International Journal of Multiphase Flow 35 (2009) 507–515

Contents lists available at ScienceDirect

International Journal of Multiphase Flow

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 / i j m u l fl o w

Flow of oil–water emulsions through a constricted capillary

S. Cobos a, M.S. Carvalho a,*, V. Alvarado b
a
PUC-Rio, Department of Mechanical Engineering, Rua Marques de São Vicente 225, Gávea, 22453-900 Rio de Janeiro, RJ, Brazil
b
University of Wyoming, Department of Chemical and Petroleum Engineering, Dept. 3295, 1000 E. University Avenue, Laramie, WY 82071, USA

a r t i c l e i n f o a b s t r a c t

Article history: The flow of oil-in-water emulsions through quartz micro-capillary tubes was analyzed experimentally.
Received 30 October 2008 The capillaries were used as models of connecting pore-throats between adjacent pore body pairs in
Received in revised form 4 February 2009 high-permeability media. Pressure drop between the inlet and outlet ends of the capillary was recorded
Accepted 23 February 2009
as a function of time, for several values of the volumetric flow rate. Several distinct emulsions were pre-
Available online 12 March 2009
pared using synthetic oils in deionized water, stabilized by a surfactant (Triton X-100). Two oils of differ-
ent viscosity values were used to prepare the emulsions, while two distinct drop size distributions were
Keywords:
obtained by varying the mixing procedure. The average oil drop size varied from smaller to larger than
Emulsions
Micro-capillary
the neck radius. The results are presented in terms of the extra-pressure drop due to the presence of
Enhanced-oil recovery the dispersed phase, i.e. the difference between the measured pressure drop and the one necessary to
Porous media drive the continuous phase alone at the same flow rate. For emulsions with drops smaller than the cap-
Pore blockage illary throat diameter, the extra-pressure drop does not vary with capillary number and it is a function of
Mobility control the viscosity ratio, dispersed phase concentration and drop size distribution. For emulsions with drops
larger than the constriction, the large oil drops may partially block the capillary, leading to a high extra
pressure difference at low capillary numbers. Changes in the local fluid mobility by means of pore-throat
blockage may help to explain the additional oil recovery observed in laboratory experiments and the
sparse data on field trials.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction by viscous emulsions during oil production operations reduces

well productivity. On the other hand, as discussed by Romero
Emulsions, namely liquid–liquid dispersions, are ubiquitous to et al. (1996) and Bai et al. (2000), the ability of emulsions to block
oil production operations. Most crude oil in the World is in fact pro- water channels can be exploited to control water production in
duced as water-in-oil (more commonly) or oil-in-water (generally at heterogeneous formations. Seright and Liang (1995) summarized
high water cut) emulsions. Certain crude oils contain enough naph- the advantages and disadvantages of different blocking agents,
thenic acid fractions as to generate natural surfactants upon addition including emulsions. According to their review, emulsions would
of alkaline components to serve as stabilizing agent, as discussed by approach placement properties similar to those of a low-viscosity
Kokal (2005). In practice, synthetic surfactants are frequently added gelant. Recent field results presented by Bai et al. (2000) showed
to oil–water systems to lower interfacial tension and hence stabilize the potential of emulsions as well conformance agents. A number
emulsions. Typical configurations of an emulsion are: oil-in-water, of authors agree that the observed blocking mechanisms can be
water-in-oil or a more complex dispersion, for instance a dis- attributed in part to the pressure drop–flow rate response of emul-
persed-phase siting inside each droplet already dispersed in a con- sion flow in porous media (Bai et al., 2000; Khambhratana et al.,
tinuous phase, such as the case oil-in-water-in-oil emulsions. 1998). Partial pore blocking by emulsions can be associated with
Emulsions play an important role in many enhanced-oil recov- the so-called straining, a mechanism directly controlled by the ra-
ery (EOR) processes, particularly in chemical flooding involving tio of droplet to pore-throat radii, or interception for the case of
alkaline components or surfactants, because emulsions are fre- small droplets. However, models for emulsion flow in porous med-
quently formed as a result of the injection of chemical blends into ia often rely on apparent rheological response, to explain the dy-
oil-bearing reservoirs. Emulsions are also important in surfactant- namic behavior of emulsion flow. It is clear that the flow in the
enhanced aquifer remediation of subsurface zones contaminated pore scale cannot be described by an effective viscosity. More real-
with organic liquids (Jain and Demond, 2002). Formation damage istic two-phase flow models to describe the emulsion flow in the
pore scale are necessary.
* Corresponding author. Tel.: +55 21 3527 1174; fax: +55 21 3527 1165. It is our contention that despite the progress in the study of
E-mail address: msc@puc-rio.br (M.S. Carvalho). emulsion flow in porous media, detailed analysis at the pore-scale

0301-9322/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijmultiphaseflow.2009.02.018
508 S. Cobos et al. / International Journal of Multiphase Flow 35 (2009) 507–515

is still required to develop reliable models of emulsion flow McAuliffe (1973a) described a field pilot test at the 5K of the Mid-
through porous media. Sarma et al. (1998) visualized emulsified way–Sunset oil field in California, in which 33,000 bbl of emulsion
solvent flooding for heavy oil recovery and clearly pointed out that were injected. This field trial is the first significant reported case of
emulsion mobility in porous media is not directly related to emul- EOR based on macroemulsion flooding. A total of additional
sion viscosity. Instead, trapping of emulsions droplets at pore 55,000 bbl of oil were attributed to 0.4% PV, 14%-oil emulsion
throats is the dominant mechanism in mobility control by oil-in- treatment. The results indicate better sweep efficiency of water
water emulsions. Results in bead-packed visual models showed chasing an emulsion bank, with lower water-oil ratio (WOR) in
the apparent mobility control exerted by the emulsified solvent, production wells.
reflected also as better heavy-oil recovery. Experiments with Otta- Early work by Alvarado and Marsden (1979) presented an anal-
wa sand packs showed that mobility control can be an important ogy between rheograms in capillary tubes and pressure drop–flow
recovery mechanism for emulsion flooding. Therefore, understand- rate (superficial velocity) response of emulsion flow in rock sam-
ing how emulsion flows through a porous media is not only rele- ples to develop a rheological model of macroemulsions in porous
vant to describe the effective single-phase flow behavior, but also media. Oil-in-water macroemulsions prepared with mineral oil,
will help shed light on the oil recovery mechanisms in emulsion water and a surfactant were shown to behave as Newtonian fluids
flooding. at a volumetric dispersed phase concentrations less than 50%. At
This paper presents visualization and pressure drop measure- higher concentration values, the emulsions became pseudo-plastic.
ments of oil-in-water emulsions flowing through a glass model of A fitting procedure was developed to bring porous media rheo-
an expansion–contraction–expansion micro-capillary tube. The grams to match capillary tube rheograms. A critical dispersed-
physical model is intended to mimic a pore-throat connecting phase concentration, between 40% and 50% in their experiments,
two adjacent pore-bodies of a high permeability porous media. was found to be a function of the emulsifier concentration. A mod-
Emulsions prepared with mineral or synthetic oils and a surfactant ified Darcy law, accounting for permeability reduction by plugging,
solution were pumped through the glass capillary and the pressure was used to describe the flow behavior of oil-in-water Newtonian
response was recorded as a function of time for a range of imposed macroemulsions in porous media.
flow rates. The extra-pressure drop due to the presence of the dis- Soo and Radke (1984) studied the flow of stable oil-in-water
persed phase was determined as a function of operating conditions emulsions through porous media, by using emulsions with average
and liquid properties, indicating the range of parameters at which droplet diameters of 2, 3, 5, 7 and 10 lm, flowing through two Ot-
partial blocking of the pores could be observed. The results can be tawa sandpacks with mean pore-throat diameters of 17.3 and
used in the development of a capillary network model to study the 29.5 lm, respectively. Micromodel experiments were also carried
flow of emulsions through a porous medium. out, in which Ottawa sand was sandwiched between glass plates.
Before describing the experimental analysis and results of this They clearly showed that, besides the straining of large droplets
work, we discuss some of the past work directly related to flow in pore throats, interception in crevices or pockets between grains
of emulsions in porous media and analysis of the flow of a single was also a capture mechanism. The authors developed a filtration
drop through straight and constricted capillaries. model for dilute emulsions (Soo and Radke, 1986; Soo et al.,
1986). Deep-bed filtration theory concepts were used to describe
1.1. Experiment and models for emulsion flow through porous media dilute, stable emulsion flow in porous media, including flow redis-
tribution and large permeability reductions (see Soo and Radke,
Romero et al. (1996) carried out a three-stage study to develop 1986). The important parameters in the model are kp (the filter
a deep-penetrating oil-in-water emulsion to improve sweep effi- coefficient) and bp (which measures the effectiveness of the re-
ciency in heterogeneous and fractured formations. The average tained drops in pores of throat diameter Dp). The former parameter
emulsion droplet size used was 2.10 lm, obtained by mixing the determines the droplet capture, while the latter one controls per-
oil with an alkaline solution. The idea of flooding the pore space meability reduction. These parameters, for lack of theoretical treat-
with droplet diameter similar to rock pore-throat size was pre- ment, were determined empirically, leading to the concept of flow
sented earlier by McAuliffe (1973b). The injectivity index in the diversion, characterized by a parameter a. Darcy law completed
experiments of Romero et al. (1996) with consolidated rock the model. Soo et al. (1986) showed how to estimate filtration
dropped 96–99% of its original value and the emulsion maintained parameters of their model from experimental data. Emulsions
its plugging ability even after the injection of 27 PV of water. ranging from 0.5% to 2.5% dispersed phase concentrations were
Bai et al. (2000) presented the results of an emulsifying treatment used to test the filtration theory for emulsion flow in porous media.
yielding viscous emulsions for water shutoff applied to 238 wells The model successfully described emulsion flow results up to
in Liaohe field, China, in the period from 1991 to 1999. An 83.6% approximately 1% concentration of the dispersed phase.
success rates was reported, with an average effective term of 4.5 Hofman and Stein (1991) presented experimental results on the
months. influence of electrostatic repulsion and emulsion stability on the
McAuliffe (1973b) carried out experiments with oil-in-water permeability impairment caused by emulsion flow through porous
emulsions, between 10% and 80% oil fraction, using three crude media. The dispersed-phase concentration was 1% (v/v) in all their
oils, and sodium hydroxide as the caustic agent (alkaline compo- experiments. Significant permeability reduction was observed for
nent). Emulsion viscosity was low for an oil fraction 650%, because average droplet diameter equal to 5.1 and 8.9 lm, respectively,
water was the continuous phase. Apparent permeability after being larger for the larger average diameter. Hofman and Sten re-
emulsion injection in a 1360 mD-core was a strong function of port this as a striking result, because the pore-size distribution had
the mean droplet size. Significant drop in permeability was ob- a large fraction of all the pores well above the average droplet size.
served for average droplet diameter of 12 lm. In all cases, plugging The results indicate that both stable and unstable emulsion caused
stopped once the reduction in permeability reached 1–10% of the permeability reduction, being the latter more effective in plugging
original rock permeability value. McAuliffe related this result to porous media.
the threshold pressure drop across the length of a pore throat nec- Islam and Farouq Ali (1994) developed a Darcy-level model for
essary to mobilize a droplet through the throat. Oil recovery with flow of stable emulsions in porous media, including in situ gener-
emulsion flooding in cores was attributed to corrections of hetero- ation. Their model was based on conservation equations for water,
geneity in the rock, and not to miscibility. This was perhaps the oil and emulsion phases, coupled to a transport equation. The fil-
first time that emulsion flooding was proposed for EOR purposes. tration model of Soo and Radke (1984, 1986) was used to account
 S. Cobos et al. / International Journal of Multiphase Flow 35 (2009) 507–515 509

for capture mechanisms. The results were compared to the exper- is determined by competing mechanisms, such as the deformation
imental data of Soo and Radke (1984). of the drop, interaction between the drop and the wall and the sim-
Abou-Kassem and Farouq Ali (1995) reviewed pseudo-plastic ple replacement of suspending liquid by a dispersed phase liquid of
fluid models of emulsion flow in porous media and discussed their different viscosity. At high capillary numbers, they explored situa-
underlying assumptions. They proposed a Darcy-law type model tions with drops less and more viscous than the suspending liquid.
based on the flow of a power-law liquid through a bundle of tubes. For all cases explored, both small and large drops, the oscillation of
The mapping of porous-media rheograms on capillary-tube type the pressure signal was very small and the extra-pressure differ-
rheograms followed from Alvarado and Marsden (1979). ence did not vary much with capillary number, but was a strong
Khambhratana et al. (1997) developed a simulator for flow of function of the viscosity ratio between the phases and the drop
emulsions in porous media and compared the results with experi- size.
mental data obtained in Berea and an Ottawa sand pack core floods The pressure oscillation due to the flow of a single drop through
presented by Khambhratana et al. (1998). The best match between a constricted capillary was predicted by the analysis presented by
simulation results and experimental data were obtained when a Tsai and Miksis (1994) using a boundary integral method.
multiphase non-Newtonian rheological model of an emulsion with It is important to note that all the experimental results dis-
interfacial tension-dependent relative permeability curves and cussed in the previous paragraphs were obtained with tubes with
time-dependent capture were used. diameter in the order of 1–5 mm, much larger than the flow chan-
As for the review presented, it appears that an accurate model nels in a porous media. Moreover, the range of capillary number
for flow of emulsions in porous media is still not available. A better explored, 102 < Ca < 1, is also outside the typical local capillary
understanding of the flow of drops immersed in a continuous li- number in a pore flow. In this work, we extend these analyses to
quid phase through a constricted capillary is crucial in this devel- micro-capillaries, with constriction diameter of 50 lm, and much
opment. In the following sub-section, we discuss some of the lower capillary numbers. Since we are interested in understanding
first attempts in this direction. the fundamental oil-recovery mechanisms upon emulsion injec-
tion, we focus our analysis on drops more viscous than the contin-
1.2. Flow of a single drop immersed in a continuous phase through uous phase, range not explored in the literature at the low capillary
straight and constricted capillaries number regime.

The flow of a single drop immersed in the liquid flowing 2. Materials and methods
through a straight capillary was analyzed experimentally by Ho
and Leal (1975) and Olbricht and Leal (1982). They measured the 2.1. Materials
extra pressure drop due to the presence of a single drop in the flow
as a function of the ratio between the drop diameter and the cap- Five oil-in-water emulsions were prepared for the experiments,
illary diameter, viscosity ratio and flow rate (capillary number). as shown in Table 1. The continuous phase was a dilute solution of
The same situation was studied using a boundary integral method Carbopol (a low molecular weight polymer), with a concentration
by Martinez and Udell (1990). For small drops, e.g. ratio of drop equal to 0.1 wt%, in deionized water. The polymer was added to in-
diameter to tube diameter less than 0.7, the extra pressure drop crease the continuous phase viscosity and therefore delay the seg-
does not depend on the capillary number. For drops larger than regation of phases due to density differences. The aqueous solution
the tube diameter, the extra pressure drop falls with capillary presented a shear thinning behavior, and its viscosity at
number and rises with viscosity ratio. The extra pressure drop is c_ ¼ 0:01 s1 was g  0:2 Pa s. A surfactant was added to the aque-
negative when the drop is less viscous than the continuous phase, ous phase to lower the oil–water interfacial tension and limit the
and positive in the reverse case. The extra pressure difference is a coalescence of oil drops. A nonionic surfactant, Triton X-100, was
strong function of the drop to capillary diameter ratio d=D. Marti- used to avoid pH changes of the polymeric solution and therefore
nez and Udell (1990) have shown that for d=D K Oð1Þ, the extra changes in its viscosity. The surfactant concentration was approx-
pressure difference scales with ðd=DÞ5 . For drops much larger than imately 10 times the critical micelle concentration. Two synthetic
the capillary diameter, i.e. d=D  1, Ho and Leal (1975) observed oils were used as the dispersed phase. The first oil was SHELL Tive-
that the increase in the extra pressure difference with the drop la 160, with density and viscosity equal to qo = 993 kg/m3 at 20 °C
diameter can be predicted simply by the taking into account the in- and lo = 350 cP at 25 °C, respectively. The second oil used was
crease in length of the large drop. SHELL Tivela 460, with density and viscosity equal to
Olbricht and Leal (1983) reported experimental results on the qo = 997 kg/m3 at 20 °C and lo = 950 cP at 25 °C, respectively. The
flow of a single drop immersed in a liquid flowing through a hori- interfacial tension of both oils with respect to the continuous
zontal tube with periodically varying diameter. As the drop flows phase was r = 5 mN/m. Each oil and water + polymer + surfactant
through the converging–diverging channel, the pressure difference mixture was sheared in a homogenizer (Ultramax). The rotation
oscillates. Each oscillation of the pressure signal corresponded to a of the dispersing tool and the time of mixture were used to control
drop’s passage through a constriction. The extra-pressure differ- the drop size distribution of the emulsion. A sample of each emul-
ence response was characterized by the arithmetic average of the sion was placed under an optical microscope and the size of
signal and the amplitude of the oscillation. The results are analyzed
separately for low and for high capillary regimes. At low capillary
number, the analysis was restricted to drops less viscous than
the continuous phase. For drops smaller than the capillary con-
striction, the pressure oscillation was smaller than the uncertainty Table 1
Main properties of the oil-in-water emulsions used in the experiments.
of the measurements, but it rises with the drop diameter, being rel-
evant when the drop is larger than the constriction diameter. The Emulsion Internal phase lo (cP) Concentration Drop size
extra-pressure difference was higher at low capillary number, 1S Tivela 160 350 30/70 Small
when all the other flow parameters were kept the same. At a fixed 1L Tivela 160 350 30/70 Large
capillary number, the extra-pressure difference was positive for 2S Tivela 460 950 30/70 Small
2L Tivela 460 950 30/70 Large
small drops, but falls and becomes negative as the drop size rises.
3S Tivela 460 950 60/40 Small
Olbricht and Leal (1983) discuss that the extra-pressure difference
510 S. Cobos et al. / International Journal of Multiphase Flow 35 (2009) 507–515

individual drops was measured to build the normalized histogram Table 2

of drop size distribution. Mean average oil drop radius and standard deviation for all the emulsion systems
used in the experiments.
One emulsion set was prepared by dispersing oil in water using
a rotation rate of 26,500 rpm for 3 min. The largest drop diameter Emulsion Mean drop diameter (lm) Standard deviation (lm)
of the dispersed phase was approximately 35 lm, smaller than the 1S 22.2 9.4
capillary throat diameter. Emulsions prepared with this procedure 1L 42.3 27.6
were labelled *S. A second emulsion batch was obtained by mixing 2S 10.3 4.2
2L 51.5 25.3
a fraction of the emulsion generated at a rotation of 6500 rpm for 3S 9.3 2.7
3 min and the remaining fraction obtained by hand-shaking an
oil–water mixture. This procedure led to a very broad dispersed
phase diameter distribution, containing drops much larger than
the capillary constriction. These emulsions were labelled *L. Fig. 1 The shear viscosity of all the fluids as a function of the shear rate
shows the normalized histograms of all the five emulsions tested. was measured using a rotational rheometer, ARES (TA Instru-
As expected from optical determinations of drop-size distributions, ments) with a grooved Couette geometry. The serrated cylinder
the frequency corresponds to drop numbers. The volume fraction was used to dissipate slipping effects at the walls, common in
of large drops is under represented, being much larger than that the characterization of emulsions and some polymeric solutions,
of small drops. The mean and the standard deviation of the drop as described by Macosko (1994). The viscosity curve of the contin-
size distribution of the five emulsions are presented in Table 2. uous phase and of emulsions 1L and 1S are shown in Fig. 2. The
The first two emulsions, labelled 1L and 1S, were prepared using polymeric aqueous solution with surfactant behaves as a shear
SHELL Tivela 160, with an oil fraction of 30%. The difference be- thinning fluid. Both emulsions also exhibit a shear thinning behav-
tween emulsions 1L and 1S was the drop size distribution. Emul- ior. The emulsion with smaller average drop size and narrower dis-
sion 1L had large drops, while 1S consisted of small drops. tribution exhibits larger viscosity values than the emulsion with
Emulsions 2L and 2S were prepared at 30% oil concentration, but larger drop size and wider drop-size distribution, but the same
using oil SHELL Tivela 460, instead. Similarly, 2L contained large phase concentration. The larger interfacial area and mutual inter-
drops and 2S small drops. In order to determine the effect of the action between different drops increases the viscosity of the emul-
dispersed phase concentration, an emulsion with small drops, sion, as explained by Becher (2001). The viscosity curves of
but with 60% of oil, was prepared. This emulsion was labelled 3S. emulsions 2L, 2S and 3S are shown in Fig. 3. Once again, the emul-
sion with smaller drop size is more viscous. As the results for emul-
sion 3S show, viscosity rises as the concentration of the dispersed
phase increases, as expected.
For all the emulsions tested, the viscosity dependence on shear
rate at shear rates larger than 1 s1 is well described by a power-
law model, namely g ¼ mc_ n1 . The parameters of the model for
the five emulsions are shown in Table 3.

2.2. Experimental procedure

The experimental apparatus used for flow visualization in this

work is shown in Fig. 4. A syringe pump is used to feed the emul-
sion through a quartz capillary at a constant volumetric flow rate.
The tubing connecting the syringe to the capillary was made as
short and as rigid as possible to minimize the expansion of the tub-
ing as the inlet pressure rises during the experiments. A port, con-

Fig. 2. Shear viscosity as a function of shear rate of the dispersed phase (carbopol
solution in water) and emulsions 1S and 1L. m and n are the power-law model
Fig. 1. Drop size distribution for the five emulsions tested. coefficients.
 S. Cobos et al. / International Journal of Multiphase Flow 35 (2009) 507–515 511

Fig. 5. Photograph of the constricted capillary used as a model of a pore body–pore

throat geometry.

the images of the emulsion flowing through the constricted capil-

lary tube were recorded. The range of flow rates explored was cho-
sen such that the average velocity of the liquid flowing through the
capillary was similar to the average velocity encountered inside
the porous space of oil reservoirs.

3. Results and discussion

Fig. 3. Shear viscosity as a function of shear rate of emulsions 2S, 2L and 3S. m and n
are the power-law model coefficients. For emulsions 1L and 1S, the flow rate explored ranged from
Q = 0.04 ml/h to Q = 0.1 ml/h. The inlet pressure during the exper-
iment for both emulsions is presented in Fig. 6. The flow rate
Table 3
Powerlaw model parameters for all emulsions tested. was kept at Q = 0.04 ml/h for t 6 58 min. At t = 58 min, the flow
rate was raised to Q = 0.07 ml/h, and at t = 110 min, it was raised
Emulsion m (Pa sn) n
once again to Q = 0.1 ml/h. For emulsion 1S, at each flow rate, the
Continuous phase 0.074 0.81 pressure stayed almost constant, with only small fluctuations
1S 0.267 0.65
around the average value. After each flow rate change, a new stea-
1L 0.171 0.73
2S 0.171 0.75 dy state was reached after approximately 9 min. This long tran-
2L 0.092 0.80 sient can be explained by the presence of the compliant tubing
3S 0.358 0.72 that connects the syringe pump to the capillary, by the small
dimensions of the capillary and extremely low flow rates explored.
Similar behavior has been observed in flows of glycerin–water
solutions through micro-capillaries (see Rodd et al., 2005). For
the emulsion with large drops 1L, at each flow rate, the pressure
difference was higher. At Q = 0.04 ml/h, the pressure difference
necessary to drive emulsion 1S through the capillary was
DP  2000 Pa, and to drive emulsion 1L was DP  4000 Pa, even
though emulsion 1S is more viscous than emulsion 1L. As expected,
from these results, it is clear that a continuous model based solely
on the emulsion viscosity cannot be used to describe the flow
when there are drops larger than the constriction diameter. The
presence of oil drops larger than the capillary throat partially ob-
structs the liquid flow. Another important difference on the behav-
ior of the flow with emulsions 1S and 1L is that the inlet pressure

Fig. 4. Photograph of the experimental setup.

nected to a pressure transducer (Validyne), was installed just up-

stream the capillary tube to measure the pressure drop along the
capillary. Pressure values were acquired on the PC by using a mul-
tiplexer slot as interface between the transducer and the computer.
Pressure data points were taken every second. Systematic error for
pressure drop acquisition was ±2%.
The capillary tube was mounted for observation on the platform
of an inverted Zeiss Axioplan microscope with a 5 objective.
Images were captured in real time and recorded by a CCD camera
attached to the microscope and connected directly to the frame
grabber, and then recorded on a video tape. The diameters of the
glass capillary tube and the neck were equal to 200 and 50 lm,
respectively. The length of the capillary was 8 cm and that of the
contraction and expansion section was equal to 4 cm. Fig. 5 shows
a photograph of the throat section of the capillary tube.
For each emulsion, the flow rate was increased in small steps. At Fig. 6. Evolution of the inlet pressure as the flow rate varies for emulsions 1S and
each value of the flow rate, the inlet pressure was measured and 1L. Range of flow rate: Q = 0.04–0.1 ml/h.
512 S. Cobos et al. / International Journal of Multiphase Flow 35 (2009) 507–515

mated by DP cap  2r=Rc ¼ 400 Pa, in this experiment. It is clear

that this effect alone does not explain the increased pressure dif-
ference necessary to drive the flow. Another mechanism responsi-
ble for the pressure rise is the higher interaction between the drop
and the capillary wall that occurs with emulsions containing large
drops. A third important mechanism is the simple replacement of a
less viscous liquid by a more viscous liquid.
The pressure at different flow rates for emulsions 2S and 2L is
shown in Fig. 9. The flow range explored went from Q = 0.04 to
Q = 0.22 ml/h. As in the previous example, the flow rate was raised
in steps as the inlet pressure was recorded. For the emulsion with
small drops 2S, the inlet pressure at each fixed flow rate was virtu-
ally constant. For the large drop emulsion 2L, pressure oscillates
even when the flow rate is held constant, as also observed in the
case of emulsion 1L. However, the amplitude of the oscillation is
much larger than that observed with emulsion 1L. This can be ex-
plained by the broader drop size distribution and by the higher vis-
cosity of the dispersed phase of emulsion 2L, when compared with
Fig. 7. Evolution of the inlet pressure at Q = 0.04 ml/h for emulsion 1L. Images of
emulsion 1L. The pressure oscillation caused by the flow of large oil
the flow near the capillary throat at the instants highlighted in this plot are shown drops through the capillary throat lasted as long as 10 min in some
in the next figure. cases. This rather long time may be explained by the fact that the
oil drop flows at a speed slower than the speed of the bulk at the
center line. The slower moving large oil drop leads to a local higher
oscillates substantially at a fixed flow rate in the flow of the emul- concentration of the dispersed phase upstream of the constriction,
sion with drops larger than the constriction, as was also observed similar to a filtration process, contributing also to the pressure rise.
by Olbricht and Leal (1983). They associated the pressure fluctua- The fact that a more viscous drop moves slower than the bulk max-
tions with the passage of a single large drop through the capillary imum velocity through a capillary has been experimentally ob-
throat. Fig. 7 shows the oscillation of the inlet pressure for emul- served by Ho and Leal (1975) and predicted by Martinez and
sion 1L at a flow rate Q = 0.04 ml/h, between t = 15 and 25 min. Udell (1990). Another possible explanation is that the dispersed
The images of the flow near the capillary throat at the instants phase drops may have coalesced in the syringe before being fed
highlighted in Fig. 7 are presented in Fig. 8. A small pressure peak into the capillary. However, we did measure the drop size distribu-
is observed as a drop somewhat larger than the throat diameter tion after the experiments and did not find any evidence of
passes through the constriction, as observed in frames (a) and (b) coalescence.
of the figure. The deformation of the drop at the constriction is The higher pressure observed with the emulsion with large
clear in frame (b). A larger pressure peak is observed when a very drops and the amplitude of the pressure oscillation at a fixed flow
large drop, much larger than the capillary throat, flows through the rate decays as the flow rate rises. At the two highest flow rates ex-
constriction, as shown in frames (c) and (d). As the tip of the drop plored, Q = 0.19 and Q = 0.22 ml/h, the average pressure with emul-
passes through the smallest diameter section, pressure reaches a sions 2L and 2S are virtually the same. As the flow rate rises, the
local maximum value and then drops as the front of the drop capillary number increases and the viscous forces becomes more
moves into the diverging portion of the constriction. A new pres- important relative to surface tension induced forces. Therefore,
sure rise occurs whenever a large drop starts to protrude the cap- the extra-pressure required to deform the large drop to squeeze
illary throat. This behavior was observed for all emulsions tested it through the capillary throat becomes negligible, leading to a
here. pressure drop close to that of the emulsion with small drops. Mar-
The higher average pressure at a fixed flow rate observed when tinez and Udell (1990) predicted that the effect of a single drop
emulsion 1L was used is caused by different mechanisms. As the oil flowing through a straight capillary tube on the pressure drop at
drop flows through the constriction, the radius of curvature of the a fixed flow rate falls with capillary number.
front of the drop falls, and consequently a higher pressure differ- The effect of the dispersed phase concentration on the flow
ence is necessary to overcome the capillary pressure across the characteristics was evaluated by analyzing the inlet pressure mea-
drop front interface at the capillary throat, that can be approxi- sured in the flow of emulsion 3S, prepared with the same oil and

Fig. 8. Images of the emulsion flow just upstream of the capillary throat at the instants highlighted in the previous plot.
 S. Cobos et al. / International Journal of Multiphase Flow 35 (2009) 507–515 513

ies along the flow direction can be written in terms of the

rheological parameters m and n, the flow rate Q and an equivalent
radius R that represents an average of the capillary radius
 1 n
Q nþ3 2mL
Dpc ¼ : ð1Þ
pR3 R
The most appropriate equivalent radius R of the capillary for a
power-law liquid is defined as
8 9 1
< L =3nþ1
R  RL : ð2Þ
: dx ;
3nþ1
0 RðxÞ

The image showing the geometry of the capillary, presented in

Fig. 5, was digitized and used to evaluate the above integral, using
the rheological parameters of the continuous phase, e.g.
m ¼ 0:074 and n ¼ 0:81. The calculated equivalent radius was
R ¼ 84:90 lm.
The characteristic viscosity of the continuous phase lc at each
flow condition was estimated by the ratio of the wall shear stress
Fig. 9. Evolution of the inlet pressure as the flow rate varies for emulsions 2S and sw and the wall shear rate c_w , which for a power-law liquid are gi-
2L. Range of flow rate: Q = 0.04–0.22 ml/h.
ven by:

RDpc
procedure of emulsion 2S, but with a oil concentration of 60%. The s_ w ¼ ; ð3Þ
2L
results are shown in Fig. 10. As in the previous flows with emul- 1
þ 3 4Q
sions containing only small drops, pressure is virtually constant c_ w ¼ n : ð4Þ
4n pR3
at each flow rate, indicating that there is no partial blocking of
the capillary throat by the oil drops. The extra-pressure difference DP þ , in units of lc V=R, as a function of
As done by previous authors studying the flow of a single drop the capillary number Ca  lc V=r for all the emulsions tested are
immersed in a liquid through a capillary, the flow rate–pressure presented in Fig. 11. The error bars indicate the standard deviation
difference relationship for the emulsion flow through a constricted of the pressure measurements at each flow rate. A large error bar
capillary is analyzed in terms of the extra pressure difference DPþ , indicates a large amplitude oscillation of the pressure signal.
defined as the measured pressure drop for the emulsion flow less The extra pressure DP þ obtained with emulsions with small
the theoretical Poiseuille pressure drop necessary to drive the con- drops, e.g. emulsions 1S, 2S and 3S, is not a function of the capillary
tinuous phase at the same flow rate. DPþ represents the pressure number and does not oscillate. This is the same behavior presented
drop due to the presence of the dispersed phase. by Olbricht and Leal (1982) and Martinez and Udell (1990) for flow
Because the continuous phase is shear thinning and the capil- of a single drop through a straight capillary and by Olbricht and
lary tube does not have a constant diameter, it is not simple to cal- Leal (1983) for flow of a single drop through a constricted capillary.
culate the Poiseuille pressure difference associated with the flow of The extra-pressure difference obtained with emulsion 2S, that
the suspending liquid and to define a characteristic value of the had the smallest average diameter, approximately 1/5 of the capil-
continuous phase viscosity, used in the definition of the capillary lary throat, and low concentration (30% by volume) was approxi-
number.
The pressure drop Dpc of the continuous phase (a power-law li-
quid) in the flow through a capillary tube with a diameter that var-

Fig. 11. Extra-pressure difference as a function of the capillary number for the five
Fig. 10. Evolution of the inlet pressure as the flow rate varies for emulsion 3S. emulsions tested. The error bars indicate the standard deviation of the measured
Range of flow rate: Q = 0.04–0.1 ml/h. pressure drop through the capillary.
514 S. Cobos et al. / International Journal of Multiphase Flow 35 (2009) 507–515

mately zero, i.e. the presence of the dispersed phase did not alter
significantly the flow, when compared to the flow of the continu-
ous phase alone. It is important to observe that the shear viscosity
of emulsion 2S, measured in a cone-and-plate fixture in a rota-
tional rheometer, was larger than the continuous phase viscosity,
as shown in Fig. 3. As expected, in this length scale, the flow of
emulsions cannot be described simply by its viscosity. The extra-
pressure rises with the average drop diameter (emulsion 1S) and
concentration of the dispersed phase (emulsion 3S).
Because of the smaller capillary diameter used here, closer to
that of a high permeability porous media, and the fact that we
are not analyzing the flow of a single drop, but of an emulsion,
the values of the extra pressure difference measured in this analy-
sis are orders of magnitude higher than those reported before in
the literature for the flow of a single drop suspended in the liquid.
The extra pressure difference DP þ obtained with emulsions that
have drops larger than the capillary throat diameter (emulsions 1L
and 2L) are a strong function of the capillary number. The extra
pressure difference falls as the capillary number rises. This behav-
ior was also reported by Olbricht and Leal (1982, 1983). Again the
values of the extra pressure difference measured here are orders of
Fig. 12. Scale factor to be used on the definition of the capillary mobility for flow of
magnitude larger because of the much smaller capillary diameter, emulsions.
capillary number and viscosity of the dispersed phase used in this
work.
The comparison between the measured extra-pressure differ-
ence with emulsions 2S and 2L reveals an interesting phenomena, in the flow due to the presence of the dispersed phase is presented
not reported before. At low capillary number, e.g. Ca < 0:008, DPþ in terms of the ratio between the pressure differences, not the dif-
with emulsion 2L is much larger than that obtained with emulsion ference between them. Further investigation is necessary in order to
2S. At capillary numbers above this critical value, the pressure gra- construct a complete description on how the scaling factor f varies
dient–flow rate relationship for both emulsions is virtually the with the emulsion concentration, viscosity ratio, drop size distribu-
same. As discussed before, at high capillary number, the surface tion and capillary number.
tension induced force is negligible and so is the extra pressure
needed to deform the drop. The partial pore blocking mechanism, 4. Summary
that represents the larger extra-pressure difference, related to the
presence of large drops on the injected emulsion, that may explain The flow of oil-in-water emulsions through a pore-throat mod-
the improved reservoir sweep observed in some experiments, only el, represented by a converging–diverging quartz capillary tube,
occurs if the local capillary number is below a critical value that is has been characterized by the pressure–drop flow rate response.
a function of the emulsion properties. In this case, it is Cac  0:008. The flow response characterization has been paired with simulta-
If the emulsion properties and process conditions are such that the neous visualization under an optical microscope. The effect of the
local capillary number is higher than this value, there is no pore diameter of the dispersed phase on the pressure drop was analyzed
blocking on an oil recovery operation. This dependence on capillary by comparing the response of two emulsions with the same volu-
number may be used in order to define the location where the par- metric dispersed phase fraction, but with different drop size distri-
tial pore blocking should occur in the reservoir. This can be accom- bution; one with an average drop diameter smaller and the other
plished by tuning the emulsion properties, such as interfacial larger than the narrowest section of the capillary. The results indi-
tension, viscosity ratio and drop size distribution, in order to have cate that even for the relatively smooth geometry of the con-
the critical capillary number below which the mobility reduction stricted capillary employed, the flow of emulsion is dominated
occurs coinciding with the local capillary number at the displace- by blocking mechanisms caused by drops larger than the capillary.
ment front. Blockage of the pore-throat or constriction in the quartz model
These results can be used in the development of a capillary net- signals abrupt pressure oscillations, which are directly associated
work model for the flow of emulsions in porous media. The flow in with the passage of large drops through the capillary throat. The
each capillary q can be described in terms of the mobility of the increased curvature of the drop tip as it approaches the narrowest
capillary and the pressure difference applied to its extremities Dp: section of the capillary tube leads to pressure peaks, followed by
pressure relaxation. However, as the capillary number is increased,
Dp
q ¼ fK c ; ð5Þ leading to larger viscous effects, the average pressure response for
L
the coarser emulsion approaches that of the smaller-radius
where K c is the mobility of the flow of the continuous phase alone emulsions.
and f is a scaling factor to account for the partial pore blocking, de- The blocking mechanism has been characterized by comparing
fined as the ratio of the Poiseuille pressure difference necessary to the average measured response to that of flow of the continuous
drive the flow of the continuous phase alone to the one associated phase alone. The results indicate the effect of the dispersed phase
with the emulsion flow at the same flow rate: in the flow and that, as expected, the flow of emulsions through
Dpc micro-capillaries cannot be described by its viscosity alone. A
f ¼ : ð6Þ two-phase flow analysis is needed.
Dp
We speculate that as emulsions used for enhanced-oil recovery
Fig. 12 shows the scaling factor as a function of the capillary num- operations flow through reservoir rocks, they act as dynamic
ber for the five emulsions tested in the experiments. These results mobility control agents, by blocking water paths and diverting
are equivalent to those presented in Fig. 11, but now the change the displacing fluid toward upswept regions of the porous medium.
 S. Cobos et al. / International Journal of Multiphase Flow 35 (2009) 507–515 515

This mechanism requires further investigation, but the results pre- Khambhratana, F., Thomas, S., Farouq Ali, S.M., 1997. In: SPE 39033, Proceedings of
the Fifth Latin American and Caribbean Petroleum Engineering Conf. and
sented here will serve to build network models to develop further
Exhibition, Rio de Janeiro, Brazil, pp. 1–16.
understanding of flow of emulsions through porous media. Khambhratana, F., Thomas, S., and Farouq Ali, S.M. 1998. In: SPE 48910, Proceedings
of the 1998 SPE Conference and Exhibition in China, Beijing, China, pp. 657–665.
Acknowledgements Kokal, S., 2005. SPE Prod. Facil. 20, 5–13.
Macosko, C.W., 1994. Rheology: Principles, Measurements, and Applications.
Wiley–VCH, New York.
S.C. would like to thank ANP for support through his doctor’s Martinez, M.J., Udell, K.S., 1990. J. Fluid Mech. 210, 565–591.
degree scholarship. This research was funded by grants from the McAuliffe, C.D., 1973a. J. Petrol. Technol. 25, 721–726.
McAuliffe, C.D., 1973b. J. Petrol. Technol. 25, 727–733.
Brazilian Research Council (CNPq) and Petrobras. Olbricht, W.L., Leal, G., 1982. J. Fluid Mech. 115, 187–216.
Olbricht, W.L., Leal, G., 1983. J. Fluid Mech. 134, 329–355.
Rodd, L.E., Scott, T.P., Boger, D.V., Cooper-White, J.J., McKinley, G.H., 2005. J. Non-
References Newton. Fluid Mech. 129, 1–22.
Romero, L., Zirit, J.L., Marín, A., Rojas, F., Mogollón, J.L., Manrique, E., Paz, F., 1996. In:
Abou-Kassem, J.H., Farouq Ali, S.M., 1995. J. Can. Petrol. Technol. 34, 30–38. SPE/DOE 35461, Proceedings of the Tenth Symposium on IOR, Tulsa, OK, USA,
Alvarado, D.A., Marsden, S.S., 1979. SPE J. 19, 369–377. pp. 611–621.
Bai, B., Han, M., Li, Y., Wei, M., Gao, Y., Cost, J.-P., 2000. In: SPE/DOE 59320, Sarma, H.K., Maini, B.B., Jha, K., 1998. J. Can. Petrol. Technol. 37, 55–62.
Proceedings of the 24th SPE/DOE Symposium on IOR, Tulsa, OK, USA, pp. 1–7. Seright, R.S., Liang, J., 1995. In: SPE 30120, Proceedings of the European Formation
Becher, P., 2001. Emulsions: Theory and Practice. Oxford University Press, Oxford. Damage Conference, The Hague, The Netherlands, pp. 431–440.
Ho, B.P., Leal, G., 1975. J. Fluid Mech. 71, 361–384. Soo, H., Radke, C.J., 1984. Ind. Eng. Chem. Fund. 23, 342–347.
Hofman, J.A.M.H., Stein, H.N., 1991. Colloids Surf. 61, 317–329. Soo, H., Radke, C.J., 1986. Chem. Eng. Sci. 41, 263–272.
Islam, M.R., Farouq Ali, S.M., 1994. J. Can. Petrol. Technol. 33, 59–63. Soo, H., Williams, M.C., Radke, C.J., 1986. Chem. Eng. Sci. 41, 273–281.
Jain, V., Demond, A.H., 2002. Environ. Sci. Technol. 36, 5426–5433. Tsai, T.M., Miksis, M.J., 1994. J. Fluid Mech. 274, 197–217.