ARTICLE IN PRESS

Chemical Engineering Science 65 (2010) 44724483

Contents lists available at ScienceDirect

Chemical Engineering Science

journal homepage: www.elsevier.com/locate/ces

Modeling the evolution and rupture of stretching pendular liquid bridges

Pirooz Darabi, Tingwen Li, Konstantin Pougatch, Martha Salcudean , Dana Grecov
Department of Mechanical Engineering, University of British Columbia, 6250 Applied Science Lane, Vancouver, BC, Canada V6T 1Z4

a r t i c l e in f o

a b s t r a c t

Article history:
Received 21 July 2009
Received in revised form
26 March 2010
Accepted 2 April 2010
Available online 8 April 2010

A simplied mathematical model and numerical simulations of the governing NavierStokes equations
are used to predict the shape evolution, rupture distance, and liquid distribution of stretching pendular
liquid bridges between two equal-sized spherical solid particles. In the simplied model, the bridge
shape is approximated with a parabola, and it is assumed that the surface tension effects dominate the
viscous, inertial, and gravitational effects. For the numerical simulations, a commercial Computational
Fluid Dynamics (CFD) software package FLUENT is used. The rupture distance predictions obtained
with both models are compared with experimental data and a reasonable agreement is found. The
results of the numerical investigations show that for simulations with negligible viscous, inertial, and
gravitational effects, the rupture distance approaches an asymptotic value, which is close to the value
predicted by the simplied model. The bridge proles predicted using the simplied model and the
numerical simulation are compared. It is found that a second-order polynomial appropriately
represents the stable bridge shape for particles with identical contact angles; however, for liquid
bridges between particles with different contact angles, the numerical simulations of the governing
NavierStokes equations should be used.
& 2010 Elsevier Ltd. All rights reserved.

Keywords:
Agglomeration
Liquid bridge
Rupture
Granulation
Mathematical modeling
Particle

1. Introduction
The statics and dynamics of liquid bridges have been studied
for a long time. These investigations started with the pioneering
works of Haines (1925) and Fisher (1926) who studied the
liquid bridges formed between two spheres on contact. A
comprehensive chronological review of the research up to 1980
concerning the pendular bridges between solid bodies was
conducted by Mehrotra and Sastry (1980), who showed that
signicant improvements could be made in this area. Following
their review, a signicant amount of research was conducted in
this area, with the following two major goals: (1) improving the
force and energy expressions associated with pendular liquid
bridges (e.g., Chan and Horn, 1985; Darabi et al., 2009; Ennis et al.,
1990, 1991; Lian et al., 1993; Mazzone et al., 1987; Mikami et al.,
1998; Pitois et al., 2000, 2001; Rossetti and Simons, 2003; Rossetti
et al., 2003; Simons et al., 1994; Soulie et al., 2006) and (2)
obtaining a better understanding of the stability and modeling the
evolution of pendular liquid bridges (Dai and Lu, 1998; De
Bisschop and Rigole, 1982; Lian et al., 1993; Mazzone et al., 1986;
Pepin et al., 2000a, 2000b; Shi and McCarthy, 2008).
Rupture of liquid bridges is an interesting phenomenon that
has been studied for many years. These studies were initially on
static liquid bridges (Coriell et al., 1977; Gillette and Dyson, 1971;

 Corresponding author. Tel.: + 1 604 822 2732; fax: + 1 604 822 2005.

E-mail address: msal@interchange.ubc.ca (M. Salcudean).

0009-2509/$ - see front matter & 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ces.2010.04.003

Haynes, 1970). The dynamics and rupture of liquid bridges was

not extensively studied until the work of Fowle et al. (1979).
Following their work, similar studies were performed on
oscillating and stretching liquid bridges (Mollot et al., 1993; Sanz,
1985; Tsamopoulos et al., 1992; Zhang et al., 1996; Zhang and
Alexander, 1990). The primary parameter to consider for the
rupture of liquid bridges is the rupture distance. Rupture distance
is an important parameter that is required for discrete element
modeling (DEM) of granules with liquid bridges in the pendular
and funicular states; moreover, an accurate estimation of this
distance leads to better prediction of the behavior and strength of
wet agglomerates. We divide the models used for calculating the
rupture distance of pendular liquid bridges into the following
three categories: (1) YoungLaplace equation, (2) simplied
model, and (3) NavierStokes equations.
The YoungLaplace equation relates the pressure deciency
and liquidsolid surface tension to the mean curvature of the
liquid bridge. The equation does not lend itself to an analytical
solution except for a few special cases (Orr et al., 1975). Therefore,
efforts have been made to solve it numerically, e.g., Erle et al.
(1971) and De Bisschop and Rigole (1982). Mazzone et al. (1986)
compared the theory developed by De Bisschop and Rigole with
experimental data and argued that the theory underestimates the
rupture distance; consequently, they developed a modied theory
that considered gravitational effects and showed better
agreement. Lian et al. (1993) studied the liquid bridge between
spherical bodies and determined the rupture distance by
minimizing the system free energy. They showed that for

ARTICLE IN PRESS
P. Darabi et al. / Chemical Engineering Science 65 (2010) 44724483

identical particles with contact angles up to 401, the rupture

distance can be written as a simple function of the liquid bridge
volume and contact angle. Later, similar expressions were
proposed by Mikami et al. (1998) for the case of two identical
particles, and for a particle and a wall. Note that using the
YoungLaplace equation for determining the rupture distance
(by minimizing the system free energy) usually requires constant
and zero liquidsolid contact angles (Pepin et al., 2000a), which is
one of the main limitations of this model.
An alternative approach is to approximate the liquid bridge
shape with a priori assumed prole, which we term the
simplied model. The advantages of the simplied model are
two-fold: rst, the solution to the YoungLaplace equation is no
longer required; and second, the model is easily extendable to
particles with varying sizes and/or with different contact angles.
Dai and Lu (1998) used a circular liquid bridge prole and
employed the theory of liquid cylinder stability to predict the
rupture distance. They used a correction factor to match their
predictions with experimental data, which somewhat reduces the
generality and reliability of their model. Pepin et al. (2000a)
employed both the toroidal and parabolic approximations to
model pendular liquid bridges in the presence of wetting
hysteresis. They showed that even though both proles have
different mathematical expressions, their approximations agree
well with their experimental data. Pepin et al. also pointed out
that the toroidal approximation is slightly less accurate and more
difcult to use since one of the parameters of the toroidal
approximation approaches innity when the bridge shape
converts from convex to concave, causing numerical difculty.
Shi and McCarthy (2008) also argued that the toroidal approximation is restricted to particles that have the same contact
angles. In conclusion, the parabolic approximation is more robust
and more general than the toroidal approximation. However,
there are a number of shortcomings attributed to the simplied
model. For instance, it assumes that rupture occurs at quasi-static
conditions and that the bridge is stable throughout its evolution.
Also, it presumes that the surface tension effects dominate the
inertial, viscous, and gravitational effects. These assumptions
require further investigation.
The nal model is the numerical solution to the governing
NavierStokes equations. Of particular importance to this study is
the work of Zhang et al. (1996), who performed extensive
numerical investigations of stretching liquid bridges between
two circular disks. Similar to Eggers and Dupont (1994) and
Papageorgiou (1995), they used the slender jet approximation to
simplify the governing NavierStokes equations. Consequently,
they derived two one-dimensional time-dependant differential
equations for the axial velocity and bridge prole. They then
simulated the dynamic response of the bridge to continuous
uniaxial stretching. Their work was on liquid bridges between two
circular disks rather than two spherical particles, and simplied
NavierStokes equations were solved one-dimensionally.
The objectives of the current work are as follows: (1) to obtain
a better understanding of the shape evolution, rupture, and liquid
distribution of stretching pendular liquid bridges between two
equal-sized spherical particles; (2) to study the problem of
interest with the simplied model and numerical simulations to
the full NavierStokes equations; and (3) to investigate the
accuracy and applicability region of the simplied model. For our
simplied model, the system of equations proposed by Shi and
McCarthy (2008) is combined with a closure criterion proposed by
Pepin et al. (2000a). Note that the simplied model is based on
several assumptions; for instance, it assumes that the inertial,
viscous, and gravitational effects are negligible compared with the
surface tension effects, and that the bridge has a parabolic shape.
In order to solve the governing NavierStokes equations numeri-

4473

cally, a commercial Computational Fluid Dynamics (CFD) software

package FLUENT is used. The main advantages of this model
are that it solves the full NavierStokes equations with interface
boundary conditions without the above-mentioned assumptions
of the simplied model. The numerical simulations are then used
to evaluate the key assumptions of the simplied model.
This paper is structured as follows: First, the problem of
interest is stated, then the simplied model, including all the key
assumptions and equations and our solution methodology, is
described. The numerical simulations, including the simulation
setup, solution methodology, and grid independence study, are
then presented, followed by a discussion of the non-dimensional
analysis of the rupture problem. The next section presents the
results and discussion of the simplied model and numerical
simulations. The limitations of the simplied model are then once
again highlighted. Finally, the summary and conclusions are
presented.

2. Problem statement
Fig. 1 schematically shows a liquid bridge connecting two
spherical particles, at its critical distance. Note that all the
parameters related to the left-hand side particle are marked
with index i, and those for the other particle are marked with
index j. It is assumed that the bridge is axisymmetric with respect
to the x-axis; thus, only its upper section is considered in this
work. Also, the particles are assumed to be smooth and
non-porous. The interface of the bridge with each particle is a
spherical cap with height h. The spheres radii (R), the length of
the liquid bridge (d), the critical separation distance, also known
as the rupture distance (Sc), and the contact angles (y) are all
labeled in this gure.

3. Simplied model
As mentioned before, our simplied model is a combination of
the system of equations developed by Shi and McCarthy (2008)
and the closure criterion by Pepin et al. (2000a). The equations
representing the parabolic shape of the bridge are presented in
the following section, followed by a discussion of the closure
criterion for determining the rupture distance.
3.1. Parabolic approximation
Assuming a parabolic shape for the gasliquid interface of
the bridge, its prole can be approximated with the following
second-order polynomial:
y ax2 bx c

where a, b, and c are unknown. As depicted in Fig. 1, the

gasliquidsolid phases intersect at two points with the following
coordinates: {0, y(0)} and {d, y(d)}. Considering the problem
geometry as given in Fig. 1, the y-coordinate of these two contact
points can be related to the cap heights (h) and particle radii (R) as
follows:
q
2
y0 R2i Ri hi 2
yd

q
R2j Rj hj 2

The cap heights can be related to the liquid bridge length and
critical separation distance with the following equation:
d Sc hi hj

ARTICLE IN PRESS
4474

P. Darabi et al. / Chemical Engineering Science 65 (2010) 44724483

Fig. 1. Pendular liquid bridge between two spherical particles at rupture instant.

By referring to Fig. 1, the contact angles (y) can be related to

the liquid bridge prole as follows:


p
y0
5
yi arctany0 0arcsin
2
Ri

yj



yd
arctany0 darcsin
2
Rj

The liquid bridge volume (V) is assumed to be constant and is

determined by
Z d
V p
y2 x dxVcap,i Vcap,j

where the volume of spherical caps can be written as

Vcap,i

Vcap,j

phi
6

phj
6

3y2 0 h2i

3y2 d h2j

In summary, in order to determine the liquid prole and its

rupture distance, the system of Eqs. (2)(7) are solved. However,
there are seven unknown variables (a, b, c, d, hi, hj, Sc) and only six
equations available, meaning that one additional equation is
required. The following section provides the closure of the above
system of equations.

Fig. 2. The liquid droplet on particle i immediately after rupture.

with the following equations:

ui

Ti2 y2i
2Ti

Ti Ti2 3y2i

11
6VL,i

hi h2i 3y2i

12

Similar equations are written for the radius and maximum cap
height of the liquid droplet on the right-hand side particle.

3.2. Rupture point

3.3. Solution methodology
Fig. 2 schematically shows the liquid droplet, immediately
after rupture, on particle i. In this gure, notations S, L, and G
stand for the solid, liquid, gas phases, respectively. We adopt
the assumption of Pepin et al. (2000a) that the rupture of
the liquid bridge occurs when the area of the bridge (before
rupture) is equal to the area of the two droplets formed on the
two particles (after rupture); this assumption is written as
follows:
Z d
q
yx 1 y_ 2 x dx 2pui Ti uj Tj
2p

10

In order to determine the rupture distance with the simplied

model, the system of equations consisting of Eqs. (2)(7) and
(10)(12) is solved using the Matlab non-linear solver.
The simplied model requires the radii of the particles, the
liquidsolid contact angles, and the liquid bridge volume as initial
conditions; with this information, the simplied model calculates
the critical separation distance (i.e., rupture distance), liquid
bridge shape (the unknown coefcients of the parabola), radii of
liquid droplets, etc.

Note that Eq. (10) is the nal equation that closes the system of
equations to determine the rupture distance using the simplied
model. In order to calculate the area of the two droplets, their size
information is required. The radius and maximum cap height of
the liquid droplet on the left-hand side particle are calculated

4. Numerical simulation
As mentioned earlier, for our numerical simulations, we
numerically solve the full NavierStokes equations to study the
evolution of stretching pendular liquid bridges. For this purpose,

ARTICLE IN PRESS
P. Darabi et al. / Chemical Engineering Science 65 (2010) 44724483

4475

we use a Computational Fluid Dynamics (CFD) software package

FLUENT. Note that our problem involves tracking the deforming
interface between the liquid and gas phases. In the current study,
the Volume of Fluid (VOF) method is used for tracking the
interface of the two uids. The VOF method has been implemented
in several commercial CFD packages and their performances have
been evaluated for several cases (Glatzel et al., 2008; Ozkan et al.,
2007). In the VOF method, the interactions of the immiscible uids
are modeled by solving a single set of momentum equations and
tracking the volume fraction of each uid throughout the domain.
In order to describe the interface deformation, the evolution
equation is solved using a special advection scheme, namely
interface reconstruction. In FLUENT, the continuum surface force
(CSF) model proposed by Brackbill et al. (1992) is used to account
for the surface tension effects; in this model, the surface tension
forces are treated as source terms in the momentum equation. To
account for wall adhesion, the liquidsolid contact angle is used to
adjust the normal vectors in cells near the wall. This results in the
adjustment of the curvature of the surface near the wall and then
is used to adjust the body-force term in the calculations. Further
description of the VOF model and the governing equations can be
found in the FLUENT documentation (FLUENT Inc., 2006).

Table 1
Values for the convergence-related parameters.

4.1. Simulation setup

The Geo-Reconstruction scheme is used for interface interpolation; this scheme represents the interface between the uids
using a piecewise-linear approach.

Fig. 3 shows the two-dimensional computational domain for

the gap between the two particles, discretized with a structured
quadrilateral grid. The gap is lled with both gas and liquid
phases. Note that since both particles are spherical, the problem is
assumed to be axisymmetric. The ow is considered to be
isothermal and incompressible. Due to the movement of the
two solid particles, the liquid bridge elongates continuously until
it ruptures. To model the domain evolution, the dynamic mesh
model, available in FLUENT, is used. In this model, as the surface
moves, layers of cells are added to the moving surfaces.
The model geometry is built and meshed with the geometry
and mesh generation software packageGAMBIT. The generated
case is then imported into FLUENT for further setup and
simulation purposes. For our setup, a segregated time-dependent
unsteady solver is used and the pressure-staggering option
(PRESTO) scheme is enabled for pressure interpolation. The
pressureimplicit with splitting of operators (PISO) algorithm,
by Issa et al. (1991), is used to determine the pressurevelocity
coupling with skewness-neighbor correction. For the momentum
equation, the QUICK differencing scheme is employed, and time
integration is performed using the rst-order implicit scheme.

Parameter

Value

Time step (s)

Maximum number of iterations per time step
Under-relaxation factor for pressure
Under-relaxation factor for density
Under-relaxation factor for body forces
Under-relaxation factor for momentum
Scaled convergence criteria for continuity and x- and
y-momentum equations

5  10  6
100
0.3
1
1
0.3
10  3

Table 2
Summary of grid independence study.
Grid

Average cell
area (mm2)

Initial number
of CVs

Predicted dimensionless
rupture distance

Coarse
Medium
Fine

6.29  10  5
2.94  10  5
1.30  10  5

5040
10800
24300

0.638
0.634
0.633

4.2. Solution methodology

The time step, the maximum number of iterations per time
step, and the relaxation factors are carefully adjusted to ensure
convergence. The common values for the convergence-related
parameters are summarized in Table 1.
The simulations are performed with no-slip boundary
conditions at particle surfaces with a constant contact angle at
the liquidsolid interface. A constant pressure boundary condition
of 1 atm is considered for the pressure outlet boundary condition,
as shown in Fig. 3. As mentioned before, an axisymmetric
boundary condition is used for the centerline. All the simulations
are performed on a Sun workstation with 4 quad-core X7350
Xeon CPUs.
4.3. Grid independence study
A grid independence study is conducted on three grids with
different resolutions. The initial numbers of control volumes (CV)
for these grids are 5040, 10,800, and 24,300, respectively. The grid
information and predicted dimensionless rupture distances are
summarized in Table 2. According to this table, as the grid
becomes ner, the difference between the predicted rupture
distances decreases, indicating grid convergence. Any of these
grids can be used for our study; for most of the simulations in this
work, the medium grid is used. Our choice represents a
compromise between computational time and accuracy.

5. Non-dimensional analysis
Suppose that the rupture distance is a function of eight
variables, according to the following equation:
13
S f R,v, r,Vb , y, s, m,g

Fig. 3. Computational domain and grid.

where Sn is the rupture distance, R is the particle radius, v is the

separation velocity, r is the liquid density, Vb is the liquid bridge
volume, y is the liquidsolid contact angle, s is the liquid surface
tension, m is the liquid viscosity, and g is the gravitational

ARTICLE IN PRESS
4476

P. Darabi et al. / Chemical Engineering Science 65 (2010) 44724483

acceleration. Selecting {R, v, r} as the scaling parameters, it can be

shown with the pi theorem that the original functional
relationship, as in Eq. (13), can be written in the following
equivalent form:


S
V
f~ 3b , y,We,Ca,Bo
14
R
R
According to Eq. (14), the dimensionless rupture distance
(Sc Sn/R) is a function of dimensionless liquid bridge volume
(VVb/R3), contact angle (y), the Weber number (WeRv2r/s),
the capillary number (Ca mv/s), and the Bond number
(BogR2r/s). Note that the latter three dimensionless numbers,
i.e., We, Ca, and Bo, respectively, represent the ratio of the inertial,
viscous, and gravitational effects over the surface tension effects.

6. Results and discussion

6.1. Simplied model
In this section, the rupture distance predictions obtained with
the simplied model are presented. Note that all results presented
in this section and the following sections are in the form of the
dimensionless rupture distance and dimensionless liquid bridge
volumes; however, to avoid repetition, the word dimensionless
is often omitted. First, the predictions of the rupture distance
using the simplied model are compared with the experimental
data available in the literature. Then, the effects of the contact
angle and liquid bridge volume on the rupture distance are
discussed.
6.1.1. Comparison between rupture distance predicted by simplied
model and experimental data
Fig. 4 shows the comparison between the rupture distance
predictions using the simplied model and the experimental data
of Mason and Clark (1965). Note that according to Mason and
Clark, the measured contact angle was slightly greater than zero
but the exact value was not provided; therefore, for comparison
purposes, the results of the simplied model are obtained for
three contact angles (01, 101, and 201). According to Fig. 4, the best
comparison is found for contact angles between 101 and 201.
Fig. 5 displays the comparison between the rupture distance
predictions using the simplied model and the experimental data
of Mazzone et al. (1986), with a measured contact angle of 101.
According to this gure, the simplied model at y 101 slightly
over-predicts the measurements of Mazzone et al. This difference
can be related to the gravitational effects that are not included
in the simplied model. Note that according to Mazzone et al.
(1986), the gravitational effects reduce the rupture distance.
However, in the simplied model, the gravitational effects
are neglected; hence, the rupture distance is over-predicted
by this model. Further discussion regarding the impact of the
gravitational effects on the rupture distance will be presented
later in this paper.
6.1.2. Effect of liquid bridge volume and contact angle on rupture
distance
Fig. 6 shows the variations of the rupture distance with respect
to the contact angle and liquid bridge volume obtained with the
simplied model. According to this gure, the rupture distance
increases as the bridge volume increases and as the contact angle
increases. These trends are consistent with the expression of Lian
et al. (1993), Eq. (15) and Mikami et al. (1998), Eq. (16). Fig. 7
shows the comparison between the results of Eqs. (15) and (16)
and the rupture distance predictions using the simplied model.
According to this gure, very good agreement is found. In

Fig. 4. Comparison of rupture distances predicted using the simplied model and
the experimental data of Mason and Clark (1965).

Fig. 5. Comparison of rupture distances predicted using the simplied model and
the experimental data of Mazzone et al. (1986) (with measured liquidsolid
contact angles of 101).

conclusion, the simplied model agrees well with other rupture

distance expressions in the explored range of dimensionless
liquid bridge volumes. Note that all the results obtained thus far
were for particles with identical contact angles. The suitability of
the simplied model for particles with different contact angles
will be discussed later in the paper:
Sc 1 0:5yV 1=3

15

Sc 0:99 0:62yV 0:34

16

6.2. Numerical simulation

Table 3 presents a summary of the simulation conditions used
in this work. In all the simulations, to impose symmetrical
conditions, both particles are considered to move with the same
velocity. The velocity of each particle is equal to half of the

ARTICLE IN PRESS
P. Darabi et al. / Chemical Engineering Science 65 (2010) 44724483

4477

Table 3
Summary of the simulation conditions.
Parameter

Value

Unit

Particle radius (R)

Liquid bridge volume (Vb)
Dimensionless liquid bridge volume (V Vb/R3)
Relative separation velocity (v)
Gravitational acceleration (g)
Liquidsolid contact angle (y)

1
1  10  10
0.1
0.001
9.81
30

mm
m3
dimensionless
m/s
m/s2
deg.

Fig. 6. Variations of the dimensionless rupture distance with respect to the

contact angle and dimensionless liquid bridge volume.

Fig. 8. Simulated bridge evolution in time in the absence of gravitational effects:

(a) t 0, (b) t 0.025, (c) t 0.05, (d) t 0.075, (e) t 0.1, (f) t 0.125, (g) t 0.1335,
(h) t 0.134, (i) t 0.1341, (j) t 0.1342, (k) t 0.1343, and (l) t 0.1344 ms.

Fig. 7. Comparisons of rupture distances predicted using the simplied model and
expressions of Lian et al. (1993) and Mikami et al. (1998).

relative separation velocity, as given in Table 3. All the simulation

conditions remained unchanged unless otherwise specied. Note
that gravity acted along the line that connects the centers of the
particles and is neglected unless otherwise specied.

6.2.1. Bridge deformation and rupture

Fig. 8 illustrates the liquid bridge evolution over time.
According to this gure, as the particles continuously move
farther apart, the bridge continuously deforms and the neck
becomes narrower until it no longer withstands any additional
elongation and breaks. Note that the gravitational effects are not
considered in this simulation. Fig. 9 shows a similar phenomenon
with the inclusion of the gravitational acceleration acting towards
the right-hand side particle. This gure shows that due to the
presence of the gravitational effects, more liquid accumulates on
the right-hand side particle.

6.2.2. Comparison between numerical simulation and experimental

data
Fig. 10 graphically compares the rupture distances predicted
using both the numerical simulations and the simplied model
with the experimental data of Mazzone et al. (1986). According to
this gure, both models slightly over-predict the experimental
data of Mazzone et al. In addition, the gure, in general, indicates
a better prediction with the numerical simulation than with the
simplied model. The maximum relative errors of the numerical
simulations and simplied model are 9% and 15%, respectively.
The better agreement obtained with the numerical simulations
can be attributed to the inclusion of the inertial, gravitational, and
viscous effects in the simulations, while the simplied model
ignores them all.

6.2.3. Variations of the rupture distance with respect to the Weber,

capillary, and Bond numbers
According to our non-dimensional analysis, the dimensionless
rupture distance depends on the dimensionless liquid bridge
volume, contact angle, Weber number (We), capillary number
(Ca), and Bond number (Bo). The relationship between the rupture
distance, liquid bridge volume and contact angle was analyzed
using the simplied model earlier in this work. Similar trends
were observed with the numerical simulations. The two main

ARTICLE IN PRESS
4478

P. Darabi et al. / Chemical Engineering Science 65 (2010) 44724483

Fig. 9. Simulated bridge evolution in time in the presence of gravitational effects: (a) t 0.015, (b) t 0.065, (c) t 0.09, (d) t 0.115, (e) t 0.117, and (f) t 0.1175 ms.

Fig. 11. Variations of simulated rupture distance with respect to Weber number.

Fig. 10. Comparison of the rupture distance predictions using both the numerical
simulations (with Bo 1.25, Ca4.4  10  4, and We 3.2  10  5) and the
simplied model with the experimental data of Mazzone et al. (1986).

objectives of this section are: (1) to investigate the relationship

between the rupture distance and the three aforementioned
dimensionless numbers and (2) to quantify the aforesaid
assumption of the simplied model (i.e., dominance of the surface
tension effects over the inertial, viscous and gravitational effects)
with the aid of our ndings. The latter is achieved by performing
three series of numerical simulations with different capillary,
Weber, and Bond numbers. To keep our analyses simple, the
gravitational effects are neglected in the rst two series of
simulations and are considered only in the last series. In addition,
in all the simulations, three cases with different liquid bridge
volumes and contact angles are studied: (1) V0.1 and y 301;
(2) V0.2 and y 201; (3) V 0.3 and y 101. Also, for comparison
purposes, the rupture distances for these cases are obtained
with the simplied model and are 0.65, 0.76, and 0.78,
respectively.
Fig. 11 shows the variations of the rupture distance with
respect to the Weber number. In all these simulations, the
capillary number is xed at Ca10  3. According to this gure,
the rupture distance increases as the Weber number increases.
Also, as the Weber number decreases, the rupture distance
approaches an asymptotic value. The asymptotic values are 0.64,

0.83, and 0.90 for cases 13, respectively, and correspond to the
rupture distance for Weber numbers equal to or less than 10  4.
Comparing these values with the predictions of the simplied
model, it is clear that the agreement between the numerical
simulations and the simplied model deteriorates when the liquid
bridge volume is increased. Nevertheless, the asymptotic values
are fairly close to the predictions of the simplied model. The
physical interpretation of the above is that at such low Weber
numbers, the surface tension effects dominate the inertial effects
such that the results of the numerical simulations agree with those
of the simplied model. Further analysis shows that for Weber
numbers up to We10  2, the predicted rupture distances are
within the 5% range of the above-mentioned asymptotic values. In
conclusion, for Weber numbers less than 10  2, the surface tension
effects dominate the inertial effects; this value (We10  2) is one
of the limiting values that can be used for quantitatively stating
the aforesaid assumption of the simplied model.
Fig. 12 shows the variations of the rupture distance with
respect to the capillary number. In all these simulations, the
Weber number is xed at We10  4. According to this gure, the
rupture distance increases as the capillary number increases. Also,
as the capillary number decreases, the rupture distance
approaches an asymptotic value. The asymptotic values are 0.62,
0.80, and 0.87 for cases 13, respectively, and correspond to the
rupture distances for capillary numbers equal to or less than
10  4. Comparing these values with the predictions of the
simplied model, it is observed that the agreement between the

ARTICLE IN PRESS
P. Darabi et al. / Chemical Engineering Science 65 (2010) 44724483

4479

Fig. 12. Variations of simulated rupture distance with respect to capillary number.

Fig. 13. Variation of simulated rupture distance with respect to Bond number.

numerical simulations and the simplied model deteriorates as

the liquid bridge volume is increased. Nevertheless, the
asymptotic values are fairly close to the predictions of the
simplied model. The physical interpretation of the above is
that at such low capillary numbers, the surface tension effects
dominate the viscous effects such that the results of the numerical
simulation agree with those of the simplied model. Further
analysis shows that for capillary numbers up to Ca 10  3, the
predicted rupture distances are within the 5% range of the abovementioned asymptotic values. In conclusion, for capillary
numbers less than 10  3, the surface tension effects dominate
the viscous effects; this value (Ca10  3) is one of the limiting
values that can be used for quantitatively stating the aforesaid
assumption of the simplied model.
Fig. 13 shows the variations of the rupture distance with
respect to the Bond number. In all these simulations, the capillary
and Weber numbers are xed at Ca10  3and We10  4.
According to this gure, the rupture distance decreases as the
Bond number increases. Also, as the Bond number decreases, the
rupture distance approaches an asymptotic value. These
asymptotic values are 0.64, 0.83, and 0.9 for cases 13,
respectively, and correspond to the rupture distances for Bond
numbers equal to or less than 10  1. We stress that comparing
these values with the predictions of the simplied model, the
agreement between the numerical simulations and the simplied
model deteriorates as the liquid bridge volume is increased.
Nevertheless, the asymptotic values are fairly close to the
predictions of the simplied model. The physical interpretation
of the above is that at such Bond numbers, the surface tension
effects dominate the gravitational effects so that the results of the
numerical simulation agree with those of the simplied model.
Further analysis shows that for Bond numbers up to Bo1, the
predicted rupture distances are within the 5% range of the abovementioned asymptotic values. In conclusion, for Bond numbers
less than 1, the surface tension effects dominate the gravitational
effects; this value (Bo1) is one of the limiting values that can be
used for quantitatively stating the aforesaid assumptions of the
simplied model.

between the two particles after rupture. Then, we determine the

liquid transfer fraction (TF) for each particle by calculating the
ratio of the liquid volume on that particle over the liquid bridge
volume.
Darhuber et al. (2001) simulated the liquid transfer fraction
between two plates at quasi-static conditions. They found that the
quasi-static condition corresponds to low Reynolds and very low
capillary numbers. The numerical simulations of Darhuber et al.
were performed with the Surface Evolver software (Brakke, 1992).
The software evolves the surface towards the minimal energy by a
gradient descent method, and is used for studying surfaces shaped
by energies such as surface tension, gravitational energy, etc. Note
that for our simulations, the full NavierStokes equations of the
two-uid interactions are solved with the VOF method, meaning
that our numerical tool and solution procedure differs from that
of Darhuber et al. In our simulations, particles have the same
sizes, and the contact angle on one particle is xed at yi 301
while that of the other particle varies in a range of yj 5601. Also,
since the liquid bridge volume was not exactly specied by
Darhuber et al., two different liquid bridge volumes are used for
our simulations, i.e., V 0.03 and 0.1. To comply with the quasistatic assumption of Darhuber et al., the capillary and Weber
numbers in our simulations are in the order of 10  3 and 10  4,
respectively. Note that in our numerical simulations the gravitational effects are neglected.
Fig. 14 graphically compares the liquid transfer fraction
predictions of our numerical simulations with those of the
numerical simulations of Darhuber et al. (2001). This gure
indicates good agreement for the simulations with a liquid bridge
volume of 0.03, and fair agreement for the simulations with the
liquid bridge volume of 0.1. The better agreement found for the
lower bridge volume corresponds with the assumption of
Darhuber et al. of a very low amount of liquid in the bridge.
According to this gure, the liquid transfer fraction is very
sensitive to the contact angles of the two surfaces. Liquid is evenly
distributed when the two surfaces have identical contact angles,
and it is no longer evenly distributed when the contact angle of
one surface differs from the other. This difference increases as the
contact angle difference increases and more liquid is transferred
to the surface with the smaller contact angle.

6.2.4. Comparison between liquid distribution predictions of our

numerical simulations and another numerical tool
The complementary phenomenon following rupture is liquid
distribution. For quantifying the matter, we assume that mass is
conserved and that the entire liquid bridge volume is partitioned

6.2.5. Effect of gravity on liquid distribution

In the previous section, the effect of the contact angle on the
liquid distribution for equal-sized particles in the absence of

ARTICLE IN PRESS
4480

P. Darabi et al. / Chemical Engineering Science 65 (2010) 44724483

Fig. 16. Variations of liquid transfer fraction with respect to liquid bridge volume.
Fig. 14. Comparison between liquid transfer fraction predictions of our numerical
simulations and the numerical simulations of Darhuber et al. (2001).

6.3. Comparison between bridge proles predicted with simplied

model and with numerical simulation

Fig. 15. Variations of liquid transfer fraction with respect to Bond number.

gravity was studied. As shown in Fig. 14, a slight difference

(even 51) between the contact angles of the two surfaces (at
plates or spherical particles) can have a drastic impact on how
liquid is partitioned between solid surfaces according to this
gure, more liquid is transferred to the surface with the lower
contact angle.
Another important factor affecting the liquid transfer fraction
is the importance of the gravitational effects compared to the
surface tension effects. Fig. 15 shows the variations of the liquid
transfer fraction with respect to the Bond number. According to
this gure, the amount of liquid transferred to the particle in the
presence of gravitational effects increases as the Bond number
increases. Fig. 16 shows the variations of the liquid transfer
fraction with respect to the dimensionless liquid bridge volume.
According to this gure, more liquid is transferred to the righthand side particle, due to the gravitational effects. The liquid
transfer fraction increases as the liquid bridge volume increases.

One of the key assumptions of the simplied model is that the

liquid bridge shape in the vicinity of the rupture instant can be
approximated with a second-order polynomial. In addition, it is
assumed that at any instant, the liquid bridge is stable, i.e., small
perturbations do not grow. In this section, these two assumptions
are analyzed in further detail.
In Fig. 17, the bridge proles predicted using the simplied
model (presented in green squares) and the numerical simulation
at the rupture instant and a few milliseconds before rupture are
shown. According to this gure, at the rupture instant, the
simplied model does not represent the simulated bridge shape
appropriately; however, a few milliseconds before rupture (here,
9 ms), the liquid bridge prole predicted using the simplied
model resembles the simulated bridge prole. In order to further
analyze the matter, the contour plots of the velocity magnitude of
the numerical simulations are compared, as shown in Fig. 18.
According to this gure, at the rupture instant, the bridge is
evolving with a very high velocity, as the maximum velocity is
orders of magnitude greater than the prescribed separation
velocity; however, a few milliseconds before rupture, the
maximum velocity magnitude (observed at the liquid-solid
interface) is comparable to the prescribed separation velocity.
Comparing these two velocity contour plots, it can be concluded
that at the rupture instant, the liquid bridge prole is unstable,
whereas a few milliseconds before rupture, a stable liquid prole
is observed.
We further analyzed the results of the numerical simulations
and tted the simulated proles with second-, fourth-, and sixthorder polynomials. The condence levels of the curve ttings for
the bridge prole at the rupture instant were 91%, 99%, and 99.9%
for the second-, fourth-, and sixth-order polynomials, respectively. However, for the bridge prole a few milliseconds before
rupture, these condence levels, in the same order, were 98%,
99.95%, and  100%. Therefore, it is concluded that a second-order
polynomial is not an appropriate choice for representing the
bridge shape at the rupture instant, while a second-order
polynomial can accurately predict the bridge shape a few
milliseconds before rupture.
In the above gures, the liquidsolid contact angle on both
particles was y 301. The impact of contact angle on the liquid

ARTICLE IN PRESS
P. Darabi et al. / Chemical Engineering Science 65 (2010) 44724483

4481

Fig. 17. Comparison between the bridge proles predicted using the simplied model and numerical simulations at: (a) rupture instant and (b) a few milliseconds before
rupture, for y 301.

Fig. 18. Contour plots of the velocity magnitude (in m/s) of the numerical simulations at: (a) rupture instant and (b) a few milliseconds before rupture, for y 301.

Fig. 19. Comparison of the bridge proles predicted using the simplied model and numerical simulation at: (a) rupture instant and (b) a few milliseconds before rupture,
with yi 301 and yj 401.

bridge shape was also studied for contact angles of 101, 201, and
401 on both particles, and similar trends were found. In
conclusion, the second-order polynomial, as used in the simplied
model, is an appropriate representation of the bridge shape only
at stable conditions, for particles with identical contact angles.
The applicability of the parabolic prole was also investigated
for particles with different contact angles. Fig. 19 shows the

comparison between the bridge proles predicted using

the simplied model (indicated in green squares) and the
numerical simulation, for particles with yi 301 and yj 401.
According to this gure, even a few milliseconds before rupture,
the bridge prole no longer resembles a second-order polynomial.
It is concluded that a second-order polynomial is not an
appropriate representation of the simulated liquid bridge prole

ARTICLE IN PRESS
4482

P. Darabi et al. / Chemical Engineering Science 65 (2010) 44724483

for particles with different contact angles; therefore, the

simplied model is not an appropriate choice for predicting the
bridge evolution between particles with different contact angles.

7. Limitations of simplied model

The liquid bridge shapes predicted using the numerical
simulation and the simplied model were compared. It was
concluded that a second-order polynomial is not an appropriate
representation of the simulated bridge shape for two conditions:
(1) unstable liquid proles and (2) liquid bridges between
particles with different contact angles. Therefore, for such
conditions another approach, such as numerical simulation,
should be used.
In addition, with the aid of the numerical simulations, the
importance of the inertial, viscous, and gravitational effects in
comparison to the surface tension effects was studied. It was
concluded that the surface tension effects dominate the viscous,
inertial, and gravitational effects for Ca o10  3, Weo10  2, and
Boo1, respectively. These limits represent the range of the
dimensionless numbers within which the simplied model is
applicable; if any of the above-mentioned dimensionless numbers
exceeds its limiting value, an alternative approach such as
numerical simulation should be employed. To sum up, the
simplied model assumes that the surface tensions effects are
dominant and does not account for any other forces.

8. Summary and conclusions

In this work, a simplied mathematical model and the
numerical simulations of the governing NavierStokes equations
were used to study the evolution of stretching liquid bridges
between two smooth equal-sized spherical particles. The
simplied model presumes a parabolic shape for the gasliquid
interface of the bridge; it also assumes that the bridge evolution is
a quasi-static phenomenon with negligible inertial, viscous, and
gravitational effects. For the numerical simulations, a commercial
Computational Fluid Dynamics (CFD) software package, FLUENT,
was used to solve the problems governing NavierStokes
equations. Since the numerical simulations were based on the
solution to the full NavierStokes equations with interface
boundary conditions, viscous, inertial, gravitational, and surface
tension effects were all considered.
Both models were used to predict the bridge shape, rupture
distance, and liquid distribution of stretching liquid bridges.
Primarily, their rupture distance predictions were compared with
the experimental data available in the literature. Both models
showed good agreement. However, the numerical simulations
provided better results than the simplied model. The simplied
model was used to investigate the impact of liquid bridge volume
and contact angle on rupture distance. It was shown that the
predictions of the simplied model were in good agreement with
the related expressions available in the literature.
The numerical simulations were initially used to predict the
bridge deformation and rupture in time. It was shown that as
particles move farther apart, the liquid bridge stretches and its
neck becomes narrower. This continuous elongation persists until
the bridge can no longer withstand any additional elongation and
ruptures. It was also illustrated that for simulations with
non-negligible gravitational effects, more liquid accumulates on
the particle due to these effects. Several numerical simulations
were used to investigate the effect of the capillary, Weber, and
Bond numbers on the rupture distance. The results showed that
for simulations with Cao10  3, Weo10  2, and Boo1, the

rupture distance is within 5% of an asymptotic value. The

asymptotic values were close to the rupture distances predicted
using the simplied model. Moreover, the predictions of our
numerical simulations were compared with the results of another
numerical tool, the Surface Evolver software package (Brakke,
1992), and good agreement was obtained. In addition, with the
aid of the numerical simulations, the effects of contact angle and
gravity were investigated on liquid distribution. It was shown that
more liquid is transferred to the particle with the smaller contact
angle. The liquid transfer fraction increased as the Bond number
or liquid bridge volume increased. Note that such effects could
not be explored with the simplied model, as the gravitational
effects are neglected in it.
The bridge proles predicted using the numerical simulations
and simplied model were compared. It was shown that under
stable circumstances, the bridge prole was appropriately
represented with a second-order polynomial; therefore, the
simplied model was appropriate for modeling purposes for
particles with the same contact angles. Furthermore, we showed
that for particles with different contact angles, the parabolic
assumption was no longer a good representation of the bridge
prole even in stable conditions.
In conclusion, the simplied model is a useful tool for studying
the evolution of stretching liquid bridges, yet it has limitations.
It is only applicable to bridges with dominant surface tension
effects and under quasi-static conditions. In particular, the
simplied model provides reasonable predictions for problems
with Cao10  3, Weo10  2, and Boo1. Although the parabolic
assumption used in the simplied model is applicable to particles
with the same contact angles, for particles with different contact
angles, it is inaccurate, thus the simplied model should not be
used. We can conclude from the above that the numerical
simulations of the governing NavierStokes equations are indeed
more generally applicable. They can be used for problems with
non-negligible viscous, inertial, and gravitational effects, without
having to prescribe a certain prole for the liquid bridge. Finally,
the numerical simulations are complementary to the simplied
model; the simplied model should also be used, while considering its limitations. Otherwise, the numerical simulations of the
governing NavierStokes equations have to be used. The results
obtained with these two models can be used to predict the
rupture distance and liquid distribution of dynamic liquid bridges.
Consequently, the results can be incorporated in DEM simulations
of particulate systems involving liquid, such as wet granulators
and rotary drum coaters.

Notation
Bo
Ca
d
g
h
R
Sn
Sc
T
TF
u
v
V
Vb
Vcap

Bond number ( rgR2/s), dimensionless

capillary number ( mv/s), dimensionless
liquid bridge length, dimensionless
gravitational acceleration, m/s2
spherical cap height, dimensionless
particle radius
dimensional rupture distance, m
dimensionless rupture distance (Sc Sn/R),
dimensionless
droplet maximum cap height, dimensionless
liquid transfer fraction, dimensionless
droplet radius, dimensionless
relative separation velocity, m/s
dimensionless liquid bridge volume ( Vb/R3),
dimensionless
dimensional liquid bridge volume, m3
spherical cap volume, dimensionless

ARTICLE IN PRESS
P. Darabi et al. / Chemical Engineering Science 65 (2010) 44724483

We
x
xmin
y

Weber number ( rv2R/s), dimensionless

x-coordinate (abscissa)
liquid bridge x-coordinate at its thinnest point
y-coordinate (ordinate)

Greek letters

m
r
s
y

liquid viscosity, Pa s
liquid density, kg/m3
surface tension, N/m
contact angle, radian or degree

Acknowledgments
The nancial support of the Natural Sciences and Engineering
Research Council (NSERC) of Canada and Syncrude Canada Ltd. are
gratefully acknowledged. The authors also acknowledge the
Canada Foundation for Innovation (CFI) for funding the computational resources. The authors are grateful to Ehsan Azadi Yazdi
and Ali Vakil for the useful discussions related to this paper.
References
Brackbill, J.U., Kothe, D.B., Zemach, C., 1992. A continuum method for modeling
surface tension. Journal of Computational Physics 100, 335354.
Brakke, K.A., 1992. The surface evolver. Experimental Mathematics 1, 141165.
Chan, D.Y.C., Horn, R.G., 1985. The drainage of thin liquid lms between solid
surfaces. Journal of Chemical Physics 83, 5311.
Coriell, S.R., Hardy, S.C., Cordes, M.R., 1977. Stability of liquid zones. Journal of
Colloid and Interface Science 60, 126136.
Dai, Z., Lu, S., 1998. Liquid bridge rupture distance criterion between spheres.
International Journal of Mineral Processing 53, 171181.
Darabi, P., Pougatch, K., Salcudean, M., Grecov, D., 2009. A novel coalescence model
for binary collision of identical wet particles. Chemical Engineering Science 64,
18681876.
Darhuber, A.A., Troian, S.M., Wagner, S., 2001. Physical mechanisms governing
pattern delity in microscale offset printing. Journal of Applied Physics 90,
36023609.
De Bisschop, F.R.E., Rigole, W.J.L., 1982. A physical model for liquid capillary
bridges between adsorptive solid spheres: the nodoid of plateau. Journal of
Colloid and Interface Science 88, 117128.
Eggers, J., Dupont, T.F., 1994. Drop formation in a one-dimensional approximation
of the NavierStokes equation. Journal of Fluid Mechanics 262, 205221.
Ennis, B.J., Li, J., Tardos, G.I., Pfeffer, R., 1990. The inuence of viscosity on the
strength of an axially strained pendular liquid bridge. Chemical Engineering
Science 45, 30713088.
Ennis, B.J., Tardos, G., Pfeffer, R., 1991. A microlevel-based characterization of
granulation phenomena. Powder Technology 65, 257272.
Erle, M.A., Dyson, D.C., Morrow, N.R., 1971. Liquid bridges between cylinders, in a
torus, and between spheres. AICHE Journal 17, 115121.
Fisher, R.A., 1926. On the capillary forces in an ideal soil; correction of formulae
given by WB Haines. Journal of Agricultural Science 16, 492505.
FLUENT Inc., 2006. Fluent 6.3 Users Guide. Lebanon, New Hampshire.
Fowle, A.A., Wang, C.A., Strong, P.F., 1979. Experiments on the stability of conical
and cylindrical liquid columns at low bond numbers. In: Proceedings of the
Third European Symposium on Material Science in Space, pp. 317325.

4483

Gillette, R.D., Dyson, D.C., 1971. Stability of uid interfaces of revolution between
equal solid circular plates. Chemical Engineering Journal 2, 4454.
Glatzel, T., Litterst, C., Cupelli, C., Lindemann, T., Moosmann, C., Niekrawietz, R., Streule,
W., Zengerle, R., Koltay, P., 2008. Computational Fluid Dynamics (CFD) software
tools for microuidic applicationsa case study. Computers & Fluids 37, 218235.
Haines, W.B., 1925. A note on the physical properties of soils. Journal of
Agricultural Science 17, 264290.
Haynes, J.M., 1970. Stability of a uid cylinder. Journal of Colloid and Interface
Science 32, 652654.
Issa, R.I., Ahmadi-Befrui, B., Beshay, K.R., Gosman, A.D., 1991. Solution of the
implicitly discretised reacting ow equations by operator-splitting. Journal of
Computational Physics 93, 388410.
Lian, G., Thornton, C., Adams, M.J., 1993. A theoretical study of the liquid bridge
forces between two rigid spherical bodies. Journal of Colloid and Interface
Science 161, 138147.
Mason, G., Clark, W.C., 1965. Liquid bridges between spheres. Chemical Engineering Science 20, 859866.
Mazzone, D.N., Tardos, G.I., Pfeffer, R., 1986. The effect of gravity on the shape and
strength of a liquid bridge between two spheres. Journal of Colloid and
Interface Science 113, 544556.
Mazzone, D.N., Tardos, G.I., Pfeffer, R., 1987. The behavior of liquid bridges
between two relatively moving particles. Powder Technology 51, 7183.
Mehrotra, V.P., Sastry, K.V.S., 1980. Pendular bond strength between unequal-sized
spherical particles. Powder Technology 25, 203214.
Mikami, T., Kamiya, H., Horio, M., 1998. Numerical simulation of cohesive powder
behavior in a uidized bed. Chemical Engineering Science 53, 19271940.
Mollot, D.J., Tsamopoulos, J., Chen, T.Y., Ashgriz, N., 1993. Nonlinear dynamics of
capillary bridges: experiments. Journal of Fluid Mechanics 255, 411435.
Orr, F.M., Scriven, L.E., Rivas, A.P., 1975. Pendular rings between solids: meniscus
properties and capillary force. Journal of Fluid Mechanics 67, 723742.
Ozkan, F., Worner, M., Wenka, A., Soyhan, H.S., 2007. Critical evaluation of CFD
codes for interfacial simulation of bubble-train ow in a narrow channel.
International Journal for Numerical Methods in Fluids 55, 537564.
Papageorgiou, D.T., 1995. On the breakup of viscous liquid threads. Physics of
Fluids 7, 15291544.
Pepin, X., Rossetti, D., Iveson, S.M., Simons, S.J.R., 2000a. Modeling the evolution
and rupture of pendular liquid bridges in the presence of large wetting
hysteresis. Journal of Colloid and Interface Science 232, 289297.
Pepin, X., Rossetti, D., Simons, S.J.R., 2000b. Modeling pendular liquid bridges with
a reducing solidliquid interface. Journal of Colloid and Interface Science 232,
298302.
Pitois, O., Moucheront, P., Chateau, X., 2000. Liquid bridge between two moving
spheres: an experimental study of viscosity effects. Journal of Colloid and
Interface Science 231, 2631.
Pitois, O., Moucheront, P., Chateau, X., 2001. Rupture energy of a pendular liquid
bridge. The European Physical Journal BCondensed Matter 23, 7986.
Rossetti, D., Simons, S.J.R., 2003. A microscale investigation of liquid bridges in the
spherical agglomeration process. Powder Technology 130, 4955.
Rossetti, D., Pepin, X., Simons, S.J.R., 2003. Rupture energy and wetting behavior of
pendular liquid bridges in relation to the spherical agglomeration process.
Journal of Colloid and Interface Science 261, 161169.
Sanz, A., 1985. The inuence of the outer bath in the dynamics of axisymmetric
liquid bridges. Journal of Fluid Mechanics 156, 101140.
Shi, D., McCarthy, J.J., 2008. Numerical simulation of liquid transfer between
particles. Powder Technology 184, 6475.
Simons, S.J.R., Seville, J.P.K., Adams, M.J., 1994. An analysis of the rupture energy of
pendular liquid bridges. Chemical Engineering Science 49, 23312339.
Soulie, F., Cherblanc, F., El Youssou, M.S., Saix, C., 2006. Inuence of liquid bridges
on the mechanical behaviour of polydisperse granular materials. International
Journal for Numerical and Analytical Methods in Geomechanics 30,
213228.
Tsamopoulos, J., Chen, T.Y., Borkar, A., 1992. Viscous oscillations of capillary
bridges. Journal of Fluid Mechanics 235, 579609.
Zhang, X., Padgett, R.S., Basaran, O.A., 1996. Nonlinear deformation and breakup of
stretching liquid bridges. Journal of Fluid Mechanics 329, 207245.
Zhang, Y., Alexander, J.I.D., 1990. Sensitivity of liquid bridges subject to axial
residual acceleration. Physics of Fluids A: Fluid Dynamics 2, 19661974.