Journal of Industrial and Engineering Chemistry 29 (2015) 382391

Contents lists available at ScienceDirect

Journal of Industrial and Engineering Chemistry

journal homepage: www.elsevier.com/locate/jiec

Emulsion ooding for enhanced oil recovery: Interactive optimization

of phase behavior, microvisual and core-ood experiments
Mohammad Saber Karambeigi a,*, Reza Abbassi b, Emad Roayaei a, Mohammad Ali Emadi a
a
b

IOR/EOR Research Institute, Tehran 1969813771, Iran

Department of Petroleum Engineering, Fars Science and Research Branch, Islamic Azad University, Marvdasht, Iran

A R T I C L E I N F O

Article history:
Received 7 March 2015
Received in revised form 21 April 2015
Accepted 21 April 2015
Available online 2 May 2015
Keywords:
Emulsion ooding
Response surface methodology
Diesel
Phase behavior
Enhanced oil recovery

A B S T R A C T

Emulsion ooding has been proved to be an effective chemical enhanced oil recovery (CEOR) method.
The performance of this technique is strongly dependant on the formulation of the emulsion slug. To
achieve the best formulation, different approaches have been introduced in the literature. Nevertheless,
a systematic workow containing the method(s) of the design of experiments (DOEs) has not been
reported as yet. In this paper, we developed a workow which is comprised mainly of three stages. In the
rst place, the phase behavior experiments of sodium dodecyl sulfate (SDS)/water/diesel (as an efcient
and economic hydrocarbon phase)/salt system were carried out using response surface methodology
(RSM) to model and optimize the emulsication process. The second stage was followed by the
characterization of optimum formulation in terms of rheological behavior and particle size distribution.
Finally, microvisual and core-ood displacement tests were performed to evaluate the efcacy of
emulsion ooding to recover the residual oil bypassed or trapped after water ooding. This paper
presents the results of experiments done in different stages of the proposed workow. The results
demonstrate the high potential of emulsion formulated systematically by DOE approach to increase oil
recovery factor.
2015 The Korean Society of Industrial and Engineering Chemistry. Published by Elsevier B.V. All rights
reserved.

Introduction
In conventional petroleum reservoirs, maximum oil production
by natural-drive (primary) mechanisms and consequently pressure maintenance (secondary) methods is merely 2060% original
oil in place (OOIP); in other words, nearly 2.0  1012 barrels of
conventional oil, which is the main objective of various enhanced
oil recovery (EOR) methods [14]. Poor sweep efciency as one of
the most underlying reasons of this issue contributes to bypassing
the substantial amount of oil in reservoirs. Weak macroscopic
displacement efciency contributes to unfavorable mobility ratio
between displacing and displaced uids as well as reservoir
heterogeneity.
From pore-scale standpoint, capillary forces are the factor
most responsible to trap oil ganglia in pore space which causes
inefcient microscopic displacement efciency and consequently

* Corresponding author. Tel.: +98 021 88874500; fax: +98 021 88661307.
E-mail address: petromazer@gmail.com (M.S. Karambeigi).

entraps an immense amount of oil in the pay zone. Capillary forces

become dominant when the pores constrain the wateroil
interface to a high degree of curvature. By contrast, viscous forces,
i.e. drag forces produced by the uid ow of continuous phase,
behave against capillary forces so that they can overcome them
and displace trapped oil. The ratio of viscous to capillary forces
which is known as capillary number determines how strongly
trapped oil is within a specic porous medium [5].
To mobilize the oil bypassed and/or trapped in the reservoir,
effective mobility ratio of displacing uid and/or signicant
reduction of interfacial tension between residual oil and displacing
uid are required. For this purpose, injection of emulsions has
been recognized as a potentially efcient chemical EOR process.
Emulsion ooding refers to a slug comprised of stable solution
of hydrocarbon, brine and one or more surfactants which is
injected into the reservoir to mobilize or solubilize the remaining
oil [69].
Emulsions (macroemulsions) are dispersion of one immiscible
liquid inside another continuous liquid phase. They are kinetically
stable and are not formed spontaneously; that is to say they

http://dx.doi.org/10.1016/j.jiec.2015.04.019
1226-086X/ 2015 The Korean Society of Industrial and Engineering Chemistry. Published by Elsevier B.V. All rights reserved.

M.S. Karambeigi et al. / Journal of Industrial and Engineering Chemistry 29 (2015) 382391

require external energy to be produced [1012]. According to the

concept of capillary number, emulsion ooding may be a unique
chemical EOR approach since both favorable microscopic and
macroscopic displacement efciencies are provided simultaneously. In other words, reduction of interfacial energy induced by
emulsion phase as well as non-Newtonian behavior of displacing
front are available concurrently. Therefore, emulsion ooding has
become one of the promising chemical EOR techniques in recent
years [1315].
Selection of hydrocarbon phase has been one of the major
challenges of emulsion/microemulsion ooding. It is of vital
importance as a large volume of emulsion slug is necessary to be
injected into the reservoir. Accordingly, to select and also provide
the hydrocarbon phase, it is imperative that different criteria be
considered such as: stability at reservoir conditions, high
efciency, prevention of formation damage through appropriate
particle size distribution, availability, cost effectiveness and
environmental compatibility as well. Regarding these factors,
different sources have been proposed in the literature: CO2 [16]
and supercritical CO2 [17], gear oil [18], used engine oil [19], palm
kernel oil [20], and residual oil in place [9].
In this work, diesel has been suggested as the hydrocarbon
phase with outstanding performance. The objective of current
study is to investigate the EOR efciency of emulsion ooding
prepared by diesel. For this purpose, a 3-stage workow was used:
rst of all, phase behavior tests were systematically planned using
design of experiments (DOEs). Then, optimum formulation of
emulsion was selected according to statistical analysis of the
responses of designed experiments. Following this, it was
characterized as the second stage in terms of rheological behavior
and particle size distribution. Finally, the optimum mixtures were
injected into a micromodel fabricated using a developed compositional method as well as a carbonate core in order to evaluate the
efciency of emulsion ooding for chemical EOR.
Phase behavior study of surfactant/hydrocarbon/brine is a set of
tests to screen and optimize the formulation of emulsion slug and
then to interpret the amount of oil recovered by optimized
emulsion ooding. It is the single most critical factor assuring the
success of any chemical ooding containing surface active
components [2123]. Generally, it is controlled by several
parameters such as the properties of the surfactant as well as
the oil phase [2427], addition of electrolyte [28,29] especially
divalent ions [25], presence of co-surfactant or co-solvent [25,30],
water/oil ratio (WOR) [31], temperature [3234], pH [35], and
pressure [3638]. A systematic change in effective parameters
during formulation can result in the most stable emulsion
(optimum mixture) being injected into candidate reservoir.
Of the vast number of investigations on the surfactant phase
behavior, only a few researches have been focused on the
systematic interaction of parameters. In this study, response
surface methodology (RSM) has been proposed to study the phase
behavior of sodium dodecyl sulfate (SDS)/brine/diesel system
based on the interaction of selected parameters in addition to their
single effects. RSM consists of a collection of mathematical and
statistical techniques. It can be well utilized when a response or a
set of responses of interest are inuenced by several associated
variables. As a statistically designed experimental protocol, it is
used for designing experiments, developing functional relationship between a response and a number of related factors through
analyzing their interactions and optimizing the levels of these
factors to attain the best process performance [3942].
Background of emulsion ooding
McAuliffe was apparently the rst person to propose dilute and
stable emulsions as mobility control agents for enhanced oil

383

recovery (EOR) and from then on many studies have been

conducted, especially in recent years to achieve a better description of this process [43]. Abdul and Farouq Ali combined an
emulsion slug as blockage agent and a polymer solution as
mobility control agent to modify the drawbacks of water ooding
in reservoirs suffering from water leg problems [44]. Cobos et al.
carried out the ow experiments of oil-in-water emulsions in
quartz micro-capillary tubes. The pore-throat mechanism of
blockage was observed in experiments, and therefore changes in
local uid mobility would explain additional oil recovery obtained
by emulsion ooding [45]. Mandal et al. characterized the oil
water emulsions produced by gear oil as hydrocarbon phase using
physicochemical properties and size distribution of dispersed oil in
water phase [18].
Rocha De Farias et al. investigated the potential of oilwater
emulsions for EOR applications of viscous oils. The hydrocarbon
phase was the residual oil in place. They compared emulsion
ooding with both water ooding and surfactant ooding [9]. Fu
and Mamora studied EOR application of emulsions generated by
engine oil as hydrocarbon phase [19]. Guillen et al. carried out pore
scale ow visualization experiments as well as core ooding tests
in parallel sandstone cores having different permeabilities to
understand mechanism(s) of incremental oil recovery achieved by
emulsion injection. The results showed emulsion ooding offers
multi-scale mechanisms which are microscopic displacement
efciency and conformance sweep efciency [13]. Guillen et al.
introduced a selective mobility control mechanism of emulsion
droplets driven by capillary forces so that strong interfacial forces
(low capillary number) have signicant effect in lowering down
the mobility of emulsion containing larger drops compared to the
diameter of throats. On the other hand, displacing uid mobility is
only slightly inuenced by emulsion drops in high capillary
numbers [46]. Moradi et al. evaluated the efciency of emulsion
injection to improve oil recovery. Furthermore, single phase ow
experiments were carried for better understanding of emulsion
ow in pore level. They showed blockage-release mechanism for
pore scale dynamics which is dependent on the emulsion drop size
as well as local capillary number [15].
Experimental
Materials
In general, there are at least four compositional variables in an
emulsion ood: water, salt, hydrocarbon as well as emulsifying
agent. Distillated water was provided by reverse osmosis method
(New Human Power I). The brine was prepared by dissolution of
sodium chloride (NaCl, Merck Co.) salt in distilled water. Diesel was
proposed as hydrocarbon phase. At ambient temperature (about
25 8C), its density was 0.8394 g/ml and its viscosity (DBRobinson
electro-magnetic viscometer) was measured as 5.15 cP.
An anionic surfactant, sodium dodecyl sulfate (SDS, purity >
85%) was used for emulsication process. It was received from
Merck Co. with molecular weight of 288.38 g. Critical micelle
concentration (CMC) of this surfactant was reported as 8  103 M
[47]. Its surface tension at CMC and ambient temperature (about
25 8C) was measured as 34.6 mN/m using Du Nouy ring method
(Kruss K10). Emulsion slugs of ooding experiments were
prepared according to the optimum formulation resulted from
RSM.
Crude oil for dynamic displacement experiments was prepared
from an Iranian offshore oileld in the Persian Gulf. Its properties
were 8API of 30.9, density of 0.8675 g/ml, and viscosity of 19.0 cP at
ambient temperature (about 25 8C). The basic uids for water
ooding in microvisual experiments and core-ood test were
distillated water and synthetic brine (distillated water containing

M.S. Karambeigi et al. / Journal of Industrial and Engineering Chemistry 29 (2015) 382391

384

3.5% NaCl), respectively. Finally, 15% NaCl was dissolved in

distillated water to prepare synthetic formation water.
Soda lime glass was utilized for fabrication of micromodels. A
carbonate core sample, 72.0 mm  38.7 mm (length  diameter)
was used for core-ood test. Accordingly, gas permeability and
porosity values were 0.89 md and 13.55%, respectively.
Methods
Emulsion preparation
According to the design of experiments, various formulations of
emulsions were prepared using the following procedure: at rst,
salt was dissolved in distillated water. Then, surfactant was added
to prepare aqueous phase. Before mixing, both aqueous and
hydrocarbon phases were separately placed in the oven for at least
60 min to equilibrate thermodynamically. Thereafter, 10 ml of
aqueous solution was mixed with 10 ml of hydrocarbon phase
(WOR = 1) using magnetic stirring (IKA, RH basic 2) at 2000 rpm for
60 min accompanied by heating near desired temperature. Final
mixture was rapidly transferred into glass tube. The end of the test
tube was sealed and the tube was stored in the oven. Phase
behavior tests included careful observation of variations in volume
of phases over a sufciently long time to reach stability. Both
temperatures of oven and stirring were set to the temperature of
each run.
Experimental design and optimization
To achieve the best results for micellar ooding, the composition of emulsion was optimized using RSM. The most popular
response surface method is the central composite design (CCD). It
consists of the following runs: (1) a 2k factorial design; (2) a 2k
axial design and (3) x0 number of center points, where k is the
number of independent factors. CCD was applied using DesignExpert1 software (version 7.0.0). The center points are designed to
determine the experimental error and reproducibility of data. In
the optimization process, response(s) can be easily correlated to
independent variables by a polynomial function which contains
quadratic terms according to the following equation:

Y b0

k
X

k X
k
X

i1

i1 j1

bi X i

bi j X i X j

(1)

where Y is the predicted response, k is the number of inputs

(factors), b0 is the model constant, bi is the linear coefcient, bij is
the quadratic or interaction coefcient, and Xi as well as XiXj are
single and interaction factors, respectively.
With regard to the design of experiments, three control factors
were selected as surfactant concentration, salt content, and
temperature. Moreover, the response was considered as the
volume of stable emulsion phase. The range of selected parameters
and specic code for each one are presented in Table 1.
In this study, 18 phase behavior tests were designed using
central composite approach. A 23 fractional factorial design with
2  3 axis points and 4 replicates at the center points leading to
eighteen phase behavior tests were planned and performed. The

Table 2
Design of experiments and corresponding results.
Run

Variable A:
salinity (%)

Variable B: surf.
concentration
(CMC)

Variable C:
temperature
(8C)

Response:
emulsion
volume (cm3)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

0.71
2.79
1.75
1.75
3.5
2.79
0.71
1.75
0
1.75
0.71
2.79
1.75
2.79
1.75
0.71
1.75
1.75

4.19
1.81
5.00
3.00
3.00
1.81
1.81
3.00
3.00
1.00
4.19
4.19
3.00
4.19
3.00
1.81
3.00
3.00

33.1
56.9
45.0
45.0
45.0
33.1
33.1
65.0
45.0
45.0
56.9
56.9
45.0
33.1
45.0
56.9
45.0
25.0

12.8
0.0
11.0
11.2
0.0
0.0
12.2
10.4
11.8
10.6
10.6
11.0
11.2
12.2
11.0
11.2
11.0
12.2

matrix of variables and corresponding responses are provided

in Table 2. The experimental tests were implemented according
to the phase behavior procedure and then obtained data was
statistically analyzed by analysis of variance (ANOVA). The
contributions and interactions of factors were estimated by tting
a quadratic model to generated dataset. The developed mathematical model was utilized to optimize the process. The best
(optimal) formulations were nally validated by performing the
conrmation tests.
Characterization of optimum emulsion phase
Optimal mixture of emulsion was characterized using
standard procedures. Particle size distribution was determined
by dynamic light scattering (SEMATech Laboratory, SEM-633) at
25 8C according to ASTM C1070-01. In this light scattering
apparatus monochromatic wavelength of polarized light beam
(Helium Neon laser lamp) was 632.8 nm in vacuum and the
average frequency of light was 120.57 kHz.
Then, rheology measurements were performed at extreme
levels of temperature by Anton Paar rheometer (MCR 300) tted
with a cone and plate geometry. A wide range of shear rates was
considered.
Micromodel
The pattern of micromodel was a thin-section image of a real
reservoir rock [48]. In this paper, the technique for micromodel
fabrication was mainly comprised of laser ablation and chemical
etching. The detailed procedure has been explained elsewhere
[49]. The physical properties of glass micromodel are presented
in Table 3, in which aspect ratio is dened as the average
diameter of pore bodies to throats. Moreover, coordination
number is the number of throats connecting the adjacent pores to
each other.

Table 1
Level of variables in the experimental design.
Factor

Low axial
(a = 1.68)

Low factorial
(1)

Center (0)

High factorial
(+1)

High axial
(+a = +1.68)

A: Salinity (%)
B: Surfactant concentration (CMC)
C: Temperature (8C)

0
1
25

0.71
1.81
33.1

1.75
3
45

2.79
4.19
56.9

3.5
5
65

M.S. Karambeigi et al. / Journal of Industrial and Engineering Chemistry 29 (2015) 382391

385

Table 3
Physical properties of glass micromodel.
Pattern type

Dimensions
(mm2)

Average depth
(mm)

Porosity
(%)

Permeability
(mm2)

Pore volume
(cc)

Aspect ratio

Coordination
number

Heterogeneous

60  60

60  10

50

6.25  105

0.108

1.66

26

Microvisual experiments
The micromodel was horizontally mounted in the holder to
minimize the gravity effect. It was vacuumed and saturated with
water. Crude oil was then injected (Vinci-D475 pump) to displace
water and establish irreducible water saturation (Swi) condition.
Water was ooded at an injection rate of 0.001 cm3/min (capillary
dominant regime) to achieve residual oil saturation (Sor).
Thereafter, emulsion slug having optimum mixture at 25 8C (low
axial level) was injected to mobilize the residual oil. System of
video capturing (Dino-Lite Premier Digital Microscope AM-4013
TL) provided live visualization of ow during dynamic displacements. At the end of each stage, high resolution images were
captured and analyzed by PhotoShop (CS 8) to calculate the
variations of the phase saturations.
Tertiary oil recovery test
Core-ood experiment was carried out to investigate the
residual oil recovery achievable through emulsion ooding in a real
reservoir rock. To prepare the core, it was rst thoroughly cleaned
by distillation-extraction method in Sohxlet column containing a
mixture (70:30) of ethanol and toluene (both industrial grade) for
1 month. The core was then stored in the oven (temperature above
100 8C) for ve days to make sure of complete drying. Thereafter, it
was horizontally placed in the core holder and conning pressure
was exerted to make a 1D frontal displacement along the sample.
The system was then vacuumed for at least 8 h and, subsequently,
fully saturated with synthetic formation brine and soaked
overnight to ensure the establishment of ion equilibration between
brine and rock surface. Following this, liquid permeability
measurements were performed at different rates and brine
permeability was calculated according to Darcy equation:
Q

KA d p
m dx

(2)

where Q is the ow rate (cm3/s), K is the liquid permeability (D), A

is the area of the core cross-section (cm2), m is the liquid viscosity
(cP) and dp/dx is the pressure gradient (atm/cm).
In the drainage step, the crude oil was injected in different rates
started from 0.05 cm3/min for an extended period of time to drain
the formation water. Injection rate was then increased moderately
to 0.1 and 0.2 cm3/min, respectively and nally the core was
bumped to the high rate of 2 cm3/min to achieve the lowest
saturation of irreducible water. For uniform distribution of uids in
the core, this step was repeated in the reverse direction.
Eventually, irreducible water saturation decreased to 2%.
Water ooding was performed at 65 8C (high axial level) to
imbibe original oil in place. Injection rate (0.05 cm3/min) was
selected near the conventional front velocity in the reservoirs
(typically 1 ft/d). Efuent uids were collected in the glass sample
tubes. A droplet of demulsier (industrial grade) was added to
every sample tube for better separation of water and oil phases.
Then, sample tubes were sealed and placed in a water bath (60 8C)
for several hours and centrifuged at 4500 rpm (Hettich EBA 8S) for
1 h. Finally, volumes of water and oil were measured and written to
calculate the variations of uid saturations in the drainage and
imbibition processes, respectively.
Water ooding was continued until oil production became
negligible (oil cut < 1%). Then tertiary oil recovery was commenced

by injection of an optimum emulsion slug (0.25 PV) at elevated

temperature and consequently water ooding was repeated to
evaluate EOR efciency induced by emulsion.
Results and discussion
Modeling of emulsication process
18 designed experiments according to CCD methodology were
done and the corresponding responses were imported into Table
2. Then, RSM was adopted to model and optimize the best
formulation of emulsion required for displacement tests. It was
done via statistical analysis of experiments using different control
parameters which were tabulated in Table 4 (column 2). A
quadratic model was tted to experimental data. According to the
results of ANOVA, the F value and prob > F of 24.70 and
0.0001 indicated the quadratic model is statistically signicant
in 99% condence level (p-value <0.01). R-squared (R2) as
coefcient of determination was calculated as 0.91, which
indicates how well the quadratic equation ts experimental data.
In general, the higher the R2, the better the model ts the data.
However, it is possible that tted models yield a poor estimation of
the new data although they have large R2 values because the
increase of R2 is dependent on the addition of variables, regardless
of their signicance or insignicance. To solve this problem, R2 is
preferentially replaced with adjusted R2(R2adj ) which does not
always have an increasing trend as variables are added to the
model. Generally, R2adj is often decreased when insignicant
(unnecessary) terms are added [50]. In this model, the values of
R2 and R2adj were reported as 0.91 and 0.87, respectively which
emphasize a satisfactory adjustment of the quadratic equation to
experimental data. Another parameter of ANOVA table is adequate
precision, which is signal to noise ratio. It was found to be
15.21. Adequate model discrimination is resulted from ratios
greater than 4. In this case, the model is able to navigate the design
space very well.
Reproducibility of the model is measured by coefcient of
variance (CV). It is dened as standard deviation expressed as a
percentage of mean value. The CV for emulsication process was
calculated as 16.48%. As a general rule, the values less than 10%
imply that the developed models are reasonably reproducible.
Measured CV for current quadratic model, however, was above the
criterion since experimental response has deviated considerably
from mean value due to zero responses of three experiments (runs
2, 5 and 6 in Table 2).

Table 4
The results of ANOVA for the response surface models.
Statistical results

Central composite
design

Historical data
design

Model F value
Model prob > F
Lack of t F value
Lack of t prob > F
R-squared
Adjusted R-squared
CV%
Adequate precision

24.70
<0.0001
243.01
0.0004
0.91
0.87
16.48
15.21

14.66
0.0043
8.16
0.0612
0.96
0.90
2.00
11.87

386

M.S. Karambeigi et al. / Journal of Industrial and Engineering Chemistry 29 (2015) 382391

Removing these 3 runs, a new historical data RSM model was

developed for further investigations. In such models, unlimited
design points can be dened using all or some of the available data.
Moreover, arrangement of factors (number of levels and their
quantity) can be adjusted according to the experimental data, in
contrast to CCD design that has xed ve levels. The statistical
analysis of this design is presented in Table 4 (column 3).
Compared to base CCD model, CV% of the new design reduced to
2.00. The way in which zero-result cases affected on the statistical
analysis of CCD approach is indicated.
The salinity of zero-result runs was higher than other ones.
Moreover, these runs had low concentration of surfactant. In such
conditions, monomer of surfactant in the micelle can be replaced
by ions of electrolyte. Thereafter, electrostatic repulsion between
micelles decreases and van der Waals forces dominate, which
result in occulation and consequently emulsion breakdown.
Regardless of CV%, it was preferred to persist with further
investigations using CCD design because zero-result runs are as a
part of the real process studied. Based on the results of statistical
analysis, the most suitable quadratic model is given by the
following equation:
V me 20:84  3:95A  2:63B  0:05C 2:34AB  1:75A2

(3)

where Vme is the emulsion volume, A, B and C are coded values

pertaining to the salinity, surfactant concentration and temperature. The p-values (prob > F) of coefcients determine their
signicance. Value less than 0.01 indicates that corresponding
coefcient is signicant in 99% condence level. Terms include: the

independent variables of A (salinity), B (surfactant concentration),

the interaction variable of AB (salinity  surfactant concentration)
as well as quadratic variables of A2 were considered as signicant
parameters of the model. The highest absolute coefcients of terms
A, AB and A2, respectively among the independent, interaction and
quadratic terms show their crucial role in emulsion stability. In
other words, salinity is the most sensitive factor in emulsication
of SDS surfactant which is in good agreement with previous studies
[51].
Fig. 1 presents the normal probability plot for the quadratic
model. The plot of the residuals illustrates a normal distribution
supporting adequacy of the least-squares t because most the
points follow a straight line. Therefore, it is a suitable model to
predict the emulsion volume of SDS surfactantdieselbrine
system. Moreover, it will be used to nd the optimum conditions
for emulsication process.
As mentioned before, RSM facilitates the ability of studying
interactions which is essential for better description of the process
behavior. According to Eq. (3), interaction of salinity and surfactant
concentration was recognized as meaningful because, as it is shown
in Fig. 2, the response was different depending on the settings of
these factors. In other words, the concentration of surfactant can
inuence the salinity scan of emulsion phase as if in low
concentration of surfactant (1.81 CMC), emulsion volume severly
reduced with increase in salinity. By contrast, high concentration of
emulsing agent (4.19 CMC) could compensate the negative effect
of salinity on emulsion stability. Presence of electrolyte cannot
effectively increase van der Waals forces of attraction between
multi-layer micelles formed in high-concentrated surfactant

Fig. 1. Normal probability plot of RSM model.

M.S. Karambeigi et al. / Journal of Industrial and Engineering Chemistry 29 (2015) 382391

387

Fig. 2. Intearction curve of salinity and surfactant concentration.

solution, however, molecules of electrolyte can be placed next to the

surfactant monomer produced by a one-layer micelle of dilute
surfactant solution, increasing the van der Waals forces of attraction
between two adjacent micelles and consequently resulting in
occulation. Therefore, the stability of emulsion containing low
concentration of surfactant reduces more sharply with increment of
salt content.
Two-dimensional contour and second-order response surface
plots for the stable emulsion phase as a function of salinity and
surfactant concentration are presented in Fig. 3. They show the joint
effects of important factors (A and B) at xed temperature of 45 8C
(level 0). The color legend in these curves shows the variation of
the volume of emulsion. The blue and red colors are representative
of the lowest and highest values, respectively. The contour plot

demonstrates that as surfactant concentration increases and

salinity decreases, the higher volume of emulsion is produced.
The 3-d plot of Fig. 3 indicates there is a nearly linear relationship
between emulsion volume and surfactant concentration, while
salinity has a nonlinear effect on the emulsiction process. As
can be seen from Figs. 2 and 3, maximum volume of emulsion
corresponds to the low salinity and high concentration of
surfactunt.
Optimization of emulsion phase
Developed model (Eq. (3)) showed that emulsion volume is
strongly inuenced by surfactant concentration, salinity, temperature and their important interactions. Therefore, optimization of

Fig. 3. Contuor and surface plots of emulsion volume versus salinity and surfactant concentration.

M.S. Karambeigi et al. / Journal of Industrial and Engineering Chemistry 29 (2015) 382391

388

Table 5
Optimum conditions of slugs required for micromodel (1) and core (2) tests.
Emulsion slug

Salinity (%)

Surfactant
concentration (CMC)

Temperature (8C)

Optimized emulsion
volume (cm3)

Real emulsion
volume (cm3)

1
2

1.93
0.71

5
1.81

25
65

15.0
12.4

15.3
12.2

these parameters can improve the efciency of emulsion ooding

for EOR applications. The RSM approach designs a sequence of
experiments to determine the optimum response. The following
goal factors were implemented in software to identify the criteria
of the RSM optimization: rst, In range was set for surfactant
concentration and salinity. Second, equal to 25 and equal to 65
were assigned as the goal factors of temperature for microvisual
and core ooding experiments, respectively. They were the
minimum and maximum levels of the temperature factor.
Finally, the goal of response optimization was set to maximize.
According to numerical optimization, the most appropriate
formulations for different objectives are presented in Table 5.
Then, two phase behavior experiments using optimum arrangement of factors were run to conrm the optimization results as
well as to supply the emulsion slug for displacement experiments.
The volume of emulsions measured from conrmation tests was
15.3 and 12.2 cm3, respectively, which was in complete agreement
with the proposed optimum mixture data in the considered
condence interval (99%).
Characterization of the optimum formulations
Particle size distribution
The colloidal interaction between the dispersed particles,
emulsion stability, and mobility through pore throats are
inuenced by particle size so that they increase when the size
decreases [52]. Particle size distribution of dispersed phase in
optimum formulation 1 is shown in Fig. 4.
The mean particle diameter (dn ) was calculated as 234.6 nm
using Eq. (4):
P
Nd
(4)
dn P i i
Ni
where di is the midpoint of class size (nm) and Ni is the
corresponding frequency (%).
Polydispersity index quanties distribution width and was
calculated as 0.318. It means a mid-range value between

monodisperse (near 0.05) and very broad (near unity) distributions. Various size distribution algorithms operate best over midrange indexes (between two extremes).
Diffusion coefcient was measured as 2.96  108 cm2/s. Using
StokesEinstein relation (Eq. (5)), hydrodynamic radius (Rh) of
dispersed phase was calculated as 76.0 nm. In practical applications, dispersed particles in continuous phase do not have
spherical shapes. For this reason, the hydrodynamic radius
calculated from the diffusional characteristics of the particle will
be representative of the apparent size of the dynamic particle.
Rh

kB T
6pm0 D

(5)

where kB is Boltzmanns constant, T is the absolute temperature, m0

is the dynamic viscosity of uid and D is the translational diffusion
coefcient.
Rheology tests
Rheological analysis of optimum formulations was carried
out. Results as shown in Fig. 5 indicate that these emulsion
mixtures exhibit shear thinning behavior, i.e. their viscosity (m)
decreases with increasing shear rate g . A power law model (Eq.
(6)) was tted and after that ow behavior index (n) as well as ow
consistency index (k) were estimated (Table 6). It is obvious that
ow behavior index decreases at in higher temperature which has
a good agreement with the literature [53]. Non-Newtonian
behavior of emulsion slug is a practical indicator because makes
it a suitable candidate for front mobility control.

mapp kg n1

(6)

Micromodel experiments
Microvisual study of emulsion ooding was done in the
heterogeneous micromodel to investigate displacement efciency
in pore level. The images of Fig. 6 were captured exactly at the end

50

Intensity (%)

40
30
20
10
0

100

200

300

400

500

600

700

Mean size (nm)

Fig. 4. Intensity wise particle size distribution of optimim formulation.

800

M.S. Karambeigi et al. / Journal of Industrial and Engineering Chemistry 29 (2015) 382391

10
10
Viscosity @ 25C
Shear stress @ 25C

1
0.01

5
0

100

Shear rate (1/s)

20

1000

15
100
10
10
Viscosity @ 65C
Shear stress @ 65C

1
0.01

Shear stress (Pa)

15
100

Viscosity (cp)

20

1000

25

10000

25

Shear stress (Pa)

Viscosity (cp)

10000

389

100

Shear rate (1/s)

Fig. 5. Rheological tests of emulsion phase with optimum formulations.

Table 6
Fitted parameters of power law model for optimum formulations.
Emulsion slug

Temperature (8C)

k (Pa sn)

Model R2

1
2

25
65

236.7
341.2

0.64
0.49

0.999
0.982

of the different steps of microvisual experiments. Each image as

the overall picture of micromodel (60 mm  60 mm) shows the
distribution of initial or residual oil saturation after different
processes. However, the images of Fig. 7 are close-up (zoomed)
pictures of Fig. 6C. They indicate the efcacy of emulsion ooding
from pore-scale point of view.
Fig. 6A illustrates the micromodel with initial oil saturation.
Green arrows show the direction of front ow. The baseline test was

water injection. Residual oil distribution at the late stage of water

oodings is presented in Fig. 6B. The recovery of water ooding was
calculated as only 29%. In other words, the residual oil saturation
was 71% which indicates that water has poorly swept the oil. Then,
emulsion slug 1 (Table 5) as the optimum formulation prepared for
micromodel experiment was injected continuously. Fig. 6C shows
oil production response at the end of emulsion ooding. The
recovery factor was increased as much as 47% of original oil in place
(OOIP). The comparison of these images can conrm the improvement of macroscopic sweep efciency due to the emulsion injection.
A close-up picture of oil solubilization due to the injection of
emulsion is presented in Fig. 7. The gradient feature of brown color
(oil) after emulsion ooding illustrates the favorable interfacial
potential of emulsion in comparison with water to partially
solubilize oil in contact front. The comparison between the surface
tension of emulsion slug (35.7 mN/m) and water (71.3 mN/m) at
25 8C is a preliminary indicator of favorable surface properties of

Fig. 6. Micromodel images (A) at initial oil saturation, (B) after water ooding and (C) after emulsion ooding; green arrow: ow direction. (For interpretation of the references
to color in this gure legend, the reader is referred to the web version of this article.)

Fig. 7. Visualization of oil solubilization by emulsion front in pore scale (yellow arrows). (For interpretation of the references to color in this gure legend, the reader is
referred to the web version of this article.)

390

M.S. Karambeigi et al. / Journal of Industrial and Engineering Chemistry 29 (2015) 382391

emulsion slug for EOR applications. Furthermore, the surface

tension of emulsion phase was approximately close to that for SDS
surfactant (34.6 mN/m). It means that the emulsion phase has
pursued the surface properties of surfactant from which it has been
produced although its rheological behavior is more favorable
than any SDS surfactant slug. In other words, it is in good agreement
with the literature that emulsions have been nominated for
conformance control rather than the improvement of microscopic
sweep efciency which is the primary role of microemulsions owing
to the fact that they show ultra-low interfacial properties.
Core-ood test
To investigate the potential of emulsion for chemical EOR, a
core ooding experiment was carried out using an optimum
formulation (emulsion slug 2 at Table 5). Brine was rst ooded
into the carbonate core, then a tertiary emulsion ood chased with
a water ood was conducted to compare incremental oil recovery
and ood performance.
The variation of cumulative oil recovery as a function of pore
volume (PV) of uids injected in secondary and tertiary modes is
depicted in Fig. 8. Moreover, pressure drop across the core was
acquired and tted with oil recovery events in this gure. The
ooding procedure was commenced by initial water injection. The
differential pressure was 95.67 psi at breakthrough point
(0.55 PV). At the end of water ooding, it uctuated around
67.21 psi. Ultimate oil recovery was 57% of OOIP by water ooding.
The viscosities of brine and oil at elevated temperature (65 8C)
were measured as 0.55 and 4.45 cP, respectively. This distinct
difference causes poor mobility of brine front during ooding.
When the recovery plot of water ooding reached a plateau, which
indicates the establishment of residual oil saturation in the core,
0.25 PV of emulsion slug was then injected and chased with water.
The pressure drop increased to 123.21 psi after injection of
emulsion slug. Then, it fell drastically during initial stage of
post-ush water injection, and nally, reached a plateau of about
75.60 psi. Analysis of pressure difference during core-ood test
indicates that the injectivity of emulsion in this low-permeable
core was favorable and any formation damage was not observed
during core-ood test. The tertiary scenario of emulsion ooding
recovered 28% incremental oil.
A dominant uctuation in the trend of pressure drop observed
during emulsion ooding followed by post-ush (Fig. 8) can easily

prove the potential of emulsion front to reduce the ow capacity of

some channels through pore blockage mechanism [13,46] and to
divert the injection front, as well as favorable mobility of emulsion
slug to displace oil banks. Furthermore, the interfacial properties of
emulsion mixture may lead to additional improvement of capillary
number and a corresponding decrease in the capillary pressure,
mobilizing and then producing the trapped ganglia through the
formation of oil bank. The connection of such discontinuous oil
droplets is essential to form a ow channel in water-wet rocks. The
results of phase behavior studies revealed that the almost
produced emulsions were water-in-oil emulsions. Previous
investigations concluded these emulsions have higher performance than oil-in-water emulsions for chemical EOR because they
behave similarly to a miscible displacement [54]. The gradual
change of oil color in the microvisual experiment of emulsion
ooding (Fig. 7) is a good indicator of this phenomenon. Therefore,
incremental oil recovery may coincide with the improvement of
both macroscopic sweep and microscopic displacement efciencies in the studied porous media.
The optimistic result of micromodel and core ooding tests
emphasize the outstanding performance of emulsion phase with
optimum formulation. The potential of emulsion solution compared to water for simultaneous improvements of interfacial
properties and rheological behaviors is very interesting for
chemical EOR, especially in carbonate rocks which have complex
nature accompanied by slightly low permeabilities. Active surface
of carbonates can increase adsorption of surfactant or cosurfactant
molecules on the rock [55]. Therefore, emulsion ooding has three
obvious priorities:
1. The adsorption of emulsion on the rock surface is lower.
2. Emulsion phase yields favorable interfacial properties compared
with simple surfactant solutions [52].
3. Emulsion has better tolerance for brine and divalent ions [56].
From rheological point of view, non-Newtonian behavior of
emulsion phase seems to be applicable for the improvement of
sweep efciency. Furthermore, appropriate particle size distribution of emulsion solution compared to typical pore size distribution of carbonate rocks prevents potential damage of low
permeability reservoirs which is known as a common problem
in conventional conformance control methods such as polymer
ooding. Moreover, there are other difculties with application of

Fig. 8. Recovery plot of core-ood experiment (DP, differential pressure).

M.S. Karambeigi et al. / Journal of Industrial and Engineering Chemistry 29 (2015) 382391

polymers such as in situ control of gelation time and complicated

interaction of polymers and surfactants in porous media [57,58]. As
a result, recommendation of emulsion phase for chemical EOR
applications can overcome some distinct drawbacks of both
surfactant and polymer slugs.
Conclusions
 Response surface methodology has been proposed for statistical
modeling and optimizing the phase behavior of surfactants to
achieve the best formulation of emulsion required for displacement ooding. An expeditious and reliable approach for optimal
designing of chemical EOR methods was found.
 A 3-stage workow for evaluation of emulsion ooding efciency
was developed and utilized consisting of the optimization of
surfactant phase behavior using design of experiments (response
surface methodology), characterization of optimal formulation
through rheological analysis and determination of particle size
distribution, and nally micromodel and core ooding experiments.
 The most signicant factors resulted from statistical analysis of
designed experiments were as follows: salinity, concentration of
surfactant and their interaction as well. This joint effect can
inuence van der Waals forces of attraction enabling them be
dominant, and therefore coalescence occurs, decreasing the
stability of the emulsion system.
 Selection of diesel as hydrocarbon phase in the system of
emulsion slug showed optimistic results. It seems that diesel can
be a good candidate as hydrocarbon component of emulsions
prepared for chemical EOR applications.
 The simultaneous mechanisms containing non-Newtonian
behavior of emulsion and favorable interfacial potential are
responsible for increasing capillary number and incremental oil
recovery observed in displacement efciency tests.

Acknowledgements
The authors are grateful to the National Iranian Oil Company
(Research and Technology Directorate) for the nancial support.
The authors are also thankful to Stat-Ease, Minneapolis for the
provision of the Design Expert package.
References
[1] A.H. Bornaee, M. Manteghian, A. Rashidi, M. Alaei, M. Ershadi, J. Ind. Eng. Chem. 20
(2014) 1720.
[2] S. Park, E.S. Lee, W.R.W. Sulaiman, J. Ind. Eng. Chem. 21 (2015) 1239.
[3] A.C. Payatakes, Annu. Rev. Fluid Mech. 14 (1982) 365.
[4] S. Thomas, Oil Gas Sci. Technol. 63 (2008) 9.
[5] J.J. Sheng, in: J.J. Sheng (Ed.), Surfactant Flooding, Gulf Professional Publishing,
Boston, 2011, pp. 239335 (Chapter 7).
[6] R. Kumar, E. Dao, K.K. Mohanty, SPE Paper 129914, in: Presented at SPE Improved
Oil Recovery Symposium, 2428 April, Tulsa, Oklahoma, USA, 2010.
[7] Q. Liu, M. Dong, S. Ma, Y. Tu, Colloids Surf. A 293 (2007) 63.
[8] Q. Liu, M. Dong, X. Yue, J. Hou, Colloids Surf. A 273 (2006) 219.

391

[9] M.L. Rocha De Farias, A.L. De Souza, M. Da Silveira Carvalho, G. Hirasaki, C. Miller,
SPE Paper 152290, in: Presented at SPE Latin America and Caribbean Petroleum
Engineering Conference, 1618 April, Mexico City, Mexico, 2012.
[10] A.A. Dehghan, M. Masihi, S. Ayatollahi, Energy Fuels 27 (2013) 5852.
[11] H. Mollet, A. Grubenmann, Colloids, Phases, Interfaces, Wiley-VCH, Verlag GmbH,
2007, pp. 158.
[12] T.F. Tadros, Emulsion Formation, Stability, and Rheology, Wiley-VCH Verlag
GmbH & Co. KGaA, 2013, pp. 175.
[13] V.R. Guillen, M.S. Carvalho, V. Alvarado, Transport Porous Med. 94 (2012) 197.
[14] H. Mendoza, S. Thomas, S.M. Farouq Ali, Petroleum Society of Canada, PETSOC-9126, in: Presented at Annual Technical Meeting, 2124 April, Banff, Alberta,
Canada, 1991.
[15] M. Moradi, M. Kazempour, J.T. French, V. Alvarado, Fuel 135 (2014) 38.
[16] Y. Soong, D. Xing, B. Wei, R.M. Enick, J. Eastoe, A. Mohamed, K. Trickett, AICHE
Annual meeting, 09AIChE, Nashville, Nashville, USA, 2009.
[17] Z. Dong, Y. Li, M. Lin, M. Li, Pet. Sci. 10 (2013) 91.
[18] A. Mandal, A. Samanta, A. Bera, K. Ojha, Ind. Eng. Chem. Res. 49 (2010) 12756.
[19] X. Fu, D. Mamora, SPE Paper 129902, in: Presented at SPE Improved Oil Recovery
Symposium, 2428 April, Tulsa, Oklahoma, USA, 2010.
[20] Z. Jeirani, B. Mohamed Jan, B. Si Ali, I.M. Noor, C.H. See, W. Saphanuchart, J. Ind.
Eng. Chem. 19 (2013) 1304.
[21] A. Bera, A. Mandal, K. Ojha, T. Kumar, J. Chem. Eng. Data 56 (2011) 4422.
[22] A. Bera, K. Ojha, T. Kumar, A. Mandal, Colloids Surf. A 404 (2012) 70.
[23] M. Kahlweit, R. Strey, Angew. Chem. Int. Ed. Engl. 24 (1985) 654.
[24] R.N. Healy, R.L. Reed, Soc. Pet. Eng. AIME J. 14 (1974) 491.
[25] R.N. Healy, R.L. Reed, D.G. Stenmark, Soc. Pet. Eng. AIME J. 16 (1976) 147.
[26] M. Kahlweit, E. Lessner, R. Strey, J. Phys. Chem. 87 (1983) 5032.
[27] J.A. Silas, E.W. Kaler, R.M. Hill, Langmuir 17 (2001) 4534.
[28] H. Kunieda, K. Nakamura, J. Phys. Chem. 95 (1991) 8861.
[29] K. Thalberg, B. Lindman, G. Karlstrom, J. Phys. Chem. 95 (1991) 6004.
[30] M. Kahlweit, R. Strey, G. Busse, J. Phys. Chem. 95 (1991) 5344.
[31] A. Pizzino, V. Molinier, M. Catte, J.F. Ontiveros, J.L. Salager, J.M. Aubry, Ind. Eng.
Chem. Res. 52 (2013) 4527.
[32] M.G. Aarra, H. Hiland, A. Skauge, J. Colloid Interface Sci. 215 (1999) 201.
[33] R.K. Mitra, B.K. Paul, J. Surf. Sci. Technol. 20 (2004) 105.
[34] P. Schrader, A. Culaguin-Chicaroux, S. Enders, Chem. Eng. Sci. 93 (2013) 131.
[35] M. Kahlweit, G. Busse, B. Faulhaber, Langmuir 16 (2000) 1020.
[36] S.B. Sandersen, E.H. Stenby, N. von Solms, Colloids Surf. A 415 (2012) 159.
[37] C.L. Sassen, L.M. Filemon, T.W. De Loos, J. De Swaan Arons, J. Phys. Chem. 93 (1989)
6511.
[38] A. Skauge, P. Fotland, SPE Reserv. Eng. (Soc. Petrol. Eng.) 5 (1990) 601.
[39] M.A. Bezerra, R.E. Santelli, E.P. Oliveira, L.S. Villar, L.A. Escaleira, Talanta 76 (2008)
965.
[40] Z. Jeirani, B. Mohamed Jan, B. Si Ali, I. Mohd Noor, C.H. See, W. Saphanuchart, J. Ind.
Eng. Chem. 19 (2013) 554.
[41] A.I. Khuri, S. Mukhopadhyay, Wiley Interdiscip. Rev. Comput. Stat. 2 (2010) 128.
[42] A.R. Naja, M.R. Rahimpour, A.H. Jahanmiri, R. Roostaazad, D. Arabian, M. Soleimani, Z. Jamshidnejad, Colloids Surf. A 82 (2011) 33.
[43] C.D. McAuliffe, JPT J. Petrol. Technol. 25 (1973) 727.
[44] H.J. Abdul, S.M. Farouq Ali, J. Can. Pet. Technol. 42 (2003) 35.
[45] S. Cobos, M.S. Carvalho, V. Alvarado, Int. J. Multiphase Flow 35 (2009) 507.
[46] V.R. Guillen, M.I. Romero, M.D.S. Carvalho, V. Alvarado, Int. J. Multiphase Flow 43
(2012) 62.
[47] M. Karagianni, A. Avranas, Colloids Surf. A 335 (2009) 168.
[48] S.A. Farzaneh, A.A. Dehghan, M.H. Ghazanfari, R. Kharrat, J. Can. Pet. Technol. 51
(2012) 383.
[49] M.S. Karambeigi, M. Schafe, M.H. Fazaelipoor, Pet. Sci. Technol. 31 (2013).
[50] F. Amiri, S.M. Mousavi, S. Yaghmaei, M. Barati, Biochem. Eng. J. 67 (2012) 208.
[51] A. Bellocq, Handbook of Microemulsion Science and Technology, Marcel Dekker,
New York, 1999p. 139.
[52] A. Bera, T. Kumar, K. Ojha, A. Mandal, Fuel 121 (2014) 198.
[53] M. Marcotte, A.R. Taherian Hoshahili, H.S. Ramaswamy, Food Res. Int. 34 (2001)
695.
[54] M.F. Destefanis, G.B. Savioli, J. Phys. Conf. Ser. 296 (2011).
[55] Z. Jeirani, B.M. Jan, B.S. Ali, C.H. See, W. Saphanuchart, Pet. Sci. Technol. 32 (2014)
180.
[56] J.H. Bae, C.B. Petrick, R. Ehrlich, SPE Paper 4749, in: Presented at SPE Improved Oil
Recovery Symposium, 2224 April, Tulsa, Oklahoma, 1974.
[57] B. Bai, L. Li, Y. Liu, H. Liu, Z. Wang, C. You, SPE Reserv. Eval. Eng. 10 (2007).
[58] B. Bai, Y. Liu, J.P. Coste, L. Li, SPE Reserv. Eval. Eng. 10 (2007).