Date post: | 08-Apr-2018 |
Category: |
Documents |
Upload: | vundavilliravindra |
View: | 221 times |
Download: | 0 times |
of 14
8/6/2019 Numerical Study on the Effect of Viscoelasticity on Drop Deformation
1/14
J. Non-Newtonian Fluid Mech. 155 (2008) 8093
Contents lists available at ScienceDirect
Journal of Non-Newtonian Fluid Mechanics
j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / j n n f m
Numerical study on the effect of viscoelasticity on drop deformation in simple
shear and 5:1:5 planar contraction/expansion microchannel
Changkwon Chung a, Martien A. Hulsen b, Ju Min Kim c, Kyung Hyun Ahn a,, Seung Jong Lee a
a School of Chemical and Biological Engineering, Seoul National University, Seoul 151-744, Republic of Koreab Department of Mechanical Engineering, Eindhoven University of Technology, PO Box 513, 5600 MB Eindhoven, The Netherlandsc Department of Chemical Engineering, Ajou University, Suwon 443-749, Republic of Korea
a r t i c l e i n f o
Article history:
Received 6 March 2008
Received in revised form 21 May 2008
Accepted 4 June 2008
Keywords:
Drop deformation
Shear flow
Contraction/expansion flow
Finite element method
Front tracking method
Oldroyd-B model
DEVSS-G
SUPG
Matrix logarithm
a b s t r a c t
We implemented a finite element-front tracking method (FE-FTM) to understand the drop dynamics in
microfluidic applications. We investigated the effect of viscoelasticity of both drop and medium. The
Oldroyd-B model was used for viscoelastic fluid, and DEVSS-G/SUPG/matrix logarithm algorithms were
applied to improve numericalstability.We first verified thereliabilityof thealgorithm by comparingwith
previous results under simple shear flow. The results were in good agreement with previous reports in a
wide range of parameters such as capillary number ( Ca) and Deborah number (De). Then we applied the
algorithm to a 5:1:5 planar contraction narrow channel expansion flow which is typical microfluidic
flows. One of the goal of this study is to explore the effect of viscoelasticity of drop and medium on
drop deformation, and to propose the strategy to control the drop shape. When a Newtonian drop was
suspendedin a viscoelastic medium, in thenarrow channel region, we observed an ellipsoid-like drop in
contrast to a bullet-like drop which is the typical shape for the case of a Newtonian drop in Newtonian
medium. When a viscoelastic drop is suspended in Newtonian medium, the extent of drop swell to the
cross-stream directionwas enhanced at the exitof the narrow channel.We explainedthese phenomena in
terms of thenormal stressdifference developedin theviscoelastic fluid. Thepresent study shows that the
viscoelasticityplays a significant role ondrop dynamicsand drop shape inparticular.We expect this studywill be helpful to understand the drop dynamics in microchannel flow, and provide useful information in
manipulating drops in complicated geometries such as microfluidic channels.
2008 Elsevier B.V. All rights reserved.
1. Introduction
Recently, microparticle fabrication attracts much attention due
to the increasing demand for potential applications such as drug
delivery [1], photonic crystal [2] and DNA multi-plexed analysis
[3]. Microfluidic channels have been extensively used due to their
inherent high-throughput way of operation [4,5]. In microparti-
cle fabrication, one important issue is to control the shape of the
microparticles, which is typically realized through two-phase fluid
systems [57] or continuous flow lithography [3,4,8]. Thus, it is
essential to understand drop dynamics in two-phase systems to
generate a targeted shape. In the conventional industries, drop
deformation problem is also quite important in polymer blends
[912] and polymer processing [13] to control the microstructure
and rheology and thus to enhance the product quality.
Corresponding author. Tel.: +82 2 880 8322; fax: +82 2 880 1580.
E-mail address: [email protected] (K.H. Ahn).
In these problems, drop dynamics is governed by diverse forces
such as surface tension and viscous forces, and the rheology of the
fluid can be important when non-Newtonian fluids are involved.
Several phenomenological models [1418] have been proposed to
explain drop dynamics in Newtonian and viscoelastic fluids which
show good agreement with experiments. However, these theories
are limited to small deformations in simple viscometric flows since
thedropshape is assumed tobe ellipsoidal.Thus,thereis a growing
demand for numerical analysis of drop deformation to gain phys-
ical insights for academic interest as well as practical purposes.
A number of numerical studies that focused on non-Newtonian
effects have been conducted for drop deformation in viscometric
flows, e.g., shear flow [1925] and extensional flow [2628]. Also,
non-Newtonian effects in complicated flow problems have been
investigated for various applications such as gas injection in a cap-
illary tube [29], bubble rising [30], pendant drop formation [31,32]
and drop deformation in contraction/expansion flows [33,34]. As
for drop deformation in contraction/expansion flows, only a few
numerical studies have been reported on axisymmetric drop defor-
mation in contraction/expansion flow [33,35] and on planar drop
0377-0257/$ see front matter 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.jnnfm.2008.06.002
http://www.sciencedirect.com/science/journal/03770257mailto:[email protected]://dx.doi.org/10.1016/j.jnnfm.2008.06.002http://dx.doi.org/10.1016/j.jnnfm.2008.06.002mailto:[email protected]://www.sciencedirect.com/science/journal/037702578/6/2019 Numerical Study on the Effect of Viscoelasticity on Drop Deformation
2/14
C. Chung et al. / J. Non-Newtonian Fluid Mech. 155 (2008) 8093 81
Table 1
Previous works for viscoelastic effects on drop deformation in shear flow
Authors Algorithm Dimension VE model Fluids Ca De
Pillapakkam and Singh [25] Level-set 2D, 3D Oldroyd-B NV, VN 0.24 0.4
0.6 8
60 8
Yue et al. [24] Diffuse interface 2D Oldroyd-B NV 0.24 0.4
Yue et al. [22] Diffuse interface 2D Oldroyd-B NV, VN 0.1, 0.2 2Yue et al. [23] Diffuse interface 2D Oldroyd-B NV 0.1 10
VN 0.1 2
Chinyoka et al. [21] Volume of fluid 2D Oldroyd-B NV, VN, VV 0.24, 0.6 0.4
NV, VN 60 8
Khismatullin et al. [20] Volume of fluid 3D Oldroyd-B, Giesekus NV 0.24, 0.6 0.4
Aggarwal and Sarkar [41] Front tracking 3D Oldroyd-B VN 0.35 3
Fig. 1. Schematic diagram for remeshing algorithm of front mesh: (a) element addition; (b) element deletion.
deformation in convergentdivergent flow [34,36]. Furthermore,
these studies were limited to Newtonian fluids [35,36] or shear-
thinning fluids modelled by the Carreau model [33], and the effect
of viscoelasticity has rarely been reported.
There are many studies on the effect of viscoelasticity on drop
deformation in simple shear flow. Boger fluids have been consid-
ered as a model fluid to investigate the effect of viscoelasticity
in shear flow [3740]. The elasticity of medium fluid helps to
align a drop along the flow direction and to enhance deforma-
tion in shear flow, whereas the elasticity of a drop contributes
to resist drop deformation [3840]. The adverse tendency of vis-
coelastic effects on drop dynamics was qualitatively reproduced in
two-dimensional simulations for small deformation (Ca = 0.1, 0.2)[22] and for large deformation (Ca = 0.6) [21,25]. Recently, three-
dimensional numerical simulations were performed [20,41]. How-
ever, a discussion for accurate viscoelastic solutions is still under
way even in two-dimensional approaches [2125]. Detailed infor-
mation for related works is presented in Table 1. Comparing our
solutions with previous works [2125] is quite meaningful since
the problem is regarded as a good model problem to validate the
numerical algorithm. We also study drop deformation in 5:1:5 pla-
nar contraction/expansion flow which is frequently encountered in
microfluidics. Though numerous experimental studies have been
conducted on drop manipulation at small length scale [42,43], the
numerical study on the effect of viscoelaticity for drop dynamics
has rarely been presented. Here, we present how viscoelasticity of
the fluid affects drop dynamics in microchannel flows.
We implemented a front tracking method [44] to dealwithtwo-
phase flow problems for various applications. The advantage of a
front tracking method is that it provides robust solutions since a
large number of front particles on the interface are advected, i.e.,
tracking interfaces into the subgrid level is available [43,45,46].
As a sharp interface method, the front tracking method is also
applicable to viscoelastic fluid problems.A front tracking algorithm
based on finite difference method (FD-FTM) has been presented
Fig. 2. Definition of unit tangent vector and unit normal vector in front elements.
8/6/2019 Numerical Study on the Effect of Viscoelasticity on Drop Deformation
3/14
82 C. Chung et al. / J. Non-Newtonian Fluid Mech. 155 (2008) 8093
Fig. 3. Schematic diagram of the drop deformation under shear flow: (a) initial drop and computational domain; (b) characterization of deformed drop.
for two-dimensional dropdeformation in periodic extensional flow
[28,47] and for three-dimensional viscoelastic drop deformation in
shear flow [41]. However, the finite difference based approaches
[28,41,44,47,48] are subjected to geometrically simple problems
with structured meshes. Here, we propose a finite element-front
tracking method (FE-FTM) to investigate the effect of viscoelas-
ticity on two-dimensional drop deformation. We believe that this
method is more adaptable to microfluidic applications since finite
element method has no geometrical constraints.
In the next section, we will explain details of the numerical
formulation of the FE-FTM algorithm. Then we apply the present
algorithm to drop deformation problems under simple shear flow
to validate the reliability of the algorithm for both Newtonian and
viscoelastic fluids. Then we investigate the dropdynamicsin a 5:1:5
Fig. 4. Mesh configurations: (a) UM1; (b) UM2; (c) UM3; (d) UC2.
8/6/2019 Numerical Study on the Effect of Viscoelasticity on Drop Deformation
4/14
C. Chung et al. / J. Non-Newtonian Fluid Mech. 155 (2008) 8093 83
planar contraction/expansion flow to show possible applications
to microfluidics. We will investigate the effect of viscoelasticity
on drop dynamics. As for effect of fluid elasticity, we will discuss
the role of the normal stress difference on drop deformation in
microchannel flow, which would be useful in controlling the drop
shape using viscoelastic fluids in microfluidic applications.
2. Numerical method
2.1. Governing equations
We consider incompressible and immiscible two-phase fluid
flow. Inertia is assumed to be negligible. The Oldroyd-B model
is adopted as a constitutive equation for the viscoelastic fluid.
Momentum, continuity and constitutive equations are denoted as
follows:
p s (u + (u)T) nl(x xp) = 0, (1a)
u = 0, (1b)
C
t+ u C (u)T C C u =
1
(C I), (1c)
wherep denotes the pressure,s the solvent viscosity, u the velocityvector, the extra stress tensor, C the conformation tensor, t the
time, the relaxation time of polymer and I the identity tensor,respectively. In Eq. (1a), the last term nl(x xp) correspondsto capillary force, whereis the surface tension coefficient, twice
the local mean curvature of the surface, nl an outward unit normal
vector fromthe surface, and(x xp) theDirac deltafunctionwhichisnon-zeroonlyatx =xp.Here,x is theposition vector in thedomain
andxp is thepositionvector to designate theinterface location.The
relationship between the extra stress tensor and conformation
tensor C is given by
=(1 )
(C I), (2)
where I is theidentitytensor. represents the solvent contributionto solution viscosity as follows:
=s
p + s=
s, (3)
where p is the polymer contribution to the solution viscosity.
2.2. Numerical formulation
Thefinite element method is employed to discretize thegovern-
ing equations Eqs. (1a)(1c). The DEVSS-G scheme [49] and SUPG
scheme [50] are adopted as stabilization schemes. We reformulate
the momentum Eq. (1a), the continuity Eq. (1b) with the DEVSS-G
scheme [49] into following weak form:
;pI + ;(u + (u)T) (1 )(GT + G)
+; {;nl(x xp)} ; t
= 0, (4a)
; u = 0, (4b)
; G (u)T = 0, (4c)
where and are bilinear andbiquadratic shape functions in two-dimension, respectively. and { } denote domain integral and line
integral along the finite element, and { }r means the line integral
along the interface, respectively. is a one-dimensional quadraticshape function, G the velocity-gradient tensor. Variables such as
p, G and are approximated in terms of bilinear shape functions,
while u is discretized with biquadratic shape function. Traction on
the boundary is t =Tnd, whereT = pI ++s(u + (u)T) and nd is
an outward normalvector from the boundary in the computational
domain. With Eq. (4c), the momentum balance Eq. (4a) is stabilized
by introducing an elliptic term [49].
In this study, we employ the matrix logarithm scheme [51] to
enhance numerical stability, such that the positive definiteness
of the conformation tensor C be preserved during computation.
The conformation tensor can be diagonalized with the relation-ship C = RcRT, where R is a matrix composed of the eigenvectors
of C and the diagonal tensor c is constructed from the corre-
sponding eigenvalues [52,53]. We replace the C-based constitutive
equation with the logarithm tensor based formulation. Thus, we
deal with the evolution equation of s = log c =dim
i=1log(ci)nini =dim
i=1sinini , where s is the logarithm tensor in the principal frame.
ci denotes the eigenvalue of the conformation tensor, si is the
eigenvalue of the logarithmtensor, ni is the principaldirection con-
jugated with ci and dim is set to 2 as the dimension number. The
evolution equation makes the problem more stable by taking the
logarithm in order to change the growth of the stress tensor form
exponetial to linear, where the exponential growth is regarded as
the source of the numerical breakdown. The time derivative ofs for
Oldroyd-B model can be written as follows [51]:
s=
dimi=1
2Gii
1
(ci Iii)
ci
nini+
dimi=1
dimj=1
si sjci cj
i /= j
(cjGij + ciGji)ninj,
(5)
Fig. 5. Steady drop deformation with increasing Ca. Viscosity of drop fluid is same
with the medium (= 1). (a) Deformation parameter D; (b) orientation angle .
8/6/2019 Numerical Study on the Effect of Viscoelasticity on Drop Deformation
5/14
84 C. Chung et al. / J. Non-Newtonian Fluid Mech. 155 (2008) 8093
Fig. 6. Contour of extensional rate N based on drop orientation (UM3). N = m1 GT
m1, where m1 is the unit vector directing major principal direction of deformed drop.
(a) Ca = 0.1; (b) Ca = 0.3; (c) Ca = 0.5; (d) Ca = 1.0.
where Gii is the velocity-gradienttensor in the principalframe. Con-
sequently, the equation described by the logarithm tensor S in the
global frame is obtained as follows:
S
t+ u S = S, (6)
where S is the transformed tensor of s through the matrix diago-
nalization as S = R s RT. Since C and S are coaxial, it is possible
to build R using the eigenvectors of C. The discrete form of Eq. (6)
with SUPG [50] scheme can be written as
+ s; Sn+1
Sn
t
+ un+1 Sn+1 = + s; Sn, (7)where s is the element-wise upwinding shape function,s =(|uchx| + |uchy|)/(2ucuc). uc denotes the velocity vector atthe center node of an element, hx and uy are the size vectors of the
element which are projected on x-axis and y-axis, respectively. In
this study, a streamline upwinding coefficient of 2 is used, fol-
Table 2
Detailed information of meshes used in this study
Name Elements Nodes DOF xmin/0.5lw
UM1 2,300 9,333 49,437 0.1
UM2 7,935 31,909 168,078 0.05
UM3 26,868 107,689 566,079 0.025
UC2 14,494 58,353 307,585 0.05
lowing some literatures [5456], though it is usually assumed as 1
in other viscoelastic fluid studies [50,57]. It is generally conceded
that proper value of is adopted depending on the problem sincesomewhat large value may introduce a serious error [57]. Super-
script n and n + 1 denote each time step. With the solution S of Eq.
(7), we need to transform it into the principal frame by s = RTSR to
get the conformation tensor C using C = RcRT= RexpsRT. Finally,
the extra stress tensor is calculated by Eq. (2). Then, the solu-
tion vector [G, u, p]T can be obtained after solving the coupled Eqs.
(4a)(4c) at every time step.
2.3. Front tracking method
2.3.1. Temporal integration of interfaces
In the front tracking method, successive front elements com-
prise the interface as in Fig. 1(a), where each front element has two
front particles. The position of front particle, xp on the interface is
Table 3
Comparison of dropdeformationat steady state (t= 3 1, Ca = 0.24,De =0.4, =0.5)
Our data (UC2 mesh) Chinyoka [68]
D D
NN 0.2674 28.90 0.2878 32.26
VN 0.2582 30.32 0.2799 32.53
NV 0.2550 25.38 0.2656 28.12
VV 0.2455 27.00 0.2598 28.22
8/6/2019 Numerical Study on the Effect of Viscoelasticity on Drop Deformation
6/14
C. Chung et al. / J. Non-Newtonian Fluid Mech. 155 (2008) 8093 85
Fig. 7. Steady drop deformation with increasing De (Ca = 0.1, = 0.5, UC2). N means
Newtonian fluid and V means viscoelastic fluid. The first character is drop fluid
and the latter one is medium fluid. (a) Deformation parameter D; (b) orientation
angle .
convected with fluid velocity. The interface moves according to the
following equation:
dxpdt
= up, (8)
Fig. 9. Steady drop deformation with increasing De (Ca =0.5, = 0.5, UC2).
where up is the particle velocity of the coordinate xp, which is
interpolated with a biquadratic shape function. New position of
the particle xn+1p is updated from the previous step xnp using the
Runge-Kutta second-order method (RK2) as follows:
xp = tup(xnp, t), (9a)
Fig.8. Shear component ofpolymerconformation tensor C basedon droporientation (Dei =0.5,i =0.5, Ca = 0.1,UC2).The shear component ofC is representedas C = m1Cm2,
where m1 and m2 are the unit vectors directing major and minor principal directions of deformed drop. (a) NV; (b) VN.
8/6/2019 Numerical Study on the Effect of Viscoelasticity on Drop Deformation
7/14
86 C. Chung et al. / J. Non-Newtonian Fluid Mech. 155 (2008) 8093
xp = tup(xnp + 0.5xp, t), (9b)
xn+1p = xnp + xp, (9c)
where xp and xp denote the intermediate position vectors of front
particles.
2.3.2. Remeshing of front mesh
The distances between two successive front nodes are initiallysetto a singleuniform value. However, these distancesmay become
nonuniform as the front moves along the flow field. Thus, we need
to keep the distance in a proper range to conserve the accuracy
of interface tracking during calculation. In this study, we conduct
front element addition/deletion as shown in Fig. 1 to keep the size
of front element, sl in the range of 2050% with respect to h, thatis the diagonal size of the smallest mesh in the domain. If a front
element longer than 0.5h is detected as in Fig. 1(a), the element is
divided into two equally sized elements by inserting a new front
particle. If a front element shorter than 0.2h is detected, two neigh-
boring elements are connected at the center of the element after
the element is removed as in Fig. 1(b).
2.3.3. Treatment of surface tension
Surface tension is treated as a force term nl(x xp) in Eq.4(a). The surface tension at finite element node, Fi is computed
using the force at a front element l,fl, with the immersed boundary
method [58]:
Fi(x) =l
fl(xp)2h(x xp). (10)
In the two-dimensional case, the smoothed delta function 2h witha range of two times mesh size h is approximated as follows:
2h(x xp) = d2h(x xp)d2h(y yp), (11)
where d2h(r) is defined as follows [59]:
d2h(r) = 1 + cos(r/2h)
4h |r| 2h0 |r| > 2h
. (12)
Surface tension at a front element fl is calculated based on the
interface topology. Here, we calculate the surface tension using
the curvature of front elements directly [46] instead of polynomial
fitting of interfaces [44,48]. Surface tension fl is calculated using
Frenet-Serret theorem [60] at xp as follows:
fl =
ds
nlds =
ds
dtl
dsds =
pbpa
dtl = (tlpb tlpa
), (13)
where denotes the surface tension coefficient, is the Dirac deltafunction which has non-zero only at the interface and ds is the
length of a front element. tl is a unit tangent vector at front par-
ticles pa and pb of the front element l as shown in Fig. 2. Thisfront element-wise integration results in inward force at the front
particles as shown in Fig. 2.
In this study, we consider two-phase system of purely immis-
cible fluids. However, we deal with this system in a way that the
material properties continuously change across the interface. We
utilize the Heaviside function H(x) to interpolate material prop-
erties of two-phase fluids. The Heaviside function is computed by
solving a Poisson equation with essential boundary condition and
a simple filtering algorithm [48]:
2H(x) = N =
ds
nc2h(x xc)ds, (14)
where N is the outward unit normal vector in the finite element
mesh, which is distributed from the front mesh (interfaces) and
nc is the outward unit normal vector at the center of a front ele-
ment as designated in Fig. 2. In this case, the immersed boundary
method [58] is also used to calculate N in the finite element mesh.
xc is the position of the center of a front element. In the present
study, fluid viscosity (), relaxation time of polymer () andsolventcontribution to solution viscosity () are interpolated as follows:
(x) = m + (d m)H(x), (15a)
(x) = m + (d m)H(x), (15b)
(x) = m + (d m)H(x), (15c)
wherethe subscriptd and m denote drop andmedium, respectively.
Fig. 10. Schematic diagram of drop deformation in contraction/expansion flow.
8/6/2019 Numerical Study on the Effect of Viscoelasticity on Drop Deformation
8/14
C. Chung et al. / J. Non-Newtonian Fluid Mech. 155 (2008) 8093 87
3. Results and discussion
3.1. Drop deformation in simple shear flow
3.1.1. Problem description
We first consider drop deformation under simple shear flow to
verifyour finite element-front tracking algorithmand to gain phys-
ical insights on dropdynamics. Thedropwiththe radius a is initially
positioned at the center of the simulation domain. We assume that
two side walls move in reverse directions with a constant velocity,
which results in a constant shear rate . The computational domainhas a width oflw and a length of 24a, as shown in Fig. 3(a). In this
study, four different cases areconsidered: Newtonian drop in New-
tonian medium (NN),Newtonian dropin viscoelastic medium (NV),
viscoelastic drop in Newtonian medium (VN) and viscoelastic drop
in viscoelastic medium (VV). The drop deformation parameter D is
defined as D = (L B)/L + B as shown in Fig. 3(b), where L and B are
the length along two principal directions m1, m2 of the deformed
drop, respectively. We define L and B as the longest and the short-
est distance from the drop center to the interface. The orientation
angle is defined as an included angle between the principal axisof the deformed drop (m1) and the flow direction as shown in
Fig. 3(b).
3.1.2. Parameters
The system is governed by the following dimensionless num-
bers: capillary number (Ca), Deborah number (De) and viscosity
ratio of drop to medium () defined as follows:
Ca =ma
, (16)
Dei=i, (17)
=dm
, (18)
where Dei is defined for drop and medium (i = d, m), respectively.
In our simulation, drop and medium are assumed to have the same
viscosity, i.e.,= 1 forconvenience, andthe relaxation times of dropand medium for the VV system are assumed to be the same. Mesh
configurations are shown in Fig. 4 with wall confinement ratio
(a/lw) of 0.25. We refined the center region with structured ele-
ments to more accurately resolve the drop deformation. Detailed
information of each mesh is given in Table 2.
3.1.3. Code verification for NN
3.1.3.1. Effect of wall confinement. The drop deformation is knownto increase as wall confinement ratio (a/lw) increases [6163]. In
Fig. 11. Mesh configurations (25,010 elements, 101,005 nodes,xmin/w = 0.05): (a) whole domain; (b) zoomed view of contraction region.
8/6/2019 Numerical Study on the Effect of Viscoelasticity on Drop Deformation
9/14
88 C. Chung et al. / J. Non-Newtonian Fluid Mech. 155 (2008) 8093
this section, we investigated how the wallconfinementaffects drop
deformation in two-dimensional flow as a preliminary study. With
an additional mesh,we confirmed thatthe dropdeformationshows
a very weak dependency on the wall confinement: e.g., 6.1% of D
deviation for Ca = 0.1 between a/lw = 0.25 and a/lw = 0.125, and 7.9%
ofD deviation for Ca = 0.5. In this study, we set a/lw as 0.25 to com-
pare our solutions with previous studies [21,24,64], following the
convention that the wall confinement effect can be neglected for
a/lw = 0.25 so that the drops do not interact with the wall [21].
3.1.3.2. Effect of Ca. We first investigated the drop deformation
problem for the NN system with unstructured meshes UM1, UM2
and UM3. A fully developed velocity profile was imposed at the
inlet. The results for D and at steady state are presented in Fig. 5.As Ca increases, D increases and decreases, i.e., the drop is morealigned with the flow direction. In Fig. 5(a), our results are con-
sistent with previous reports [19,21,24,64] in a wide range of Ca.
Hence, it can be concluded that the present FE-FTM is reliable. For
small deformation, Ca 0.3 in Fig. 5(b), coincides with the pre-diction from the perturbation theory [65,66]. Though the theory is
based on three-dimensional drop deformation, it coincides with
the present and existing results, which implies that the viscous
stress to the vorticity direction gives little effect on the orienta-
tion angle. In the previous study [64], the drop did not show the
steady-state shape over critical Ca (Cac) 0.875, i.e., the drop was
continuously extended for Ca > Cac. However, in our simulations,
the steady-state shape is maintainedup to Ca 1. Major differences
between the study in Ref. [64] and the present study are the size
of the computational domain and the inlet boundary condition. In
Ref. [64], the domain wassmaller and periodic boundaryconditions
were imposed, therefore, closely packed drops were considered,
while a steady velocity profile was imposed in the present study,
i.e., we investigate singledrop dynamicsexcluding dropdrop inter-
action. In addition, it is noteworthy that there were no differences
between fully developed condition and periodic condition, which
implies thatour domain is large enoughalong the streamwisedirec-
tion.Tounderstand the kinematic characteristics of the sheared drop,
we decoupled the extensional and rotational component of the
velocity-gradient tensor since shear flow is kinematically a mix-
ture of stretching and rotation [67]. Following Yue et al. [22], the
extensional rate N was estimated as N = m1 GT
m1, where G
is the velocity-gradient tensor and m1 is the unit vector in the
major principal direction as shown in Fig. 3(b). The drop shape
is related to the extensional rate N since the extensional stressexerted in the orientation direction is considered as a primary
factor on drop deformation. The contours of the extensional rate
are shown for various Ca in Fig. 6. For weakly deformed drops
(Ca = 0.10.5; Fig. 6(a)(c)), a compressive rate (equivalent to nega-
tive N) is observed inside the drop, while a high extensional rate is
found near the tip; therefore the ellipsoidal shape is developed atmoderate Ca. For a highly extended drop (Ca = 1; Fig. 6(d)), a com-
pressive rate is seen atbothends insidethe drop,and an extensional
rate is observed in the drop center for Ca = 1, i.e., the extensional
stress at the center tends toseparatethe dropin two drops, and the
compressive stress at both ends tends to develop bulbous shapes.
Therefore, we can conclude that the drop shape is determined by
the distribution of the extensional stress.
3.1.4. Code verification for effect of viscoelasticity
3.1.4.1. Effect of viscoelasticity (Ca = 0.24). The effect of viscoelas-
ticity on drop deformation was first studied with numerical
simulation by Pillapakkam and Singh [25] using the level-set
method. They employed the Oldroyd-B model with = 0.5 and
obtained D =0.48 for Ca = 0.24 and De = 0.4 when both drop and
medium are viscoelastic. Later, Yue et al. [24] considered the same
problem using the diffuse-interface method and showed a D value
similar to Pillapakkam and Singh [25]. Recently, Chinyoka et al. [21]
revisited the same problem using the volume of fluid method and
obtained D = 0.26. In the present study, the UC2 mesh (Fig. 4(d))
was used to compare steady D values for Ca =0.24 and De = 0.4.
The number of degrees of freedom (DOF) is provided in Table 2.
Periodicboundaryconditions for u, G and S were imposed, connect-
ing the inflow and outflow boundary. With the above conditions,
our result is similar to that of Chinyoka et al. [68]. Other solu-
tions for NN, NV and VN cases are also compared with Ref. [68]
in Table 3.
3.1.4.2. Effect of viscoelasticity (Ca= 0.1 and Ca = 0.5). In order to
investigate the effect of fluid elasticity on drop deformation,
the simulations were conducted at Ca = 0.1 with the UC2 mesh
(Fig. 4(d)). Calculated D values for NV and VN agree with Yue et al.
[22] as shown in Fig. 7(a). For instance, D was 0.122 (NV) and 0.103
(VN) for Dei = 2 in Ref. [22], while it was 0.122 (NV) and 0.109 (VN),
respectively in our simulation. Comparing NV with VN in Fig. 7,
an increase of medium elasticity (Dem) results in an increase of D
and decrease of. Comparing VV with NV, D was predicted some-what lower than NV at all De, while was almost constant, which
implies that the medium elasticity plays a dominant role in drop
orientation and the drop elasticity tends to resist drop deformation
at small Ca.
In order to understand the drop dynamics in shear flow more
quantitatively, the extensional componentof the conformation ten-
sor was separated from drop orientation through C = m1Cm1. We
Fig. 12. Drop deformation along the position of moving drop (=100, Ca = 0.01,
d = 50): (a) drop shapes and D; (b) drop size (L, B).
8/6/2019 Numerical Study on the Effect of Viscoelasticity on Drop Deformation
10/14
C. Chung et al. / J. Non-Newtonian Fluid Mech. 155 (2008) 8093 89
also separated the shear component of the conformation tensor
from through C = m1Cm2, where m2 is the unit vector towardsthe minor principal direction of the deformed drop as in Fig. 3(b).
C is closely related to the drop orientation. For instance, a positive
shear stress in the upward flow region means the stress towards
right (positive x-direction), while a negative shear stress in the
upward flow region means the stress towards left (negative x-
direction). With these concepts, we denote the direction of the
stresswith respect tothe flow directionas an arrow in Fig.8. Forthe
NV case as in Fig. 8(a), the contribution ofC near the drop makesthe drop align to the shear flow, i.e., decreases. On the contrary,for VNcase asin Fig. 8(b), the stress exerts to the opposite direction
(denoted as white arrows) against the flow direction since the fluid
inside the drop flows clockwise. Therefore, the medium elasticity
plays a role to rotate the drop more to the flow direction, whereas
the drop elasticity resists the orientation to the flow direction. For
NN case, it was found thatthe viscous shear stress of both drop and
medium tended to reduce the orientation angle, andconsequently decreases as Ca increases as in Fig. 5(b).
In case of Ca =0.5 (Fig. 9), a different pattern of D and was
observed compared to Ca =0.1 (Fig. 7). For all cases (NV, VN, VV),
D was less than that of NN case. The only similarity between
Ca = 0.5 and Ca = 0.1 is the VN case, i.e., slight decrease of D and
slight increase of as De increases, which is qualitatively inagreement with Ref. [25] for Ca =0.6 even though Ca is not the
same.
3.2. Drop deformation in 5:1:5 planar contraction/expansion flow
3.2.1. Problem descriptionHere, we consider drop deformation in contraction/expansion
flow as a typical microfluidicproblem. Schematic diagram for5:1:5
contraction/expansion channel is presented in Fig. 10. The width of
the narrow channel isw, and the radius of curvature at the contrac-tion/expansioncorner was set as r= w/2. There are two reasons forrounding the corner: first to prevent the interface from going out
of the corner during temporal integration of the interface and sec-
ond to avoid the ambiguity due to stress singularity at the abrupt
corner. The length of the narrow and expansion channel was long
enoughcompared tothe dropsize. Initiallya dropwith the diameter
d was positioned at the center of the upstream region, namely 5wupward from the entrance of the narrow channel. Fully developed
profiles of velocity and extra stress were imposed at the inlet and
outlet.
Fig.13. Dropshapes in theentrance andexitregion dependingon drop size (= 100, Ca =0.01). (a) d =50,yd/w = 3.50, 5.15, 7.08, (b)d =60,yd/w = 2.73,4.85,7.30, (c)d =70,
yd/w = 2.45, 5.10, 8.15, (d) d =50, yd/w = 13.71, 15.47, 17.14, (e) d =60, yd/w = 13.33, 15.57, 17.40, (f) d =70, yd/w = 12.77, 15.60, 17.73.
8/6/2019 Numerical Study on the Effect of Viscoelasticity on Drop Deformation
11/14
90 C. Chung et al. / J. Non-Newtonian Fluid Mech. 155 (2008) 8093
3.2.2. Parameters
In this problem, we defined the dimensionless numbers with
the characteristic variables in the narrow channel as follows:
Ca =mV
, (19)
Dei = iV
w/2, (20)
=dm
, (21)
wherem denotes the medium viscosity,the surface tension coef-ficient,i the relaxation timeof viscoelastic fluidi (i = d, m)and Vthemean velocity of thefluid in thenarrow channel. Here, theviscosity
ratio of drop to medium was set as = 100 to consider the dynam-ics of more viscous drops [5]. The mesh configuration is provided
in Fig. 11. Center regions in the upstream and downstream and the
narrow channel were refined with structured elements, where the
characteristic mesh size was set xmin/w = 0.05 to resolve dropdynamics as accurately as possible with affordable computation
time. The mesh was composed of 25,010 elements with 101,005
nodes.
In this study, we re-defined the smoothed delta function h(x xp) in place of Eq. (12) to treat the surface tension using the
immersed boundary method [58]:
h(x xp) = dh(x xp)d
h(y yp)
=1
h2
1
|x xp|
h
1
|y yp|
h
(22)
where h =xmin =ymin. In this application, we should considerthe case that the drop interface exists near the wall within a mesh
size, especially in the narrow channel. In order to calculate the sur-
face tension along the wall where the normal direction to the wall
is the x-direction, a half-ranged delta function dh/2
(x xp) should
be substituted for dh
(x xp) in Eq. (22) to avoid a loss of surface
tension transferred out of the domain [69], i.e., using d
h/2(x xp)means that the smoothed region becomes half in the x-direction
and the weighting factor is doubled in the region. For the same
reason, we also applied a modified smoothed delta function as
h/2
(x xp) = dh/2(x xp)dh/2
(y yp) along the rounded wall at
both the contraction and expansion region.
3.2.3. Newtonian drop/Newtonian medium (NN)
We investigated the dynamics of a viscous drop (= 100) forCa = 0.01 to reproduce the experimental conditions as far as pos-
sible [5]. Drop shape and deformation parameter D are presented
in Fig. 12(a) for the drop with d = 50, which is larger than the nar-
row channel width (w = 40). Dimensionless position of themovingdrop, yd/w is designated on the x-coordinate in the graph, where
the region of the narrow channel corresponds to from 5 to 15.The position of the drop is the mean value of both leading and
trailing meniscus ends. The drop exhibits an oblong shape at the
contraction region and a bullet shape in the narrow channel at
steady state. The drop is swelled in the cross-stream direction (x-
direction in Fig. 10) just after a drop escapes from the exit region,
where D is negative since L < B. Surface tension tends to relax the
deformed dropbackto theinitialshapeof thecircle atthe expanded
channel. In Fig. 12(b), L exhibits an overshoot at the contraction
region where the extensional rate shows a sharp peak along the
center line, whereas B is maintained as a constant due to the
confinement effect by channel walls. At the exit region, a reverse
phenomenon is observed such that B is larger than L, which shows
the drop swell to the cross-stream direction. It is to be noted
that L and B decrease below the initial length due to numerical
Fig. 14. Effect of fluid elasticity on deformation parameter along the position of
moving drop (= 100, Ca = 0.01, =0.5, d = 50): (a) effect of drop elasticity (VN); (b)
effect of medium elasticity (NV).
error. We observed a 2.5% reduction of the radius at the final state
in Fig. 12(b). It was confirmed that this error reduces with mesh
refinement.
The shapes of larger drops of d =50, 60, 70 are presented in
Fig. 13 to investigate the effect of drop size. Drop deformation
seems to be insignificant at the entrance region and in the nar-
row channel, i.e., the oblong shape is commonly observed due to
the geometric constraint as in Fig. 13(a)(c). A slight alleviation of
drop swell is observed at exit region as d increases as evidenced
in Fig. 13(d)(f). It might originate from a slight difference of cap-
illary force (ds nds) since a larger drop at exit region shows arelatively smaller interface curvature of the trailing meniscus.
3.2.4. Effect of viscoelasticity
The drop deformation parameter was also studied for a vis-
coelastic drop in a Newtonian medium (VN) and for a Newtonian
drop in a viscoelastic medium (NV) as in Fig. 14. Increase of D in
the narrow channel is clearly observed as the elasticity of medium
(Dem) increases in Fig. 14(b), whereas the effect of drop elastic-
ity (Ded) is relatively small on drop deformation in Fig. 14(a). D
increases at lower Dem = 0.2 and 0.4, and the drop swells to the
cross-stream direction at the exit region (at around 15 on the x-
coordinate in Fig. 14(a)). For the VN case, the drop-swell is clearly
observed and the recovery process to the initial shape is faster than
for the Newtonian drop, while the drop-swell is remarkably alle-
viated for the NV case. The drop shape and velocity vectors in the
8/6/2019 Numerical Study on the Effect of Viscoelasticity on Drop Deformation
12/14
C. Chung et al. / J. Non-Newtonian Fluid Mech. 155 (2008) 8093 91
Fig. 15. Drop shape with velocity vectors in the narrow channel: (a) NN, f/0.5w = 0.103; (b) VN (Ded =2), f/0.5w = 0.113; (c) NV (Dem = 0.4),f/0.5w = 0.186.
narrow channel are provided in Fig. 15. For the NV case, the drop
shows an ellipsoid-like shape with symmetric curvatures of lead-
ing and trailing menisci as shown in Fig. 15(c) which leads to a
D increase, whereas a bullet shape with a relatively flat trailing
meniscus is developed for the NN and VN cases. It was reported
that there is a film of liquid of uniform thickness between drop
interface and tube wall, which is in a region of uniform pressure
with no tangential stress on the interface [70]. Here, we estimated
the dimensionless film thickness between dropinterfaceand chan-
nel wall as f/0.5w = 0.103 (NN), 0.113 (VN), 0.186 (NV), where
Fig. 16. Contour of normal stress difference near drop in the narrow channel (=100, Ca = 0.01, =0.5, d = 50). (a) yy xx (VN, Ded = 2); (b) yy xx (NV, Dem =0.4).
8/6/2019 Numerical Study on the Effect of Viscoelasticity on Drop Deformation
13/14
92 C. Chung et al. / J. Non-Newtonian Fluid Mech. 155 (2008) 8093
the thickest one is the NV case. To understand the differences in
drop shape and film thickness, we investigated the contours of
normal stress difference for the VN and NV cases in Fig. 16. For
VN in Fig. 16(a), the normal stress difference develops inside the
drop near the wall, which is considered to help drop swell to the
cross-stream direction at the exit region since the normal stress
difference is exerting to the wall direction during the confinement
in the narrow channel. For the NV case, the normal stress differ-
ence is highly developed in the film region in Fig. 16(b), which is
regarded as a primary factor for the increase of drop deformation
in the narrow channel and for alleviation of drop swell at the exit
region. Therefore, it might be concluded that the normal stress dif-
ference strongly affects drop dynamics including drop shape, film
thickness in the narrow channel and drop swell at the exit region.
4. Concluding remarks
In this paper, we investigate viscoelastic effects on drop dynam-
ics utilizing a finite element-front tracking method (FE-FTM) with
stabilizing schemes as DEVSS-G/SUPG/ML. First, drop deformation
in shear flow is considered. We compare our steady solutions with
previous results [19,21,24,6466] for Newtonian system (NN) to
verify our algorithm, where we present steady solutions up to
Ca = 1. We also validate our viscoelastic solutions in comparison
with previous results [22,25,68]. We explain drop dynamics with
extensional components of Newtonian viscous fluid (N) and ofpolymer conformation tensor (C), and shear component of poly-mer conformation tensor (C) based on drop orientation. Since the
shear component is considered to relate to the drop rotation, a vis-
coelastic mediumtends to rotate a drop to the flow direction, while
a viscoelastic drop tends to resist the rotation to the shear flow.
We also treat drop deformation in 5:1:5 planar contrac-
tion/expansion flow which is a typically confronted problem in
microfluidics. We confirmthat FE-FTM describes physical phenom-
ena qualitatively by presenting the effect of initial drop size in a
Newtonian system (= 100, Ca = 0.01). In the NN and VN cases, the
drop shows two distinctive features: a bullet shape in the nar-row channel and a drop swell to the cross-stream direction at
the exit region. Especially, for the VN case, the normal stress dif-
ference developed inside the drop contributes to the drop swell
at the exit region. Meanwhile, in the NV case, the drop is more
deformed by the normal stress difference of the medium in the
narrow channel. Normal stress differences developed in the film
region between drop interface and channel wall seems to make a
drop into an ellipsoid-like shape instead of a bullet shape, which
induces D enhancement in the narrow channel. Alleviation of the
drop swell to the cross-stream direction at the exit region is also
attributed to the normal stress difference of the medium. These
results tell us that viscoelastic fluids canbe usefulto control shapes
of microparticles in microfluidics.
It is expected that knowledge as stated above on the dropdynamics in shear flow or in mixed flow would be helpful to
manipulate drops in microfluidics. In both cases, the normal stress
difference near the drop plays a significant role for drop shapes
and deformation. We infer that viscoelastic drop is unlikely to be
deformed easily while viscoelastic medium enhances drop defor-
mation since drop is more deformed in the NV case than the VN
case. Wehopethatthis studywould give physicalinsightsto control
the drop shapes for various applications.
Acknowledgements
This work was supported by the Korea Research Foundation
Grant funded by the Korean Government (MOEHRD) (KRF-2005-
213-D00033). The authors wish to acknowledge the National
Research Laboratory Fund (M1030000 0159) of the Ministry of
Science and Technology in Korea. The authors would like to
acknowledge the support from KISTI Supercomputing Center (KSC-
2007-S00-1021).
References
[1] J.A.Champion, Y.K. Katare, S. Mitragotri,Particleshape:a newdesignparameter
for micro- and nanoscale drug delivery carriers, J. Control. Release 121 (2007)39.
[2] M. Seo, Z.H. Nie, S.Q. Xu, P.C. Lewis, E. Kumacheva, Microfluidics: fromdynamic lattices to periodic arrays of polymer disks, Langmuir 21 (2005)47734775.
[3] D.C. Pregibon, M. Toner, P.S. Doyle, Multifunctional encoded particles for high-throughput biomolecule analysis, Science 315 (2007) 13931396.
[4] D. Dendukuri, D.C. Pregibon, J. Collins, T.A. Hatton, P.S. Doyle, Continuous-flowlithography for high-throughput microparticle synthesis, Nat. Mater. 5 (2006)365369.
[5] D. Dendukuri, K. Tsoi, T.A. Hatton, P.S. Doyle, Controlled synthesis of nonspher-ical microparticles using microfluidics, Langmuir 21 (2005) 21132116.
[6] J. Collins, A.P. Lee, Control of serial microfluidic droplet size gradient by step-wise ramping of flow rates, Microfluid Nanofluid 3 (2007) 1925.
[7] Y.C. Tan, V. Cristini, A.P. Lee, Monodispersed microfluidic droplet genera-tion by shear focusing microfluidic device, Sens. Actuators B 114 (2006)350356.
[8] Y.K. Cheung, B.M. Gillette, M. Zhong, S. Ramcharan, S.K. Sia, Direct patterning
of composite biocompatible microstructures using microfluidics, Lab Chip 7(2007) 574579.[9] L.A. Utracki, Polymer Alloys and Blends, Hanser, Munich, 1989.
[10] H.A. Stone, Dynamics of drop deformation and breakup in viscous fluids, Ann.Rev. Fluid Mech. 26 (1994) 65102.
[11] J.M.Rallison,The deformationof smallviscous drops andbubblesin shear flows,Ann. Rev. Fluid Mech. 16 (1984) 4566.
[12] J.S. Hong, K.H. Ahn, S.J. Lee, Strain hardening behavior of polymer blends withfibril morphology, Rheol. Acta 45 (2005) 202208.
[13] C.D. Han, Multiphase Flow in Polymer Processing, Academic press, New York,1981.
[14] K. Verhulst, P. Moldenaers, M. Minale, Drop shape dynamics of a Newtoniandrop in a non-Newtonian matrix during transient and steady shear flow, J.Rheol. 51 (2007) 261273.
[15] W. Yu, C.X. Zhou, M. Bousmina,Theory of morphology evolution in mixtures ofviscoelastic immiscible components, J. Rheol. 49 (2005) 215236.
[16] M. Minale, Deformation of a non-Newtonian ellipsoidal drop in a non-Newtonian matrix: extension of Maffettone-Minale model, J. Non-Newton.Fluid Mech. 123 (200 4) 151160.
[17] P.L. Maffettone, F. Greco, Ellipsoidal drop model for single drop dynamics withnon-Newtonian fluids, J. Rheol. 48 (2004) 83100.
[18] F. Greco, Drop deformation for non-Newtonian fluids in slow flows, J. Non-Newton. Fluid Mech. 107 (2002) 111131.
[19] G.I. Taylor, The formation of emulsions in definable fields of flow, Proc. R. Soc.Lond. Ser. A 146 (1934) 501523.
[20] D. Khismatullin, Y. Renardy, M. Renardy, Development and implementation ofVOF-PROST for 3D viscoelastic liquidliquid simulations, J. Non-Newton. FluidMech. 140 (2006) 120131.
[21] T.Chinyoka, Y.Y. Renardy, A. Renardy, D.B.Khismatullin, Two-dimensionalstudyof drop deformation under simple shear for Oldroyd-B liquids, J. Non-Newton.Fluid Mech. 130 (2005) 4556.
[22] P.T. Yue, J.J. Feng, C. Liu, J. Shen, Viscoelastic effects on drop deformation insteady shear, J. Fluid Mech. 540 (2005) 427437.
[23] P. Yue, J.J.Feng, C. Liu,J. Shen, Transient dropdeformationupon startup of shearin viscoelastic fluids, Phys. Fluids 17 (2005) 12310111231016.
[24] P.T. Yue, J.J. Feng, C. Liu, J. Shen, A diffuse-interface method for simulating two-phase flows of complex fluids, J. Fluid Mech. 515 (2004) 293317.
[25] S.B. Pillapakkam, P. Singh, A level-set method for computing solutions to vis-coelastic two-phase flow, J. Comput. Phys. 174 (2001) 552578.
[26] R.W. Hooper, V.F. de Almeida, C.W. Macosko, J.J. Derby, Transient polymericdrop extension and retraction in uniaxial extensional flows, J. Non-Newton.Fluid Mech. 98 (2001) 141168.
[27] R. Hooper, M. Toose, C.W. Macosko, J.J. Derby, A comparison of boundary ele-ment and finite element methods for modeling axisymmetric polymeric dropdeformation, Int. J. Numer. Meth. Fluids 37 (20 01) 837864.
[28] K. Sarkar, W.R. Schowalter, Deformation of a two-dimensional viscoelasticdrop at non-zero Reynolds number in time-periodic extensional flows, J. Non-Newton. Fluid Mech. 95 (2000) 315342.
[29] D.A. de Sousa, E.J. Soares, R.S. de Queiroz, R.L. Thompson, Numerical investiga-tionon gas-displacement of a shear-thinningliquid anda visco-plastic materialin capillary tubes, J. Non-Newton. Fluid Mech. 144 (2007) 149159.
[30] C. Malaga, J.M. Rallison, A rising bubble in a polymer solution, J. Non-Newton.Fluid Mech. 141 (2007) 5978.
[31] M.R. Davidson, J.J. Cooper-White, Pendant drop formation of shear-thinningand yield stress fluids, Appl. Math. Model. 30 (2006) 13921405.
[32] M.R. Davidson, D.J.E. Harvie, J.J. Cooper-White, Simulations of pendant drop
formation of a viscoelastic liquid, Korea-Aust. Rheol. J. 18 (2006) 4149.
8/6/2019 Numerical Study on the Effect of Viscoelasticity on Drop Deformation
14/14
C. Chung et al. / J. Non-Newtonian Fluid Mech. 155 (2008) 8093 93
[33] D.J.E. Harvie, M.R. Davidson, J.J. Cooper-White, M. Rudman, A parametric studyof droplet deformation through a microfluidic contraction: shear thinning liq-uids, Int. J. Multiphase Flow 33 (2007) 545556.
[34] R.E. Khayat, A. Luciani, L.A. Utracki, F. Godbille, J. Picot, Influence of shear andelongation on drop deformation in convergentdivergent flows, Int. J. Multi-phase Flow 26 (2000) 1744.
[35] D.J.E. Harvie, M.R. Davidson, J.J. Cooper-White, M. Rudman, A parametric studyof droplet deformation through a microfluidic contraction: low viscosity New-tonian droplets, Chem. Eng. Sci. 61 (2006) 51495158.
[36] R.E. Khayat, A. Luciani, L.A. Utracki, Boundary-element analysis of planar drop
deformation in confined flow. Part1. Newtonian fluids, Eng.Anal. Bound. Elem.19 (1997) 279289.
[37] P.L. Maffettone, F. Greco, M. Simeone, S. Guido, Analysis of start-up dynamicsof a single drop through an ellipsoidal drop model for non-Newtonian fluids, J.Non-Newton. Fluid Mech. 126 (2005) 145151.
[38] S. Guido, M. Simeone, F. Greco, Deformation of a Newtonian drop in a vis-coelasticmatrix under steady shear flowexperimentalvalidation of slowflowtheory, J. Non-Newton. Fluid Mech. 114 (2003) 6582.
[39] W. Lerdwijitjarud, R.G. Larson, A. Sirivat, M.J. Solomon, Influence of weakelasticity of dispersed phase on droplet behavior in sheared polybutadi-ene/poly(dimethyl siloxane) blends, J. Rheol. 47 (2003) 3758.
[40] F.Mighri, P.J.Carreau, A. Ajji, Influenceof elastic propertieson dropdeformationand breakup in shear flow, J. Rheol. 42 (1998) 14771490.
[41] N. Aggarwal, K. Sarkar, Deformation and breakup of a viscoelastic drop in aNewtonian matrix under steady shear, J. Fluid Mech. 584 (2007) 121.
[42] T.M. Squires, S.R. Quake, Microfluidics: fluid physics at the nanoliter scale, Rev.Mod. Phys. 77 (2005) 9771026.
[43] V. Cristini, Y.C. Tan, Theory and numerical simulation of droplet dynamics in
complex flowsa review, Lab Chip 4 (2004) 257264.[44] S.O. Unverdi, G. Tryggvason, A front-tracking method for viscous, incompress-ible, multi-fluid flows, J. Comput. Phys. 100 (1992) 2537.
[45] J.A. Sethian, P. Smereka, Level set methods for fluid interfaces, Annu. Rev. FluidMech. 35 (2003) 341372.
[46] S.Shin, D. Juric, Modelingthree-dimensional multiphase flowusinga level con-tour reconstruction method for front tracking without connectivity, J. Comput.Phys. 180 (2002) 427470.
[47] K. Sarkar,W.R.Schowalter, Deformation of a two-dimensional dropat non-zeroReynolds number in time-periodic extensional flows: numerical simulation, J.Fluid Mech. 436 (2001) 177206.
[48] G. Tryggvason, B. Bunner, A. Esmaeeli,D. Juric, N. Al-Rawahi, W.Tauber,J. Han,S.Nas, Y.J. Jan, A front-tracking method for the computations of multiphase flow,
J. Comput. Phys. 169 (2001) 708759.[49] A.W. Liu,D.E. Bornside,R.C.Armstrong,R.A. Brown,Viscoelasticflowof polymer
solutions around a periodic, linear array of cylinders: comparisons of predic-tions for microstructure and flow fields, J. Non-Newton. Fluid Mech. 77 (1998)153190.
[50] A.N. Brooks, T.J.R. Hughes, Streamline upwind Petrov-Galerkin formulations
for convection dominated flows with particular emphasis on the incom-
pressible Navier-Stokes equations, Comput. Meth. Appl. Mech. Eng. 32 (1982)199259.
[51] M.A. Hulsen, R. Fattal, R. Kupferman, Flow of viscoelastic fluids past a cylinderat high Weissenberg number: stabilized simulations using matrix logarithms,
J. Non-Newton. Fluid Mech. 127 (2005) 2739.[52] R. Fattal, R. Kupferman, Time-dependent simulation of viscoelastic flows at
high Weissenberg number using the log-conformation representation, J. Non-Newton. Fluid Mech. 126 (2005) 2337.
[53] R. Fattal, R. Kupferman, Constitutive laws for the matrix-logarithm of the con-formation tensor, J. Non-Newton. Fluid Mech. 123 (2004) 281285.
[54] J. Ramirez, M. Laso, Size reduction methods for the implicit time-dependentsimulation of micromacro viscoelastic flow problems, J. Non-Newton. FluidMech. 127 (2005) 4149.
[55] J.M. Kim, C. Kim, K.H. Ahn, S.J. Lee, An efficient iterative solver and high-resolution computations of the Oldroyd-B fluid flow past a confined cylinder,
J. Non-Newton. Fluid Mech. 123 (2004) 161173.[56] F.P.T. Baaijens, Mixed finite element methods for viscoelastic flow analysis: a
review, J. Non-Newton. Fluid Mech. 79 (1998) 361385.[57] G.C. Georgiou, M.J. Crochet, The simultaneous use of 4 4 and 2 2 bilinear
stress elements for viscoelastic flows, Comput. Mech. 11 (1993) 341354.[58] R. Mittal, G. Iaccarino,Immersedboundarymethods, Annu. Rev. FluidMech. 37
(2005) 239261.[59] C.S. Peskin, Numerical-analysis of blood-flow in heart, J. Comput. Phys. 25
(1977) 220252.[60] R.S. Millman, G.D. Parker, Loval Curve Theory, Elements of Differential Geome-
try, Prentice-Hall, 1977, pp. 1348, Chapter 2.[61] A. Vananroye, P.J.A. Janssen, P.D. Anderson, P. Van Puyvelde, P. Moldenaers,
Microconfined equiviscous droplet deformation: comparison of experimental
and numerical results, Phys. Fluids. 20 (2008) 013101-1-013101-10.[62] P.J.A. Janssen, P.D. Anderson, Boundary-integral method for drop deformationbetween parallel plates, Phys. Fluids. 19 (2007) 043602-1-043602-11.
[63] V. Sibillo, G. Pasquariello, M. Simeone, V. Cristini, S. Guido, Drop deformationin microconfined shear flow, Phys. Rev. Lett. 97 (2006) 054502-1-054502-4.
[64] H. Zhou, C. Pozrikidis, Theflow of suspensions in channelssingle filesof drops,Phys. Fluids A 5 (1993) 311324.
[65] P.L. Maffettone, M. Minale, Equation of change for ellipsoidal drops in viscousflow, J. Non-Newton. Fluid Mech. 78 (1998) 227241.
[66] C.E.Chaffey, H. Brenner, A second-order theory for shear deformation of drops,J. Colloid Interface Sci. 24 (1967) 258269.
[67] C.W. Macosko, Rheology: Principles, Measurement, and Applications, ViscousLiquid, VCH Publishers Inc., 1994, pp. 65108, Chapter 2.
[68] T. Chinyoka, Numerical simulation of stratified flows and droplet deformationin 2D shear flow of Newtonian and viscoelastic fluids, PhD Thesis, VirginiaPolytechnic Institute and State University, 20 04.
[69] J.Zhang, Thedynamics ofa viscous dropwitha movingcontact line,PhDThesis,Northwestern University, 2003.
[70] F.P. Bretherton, The motion of long bubbles in tubes, J. Fluid Mech. 10 (1961)
166188.