Raiola Marco, Lopez-Nunez Elena, Cafiero Gioacchino, Discetti Stefano Discetti (2020) Adaptive ensemble PTV,Measurement Science and Technology
Ensemble Particle Tracking Velocimetry (EPTV) is a method to extract high-resolution
statistical information on flow fields from Particle Image Velocimetry (PIV) images. The
process is based on tracking particles and extracting the velocity probability distribution
functions of the image ensemble in averaging-regions deemed to contain a sufficient number of
particle pairs/tracks. The size of the averaging regions depends on the particle density and the
number of snapshots. An automatic adaptive variation of the ensemble PTV is presented to
further push the spatial resolution of the method. The proposed Adaptive-EPTV is based on
stretching and orienting the averaging regions along the direction of maximum curvature of the
velocity fields. The process requires a predictor calculation with isotropic-window EPTV to
compute the second derivatives of the mean velocity components. In a second step, the principal
directions of the Hessian tensor are calculated to tune the optimal orientation and stretch of the
averaging regions. The stretching and orientation are achieved using a Gaussian windowing
with different standard deviation along the local principal direction of the Hessian tensor. The
algorithm is first validated using three different synthetic datasets: a sinusoidal displacement
field, a channel flow and the flow around a NACA 0012 airfoil. An experimental test case of an
impinging jet equipped with a fractal grid at the nozzle outlet is also carried out.
We address the important point of the proportionality between the longitudinal integral lengthscale (L) and the characteristic mean flow width (´) using experimental data of an axisymmetric wake and a turbulent planar jet. This is a fundamental hypothesis when deriving the self-similar scaling laws in free shear flow. We show that L/´ is indeed constant, at least in a range of streamwise distances between 15 and 50 times the characteristic inlet dimension. We revisit turbulence closure models such as the Prandtl mixing length and the eddy viscosity in the light of the non-equilibrium dissipation scaling. We show that the mixing length model does not comply with the scalings stemming from the non-equilibrium version of the theory even if it does comply with the theory's equilibrium version. Similarly, the eddy viscosity model holds in the case of the non-equilibrium version of the theory provided that the eddy viscosity is constant everywhere. We conclude by comparing the results of the different models with each other and with experimental data and with an improved model (following Townsend) that corrects for the eddy viscosity by considering the intermittency of the flow.
An experimental investigation focused on the manipulation of the wake generated by a square back car model is presented. Four continuously-blowing rectangular slot jets were mounted on the rear face of a 1:10 commercial van model. Load cell measurements evidence drag reduction for different forcing configurations, reaching a maximum of 12% for lateral and bottom jets blowing. The spectral analysis of the pressure fluctuations evidence, for all forced cases, an energy attenuation with respect to the natural case, especially close to the shedding frequency. An energy budget highlighted the most efficient forcing configurations accounting for both the drag reduction and the power required to feed the blowing system. Two main configurations are considered: the maximum drag reduction and the best compromise, yielding 5% drag reduction and a convenient energy balance. Particle Image Velocimetry (pPIV) and stereoscopic PIV (sPIV) experiments were performed allowing the three-dimensional reconstruction of the wake in the three considered configurations. Consistently with static and fluctuating pressure measurements, sPIV results reveal a dramatic change in the wake structure when the jets blow in the maximum drag reduction configuration. Conversely, the best compromise configuration reveals a wake structure similar to the natural one.
We report on a series of laboratory experiments in which we investigate the mixing in a wake produced downstream of an obstacle in a uniform flow. The fluid is confined within a channel of finite width, and the water depth is small compared with the channel width. The mixing appears to be dominated by dispersion caused by the circulation of the eddies that are shed alternately from each side of the obstacle. However, due to bottom friction, these eddies gradually dissipate downstream. In turn, the intensity of the cross-stream mixing of the tracer decays in the downstream direction, limiting the cross-stream extent of the tracer. We present a time-averaged picture of the experiments which illustrates the deviation of the time-averaged flow in the wake relative to the uniform flow upstream. We then develop a time-averaged model for the flow, using mixing length theory to account for the cross-channel momentum transfer as an eddy viscosity , where is the cross-channel integral of the perturbation in the along-channel speed associated with the wake. We also include a frictional stress to account for the bottom friction. The model predicts a similar pattern of variation of the along-channel velocity in both the along- and cross-channel directions to our experimental data. By matching the cross-channel data with the model, we find that the constant has value 0.2. We also analyse our experimental data to develop a time-dependent picture of the mixing of a stream of dye released into the wake. Using the model for the evolution of the flow, we develop a model for the time-averaged mixing, again based on mixing length theory. The model predicts a similar spatial distribution for the tracer in both the cross-stream and along-stream directions to that seen in our experimental data. By quantitative comparison of the model with the data, we find that the best fit of the empirical eddy diffusivity, , with the data occurs with . We discuss implications of our results for modelling cross-stream mixing in shallow turbulent flow.
A new passive method for the heat transfer enhancement of circular impinging jets is proposed and tested. The method is based on enhancing the mainstream turbulence of impinging jets using square fractal grids, i.e. a grid with a square pattern repeated at increasingly smaller scales. Fractal grids can generate much higher turbulence intensity than regular grids under the same inflow conditions and with similar blockage ratio, at the expense of a slightly larger pressure drop. An experimental investigation on the heat transfer enhancement achieved by impinging jets with fractal turbulence promoters is carried out. The heated-thin foil technique is implemented to measure the spatial distribution of the Nusselt number on the target plate. The heat transfer rates of impinging jets with a regular grid and a fractal grid insert are compared to that of a jet without any turbulator under the same condition of power input. A parametric study on the effect of the Reynolds number, the nozzle-to-plate distance and the position of the insert within the nozzle is carried out. The results show that a fractal turbulence promoter can provide a significant heat transfer enhancement for relatively small nozzle-to-plate separation (for a distance equal to 2 diameters an increase of 63% at the stagnation point, and 25% if averaged over an area of radius equal to 1 nozzle diameter, are respectively found with respect to the circular jet, against 9% and 6% for the regular grid in the same conditions of power input).
A method to extract turbulent statistics from three-dimensional (3D) PIV measurements via ensemble averaging is presented. The proposed technique is a 3D extension of the ensemble particle tracking velocimetry methods, which consist in summing distributions of velocity vectors calculated on low image density samples and then extract the statistical moments from the velocity vectors within sub-volumes, with the size of the sub-volume depending on the desired number of particles and on the available number of snapshots. The extension to 3D measurements poses the additional difficulty of sparse velocity vectors distributions, thus requiring a large number of snapshots to achieve high resolution measurements with a sufficient degree of accuracy. At the current state, this hinders the achievement of single-voxel measurements, unless millions of samples are available. Consequently, one has to give up spatial resolution and live with still relatively large (if compared to the voxel) sub-volumes. This leads to the further problem of the possible occurrence of a residual mean velocity gradient within the sub-volumes, which significantly contaminates the computation of second order moments. In this work, we propose a method to reduce the residual gradient effect, allowing to reach high resolution even with relatively large interrogation spots, therefore still retrieving a large number of particles on which it is possible to calculate turbulent statistics. The method consists in applying a polynomial fit to the velocity distributions within each sub-volume trying to mimic the residual mean velocity gradient.
A self-excited precessing jet PJ generated by a 5:1 expansion of a circular jet issuing at Red = 42,500 into a short coaxial cylindrical chamber has been investigated in a water facility by means of tomographic Particle Image Velocimetry. Two inflow conditions, using either simply a short-pipe nozzle, i.e. jet without grid JWG or placing a regular grid RG in correspondence of the short pipe exit, have been considered. A statistical analysis has been conducted for both configurations revealing that the entrainment region extends along the entire length of the cylindrical chamber. A modal analysis using proper orthogonal decomposition, conducted in the sudden expansion region SER at the basis of the cylindrical chamber, indicates the dominance of the large-scale precessing motion for both configurations. It is found that the entrainment process influences the instantaneous organization of the large-scale coherent structures in SER during the precessing motion. Under the influence of both the entrainment and the induced swirling motion, the helical coherent structures, detected within the first three diameters from the nozzle exit, undergo an asymmetric reduction of their convective velocity, causing in turn the bending of the jet axis. At the middle height of the cylindrical chamber, the crosstalk between the entrainment and the jet column produces vortex filaments with zero convective velocity observed around the jet column itself. At the exit of the cylindrical chamber, the entrainment process from one side is sustained by the fluid coming from the external ambient and from the other by the interaction of the emerging jet with the quiescent fluid, leading to the formation of a recirculation region.
The investigation focuses on the forcing of a fully developed turbulent channel flow through a linear array of synthetic jets injected tangentially to the wall and orthogonal to the mean flow direction. Forcing configurations are varied by differently combining the number of actuated jets working in an opposing blowing?suction configuration. Instantaneous wall shear stress and streamwise velocity fluctuations evidence drag reductions as well as turbulence attenuation up to 20%. The forcing effects are persistent up to at least a 150 half-channel height downstream of the injection section. Particle image velocimetry investigations in planes perpendicular to the channel axis highlight the presence of a large-scale streamwise vortical structure covering the whole height of the channel. This structure is thought to be responsible for the significant drag reduction, which is similar to the typical behavior evidenced in the case of colliding jets. The nondimensional forcing frequency of the synthetic jets producing the maximum drag reduction and turbulence attenuation is 0.0074 for the investigated Reynolds number (ReÄ=180). A statistical analysis of the near-wall structures demonstrates that the control mechanism acts in a way to reduce them in the forced configuration. It is conclude that the effect of the forcing is such that the near-wall structures merge and become less prone to inducing new structures, thus effectively reducing their number, and consequently the near-wall turbulence activity.
Passive methods are recognized as one of the most efficient means to achieve high heat and mass transfer in impinging jets. In a recent study, Cafiero et al. (2014) demonstrated the effectiveness of square fractal grids (SFGs, obtained repeating the same square pattern at increasingly smaller scales) in terms of heat transfer enhancement when locating the grid in correspondence of the nozzle exit section. Indeed, the capability of producing turbulence at multiple scales and the possibility of tuning the peak in the turbulence intensity profile as a function of the grid geometric parameters are both extremely appealing for heat transfer enhancement purposes. In this study, the effect of the grid geometry on the convective heat transfer rate of impinging jets is assessed and discussed. Three main effects are taken into account: the grid thickness ratio (obtained by varying the thickness of the first iteration of the SFG), the effect of the secondary grid iterations and the choice of the initial pattern. It is demonstrated how a larger thickness ratio, which in the present case corresponds to an anticipated location of the peak in the turbulence intensity profile, is beneficial to get a spotted high convective heat transfer rate at short nozzle to plate distances. Either the use of a single square grid, or the choice of a different initial pattern (for example a circular fractal grid) is instead indicated when it is desirable a uniform distribution of the convective heat transfer rate.
An experimental investigation of the flow field features of a round air jet equipped with a fractal (FG) and a regular (RG) grid insert impinging on a flat surface is carried out by means of 2D-2C Particle Image Velocimetry (PIV). The results are compared to those for a round jet without any grid (JWT). The test Reynolds number is set to 10, 000. The average flow fields and the turbulent kinetic energy distributions are presented. In particular, the effect of the presence of the fractal grid on the turbulence intensity distribution and on the planar component of the Reynolds stress is analyzed. Some differences between the location of the maximum of the turbulence intensity profile and the data reported in the literature are found. A possible interaction process between the wakes of the grids and the growing shear layer of the jet might be responsible of this discrepancy. A comparison between the flow field and the heat transfer results obtained by the authors in a previous work is also carried out. What is underlined is that both an higher turbulence level and a much stronger axial velocity cause an increment in the heat transfer rate.
Fractal grids (FGs) have been recently an object of numerous investigations due to the interesting capability of generating turbulence at multiple scales, thus paving the way to tune mixing and scalar transport. The flow field topology of a turbulent air jet equipped with a square FG is investigated by means of planar and volumetric particle image velocimetry. The comparison with the well-known features of a round jet without turbulence generators is also presented. The Reynolds number based on the nozzle exit section diameter for all the experiments is set to about 15 000. It is demonstrated that the presence of the grid enhances the entrainment rate and, as a consequence, the scalar transfer of the jet. Moreover, due to the effect of the jet external shear layer on the wake shed by the grid bars, the turbulence production region past the grid is significantly shortened with respect to the documented behavior of fractal grids in free-shear conditions. The organization of the large coherent structures in the FG case is also analyzed and discussed. Differently from the well-known generation of toroidal vortices due to the growth of azimuthal disturbances within the jet shear layer, the fractal grid introduces cross-wise disturbs which produce streamwise vortices; these structures, although characterized by a lower energy content, have a deeper streamwise penetration than the ring vortices, thus enhancing the entrainment process.
An experimental investigation was performed on the effect of fractal endplates on the wingtip vortex of a NACA 0012 semi span wing at a Reynolds number of 2 x 105. The endplates were obtained by introducing three different fractal patterns. Constant temperature anemometry and stereoscopic particle image velocimetry were employed to assess both the local flow properties as well as the spatial organization of the wingtip vortex. The results show that the introduction of a fractal endplate strongly affects both the geometry and the turbulence features of the vortex. In particular, it is found that the fractal geometry weakens the vortex by spreading the turbulent kinetic energy over a broader range of frequencies. We relate this loss of coherence to a faster dissipation of the vortex, thus paving the way to the employment of fractal endplates to reduce the hazard associated to such flow features.
We study the self-similarity and dissipation scalings of a turbulent planar jet and the theoretically implied mean flow scalings. Unlike turbulent wakes where such studies have already been carried out (Dairay et al. 2015 J. Fluid Mech. 781, 166-198. (doi:10.1017/jfm.2015.493); Obligado et al. 2016 Phys. Rev. Fluids 1, 044409. (doi:10.1103/PhysRevFluids.1. 044409)), this is a boundary-free turbulent shear flow where the local Reynolds number increases with distance from inlet. The Townsend-George theory revised by (Dairay et al. 2015 J. Fluid Mech. 781, 166-198. (doi:10.1017/jfm.2015.493)) is applied to turbulent planar jets. Only a few profiles need to be self-similar in this theory. The self-similarity of mean flow, turbulence dissipation, turbulent kinetic energy and Reynolds stress profiles is supported by our experimental results from 18 to at least 54 nozzle sizes, the furthermost location investigated in this work. Furthermore, the non-equilibrium dissipation scaling found in turbulent wakes, decaying grid-generated turbulence, various instances of periodic turbulence and turbulent boundary layers (Dairay et al. 2015 J. Fluid Mech. 781, 166-198. (doi:10.1017/jfm.2015.493); Vassilicos 2015 Annu. Rev. Fluid Mech. 95, 114. (doi:10.1146/ annurev-fluid-010814-014637); Goto & Vassilicos 2015 Phys. Lett. A 3790, 1144-1148. (doi:10.1016/j.physleta. 2015.02.025); Nedic et al. 2017 Phys. Rev. Fluids 2, 032601. (doi:10.1103/PhysRevFluids.2.032601)) is also observed in the present turbulent planar jet and in the turbulent planar jet of (Antonia et al. 1980 Phys. Fluids 23, 863055. (doi:10.1063/1.863055)). Given these observations, the theory implies new mean flow and jet width scalings which are found to be consistent with our data and the data of (Antonia et al. 1980 Phys. Fluids 23, 863055. (doi:10.1063/1.863055)). In particular, it implies a hitherto unknown entrainment behaviour: the ratio of characteristic cross-stream to centreline streamwise mean flow velocities decays as the -1/3 power of streamwise distance in the region, where the non-equilibrium dissipation scaling holds.
A method to enhance the quality of the tomographic reconstruction and, consequently, the 3D velocity measurement accuracy, is presented. The technique is based on integrating information on the objects to be reconstructed within the algebraic reconstruction process. A first guess intensity distribution is produced with a standard algebraic method, then the distribution is rebuilt as a sum of Gaussian blobs, based on location, intensity and size of agglomerates of light intensity surrounding local maxima. The blobs substitution regularizes the particle shape allowing a reduction of the particles discretization errors and of their elongation in the depth direction. The performances of the blob-enhanced reconstruction technique (BERT) are assessed with a 3D synthetic experiment. The results have been compared with those obtained by applying the standard camera simultaneous multiplicative reconstruction technique (CSMART) to the same volume. Several blob-enhanced reconstruction processes, both substituting the blobs at the end of the CSMART algorithm and during the iterations (i.e. using the blob-enhanced reconstruction as predictor for the following iterations), have been tested. The results confirm the enhancement in the velocity measurements accuracy, demonstrating a reduction of the bias error due to the ghost particles. The improvement is more remarkable at the largest tested seeding densities. Additionally, using the blobs distributions as a predictor enables further improvement of the convergence of the reconstruction algorithm, with the improvement being more considerable when substituting the blobs more than once during the process. The BERT process is also applied to multi resolution (MR) CSMART reconstructions, permitting simultaneously to achieve remarkable improvements in the flow field measurements and to benefit from the reduction in computational time due to the MR approach. Finally, BERT is also tested on experimental data, obtaining an increase of the signal-to-noise ratio in the reconstructed flow field and a higher value of the correlation factor in the velocity measurements with respect to the volume to which the particles are not replaced.
An experimental analysis of the flow field generated by an impinging jet equipped with a fractal grid insert located at the nozzle exit is carried out by means of Tomographic Particle Image Velocimetry. The Reynolds number based on the nozzle exit section diameter d is set to 15,000. The presence of the grid leads to a non uniform curvature of the jet shear layer. As a consequence, azimuthal instabilities are triggered in proximity of the nozzle exit section, causing the production of streamwise vortices. Furthermore, the organization of the coherent structures in proximity of the impinged wall is discussed and related to the convective heat transfer distribution. Owing to the presence of counter rotating wall vortices along the diagonals of the on-plate imprint of the fractal grid, a region of minimum in the scalar transfer map can be detected. In addition to that, similarly to the well known vortex rings that characterize the case of round jet without turbulence promoter, the presence of azimuthally coherent structures that might be generated by Kelvin-Helmholtz instability of the jet shear layer is presented and discussed.
In this work the effects due to the introduction of a 3-iterations square fractal grid inserted at the exit section of a synthetic jet actuator are experimentally analysed and discussed. The contributions related to the secondary iterations are assessed comparing the results with those obtained inserting a single-square grid (obtained by removing the 2nd and the 3rd iteration bars). The flow field generated by an axisymmetric synthetic jet is also investigated as a reference.
The synthetic jet is produced under three different dimensionless stroke lengths, ?0/D = 7.1, 19.2 and 50, covering flow regimes ranging from the nearly pure vortex to the nearly continuous jet. Results are reported both in terms of time-averaged and phase-averaged statistics. Notwithstanding the different nature of the jet flow, fractal turbulators confirm the interesting feature of an extended range of turbulence production, as already documented for continuous jets.
Phase-averaged results show that the vorticity field generated by the presence of the grid interacts with the vortex produced in the jet shear layer, causing its azimuthal distortion. This phenomenon can be prevented in the single-square grid case at lower ?0/D values, but it is still clear when the fractal one is located at the nozzle exit section.
Regardless of the inlet conditions, the outer vortex position scales with the stroke length. However, the introduction of the grid shows an effect on the lifetime of the vortex, with a faster dissipation in the fractal grid case.
In this work the effects due to the introduction of a 3-iterations square fractal grid inserted at the exit section of a synthetic jet actuator are experimentally analysed and discussed. The contributions related to the secondary iterations are assessed comparing the results with those obtained inserting a single-square grid (obtained by removing the 2nd and the 3rd iteration bars). The flow field generated by an axisymmetric synthetic jet is also investigated as a reference. The synthetic jet is produced under three different dimensionless stroke lengths, L0/D=7.1,19.2 and 50.0, covering flow regimes ranging from the nearly pure vortex to the nearly continuous jet. Results are reported both in terms of time-averaged and phase-averaged statistics. Notwithstanding the different nature of the jet flow, fractal turbulators confirm the interesting feature of an extended range of turbulence production, as already documented for continuous jets. Phase-averaged results show that the vorticity field generated by the presence of the grid interacts with the vortex produced in the jet shear layer, causing its azimuthal distortion. This phenomenon can be prevented in the single-square grid case at lower L0/D values, but it is still clear when the fractal one is located at the nozzle exit section. Regardless of the inlet conditions, the outer vortex position scales with the stroke length. However, the introduction of the grid shows an effect on the lifetime of the vortex, with a faster dissipation in the fractal grid case.
A circular jet flow past an abrupt expansion under some conditions switches intermittently between two states: quasi-axisymmetric expansion and gyroscopic-like precessing motion. In this work, an experimental investigation into the self-excited precessing flow generated by a 5:1 expansion of a round jet in a coaxial cylindrical chamber is carried out by means of tomographic particle image velocimetry. The experiments are performed on a jet issued from a short pipe at a Reynolds number equal to 150,000. Proper orthogonal decomposition (POD) is applied to extract information on the organization of the large coherent structures of the precessing motion. The application of this technique highlights the dominance of three modes: the most energetic two are associated with the jet precession; the third one is representative of the axial motion. An estimate of the precession probability based on the modal energy obtained from the application of POD is proposed. The precession frequency is extracted using a low-order reconstruction (LOR) of a subset of the POD modes. The reconstructed flow field topology obtained by the LOR highlights an underlying mechanism of swirl generation in proximity of the inlet nozzle; the phenomenon is closely related to the interaction between the entrainment in the far field and the recirculation regions in the near field. The application of a stability criterion shows that the self-induced swirl flow results to be unstable. The instability is responsible for the generation of helicalshaped vortices in the near field, even though the dominant feature for the unconfined jet issued from the same nozzle is the axisymmetric ring-vortices generation.