Professional Documents
Culture Documents
Synopsis
The dynamics of single droplets dispersed in a second, immiscible liquid undergoing a controlled
mixture of shear and elongational flow has been studied using a home made eccentric cylinder
device. The model system consists of polydimethyl siloxane droplets in a polyisobutylene
matrix, both Newtonian liquids at room temperature. In continuation of previous work Boonen
et al., Rheol. Acta 48, 359371 2009, the effect of changing the balance of shearing and
elongational flow components and varying viscosity ratio on the deformation and orientation of
the droplets has been systematically investigated under sub-critical flow conditions. The
experimental results obtained from optical microscopy are compared with theoretical predictions of
the phenomenological model by Maffettone and Minale J. Non-Newtonian Fluid Mech. 78,
227241 1998, obtained using the transient form of the model and incorporating a flow type
parameter that accounts for the relative amount of extension in the flow. Overall, a fair agreement
was found between the model predictions and the experimental results for all sub-critical mixed
flows applied and all viscosity ratios investigated here. This work provides an experimental
reference data set which can be used to guide future modeling efforts. 2010 The Society of
I. INTRODUCTION
The behavior of droplet dispersions is of interest for many industrial applications, e.g.,
food emulsions, cosmetics, pharmaceuticals, and polymer blending. An understanding of
the droplet dynamics in such systems helps explaining the rheological behavior of the
flowing emulsions, such as viscoelasticity even if the components are Newtonian and a
shear-dependent viscosity. Conversely, from the rheology of the dispersion substantial
information about the morphology development can be obtained e.g., Vinckier et al.
1996. Since the pioneering work of Taylor 1932, 1934, a large number of experimental and theoretical studies on droplet dispersions have been carried out. These are well
described in several recent reviews Guido and Greco 2004; Ottino et al. 1999; Stone
1994; Tucker and Moldenaers 2002. Most studies have focused on the structure
development in either purely elongational flow or simple shear flow. However, since most
0148-6055/2010/546/1285/22/$30.00
1285
1286
FIG. 1. a Schematic of eccentric cylinder device. b Sketch of top view of eccentric cylinder geometry.
1287
0 23 C
Material
Pa s
p = PDMS / PIB
23 C
Activation energy Ea
kJ/mole
PIB
Glissopal 1300
PDMS
Rhodorsil 47V60.000
PDMS
Rhodorsil 47V500.000
PDMS
Rhodorsil 47V5000
51.6
Matrix
60.1
60.4
1.17
15.4
501.6
9.72
15.3
0.10
15.2
5.10
nents, only three dimensionless parameters play a role for the dynamics in slow flows: the
viscosity ratio p = d / m, where d is the viscosity of the drop fluid and m is the
viscosity of the matrix, the capillary number Ca which represents the ratio of viscous to
interfacial tension stresses, and the flow type which can be represented by the flow type
parameter see Sec. II. Ca can be defined as mR0E / , where R0 is the radius of the
undistorted spherical droplet, is the interfacial tension, and E is the flow intensity
which equals 4 2 + 2 for 2D flows. In Boonen et al. 2009, the experimental results for
a viscosity ratio of p = O1 have been compared with predictions obtained using the
transient form of the model of Maffettone and Minale 1998 adapted to complex flows
see also Sec. II. Under sub-critical flow conditions, good agreement was found between
model predictions and experimental data, providing a quantitative assessment of drop
shape predictions in controlled complex flows. In the present study, a systematic investigation is made of the effect of viscosity ratio on the droplet dynamics for a wide range
of viscosity ratios ranging from 0.1 to 10. In addition, we will explore more extreme
flow conditions with an enhanced extensional contribution as compared to the previous
study Boonen et al. 2009.
II. EXPERIMENT
A. Materials
The materials used in this work are polyisobutylene PIB Glissopal from BASF
and polydimethyl siloxane PDMS Rhodorsil from Rhodia. Since Guido et al. 1999
reported that PIB is slightly soluble in PDMS, PIB is chosen as the matrix phase. In order
to vary the viscosity ratio, three different grades of PDMS have been selected as the
droplet phase. Table I gives the zero shear viscosities 0 and the corresponding viscosity
ratios p at 23 C. All fluids are transparent at room temperature and exhibit Newtonian
behavior over the range of strain rates investigated here. The interfacial tension for the
PIB/PDMS system was measured by three methods small deformation theory, droplet
retraction measurements, and pendant drop technique and was found to be
2.8 0.1 mN/ m. Moreover, this value was measured to be independent of the molecular
weight for the three grades of PDMS used here. All experiments were performed at
ambient temperature 23 C; as the viscosity of PIB is very sensitive to temperature,
the temperature of the sample was directly monitored by immersing a fine thermocouple
needle in the continuous phase. The viscosities m and d, and the resulting viscosity
ratio p could be back-calculated using an Arrhenius equation.
1288
The density difference between both polymers is quite small about 0.08 g / cm3 so
that in combination with the high viscosity of the matrix, gravitational effects can be
neglected. Furthermore, the difference in refractive indices is high enough to have good
contrast for observations by optical microscopy.
B. Methods
The experiments are performed using a home built ECD which is equipped with a
microscope and camera system to enable visualization of droplet dynamics in controlled
complex flows. A detailed discussion of the experimental set-up has been given in
Boonen et al. 2009. In Fig. 1a, a schematic of the basic ECD is shown. In brief, the
apparatus allows applying controlled mixed flow conditions, i.e., a combination of
shear and elongational components, in the narrowing and expanding areas of the gap
between the rotating, eccentrically positioned cylinders. Figure 1b shows a 2D sketch of
the geometry, consisting of an inner and an outer cylinder with radii Ri and Ru and
angular velocities i and u, respectively. The axes of these cylinders can be displaced by
a certain distance e, the eccentricity. The eccentricity ratio X of the system is then defined
as e / Ru Ri. This is an important, adjustable parameter that influences the relative, as
well as the absolute magnitudes of shear and elongational strain rates along the streamlines of the flow field. In addition, the flow configuration, i.e., inner cylinder rotating,
outer cylinder rotating, counter-rotating, or co-rotating, also has a profound effect on the
type of mixed flow obtained see further. For the experiments conducted in this study,
the ECD has been used in two configurations: configuration A in which only the outer
cylinder is rotating with a constant speed, and configuration B where both cylinders are
co-rotating with the ratio of outer to inner cylinder velocity u / i equal to 1.5. For both
configurations the eccentricity ratio X is set to 0.2. In configuration A, a shear dominated
type of flow is obtained, while configuration B produces a flow with an enhanced extensional contribution, which allows exploring more extreme flow conditions.
Single PDMS droplets with initial radii R0 in the range of 300 700 m were introduced in the matrix material by using a home made injection system. The spherical drop
is injected in the widest part of the gap and positioned so that its center is at a radial
distance around 30 mm half the clearance from the axis of the inner cylinder. Upon
start-up of the flow, the droplet deformation and orientation is visualized during several
revolutions using optical microscopy Olympus SZ61-TR stereo microscope. The microscope, equipped with a CCD camera Basler A301f, is mounted on top of the flow cell
so that images are captured in the velocity-velocity gradient plane see Fig. 2 using the
Streampix Digital Video Recording Software NORPIX. Due to the limited field of view
of the microscope and difficulties in illumination, images of the deforming drop could, in
a single experiment, only be captured in the first two I and II or the last two III and IV
quadrants of the Cartesian coordinate system, defined in Fig. 1b. In order to obtain the
shape of the droplet, an ellipse is fitted to the recorded images after performing basic
image processing, including a convolution operation and threshold using IMAGEJ software
for Windows Rasband 1997. Throughout the experiments, the flow intensity is kept
below the critical conditions for breakup, so that no significant deviations from the
ellipsoidal shape are present. Using this approach, it is then possible to obtain the deformation and orientation of the droplets, i.e., the length L of the long axis, the length B of
the short axis, and the angle between the long axis and the flow direction, as shown in
Fig. 2.
The experimental results are compared with the predictions of the MaffetoneMinale
model Maffettone and Minale 1998, 1999, which has been applied to complex flows
1289
Boonen et al. 2009. This phenomenological model assumes that during flow the drop
shape is ellipsoidal at all times and that its volume is preserved. The drop shape can
therefore be described by a symmetric, positive definite second order tensor S, which
eigenvalues represent the square semi-axes of the ellipsoid. The evolution equation of S,
resulting from the competing actions of the hydrodynamic forces and the interfacial
tension, is given by Maffettone and Minale 1998
f MM
dS
W S S W = 1 pS gMMSI + f MM
2 p,CaD S + S D,
dt
where the characteristic time equals mR0 / , I is the second rank unit tensor, and D
and W are the deformation rate and vorticity tensor of the flow field. The characteristic
time will be used to define a dimensionless time t = t / in which t is the time since the
start-up of the flow. The scalar function g MM is required to preserve drop volume, while
the functions f 1MM and f 2MM have been determined to recover the asymptotic analytic
limits Maffettone and Minale 1998; Taylor 1934:
f MM
=
1
f 02 =
5
,
3 + 2p
40p + 1
3 + 2p16 + 19p
f c2 =
1
3Ca2
,
2 + 6Ca2+ 1 + p2
f MM
= f 02 + f c2 ,
2
gMM =
3IIIS
.
IIS
Here, IIS is the second scalar invariant of S, and and are the small positive numbers
which, for all practical purposes when Ca is not too large and p is far from infinity, are
set to zero. The flow field generated in the ECD can be represented by a deformation rate
and vorticity tensor of the form
1290
D=G
1
0
2
1
2
W=G
1
0
2
1 +
2
Here, the flow type parameter expresses the relative amount of elongation present in
the flow and is defined as Feigl et al. 2003
=
the sum of absolute shear and elongation rates and 1 1. Mixed flow
with G
t
conditions between eccentric cylinders can thus be represented by the flow strength G
and the flow parameter t which in general vary in time. In this work, we will then refer
to the MM model as the model predictions according to the evolution equation, Eq. 1,
with the parameters given by Eq. 2 and with the flow field represented by Eq. 3.
III. RESULTS AND DISCUSSION
In Sec. III A and III B, the experimental results on droplet deformation obtained in
two types of sub-critical complex flows with different relative contributions for shear and
elongation, and for three viscosity ratios, are presented and discussed. These results will
be compared to the model predictions of the MM model as described above to help
elucidate the effect of flow type and viscosity ratio in these mixed flow conditions.
A. Effect of shear/elongation balance
First, the dynamics of droplets with a constant viscosity ratio of about 1.2 in two
different types of mixed flow are considered: shear dominated flow in configuration A of
the ECD Figs. 3 and 4 and a flow field with an enhanced extensional component
obtained in configuration B Figs. 5 and 6. In both cases, the transient droplet deformation is observed in the first two quadrants of the flow field Fig. 1b for a number of
revolutions during start-up of the flow. Furthermore, by varying the velocities of the
cylinders and changing the initial droplet radius R0, a range of capillary numbers can be
explored.
1. Configuration A: Shear dominated flow
Due to the motion through the ECD, the drop is subjected to a periodic flow field.
Hence a new characteristic time-scale, the oscillation period of the imposed flow, comes
into play. The droplet dynamics as reported in this paper are not influenced by the
presence of this additional time-scale. For instance, in the experiments presented in Fig.
3, the droplet relaxation time is about 0.7 s, whereas the ECD rotation period is 250 s.
These time-scales are well separated indicating that the droplet can rapidly respond to the
applied deformation and that the droplet shape is fully determined by the deformation at
that instant.
For the shear dominated flow of configuration A, the results at a viscosity ratio of 1.2
have been discussed in detail elsewhere Boonen et al. 2009. Typical profiles for the
strain rates and the corresponding flow parameters Cat and t for a PDMS drop with
a representative initial radius of 375 m are shown in Fig. 3. The shear rate is seen to
1291
FIG. 3. Example of flow parameters for configuration A with i = 0, u = 0.025 rad/ s and eccentricity ratio
X = 0.2, along streamline starting at x = 30 mm from the axis of the inner cylinder: a strain rate profiles, shear
and elongation ; b and c evolution of Ca and for a PDMS drop with initial radius of 375 m; d
example of a deforming droplet at different points along the trajectory.
reach a sharp maximum when the gap is at its narrowest point C and displays a broad
minimum in the widest part of the gap point A. These features also show up in the
Ca-profile shown in Fig. 3b. Here, the flow conditions are mostly shear dominated,
which is also evident from the values of the flow type parameter in Fig. 3c. The
average amount of elongation avg,III over the first two quadrants III in Fig. 1b is
1292
FIG. 4. Droplet dynamics as a function of dimensionless time t for the flow conditions of Fig. 3 and for a
viscosity ratio of p 1.2. The symbols represent the experimental results, whereas the lines are the predictions
according to the MaffettoneMinale model.
For configuration B, in which both cylinders are co-rotating with the ratio u / i set to
a value of 1.5, the flow field comprises an enhanced extensional contribution. Here, the
flow conditions are drastically different from the one previously discussed, as shown in
Fig. 5. In this case, the shear rate see Fig. 5a displays a minimum peak at the point
where the gap is the narrowest point C and a broad maximum in the widest part of the
gap point A. For this specific flow configuration, in some parts of the gap, the velocity
profiles in the radial direction appear the show a nonmonotic behavior reaching an extremum value. This could explain this counterintuitive finding that the shear rate is
1293
FIG. 5. Example of flow parameters for configuration B with i = 0.02, u = 0.03 rad/ s and eccentricity ratio
X = 0.2, along streamline starting at x = 35 mm from the axis of the inner cylinder: a strain rate profiles, shear
and elongation ; b and c evolution of Ca and for a PDMS drop with initial radius of 500 m.
FIG. 6. Droplet dynamics as a function of dimensionless time t for the flow conditions of Fig. 5 and for a
viscosity ratio of p 1.2. The symbols represent the experimental results, whereas the lines are the predictions
according to the MaffettoneMinale model.
1294
maximum in the widest part of the gap. The elongation rate on the other hand shows the
same type of symmetric profile as in Fig. 3a, being positive in the converging part and
negative in the diverging part of the gap. The main difference as compared to the previous configuration is that the extensional strain rate has the same order of magnitude as the
shear rate during a substantial part of the profile. This results in relatively large values for
the flow type parameter shown in Fig. 5c. This parameter follows the same symmetric profile as in Fig. 3c, but with significantly higher maximum values of about
0.50 in quadrants II and III. As a consequence, the average amount of elongation
avg,III in the first two quadrants is now increased to approximately 30% of the sum of
strain rates, as compared to only 10% for the previous configuration. The profile for the
capillary number Ca also differs substantially from that in Fig. 3, as can be seen in Fig.
5b. The Ca-number is more or less constant in the first quadrant of the flow field,
subsequently drops rapidly to reach a sharp minimum at the end of quadrant II point C,
and symmetrically continues in quadrant III and IV. It will be verified now to what extent
the drastically changed flow field will affect the droplet dynamics.
Figure 6 shows the results for droplet deformation observed in quadrants III during
start-up, as well as the model predictions according to the MM model. It can be seen that
the deformation parameters L / 2R0 and B / 2R0, and the orientation angle again seem to
follow an oscillating pattern for subsequent revolutions, in nice agreement with the model
predictions. In this case, however, the deformation strongly relaxes near the end of the
second quadrant as shown by the decrease and increase especially for the lower Ca
values of the major axis L and the minor axis B of the droplet, respectively. Furthermore,
the orientation angle starts to increase at the end of the second quadrant, indicating the
droplet rotates away from the velocity direction. This decreased deformation and orientation corresponds to the steep decline of the capillary number Ca at the end of quadrant
II see Fig. 5b. As the Ca-number drops, the driving force for deformation and orientation of the droplets decreases, and due to the action of the interfacial tension the droplet
relaxes to a less deformed and less oriented state. For the orientation angle, the model
predictions even indicate for this range of Ca-numbers that the droplet rotates past 45,
which is the asymptotic limit for small Ca in simple shear flow Taylor 1934. This
peculiar effect cannot be explained entirely by the evolution of Ca. To investigate this
phenomenon more thoroughly, experiments were also performed in the third and fourth
quadrants of the flow field, the droplet still being injected at point A.
The results are shown in Fig. 7; the relaxation of the droplet is clearly confirmed in the
evolutions of L / 2R0 and B / 2R0 depicted in Figs. 7a and 7b. In addition, the orientation angle quickly rises from the velocity direction toward the perpendicular direction to
a maximum value above 45, as predicted by the MM model. This effect can be attributed
to the local change in flow type the droplet experiences in this flow field. At the end of
the second quadrant the elongation rate changes sign and becomes negative compression
flow, while the shear rate remains quite low for some time see Fig. 5a. As a consequence a local flow field with a large, negative -value exists for some time. This tends
to rotate the drop toward the 90 limit, resulting in a maximum orientation angle above
45. In summary, the balance of elongational and shear components in the flow field
clearly affects droplet dynamics in transient mixed flow conditions. Furthermore, it has
been shown that this effect can be correctly predicted by the MM model.
B. Effect of viscosity ratio
It is known that the viscosity ratio is a crucial parameter for droplet dynamics in
simple flow conditions. Here, we want to investigate the effect of the viscosity ratio p on
1295
FIG. 7. Droplet dynamics as a function of dimensionless time t for the flow conditions of Fig. 5 and a
viscosity ratio of p 1.2, with observations made in quadrants III and IV of the flow field. The symbols
represent the experimental results, whereas the lines are the predictions according to the MaffettoneMinale
model.
droplet dynamics in mixed flow conditions. Hereto, droplets with viscosity ratios of
O10 and O0.1 have been observed during the type of complex flow obtained in
configuration B of the ECD with enhanced extensional contribution see Fig. 5.
1. High viscosity ratio drops
Figure 8 shows the droplet dynamics for drops with a high viscosity ratio of p 10.
Note that observations have only been made in the first two quadrants of the flow field,
and results are for low to intermediate Ca higher Ca-numbers will be discussed further
on. Qualitatively, the same behavior is observed as for a viscosity ratio around 1.2 see
Fig. 6. The deformation and orientation increase in the first part of the flow field and
start to relax again near the end of the second quadrant, when the capillary number
quickly drops to a lower value see Fig. 5. It seems though that the relaxation of the
droplet starts somewhat later for this high viscosity ratio case as compared to p 1.2.
This is more clearly illustrated in Fig. 9 where the droplet dynamics for p 1.2 and p
10 are compared for about the same average Ca-number. As expected, the high viscosity ratio drops are less deformed lower L, higher B. They are also slightly more
oriented toward to the flow direction lower . Furthermore, nice agreement is obtained
between the model predictions and the experimental results for the low to intermediate
Ca-range for these high viscosity ratio drops.
For higher Ca, as depicted in Fig. 10, deviations with respect to the predictions of the
MM model start to appear for the major and minor axes of the drop, while the predictions
for the angles are still reasonable. Furthermore, for these high capillary numbers the
effect of p on the orientation angle becomes clearer, as shown in Fig. 11. Here, the
experimental results for the orientation of droplets with p around 10 and 1.2 are com-
1296
FIG. 8. Droplet dynamics as a function of dimensionless time t for the flow conditions of Fig. 5 for a viscosity
ratio of p 10 and intermediate capillary numbers Ca. The symbols represent the experimental results, whereas
the lines are the predictions according to the MaffettoneMinale model.
pared for an average capillary number of about 0.33. A high viscosity ratio obviously
promotes orientation toward the flow direction, something that is also seen in simple
shear flow e.g., Rumscheidt and Mason 1961. If we apply even higher Ca-numbers, as
the ones shown in Fig. 12, quantitative deviations between model predictions and experimental results grow even larger for L / 2R0 and B / 2R0, while the angle predictions still
coincide with the experimental data very well. Qualitatively, the predictions for the major
and minor axes still describe the experimental results. In addition, Fig. 12 illustrates that
at this high viscosity ratio, the amplitude of the oscillating deformation parameters varies
in time, as also seen in the model predictions. This peculiar transient oscillatory behavior
for high p in mixed flow conditions is similar to the damped oscillatory behavior observed during start-up of simple shear flow for high viscosity ratios e.g., Torza et al.
FIG. 9. Comparison of droplet deformation and orientation for the flow conditions of Fig. 5, for a viscosity
ratio of p 10 and p 1.2, respectively, for the same average capillary number Caavg. The symbols represent
the experimental results, whereas the lines are the predictions according to the MaffettoneMinale model.
1297
FIG. 10. Droplet dynamics as a function of dimensionless time t for the flow conditions of Fig. 5 for a
viscosity ratio of p 10 and high capillary numbers Ca. The symbols represent the experimental results,
whereas the lines are the predictions according to the MaffettoneMinale model.
1972 and is not fully understood at present. Note that in Fig. 12 in addition to the MM
model predictions represented by the solid lines also model predictions according to the
so called Minale model corresponding to the dashed lines are shown. The Minale
model Minale 2004 is an extension of the MaffettoneMinale model to account for
elastic effects in the matrix and/or droplet phase. It has an extra term in the evolution
equation of the droplet shape tensor S with the coefficient f 3. This extra term allows for
oblate droplet configurations, and for instance, enables this model to predict the widening behavior seen for p 1 in transient simple shear flow Cristini et al. 2002. In the
FIG. 11. Comparison of droplet orientation for the flow conditions of Fig. 5, for viscosity ratios of p 10 and
p 1.2, respectively, for high capillary numbers Ca.
1298
FIG. 12. Droplet dynamics as a function of dimensionless time t for the flow conditions of Fig. 5 for a
viscosity ratio of p 10 and very high capillary numbers. The symbols represent the experimental results,
whereas the lines are the predictions according to the MaffettoneMinale model solid line and the Minale
model dashed line.
case of Newtonian components studied here, Minale 2004 showed that the predictions
for steady simple shear flow almost coincide with the MM model over a large range of
Ca and only differ at very high capillary numbers. The evolution equation for S for
Newtonian components is given as
f MM
dS
W S S W = 1 pS gMinaleSI + f Minale
p,CaD S + S D
2
dt
f Minale
S:I
3
p,CaD S S + S S D D S + S D
,
R0
3
where the extra coefficient f 3 depends on the applied Ca and the viscosity ratio p note
that gMinale is again a term to preserve the volume and f 2Minale is a slightly adapted form
of f 2MM using a weighting factor for f 2c, as defined in Eq. 2. In our study at high
viscosity ratio and high capillary number, the extra term in the Minale model seems to
give different predictions as compared to the MM model, as shown in Fig. 12. From this
figure, it can be concluded that the Minale model predictions show somewhat better
agreement with the experimental profiles of L / 2R0 and B / 2R0, although not quite satisfactorily yet. For the orientation angle on the other hand, the predictions of the Minale
model and the MM model are practically on top of each other and agree very well with
the experimental values.
Considering both models in more detail, one might suspect that the discrepancy is
related to the behavior of another almost mutual coefficient of both models: f 2. The
1299
FIG. 13. Effect of parameter on droplet dynamics for the flow conditions of Fig. 5, a viscosity ratio of p
10 and large capillary number Ca. The symbols represent the experimental results, whereas the lines are the
predictions according to the MaffettoneMinale model solid line, the Minale model dashed line, and the
Minale model with = 0.001 dotted line and = 0.01 dash-dotted line.
parameters f 2MM and f 2Minale depend on p as well as Ca and consists of two terms see
Eqs. 2. One term, f 20, is determined to recover the asymptotic limits of Taylor 1934.
The other term, f 2c, is an empirical factor added by Maffettone and Minale to improve the
predictions in simple shear flow and used in the Minale model with a weighting factor of
0.75. The factor f 2c contains two empirical constants and , which for all practical
purposes, when Ca and p are far from infinity, have been set to zero by Maffettone and
Minale. In theory though, these parameters have to be small, positive numbers to be able
to recover the asymptotic limits of p , and of affine behavior for p = 1 and Ca ,
respectively. This could mean that, in our case of high p, a small but nonzero value of the
parameter has to be chosen in order to reflect the behavior of the high viscosity ratio
droplets. In Fig. 13, we have verified the effect of the parameter on the predictions of
droplet dynamics by the Minale model for the high viscosity ratio drops p 10 for a
large Ca. By assigning a small positive value to 103 and 102 in Fig. 13, the
predictions for L / 2R0 and B / 2R0 are indeed improved as compared to the normal MM
and Minale model predictions. For the predictions of the orientation angle, on the other
hand, hardly any effect of is observed. This indicates that it is the high viscosity ratio
effect that causes deviations from the MM model predictions because we are reaching the
asymptotic limit for high p for which a small, nonzero value of the parameter has to be
set in the coefficient f 2MM of the model.
Furthermore, it is known e.g., Grace 1982 that in simple shear flow when p 4 a
steady deformation limit, independent of Ca, exists for large Ca. In that case, the deformation parameter D of the droplet, defined as L B / L + B, tends toward a constant
value equal to 5 / 4p and independent of Ca Taylor 1934. To check if similar behavior can occur in transient mixed flows, the deformation results for several large, but
different Ca-number experiments for p = O10, are compared in Fig. 14. Comparing the
1300
FIG. 14. Limiting behavior of droplet deformation for the flow conditions of Fig. 5, a viscosity ratio of p
10 and large capillary numbers Ca: a Dimensionless droplet axes L / 2R0 and B / 2R0. b Droplet deformation parameter D = L B / L + B. c Average deformation parameter Davg as a function of the number of
cycles.
results for two experiments for subsequent revolutions, the deformation, in terms of L and
B see Fig. 14a, appears to level off to the same value, independent of Ca, similar to
the limiting behavior seen in steady simple shear flow for p 4. In the case of mixed flow
conditions, Fig. 14b demonstrates the evolution of the deformation parameter D, for the
same two experiments shown in Fig. 14a. Here, two additional experiments have been
included with about the same average capillary numbers as the first two ones, to illustrate
the reproducibility. Inspecting Fig. 14b, the deformation indeed seems to level off to a
steady value. To show this more clearly, the average values of the experimental deformation parameter Davg were calculated using t0Dtdt / t for each revolution after
start-up of the flow. In Fig. 14c, Davg is plotted as a function of the number of cycles.
Obviously, apart from the initial start-up transients, the average deformation for all four
experiments appears to go to a constant value, although the observations were only made
for a limited number of cycles. To confirm this steady limiting behavior for high p, the
experiments were repeated for a larger number of revolutions, the results of which are
shown in Fig. 15. In this figure, clear evidence is given for the limiting behavior for high
viscosity ratios p = O10 in mixed flow conditions. Note that for steady mixed flows
with a constant value of the flow type parameter , we can also derive a steady deformation limit for p from the analytical solution of the MaffettoneMinale model see
Appendix. This is represented by the solid line in Fig. 15 for a steady 2D flow with a
constant value of equal to the average value avg present in the first two quadrants of
the flow field. Clearly, the experimental deformation limit is close to the limit for steady
2D flows obtained from the MM model. The existence of a steady deformation limit for
high viscosity ratios in mixed flow conditions was already reported by Bentley and Leal
1986 for steady flows with a constant -value. For instance, they found that for
1301
FIG. 15. Average droplet deformation for a viscosity ratio of p 10 and very high Ca for the flow conditions
of Fig. 5 as function of the number of cycles. The solid line is the deformation limit for p for steady mixed
flows with a constant, positive value of according to the MaffettoneMinale model.
0.36, a limiting deformation exists for p-values larger than about 27. In the present
case, however, this steady limiting behavior is already observed for p 10. Probably, this
difference is related to the transient nature of the mixed flow applied with the eccentric
cylinder device. Nevertheless, the average deformation observed for high p in the transient complex flow of the ECD is approximately the same as for a steady mixed flow with
the same average value of the flow type parameter .
2. Low viscosity ratio drops
In the last part of this study, the effect of a low viscosity ratio on the droplet dynamics
in complex flows was investigated, with p-values of about 0.1 at 23 C see Table I. The
resulting dynamics for these low viscosity ratio drops are shown in Fig. 16for low to
intermediate capillary numbers Ca and in Fig. 17 for high Ca. Again, qualitatively the
same profiles as for p 1.2 are obtained in Fig. 16 compare with Fig. 6; the deformation and orientation increase during the first quadrant of the flow field and subsequently
relax near the end of the second quadrant. As before, good quantitative agreement is
found between experimental results and predictions according to the MM model. Deviations only start to appear for higher Ca, as illustrated in Fig. 17. In this Ca-range we are
near the critical conditions for droplet break-up where the MM model obviously fails. In
fact, for the highest Ca explored here, it was found experimentally that the droplet indeed
breaks up after many revolutions outside the time scale of Fig. 17, as indicated by the
arrows in the L / 2R0-profile of Fig. 17.
To quantify the effect of viscosity ratio in more detail, the results for p 0.1 and p
1.2 are compared in Fig. 18. Note the difference in time-scales which is due not only
to differences in revolution period but also to differences in R0 and m for the different
experiments. These differences only become visible and pronounced at larger t more
revolutions. First, Fig. 18a shows the comparison for intermediate Ca Caavg 0.2;
here, no significant effect of the low viscosity ratio is visible, and the deformation and
orientation are approximately at the same level for both viscosity ratios. In Fig. 18b
results are given for a higher Ca Caavg 0.4, where some differences start to appear.
The low viscosity ratio drops exhibit a somewhat larger deformation e.g., visible in the
evolution of the major axis L and slightly less orientation e.g., a larger maximum value
1302
FIG. 16. Droplet dynamics as a function of dimensionless time t for the flow conditions of Fig. 5, for a
viscosity ratio of p 0.1 and intermediate capillary numbers Ca. The symbols represent the experimental
results, whereas the lines are the predictions according to the MaffettoneMinale model.
FIG. 17. Droplet dynamics as a function of dimensionless time t for the flow conditions of Fig. 5, for a
viscosity ratio of p 0.1 and high capillary numbers Ca. The symbols represent the experimental results,
whereas the lines are the predictions according to the MaffettoneMinale model.
1303
FIG. 18. Comparison of droplet dynamics for the flow conditions of Fig. 5, for viscosity ratios of p 0.1 and
p 1.2, respectively, for a intermediate capillary number Ca, b high capillary number Ca, and c very high
capillary number Ca.
for at the beginning of first quadrant. These observations are also in agreement with
the model predictions according to the MM model not shown in Fig. 18. Finally, Fig.
18c compares the data at p 0.1 and p 1.2 for a large capillary number Caavg
0.5. Clearly, the low viscosity ratio drops are more deformed, especially the major
axis of the droplet L. In addition, they show significantly less orientation toward the flow
direction as evident from the average level of the orientation angle . In summary, the
effect of low viscosity ratio p = O0.1 in transient mixed flow conditions as applied
with the ECD only becomes apparent for high Ca, leading to increased deformation and
decreasing orientation, as compared to the case for p 1.2.
IV. CONCLUSIONS
In this work, we have studied the dynamics of single Newtonian droplets suspended in
an immiscible Newtonian matrix, which is subjected to transient mixed flow conditions.
The controlled complex flows are applied using a home built ECD, as described in
Boonen et al. 2009. By exploring different operation modes for the ECD, and varying
the viscosity ratio p over a range of 2 decades from 0.1 to 10, the effect of shear/
elongation balance and viscosity ratio has been systematically explored. In addition to the
experimental results on droplet dynamics obtained from optical microscopy, model predictions have been used for comparison. To this end, the model of Maffettone and Minale
1998 was used, adapted to complex flows by incorporating a flow type parameter that
accounts for the relative amount of elongation in the flow field.
For the different types of sub-critical flow applied and all viscosity ratios explored in
this work, good agreement is found between the experimental results and model predictions. The effect of varying the balance of shear and elongational components is predicted, as evident from the comparison between two different flows obtained in the ECD,
using two different configurations shear dominated versus enhanced extension. Next,
the experimentally observed effect of a high p = O10 and low p = O0.1 viscosity
ratio, respectively, as compared to the base case of p 1.2, is also in agreement with the
model predictions. Quantitative deviations only start to appear for higher capillary numbers Ca, for p 0.1 and p 1.2 because we are near the critical conditions for droplet
1304
break-up, where the MM model obviously is not valid anymore. For p 10, on the other
hand, a deformation limit exists, independent of the applied Ca. This limiting behavior
can only be incorporated into the MM model by assigning a nonzero value to the empirical factor in the definition of the coefficient f 2MM in the model.
All these findings provide evidence that the rather simple phenomenological of Maffettone and Minale performs quite well in predicting single droplet dynamics for arbitrary
sub-critical mixed flow fields. Hence, it can be regarded as a useful tool when predicting
the morphology development in more complex systems e.g., incorporating elastic and
concentration effects, and real processing flows. The extensive data set provided here
can serve as a reference to guide and evaluate future modeling efforts.
ACKNOWLEDGMENTS
The authors would like to thank Ir. Bart Caerts for his help with the design and
construction of the ECD. This work has been financially supported by the Onderzoeksfonds KULeuven Grant Nos. GOA 03/06 and GOA 09/002.
APPENDIX
The non-dimensional form of the MM model Maffettone and Minale 1998 is given
by
dS
MM
CaW S S W = f MM
S I + Caf MM
1 pS g
2 p,CaD S
dt
+ S D,
A1
where S = S / R02, t = t / , and D and W have been made dimensionless with the flow
intensity E:
D =
W =
42 + 1 2
42 + 1 2
1
0
2
1
2
1
0
2
1 +
2
A2
The steady state values for the dimensionless axes L / 2R0 and B / 2R0 can then be calculated from Eqs. A1 and A2 to be
L
2R0
B
2R0
1305
A3
A4
Taking the limit for 1 / p 0 where f 1 goes to 20/ 19p and f 2 to 5 / 2p see Eqs. 2 for
, 0, finally we obtain
Dlim =
5 1 2 + 42
.
4p
1
A5
Hence, Dlim represents the steady deformation limit for steady 2D mixed flows with a
constant . For = 0 Dlim equals 5 / 4p, the Taylor limit for simple shear flow Taylor
1934; for = 1, on the other hand, no steady limit exists as a pure extensional flow can
break-up a drop of any p.
References
Ballal, B. Y., and R. S. Rivlin, Flow of a Newtonian fluid between eccentric rotating cylindersInertial
effects, Arch. Ration. Mech. Anal. 62, 237294 1976.
Bentley, B. L., and L. G. Leal, An experimental investigation of drop deformation and break-up in steady
two-dimensional linear flows, J. Fluid Mech. 167, 241283 1986.
Boonen, E., P. Van Puyvelde, and P. Moldenaers, Droplet dynamics in sub-critical complex flows, Rheol. Acta
48, 359371 2009.
Cristini, V., R. W. Hooper, C. W. Macosko, M. Simeone, and S. Guido, A numerical and experimental
investigation of lamellar blends morphologies, Ind. Eng. Chem. Res. 41, 63056311 2002.
Diprima, R. C., and J. T. Stuart, Flow between eccentric rotating cylinders, ASME J. Lubr. Technol. 94,
266274 1972.
Egholm, R. D., P. Fischer, K. Feigl, E. J. Windhab, R. Kipka, and P. Szabo, Experimental and numerical
analysis of droplet deformation in a complex flow generated by a rotor-stator device, Chem. Eng. Sci. 63,
35263536 2008.
Feigl, K., S. F. M. Kaufmann, P. Fischer, and E. J. Windhab, A numerical procedure for calculating droplet
deformation in dispersing flows and experimental verification, Chem. Eng. Sci. 58, 23512363 2003.
Godbille, F. D., and J. J. C. Picot, Drop break-up in combined shear and elongational flow conditions, Adv.
Polym. Technol. 19, 1421 2000.
Grace, H. P., Dispersion phenomena in high-viscosity immiscible fluid systems and application of static mixers
as dispersion devices in such systems, Chem. Eng. Commun. 14, 225277 1982.
Guido S., and Greco F., Dynamics of a liquid drop in a flowing immiscible liquid, Rheology Reviews 2004,
The British Society of Rheology 2004, pp. 99142.
Guido, S., M. Simeone, and M. Villone, Diffusion effects on the interfacial tension of immiscible polymer
blends, Rheol. Acta 38, 287296 1999.
Kaufmann, S. F. M., P. Fischer, and E. J. Windhab, Investigation of droplet dispersing processes in shear and
elongational flow, Second International Symposium on Food Rheology and Structure, Laboratory of Food
Process Engineering, Zurich, Switzerland 2000, edited by P. Fischer, I. Marti, and E. J. Windhab, pp.
404405.
Khayat, R. E., A. Luciani, L. A. Utracki, F. D. Godbille, and J. J. C. Picot, Influence of shear and elongation
on drop deformation in convergent-divergent flows, Int. J. Multiphase Flow 26, 1744 2000.
1306
Maffettone, P. L., and M. Minale, Equation of change for ellipsoidal drops in viscous flow, J. Non-Newtonian
Fluid Mech. 78, 227241 1998.
Maffettone, P. L., and M. Minale, Equation of change for ellipsoidal drops in viscous flows vol 78, pg 227,
1998, J. Non-Newtonian Fluid Mech. 84, 105106 1999.
Minale, M., Deformation of a non-Newtonian ellipsoidal drop in a non-Newtonian matrix: Extension of the
Maffetone-Minale model, J. Non-Newtonian Fluid Mech. 123, 151160 2004.
Ottino, J. M., P. De Roussel, S. Hansen, and D. V. Khakhar, Mixing and dispersion of viscous liquids and
powdered solids, Adv. Chem. Eng. 25, 105204 1999.
Rasband, W. S., IMAGEJ, U. S. National Institutes of Health, Bethesda, MD 1997.
Rumscheidt, F. D., and S. G. Mason, Particle motions in sheared suspensions. XII. Deformation and burst of
fluid drops in shear and hyperbolic flow, J. Colloid Sci. 16, 238261 1961.
Stone, H. A., Dynamics of drop deformation in viscous liquids, Annu. Rev. Fluid Mech. 26, 65102 1994.
Stone, H. A., B. L. Bentley, and L. G. Leal, An experimental study of transient effects in the break-up of
viscous drops, J. Fluid Mech. 173, 131158 1986.
Stone, H. A., and L. G. Leal, The influence of initial deformation on drop break-up in sub-critical timedependent flows at low Reynolds numbers, J. Fluid Mech. 206, 223263 1989.
Taylor, G. I., The viscosity of a fluid containing small drops of another fluid, Proc. R. Soc. London, Ser. A
138, 4148 1932.
Taylor, G. I., The formation of emulsions in definable fields of flow, Proc. R. Soc. London, Ser. A 146,
501523 1934.
Torza, S., R. G. Cox, and S. G. Mason, Particle motions in sheared suspensions. XXVII. Transient and steady
deformation and burst of liquid drops, J. Colloid Interface Sci. 38, 395411 1972.
Tucker, C. L., and P. Moldenaers, Microstructural evolution in polymer blends, Annu. Rev. Fluid Mech. 34,
177210 2002.
Van Puyvelde, P., A. Vananroye, R. Cardinaels, and P. Moldenaers, Review on morphology development of
immiscible blends in confined shear flow, Polymer 49, 53635372 2008.
Vananroye, A., P. Van Puyvelde, and P. Moldenaers, Structure development in confined polymer blends:
Steady-state shear flow and relaxation, Langmuir 22, 22732280 2006.
Vinckier, I., P. Moldenaers, and J. Mewis, Relationship between rheology and morphology of model blends in
steady shear flow, J. Rheol. 40, 613631 1996.
Wannier, G. H., A contribution to the hydrodynamics of lubrication, Q. Appl. Math. 8, 132 1950.
Windhab, E. J., M. Dressler, K. Feigl, P. Fischer, and D. Megias-Alguacil, Emulsion processingFrom
single-drop deformation to design of complex processes and products, Chem. Eng. Sci. 60, 21012113
2005.