Group Title: 7th International Conference on Multiphase Flow - ICMF 2010 Proceedings
Title: 10.7.3 - Lagrangian particle tracking of aggregate breakage
Full Citation
Permanent Link:
 Material Information
Title: 10.7.3 - Lagrangian particle tracking of aggregate breakage Collision, Agglomeration and Breakup
Series Title: 7th International Conference on Multiphase Flow - ICMF 2010 Proceedings
Physical Description: Conference Papers
Creator: Saha, D.
Lüthi, B.
Holzner, M.
Soos, M.
Liberzon, A.
Tsinober, A.
Kinzelbach, W.
Publisher: International Conference on Multiphase Flow (ICMF)
Publication Date: June 4, 2010
Subject: Lagrangian particle tracking
Abstract: An experimental investigation has been carried out of the breakup of statically grown polystyrene latex of arbitrary shape and size subjected to a laminar extensional ow. Three dimensional particle tracking velocimetry (3D-PTV), a non intrusive Lagrangian ow diagnostic technique, was employed to gain ow eld characteristics but also to track the motion of colloids and detect their breakup. Extensional ow was made to emerge from an axially symmetric test section converging to an ori ce of 3 3 mm2. It is found that ocs were broken apart in the high strain zone that extends 2 mm from the outset of the ori ce to approximately 35 mm inside the ori ce. The feasibility of detecting breakage events from the observed particle trajectories is tested.
General Note: The International Conference on Multiphase Flow (ICMF) first was held in Tsukuba, Japan in 1991 and the second ICMF took place in Kyoto, Japan in 1995. During this conference, it was decided to establish an International Governing Board which oversees the major aspects of the conference and makes decisions about future conference locations. Due to the great importance of the field, it was furthermore decided to hold the conference every three years successively in Asia including Australia, Europe including Africa, Russia and the Near East and America. Hence, ICMF 1998 was held in Lyon, France, ICMF 2001 in New Orleans, USA, ICMF 2004 in Yokohama, Japan, and ICMF 2007 in Leipzig, Germany. ICMF-2010 is devoted to all aspects of Multiphase Flow. Researchers from all over the world gathered in order to introduce their recent advances in the field and thereby promote the exchange of new ideas, results and techniques. The conference is a key event in Multiphase Flow and supports the advancement of science in this very important field. The major research topics relevant for the conference are as follows: Bio-Fluid Dynamics; Boiling; Bubbly Flows; Cavitation; Colloidal and Suspension Dynamics; Collision, Agglomeration and Breakup; Computational Techniques for Multiphase Flows; Droplet Flows; Environmental and Geophysical Flows; Experimental Methods for Multiphase Flows; Fluidized and Circulating Fluidized Beds; Fluid Structure Interactions; Granular Media; Industrial Applications; Instabilities; Interfacial Flows; Micro and Nano-Scale Multiphase Flows; Microgravity in Two-Phase Flow; Multiphase Flows with Heat and Mass Transfer; Non-Newtonian Multiphase Flows; Particle-Laden Flows; Particle, Bubble and Drop Dynamics; Reactive Multiphase Flows
 Record Information
Bibliographic ID: UF00102023
Volume ID: VID00270
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: 1073-Saha-ICMF2010.pdf

Full Text

7th International Conference on Multiphase Flow,
ICMF 2010, Tampa, FL, May 30 June 4, 2010

Lagrangian Particle Tracking of Aggregate Breakage

D.Saha* B. Lithi M. Holzner* M. Soost A. Liberzonj A. Tsinobert and W. Kinzelbach*

Institute of Environmental Engineering, ETH Zurich, 8093 Zurich, Switzerland
t Institute of Chemical and Bio Chemical Engineering, ETH Zurich, 8093 Zurich, Switzerland
School of Mechanical Engineering, Tel Aviv University, Ramat Aviv 69978, Israel
Keywords: Lagrangian particle tracking, strain, colloid, agglomerate, breakage


An experimental investigation has been carried out of the breakup of statically grown polystyrene latex of arbitrary
shape and size subjected to a laminar extensional flow. Three dimensional particle tracking velocimetry (3D-PTV), a
non intrusive Lagrangian flow diagnostic technique, was employed to gain flow field characteristics but also to track
the motion of colloids and detect their breakup. Extensional flow was made to emerge from an axially symmetric test
section converging to an orifice of 3 x 3 mm2. It is found that flocs were broken apart in the high strain zone that
extends 2 mm from the outset of the orifice to approximately 35 mm inside the orifice. The feasibility of detecting
breakage events from the observed particle trajectories is tested.


Colloids are constellations of tiny particles on the or-
der of micron scale. Aggregation and breakage of col-
loids are quite prevalent phenomena in a variety of flows
spanning from industrial processes such as crystalliza-
tion, separation and reaction in multi phase systems to
environmental flows like the bio geochemical cycle of
marine colloid and transport of toxic elements.
The coagulation mechanism of particles in solutions
has been studied to significant extent and clarified fairly
well by Hunter et al. (1987) while the breakup mecha-
nism of coagulated particles, especially large aggregates
of arbitrary shape remains poorly understood up to now
as reported by Soos et al. (2008). Detailed knowledge
of the turbulent flow field around a moving colloid dur-
ing its breakage is essential to understand the physical
mechanism of the underlying process. Therefore, it is
necessary to have complete access to the properties of
the turbulent flow around the colloid before, during and
after its breakage. In other words we need to know the
full Lagrangian history of the fluid flow field and of
the colloid motion. This means the experimental tool
has to be capable of tracking the colloid along its La-
grangian trajectory and at the same time to provide the
full information about velocity gradients in its proxim-
ity. Recently, Luethi et al. (2005) developed a technique
to measure all the nine components of the velocity gra-

dient tensor in a turbulent flow field via 3D-PTV. Our
aim is to apply this technique to analyze colloid breakup.
Before analyzing the effect of turbulence, the problem of
breakage is investigated in a simple flow, i.e. in a lami-
nar flow converging to an orifice of 3 x 3 mm2. This is
what we intend to do in this proposed work by using the
Lagrangian measurement technique of 3D-PTV.
Extensive efforts have been invested from both ex-
perimental and simulation side for in depth analysis of
breakage dynamics of a variety of flocs under various
flow conditions. Different theoretical propositions and
experimental tools were surfaced over the decades to
gain an unequivocal interpretation of aggregate break-
age. As for the experimental investigations, Sonntag
and Russel (1986), Higashitani et al. (1991) and Blaser
(2000) performed (theoretical and experimental) stud-
ies on breakup of latex flocs subjected to a contraction
flow. Sonntag and Russel and Higshitani et al. ob-
tained the size of the broken flocs as a function of exten-
sional rates in converging flows. Also, Higashitani and
Blaser conducted directly observed breakup and defor-
mation of flocs formed with polymeric and precipitated
coagulants. They found that flocs were disrupted at the
close vicinity of the inlet of an orifice because of the
extremely high elongation rate. Parker, Kaufman and
Jenkins (1972) investigated the breakup mechanism in
shear flows and clarified that aggregates exhibit not only
surface erosion of constituent particles from the parent

aggregates but also splitting of a parent aggregate into
smaller fragments.
As for theoretical investigations, various methods
have been proposed to predict the floc behavior in flows.
Adler and Miles (1979) performed a calculation to ob-
tain the critical size of aggregates in shear flow, approx-
imating the spherical aggregate as a porous body. The
relative motion between a pair of spherical particles in
flows has been investigated extensively by Batchelor and
Green et al. (1972) but only few studies are performed
on the behavior of coagulated particles. Doi and Chen
(1989) proposed the sticky-sphere model for the motion
of coagulated particles. In this model the Stokes' hy-
drodynamic drag force was assumed to act on all the
particles comprising the aggregate, even though in re-
ality they are not necessarily all exposed to the flow di-
rectly. Hence this model is applicable only to flocs com-
posed of a small number of particles in which almost
all particles are exposed directly to the fluid. Bossis and
Brady (1984) proposed the Stokesian dynamics in which
the hydrodynamic movement of individual particles in a
floc is calculated rigorously. But this method requires an
extremely long computational time to simulate the 3D
movement of a sufficiently large floc.
Measurements addressing breakage have been per-
formed in various types of flows. Sonntag and Russel
(1986) performed their experiment in simple shear flow,
Blaser (2000) employed 2D straining flow generated in
four roll mill apparatus, Kobayashi (1999) used exten-
sional flow generated by a sudden contraction of a pipe
exemplify the effort to get a comprehensive insight of
the dynamics. Glasgow et al. (1982) and Soos et al.
(2008) studied turbulent jets and a stirred tank equipped
with different types of impellers. Stirred tanks operated
under turbulent conditions are closest to the industrial
conditions. However, up to now there is no direct study
of the three dimensional breakage event because experi-
mental techniques were not developed sufficiently to of-
fer the required resolution allowing the observation of
the motion of the particles of small size and simultane-
ously accessing the details of the turbulent flow field car-
rying them.
Therefore the intricacy of the problem clearly pushes
the choice towards the units operating under laminar
conditions where a narrower distribution of the hydro-
dynamic stress exists, e.g., Blaser et al. (2000) con-
ducted an experiment in rheometers and four roll mills.
Kobayashi et al. (1999) made an experiment in contract-
ing nozzles. The contractile flow was found to be ef-
fective in the breakup of flocs. Contractions in flow
are typical situations where high velocity gradients oc-
cur and therefore contribute to the breakup of flocs. In
the contractile flow, flocs near the centerline and wall
will be broken by the elongational and shear stresses

7th International Conference on Multiphase Flow,
ICMF 2010, Tampa, FL, May 30 June 4, 2010

respectively as concluded by Higashitani et al. (1991).
Moreover implementing high speed cameras enables the
detection of individual breakage events. Glasgow and
Hsu (1982) and Liu and Glasgow (1997) applied this
method to study the breakup of aggregates in turbulent
jet (600>Rejt>5400). The objective of their study was
to observe the disintegration of the individual flocs pho-
tographically and deduce the magnitude of the force re-
quired to produce the break up. It was attempted to mea-
sure the gradients of the flow field and track the colloid
as well as the broken fragments simultaneously.
In the present experimental effort we deal with stati-
cally grown polystyrene latex aggregates. However, due
to the technical limitations outlined in detail below, ag-
gregates and aggregate fragments have to be relatively
large (several tens of microns). The prime purpose of the
present work is to substantiate our notion of the break-
age behavior through a relatively simple experiment in
laminar flow by employing sophisticated imaging hard-
ware like high speed cameras, strong illumination source
(laser), as well as generic optical elements (lens, beam
expander, pi shaper etc.) alongside with the conventional
laboratory equipment like pumps. On this course, we
attempt to carefully orchestrate the existing 3D-PTV al-
gorithm in a way to enhance its ability i.e., to make the
code realize the breakage event visible in reality with
high accuracy. This will certainly advocate the compe-
tence and utility of 3D-PTV as a reliable tool not only
for tracking a lonely particle in a flow but also to track
detached particles from the original one. First we as-
sess the feasibility to track breakage. Flow field and gra-
dients are studied separately and thereafter we proceed
to extend our understanding of breakage based on a al-
ready well perceived laminar flow field which will pave
the way for further analysis in increasingly complicated
turbulent flow close to realistic environment.

Materials and Methods

3D-PTV (Three dimensional particle tracking ve-
locimetry) is a non intrusive Lagrangian flow measure-
ment technique. 3D-PTV can be divided into two ma-
jor parts: determination of particle positions in space
coordinates and tracking of individual particles in time.
Through a stereoscopic principle and careful calibration
of the camera position and orientation, 3D positions are
computed from 2D images of the tracer particles This
is very similar to what our eyes do in daily life to judge
distances and positions of objects and persons. Individ-
ual particles from the object space are recognized into
the image space of the camera. This is referred to as
detection phase. Then in the correspondence phase, 2D
particle coordinates of the species are linked in 3D carte-

7th International Conference on Multiphase Flow,
ICMF 2010, Tampa, FL, May 30 June 4, 2010

sian space based on epipolar constraints. Having 3D po-
sitions established, the next step goes on to track corre-
sponding particles through consecutive time steps. This
is called tracking phase. Trajectories are post processed
and velocity gradients are obtained by using the methods
described by Luethi et al. (2005).
Typically two to four digital cameras are focused
onto an illuminated volume. While two cameras would
be enough, four cameras give some welcome redun-
dancy. The field of view is 14 x 14 mm2 and illumi-
nated strongly by a 20 watt Ar-Ion laser (514 nm of
wavelength). Four cameras record at a frame rate of
430 Hz for 30 seconds which produces approximately
170 GB of data. We utilized four MC132X high speed
CMOS GigE vision cameras (Mikrotron GmbH, Ger-
many) to capture the image sequence in time. The
flow was seeded with two types of particles: flocs
(aggregates) and neutrally buoyant tracers which were
polyamide particles with a mean diameter of 80pm
(trade name: Vestosint, manufactured by Evonik Indus-
tries, Germany). In the particle tracking system that
is available at the laboratory of IfU (Institute for En-
vironmental Engineering) this problem is solved by us-
ing many hard disks simultaneously to write out the im-
age stream. Required hardware and related software for
recording (trade name: Streams) was manufactured by
1O Industries, Canada.
Flow chamber
The setup is made of PVC. It has dimensionally iden-
tical twin chambers (75 mmx25 mm) set apart by a
20 mm long rectangular channel (3 mm x3 mm) with
sharp edges both in entrance and exit (see Fig. 1 and
Fig. 2). This narrow channel develops the contraction
regime. PVC reflects light and is also liable to melt
upon even intermediate laser intensity within less than
a minute. To prevent reflection, the observation volume
was painted with ordinary black marker. Quality spray
paint is not suitable in this case as it tends to increase the
surface height by a few percent of a millimeter which as
a consequence does not allow to accommodate the cal-
ibration target inside the chamber. Risk of leakage was
eliminated by putting a number of closely spaced screws
tightening the glass cover on the face of the apparatus
and as an extra precaution a thin rubber thread was put
in place in between the line of contact of glass and PVC.
Floc and carrier fluid preparation
Aggregates were prepared from surfactant free mono
disperse white sulphate polystyrene latex of diameter
420 nm manufactured by Interfacial Dynamics Corp.
(IDC), Portland, OR (USA) (Product- No: 1-800, cumu-
lative variation = 2%, batch no: 642, 4, solid% = 8.0).
15 mL of 2.5 molar NaCl solution was mixed with 15
mL of primary particle solution (0.3 mL of latex solu-
tion dissolved in 15 mL deionized water) to get at the

Figure 1: Schematic showing the complete flow geom-

Figure 2: Schematic showing enlarged view of the ori-

end 1.25 molar salt solution with concentration of pri-
mary particles 10 4. This two solutions are prepared in
two small vials and then mixed together by gentle shak-
ing. Approximately 12 hours are required to complete
the aggregation process. Flocs appear as fluffy little
chunks with random shape and size suspended in the so-
lution. Besides the destabilizer (NaC1) no added chemi-
cal agent was used to enhance aggregation or strengthen
the floc structure. To reduce the sedimentation or buoy-
ancy of produced aggregates, the concentration of the
salt solution was selected such to have the same den-
sity as the polystyrene particles (1.05 kg/m3). There-
fore, flocs produced by static aggregation are vulnerable
to moderate external physical disturbance leading to any
undesirable breakage. On one hand their susceptibility
to disintegrate facilitates breakage in laminar converg-
ing flow but on the other hand this floc fragility asks for
additional attention while connecting to the carrier flow
to avoid breakage far beyond the investigation volume.
Arbitrary initial floc size distribution is attributed to sta-
tic aggregation that lacks any chemical coagulantt to in-
fluence strength and size) or mechanical (stirring with a

7th International Conference on Multiphase Flow,
ICMF 2010, Tampa, FL, May 30 June 4, 2010

Table 1: Design values of the experimental setup.
Dimension Value Units
Chamber length 75 mm
Chamber width 25 mm
Chamber height 12 mm
Channel length 20 mm
Channel width 3 mm
Channel height 3 mm
Inlet nozzle diameter 15 mm

Table 2: Operating parameters.
Parameter Value Units
Syringe pump flow rate 88 ml/min
Peristaltic pump flow rate 44 ml/min
Laser power 20 watt
Camera frequency 430 Hz
Spatial resolution 1280x 1024 pixel

prescribed rpm) control over the floc population.

Since flocs were quite delicate in nature, the aggregate
solution was gently introduced into the flow chamber
from the bottom part using a syringe (120 ml in volume)
mounted on a syringe pump (Lambda VIT-FIT Syringe
Pump, Lambda Laboratory Instrument, Switzerland) at a
volumetric flow rate of 47 ml/min. Under this flow rate
the corresponding Reynolds number inside the chamber
(20 x 25 mm2 ) and orifice (3 x 3 mm2) is equal to 65 and
488 respectively. A steel tube capable of smoothly slid-
ing back and forth in the chamber was specially tailored
to act as a moving inlet for the aggregate. Carrier fluid,
in this study a simple electrolyte solution of 1.25 molar
NaC1, seeded with the tracer particle was pumped in by
a peristaltic pump (ISMATEC Laboratory Instrument,
Switzerland) at a volumetric flow rate of 88 ml/min from
both sides of the chamber to ensure uniform flow. The
inlet tube was chosen long enough to dampen the rel-
atively mild pulsating effect exerted on the flow by the
pump. Above this flow rate, the fluid velocity inside the
orifice becomes faster than what our system is able to
track at the frequency of 430 Hz and below this flow
rate colloid takes longer time to be transported into the
observation regime. Longer recording time is inevitable
to observe the paths of many particles at a low flow rate
in this case and as a consequence data swells in no time
loading the camera memory. That is why we needed
to synchronize our optimum experimental demand with
available resources to analyze the result.

Figure 3: Laboratory arrangement showing the setup
recorded by four cameras and illuminated by a laser


The first part of this section is dedicated to the charac-
terization of the contracting flow where we show the re-
sults obtained from a 3D-PTV experiment using tracer
particles and in the second part the focus is on colloid
breakage where the results are taken from an experiment
where both tracer particles and flocs were seeded into the
Part I: Flow field characterization
The motion of tracer particles converging into the ori-
fice was tracked via 3D-PTV. The Lagrangian measure-
ments of the instantaneous velocity and its spatial deriv-
atives are also interpolated onto a regular Eulerian grid
with a spacing of 0.7 mm. Fig. 4 shows an ensemble of
tracer trajectories that could be tracked over a time that is
longer than 20 mm. It is visible that many particles could
be tracked over the whole measurement domain. Due to
the contraction, the measured velocity grows from about
4 mm/sec in the chamber to approximately 20 cm/sec in
the channel, which is consistent with the decrease of the
cross sectional area. Fig. 7 shows velocity contours in
the channel.
Due to the sharp increase of the velocity it was more
difficult to track particles in the proximity of the orifice.
Fig. 6 shows a Probability Density Function (PDF) of
the length of the measured trajectories. The trajectories
shown in Fig. 4 were generated under certain geometric
constraints. For the current experiment, the field of view
started at 4.5 mm upstream of the orifice and extended
approximately up to 14 mm downstream from the orifice
entrance. The trajectories, which travelled at least 2 mm
from both sides of the orifice entrance are taken into ac-
count. So, longer trajectories coming from deep in the
flow chamber, a region which is out of the field of view,
are not considered and hence have no contribution to the

7th International Conference on Multiphase Flow,
ICMF 2010, Tampa, FL, May 30 June 4, 2010

PDF. But, this declining trend of the trajectory length is
quite typical due to several reasons.
The prime suspect of unwanted trajectory termination
is abrupt increment of velocity that challenged the tem-
poral resolution of the cameras. Operating camera fre-
quency (430 Hz) apparently can not cope with this ve-
locity gradient all over the experiment run time. Other
culpable factors causing this unwanted cease are associ-
ated with intensity fluctuation of the illumination source,
out of focus problem and the flow geometry itself. How-
ever, all these difficulties did not prove to be alarming
as we have scores of long enough trajectories travelling
through the entire observation domain.
Contour slices of the measured rate of strain, s2
sij sij, where sij are the nine components of the rate of
strain tensor, is depicted in Fig. 5, where the colorbar is
on logarithmic scale. The figure shows that the contrac-
tion is characterized by an increase of s2 over several
orders of magnitude. More precisely, in the chamber s2
is on the order of 0.1-1 s 2 and reaches values on the
order of 104 s 2 in the center of the orifice.
One can estimate from the laminar profile that at
the orifice walls the boundary layer is associated with
a shear stress of the same order of magnitude, s2
104s-2. At the orifice, both the contraction and the wall
shear give similar contributions to the total stress. Be-
yond the orifice and deeper inside the small channel, as
the velocity profile converges to the laminar profile of a
rectangular duct, the contribution of the boundaries will
be dominating.
Fig. 7 shows the velocity contours inside the inves-
tigated zone. The very first slice perpendicular to the
stream wise direction corresponds to the cross-section
at the orifice location. The velocity gains its maximum
value, which is approximately 17 cm/sec, from 1 mm
upstream of the orifice and stays beyond 10 mm down-
stream. The orifice length is 20 mm, which means
that this maximum velocity should exist up to 45 mm
in stream wise direction. From Fig. 7 it might seem
that velocity decays even inside the orifice after 35 mm.
Here, it should be carefully noted, our investigation do-
main (field of view of the cameras) ends at around 35
mm downstream and that is why maximum velocity af-
ter this length is not observed as the tracer particles are
not tracked anymore. However, comparing the velocity
of near wall region of the orifice with the velocity in-
side it, we found that within a millimeter range in span
wise distance, velocity is increased by a factor five. This
high velocity gradient causes an immense shear and is
responsible for the violent breakage of the colloids pass-
ing close by the walls as seen from the experiment.
Part II: Breakage tracking
The results in this part are obtained from a 3D-PTV
experiment where both colloidal aggregates and tracer

0.o01 0003 0.04
0.015 0.02 0.025
x (m) y (m)

Figure 4: Trajectories measured through PTV converg-
ing into the orifice. The trajectories stem from 30 sec-
onds of recording at a recording rate of 430 Hz.

0.02 10
0.015 0
y (m) x (m)

x 103

Figure 5: Strain (1/s2) contours on loglO scale, the iso-
surface renders the zone of strain which lies in the vicin-
ity of the orifice. This zone is expected to be the break-
age region for colloids and thereby is in the special focus
of this study.

Trajectory length (mm)

Figure 6: Pdf of trajectory length

7th International Conference on Multiphase Flow,
ICMF 2010, Tampa, FL, May 30 June 4, 2010

0 01- -
0 005

x 10

-- 003
12 002 0025
0 02
x(m) y(m)

Figure 7: Velocity (m/sec) slice inside the observation

particles were seeded into the flow. The results are pre-
liminary and focus on one single breakage event as a
representative example. Since colloids are breaking in
extensional flow, the evolution of the stream wise ex-
tension of the particle image drops after the event and
appears to be a good indicator of breakage. Fig. 8 (left)
shows a snapshot of a breakage event on an image from
one of the cameras. The particle is marked on the im-
age with a white cross and the snapshot corresponds to
the time instance when the colloid visually appears al-
ready broken as two parts which can be clearly distin-
guished. They separate more clearly at later times. On
the snapshot (Fig. 8 left) the two particles are not to-
tally disconnected, that is, the two parts are connected
by a few pixels with non-zero pixel value. Two parti-
cles can appear as a single one for several reasons, the
most plausible one in this case being that the flow ve-
locities in the orifice are high and therefore the two par-
ticles leave streaklines on the image that may connect
to each other. Fortunately our PTV algorithm is able
to cope with this problem. Namely, the algorithm in-
cludes a feature that is based on the property that grey-
values drop continuously away from the center of grav-
ity of a particle. Maas et al. (1993) used this principle
and developed a modified anisotropic thresholding oper-
ator, which searches for discontinuities in the grey val-
ues. The motivation of Maas was that in a PTV experi-
ment with high seeding densities of tracer particles am-
biguities arise frequently due to the problem of particles
optically blocking or overlapping each other in one or
more views. Now this operator turns out to be very use-
ful for our breakage problem. Fig. 8 (centre and right)
shows how the algorithm behaves with two choices of
the "tolerable discontinuity" threshold on the greyvalue.
For a rather low greyvalue threshold of 30 (Fig. 8 cen-
ter) the two broken parts of the colloid are successfully
detected and tracked. For a higher value of 80 (Fig. 8
right) the two parts are recognized as a single particle
and only one trajectory is measured. Fig. 9 and 10 show
the evolution of the streamwise length (in pixels) of the
particles for the two selected thresholds.

Figure 8: Figure showing how a certain parameter
called "Tolerable discontinuity" in the PTV code re-
sponse to a breakage event. (Left) A snap shot of the
a colloid at the time it breaks into two, (middle) at tol-
erable discontinuity = 30, two separate trajectories are
generated from two broken parts as expected and (right)
at tolerable discontinuity = 80 a single trajectory holds
two broken parts as if they were an unified body

Fig. 9 shows that the breakage of a single colloid set-
ting tolerable discontinuity of 80. At this high tolerance,
the event is not recognized as breakage and an increase
of pixel value from 30 to 55 represents only an elonga-
tion of the colloid masking the true event. On the other
hand, Fig. 10 shows that when the same event is ana-
lyzed at tolerable discontinuity of 30, a single colloid
having an initial pixel value of 35 is broken into two
parts having pixel values of approximately 10 and 20 re-
spectively. Fig. 9 also shows that part of the trajectory
is missing, i.e., the orifice starts from 24.5 mm as high-
lighted by deep green dotted line and pixel information
from 24.15 mm to 25.89 mm in downstream direction is
lost. This is due to the fact that the 3D stereomatching
was not successful in that locations. A possible reason is
that the two parts were still recognized as a single one in
one of the cameras and further checks and possibly in-
dividual tolerable discontinuity thresholds for the single
cameras need to be implemented.
This problem was cured by another supplementary al-
gorithm that will enable automatic detection of break-
age, which is still in the development phase. This part of

7th International Conference on Multiphase Flow,
ICMF 2010, Tampa, FL, May 30 June 4, 2010




0.0107 -


0.023 0.0245 0.026
Y (m)


Figure 9: Change of stream wise mean pixel value dur-
ing a breakage event at tolerable discontinuity = 80. Y
coordinate refers to stream wise position and Z refers
to perpendicular to the flow direction. Bold green line
chalks out the beginning of the orifice position.






/ I


0.022 0.0245 0.026
Y (m)


Figure 10: Change of stream wise mean pixel value dur-
ing a breakage event at tolerable discontinuity = 30. Y
coordinate refers to stream wise position and Z refers
to perpendicular to the flow direction. Bold green line
chalks out the beginning of the orifice position.

the detection code should be fine tuned in a way to dis-
card the phenomena where two trajectories may give rise
to a false implication of breakage. However, for our rel-
atively simple and single breakage event, the code ma-
tured so far performs quite well. The principle of this
detection routine is based on recursive neighborhood
search of new trajectories around the candidate trajec-
tories of the broken parts of colloid. If there exists such
a close neighbor trajectory then the detached trajectories
from the broken parts are back extrapolated in time with
an expectation to meet another trajectory which is likely
to reach its terminal point. This certainly raises the pos-
sibility of finding a breakage event with increased accu-
racy. To make things sure, this trajectory from its dead


-10 -5


5 10 15

Figure 11: Normalized pixel area is plotted against
frame. Here frame 0 stands for the mid point of the ex-
trapolation path.

end is then forward extrapolated and if it meets with the
neighbor of the daughter trajectories then the event is
identified as breakage. Fig. 12 shows the change of to-
tal particle area along the trajectories and we see that the
two fragments have an area of a bit less than half the area
of the original agglomerate.


Flocs exposed to a contraction are elongated and aligned
with the streamline and disintegrated at the beginning of
the orifice where the flow velocity increases abruptly.
Detached parts from the parent aggregate stay at a
span wise distance of 0.05 mm. Since the colloids are
snapped on the image space in consecutive time steps
and preserved as pixels, 3D-PTV can determine size dis-
tribution in two dimensions as well. The most notewor-
thy feature of this technique as the results showed in this
study is, it can simultaneously determine (detection and
tracking) the full Lagrangian history of breakage event
with appreciable accuracy in space and time and char-
acterize the flow field. In the present study we utilized
larger flocs and observed them to break in a converging
laminar flow. We are continuously trying to optimize our
experimental conditions, i.e., better camera calibration,
improved focusing, higher resolution together with other
necessary details related to the laboratory environment.
With regard to data analysis, we are in the process of
modifying the current 3D-PTV code such that breakage
can be detected reliably and automatically.

* **

AA. .'

7th International Conference on Multiphase Flow,
ICMF 2010, Tampa, FL, May 30 June 4, 2010

Acknowledgements Liu S.X. and Glasgow L.A., Aggregate disintegration
in turbulent jet, Water, Air and Soil Pollution, Vol.95,
This study is supported by SNF (Swiss National Science pp.257-275, 1997
Foundation) research grant SNF Nr.: 200021-119815
Mass H.G., Gruen A. and Papantoniou D., Particle
References tracking velocimetry in three-dimensional flows, Exper-
iments in fluids, Vol.15, pp.133-146, 1993
Liithi B., Tsinober A. and Kinzelbach W., Lagrangian
measurement of vorticity dynamics in turbulent flow,
Journal of Fluid Mechanics, Vol. 528, pp. 87-118, 2005.

Hunter R.J., Foundation of colloid science, Vol.I, Ox-
ford:Clarendon press, 1987

Sonntag R.C. and Russel W.B., Structure and breakup
of flocs subjected to fluid stresses, Colloid interface sci-
ence, Vol.113, pp. 399, 1986 Vol.166, pp.215-223, 1999

Parker D.S., Kaufman W.J. and Jenkins D., Floc breakup
in turbulent flocculation process, Proceedings of the
Americal Society of Civil Engineers, Journal of sanitary
engineering division, Vol. 98(SA1), pp. 79-99, 1972

Higashitani K., Inada N. and Ochi T., Floc breakup
along centerline of contractile flow to orifice, Colloids
and surfaces Vol.56, pp.13-23, 1991

Adler PM. and Miles P.M., Motion and rapture of a
porous sphere in a linear flow field, Journal of Rheol-
ogy, Vol.23, pp.25-37, 1979

Batchelor G.K and Green J.T., The hydrodynamic inter-
action of two freely moving spheres in linear flow field,
Journal of Fluid Mechanics, Vol.56, pp.375-400, 1972

Doi M. and Chen D., Simulation of aggregating col-
loids in shear flow,Joumal of chemical physics, Vol.90,
pp.5271-5279, 1989

Bossis G. and Brady J.F., Dynamic simulation of
sheared suspension.I.General method, Journal of chem-
ical physics, Vol.80, pp.5141-5154, 1984

Blaser S., Flocs in shear and strain flows,Joumal of col-
loid and interface science, Vol.225, pp.273-284, 2000

Kobayashi M. and Adachi Y., Breakup of fractal flocs in
a turbulent flow, Langmuir, Vol.166, pp.215-233, 1999

Glasgow L.A. and Hsu J.P, An experimental study of
floc strength, AIChE journal, Vol.28, pp.779-785, 1982

Soos M., Moussa A., Ehrl L., Sefcik J., Wu H. and
Morbidelli M., Effect of shear rate on aggregate size
and morphology investigated under turbulent conditions
in stirred tank, Journal of colloid and interface science,
Vol.319, pp.577-589, 2008

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - Version 2.9.7 - mvs