naturally Aspirated Water Heater

The time-averaged characteristics of a fuel jet have been measured via acetone planar laser-induced fluorescence (PLIF) and particle image velocimetry (PIV) in the primary mixing region of an operating gas-fired water heater. These measurements allow for experimental characterization of the cross-sectional scalar and velocity fields as well as the estimation of the mass entrainment as the flow enters the burner in a practical system. In these experiments, reasonable results were obtained when only the fuel jet was seeded with acetone or PIV particles rather than the entire flow, thus demonstrating the potential for simplified experimental configurations in some applications where controlling or seeding the entire flow may be difficult. The entrainment characteristics of the fuel jet are compared with benchmarks from literature. The commercial device exhibits a larger mass entrainment rate than is found in typical free jets that have been studied in the literature. This may be a result of the jet's low Reynolds number (9,600) in comparison with other literature studies, and a result of initial conditions.