Hostname: page-component-7b7567c8c-2n9c9 Total loading time: 0 Render date: 2026-11-16T08:09:20.678Z Has data issue: false hasContentIssue false

Experimental study of negatively buoyant finite-size particles in a turbulent boundary layer up to dense regimes

Published online by Cambridge University Press:  13 March 2019

Lucia J. Baker*
Affiliation:
Department of Aerospace Engineering and Mechanics, University of Minnesota, Minneapolis, MN 55455, USASt. Anthony Falls Laboratory, University of Minnesota, Minneapolis, MN 55414, USA
Filippo Coletti
Affiliation:
Department of Aerospace Engineering and Mechanics, University of Minnesota, Minneapolis, MN 55455, USASt. Anthony Falls Laboratory, University of Minnesota, Minneapolis, MN 55414, USA
*
†Email address for correspondence: bake0616@umn.edu

Abstract

We experimentally investigate the two-phase interplay in an open-channel turbulent boundary layer laden with finite-size particles at global volume fractions between 4 and 25 %. The working fluid (water) and the dispersed phase (hydrogel spheres) have closely matched refractive indices, allowing us to measure the properties of both phases using particle image velocimetry and particle tracking velocimetry, respectively. The particles have a diameter of approximately 9 % of the channel depth and are slightly denser than the fluid. The negative buoyancy causes a strong vertical concentration gradient, characterized by discrete and closely spaced particle layers parallel to the wall. Even at the lowest considered volume fractions, the near-wall fluid velocity and velocity gradients are strongly reduced, with large mean shear throughout most of the channel height. This indicates that the local effective viscosity of the suspension is greatly increased due to the friction between particle layers sliding over one another. The particles consistently lag the fluid and leave their footprint on its mean and fluctuating velocity profiles. The turbulent activity is damped near the wall, where the nearly packed particles disrupt and suppress large-scale turbulent fluctuations and redistribute some of the kinetic energy to smaller scales. On the other hand, in the outer region of the flow where the local particle concentration is low, the mean shear produces strong Reynolds stresses, with enhanced sweeps and ejections and frequent swirling events.

Information

Type
JFM Papers
Copyright
© 2019 Cambridge University Press 
Figure 0

Figure 1.Schematic diagram of the water channel showing key components and dimensions. The bold arrow indicates the direction of the flow.

Figure 1

Figure 2.Instantaneous realization of the particle-laden flow. The raw image is shown in (a), where both the large spherical particles and the small flow tracers are visible. The velocity field and detected particles, obtained using the methods described in the text, are shown in (b).

Figure 2

Table 1.Properties of the hydrogel particles. $D_{p}$ is the mean diameter, $D_{p}^{+}$ is the diameter normalized by the viscous length scale of the unladen flow, $\unicode[STIX]{x1D70C}_{p}$ is the density, $e$ is the coefficient of restitution, $V_{t}$ is the terminal velocity, $Re_{p,V_{t}}$ is the Reynolds number based on the terminal velocity, $m$ is the approximate coefficient of friction for the range of sliding velocities experienced by the particles in this experiment from Cuccia (2017), $E$ is the elastic modulus and $\unicode[STIX]{x1D6F7}_{0}$ is the global volume fraction for the different cases in this study.

Figure 3

Figure 3.Probability density function of hydrogel particle diameters. The standard deviation is 9 % of the mean value.

Figure 4

Figure 4.Vertical position (a) and velocity (b) time series of hydrogel particles dropped from rest in a tank of quiescent water and rebounding off the bottom, averaged over 10 trials.

Figure 5

Table 2.Imaging and PIV processing parameters.

Figure 6

Figure 5.Component of the random error in the PIV of the fluid velocity, due to the decreased imaged quality in the presence of the hydrogel particles (see Zhang & Rival 2018).

Figure 7

Figure 6.Threshold method for particle detection used for the near-wall data: (a) original image, (b) low-pass filtered image, (c) binarized image, (d) particle detected.

Figure 8

Figure 7.Mean streamwise velocity profile of the unladen flow, in (a) outer units and (b) wall units. Grey and black symbols represent data from the 50 mm lens and 200 mm lens measurements, respectively. The dashed line in (b) indicates the logarithmic law fit. Error bars on the velocity are smaller than the symbol width and are not shown. The profiles are compared with Tachie et al. (2003) in (a) and De Graaff & Eaton (2000) in (b), both shown in blue open circles.

Figure 9

Table 3.Physical parameters of the water channel and unladen boundary layer properties. $U_{\infty }$ is the free-stream velocity, $\overline{U}$ is the depth-averaged velocity, $H$ is the water depth, $w$ is the channel width, $\unicode[STIX]{x1D6FF}_{99}$ is the boundary layer thickness and $u_{\unicode[STIX]{x1D70F}}$ is the friction velocity. The depth-averaged velocity is defined as $\overline{U}=1/H\int _{0}^{H}u(y)\,\text{d}y$, the boundary thickness is defined such that $u(\unicode[STIX]{x1D6FF}_{99})=0.99U_{\infty }$ and the friction velocity is determined by fitting (3.1) to the mean fluid velocity profile. $Re_{b}=\overline{U}H/\unicode[STIX]{x1D708}$, $Re_{\unicode[STIX]{x1D70F}}=u_{\unicode[STIX]{x1D70F}}\unicode[STIX]{x1D6FF}_{99}/\unicode[STIX]{x1D708}$ and $Re_{\unicode[STIX]{x1D703}}=U_{\infty }\unicode[STIX]{x1D703}/\unicode[STIX]{x1D708}$ are the depth-averaged, friction and momentum thickness Reynolds numbers, respectively. Standard water properties at $22\,^{\circ }\text{C}$ are used in the calculations.

Figure 10

Figure 8.Profiles of streamwise turbulent normal stress (a), wall-normal normal stress (b) and shear stress (c) of the unladen fluid in wall units. Grey and black symbols represent data from full-channel imaging and near-wall imaging, respectively. Error bars on the velocity are smaller than the symbol width and are not shown. The profiles are compared with De Graaff & Eaton (2000) shown in blue open circles.

Figure 11

Figure 9.Sample raw images of the particle-laden cases: (a) $\unicode[STIX]{x1D6F7}_{0}=4\,\%$, (b) $\unicode[STIX]{x1D6F7}_{0}=8\,\%$, (c) $\unicode[STIX]{x1D6F7}_{0}=13\,\%$, (d) $\unicode[STIX]{x1D6F7}_{0}=20\,\%$ and (e) $\unicode[STIX]{x1D6F7}_{0}=25\,\%$.

Figure 12

Figure 10.Profiles of mean local volume fraction for the particle-laden cases.

Figure 13

Figure 11.Spatial autocorrelations of instantaneous particle concentration. The autocorrelation is compared between (a) different volume fractions in the bottom particle layer ($y/H=0.04$), and (b) different wall-normal locations for the $\unicode[STIX]{x1D6F7}_{0}=13\,\%$ case. The five wall-normal coordinates, when normalized by the particle diameter, are approximately $y/D_{p}=0.5$, 1.5, 2.5, 3.5 and 4.5.

Figure 14

Figure 12.Profiles of inertial number for the different volume fraction cases. The vertical dashed line shows the location of $I=1$ where the dynamics transitions from frictional to collisional.

Figure 15

Figure 13.Mean fluid streamwise velocity profiles. The velocities are normalized by the free-stream velocity of the unladen case, $U_{\infty ,0}$.

Figure 16

Figure 14.Profiles of Bagnold number for the different volume fraction cases. The vertical dashed line shows the location of $Ba=450$ above which the dynamics transitions to the grain-inertia regime.

Figure 17

Figure 15.Compensated mean streamwise velocity profiles in single-phase wall units showing the existence of a logarithmic region in all cases (horizontal dashed line). The vertical dashed lines indicate the location of the first two particle layers. The superscript 0 indicates the use of single-phase wall units.

Figure 18

Figure 16.Mean particle velocity profiles (triangles) compared with the mean fluid velocity profiles (dots). The velocities are normalized by the free-stream velocity of the unladen case.

Figure 19

Figure 17.Profiles of particle Reynolds number based on (a) the local slip velocity and (b) local shear rate.

Figure 20

Figure 18.Profiles of the fluid streamwise (a) and wall-normal (b) r.m.s. velocity fluctuations and Reynolds shear stresses (c), normalized by the depth-averaged velocity.

Figure 21

Figure 19.Profiles of shear production normalized by the depth-averaged velocity and the channel height.

Figure 22

Figure 20.Profiles of streamwise (a) and wall-normal (b) r.m.s. velocity fluctuations of the particles (triangles) compared to the fluid (dots).

Figure 23

Figure 21.The particle-centred mean relative velocity field, in particle-centred coordinates, for particles from the $\unicode[STIX]{x1D6F7}_{0}=4\,\%$ case whose wall-normal position is in the range $0, and (b) detailed view of the region indicated by the black rectangle in (a).

Figure 24

Figure 22.Joint PDFs of streamwise and wall-normal fluctuating fluid velocities at four different wall-normal distances throughout the channel height. Contour units are arbitrary, and contours are cut off at the same level in all plots.

Figure 25

Figure 23.(a) PDFs of signed swirling strength $\unicode[STIX]{x1D706}_{ci,s}$ for each case, where $\unicode[STIX]{x1D706}_{ci,s}>0$ represents retrograde motion and $\unicode[STIX]{x1D706}_{ci,s}<0$ represents prograde motion, and (b) the fraction of prograde events plotted for each case as a function of wall-normal distance.

Figure 26

Figure 24.Spatial correlations of the streamwise fluctuating velocity plotted at four wall-normal distances for each case. Contours are cut off at a level of 0.7 in all plots.

Figure 27

Figure 25.Normalized second-order velocity structure functions each case at (a) a near-wall location ($y/H=0.1$, $y/D_{p}=1.2$) and (b) a location away from the wall ($y/H=0.7$, $y/D_{p}=8.1$).