Professional Documents
Culture Documents
Yuan Yu (余愿), Haihu Liu (刘海湖) , Dong Liang (梁栋), and Yonghao Zhang (张勇豪)
COLLECTIONS
Transient evolution of the heat transfer and the vapor film thickness at the drop impact in
the regime of film boiling
Physics of Fluids 30, 122109 (2018); https://doi.org/10.1063/1.5059388
© 2019 Author(s).
Physics of Fluids ARTICLE scitation.org/journal/phf
Yuan Yu (余愿),1 Haihu Liu (刘海湖),2,a) Dong Liang (梁栋),1,3 and Yonghao Zhang (张勇豪)4
AFFILIATIONS
1
School of Engineering, Sun Yat-Sen University, Guangzhou 510006, China
2
School of Energy and Power Engineering, Xi’an Jiaotong University, Xi’an 710049, China
3
Guangdong Provincial Key Laboratory of Fire Science and Technology, Guangzhou 51006, China
4
Department of Mechanical and Aerospace Engineering, University of Strathclyde, Glasgow G1 1XJ, United Kingdom
a)
Author to whom correspondence should be addressed: haihu.liu@mail.xjtu.edu.cn
ABSTRACT
We propose a lattice Boltzmann color-gradient model for immiscible ternary fluid flows, which is applicable to the fluids with a
full range of interfacial tensions, especially in near-critical and critical states. An interfacial force for N-phase systems is derived
and then introduced into the model using a body force scheme, which helps reduce spurious velocities. A generalized recoloring
algorithm is applied to produce phase segregation and ensure immiscibility of three different fluids, where an enhanced form of
segregation parameters is derived by considering the existence of Neumann’s triangle and the effect of the equilibrium contact
angle in a three-phase junction. The proposed model is first validated by two typical examples, namely, the Young-Laplace test
for a compound droplet and the spreading of a droplet between two stratified fluids. It is then used to study the structure
and stability of double droplets in a static matrix. Consistent with the theoretical stability diagram, seven possible equilibrium
morphologies are successfully reproduced by adjusting the interfacial tension ratio. By simulating near-critical and critical states
of double droplets where the outcomes are very sensitive to the model accuracy, we show that the present model is advantageous
to three-phase flow simulations and allows for accurate simulation of near-critical and critical states. Finally, we investigate the
influence of interfacial tension ratio on the behavior of a compound droplet in a three-dimensional shear flow, and four different
deformation and breakup modes are observed.
Published under license by AIP Publishing. https://doi.org/10.1063/1.5056765
studied the droplet morphology in terms of spreading coef- immiscible ternary fluids; see, for example, Refs. 50–59.
ficients and obtained the theoretical relationship between the However, the interparticle-potential model for three-phase
droplet morphology and spreading coefficients. They exper- flows suffers from high spurious velocities, lack of flexibil-
imentally observed three equilibrium morphologies of dou- ity in adjusting interfacial tensions and poor stability for
ble droplets, i.e., complete engulfing, partial engulfing, and unequal fluid viscosities. The phase-field-based model can-
non-engulfing, which correspond to three sets of spreading not conserve the mass for each fluid and produces mobility-
coefficients. Beyond these three equilibrium states, Pannacci dependent numerical results, like those in the phase-field
et al.14 identified several new morphologies of double droplets method. These limitations can be overcome by the color-
and found that the non-equilibrium morphologies can have gradient models,56,57 which are well suited to exploring the
long lifetimes controlled by hydrodynamics, which facilitates dynamic processes occurring in ternary fluid systems as pre-
the use of double droplets to produce encapsulated particles viously demonstrated by Fu et al.58 and Jiang and Tsuji.60 The
at early times and Janus particles at longer times. Guzowski existing color-gradient models for three-phase flows, how-
et al.15 presented a detailed theoretical analysis on the possi- ever, commonly suffer from a problem, i.e., three interfacial
ble equilibrium morphologies of double droplets and designed tensions should satisfy a Neumann’s triangle. In industrial pro-
the structure of double emulsions by tuning the volumes of cesses, surfactants are often added to emulsions to stabilize
the constituent segments experimentally. As a supplement to them against droplet coalescence. The presence of surfactants
theoretical and experimental studies, numerical modeling and could significantly modify the interfacial tensions so that the
simulations are becoming increasingly popular in investigation interfacial tensions do not always yield a Neumann’s trian-
of the behavior of Janus/double emulsions, which are typical gle. To correctly predict the dynamical behavior of emulsions,
of three-phase flow problems. thereby allowing precise control over the droplet geometry
and composition, it is necessary for a numerical model to be
Traditionally, three-phase flows are simulated by solv-
capable of simulating ternary fluids with a full range of inter-
ing the macroscopic Navier-Stokes equations together
facial tensions. On the other hand, it is challenging to simulate
with various approaches to capturing or tracking the
the near-critical and critical states of a ternary fluid system
interfaces between fluids. Among these approaches, the
where the largest interfacial tension is close to the sum of
front-tracking,16 volume-of-fluid (VOF),17 level-set,18–20 and
the other two as the outcomes are very sensitive to the model
phase-field21–26 methods are commonly used. However, the
accuracy.
front-tracking method is not suitable for simulating inter-
face breakup and coalescence; the VOF and level-set methods In this paper, we develop a LB color-gradient model for
require either sophisticated interface reconstruction algo- simulating immiscible ternary fluids with a full range of inter-
rithms or unphysical re-initialization processes to represent facial tensions. Based on the perturbation operator developed
the interfaces; and the phase-field method yields an inter- by Leclaire et al.,57 an interfacial force formulation is derived
face thickness far greater than its actual value,27,28 which may to describe the interactions among different fluids and is then
lead to unphysical dissolution of small droplets and mobility- introduced into the model using a body force scheme, which
dependent numerical results.29 It still remains an open ques- is found to effectively reduce spurious velocities. In addi-
tion for the phase-field method to choose an optimal mobility, tion, the recoloring algorithm proposed by Spencer et al.56 is
even for a two-phase flow problem.30 applied to maintain the interfaces and ensure immiscibility of
three different fluids, where an enhanced form of segregation
In the past decades, the lattice Boltzmann (LB) method
parameters is proposed by considering both the existence of
has developed into a promising alternative to the tradi-
Neumann’s triangle and the effect of the equilibrium contact
tional Navier-Stokes-based solvers, for simulating complex
angle in a three-phase junction. The capability and accuracy of
flow problems.31–35 It is a pseudo-molecular method track-
this model are first assessed by simulating the Young-Laplace
ing evolution of the distribution function of an assembly of
test for a compound droplet and the spreading of a droplet
molecules, built upon microscopic models and mesoscopic
between two stratified fluids. It is then used to study the
kinetic equations.36 The LB method has several advantages
structure and stability of double droplets in a static matrix
over the traditional Navier-Stokes-based solvers, e.g., the
fluid, where we emphasize the model’s capability for simu-
algorithm simplicity and parallelizability, and the ease of han-
lating ternary fluid flows in near-critical and critical states.
dling complex boundaries.37 In addition, its kinetic nature
Finally, the model is used to simulate a compound droplet in a
allows a simple incorporation of microscopic physics with-
three-dimensional shear flow, and different droplet behaviors
out suffering from the limitations in terms of length and time
are exhibited and explained.
scales typical of molecular dynamics simulations.38 Thus, the
LB method is particularly useful in the simulation of mul-
tiphase flows. The existing LB models for multiphase flows
can be generally classified into four categories: color-gradient II. NUMERICAL METHOD
model,30,39,40 interparticle-potential model,41–45 phase-field- The two-phase color-gradient LB model of Liu et al.30,61
based model,46–48 and mean-field theory model.49 These is extended to the simulation of immiscible ternary fluids. The
models have shown great success as in dealing with two- ternary color-gradient model consists of three steps, i.e., the
phase flow problems, and all of them except the mean-field collision step, the recoloring step, and the streaming step.
theory model have been extended to the modeling of In the collision step, an interfacial force that describes the
will be shown later. Recently, Fu et al.58 seemed to have also domain is taken as 160 × 160, and it is filled with three different
noticed that Eqs. (19) and (20) do not always produce convinc- fluids, which are initialized as
ing results in three-phase simulations, so they simply selected
a constant β kl , i.e.,
ρr = 1, ρg = ρb = 0 (x − 80)2 + (y − 80)2 ≤ R2r
βkl = β 0 .
(21)
ρg = 1, ρr = ρb = 0 R2r < (x − 80)2 + (y − 80)2 ≤ R2g
(27)
It is evident that the dependence of β kl on ϕ kl is not considered
in Eq. (21), and thus, incorrect results may be obtained, e.g., in ρb = 1, ρr = ρg = 0
otherwise,
the critical state.
with Rg = 2Rr . This gives the initial condition that a com-
To overcome the aforementioned drawbacks associated
pound droplet is located in the center of the computational
with the existing β kl , an enhanced form of segregation param-
domain. The interfacial tensions and the fluid viscosities are
eters is proposed. First, we determine whether Neumann’s
taken as σ rg = σ rb = σ gb = 0.01 and ν r = ν g = ν b = 0.1, respec-
triangle exists by calculating
tively, and the periodic boundary conditions are used in both
σmk
2 + σ2 − σ2 x and y directions. According to Young-Laplace’s law, when
ml kl
Xkl = . (22) the system reaches the equilibrium state, the pressure differ-
2σmk σml
ence ∆p across an interface is related to the interfacial tension
It is easily seen from Eq. (18) that Neumann’s triangle will exist σ by
if |Xkl | < 1 for all kl. Then, the segregation parameter β kl is σ
∆p = , (28)
defined as a continuous function of Xkl , R
where R is the curvature radius of the interface. Equation (28)
35ρr ρg ρb
!
βkl = β 0 + β 0 min , 1 g(Xkl ), (23) allows us to quantify the modeling accuracy of interfacial
ρ3 tensions through the relative error
where
∆pgb Rg + ∆prg Rr − σgb + σrg
= × 100%. (29)
1 Xkl < −1 σgb + σrg
1 − sin (arccos(Xkl )) −1 ≤ Xkl < 0
Table I shows the relative errors of interfacial tensions for
g(Xkl ) = . (24)
sin (arccos(Xkl )) − 1 0 ≤ Xkl ≤ 1 different values of Rr . All the relative errors are below 1.5%,
suggesting that our numerical results are in excellent agree-
1 < Xkl
−1
ment with Young-Laplace’s law. In addition to the present
forcing method, i.e., Eqs. (11) and (12), the interfacial tension
It should be noted in the three-phase junction that
effects can also be realized by the forcing method of Leclaire
Eqs. (23) and (24) can be derived based on the following
et al.,57 i.e., Eqs. (8) and (9). It is of interest to compare the
relationship:
effect of these two different forcing methods on spurious
βrg βrb βgb velocities. Table II shows the maximum spurious velocities
= = , (25) (|u| max ) for various Rr , where the values of |u| max are magnified
sin(ϕrg ) sin(ϕrb ) sin(ϕgb )
by 105 times. It is seen that the maximum spurious veloci-
and the detailed process is shown in the Appendix. Clearly, the ties are almost independent of Rr for either forcing method
proposed β kl is consistent with the nature of diffuse interfaces, and that the present spurious velocities are always smaller
thus leading to more accurate results than using other forms than those obtained with the forcing method of Leclaire
of β kl . Moreover, it works well no matter if Neumann’s triangle et al.57
exists or not.
After the recoloring step, the red, green, and blue distri- TABLE I. The relative errors of interfacial tensions for various values of Rr .
bution functions propagate to the neighboring lattice nodes,
known as the propagation or streaming step, Rr 15 20 25 30
‡
fi,k (x + ei δt , t + δt ) = fi,k (x, t), k = {r, g, b}, (26) (%) 1.3 0.95 0.83 0.57
TABLE III. Comparison between the analytical predictions and simulated results for the geometrical sizes of the liquid lens.
Case Da ha1 ha2 Ds hs1 hs2 E(D) (%) E(h1 ) (%) E(h2 ) (%)
(a) 55.34 15.97 15.97 53.59 15.92 15.92 3.17 0.37 0.38
(b) 65.18 8.73 18.81 61.79 8.79 18.97 5.21 0.59 0.82
(c) 45.84 13.23 22.92 45.39 13.35 22.51 0.99 0.85 1.80
(d) 50.99 6.83 25.49 50.93 6.77 25.22 0.13 0.94 1.09
IV. STRUCTURE AND STABILITY OF DOUBLE A. Stability diagram for double droplets
DROPLETS Consider a pair of equal-sized droplets, consisting of
Double emulsions have received considerable atten- red and green fluids and initially sitting next to each other,
tion because of their potential applications in food science, immersed in the third fluid (blue fluid). Based on the theoret-
cosmetics, pharmaceuticals, and medical diagnostics. Since ical analysis, Guzowski et al.15 presented a stability diagram
emulsion properties and functions are related to the droplet that describes the possible topologies of double droplets and
geometry and composition, it is of great importance, from a their transitions in terms of two ratios of the interfacial ten-
numerical point of view, to accurately predict the topolog- sions (see the left panel of Fig. 5). In the stability diagram,
ical structure of double droplets when dispersed in another seven typical cases (represented by the solid points) are simu-
immiscible fluid. lated to examine if the present model is able to reproduce the
TABLE IV. The fluid properties in lattice unit, the height-to-length ratios n, and ξ = 1/(6kβ 0 ),68 where k is a geometric constant that is deter-
the relative errors E n on different grid resolutions. The relative error is defined as mined by69
n−n
En = n R=40 × 100%. Note that the fluid properties in lattice unit usually should
R=40 1 X wi ei ei
be changed on different grid resolutions in order to correspond to the same physical k= . (34)
system. 2 |ei |
i
R ρk νk σ kl n En (%) For the D2Q9 model, one can obtain k ≈ 0.1504 from Eq. (34),
and thus, ξ ≈ 1.5831 for β 0 = 0.7. The periodic boundary con-
80 × 80 10 1 0.05 0.005 0.613 4.8 ditions are used in both the x and y directions. All the fluids
160 × 160 20 1 0.10 0.010 0.594 1.5
are assumed to have equal viscosity of 0.1, and the interfacial
320 × 320 40 1 0.20 0.020 0.585 0
tension σ rg is fixed at 0.01. The simulations are run until an
equilibrium state is reached, and the equilibrium morpholo-
gies of double droplets for the seven cases are depicted in
correct morphologies of double droplets. These typical cases the right panel of Fig. 5. It is seen that seven different equilib-
σ σ σ σ
are (i) σgb
rg
= 1.7 and σrb
rg
= 0.5, (ii) σgb
rg
= 2 and σrb
rg
= 1, (iii) rium morphologies are exhibited, and they can be described as
σgb σrb σgb σrb σgb
= 0.4 and = 0.4, (iv) complete engulfing of the green fluid by the red fluid (i), crit-
σrg σrg σrg = 0.5 and σrg = 0.5, (v) σrg =1
σrb σgb σ σgb ical engulfing of the green fluid by the red fluid (ii), separate
and σrg = 1, (vi) σrg = 1 and σrb = 2, and (vii) σrg = 0.5 and
σrb
rg dispersion or non-engulfing (iii), kissing (iv), partial engulfing
σrg = 1.7, which cover all the possible morphologies identified (v), critical engulfing of the red fluid by the green fluid (vi),
by Guzowski et al.15 and complete engulfing of the red fluid by the green fluid
(vii). These simulation results are consistent with the theo-
The computational domain is taken to be [1, 120] × [1, 120],
retical predictions by Guzowski et al.15 By contrast, it still
and the initial fluid distributions are
remains a challenge to use the LB phase-field-based mod-
els for correctly reproducing all the equilibrium morphologies
q
2 2
R − (x − 60.5) + (y − 60.5 − R)
ρr (x, y) = 0.5 + 0.5 tanh shown in Fig. 5. This is because when the largest interfacial
ξ , (31)
tension is greater than the other two, the free-energy den-
sity becomes negative,52,53 which often leads to numerical
q instability.
R − (x − 60.5)2 + (y − 60.5 + R)2
ρg (x, y) = 0.5 + 0.5 tanh , (32) B. Near-critical and critical states
ξ
For double droplets immersed in a static matrix, the criti-
cal state occurs when the largest interfacial tension equals the
ρb (x, y) = 1 − ρr (x, y) − ρg (x, y), (33) sum of the other two. As previously shown in Fig. 5, the criti-
cal state of double droplets can be subdivided into the kissing
where the droplet radius R = 20 lattices. The parameter ξ
state [(iv) in Fig. 5] and the critical engulfing state [(ii) or (vi)
is a measure of the interface thickness related to β 0 by
in Fig. 5]. For the convenience of description, we define the
near-critical state as the state where the largest interfacial
tension is close to the sum of the other two. It is challeng-
ing to accurately simulate the critical and near-critical states,
where a slight inaccuracy in modeling could lead to significant
simulation errors.
To highlight the strength of the present model for crit-
ical scenarios, we consider the kissing/near-kissing states
and the critical/near-critical engulfing states in a rectangu-
lar domain of [1, 300] × [1, 300]. The boundary conditions and
fluid viscosities are set the same as those in Sec. IV A. In the
kissing/near-kissing states, two equal-sized droplets with the
radii of Rr = Rg = 60 are initially placed with a distance of
drg , and they are symmetric with respect to the centerline
y = 150.5. The simulations are performed for a constant σ rg
of 0.01 but varying σ gb (=σ rb ), which is varied around the criti-
cal value of 0.005 with an increment of 2 × 10−4 . Note that the
initial distance drg depends on the value of σ gb and is given by
FIG. 5. Stability diagram representing possible morphologies of double droplets its analytical value in equilibrium as
(left panel) and equilibrium shapes of the droplets for the typical cases marked in
the stability diagram (right panel). The lines represent the critical morphologies or
the transitions between the regions of complete engulfing, partial engulfing, and Rr σrg /σgb
if σgb > 0.005;
drg = (35)
non-engulfing. R + R = 120 otherwise.
r g
In the critical/near-critical engulfing states, we consider results as the one of Fu et al.,58 so the fluid distributions from
a green droplet with Rg = 80 entirely or partially engulf- it are not shown in the figure. Clearly, both critical states are
ing a red droplet with Rr = 60 for σ gb = σ rg = 0.01. σ rb is successfully reproduced by the present model but not by the
varied around the critical value of 0.02 with an increment model of Fu et al.58 These results indicate that the present
of 2 × 10−4 . With these parameters, we are able to analyti-
cally compute other geometric parameters, which are given
R sin θ
q
by drg = R2r + R2g − 2Rr Rg cos α, Rb = gsin θ g , and dgb = Rg cos θ g
b
− Rb cos θ b for σ rb < 0.02, and by drg = Rg − Rr = 20 for
σrb R2g +d2gr −R2r
σ rb ≥ 0.02. Herein, cos α = 2σgb , θg = arccos 2Rg dgr
, and
θ b = θ g + 2α. Again, we initialize the fluid distribution such that
it follows the analytical geometric parameters.
In addition to the present model, we also use the model
of Fu et al.58 and the model of Leclaire et al.57 for the simula-
tions. When the simulations reach the steady state, we quan-
tify the interface lengths Lrg in the kissing/near-kissing states
and Lrb in the critical/near-critical engulfing states. Figure 6
compares the simulated results from the present model with
those from the model of Fu et al.58 and the model of Leclaire
et al.,57 and the analytical solutions. It is seen that for either
kissing/near-kissing states or critical/near-critical engulfing
states, the simulated results from the present model are in
good agreement with the analytical solutions, while the sim-
ulated results from the other two models significantly deviate
from the analytical solutions. Figure 7 shows the final fluid dis-
tributions obtained by the present model and the model of
Fu et al.,58 in both kissing and critical engulfing states. Note FIG. 8. The dimensionless interface lengths Lrg /Rr and Lrb /Rr versus Rr in the
that the model of Leclaire et al.57 produces almost the same kissing state and the critical engulfing state.
model is advantageous to simulate critical state in ternary of the entire flow domain is set as L × W × H = 400 × 150
fluids. × 120. The periodic boundary conditions are applied in the
x-direction, while the moving velocity boundary conditions
Finally, it is interesting to investigate how the grid res-
are imposed on the top and bottom walls using the scheme
olution influences the accuracy of simulating the kissing and
proposed by Zou and He.70 Considering the symmetry, only a
critical engulfing states for the present model and the model
half of the flow domain in the y-direction is simulated, with
of Fu et al.58 Six different resolutions are considered, i.e.,
the implementations following Liu et al.29
Rr = 20, 40, 60, 80, 100, and 120. Note that the simulation
results with the grid resolution of Rr = 60 have been shown When a compound droplet is subject to a simple shear
in Fig. 6 (see only the critical state) and Fig. 7. For different flow, two primary forces, i.e., the viscous force and the cap-
grid resolutions, the fluid properties in lattice unit and the illary force, are exerted on the droplet. The droplet behav-
lattice domain should be adjusted accordingly to correspond ior can be characterized by several important dimensionless
to the same physical system. Figure 8 shows the variation of numbers, including the Reynolds number (Re), the capillary
the dimensionless interface lengths Lrg /Rr and Lrb /Rr with number (Ca), the viscosity ratio (λvis ), and the interfacial ten-
Rr for both models. It is seen that increasing grid resolu- γ̇ ρb R2g
sion ratio (λint ), which are defined as follows: Re = µb ,
tion can improve the accuracy of simulation results, but the γ̇ µb Rg
improvements are negligibly small when Rr ≥ 60. Unlike the Ca = σgb , λvis = ν r :ν g :ν b , and λint = σ gb :σ rb :σ rg . In this study,
present model, the model Fu et al.58 cannot produce the cor- the effect of interfacial tension ratio λint is investigated for
rect outcomes of both critical states even for very high grid Re = 0.1, Ca = 0.45, and λvis = 1:1:1.
resolutions.
Seven different interfacial tension ratios are simulated,
i.e., (a) 1:0.4:0.4, (b) 1:0.5:0.5, (c) 1:1:3, (d) 1:1:2, (e) 1:1:1, (f) 1:2:1,
V. A COMPOUND DROPLET IN A
THREE-DIMENSIONAL SHEAR FLOW and (g) 1:3:1, which correspond to seven equilibrium morpholo-
gies shown previously in Fig. 5. All simulations are run until
In this section, we investigate the dynamical behavior of a the droplet reaches stable deformation or breaks up. The final
compound droplet when subject to a three-dimensional shear shapes of the compound droplet are shown in Fig. 10. Four
flow. As illustrated in Fig. 9, a spherical compound droplet, different modes are clearly seen for these interfacial tension
consisting of a red fluid core (Rr = 20) and a green fluid shell ratios considered. Cases (a) and (b) show the first breakup
(Rg = 30) immersed in a blue fluid, is initially placed in the cen- mode, where the red fluid in the core comes out and wraps
ter of two parallel walls which move at an equal speed U but around the green fluid like a thin film, and the droplet breaks
in opposite directions. The distance between these two walls up into three parts. Case (c) shows the second breakup mode,
is H = 120, and the resulting shear rate is γ̇ = 2U H . The size where the red fluid partly comes out of the green fluid and
splits into three parts, with one staying in the center with the
green fluid (which is disintegrated into two equal parts) and
the other two escaping from the center. Cases (d) and (e) show
the third breakup mode, where the red fluid comes out, splits
into two equal-sized parts, and separates and moves away
from the green droplet. In cases (f) and (g), the red fluid keeps
staying inside the green fluid, and the Janus droplet reaches a
stable deformation eventually.
The above four modes can be explained by the results in
Fig. 5 to some extent. For λint = 1:0.4:0.4 and 1:0.5:0.5, which
correspond to the morphologies (i) and (ii) in Fig. 5, the red
fluid favors to engulf the green fluid, whereas for λint = 1:2:1
and 1:3:1, which correspond to morphologies (vi) and (vii) in
Fig. 5, the green fluid favors to engulf the red fluid. For
λint = 1:1:1, the double droplets tend to form a Janus droplet
[see morphology (v) in Fig. 5], and the shear flow is strong
enough to overcome the adhesive force between the red and
green fluids. For λint = 1:1:2, the so-called kissing state occurs
[see morphology (iv) in Fig. 5], where there is no adhesive or
repulsive force between the red and green fluids, so the shear
flow can easily separate the red fluid from the green one. For
λint = 1:1:3, there exists a repulsive force between the red and
green fluids; such a repulsive force, on the one hand, can pro-
mote the separation between the red and green fluids and, on
FIG. 9. Illustration of a compound droplet in a three-dimensional shear flow. Due the other hand, can balance with the shear force so that the
to the symmetry, only a half of the flow domain in the y-direction is considered. red fluid can steadily stay in the center with the green fluid.
All the above results indicate that the interfacial tension ratio with the theoretical stability diagram by Guzowski et al.15 We
determines the interactions among different fluids, and thus then simulate the near-critical and critical states of double
it has a great effect on the dynamical behavior of a compound droplets, which is challenging since the outcomes are very
droplet. sensitive to the model accuracy. It is found that the simulated
results from the present model agree well with the analyt-
ical solutions, while the simulated results from the existing
VI. CONCLUSIONS
color-gradient models significantly deviate from the analytical
A LB color-gradient model is proposed to simulate immis- solutions, especially in critical states. Finally, we investigate
cible ternary fluids with a full range of interfacial tensions. the influence of the interfacial tension ratio on a compound
An interfacial force formulation for N-phase (N ≥ 3) systems droplet subject to a three-dimensional shear flow. Four differ-
is derived and then introduced into the model using a body ent deformation and breakup modes are observed. The inter-
force scheme, which is found to effectively reduce spurious facial tension ratio is identified as an important factor to affect
velocities. A recoloring algorithm proposed by Spencer et al.56 the dynamical behavior of a compound droplet. In summary,
is applied to produce the phase segregation and ensure the the present work provides the first LB multiphase model that
immiscibility of three different fluids, where an enhanced form allows for accurate simulation of ternary fluid flows with a full
of segregation parameters is proposed by considering the range of interfacial tensions. Although this model is limited to
existence of Neumann’s triangle and the effect of the equi- ternary fluids with similar densities, it is of great potential in
librium contact angle in a three-phase junction. The model’s the study of droplet dynamics in microfluidic devices, where
capability in modeling interfacial tensions and reducing spu- the Bond number is so small that the effect of the density
rious velocities is first validated by the Young-Laplace test for ratio can be negligible compared to the capillary and viscous
a compound droplet, and its overall performance is assessed effects.
by simulating the spreading of a droplet between two strat-
ified fluids, where both partial and complete spreadings are
predicted with satisfactory accuracy. ACKNOWLEDGMENTS
The present model is then used to study the stability This work was supported by the National Natural Sci-
and structure of double droplets in a static matrix over a ence Foundation of China (Nos. 51506168 and 51711530130), the
wide range of interfacial tensions. By changing two ratios National Key Research and Development Project of China (No.
of the interfacial tensions, seven possible equilibrium mor- 2016YFB0200902), the China Postdoctoral Science Foundation
phologies are successfully reproduced, which are consistent (No. 2016M590943), Guangdong Provincial Key Laboratory
14 39
N. Pannacci, H. Bruus, D. Bartolo, I. Etchart, T. Lockhart, Y. Hen- A. K. Gunstensen, D. H. Rothman, S. Zaleski, and G. Zanetti, “Lat-
nequin, H. Willaime, and P. Tabeling, “Equilibrium and nonequilibrium tice Boltzmann model of immiscible fluids,” Phys. Rev. A 43, 4320–4327
states in microfluidic double emulsions,” Phys. Rev. Lett. 101, 164502 (1991).
(2008). 40
T. Reis and T. N. Phillips, “Lattice Boltzmann model for simulating
15
J. Guzowski, P. M. Korczyk, S. Jakiela, and P. Garstecki, “The structure and immiscible two-phase flows,” J. Phys. A: Math. Theor. 40, 4033–4053
stability of multiple micro-droplets,” Soft Matter 8, 7269–7278 (2012). (2007).
16 41
M. Muradoglu and S. Tasoglu, “A front-tracking method for computa- X. Shan and H. Chen, “Lattice Boltzmann model for simulating flows with
tional modeling of impact and spreading of viscous droplets on solid walls,” multiple phases and components,” Phys. Rev. E 47, 1815 (1993).
Comput. Fluids 39, 615–625 (2010). 42
X. Shan and H. Chen, “Simulation of nonideal gases and liquid-gas phase
17
R. Bonhomme, J. Magnaudet, F. Duval, and B. Piar, “Inertial dynamics of transitions by the lattice Boltzmann equation,” Phys. Rev. E 49, 2941–2948
air bubbles crossing a horizontal fluid–fluid interface,” J. Fluid Mech. 707, (1994).
405–443 (2012). 43
M. Sbragaglia, R. Benzi, L. Biferale, S. Succi, K. Sugiyama, and F. Toschi,
18
H.-K. Zhao, T. Chan, B. Merriman, and S. Osher, “A variational level set “Generalized lattice Boltzmann method with multirange pseudopotential,”
approach to multiphase motion,” J. Comput. Phys. 127, 179–195 (1996). Phys. Rev. E 75, 026702 (2007).
19
K. A. Smith, F. J. Solis, and D. Chopp, “A projection method for motion of 44
M. Sbragaglia, R. Benzi, M. Bernaschi, and S. Succi, “The emergence of
triple junctions by level sets,” Interfaces Free Boundaries 4, 263–276 (2002). supramolecular forces from lattice kinetic models of non-ideal fluids: Appli-
20
R. I. Saye and J. A. Sethian, “The Voronoi implicit interface method for cations to the rheology of soft glassy materials,” Soft Matter 8, 10773–10782
computing multiphase physics,” Proc. Natl. Acad. Sci. U. S. A. 108, 19498– (2012).
19503 (2011). 45
B. Dollet, A. Scagliarini, and M. Sbragaglia, “Two-dimensional plastic flow
21
F. Boyer and C. Lapuerta, “Study of a three component Cahn-Hilliard flow of foams and emulsions in a channel: Experiments and lattice Boltzmann
model,” ESAIM: Math. Modell. Numer. Anal. 40, 653–687 (2006). simulations,” J. Fluid Mech. 766, 556–589 (2015).
22 46
F. Boyer, C. Lapuerta, S. Minjeaud, B. Piar, and M. Quintard, “Cahn- M. R. Swift, E. Orlandini, W. Osborn, and J. Yeomans, “Lattice Boltzmann
Hilliard/Navier-Stokes model for the simulation of three-phase flows,” simulations of liquid-gas and binary fluid systems,” Phys. Rev. E 54, 5041
Transp. Porous Media 82, 463–483 (2010). (1996).
23 47
J. Kim and J. S. Lowengrub, “Phase field modeling and simulation of three- Y. Wang, C. Shu, H. Huang, and C. Teo, “Multiphase lattice Boltzmann
phase flows,” Interfaces Free Boundaries 7, 435–466 (2005). flux solver for incompressible multiphase flows with large density ratio,”
24 J. Comput. Phys. 280, 404–423 (2015).
J. Kim, “Phase field computations for ternary fluid flows,” Comput. Meth-
ods Appl. Mech. Eng. 196, 4779–4788 (2007). 48
Y. Wang, C. Shu, and L. M. Yang, “An improved multiphase lattice Boltz-
25 mann flux solver for three-dimensional flows with large density ratio and
J. Kim, “A generalized continuous surface tension force formulation for
phase-field models for multi-component immiscible fluid flows,” Comput. high Reynolds number,” J. Comput. Phys. 302, 41–58 (2015).
Methods Appl. Mech. Eng. 198, 3105–3112 (2009). 49
X. He, S. Chen, and R. Zhang, “A lattice Boltzmann scheme for incompress-
26 ible multiphase flow and its application in simulation of Rayleigh-Taylor
J. Kim, “Phase-field models for multi-component fluid flows,” Commun.
Comput. Phys. 12, 613–661 (2012). instability,” J. Comput. Phys. 152, 642–663 (1999).
27 50
R. Folch and M. Plapp, “Quantitative phase-field modeling of two-phase H. Chen, B. M. Boghosian, and P. V. Coveney, “A ternary lattice Boltz-
growth,” Phys. Rev. E 72, 011602 (2005). mann model for amphiphilic fluids,” Proc. R. Soc. A 456, 2043–2057
28 (2000).
R. Spatschek and B. Eidel, “Driving forces for interface kinetics and phase
field models,” Int. J. Solids Struct. 50, 2424–2436 (2013). 51
O. Shardt, J. J. Derksen, and S. K. Mitra, “Simulations of Janus droplets at
29 equilibrium and in shear,” Phys. Fluids 26, 012104 (2014).
H. Liu, Y. Ba, L. Wu, Z. Li, G. Xi, and Y. Zhang, “A hybrid lattice Boltz-
mann and finite difference method for droplet dynamics with insoluble 52
H. Liang, B. C. Shi, and Z. Chai, “Lattice Boltzmann modeling of three-
surfactants,” J. Fluid Mech. 837, 381–412 (2018). phase incompressible flows,” Phys. Rev. E 93, 013308 (2016).
30 53
H. Liu, A. J. Valocchi, and Q. Kang, “Three-dimensional lattice Boltzmann C. Semprebon, T. Kruger, and H. Kusumaatmaja, “Ternary free energy lat-
model for immiscible two-phase flow simulations,” Phys. Rev. E 85, 046309 tice Boltzmann model with tunable surface tensions and contact angles,”
(2012). Phys. Rev. E 93, 033305 (2016).
31 54
A. Mazloomi Moqaddam, S. S. Chikatamarla, and I. V. Karlin, “Simulation of M. Wöhrwag, C. Semprebon, A. M. Moqaddam, I. Karlin, and H. Kusumaat-
binary droplet collisions with the entropic lattice Boltzmann method,” Phys. maja, “Ternary free-energy entropic lattice Boltzmann model with a high
Fluids 28, 022106 (2016). density ratio,” Phys. Rev. Lett. 120, 234501 (2018).
32 55
X.-D. Niu, Y. Li, Y.-R. Ma, M.-F. Chen, X. Li, and Q.-Z. Li, “A mass- Y. Shi, G. Tang, and Y. Wang, “Simulation of three-component fluid flows
conserving multiphase lattice Boltzmann model for simulation of multi- using the multiphase lattice Boltzmann flux solver,” J. Comput. Phys. 314,
phase flows,” Phys. Fluids 30, 013302 (2018). 228–243 (2016).
33 56
F. Qin, A. Mazloomi Moqaddam, Q. Kang, D. Derome, and J. Carmeliet, T. Spencer, I. Halliday, and C. M. Care, “Lattice Boltzmann equation
“Entropic multiple-relaxation-time multirange pseudopotential lattice method for multiple immiscible continuum fluids,” Phys. Rev. E 82, 066701
Boltzmann model for two-phase flow,” Phys. Fluids 30, 032104 (2010).
(2018). 57
S. Leclaire, M. Reggio, and J. Trepanier, “Progress and investigation on
34
E. Ezzatneshan, “Study of surface wettability effect on cavitation incep- lattice Boltzmann modeling of multiple immiscible fluids or components
tion by implementation of the lattice Boltzmann method,” Phys. Fluids 29, with variable density and viscosity ratios,” J. Comput. Phys. 246, 318–342
113304 (2017). (2013).
35 58
H. Yuan, Y. Wang, and C. Shu, “An adaptive mesh refinement-multiphase Y. Fu, S. Zhao, L. Bai, Y. Jin, and Y. Cheng, “Numerical study of dou-
lattice Boltzmann flux solver for simulation of complex binary fluid flows,” ble emulsion formation in microchannels by a ternary lattice Boltzmann
Phys. Fluids 29, 123604 (2017). method,” Chem. Eng. Sci. 146, 126–134 (2016).
36 59
X. He and L.-S. Luo, “A priori derivation of the lattice Boltzmann equation,” R. H. H. Abadi, M. H. Rahimian, and A. Fakhari, “Conservative phase-field
Phys. Rev. E 55, R6333 (1997). lattice-Boltzmann model for ternary fluids,” J. Comput. Phys. 374, 668–691
37 (2018).
S. Chen and G. D. Doolen, “Lattice Boltzmann method for fluid flows,”
Annu. Rev. Fluid Mech. 30, 329–364 (1998). 60
F. Jiang and T. Tsuji, “Estimation of three-phase relative permeability by
38 simulating fluid dynamics directly on rock-microstructure images,” Water
C. K. Aidun and J. R. Clausen, “Lattice-Boltzmann method for complex
flows,” Annu. Rev. Fluid Mech. 42, 439–472 (2010). Resour. Res. 53, 11–32, https://doi.org/10.1002/2016wr019098 (2017).
61 66
H. Liu and Y. Zhang, “Modelling thermocapillary migration of a microflu- M. Latva-Kokko and D. H. Rothman, “Diffusion properties of gradient-
idic droplet on a solid surface,” J. Comput. Phys. 280, 37–53 (2015). based lattice Boltzmann models of immiscible fluids,” Phys. Rev. E 71, 056702
62
Z. Guo, C. Zheng, and B. Shi, “Discrete lattice effects on the forcing term (2005).
67
in the lattice Boltzmann method.” Phys. Rev. E 65, 046308 (2002). J. S. Rowlinson and B. Widom, Molecular Theory of Capillarity (Courier
63
I. Halliday, C. M. Care, and A. P. Hollis, “Improved simulation of drop Corporation, 2013).
68
dynamics in a shear flow at low Reynolds and capillary number,” Phys. Rev. E A. Riaud, S. Zhao, K. Wang, Y. Cheng, and G. Luo, “Lattice-Boltzmann
73, 056708 (2006). method for the simulation of multiphase mass transfer and reaction of dilute
64
Y. Ba, H. Liu, Q. Li, Q. Kang, and J. Sun, “Multiple-relaxation-time color- species,” Phys. Rev. E 89, 053308 (2014).
69
gradient lattice Boltzmann model for simulating two-phase flows with high H. Liu, L. Wu, Y. Ba, and G. Xi, “A lattice Boltzmann method for axisym-
density ratio,” Phys. Rev. E 94, 023310 (2016). metric thermocapillary flows,” Int. J. Heat Mass Transfer 104, 337–350
65
U. D’Ortona, D. Salin, M. Cieplak, R. B. Rybka, and J. R. Banavar, “Two-color (2017).
70
nonlinear Boltzmann cellular automata: Surface tension and wetting,” Phys. Q. Zou and X. He, “On pressure and velocity boundary conditions for the
Rev. E 51, 3718 (1995). lattice Boltzmann BGK model,” Phys. Fluids 9, 1591–1598 (1997).