Sunteți pe pagina 1din 19

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/305694833

Lattice-boltzmann analysis of capillary rise

Article  in  Journal of Porous Media · January 2016


DOI: 10.1615/JPorMedia.v19.i5.60

CITATIONS READS
5 1,123

3 authors:

Mohamed El amine Ben Amara Patrick Perre


National Engineering School of Monastir CentraleSupélec
7 PUBLICATIONS   33 CITATIONS    301 PUBLICATIONS   5,239 CITATIONS   

SEE PROFILE SEE PROFILE

Sassi Ben Nasrallah


National Engineering School of Monastir
509 PUBLICATIONS   6,684 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

viscous dissipation effects on heat transfer in composite system Fluid/porous View project

Wind Energy Research View project

All content following this page was uploaded by Sassi Ben Nasrallah on 29 August 2016.

The user has requested enhancement of the downloaded file.


begell
Begell House, Inc. Publishers

Journal Production
50 North Street
Danbury, CT 06810
Phone: 1-203-456-6161
Fax: 1-203-456-6167
Begell House Production Contact : journals@begellhouse.com

Dear Corresponding Author,

Attached is the corresponding author pdf file of your article that has been published.

Please note that the pdf file provided is for your own personal use and is not to be posted on
any websites or distributed in any manner (electronic or print). Please follow all guidelines
provided in the copyright agreement that was signed and included with your original
manuscript files.

Any questions or concerns pertaining to this matter should be addressed to


journals@begellhouse.com

Thank you for your contribution to our journal and we look forward to working with you again
in the future.

Sincerely,
Michelle Amoroso
Michelle Amoroso
Production Department
Journal of Porous Media, 19 (5): 453–469 (2016)

LATTICE-BOLTZMANN ANALYSIS OF CAPILLARY


RISE

Mohamed El Amine Ben Amara,1,∗ Patrick Perré,2 & Sassi Ben Nasrallah1

1
Laboratoire d’Études des Systèmes Thermiques et Énergétiques, École Nationale d’Ingénieurs de
Monastir, Monastir 5019, Tunisia
2
Laboratoire de Génie des Procédés et des Matériaux, École Centrale Paris, CentraleSupélec,
campus de Châtenay-Malabry Grande Voie des Vignes F-92 295 Châtenay-Malabry Cedex,
Paris, France


Address all correspondence to: Mohamed El Amine Ben Amara, E-mail: ben.amara.amin@gmail.com

Original Manuscript Submitted: 3/25/2015; Final Draft Received: 10/22/2015

The current study presents an investigation of capillary flow by using the two-phase lattice-Boltzmann method. The
Shan-Chen single-component multiphase model was applied to simulate capillary rise in straight and sinusoidal cap-
illaries. In order to validate our code, three test cases are considered: (a) The Laplace law is tested for various droplets,
(b) the contact angle was verified by comparing the ratio of droplet wet length to droplet height at various adhesion
parameters and (c) the verification of the Washburn equation for long times. The density distribution was presented
for different geometry configurations and liquid front position was plotted as a function of time. The numerical results
showed that the Washburn equation is not valid for short times. The internal structure of the flow inside the capillaries
was described, thus the analysis shows the forming of recirculation zones near the inlet region and a decelerating effect
of the varying path on the meniscus movement. For sinusoidal capillaries, we notice that the meniscus reaches different
equilibrium positions.

KEY WORDS: capillary rise, sinusoidal capillary, lattice-Boltzmann method, Shan and Chen model

1. INTRODUCTION

Many applications, such as water transport in a gas diffusion layer of fuel cell, oil recovery, microfluidics for chemical
analysis, or biological assay and flow in porous media involve wicking in one or multiple capillaries of complex
geometry. To a large degree, understanding and controlling the wicking are useful to improve the efficiency of the
related device. The wicking is based on the well-known phenomenon of capillary rise, which happens when we put a
capillary of radius R into contact with a bath of a wetting liquid topped with a gas, which causes the liquid rising in
the tube until an equilibrium height heq due to the pressure drop over the curved interface between the two immiscible
fluids. Neglecting the effects of gravity on the interface, we suppose that the curvature radius r is constant and equal
to R/cos θ, where θ is the contact angle between the interface and the solid wall. It depends on the nature of the
fluid and the wall. The Laplace equation leads to pc = 2σ cos θ/R, which is the capillary pressure. At the equilibrium
height heq , the capillary pressure is balanced by the hydrostatic pressure Pc = ∆ρgheq , where g is the acceleration
of gravity and ∆ρ is the density difference between both air and liquid density. By equalizing the two last relations
we can calculate the equilibrium height heq = 2σ cos θ/R∆ρg. The previous relationship is known as the Jurin’s
law (Quéré, 1997). This phenomenon has been a subject of research and debate since Washburn (1921) described the
dynamics of capillary rise. Considering that the flow is confined and the speed involved is small, the balance between
capillary force and viscous force leads to the Washburn law,

1091–028X/16/$35.00 ⃝
c 2016 by Begell House, Inc. 453
454 Ben Amara, Perré, & Ben Nasrallah

NOMENCLATURE

ci particle speed (lu t s−1 ) V characteristic velocity (lu t s−1 )


cs speed of sound (lu t s−1 ) wi weighting factor
fi distribution function z meniscus height (m)
fieq equilibrium distribution function
G cohesion parameter Greek Symbols
Gads adhesion parameters Ω collision operator
g gravity (lu ts−2 – m s−2 ) ⃗ξ molecular velocity
heq balance height (m) ν kinematic viscosity (lu2 t s−1 )
i lattice streaming vector direction ∆x lattice spacing (lu)
Lcap capillary length (lu) ∆t time step (ts)
lu lu lattice unit ρ density (mu lu−3 – kg m−3 )
mu mass unit ψ potential function
P pressure (mu lu−1 t s−2 ) τ relaxation time (ts)
R droplet radius (lu) θ contact angle (rad)
R(z) capillary radius (lu – m) θd dynamic contact angle (rad)
Rmin minimum radius of the capillary (lu) λ undulation (lu)
Rmax maximum radius of the capillary (lu) µ viscosity (mu lu−1 t s−1 – kg m−1 s−1 )
r specific gas constant σ liquid surface tension (lu mu ts−2 – J m−2 )
T temperature (K)
ts time step Abbreviations
u macroscopic velocity (m s−1 ) Ca capillary number


z(t) = (Rσ cos θ/2µ)t (1)
where µ is the dynamic viscosity of the liquid.
This law is widely used and tested throughout the years to study the capillary flow and features porous media.
However, it is only valid when the weight of the liquid column is negligible; otherwise, the weight of the column
should be taken into consideration and the liquid asymptotically reaches equilibrium height following the Lucas-
Washburn law (Zhmud et al., 2000) given by the following formula:

z(dz/dt) = (R2 /8µ) [(2σ cos θ/R) − ρgz] (2)



At times very short Washburn law leads to an infinite velocity u(t) = Rσ cos θ/8µ t. To avoid this problem, we
apply (Ben Amara et al., 2011) Newton’s second law for the liquid column as follows:
d [ ]
(πρR2 z)dz/dt = −8µzdz/dt + 2πRσ cos θ − πρR2 gz (3)
dt
This equation highlights all forces acting on the liquid column, including the gravity force πρR2 gz, the capillary force
2πRσ cos θ, the viscosity force 8πµzdz/dt, and √ the inertia. The resolution of this equation requires two boundary
conditions. We should take z(0) = 0 and z ′ (0) = 2σ cos θ/ρR as the initial conditions; the second one is chosen to
avoid the singularity at t = 0. Therefore, the addition of inertia to the forces acting on the liquid has revealed the exis-
tence of an oscillatory regime around the equilibrium position of the meniscus; it was demonstrated by Quéré (1997)
and Zhmud et al. (2000) that the oscillations appear
√ over a limit radius Rlim given by Rlim = (8σµ2 cos θ/ρ3 g 2 )1/5
and over a period T written as follows: T = 6 2σ cos θ/ρR/g. The different aspects of this model were well studied
by Fries (Fries and Dreyer, 2008) including an analytical approach, a detailed dimensional analysis, and an intro-
duction to the different time stages. Szekely et al. (1971) developed an analysis of capillary rise considering both

Journal of Porous Media


Lattice-Boltzmann Analysis of Capillary Rise 455

inertial and entrance effects by adding a term to the inertia force. Due to the nonuniformity of the porous medium,
it is modeled as a bundle of capillary tube with variable cross sections; this model can correct the approximation of
the straight capillary model, even if it is a poor representation of the complex pore geometries characteristic of rocks.
Many authors studied the behavior of liquid in a nonuniform capillary. In the work of Samuel et al. (2003), the authors
gave a formulation of the balance height in a variable-section capillary, and Staples and Shaffer (2002) studied the
dynamics of the rise using the Poiseuille equation given by

∆P /z = 8µ(dz/dt)/R2 (4)

to find a generalized Lucas-Washburn equation, which is written for a variable-section capillary tube as
/ ∫z

dz 8µR2 (z) 1/R4 (ζ)dζ
= [(2σ cos θ/R(z)) − ρgz] (5)
dt
0

However, this model did not take into consideration the inertia and the variation of the curvature radius. Czachor
(2006) modified the previous equation, taking into account the variation in the curvature radius along a sinusoidal
capillary. The analysis of the capillary rise-or descent in tapered capillaries has been extended by Tsori (2006). Show-
ing the unexpected behavior of the liquid flow under specific conditions, he shows that apart from the contact angle,
it depends on the size sufficiency of the pipette; whether it is hydrophobic or hydrophilic, the meniscus will stay at
the bottom. No liquid will penetrate the pipette unless a negative pressure is established in the pipette. Ben Amara
and Ben Nasrallah (2011) provide numerical solutions for the dynamics of penetration of liquids into nonuniform
capillary tubes using a generalized equation taking into account inertia effects and the variation in the curvature ra-
dius. To summarize the previous works concerning the wicking, we have used a global model able to describe the
detailed flow structures in the capillaries and the detailed dynamic of meniscus. The flow in the capillaries is simply
considered fully developed Poiseuille flow and the interface shape is also considered spherical or cylindrical. There
are theoretical and numerical difficulties in computing the detailed wicking flow and the meniscus dynamics in com-
plex geometries. These include a lack of a good model for the moving contact line, the need to capture dynamically a
moving and deforming interface, morphological singularities in coalescence and rupture of the interface, and the com-
plex flow geometries in practically interesting problems. The lattice-Boltzmann method (LBM) is successfully tested
in the context of multiphase flow and in complex geometries. This method has the following advantages: flexibility
for complex boundary geometries, ease of parallel computing, and no tracking of interfaces is necessary. It enables
a detailed and local description of the multiphase flows. Raiskinmäki et al. (2002) studied capillary rise dynamics
in a circular tube using the lattice-Boltzmann method with the Shan-Chen multiphase model. They showed that the
Washburn equation was satisfied by the LBM simulations for large capillary tubes of typically at least 30 lattice units.
Also, they demonstrated the dependence of the cosine of the dynamic contact angle on the capillary number Ca. In
this work, a two-phase lattice-Boltzmann method has been used for the study of the capillary rise in variable capillary
cross sections. The Shan-Chen single-component multiphase model was used. The aim of this research is to show
the capability of the lattice-Boltzmann method to model the wicking and to describe the detailed flow structures and
the dynamic of meniscus. In what follows, we are going to outline the essential background of the lattice-Boltzmann
method and the Shan-Chen model. Next, we are going to present the simulation results showing the capability of the
proposed method in solving benchmark cases such as Laplace law. Finally, capillary rise simulations were carried out
in both straight and sinusoidal tubes and the dynamics of the flow has been studied.

2. THE LATTICE-BOLTZMANN METHOD


The lattice-Boltzmann method originated from the lattice gas automata (LGA) method which has been developed
to model fluid flow based on the statistical physics and gas kinetics. The fundamental equation of this development
is the Boltzmann transport equation, reflecting the fact that the macroscopic behavior of a fluid depends directly
on its microscopic behavior. The originality of this method compared to conventional methods is that it does not
consider the fluid from a macroscopic point of view but based on its mechanics at the microscopic level. Despite the

Volume 19, Issue 5, 2016


456 Ben Amara, Perré, & Ben Nasrallah

apparent complexity of simulating microscopic fluid scales, the lattice-Boltzmann method has many advantages. The
formulation based on molecule packets at a mesoscopic level associated with a simplification of the collision model
makes it easier to implement. This simplicity allows the resolution of complex flows such as two-phase flows. It has
also a simple treatment of boundary conditions; this facilitates the processing of problems in complex geometries such
as porous flow.

2.1 Boltzmann Equation


The Boltzmann equation, which gives information regarding the single-particle distribution function f , takes the
following form:
∂f
+ ⃗v · ∇f
⃗ = Ω(f ) (6)
∂t
where ⃗v is the speed of a particle at a position ⃗x and time t and Ω(f ) is the collision operator controlling the rate of
change in the distribution function f during the collision. The term ⃗v · ∇f
⃗ models the modification in the distribution
function due to the spread of the particles during their movement. The collision term Ω(f ) is usually represented
by a complex integral form, in most applications dealing with the kinetic theory of fluid dynamics; this term is
approximated by simple expressions. One of these is the so-called “BGK” approximation proposed by Bhatnagar,
Gross, and Krook, which was integrated by Qian et al. (1992). Consequently, the integral term is replaced by a simpler
one given by the following equation:
1 eq
Ω(f ) = [f (⃗x, ⃗v , t) − f (⃗x, ⃗v , t)] (7)
τ
This operator models the effect of the collision as a relaxation of the distribution function f to a Maxwellian steady
state f eq (⃗x, ⃗v , t), and the parameter τ represents a characteristic time that controls the frequency of relaxation of
the distribution function toward the equilibrium. The BGK approximation is sometimes called the single-relaxation
time model and it is the most popular in the scientific literature. Other forms of the collision operator are used as the
multiple relaxation time model, as seen in works of D’Humières et al. (2002).

2.2 Lattice BGK Equation


Generally, the velocity space is continuous, i.e., a particle can move freely at a random speed. A first step toward the
discretization of the Boltzmann equation is the restriction of the velocity from continuous space into a finite set of
velocities V⃗ := {⃗c0 , ⃗c1 , ..., ⃗cn−1 }; this means that a particle located at a point x must have one of the n speeds called
⃗ci . Thus the distribution function f (⃗x, ⃗v , t) is reduced to fi (⃗x, t), which describes the distribution along a finite lattice.
The Boltzmann equation under the BGK approximation becomes the discrete Boltzmann equation (8) given by

∂fi 1
+ ⃗ci · ∇fi = (fieq − fi ) (8)
∂t τ
This equation is discretized with a space step ∆x and a time step ∆t which are linked by the following relation
∆x/∆t = ci . This discretization ensures that the particles in a node x move while ∆t moves to a neighbor node
x + ci ∆t along the vector ⃗ci . If we take the time step ∆t = 1, the lattice Bhatnagar-Gross-Krook (LBGK) equation
becomes
1
fi (⃗x + ⃗ci , t + 1) − fi (⃗x, t) = [fieq (⃗x, t) − fi (⃗x, t)] (9)
τ
To further simplify the LBGK equation, it is divided into the following two stages:
•Collision—The arriving particles at the points interact with one another and change their velocity directions; thus
at time t the particles at node x come into collision with each other, which changes the distribution function from
fi (⃗x, t) to
1
fi∗ (⃗x, t) = fi (⃗x, t) + [fieq (⃗x, t) − fi (⃗x, t)] (10)
τ

Journal of Porous Media


Lattice-Boltzmann Analysis of Capillary Rise 457

•Streaming—Particles move during the time step ∆t, along lattice bonds to the neighboring lattice nodes and the
distribution function fi∗ (⃗x, t) spreads along the vector ⃗ci , more formally,

fi (⃗x + ⃗ci , t + 1) = fi∗ (x, t) (11)

Finally, in order to obtain the macroscopic variables ρ and ⃗u, we use some linear combination of the distribution
function, thus ∑
ρ(x) = fi (12)
i

and ∑
ρ(x)⃗u(x) = ci fi . (13)
i

2.3 Lattice Pattern


In the lattice-Boltzmann method many types of lattice are used, and they all have the same naming scheme Dn Qm ,
where n denotes the dimension of the lattice model and m is the number of the discrete velocities. For example, we
have D1 Q3 and D1 Q5 for a one-dimensional configuration, for a two-dimensional system we find the D2 Q5 , D2 Q7 ,
or D2 Q9 , and in a three- dimensional configuration we have D3 Q15 , D3 Q19 , and D3 Q27 (Wolf Gladrow, 2000).
In two-dimensional problems, the most popular scheme is the D2 Q9 model as shown in Fig. 1. It consists of √ a null
vector for particles at rest, four vectors of unit length in the Cartesian directions, and four vectors of length 2 in
the diagonal directions. We attribute to each direction a weighting factor while respecting the isotropy of the system,
so we set Wi the weighting factor of the speed ci , and we also presume that these
∑ factors are the same for directions
having the same velocity. These variables satisfy the following relationship: i Wi = 1. For D2 Q9 , the weighting
factors are defined by 
 4/9 i=0
Wi = 1/9 i = 1, 3, 5, 7 . (14)

1/36 i = 2, 4, 6, 8

2.4 Equilibrium Distribution Function


The equilibrium distribution function fieq depends on the density and the local fluid velocity and can be written under
the following implicit form:
fieq = fieq [ρ(⃗x, t), ⃗u(⃗x, t)] (15)

FIG. 1: D2 Q9 model

Volume 19, Issue 5, 2016


458 Ben Amara, Perré, & Ben Nasrallah

A general form of the equilibrium distribution function is obtained by a Taylor expansion of the Maxwell-Boltzmann
equilibrium function [ ]
n |⃗ξ − ⃗u |
F (⃗ξ) = exp − (16)
(2πrT )3/2 2rT

with r the specific gas constant, T the static temperature, n the particle density, ⃗ξ the molecular velocity components,
and ⃗u the macroscopic velocity vector. Thus fieq can be written as (Chen and Doolen, 1998; Qian et al., 1992).
[ ]
ci · ⃗u (ci · ⃗u)2 ⃗u 2
fieq = ρWi 1 + 2 + − (17)
cs 2c4s 2c2s

where cs is the isothermal speed of sound obtained by the Chapman-Enskog expansion (Chen et al., 1992; Frisch et
al., 1987; He and Luo, 1997) of the Boltzmann equation near the incompressible limit (|⃗u|/cs ≪ 1). This expansion
makes the relation between the Navier-Stokes equation and the Boltzmann equation. Such development gives us the
relation between the sound speed and the collision frequency through this formula:
1 2
ν= c (2τ − 1) (18)
2 s
The positivity of the kinematic viscosity requires that τ > 1/2. The speed of sound is
1
cs = √ c (in the D2 Q9 model) (19)
3
and the pressure is given by
P = c2s ρ. (20)

2.5 Solid Boundary Condition


In the lattice-Boltzmann method, the solid boundary condition is called “the bounce-back boundary conditions” (Chen
and Doolen, 1998). It is considered as the most attractive advantage of the LBM. The basic idea behind the bounce-
back rule is that each particle hitting the wall in the streaming step reverses its direction; this condition ensures the
nonslip boundary condition at the walls, as shown in Fig. 2. In terms of distribution functions and referring to Fig. 1,

FIG. 2: Bounce-back at the wall

Journal of Porous Media


Lattice-Boltzmann Analysis of Capillary Rise 459

we can write for the bottom wall, since f6 , f7 , and f8 are known from the streaming process, then f2 = f6 , f3 = f7
and f4 = f8 ; for the left side f1 = f5 , f2 = f6 and f8 = f4 ; and for the right side f5 = f1 ; f6 = f2 and f4 = f8 .
To simulate a curved boundary, many methods are used, as clearly treated by Renwei et al. (1999, 2000), in which
they solved the Boltzmann equation in curvilinear coordinates. In this work we use the simple halfway bounce-back
boundary conditions in which we define the boundary node as a point halfway in an exterior and interior lattice site,
as shown in Fig. 3. To increase the accuracy, the use of a very fine resolution is a necessary advisable, but then this
leads to an increase in the simulation time.

2.6 Lattice-Boltzmann Model for Multiphase Flow


A multiphase system involves many “forms” of a given fluid, for example, the liquid and vapor phases of water.
Such a system is ruled by a nonideal equation of state which is completely defined in terms of pressure, volume, and
temperature. For example, for the ideal gas the thermodynamic equation is P V = nRT , where P is the pressure, V
is the volume, R is the gas constant, T is temperature of the gas, and n is the number of moles of the gas. The Van
der Waals equation of state provides a better description of gases at higher pressure than does the ideal gas law. It
introduces two empirical parameters to correct for the interaction between gas atoms or molecules and for the volume
occupied by the gas molecules, which are mentioned in the Van der Waals equation of state as the following:

(P + a/V 2 )(V − b) = RT (21)

The term a/V 2 is a correction for the mutual attraction of the molecules, and b is a correction for the actual
volume of the molecules themselves. The attractive forces between molecules are known as Van der Waal’s forces.
Many lattice-Boltzmann simulation models of multiphase flows have been presented in several publications. We can
cite the Shan and Chen model (Shan and Chen, 1993), the free-energy model developed by Swift et al. (1996), the
mean-field theory model proposed by He et al. (1998), and the color-fluid model developed by Gunstensen et al.
(1991). The model proposed by Shan and Chen imposes a no local interaction between the fluid particles.The central
point of this model is the definition of a so-called particle-particle pseudopotential V , which has the following formula:

V (x, x′ ) = −Gαᾱ (x, x′ )ψα ψᾱ (22)

with x′ = x + ci the next neighbor of cell x on the grid. Gαᾱ is the Green’s function that controls the interaction
strength between phases or components α and ᾱ. The Green’s function Gαᾱ (x, x′ ) is defined as

FIG. 3: Solid–fluid interfaces located halfway between adjacent lattice nodes

Volume 19, Issue 5, 2016


460 Ben Amara, Perré, & Ben Nasrallah


 G, √ if |x − x′ | = 1√
Gαᾱ (x, x′ ) = G/ 3, if |x − x′ | = 3 (23)

0, otherwise
where G is the fluid-fluid interaction coefficient, usually chosen as a constant. ψα ψᾱ are the effective masses of the
phase or component α and ᾱ. ψ is a function of the density and can take many different forms, e.g.,
( )
ψ(ρ) = ψ0 1 − e(−ρ0 /ρ) (24)

where ψ0 and ρ0 are constants. The interparticle force for a single-component multiphase flow is written as

F α (x, t) = −Gψα (x, t) Wi ψα (x + ci ∆t, t)ci (25)
i

The adhesive force between the fluid at site x and the solid wall at site x′ is formulated in Martys and Chen’s work
(1996) as ∑
F αs (x, t) = −Gads ψα (x, t) Wi s(x + ci ∆t, t)ci (26)
i

where Wi is a weighting factor that depends on the particular lattice-Boltzmann model chosen. s is 1 for a solid and
0 for a fluid. The interaction strength between the fluid and the wall can be adjusted by the parameters Gads . Varying
Gads allows determination of contact angles as described in Huang et al. (2007). With these definitions of interaction
forces, the rate of change of momentum taking into account the effect of relaxation is given by

ρα (x, t + ∆t)uα (x, t + ∆t) = ρα (x, t)uα (x, t) + τα [F α (x) + F αs (x)] ∆t (27)

According to Shan and Chen, the equation of state for a single-component, multiphase flow is given through the Taylor
expansion by
GRT 2
P = ρRT + ψ (ρ) (28)
2
In what follows we are interested in the dynamic of the capillary rise using the Shan-Chen model to study the
capillary rise phenomena. The implementation of the lattice-Boltzmann scheme can be done in many different ways.
In this study, the numerical algorithm will be applied following one described in Fig. 4.

3. RESULTS AND DISCUSSION


We implement the lattice-Boltzmann method for the D2Q9 model to simulate two-phase flow in straight and sinusoidal
capillaries for different radii and undulations as it is shown in Fig. 5.
The radius of the sinusoidal capillaries is given by R(z) = Rmin + (Rmax − Rmin )/2 [1 + cos (2πz/λ)], where
Rmin is the throat diameter, Rmax the cell diameter, and λ is the wavelength of undulations in the vertical z direction.
To test our lattice-Boltzmann simulation against the Laplace law (∆P = 2σ/R), the simulation was conducted for
several initial droplet radii. Then the change of pressure differences with respect to 1/R are plotted (Fig. 6). Drops
with different radii are generated inside a computational domain using the Shan-Chen model. A plot of P versus 1/R
must be a straight line and its slope will be σ, as shown in Fig. 6.
The Laplace law is verified and the surface tension can be deduced by a linear interpolation method. For
∆ρ = 438.686 mu lu−2 , G = −120, ψ0 = 4, and ρ0 = 200 mu lu−2 , ∆P = 14.473(1/R) + 0.00181. Here the
surface tension is equal to 14.473 lu mu ts−2 . To validate the contact angle value obtained by the LBM, contact angle
is compared with the analytical results given by Dullien (1992) (see Fig. 7). If H and D represent the height and the
diameter of the drop, the contact angle θ is defined by θ = 2arctg (2H/D). The curve representing the variation of
the contact angle versus the ratio H/D was plotted and compared with the theoretical curve given by Dullien (1992).
The numerical results show good agreement with the theoretical values, as we can see in Fig. 8. The LBM simulations
of the capillary rise between two vertical parallel plates are also performed for an xy-periodic domain of size 500 ×

Journal of Porous Media


Lattice-Boltzmann Analysis of Capillary Rise 461

FIG. 4: Algorithm of the lattice-Boltzmann method

FIG. 5: Layout configurations of capillaries used in this study

800 lattice units (lu). At the solid surface, bounce-back boundary conditions were imposed. The results are compared
to a numerical solution of the corresponding equation:
d 12µzdz/dt 2σ cos θd
(zdz/dt) = − + − gz (29)
dt R2 R
where θd is the dynamic contact angle, and θd as given by Seeberg’s model (Seeberg and Berg, 1992) is
cos θ − cos θd
2.24Ca0.54 = (30)
cos θ + 1
Here θ is the static contact angle, Ca is the capillary number defined as Ca = µV /σ, and V is the contact line velocity.
The calculations permit the determination of the densities by initializing the domain randomly and performing a phase
transition. Afterward, the density within the liquid and gaseous phases is evaluated.

Volume 19, Issue 5, 2016


462 Ben Amara, Perré, & Ben Nasrallah

FIG. 6: Verification of Laplace law using the Shan-Chen model

FIG. 7: Static angle geometry

FIG. 8: Contact angle vs droplet ratio

Journal of Porous Media


Lattice-Boltzmann Analysis of Capillary Rise 463

We report here (Fig. 9) two-phase flow in a capillary with 18 slit lattices initially located at the center of the
domain. The wetting fluid is colored red while the nonwetting fluid is colored blue. In Fig. 10 we show the position
of the advancing liquid front inside the capillary as a function of time; the solution of the Washburn equation is also
reported to be compared with our numerical results. We find that the results nearly match up for long times.
The solution of the Washburn equation describes the dynamics of capillary rise given by the Shan-Chen model.
Initially, the meniscus quickly descends below the limit z = 0 and rises steadily to the equilibrium height. This
is explained by the descent that the meniscus made abruptly curve when entering the liquid in the capillary. This
observation is not noticed by using the Washburn equation. The Washburn equation does not hold true for the initial
stage of the capillary rise.
We carried out a domain independence test in order to verify that the capillary dynamics is little influenced by the
grid size. We have considered three domain sizes, 200 × 300, 400 × 600, and 500 × 800 lu2 , and the capillary radius
is set the same for each configuration. The effect of lattice size on the capillary rise height with time can be seen in
Fig. 11. From this we are able to observe that there are small differences in the capillary dynamics. We choose the fine
grid (500 × 800 lu2 ) in order to improve the resolution of the flow. It is interesting to study the flow structure inside

FIG. 9: Snapshot of capillary rise at a simulation time of 50,000 ts in a domain size of 500 × 800 lu2 and using
G = −120, ρl /ρg = 6.11, θ = 24.18474◦ , R = 18 lu

FIG. 10: Position of the liquid meniscus LBM and analytical simulations (ρl /ρg = 6.11, θ = 24.18474◦ , g = 2 ×
10−8 lu s−2 , R = 18 lu)

Volume 19, Issue 5, 2016


464 Ben Amara, Perré, & Ben Nasrallah

FIG. 11: Grid independence test with three grid systems, 200 × 300, 400 × 600, and 500 × 800 lu2 , under gravity
force g = 8 × 10−8 lu s−2 , R = 18 lu, σ = 14.473 lu mu ts−2 , and θ = 30◦

the capillary. In the inlet region of the capillary we observe the formation of recirculation zones at the walls. Near to
the meniscus, the capillary flow follows the Poiseuille flow; in the gas phase the Poiseuille flow is performed but in
the opposite direction; and at the level of the meniscus two recirculations zones are formed.
Figure 12 shows streamlines and velocity vectors in the capillary. Velocity profiles along the capillary (for y =
0.1Lcap , 0.13Lcap , 0.16Lcap , 0.2Lcap , with Lcap = 375 lu) are displayed in Fig. 13. The profile is found approxi-
mately parabolic for all positions. The effect of the presence of a liquid film on the capillary walls can be observed on
the velocity profile, which presents a small parabolic variation close to the sidewalls. Generally, in analytical studies
of capillary rise, many methods of analysis are based on the supposition that the flow inside the capillary is fully
parabolic and does not take into account the effect of liquid films on the velocity profile. To emphasize the existence

FIG. 12: (a) Fluid structure inside straight capillary and (b) flow vectors at the entrance region of the capillary

Journal of Porous Media


Lattice-Boltzmann Analysis of Capillary Rise 465

FIG. 13: Velocity profile uy (x) vs x for different cuts y = 0.1Lcap , 0.13Lcap , 0.16Lcap , and 0.2Lcap . These y
positions are considered from the beginning of the capillary with total length Lcap = 375 lu and a width of 61 lu

of oscillations at large capillary radii as mentioned by Quéré (1997), we simulate capillary rise between two parallel
plates distanced 61 lu from each other. The interface position with time is shown in Fig. 14. We observed that in
addition to a slight oscillation around the equilibrium height, irregular oscillations in the initial stage of the capillary
rise under the effect of surface tension.
Now we consider a sinusoidal capillary with narrow throats separating large cavities. Such a figure can be de-
scribed as a rotation of a sine wave about a line outside its range (see Fig. 5). The following function describes the
capillary radius:
Rmax − Rmin
R(z) = Rmin + [1 + cos (2πz/λ)] (31)
2

FIG. 14: Oscillatory behavior of liquid column. Here, density ratio, capillary radius, contact angle, and surface tension
are 6.1, 61 lu, 24◦ , and 14.473 lu mu ts−2 , respectively

Volume 19, Issue 5, 2016


466 Ben Amara, Perré, & Ben Nasrallah

Here Rmin is the throat diameter, Rmax the cell diameter, and λ the wavelength of undulations in the vertical z
direction. Rmax is always greater than Rmin . Details of the density field in a sinusoidal capillary are shown in Fig. 15.
In the case of a nonuniform capillary it is interesting to investigate the dynamic of meniscus height to better
understand the geometry effect on the capillary rise. A representative plot of meniscus position versus time is shown
in Fig. 16. We note that in the case of a sinusoidal capillary we observe an evolution in staircase. This is due to the
acceleration on the level of narrowing, i.e., when the fluid passes by Rmin , and the deceleration of velocity in the
divergent section. Different equilibrium heights are therefore possible.
The rises in various capillaries with constant radius (Rmin = 9 lu, Rmax = 18 lu) and with sinusoidal radius
variation [R(z); λ = 28 lu] are presented in Figs. 17 and 18. It is clear that the sinusoidal shape slows down the
capillary rise. The equilibrium height in the sinusoidal capillary is less important than in the case of straight ones.

FIG. 15: Capillary rise in sinusoidal capillary after 50,000 time steps with G = −120, ρl /ρg = 6, θ = 24.18474◦ ,
g = 2× 10−8 lu s−2 , Rmax = 18 lu; Rmin = 9 lu and λ = 28 lu

FIG. 16: Capillary rise kinetic in the sinusoidal capillary (G = −120, ρl /ρg = 6, θ = 24.18474◦ ,
g = 2 × 10−8 lu s−2 , Rmax = 18 lu; Rmin = 9 lu, and λ = 28 lu) after 50,000 time steps

Journal of Porous Media


Lattice-Boltzmann Analysis of Capillary Rise 467

FIG. 17: Density distribution after 100,000 time steps for the three cases R = R(z), R = Rmax = 18 lu, and
R = Rmin = 9 lu, G = −120, g = 2 × 10−8 lu s−2 , θ = 43.3841◦

FIG. 18: Dynamics of the capillary rise in sinusoidal capillary and two straight capillaries after 100,000 time steps
(R = Rmax , Rmin ) G = −120, g = 2 × 10−8 lu s−2 , θ = 43.3841◦

This is explained by the fact that the curvature radius in a sinusoidal capillary (Ben Amara and Ben Nasrallah, 2011)
is much bigger. The influence of the the undulation on the rise kinetic was examined for different wavelengths.
The plots shown in Figs. 19 and 20 illustrate the main effects of waviness. It is shown that when λ increases,
the number of stages decreases but the equilibrium height does not necessarily increase, since that is related to the
position where capillary force balances weight. More information about equilibrium height can be found in the work
of Tsori (2000).

4. CONCLUSION

This work presents numerical solutions for the dynamics of capillary rise in straight and sinusoidal capillaries. Our
simulations are based on the lattice-Boltzmann scheme for incompressible multiphase flow, which proves to be com-
petitive with traditional computational tools. The results are summarized as follows:

Volume 19, Issue 5, 2016


468 Ben Amara, Perré, & Ben Nasrallah

FIG. 19: Two-phase flow in sinusoidal capillaries with different wavelengths (λ1 = 28 lu, λ2 = 42 lu, λ3 = 16 lu,
θ = 24.18474◦ , G = −120, g = 2 × 10−8 lu s−2 )

FIG. 20: Dynamics of the capillary rise in sinusoidal capillaries with different undulations (Rmin = 9 lu, Rmax =
18 lu, θ = 24.18474◦ )

(i) The capillary rise dynamics is different from those involved in classical studies, and the Washburn equation is
not able to properly describe the capillary dynamics at the initial stage.
(ii) Sufficiently far from the start of the capillary and the walls, flow is parabolic.
(iii) Recirculation was observed in the inlet of the capillary due to the Venturi effect.
(iv) Below the meniscus we observe a development of recirculation.
(v) The sinusoidal shape of the channel walls diminishes the rise dynamic due to the higher friction and the menis-
cus jump while passing a narrowing of the passage.

Further work must be done to fully understand the internal structure of the flow inside the capillary and oscillatory
behavior of the liquid column.

Journal of Porous Media


Lattice-Boltzmann Analysis of Capillary Rise 469

REFERENCES
Asekomhe, S. O. and Elliott, J. A. W., The effect of interface deformation due to gravity on line tension measurement by the
capillary rise in a conical tube, Colloids Surf., A, vol. 220, pp. 271–278, 2003.
Ben Amara, M. E.-A. and Ben Nasrallah, S., Capillary rise in a non-uniform tube, J. Porous Media, vol. 14, no. 5, pp. 411–422,
2011.
Chen, S. and Doolen, G. D., Lattice Boltzmann method for fluid flows, Ann. Rev. Fluid Mech., vol. 30, pp. 329–364, 1998.
Chen, H., Chen, S., and Matthaeus, W. H., Recovery of the Navier-Stokes equations using a lattice-gas Boltzmann method, Phys.
Rev. A, vol. 45, pp. 5339–5342, 1992.
Czachor, H., Modelling the effect of pore structure and wetting angles on capillary rise in soils having different wettabilities, J.
Hydrol., vol. 328, pp. 604–613, 2006.
D’Humières, D., Ginzburg, I., Krafczyk, M., Lallemand, P., and Luo, L.-S., Multiple-relaxation-time lattice Boltzmann models in
three dimensions, Philos. Trans. R. Soc., A, vol. 360, pp. 437–451, 2002.
Dullien, F. A. L., Porous Media Fluid Transport and Pore Structure, Academic Press, Salt Lake City, UT, 1992.
Fries, N. and Dreyer, M., An analytic solution of capillary rise restrained by gravity, J. Colloid Interface Sci., vol. 320, pp. 259–263,
2008.
Frisch, U., D’Humières, D., Hasslacher, B., Lallemand, P., Pomeau, Y., and Rivet, J.-P., Lattice gas hydrodynamics in two and three
dimensions, Complex Syst., vol. 1, pp. 649–707, 1987.
Gunstensen, A. K., Rothman, D. H., Zaleski, S., and Zanetti, G., Lattice Boltzmann model of immiscible fluids, Phys. Rev. A, vol.
43, no. 8, pp. 4320–4327, 1991.
He, X. and Luo, L.-S., Theory of the lattice Boltzmann method from the Boltzmann equation to the lattice Boltzmann equation,
Phys. Rev. E, vol. 56, pp. 6811–6818, 1997.
He, X. Y., Shan, X. W., and Doolen, G. D., Discrete Boltzmann equation model for nonideal gases, Phys. Rev. E, vol. 57, no. 1, pp.
R13–R16, 1998.
Huang, H., Thorne, D. T., Schaap, M. G., and Sukop, M. C., Proposed approximation for contact angles in the Shan-and-Chen-type
multicomponent multiphase lattice Boltzmann models, Phys. Rev. E, vol. 76, 066701, 2007.
Martys, N. S. and Chen, H., Simulation of multicomponent fluids in complex three-dimensional geometries by the lattice Boltzmann
method, Phys. Rev. E, vol. 53, no. 1, pp. 743–750, 1996.
Qian, Y. H., D’Humieres, D., and Lallemand, P., Lattice BGK models for Navier-Stokes equation, Europhys. Lett., vol. 17, pp.
479–484, 1992.
Quéré, D., Inertial capillarity, Europhys. Lett., vol. 39, no. 5, pp. 533–538, 1997.
Raiskinmäki, P., Shakib-Manesh, A., Jäsberg, A., Koponen, A., Merikoski, J., and Timonen, J., Lattice Boltzmann simulation of
capillary rise dynamics, J. Stat. Phys., vol. 107, pp. 143–158, 2002.
Renwei, M., Luo, L.-S., and Shyy, W., An accurate curved boundary treatment in the lattice Boltzmann method, J. Comput. Phys.,
vol. 155, pp. 307–330, 1999.
Renwei, M., Shyy, W., Dazhi, Y., and Luo, L.-S., Lattice Boltzmann method for 3-D flows with curved boundary, J. Comput. Phys.,
vol. 161, pp. 680–699, 2000.
Seeberg, J. E. and Berg, J. C., Dynamic wetting in the flow of capillary number regime, Chem. Eng. Sci., vol. 47, pp. 4455–4464,
1992.
Shan, X. and Chen, H., Lattice Boltzmann model for simulating flows with multiple phases and components, Phys. Rev. E, vol. 47,
no. 3, p. 1815, 1993.
Staples, T. L. and Shaffer, D. G., Wicking flow in irregular capillaries, Colloids Surf., A, vol. 204, pp. 239–250, 2002.
Swift, M. R., Orlandini, E., Osborn, W. R., and Yeomans, J. M., Lattice Boltzmann simulations of liquid-gas and binary fluid
systems, Phys. Rev. E, vol. 54, no. 5, p. 5041, 1996.
Szekely, J., Neumann, W., and Chuang, Y., The rate of capillary penetration and the applicability of the Washburn equation, J.
Colloid Interface Sci., vol. 35, no. 2, pp. 273–278, 1971.
Tsori, Y., Discontinuous liquid rise in capillaries with non-uniform cross-sections, Langmuir, vol. 22, no. 21, pp. 8860–8863, 2006.
Wolf-Gladrow, D., Lattice-Gas Cellular Automata And Lattice Boltzmann Models, Springer, New York, 2000.
Zhmud, B. V., Tiberg, F., and Hallstensson, K., Dynamics of capillary rise, J. Colloid Interface Sci., vol. 228, pp. 263–269, 2000.

Volume 19, Issue 5, 2016

View publication stats

S-ar putea să vă placă și