You are on page 1of 22

Experimental investigation of interface deformation in free surface flow of

concentrated suspensions
A. Ashok Kumar, Bhaskar Jyoti Medhi, and Anugrah Singh

Citation: Phys. Fluids 28, 113302 (2016); doi: 10.1063/1.4967739


View online: http://dx.doi.org/10.1063/1.4967739
View Table of Contents: http://aip.scitation.org/toc/phf/28/11
Published by the American Institute of Physics
PHYSICS OF FLUIDS 28, 113302 (2016)

Experimental investigation of interface deformation in free


surface flow of concentrated suspensions
A. Ashok Kumar,a) Bhaskar Jyoti Medhi, and Anugrah Singhb)
Department of Chemical Engineering, Indian Institute of Technology,
Guwahati 781039, India
(Received 21 June 2016; accepted 31 October 2016; published online 21 November 2016)

It is well known that during the free surface flow of concentrated suspension of non-
colloidal particles, the suspension-air interface becomes highly corrugated. This sur-
face corrugation changes the interfacial area which could have important implications
in various applications involving heat and mass transfer across the interface. Surface
corrugation in free surface flow has been studied in the past, but its mechanism is
not fully understood. We report detailed experiments on quantitative measurement
of the surface deformation of concentrated suspension of non-colloidal particles in
open channel flow. The motion and location of the interface and the velocity field
of the bulk flow beneath the free surface were measured using the particle image
velocimetry technique. Experiments were performed to study the effect of particle
size, particle concentration, and viscosity of suspending fluid on the corrugation.
The interface fluctuation was found to increase linearly with the flow rate. The
deformation of the interface increased with increase in particle concentration until
an optimum concentration is reached and thereafter it decreases. Our observation
supports the previous studies on surface corrugation interpreted from the power
spectra of the reflected light from the interface. Suspension of larger particles and
less viscous fluid gives larger deformations of the suspension-air interface. These
results can be used to determine the optimum parameters to control the interfacial
area in free surface flow of concentrated suspensions. Published by AIP Publishing.
[http://dx.doi.org/10.1063/1.4967739]

I. INTRODUCTION
The deformation of interface during free surface flow of dense suspension appears to be
irrelevant when the interest is only confined to the bulk motion. However, there are numerous
applications where the changes in interfacial area could have a profound effect on transport pro-
cesses involving heat and mass transfer. It is the interface where the heat and mass transfer can be
augmented by increasing the free-surface area.1 The applications could range from food processing
to material processing as well as many biological processes. For example, there are many aerobic
bacteria living in thin film layers near the air-liquid interface and one way to increase the uptake of
oxygen into the suspension would be to increase the interface area of the bacterial suspension. In
some biotechnology applications, the goal is to produce structures with locally high concentration
of cells in bacterial suspensions through the interplay of chemicals and diffusion of nutrients.2 In
such applications, it is desired to know the parameters that could increase the transport through the
interface.
There are several studies that have investigated the free surface interaction with the bulk motion
of pure fluids. Hirsa et al.3 experimentally studied the vortex flow in a tank containing water whose
free surface was contaminated with surfactants. A laser-based technique of second-harmonic-
generation (SHG) was used to measure the surfactant concentration at the free surface. The intensity

a) Present address: Department of Chemical Engineering, Periyar Maniyammi University, Tamilnadu 613403, India.
b) Author to whom correspondence should be addressed. Electronic mail: anugrah@iitg.ernet.in

1070-6631/2016/28(11)/113302/21/$30.00 28, 113302-1 Published by AIP Publishing.


113302-2 Kumar, Medhi, and Singh Phys. Fluids 28, 113302 (2016)

of the second harmonic signal reflected from the interface correlates with the concentration of the
surfactants at the free surface. In their experiments, the digital particle image velocimetry (DPIV)
was used to measure the velocity field beneath the interface and this enabled them to find out the
effect of vortex pair interaction on surface concentration. In another study, they also determined the
location of the free surface from the digital images of the seeded flow using the Fourier transform
based method of surface elevation mapping.4 This method is based on the total internal reflection
at the interface. Unlike the conventional DPIV method, the cross-correlations for velocimetry were
performed on windows that were located on the deformable interface. However, in this technique
the distortion of the reflection can be caused by the surface curvature. Moreover, the light source
and camera can also cause distortion of the reflected image as well as a difference between the
brightness of the particle and its surface reflection. Kumar et al.5 have experimentally studied
the structural features of free surface turbulence in open channel flow. The population densities
and the persistence times of the various structures were measured for different flow conditions.
The interface features were visualized by sprinkling neutrally buoyant micro-particles on the free
surface and their motion was tracked in a frame moving with the mean velocity of the free-surface.
They concluded that the physical parameters characterizing the structures at the interface scale with
the wall shear stress and viscosity. There are several studies involving the particle image velocime-
try (PIV) technique to determine the velocity field but not many on the dynamics of the interfaces.
Tsuei and Savas6 developed a modification of digital particle image velocimetry (DPIV) method
for the free-surface boundary layer measurements. A variation of this technique was developed for
the simultaneous measurements of free surface deformation and near-surface velocity.4 The free
surface coordinates were determined directly from the seeded flow field images via a fast Fourier
transformation approach. Dabiri and Gharib7 combined the DPIV and the reflective mode of the
free-surface gradient detector (FSDG) technique to construct correlations between small-sloped
free-surface deformations and near-surface velocities.
The dynamics of air-liquid interface shows additional complexities for the case of suspension
flow. Free surface flow of dense suspension is often encountered in nature such as flood waves
carrying large amount of sediments and also in several material processing industries which handle
slurry flow in open channels. Other applications include processing of paints and debris flow. Bon-
noit et al. have demonstrated that free surface flow of dense suspension down an inclined plane
can be used as an effective rheometer.8 Examination of the shape of the interface in a tilted trough
flow has been effectively used to measure the second normal-stress difference in dense suspensions
by Couturier et al.9 Presence of particles not only changes the rheology of the mixture but in free
surface flow the particle dynamics can also influence the stability of the interface. Federico10 has
theoretically studied the unsteady flow of a mixture of fluid and solid particles with high sedi-
ment concentration flowing down on an inclined plane using a non-Newtonian rheological model
and determined the possible surface profiles. Li and Pozrikidis11 have studied the film flow of a
two-dimensional suspension of neutrally buoyant liquid drops down an inclined plane by numerical
simulations. A single drop was found to migrate toward an equilibrium position located between
the free surface and the wall; whereas the motion of a periodic file of drops was found to be
unstable to periodic perturbations, which leads to periodic accumulation of drops within bands
developing in the streamwise direction. Pozrikidis12 using the boundary integral technique studied
the motion of a spherical particle suspended in a gravity-driven film flow at low Reynolds number
assuming that the free surface deformation is infinitesimal. The particle translational velocity was
found to increase as it approached the free surface, whereas the particle rotational velocity reached
maximum value at certain intermediate position. Very recently Ancey et al.13 have studied the
flow of concentrated suspension of neutrally buoyant particles in Newtonian fluid down an inclined
flume. They have conducted experimental measurements of velocity, concentration, front position,
and flow depth profile during dam-break problem. Experimental results were compared with predic-
tions from lubrication theory and particle migration model of Phillips et al.14 Wang has reported
that non-Newtonian laminar flow exhibits free surface instability such as intermittent viscous debris
flows and fluctuation in mudflow.15 Brady and Carpen using stability analysis showed that a jump
in normal stress across the interface between the two fluids causes instability.16 They have reported
that the interface between the two fluids with a negative second normal stress difference can become
113302-3 Kumar, Medhi, and Singh Phys. Fluids 28, 113302 (2016)

unstable with respect to transverse and spanwise perturbations. Based on this theory, they predicted
instability in two-layer Couette flow and falling film flow of concentrated suspensions.
An interesting phenomenon related to the free surface flow of dense suspension at low Reynolds
number was observed by Loimer et al.17 When suspension was sheared between two parallel belts
moving with equal velocity in opposite directions, additional perturbation of the suspension-air inter-
face was induced which is not seen in the case of pure suspending fluid. The air-suspension interface
was reported to be highly corrugated, and the degree of corrugation depends on the size of parti-
cle, bulk concentration of suspension, viscosity, and surface tension of the suspending fluid. The
power spectra of the reflected light from the surface revealed that the corrugation structures at the
interface are present at various sizes and frequencies that resemble that of kinetic energy spectra of
turbulent flows. Later Timberlake and Morris18 studied the surface corrugation in gravity driven film
flow, and Singh et al.19 conducted experiments to study flow structures associated with the surface
corrugation in open channel flow of concentrated suspensions. These studies have now confirmed
that the surface corrugation during the free surface flow of concentrated suspensions is observed
for both homogeneous and inhomogeneous shear flows. Medhi et al. studied the velocity field at
the free surface in open channel flow with an objective to measure the wall slip.20 They also stud-
ied the surface roughness by analysis of the spectra of the reflected light from the free surface and
observed that the wall slip affects the velocity profile close to the wall as well as blunting of profile
but not the surface corrugation. These studies suggest that during the free surface flow of suspension
the random fluctuation of particles beneath the free surface creates the deformation of the interface.
The energy exchange between the fluid and solid particles depends on several parameters like par-
ticle size, viscosity, surface tension, and particle concentration. It is to be noted that Loimer et al.
reported maximum disorder of the surface at particle volume fraction (φ) between 0.40 and 0.45.
This observation was also supported by the study of Singh et al.19 and later by Medhi et al.20 who
studied the auto-correlation of power spectra from the refracted light intensity. The auto-correlation
function gives an idea as to how the flow structure at a location is correlated with that at nearby
location. It was observed that the flow structures uncorrelated faster for φ = 0.45 compared to that
of 0.40. Again for φ = 0.50, the auto-correlation was reported to decay slower compared to 0.45.
The rate of decay of auto-correlation is related to the disorder (corrugation) of the interface. Singh
et al. suggested that the particle fluctuations are weak for dilute suspension and the flow structure
at the free surface simply moves with the mean flow.19 At higher particle concentrations, the parti-
cles again follow ordered motion due to decreased fluctuation. This suggested that there could be an
intermediate particle concentration for which the fluctuations produce highest surface corrugation.
These observations motivated us to carry out the direct measurement of interface fluctuations during
the free surface flow of concentrated suspensions. Unlike previous studies on corrugation using the
indirect approach of analyzing the power spectra of the reflected light from the surface, we have
directly measured the position and motion of the interface in the velocity-velocity gradient plane.
Our method of determining the location of the interface is similar to that of Law et al.21 The velocity
field beneath the interface was measured by the particle image velocimetry (PIV). In Section II, we
describe the experimental method, and the results of experiments are discussed in Section III followed
by discussion in Section IV.

II. EXPERIMENTAL METHOD


A. Channel and flow visualization apparatus
The schematic diagram of our experimental setup is shown in Fig. 1. The flow channel was
made from 5 mm thick glass plates. The channel was 0.436 m long, 0.022 m wide, and 0.055 m
deep. The suspension in the channel was circulated using a screw pump. The motion of the screw
in the barrel of the pump gives rise to small oscillation in the flow. To overcome this, two ends
of the channel were connected to reservoirs which had several baffles to provide uniform mixing
and remove any unsteadiness in the flow before it enters the open channel. These reservoirs were
connected to the inlet and outlet of the screw pump via circular tubes. The pump was driven by
an AC frequency controlled motor whose speed could be adjusted to get the desired flow. The
113302-4 Kumar, Medhi, and Singh Phys. Fluids 28, 113302 (2016)

FIG. 1. Schematic diagram of the experimental setup and optical arrangements for PIV study.

whole channel was secured on an optical table in order to isolate the experimental measurement
from external vibrations. In Fig. 1, x is the span wise direction, y is the flow direction, and z is
the gravitational direction. We have also shown in Fig. 1 the optical arrangements to capture the
images of the interface in the y-z plane. The light source of the optical system was an argon ion
continuous laser from Spectra Physics (power = 1 W). A cylindrical lens was used to generate a
vertical light sheet as shown by the shaded region in the figure. The images were captured using a
1360 × 1024 pixel CCD camera (PixFly HiRes from PCO). The camera captured the images at the
frame rate of 19 frames/s, and the inter-framing time was 52 ms. Experiments were conducted at
different flow speeds. It was observed that the centerline velocity in the flow direction (measured at
x = 0.011 m, y = 0.273 m) at the free surface varied from 0.005 m/s to 0.0125 m/s.
Figure 2(a) shows a representative image (in the x- y plane) of the free surface corrugation
taken by placing the camera on the top of the channel. Crest and valleys resulting from surface
corrugation can be clearly seen in this image. To quantify the vertical interface fluctuations of

FIG. 2. (a) A sample image of the free surface in the x-y plane showing the surface corrugation in open channel flow of
concentrated suspension. (b) Schematic of optical arrangements for tracking the vertical (z) location of the interface and the
velocity field beneath the interface in the y-z plane.
113302-5 Kumar, Medhi, and Singh Phys. Fluids 28, 113302 (2016)

this interface, it is required to precisely locate the suspension-air interface. To find out the vertical
position (z) of the interface, the CCD camera was placed on the side of the channel and images were
captured in the y-z plane. However, it was noticed that due to the surface tension, the interface near
the wall of the channel was little raised and this obstructed the normal view of the interface away
from the wall. Therefore, to get a clear view of the interface, the camera was inclined to a small
angle (5◦) as shown in Fig. 2(b). The calibration of images at this inclination was performed for
proper analysis.

B. Preparation of suspensions
Different suspensions were prepared to study the effect of particle size, particle volume frac-
tion, and viscosity of suspending fluid on the interface dynamics. The suspension was required
to be neutrally buoyant with good match between the density of fluid (ρ f ) and density of parti-
cles (ρ p ). Two types of particles (polystyrene and PMMA) were used to prepare the suspensions.
Polystyrene particles were purchased from Microbeads AS (Norway) and PMMA particles from
Polysciences, Inc. (USA). These particles were chosen not for their commercial availability, but
they also allow preparing several combinations of suspending fluid for density matched suspension.
For the suspensions of polystyrene particles, the suspending fluid was mixture of glycerol and
water, and for the suspensions of PMMA particles, we have chosen two different fluid mixtures.
One was mixture of Triton X-100, zinc chloride, and water and the other was a mixture of glyc-
erol and water. These fluids have Newtonian behavior and are known to be chemically unreactive
with the particulate phase. To study the effect of particle size, three different suspensions were
prepared by dispersing polystyrene microspheres of mean diameter (d p ) equal to 80 µm, 250 µm,
and 500 µm in glycerol-water mixture. To achieve density matching with polystyrene particles
(density = 1.05 g/cc), the suspending fluid was prepared by mixing of 24% glycerol and 76% water
(volume %). The viscosity (η 0) and surface tension (γ) of this fluid mixture at 26 ◦C were found
to be 2.05 cP and 0.0484 N/m, respectively. To study the effect of suspending fluid viscosity on
interface fluctuations, two different suspensions of 200 µm PMMA particles (density = 1.18 g/cc)
were prepared. In the first case the density matched suspending fluid was mixture of 74% glycerol
and 26% water. The viscosity and surface tension of this fluid mixture at 26 ◦C were found to be
19.5 cP and 0.0604 N/m, respectively. In the second case, the suspending fluid that matched the
density of PMMA particles was a mixture of 76% Triton-X-100, 16.2% zinc chloride (ZnCl2), and
7.8% water, with percentage based on mass. The viscosity and surface tension of this fluid at 26 ◦C
was found to be 4000 cP and 0.0342 N/m, respectively. The particle and fluid properties of all the
five different suspensions are shown in Table I. With all the above mentioned suspensions, we have
conducted experiments with the particle volume fraction (φ) ranging from 0.10 to 0.50. The range
of centerline velocity (in the flow direction) near the interface in these experiments is also shown
in the table. The suspension was prepared by mixing the required amount of particles and liquid
in a beaker and thoroughly stirring to achieve homogeneous dispersion. To remove the air bubbles
formed during the mixing, the suspension was left out with the mouth of the beaker sealed for over
night, allowing the bubbles to rise up. Due to buoyancy the bubbles accumulated at the top layer,
which was skimmed out before transferring the suspension into the channel.

C. Measurement of interface location and velocity field beneath the interface


To visualize the free surface and the bulk flow beneath it we have seeded the flow with two
types of tracer particles. The air-suspension interface was visualized by seeding the flow with floater
particles, which were essentially hollow glass balloons of diameter 10–25 µm (supplied by ICI Ltd.,
India). These particles are very light and it was found that due to buoyancy they quickly come to
occupy positions near the interface as shown in the schematic diagram of Fig. 2(b). When the inter-
face was illuminated by a laser light sheet, these particles produced bright luminance. The location
of the interface was determined from the luminance contrast at the air-suspension interface. This is
discussed in more detail in Sec. III. To measure the velocity field beneath the interface, the suspen-
sion was seeded with polystyrene tracer particles of 10 µm diameter. The amount of added tracer
113302-6
Kumar, Medhi, and Singh
TABLE I. Particle and fluid properties.

Particle Particle Suspending Suspending Minimum Maximum


Type of diameter (d p ) density (ρ p ) fluid viscosity fluid density Surface tension centreline velocity centreline velocity
suspension Particle type (µm) (kg/m3) Suspending fluid (η 0) (Pa s) (ρ f ) (kg/m3) (γ) (N/m) (Vmin) (m/s) (Vmax) (m/s)

I Polystyrene 80 1050 24% glycerol + 76% water 0.002 05 1050.132 0.0484 0.0051 0.0121
II Polystyrene 250 1050 24% glycerol + 76% water 0.002 05 1050.132 0.0484 0.005 0.0110
III Polystyrene 500 1050 24% glycerol + 76% water 0.002 05 1050.132 0.0484 0.005 0.012
IV PMMA 200 1180 74% glycerol + 26% water 0.0195 1180.113 0.0642 0.005 0.0105
V PMMA 200 1180 76% Triton-X100 + 16.2% 4.0 1180.113 0.0342 0.005 0.012
ZnCl2 + 7.8% water

Phys. Fluids 28, 113302 (2016)


113302-7 Kumar, Medhi, and Singh Phys. Fluids 28, 113302 (2016)

FIG. 3. (a) A sample PIV image of tracer particles. (b) Vector map of velocity field beneath the interface.

particles was negligibly small in comparison to the volume of suspension so that it is not expected to
influence the flow behavior of bulk suspension. The images of these tracer particles illuminated by
laser light sheet were captured using the CCD camera and saved in 680 × 512 pixels, 12 bits format
gray scale images on a computer. The captured view of the flow field was 15.5 mm × 11.5 mm,
and the size of the tracer particle images varied between 2 and 5 pixels. These images were used
in the PIV analysis to determine the velocity field. In PIV analysis, each captured image is divided
into small subareas called interrogation window. Cross-correlation between interrogation windows
of two consecutive images gives the velocity field. To obtain the velocity field of suspension in
the vertical plane ( y-z), the PIV analysis software PIV SLEUTH22 was used by considering the
two-frame cross correlation technique.23 The inter-framing time between pair of images was 52 ms.
The spot size for flow field interrogation was 128 × 128 pixels with 75% overlap between interroga-
tion spots. Figure 3(a) shows a sample PIV image and the resulting velocity vector map is shown in
Fig. 3(b). The velocity component in the flow direction (Vy) at the top is referred as the centerline
velocity. It should be noted that the velocity vectors from PIV could not be obtained right up to the
interface since the camera view for PIV measurements was kept slightly below the free surface due
to obstruction caused by the curved interface near the channel walls.

III. RESULTS
A. Velocity field beneath the interface
In this section, we present the velocity profiles for the suspensions of 200 µm PMMA parti-
cles in Triton-X100, zinc chloride, and water mixture in the vertical plane ( y-z) beneath the
air-suspension interface. The position of the vertical laser sheet that illuminated the tracer parti-
cles was in the middle of the channel, i.e., 0.01 m from the channel wall. The CCD camera was
positioned on the side of the channel at an axial distance of 0.273 m from the entrance. At this
location, it was observed that the flow was fully developed.20 An average over 100 velocity vectors
was taken to get the velocity profile at a given location. Figure 4 shows the velocity profile of
the pure suspending fluid and suspensions with particle volume fractions (φ) of 0.2, 0.3, and 0.4.
The measurements were carried out at three different flow rates and the corresponding centerline
velocity is also shown. It can be observed that the velocity is zero near the wall (due to no-slip
condition there) but increases towards the interface where it has maximum velocity.
We observe that the normalized velocity profiles for all the cases closely match with the analyt-
ical profile (parabolic law) of Newtonian fluid in an open channel flow. This indicates that there was
no significant migration of the particles from the bottom of the channel to the free surface. This is
expected since the length of the channel in our experiments was much smaller than the characteristic
length scale required for shear induced migration.24 There is also no significant difference in the
velocity profiles for all the three flow rates considered. This shows that the velocity profiles at the
measurement locations were fully developed. PIV analysis with particle volume fraction beyond 0.4
could not produce correct velocity field due to mismatch in the refractive index of the suspending
fluid and particles.
113302-8 Kumar, Medhi, and Singh Phys. Fluids 28, 113302 (2016)

FIG. 4. Velocity profile in the y-z plane for (a) suspending fluid (viscosity, η = 4000 cP) and suspensions of 200 µm PMMA
particles in density matched suspending fluid with various particle fractions: (b) φ = 0.2, (c) φ = 0.3, and (d) φ = 0.4. The
analytical profile is also shown for comparison.

In open channel flow, different forces such as capillary, viscous, inertia, and gravity can be pres-
ent. To quantify the relative importance of these forces we have computed various dimensionless
numbers, which are shown in Table II. The expressions used to calculate these dimensionless num-
bers are also shown in the table. The particle Reynolds number (Re p ) which compares the inertial
force to viscous force on the particles was based on the diameter of particle (d p ), centerline velocity
(V), and fluid viscosity (η0). It is observed that except for the experiments with 500 µm particles,
in most of the experiments Re p < 1 and would be even smaller if effective viscosity of suspension
was considered. For bulk flow of suspension, the relative importance of inertial and viscous forces
can be characterized by the flow Reynolds number (Re f ). The length scale in the calculations of Re f
was taken as the depth of the film (H = 1.15 cm). The Froude number (Fr ) is the ratio of inertial
and gravitational forces, which is also a measure of flow and depth interactions. Fr ≪ 1 corresponds

TABLE II. Dimensionless numbers.

Particle Reynolds Flow Reynolds Froude Capillary Bond number Weber


number (Re p ) number (Re f ) number (Fr) number (Ca) (Bo) number (We)
ρf V dp ρfV H η V 2 ρ f HV 2
Re p = η0 Re f = η0
Fr = √ V
gH Ca = 0γ Bo = ∆ρgγ H We = γ

Type of suspension Min Max Min Max Min Max Min Max Min Max Min Max

I 0.208 0.495 30.04 71.27 0.015 0.036 0.0002 0.0005 0.0033 0.0035 0.0062 0.0365
II 0.640 1.40 29.45 64.7 0.015 0.033 0.0002 0.0005 0.0033 0.0035 0.0062 0.0302
III 1.280 3.073 29.45 70.68 0.015 0.036 0.0002 0.0005 0.0033 0.0035 0.0062 0.0359
IV 0.060 0.127 3.47 7.30 0.015 0.031 0.0016 0.0034 0.0024 0.0024 0.0056 0.0248
V 0.0003 0.0007 0.017 0.041 0.015 0.031 0.585 1.403 0.0041 0.0042 0.0099 0.0571
113302-9 Kumar, Medhi, and Singh Phys. Fluids 28, 113302 (2016)

to lower energy state indicating that interface is free from any standing waves. Capillary number
(Ca ) compares the viscous forces and the surface tension forces acting across the suspension-air
interface. We observed that in all except one of our experiments Ca ≪ 1. Low capillary number
indicates that the viscous forces are much smaller compared to the interfacial tension. Bond number
(Bo ) is the ratio of buoyancy force and surface tension force. Since the density difference (∆ρ)
between the particle and fluid was very small, Bo ≪ 1. Another dimensionless number, which is
used in analyzing the free surface flow, is the Weber number (We ) defined as the ratio of inertial
force to the surface tension force. High inertia forces may cause increase in the deformation of the
interface. On the other hand, surface tension forces oppose the increase in surface area caused by
the deformation. Very low Weber number in all of our experiments indicates that the particles were
not able to leave the interface at any time and that the deformation of the interface is purely due to
hydrodynamic fluctuations.

B. Interface fluctuations
The contrast between the suspension and air, which indicates the location of the interface, was
determined using an edge detection method. As mentioned earlier, the interface was visualized by
illuminating the hollow glass particles. When a vertical sheet of laser light illuminates the surface,
the region of the suspension produces a luminance, which contrasts very sharply with the air side.
The edge of this contrast is precise location of the interface. The advantage of this method is
that the interface will always show up as a very distinctive and continuous edge. Since the screw
pump was circulating the suspension continuously the tracer particles were well mixed with the
suspension. The suspension side is bright due to illuminating tracer particles and air side is dark.
Therefore, interface can be located by determining the edge of the luminance contrast between the
air and suspension. Figure 5(a) shows a sample image of the interface visualized by the hollow glass
particles. The images were stored in gray scale whose intensity varied between 0 and 255. At a
given axial location, the intensity of the pixels along the vertical direction is shown in Fig. 5(b). To
determine the location of the interface at this axial location, the gradient of intensity was computed.
The position of maximum gradient locates the interface as shown in Fig. 5(b). The first point of
inflection was taken as the position of the interface. The uncertainty in the measurement would be
the difference between the first and second inflection points. The maximum uncertainty in these
measurements was not more than 3 pixels. Sweeping across all the pixels in axial direction, we

FIG. 5. (a) A sample image of the air-suspension interface. (b) Intensity variation in the vertical direction (z) at a given axial
location. (c) Location of the interface along the flow direction (y) at a given instant of time.
113302-10 Kumar, Medhi, and Singh Phys. Fluids 28, 113302 (2016)

FIG. 6. Root mean square (RMS) interface fluctuations plotted with time for various suspending fluids. The average
centerline velocity was 0.019 58 m/s.

were able to determine the profile of the interface as shown in Fig. 5(c). From this surface elevation
profile, the root mean square (RMS) fluctuation of the interface about its mean position at any given
time instant is given as

N 2
1 (Z i − Z o )
Z m (t i ) = , (1)
N
where Zi is the vertical location of the interface, Zo is the mean vertical position of the interface,
and N is number of pixels of the image in the flow direction. In our experiments, 444 continuous
images were recorded with the time interval of 52 ms. From these images, we were also able to
determine the time averaged root mean square position ( Z̄ m ) of the interface.
Figure 6 shows the time trace of the root mean square fluctuation of the interface for various
suspending fluid mixtures considered in this work. The average centreline velocity (along the flow
direction) was 0.019 58 m/s in these experiments. It was observed that for pure suspending fluid the
RMS fluctuations of the interface were detectable only for the fluid with lowest viscosity (2.05 cP).
For more viscous fluids, the fluctuations were not detectable. These fluctuations could be due to
pump noise.
Figure 7 shows the time trace of the root mean square fluctuation of the interface for three
different flow speeds for suspension of 80 µm polystyrene particles having volume fraction (φ)
equal to 0.10 in density matched glycerol-water mixture of viscosity 2.05 cP. It can be observed that
the perturbation of the interface increases with the flow rate. We also notice that the perturbation is
low at small flow speed but increases at higher flow rates.
The effect of particle concentration on the interface fluctuations is shown in Fig. 8. Here we
have plotted the RMS value of the fluctuation with time for suspensions of 250 µm polystyrene
particles for φ ranging from 0.10 to 0.50. The suspending fluid was density matched glycerol-water
mixture of viscosity 2.05 cP. In all these experiments, we tried to maintain the same flow rate.
However, it was difficult to maintain the same velocity in all the cases. The centreline velocities
in these experiments were in the range of 0.0069–0.0076 m/s. We observe that up to φ = 0.40
the mean fluctuation of the interface increases, but thereafter it shows decrease in magnitude for
φ = 0.45 and 0.50.
The effect of particle size on the interface fluctuation is shown in Fig. 9, where we have
plotted the time trace of the RMS value of the fluctuation for suspensions of 80, 250, and 500 µm
polystyrene particles in density matched glycerol water mixture of viscosity 2.05 cP. The particle
volume fraction in all the three cases was fixed at 0.30 and the centreline velocities of the interface
in the flow direction were in the range of 0.0074–0.0079 m/s. It can be clearly observed that the
fluctuations increase with the size of particles and their magnitude is of the order of particle size.
113302-11 Kumar, Medhi, and Singh Phys. Fluids 28, 113302 (2016)

FIG. 7. RMS interface fluctuation plotted with time for three different flow rates for suspension of 80 µm polystyrene
particles dispersed in density matched glycerol water mixture (η = 2.05 cP). The volume fraction of suspension (φ) was 0.10.

For φ = 0.40, this can give rise to over 50% increase in the interfacial area. Next, we have studied
the effect of the viscosity of suspending fluid on the interface fluctuations.
Figure 10 shows the RMS value of the fluctuations plotted with time for two different suspen-
sions. In the first case (Fig. 10(a)), the suspending fluid was glycerol-water mixture of viscosity
19 cP whose density matched with the 200 µm PMMA particles. In the second case (Fig. 10(b)),
the suspending fluid was density matched mixture of Triton X-100, zinc chloride, and water and its
viscosity was measured to be 4000 cP. The volume fraction of PMMA particles in both the cases
was 0.30. The centreline velocity in case (a) was 0.0071 m/s and that of case (b) was 0.0073 m/s.
The fluctuation of the interface for the second case (higher suspending fluid viscosity) is much
lower compared to that of first case.
We have averaged the RMS fluctuations of the interface over 444 images to obtain the time
averaged fluctuations; in Fig. 11, we show its variation with centreline velocity for density matched
suspensions of 80, 250, and 500 µm polystyrene particles at various values of φ. We observe a linear
variation of the fluctuation with the flow rate for all the suspensions. As mentioned previously, it is
noticed that the fluctuation first increases with the particle concentration and thereafter decreases.
The rate of increase in the RMS value of the fluctuation is also found to slow down after the
optimum concentration. The value of optimum concentration appears to depend on the particle size.
The same trend is also observed in Fig. 12, which shows the time averaged fluctuations for
two different suspending fluids as was the case in Fig. 10. Since for a given concentration the time
averaged fluctuation appears to increase linearly with flow rate, we have plotted the slope of the
curves of Figs. 11 and 12 against the particle fraction. This is shown in Fig. 13. It can be observed
that the mean squared fluctuation first increases with particle concentration and then at certain
optimum particle fraction it reaches maxima and thereafter again decreases. This optimum value
of particle volume fraction which depends on the particle size and suspension viscosity lies in the
range of 0.30 and 0.40 for all kind of suspensions studied in this work. As mentioned previously, it
can be noticed that the fluctuation increases with increase in the particle size, whereas it decreases
with the increase in the viscosity of suspending fluid. The surface tension of the suspending fluid
also affects the corrugation structure. Higher surface tension value should dampen the fluctuations.
Though it was desired to have the same value of surface tension for fluids with 19.5 cP and 4000 cP
113302-12 Kumar, Medhi, and Singh Phys. Fluids 28, 113302 (2016)

FIG. 8. RMS value of interface fluctuation plotted with time for suspensions of 250 µm polystyrene particles in density
matched glycerol water mixture (η = 2.05 cP) at various volume fractions of particles: (a) φ = 0.10, (b) φ = 0.20, (c) φ = 0.30,
(d) φ = 0.40, (e) φ = 0.45, and (f) φ = 0.50. The centerline velocities in all these cases were in the range of 0.0069–0.0076 m/s.

viscosity, this was not achieved. The values of surface tension for these fluid mixtures were 0.0642
and 0.0342 N/m, respectively. It is to be noted that for more viscous suspending fluid (4000 cP), we
were unable to perform experiments beyond φ = 0.40 due to difficulty in handling the suspension
flow in the channel.

C. Vertical component of the interface velocity


In this section, we present the vertical component of interface velocity and its dependence on
the volume fraction of the suspension, which has important impact on the interface topology and its
113302-13 Kumar, Medhi, and Singh Phys. Fluids 28, 113302 (2016)

FIG. 9. RMS value of interface fluctuation plotted with time for suspensions of (a) 80 µm, (b) 250 µm, and (c) 500 µm
polystyrene particles in density matched glycerol water mixture (η = 2.05 cP). The particle volume fraction in all the three
cases was fixed at 0.30 and the centerline velocities were in the range of 0.0074–0.0079 m/s.

stability. The dynamics of particles at the interface increases the size of the exposed surface signifi-
cantly and influences the amount of energy there. From the knowledge of the vertical component of
the velocity (Vz), the fluctuating kinetic energy can be computed. From the evolution of free surface
location with time the vertical component of the interface velocity was computed. This requires the
knowledge of vertical displacement of the interface between two successive images. The difference
in the vertical position of the interface between two consecutive frames divided by the inter-framing
time was taken as the measure of the vertical velocity of the interface. Figure 14 shows the time
trace of the velocity of the interface in the vertical direction (Vzm) at a given axial location for

FIG. 10. RMS value of interface fluctuation plotted with time for suspensions of 200 µm PMMA particles in (a) density
matched glycerol water mixture (η = 19 cP), and (b) density matched mixture of Triton X-100, zinc chloride and water
(η = 4000 cP). The particle volume fraction was 0.30 and the centerline velocity in case (a) was 0.0071 m/s, and that of case
(b) was 0.0073 m/s.
113302-14 Kumar, Medhi, and Singh Phys. Fluids 28, 113302 (2016)

FIG. 11. Time averaged RMS fluctuation of the interface location plotted against centerline velocity at various particle
fractions for density matched suspensions of (a) 80 µm, (b) 250 µm, and (c) 500 µm polystyrene particles in glycerol-water
mixture (η = 2.05 cP).

three different flow rates for suspension of 500 µm polystyrene particles in glycerol-water mixture
of viscosity 2.05 cP. The volume fraction of suspension was 0.2. The vertical component of the
velocity of interface shows wide fluctuations whose magnitude increase with flow speed.
Figure 15 shows the vertical component of interface velocity plotted against time for six
different volume fractions of particles in the range of 0.10–0.50. The suspensions taken were
500 µm polystyrene particles dispersed in density matched glycerol water mixture (η = 2.05 cP).
Similar to the observation of fluctuation of the interface (Fig. 8), we observe that the Vzm first
increases with particle concentration and shows maximum for φ = 0.4, thereafter it decreases. The
range of values for φ = 0.5 is nearly same as that for φ = 0.1.

FIG. 12. Time averaged RMS fluctuation of the interface location plotted against centerline velocity at various particle
fractions for density matched suspensions of 200 µm PMMA particles in (a) glycerol-water mixture (η = 19 cP) and (b)
mixture of Triton X-100, zinc chloride and water (η = 4000 cP).
113302-15 Kumar, Medhi, and Singh Phys. Fluids 28, 113302 (2016)

FIG. 13. RMS value of interface fluctuation (scaled with the centerline velocity) plotted against the particle fraction for
(a) three different suspensions of polystyrene particles (80 µm, 250 µm, and 500 µm) suspended in density matched
glycerol-water mixture (η = 2.05 cP), and (b) two different suspensions of 200 µm PMMA particles in suspending fluids
of different viscosities (glycerol-water mixture, η = 19 cP and Triton X-100, zinc chloride, water mixture, η = 4000 cP).

The characteristic values of the fluctuating velocities in the vertical direction are represented
by the time averaged root mean square (RMS) values which are represented as V z m . To understand
the effect of particle size and viscosity of suspending fluid on the vertical component of the velocity
of the interface, we performed similar analysis as was the case for Figs. 9-13. Interestingly it was
noted that profiles of vertical velocity also exhibits the same behavior as the interface location,
i.e., the surface normal velocity also increases with particle fraction up to certain optimum value
and thereafter it decreases. It was observed that for pure fluid the vertical velocity of interface was
almost zero (not shown here). Figure 16 shows the time averaged RMS value of vertical velocity
for suspensions of three different sizes (80, 250, and 500 µm) of polystyrene particles for various
particle fractions. Here also we observe that the RMS value increases almost linearly with the flow
speed.

FIG. 14. Time trace of the vertical component (Vzm) of interface velocity for three different flow rates for suspension of
500 µm polystyrene particles dispersed in density matched glycerol water mixture (η = 2.05 cP). The volume fraction of
suspension was 0.2. The maximum centerline velocity is shown in the plots.
113302-16 Kumar, Medhi, and Singh Phys. Fluids 28, 113302 (2016)

FIG. 15. Time trace of the vertical component (Vzm) of interface velocity for suspensions of 500 µm polystyrene particles in
density matched glycerol water mixture (η = 2.05 cP) at various particle fractions: (a) φ = 0.10, (b) φ = 0.20, (c) φ = 0.30, (d)
φ = 0.40, (e) φ = 0.45, and (f) φ = 0.50. The centerline velocities in all these cases were in the range of 0.0068–0.0081 m/s.

The effect of viscosity of suspending fluid on the time averaged RMS value of the velocity
fluctuations is shown in Fig. 17. Velocity fluctuations were analyzed for suspensions of 500 µm
PMMA particles in density matched fluid. In one case, the suspending fluid was glycerol-water
mixture of viscosity 19 cP, and for other it was a mixture of Triton X-100, zinc chloride, and water
whose viscosity was measured to be 4000 cP. In Fig. 18, we have normalized the time averaged
RMS velocity of the interface with the centerline velocity and plotted against the particle fraction.
Figure 18(a) shows the effect of particle size and Fig. 18(b) shows the effect of suspending
fluid viscosity on interface velocity fluctuations. The results are as per the expectations and follow
the same trend as observed in Fig. 13. From these experimental observations, it can be argued
that at low concentration there are few inter particle interactions and dispersed particles typically
follow the mean flow. Therefore, the perturbation of the interface is also small. As the concentration
increases, the particle-particle interaction causes fluctuations of the interface due to random fluctu-
ations of particles. However, at higher concentrations the particles are close to each other and their
movement is hindered by neighboring particles. The suspended particles again follow the mean
motion of the fluid due to the crowding effect. As a result, we observe decrease in surface normal
velocity. These profiles are qualitatively same for any particle size of dispersed particles and any
viscosity of suspending fluid. However, the surface corrugation and velocity fluctuations are high
for suspension of larger particles and less viscous fluids.

IV. DISCUSSION
The fluctuating motion of the particles near the free surface causes projections, cusps, and
dimples, which is called as surface corrugation. While the surface tension acting on the particle can
113302-17 Kumar, Medhi, and Singh Phys. Fluids 28, 113302 (2016)

FIG. 16. Time averaged RMS value of the vertical component (V z m ) of interface velocity plotted with centerline velocity
at various particle fractions for density matched suspensions of (a) 80 µm (b) 250 µm, and (c) 500 µm polystyrene particles
in glycerol-water mixture (η = 2.05 cP).

significantly dampen the fluctuations near the free surface, in a concentrated suspension the particle
fluctuations will try to enhance the normal perturbations. Brady and Carpen16 have shown that the
jump in normal stresses acts to destabilize the interface. The first normal stress difference governs
the behavior due to perturbation in the flow direction, and the second normal stress difference
impacts the transverse perturbation. The presence of second normal stress difference also leads
to many interesting phenomena in non-Newtonian fluids as well as concentrated suspensions.25
Ramachandran and Leighton26 have shown that second normal stress difference is responsible for
secondary flow of suspension in non-axisymmetric geometries. The secondary flow in open channel
flow may also have an effect on the dynamics of the free surface.27 Whether similar instability

FIG. 17. Time averaged RMS value of the vertical component (V z m ) of interface velocity plotted against centerline velocity
at various particle fractions for density matched suspensions of 200 µm PMMA particles in (a) glycerol-water mixture
(η = 19 cP), and (b) mixture of Triton X-100, zinc chloride, and water (η = 4000 cP).
113302-18 Kumar, Medhi, and Singh Phys. Fluids 28, 113302 (2016)

FIG. 18. Time averaged RMS value of the vertical velocity of the interface (scaled by the centerline velocity) plotted against
the particle fraction for (a) three different suspensions of polystyrene particles (80 µm, 250 µm, and 500 µm) suspended
in density matched glycerol-water mixture (η0 = 2.05 cP), and (b) two different suspensions of 200 µm PMMA particles in
suspending fluids of different viscosities.

in free surface flow results in the form of corrugation structures due to random perturbation in
rheological properties is a relevant question. Initial studies have linked this phenomenon to the
shear induced migration.17,18 At present, the complex interaction between the surface corrugation
and suspended particles is far from being clearly understood. Our study was undertaken to give
some insight into the dynamics of surface profile. Let us first analyze the corrugation from the
surface instability point of view. If the first normal stress difference is associated with perturbation
in the flow direction and second normal stress difference to that in the transverse direction then
we would have expected much weaker corrugation in the x-y plane compared to the y-z plane. All
the previous studies in the flow and spanwise directions report corrugation structure whose size
is much larger compared to the particle size. In this study, we measured the perturbation in the
transverse plane (y-z) and found it to be of the order of particle size. The question then comes to
mind is whether the surface corrugation is property of bulk motion or a particle level phenomenon.
Medhi et al.20 conducted experiments in open channel flow of concentrated suspensions under slip
and no-slip conditions of the channel wall and observed no noticeable difference in corrugation
structure. This led them to speculate that free surface corrugation is a response of suspended parti-
cles to the bulk flow and it does not arise from boundary problem such as development of slip
layer.
We now analyze the observation of maxima in the plot of vertical fluctuation of the interface
against the particle volume fraction. At low concentration, the particles will follow a streamlined
motion and there should not be much transverse displacement associated with particle-particle inter-
actions. At higher concentrations, the many-body effects play a role and the suspended particles
undergo fluctuating motion due to frequent interaction with other particles. When larger particles
are present, the impact and interactions between them become even more significant. This gives rise
to perturbation of the interface. However, the question still remains as why do we see decrease in
corrugation and fluctuations at higher concentrations? The velocity profiles in Fig. 4 shows that the
suspension is sheared across the whole depth, which rules out possibility of jammed state in upper
flow layers. In gravity-driven flow down an inclined plane, Ancey et al.13 have shown that the bulk
behaved like a Newtonian fluid for solid fractions as large as 0.52. Therefore, the observation of
maxima appears to result from competing effect of particle fluctuation motion and viscosity. The
fluctuations grow with increase in particle concentration so is the suspension viscosity. The increase
in suspension viscosity will itself dampen the fluctuations, and above a certain concentration there
will be a decrease in the surface corrugation. Decrease in surface fluctuations may also be due
to the layering of particles beneath the interface. The non-Brownian suspensions in Newtonian
fluids with volume fractions greater than 0.3 are known to show shear thinning behavior, which
is related to the layering of particles in the flow directions.28 Brown and Jaeger29 have studied
113302-19 Kumar, Medhi, and Singh Phys. Fluids 28, 113302 (2016)

the role of dilation of air-suspension interface on the rheology of shear thickening suspensions
where stresses are transmitted through frictional interactions. Shear thickening behavior is usually
observed at much higher concentration of particles, whereas all of our experiments were for much
lower concentrations. We have not performed any measurement on the rheology of suspensions in
this work, but previous measurements with same lot of particles and suspending fluids did show
shear thinning behavior.20,30 The degree of shear thinning increases as the particle volume fraction is
increased. If we take a closer look of Figure 16, it can be observed that for particle fractions greater
than 0.4, the rate of increase in fluctuation with flow rate is smaller at higher velocities. However,
our velocity range was too narrow to elucidate this behavior. Therefore, we speculate that the shear
induced layering of particles may also be responsible for decrease in surface corrugation at higher
concentration.
Loimer et al.17 have presented a model for free surface corrugation, which accounts for the
shear induced flux of particles towards the interface and restoring force due to surface tension. It is
argued that even though the bulk of suspension remains homogeneous, at the free surface there is
sharp normal gradient in the particle concentration due to a jump from the bulk concentration, φ, to
zero across a thin layer. The resulting flux of particles to the surface is
∂φ
qout = −D (φ) Ga2 , (2)
∂z
where G is the shear rate near the free surface, a is the particle radius, and D(ϕ) is dimensionless
self diffusivity of suspension. The restoring inward flux of particles due to capillary force can be
approximated as

1 γl 2 − a φ
l

qin = . (3)
3 aη r (φ)
In the above equation, l is the height (beyond the undisturbed surface level) of a particle sticking
out to the interface, γ is the surface tension of the liquid, and η r (ϕ) is the relative viscosity of
suspension (ratio of effective suspension viscosity and suspending fluid viscosity). From the balance
of these two fluxes and approximating the concentration gradient at the interface to be ϕ/2a, Loimer
et al. obtained the following approximation for the relative roughness (l/a) of the surface resulting
from the corrugation:
( )
l 3
= η r (φ) D (φ) Ca , (4)
a 4

where Ca is the capillary number. The above expression predicted that the corrugation will mono-
tonically increase with particle concentration since they assumed both relative suspension vis-
cosity η r (φ) and dimensionless diffusion coefficient D(φ) to be monotonically increasing func-
tion of ϕ. This was not observed in our experiments, which showed non-monotonic behavior of
corrugation with particle concentration. The qualitative difference between our experiments and
the model predictions could be due to the approximation of self diffusion coefficient D(φ) as
a linear function of particle concentration. This was based on the experiments of Leighton and
Acrivos31 who calculated the diffusion coefficients from long-term viscosity decrease resulting
from the particle migration in cylindrical Couette flow. Sierou and Brady32 have computed the
self-diffusion coefficients for suspension of non-Brownian particles in simple shear flow using
accelerated Stokesian dynamics simulations. The self-diffusivities were calculated for volume frac-
tion (φ) in the range of 0.10–0.50 and compared them to be in good qualitative agreement with
the experimental results of Breedveld et al.33 It is interesting to note that in the simulations as
well as experiments, the self diffusivity in the velocity gradient direction was observed to increase
with volume fraction up to φ = 0.40 and decrease thereafter. This could possibly explain the
non-monotonic behavior of corrugation in our experiments. Simultaneous measurement of self
diffusivity near the free surface and interface corrugation is required to further understand this
behavior.
113302-20 Kumar, Medhi, and Singh Phys. Fluids 28, 113302 (2016)

V. CONCLUSIONS
In this work, we have described experimental method to locate and measure position of
air-suspension interface, velocity of bulk flow beneath the interface, and vertical velocity of the
interface for concentrated suspension of non-colloidal particles in open channel flow. The lumi-
nance contrast between suspension and air indicates the location of the interface, which was deter-
mined using edge detection method. Fairly standard PIV routine was used to obtain the velocity
field beneath the free surface. It was observed that the interface fluctuations are very small at low
volume fraction of particles, but increase rapidly with the increase in particle volume fraction.
There is an optimum value of volume fraction at which the interface fluctuation is the largest
and thereafter it decreases. It was found that for a given concentration of particles, the interface
fluctuation increases linearly with the flow speed. The results of mean fluctuation and mean surface
normal velocity of the interface followed almost similar behavior. This indicates that during the
free surface flow of concentrated suspensions, interaction between the dispersed particles causes
the distortion and fluctuation in interface resulting into corrugated surface. Due to this corrugation,
the interface area also changes. These surface corrugations are more for bigger particles and less
viscous suspending fluids. Interestingly, the interface fluctuation or surface corrugation does not
affect the velocity profile beneath the free surface. The results from this study can be very useful in
predicting the interface area in such systems as in many material processing applications, it is the
interface where heat and mass transfer plays an important role.

ACKNOWLEDGMENTS
We acknowledge the funding from the Department of Science and Technology, India through
the Project No. SR/S3/CE/38/06.
1 C. Ancey, in Gravity Flow on Steep Slope in Buoyancy Driven Flows, edited by E. Chassignet, C. Cenedese, and J.
Verron (Cambridge University Press, 2012).
2 F. G. Ochoa and E. Gomez, “Bioreactor scale-up and oxygen transfer rate in microbial processes: An overview,” Biotechnol.

Adv. 27, 153 (2009).


3 A. Hirsa, G. M. Korenowski, L. M. Logory, and C. D. Judd, “Velocity field and surfactant concentration measurement

technique for free-surface flows,” Exp. Fluids 22, 239 (1997).


4 A. H. Hirsa, M. J. Vogel, and J. D. Gayton, “Digital particle velocimetry technique for free-surface boundary layer measure-

ments: Application to vortex pair interactions,” Exp. Fluids 31, 127 (2001).
5 S. Kumar, R. Gupta, and S. Banerjee, “An experimental investigation of the characteristics of free-surface turbulence in

channel flow,” Phys. Fluids 10, 437 (1998).


6 L. Tsuei and O. Savas, “Treatment of interfaces in particle image velocimetry,” Exp. Fluids 29, 203 (2000).
7 D. Dabiri and M. Gharib, “Simultaneous free-surface deformation and near-surface velocity measurements,” Exp. Fluids

30, 381 (2001).


8 C. Bonnoit, T. Darnige, E. Clement, and A. Lindner, “Inclined plane rheometry of a dense granular suspension,” J. Rheol.

54, 65 (2010).
9 E. Couturier, F. Boyer, O. Pouliquen, and E. Guazzelli, “Suspensions in a tilted trough: Second normal stress difference,”

J. Fluid Mech. 686, 26 (2011).


10 V. D. Federico, “Free-surface flow of hyper concentrations,” Fluid Dyn. Res. 24, 23 (1999).
11 X. Li and C. Pozrikidis, “Film flow of a suspension of liquid drops,” Phys. Fluids 14, 61 (2002).
12 C. Pozrikidis, “Motion of a spherical particle in film flow,” J. Fluid Mech. 566, 465 (2006).
13 C. Ancey, N. Andreini, and G. Epely-Chauvin, “The dam-break problem for concentrated suspensions of neutrally buoyant

particles,” J. Fluid Mech. 724, 95 (2013).


14 R. J. Phillips, R. C. Armstrong, R. A. Brown, A. L. Graham, and J. R. Abott, “A constitutive equation for concentrated

suspensions that accounts for shear-induced particle migration,” Phys. Fluids A 4, 30 (1992).
15 Z. Y. Wang, “Free surface instability of non-Newtonian laminar flows,” J. Hydraul. Res. 40, 449 (2002).
16 J. F. Brady and I. C. Carpen, “Second normal stress jump instability in non-Newtonian fluids,” J. Non-Newtonian Fluid

Mech. 102, 219 (2002).


17 T. Loimer, A. Nir, and R. Semiat, “Shear-induced corrugation of free interfaces in concentrated suspensions,” J. Non-

Newtonian Fluid Mech. 102, 115 (2002).


18 B. D. Timberlake and J. F. Morris, “Particle migration and free-surface topography in inclined plane flow of a suspension,”

J. Fluid Mech. 538, 309 (2005).


19 A. Singh, A. Nir, and R. Semiat, “Free surface flow of concentrated suspensions,” Int. J. Multiphase Flow 32, 775 (2006).
20 B. J. Medhi, A. A. Kumar, and A. Singh, “Apparent wall slip velocity measurements in free surface flow of concentrated

suspensions,” Int. J. Multiphase Flow 37, 609 (2011).


21 C. N. S. Law, B. C. Khoo, and T. C. Chew, “Turbulence structure in the immediate vicinity of the shear-free air-water

interface induced by a deeply submerged jet,” Exp. Fluids 27, 321 (1999).
113302-21 Kumar, Medhi, and Singh Phys. Fluids 28, 113302 (2016)

22 K. T. Christensen, S. M. Soloff, and R. J. Adrian, Technical Report 943, Department of Theoretical and Applied Mechanics,
University of Illinois at Urbana-Chamaign, 2000.
23 R. J. Adrian, R. D. Keane, and Y. Zhang, “Super resolution particle-imaging velocimetry,” Meas. Sci. Technol. 6, 754 (1995).
24 P. R. Nott and J. F. Brady, “Pressure driven flow of suspensions: Simulation and theory,” J. Fluid Mech. 275, 157 (1994).
25 I. E. Zarraga, D. A. Hill, and D. T. Leighton, “The characterization of the total stress of concentrated suspensions of

non-colloidal spheres in Newtonian fluids,” J. Rheol. 44, 185 (2000).


26 A. Ramachandran and D. Leighton, Jr., “The influence of secondary flows induced by normal stress differences on the

shear-induced migration of particles in concentrated suspensions,” J. Fluid Mech. 603, 207 (2008).
27 M. Tirumkudulu, A. Tripathi, and A. Acrivos, “Particle segregation in monodisperse sheared suspensions,” Phys. Fluids 11,

507 (1999).
28 C. Voltz, M. Nitsche, L. Heymann, and I. Rehberg, “Thixotropy in macroscopic suspension of spheres,” Phys. Rev. E 65,

051402 (2002).
29 E. Brown and H. M. Jaeger, “The role of dilation and confining stresses in shear thickening of dense suspensions,” J. Rheol.

56, 875 (2012).


30 A. Ahuja and A. Singh, “Slip velocity of concentrated suspensions in Couette flow,” J. Rheol. 53, 1461 (2009).
31 D. Leighton and A. Acrivos, “The shear-induced migration of particles in concentrated suspensions,” J. Fluid Mech. 181,

415 (1987).
32 A. Sierou and J. F. Brady, “Shear-induced self-diffusion in non-colloidal suspensions,” J. Fluid Mech. 506, 285 (2004).
33 V. Breedveld, D. Van Den Ende, A. Tripathi, and A. Acrivos, “The measurement of the shear-induced particle and fluid

tracer diffusivities by a novel method,” J. Fluid Mech. 375, 297 (1998).

You might also like