7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
TimeResolved PIV Measurements of ParticleFlow Interactions in a Turbulent Boundary
Layer
Rene van Hout
Technion Israel Institute of Technology, Faculty of Mechanical Engineering
Technion city, Haifa, 32000, Israel
rene@ttechnion.ac.il
Keywords: Timeresolved PIV, Particleflow interaction, polystyrene spheres, Vorticity, Swirling strength
Abstract
Timeresolved (highspeed) Particle Image Velocimetry (PIV) measurements were carried out in a dilute particleladen flow to
track both the movement of polystyrene spheres (D, = 583.0 14.4 jim, p, = 1050 kg/m3) as well as resolve the instantaneous velocity field
of the turbulent water flow transporting the spheres. PIV images were taken at a sampling rate of 1 kHz, corresponding to a time scale one
order of magnitude smaller than the flow's Kolmogorov time scale. The measurements were performed in a closed loop, transparent, square
water channel facility with cross section of 50x50 mm2 and a length of 2 m. Measurements were taken in a vertical plane (29.3x29.3 mm2)
oriented along the channel's centerline at a bulk water velocity of 0.3 m/s. The polystyrene spheres were imaged together with the flow
tracers (hollow glass spheres, 10gm) and were discriminated based on their size difference. Results show that particle Reynolds numbers
based on the relative velocity between the particle and the fluid are below 40, indicating a closed wake without flow separation. Results
further indicate a clear effect of the instantaneous vortical structures on the particle movement, ejections pushing particles upwards and
sweeping motions directing particles downwards in agreement with DNS studies. Instantaneous sequences of polystyrene spheres plotted
on the spatial velocity, vorticity and swirling strength distribution clearly show the effect of turbulent flow structures and resulting particle
movement. Examples are provided of a particle swept upwards by a hairpin structure and another particle pushed upwards by a
counterrotating vortex pair.
Introduction
Understanding the dynamics of particles (solids, droplets or
bubbles) in turbulent flows and their interaction is crucial in
many industrial and environmental processes. Some
examples are the flow in chemical reactors and combustion
chambers, drag reduction by microbubbles, deposition of
sediments on riverbanks, aerosol dispersal and cloud
physics (e.g. Eaton & Fessler 1994, Nezu & Azuma 2004,
Ferrante & Elgobashi 2004, Soldati 2005).
Particle dispersion has traditionally been treated in a
statistical sense assuming Fickian diffusion. This type of
time averaged analysis causes an initially uniform particle
distribution to remain uniformly (Poisson) distributed.
However, it is well known that turbulence is characterized
by semiorganized structures (Robinson 1991) that can
cause nonuniformity in the particle density field even for
initially uniformly distributed particles (e.g. Wang & Maxey
1993, Aliseda et al. 2002, Yang & Shy 2005). In addition,
particles can either increase or decrease the turbulence level
of the carrier fluid (Kaftori et al. 1995a & b, Crowe 2000).
Interactions between particles and turbulence are complex
and depend on several parameters like the ratio of particle
diameter to a typical flow length scale, the particle Reynolds
number, the Stokes number and the mass loading of the
particles. A combination of these dictates the nature of the
coupling between particles and the turbulent flow field. One
of the most important parameters is the Stokes number, St,
defined as the ratio between the particle's response time, rp,
and a characteristic time scale of the flow, e.g. the
Kolmogorov time scale k, St = /rk. The Stokes number
indicates how well a particle follows the fluid streamlines,
i.e. particles with small Stokes numbers will act as passive
flow tracers while particles with very large Stokes numbers
will be unaffected by turbulent fluctuations. The way that
particles and turbulence interact has been classified into
different regimes occurring with increasing particle
concentration (Elgobashi 1994):
(i) Oneway coupling: the particles are entrained by the
fluid but do not affect the carrier fluid turbulence.
(ii) Twoway coupling: the turbulence structure of the
carrier fluid is altered by the momentum transfer of the
particles.
(iii) Fourway coupling: particleparticle collisions take
place in addition to fluidparticle interactions.
Oneway coupling in particular and also twoway coupling
have been widely studied using numerical simulations such
as DNS and LES but to a much lesser extent by
experimentalists. DNS is limited to small Reynolds number
flows (Marchiola & Soldati 2002; Portela & Oliemans
2003; Soldati 2005) while LES can be used for high
Reynolds number flows and is commonly applied to model
atmospheric boundary layer flows (Piomelli & Balaras
2002). However, in both methods, the particlefluid
coupling needs to be modeled, mostly by introducing an
additional force term in the fluid momentum equation that
acts on the fluid while particle trajectories are determined
by Lagrangian particle tracking (e.g. Yamamoto et al. 2001;
Portela & Oliemans 2003; Ferrante & Elghobashi 2004).
The correct specification of the fluidparticle forcing is
essential to accurately model particle trajectories. During
the last few years, DNS studies have revealed that coherent
structures are instrumental in the entrainment of inertial
particles. Soldati (2005), who used DNS and Lagrangian
particle tracking showed a strong correlation between
sweeps and inward particle motion and ejections and
outward particle motion. However, much remains to be
answered about the underlying physical mechanisms by
which particles interact with the turbulent flow structure
(Maxey et al. 1996; Kaftori et al. 1998; Soldati 2005). For
example, it has been proposed that an increase in turbulence
level is achieved by enhanced mixing either caused by
inertial particles moving from one eddy to another (e.g. Mei
et al. 1991) or by vortex shedding (Hetsroni 1989; Crowe
2000). Similarly, different mechanisms have been suggested
for the suppression of turbulence (Crowe 2000).
Despite the significance of particleladen flows, relatively
few experimental studies have been performed due to the
difficulty of both temporally and spatially resolving the flow
as well as the particle motion. In recent years, the
introduction of Phase Doppler Anemometry (PDA) has
enabled the simultaneous measurement of particle and fluid
velocities (e.g. Kussin & Sommerfeld 2002; Aliseda et al.
2002). However, PDA is limited to spherical particles and
since it is inherently a point measurement, it does not
spatially resolve the instantaneous velocity field. It therefore
does not provide any information on spatial derivative based
quantities like vorticity and strain rate.
Spatial measurement techniques such as Particle Image
Velocimetry (PIV) are especially suited to provide velocity
field information. PIV has been applied by a few researchers
to particleladen flows and has the potential to spatially and
temporally resolve particle turbulence interactions in a
measurement plane (Khalitov & Longmire 2003, Yang &
Shy 2005, Borowski & Wei 2006).
Two experimental studies that highlight the potential of
applying spatially resolved 2DPIV to particleladen
turbulent boundary layer flows were carried out by Kiger &
Pan (2002) and Khalitov & Longmire (2003). Particles and
fluid tracers were discriminated based on their image size.
Khalitov & Longmire (2003) demonstrated the Stokes
number ranges and spatial ranges over which particles were
correlated with turbulent structures. Kiger & Pan (2002)
showed that turbulence properties of the carrier fluid were
noticeably changed even for dilute suspensions and they
related upward moving particles with ejection events. Both
Kiger & Pan (2002) and Khalitov & Longmire (2003) did
not temporally resolve the particle fluid motions.
There are very few experimental studies that have quantified
both spatially and temporally the interaction of the particles
with the turbulent structures. An exception is Sato et al.
(2000) who measured both particle and fluid motion using a
highspeed 2D PIV setup in which the camera moved at the
mean particle velocity thus providing a Lagrangian frame of
reference. The experiments were carried out in a downward
flow, vertical water channel facility. Turbulence modulation
was related to the ratio of particle size to kolmogorov scale
and to strain rate and vorticity in the vicinity of the
particles.
The goal of this research is to gain further insight into
particle behavior in turbulent boundary layers by studying
turbulence structures and particle dynamics. Time resolved
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
PIV measurements were performed to study the entrainment
and deposition of polystyrene spheres in a turbulent
boundary layer, resolving both the flow field as well as the
particle motion.
Nomenclature
Diameter of polystyrene spheres (m)
Stokes number
Time (s)
Fluctuating streamwise velocity (m s')
Instantaneous streamwise velocity (m s')
Fluctuating wallnormal velocity (m s')
Instantaneous transverse velocity (m s')
Vortex core
Streamwise coordinate (m)
Wall normal coordinate (m)
Transverse coordinate (m)
Greek letters
L Swirling strength (S2)
A Directional swirling strength (S2)
p Density (kg m 3)
T Time scale (s)
co Vorticity (s')
Subscripts
14 Vortex indices
ci Complex, imaginary
f Fluid
k Kolmogorov
p particle
co Free stream
Abbreviations
DNS Direct numerical simulations
FOV Field of view
LES Large eddy simulations
PIV Particle Image Velocimetry
Experimental Facility
The present experiments were performed in a square, closed
loop, water channel flow facility consisting of a centrifugal
pump, magnetic flow meter and an inlet section comprised
of a honeycomb and contraction section (9:1, see Fig. 1).
Optics
S ID \
Highspeed
laser (Nd:Yag)
(b) 0
50 mm
FOV
r
(c) 29.3 mm
(c) z ;'
Figure 1: Schematic of experimental setup (a) top view
(b) crosssectional view of square channel (c) coordinate
system
The 2 m long test section had an internal cross sectional area
of 50 x 50 mm2 and was made of glass to ensure optical
access from all sides. At the end of the test section an end
diffuser was connected to a large tank. Access into the test
section was provided by two, internally flush mounted
acrylic lids on top of the test section. A frequency controlled
centrifugal pump was used to set the test section flow
velocity to U0 = 0.3 m/s. Uncertainties in the free stream
velocity were less than 2%. The employed coordinate
system is presented in Figure Ic where x, y and z are the
streamwise, wallnormal and transverse directions,
respectively. The instantaneous streamwise and wallnormal
velocities are denoted by U and V, respectively, and the
fluctuating velocities by u and v.
Experiments were carried out using a highspeed PIV
system (LaVision GmbH) consisting of a CMOS camera
(Photron, 1024x1024 pixels @ 1000 fps), an Nd:YAG laser
(New Wave) and laser sheet optics (see Figure 1). Neutrally
buoyant hollow glass spheres (10 utm) were used as PIV
flow tracers. Measurements of both the flow field and the
particles' motion were made in a vertical plane positioned at
the channel's centerline in the streamwise direction. The
Field Of View (FOV) was 29.3 x 29.3 mm2. The camera's
frame rate was set to 1000 fps to ensure PIV time series
data.
Data Processing
An example of a part of a PIV image that includes
polystyrene spheres is displayed in Figure 2. Initially, the
polystyrene spheres were separated from the PIV tracers by
applying a median filter and subsequently subtract the result
from the original PIV image. This procedure also reduces
background noise and spatial nonuniformities. In the next
step, spheres entering the FOV were identified and tracked
using a direct crosscorrelation procedure: polystyrene
sphere's images were cropped in successive PIV images,
median filtered to remove the small PIV flow tracers and
subsequently crosscorrelated. The crosscorrelation values
in the vicinity of the peak value were three point Gaussian
fitted in order to obtain subpixel accuracy (Raffel 1998).
After tracking all spheres, the spheres' centroid position
information was used to generate circular position masks
having twice the sphere's radius in order to cancel out near
sphere effects on the surrounding flow. This mask was
multiplied with the median filtered PIV images.
Figure 2: Part of PIV
polystyrene spheres.
image including multiple
PIV data processing was then carried out using a threepass
cross correlation algorithm (FFTbased, LaVision GmbH)
with decreasing interrogation window size; the final step
32 x 32 pixel interrogation windows with 50% overlap
resulting in a vector spacing of 0.311 mm. Instantaneous
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
vectors had an error of less than 0.1 pixels. The resulting
vector maps were used to calculate the outofplane
component of the vorticity (from now on termed as the
vorticity), co =(9V/9x 9U/Ty) Derivatives were
calculated using a centerdifference scheme. As a means to
detect vortex cores, we used the swirling strength (e.g. Zhou
et al. 1999), which is a well accepted parameter to detect
vortical structures in turbulent flows. It is based on the
imaginary part of the eigenvalue of the velocity gradient
tensor being larger than zero, 2c, > 0 (Zhou et al. 1999). In
the case of 2DPIV, not all terms of the local velocity
gradient tensor are known and in this case the eigenvalue is
determined from an equivalent 2D tensor (Adrian et al.
2000). Here, we follow the work of Wu & Christensen
(2006) and define a directional swirling strength parameter
(from now on termed as the swirling strength) that retains
the vortex rotation direction, Ac, = lc,a/l l. Note that the
swirling strength is preferable over the vorticity for the
detection of vortical motions since it does not reveal areas
that have significant vorticity such as shear layers, but lack
any local swirling motion.
Results and Discussion
As a typical example, we track three different particles, P1
to P3 (indicated by the colors red, blue and green,
respectively) as they are advected by the turbulent water
flow. The trajectories of these particles are displayed in
Figure 3. P1 and P3 clearly move upwards while P2 remains
more or less at the same height although being initially quite
close to P1.
7
6  t = 0.065s
7 \
5
 4
t =Os
t = 0.098s
E
3  t = Os _ll. __":1lY'
2 t = 0.146s
0 o P3 A P1 P2
t1 = 0.016s
0 5 10 15 20 25 30
x [mm]
Figure 3: Example of three particle trajectories.
A time sequence at reduced temporal resolution (At = 0.020
s) of particle positions plotted on top of the fluctuating fluid
velocity distributions (Reynolds decomposed), u(x,y) and
v(x,y), is displayed in Figure 4. For clarity every second
vector is displayed. Particles P1 and P2 enter the field of
view in Figure 4a and it can be observed that P1 is subjected
to higher fluctuating velocities than P2 that is located just
underneath it. As the two particles are advected downstream
P1 is pushed upwards by about one and a half diameters
while P2 remains approximately at the same level (Figure 3).
We will see in the swirling strength maps that P1 seems to
be pushed upwards by a slightly downstream located
counterrotating vortex pair, possibly the spatial signature of
a vortex ring. Note that although P2 is located closer to the
 ... .... ............ n ..... ..............
wall, it moves faster than P1 in the streamwise direction. P1
moves out of the laser sheet plane between Figure 4d and e.
Polystyrene sphere P3 enters the field of view in Fig. 4b and
is located about 3 particle diameters above the channel
bottom. It can be observed that P3 is continuously subjected
to a sweeping motion, u < 0 and v > 0 and during the
tracking time it is lifted up by about two particle diameters.
15
U 0 05 m/s . ,, .
10
5 ...... i i I . ,
5_ ........ ...... ..... ...
(a) ..
15 , \ ... .. 
UO 05 ms . , .. , ,
I I\ \ N\ \ ^ ~.. . t t ..
15 ' .  
(b) o0~ .. .
15
U= 05 m/s .,
''.\' N' N N> s\.\
.. \N** _..; .... . \
15. .., ,_'t, . .... .. .,
10
. . . . ..  .. l \ .\ . . .
E
S 
(e) o
U 0 m '5 m
10 5 10 15 20 25
x [mm]
Figure 4: Example of sequence (At = 0.020 s) of particle
positions plotted on top of velocity fluctuations. Mean flow
_ __ *' ^ ^^^^^_
direction is from left to right. Particles are imaged by filled
circles, red: P1, blue P2, green P3.
The instantaneous streamwise and wallnormal particle
velocities and fluid velocities interpolated to the particle's
,;^ > ,... ^ ^
position are shown in Fig. 5 for P3. The instantaneous
velocities show quite some scatter and are smoothed using a
weighted running average algorithm (Matlab, "rloess") that
does not take outliers into account. Running averaged
velocities are indicated by <..>. It is seen that the
instantaneous particle streamwise velocity lags behind the
fluid streamwise velocity for most of the tracking time. The
wall normal particle and fluid velocities are small, upward
directed and more or less of the same magnitude. Particle
Reynolds numbers based on the relative velocity are below
40, indicating a closed wake structure. Note that velocity
40, indicating a closed wake structure. Note that velocity
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
maps presented in Figure 4 are an inconvenient means for
visualizing the instantaneous flow structures (Adrian et al.
2000) and we therefore present the instantaneous vorticity
and swirling strength distributions to visualize the shear
layers and vortex cores.
0.3
0.2 
(0 0
a Uf A
D Up
0.05 t [s] 0.1
0.05
0
0.05
(b) 0.1
Figure 5: Instantaneous particle and fluid velocities for P3.
(a) Streamwise velocity (b) wall normal velocity. Blue and
red symbols represent running averaged data sets.
Figures 6 and 7 display an instantaneous sequence (At =
0.020 s) of the vorticity, ct, and the swirling strength, Ac,,
respectively, together with the particle positions. Particles P1
and P2 are advected downstream and it can be observed that
slightly downstream of P1 a counterrotating vortex pair is
located (V1, V2 indicated by dashed rectangle), clearly
observed in both the vorticity and swirling strength maps of
Figs. 6c,d and 7c,d. This counterrotating vortex pair is
possibly the spatial signature of a vortex ring pinchedoff
from near wall hairpin structures (e.g. Wu & Christensen
2006). The vortex pair is located such that the flow "pushed
out" by the vortex pair affects P1 and "pushes" it in the
upward direction as indicated by the fluctuating velocity
distribution presented in Fig. 4. On the other hand, P2 is
positioned slightly below the counterclockwise rotating
vortex (V2) and is not exposed to the vortex ring outflow.
Particle P3 is positioned close to the high shear layer near to
the bottom wall. Interestingly, just upstream of P3, a near
wall structure that lifts up from the bottom wall can be
observed in the vorticity maps (Figs. 6ce). This structure is
reminiscent of hairpin structures that are commonly
observed in turbulent boundary layers (Robinson 1991).
Note that the clockwise rotating vortex V3 is most probably
associated with the uplifted hairpin structure. It is clearly
observed in Fig. 6 that the hairpin moves faster than the
particle and "envelopes" it further downstream. As a result,
P3 is exposed to an ejection motion induced by the legs of
the hairpin, causing it to move away from the wall.
0 Vf 0
o Vp o
0.05 0.1 0.
t[s]
0.1
S0. 1
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
150
10
5 <
0
15 ,
0
15 
0 C
0
lo O V
V3
structure V
"Hairpin" V3
5
0 5 10 15 20 25
x [mm]
Figure 6: Example of sequence of particle advection
plotted on top of vorticity maps (At = 0.020 s). Particles are
imaged by filled circles, red: P1, blue P2, green P3.
Summary and Conclusions
Time resolved Particle Image Velocimetry measurements of
transport of polystyrene spheres (D, = 583.0 + 14.4 utm) in a
turbulent boundary layer were performed. The spheres were
discriminated from the flow tracers based on size and the
sphere positions were tracked by crosscorrelating
subsequent particle images. At the determined sphere
positions, the PIV images were masked and subsequently
processed in order to obtain the turbulent flow field. The
effect of instantaneous flow structures was shown by
plotting the polystyrene sphere positions on the
instantaneous vorticity and swirling strength distributions. It
was shown that a counterrotating vortex pair, possibly the
spatial signature of a vortex ring was instrumental in
"pushing" upwards one of the spheres. In another case, a
near wall particle was "caught" by a hairpin structure and
10
00
15 0
10
3 5g
5 o*
0 5 10 15 20 25
x [mm]
Figure7: Example of sequence of particle advection
plotted on top of swirling strength maps (At = 0.020 s).
Particles are imaged by filled circles, red: P1, blue P2, green
P3.
exposed to an ejection motion resulting in movement away
from the bottom wall.
Acknowledgements
This work was supported in part by the Edmund J. Safra
Philanthropic Foundation, the Wolfson Family Charitable
Trust and the Technion Fund for Promotion of Research.
References
Adrian, R.J., Christensen, K.T. & Liu, Z.C. Analysis and
interpretation of instantaneous turbulent velocity fields. Exp.
Fluids, Vol. 29, 275290 (2000)
Aliseda, A., Cartellier, A., Hainaux, F. & Lasheras, J.C.
Effect of preferential concentration on the settling velocity
of heavy particles in homogeneous isotropic turbulence. J.
Fluid Mech., Vol. 468, 77105 (2002)
Borowsky, J. & Wei, T Simultaneous
Velocimetry/accelerometry measurements in a turbulent
twophase pipe flow. Exp. Fluids, Vol. 41, 1320 (2006)
Crowe, C.T. On models for turbulence modulation in
fluidparticle flows. Int. J. Multiphase Flow, Vol. 26,
719727 (2000)
Eaton, J.K. & Fessler, J.R. Preferential concentration of
particles by turbulence. Int. J. Multiphase Flow, Vol. 20,
169209 (1994)
Elghobashi, S. On predicting particleladen turbulent flows.
Appl. Sci. Res., Vol. 52, 309329 (1994)
Ferrante, A. & Elghobashi, S. On the physical mechanisms
of drag reduction in a spatially developing turbulent
boundary layer laden with microbubbles. J. Fluid Mech., Vol.
503, 345355 (2l "14)
Hetsroni, G. Particleturbulence interaction. Int. J.
Multiphase Flow, Vol. 15, 735746 (1989)
Kaftori, D., Hetsroni, G. & Banerjee, S. Particle behavior in
the turbulent boundary layer. I. Motion, deposition and
entrainment. Phys. Fluids, Vol. 7, 10951106 (1995a)
Kaftori, D., Hetsroni, G. & Banerjee, S. Particle behavior in
the turbulent boundary layer. II. Velocity and distribution
profiles. Phys. Fluids, Vol. 7, 11071121 (1995b)
Kaftori, D., Hetsroni, G. & Banerjee, S. The effect of
particles on wall turbulence. Int. J. Multiphase Flow, Vol. 24,
359386 (1998)
Khalitov, D.A. & Longmire, E.K. Simultaneous twophase
PIV by twoparameter phase discrimination. Exp. Fluids,
Vol. 32, 252268 (2002)
Khalitov, D.A. & Longmire, E.K. Effect of particle size on
velocity correlations in turbulent channel flow. Proceedings
of the 4th ASME/FED and JSME Joint Fluids Conference,
July 610, Honolulu, Hawaii (2003)
Kiger, K. T & Pan, C. Suspension and turbulence
modification effects of solid particulates on a horizontal
turbulent channel flow. Journal of Turbulence, Vol. 3, 121
(2002)
Kussin, J. & Sommerfeld, M. Experimental studies on
particle behavior and turbulence modification in horizontal
channel flow with different wall roughness. Exp. Fluids, Vol.
33, 143159 (2002)
Marchioli, C. & Soldati, A. Mechanisms for particle transfer
and segregation in a turbulent boundary layer. J. Fluid
Mech., Vol. 468, 283315 (2002)
Maxey, M.R., Chang, E.J. & Wang, LP Interactions of
7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30June 4, 2010
particles and microbubbles with turbulence. Exp. Therm.
Fluid Sci., Vol. 12, 417425 (1996)
Mei, R., Adrian, R.J. & Hanratty, T.J. Particle dispersion in
isotropic turbulence under Stokes drag and Basset force
with gravitational settling. J. Fluid Mech., Vol. 225, 481495
(1991)
Nezu, I. & Azuma, R. Turbulence characteristics and
interaction between particles and fluid in particleladen
open channel flows. J. Hydr. Eng. ASCE, 9881001 (2'" 14)
Piomelli, U. & Balaras, E. Walllayer models for
LargeEddy Simulations. Annu. Rev. Fluid Mech., Vol. 34,
349374 (2002)
Portela, L.M. & Oliemans, R.VA. EulerianLagrangian
DNS/LES of particleturbulence interactions in
wallbounded flows. Int. J. Numer. Meth. Fluids, Vol. 43,
10451065 (2003).
Raffel, M., Willert, C. E., & Kompenhans, J. Particle Image
Velocimetry: a practical guide. SpringerVerlag Berlin
Heidelberg, 253 pp. (1998)
Robinson, S.K. Coherent motions in turbulent boundary
layers. Annu. Rev. Fluid Mech., Vol. 23, 601639 (1991)
Sato, Y., Fukuichi, U. & Hishida, K. Effect of interparticle
spacing on turbulence modulation by Lagrangian PIV. Int. J.
of Heat and Fluid flow, Vol. 21, 554561 (2000)
Soldati, A. Particles turbulence interactions in boundary
layers. Z. Angew. Math. Mech., Vol. 85, 683699 (2005)
Wang, L. P & Maxey, M. R. Settling velocity and
concentration distribution of heavy particles in
homogeneous isotropic turbulence. J. Fluid Mech., Vol. 256,
2768 (1993)
Wu, Y. & Christensen, K. T Population trends of spanwise
vortices in wall turbulence. J. Fluid Mech., Vol. 568, 5576
(2006)
Yamamoto, Y, Potthoff, M., Tanaka, T, Kajishima, T &
Tsuji, Y. Largeeddy simulation of turbulent gasparticle
flow in a vertical channel: effect of considering
interparticle collisions. J. Fluid Mech., Vol. 442, 303334
(2001)
Yang, T.S. & Shy, S.S. Twoway interaction between solid
particles and homogeneous air turbulence: particle settling
rate and turbulence modification measurements. J. Fluid
Mech., Vol. 526, 171216 (2005)
Zhou, J., Adrian, R. J., Balachandar, S. & Kendall, T M.
Mechanisms for generating coherent packets of hairpin
vortices in channel flow. J. Fluid Mech., Vol. 387, 353
(1999).
