-
Heavy Crude Oils/Particle Stabilized Emulsions
. . .
. . .3.2.1. Particle stabilized emulsio
le-stabilsionspitatedmodel. . .e diagraperties. . .. . .
Advances in Colloid and Interface Science 169 (2011) 106127
Contents lists available at SciVerse ScienceDirect
Advances in Colloid an
.e4. Conclusion . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . .Acknowledgments . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . .References . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
1. Introduction components are often categorized into the
group-type classes [1] of
126126126Heavy and extra heavy oils are strategicchallenges to
the oil industry. One major prcosity which makes them difcult to
tran
Corresponding author. Tel.: +47 73591605; fax: +E-mail address:
[email protected] (I. Kr
0001-8686/$ see front matter 2011 Elsevier B.V.
Alldoi:10.1016/j.cis.2011.09.001. . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .e
model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .lts . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1211221223.7.2. Notes on the lm drainag3.7.3. Separation
modeling resu3.7. Separation modeling . . .3.7.1. Model
description3.2.2. Rheology of partic3.3. Asphalthene stabilized
emu
3.3.1. Asphaltenes preci3.4. Polyaromatic surfactants as3.5. Wax
in crude oils . . . .3.6. Naphthenic acids and phas
3.6.1. Langmuir lm prons . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .lized
emulsions . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .by
multi-step precipitation procedure. . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .compounds for asphaltenes . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .ms . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
111112113115116116119120121reserves but pose newoblem is their
high vis-sport. Heavy crude oil
saturates, aromahigh-molecular-wbilize asphaltenobtained
duringseparation by pohydrocarbons anmolecular-weighare considered
b
47 73594080.alova).
rights reserved.. . . . . . . . . . . . . . . . . . . . . . . .
. . .109
3.1. Crude oil components . .3.2. Crude oil emulsions . . .. . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . .1082. Separation
technology . . . . . . . . . . . . . . . . . . . . . .3. Crude oils
and their emulsions. . . . . . . . . . . . . . . . . . .Iva Kralova
, Johan Sjblom, Gisle ye, Sbastien Simon, Brian A. Grimes,
Kristofer PasoUgelstad Laboratory, Norwegian University of Sciences
and Technology (NTNU), Department of Chemical Engineering, Sem
Slandsvei 4, N-7491 Trondheim, Norway
a b s t r a c ta r t i c l e i n f o
Available online 6 October 2011
Keywords:ElectrocoalescenceGelationNaphthenic
acidsParticle-stabilized emulsionsPhase diagramsSeparation
modeling
Fluid characterization is a key technology for success in
process design for crude oil mixtures in the future off-shore. In
the present article modern methods have been developed and
optimized for crude oil applications.The focus is on
destabilization processes in w/o emulsions, such as
creaming/sedimentation and occulation/coalescence. In our work, the
separation technology was based on improvement of current devices
to promotecoalescence of the emulsied systems. Stabilizing
properties based on particles was given special attention. Avariety
of particles like silica nanoparticles (AEROSIL), asphalthenes, wax
(parafn) were used. The behaviorof these particles and
corresponding emulsion systems was determined by use of modern
analytical equip-ment, such as SARA fractionation, NIR,
electro-coalescers (determine critical electric eld), Langmuir
technique,pedant drop technique, TG-QCM, AFM.
2011 Elsevier B.V. All rights reserved.
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . .
106107107j ourna l homepage: wwwd Interface Science
l sev ie r .com/ locate /c istics, resins, and asphaltenes
(SARA). The resins areeight polar hydrocarbons, which are known to
sta-
es in petroleum uids. Resin fractions are easilyhigh-performance
liquid chromatography (HPLC)lar solvent extraction. Asphaltenes are
polyaromaticd are often precipitated from crude oils using lowt
alkane solvents, such as pentane or hexane. Theyy most researchers
[2,3] together with resins to be
-
model system and the process parameters (ow rate, pressuredrop
and water cut). It was observed that increased pressuredrops
resulted in droplets with smaller mean diameter and amore narrow
size distribution and an increasing the water cut ata certain
pressure drop for a certain system gave an increase inthe mean
diameter of the droplets.
3. Crude oils and their emulsions
The proper understanding of the mechanisms and the
chemistrybehind the stability of crude oil based emulsions is an
essentialstep in crude oil processing. The stability mechanism is,
of course, re-lated to the composition of crude oil such as the
nature and the con-centration of surface-active components present
in oil, the physicalproperties of oil (viscosity, density for
instance) and the interfacialrheology of the interface around water
droplet which informs
107I. Kralova et al. / Advances in Colloid and Interface Science
169 (2011) 106127responsible for the high stability of
water/in/crude oil emulsions,mainly because of their capacity to
form a stable thick network atthe interface [4]. In addition, solid
particles [5] originating from res-ervoir adsorbed at the water/oil
interface may also contribute to thestability of the water droplets
together with waxes [6] and naph-thenic acids [7]. The combination
of these components creates acomplex picture of several
contributing mechanisms to the stabilityof water-in-oil emulsions
[3]. Co-production of brine and crude oiloften results in the
formation of stable water-in-oil emulsionswhen turbulent mixing
conditions are encountered during the trans-portation process. When
the crude oil is processed from the wellhead to the manifold, there
is usually a substantial pressure reduc-tion with a pressure
gradient over chokes and valves where the mix-ing of oil and water
can be intense. After this the well-stream isentering the separator
where most of the water is separated fromthe crude. The nal
treatment normally takes place in the electrocoa-lescer after which
the level of water should be below 0.5%. The pro-cessing of the oil
and water offers several possibilities to mix thephases and create
an emulsion [3]. The life time of emulsions de-pends on the kind of
stability mechanisms involved and compositionof interfacial
material. Due to the important role of solid particlesduring
formation of the interfacial lms, the presence and solubilitystate
of waxes, asphaltenes, resins and naphthenates is a key
factorinuencing emulsion stability and separation [8].
In order to understand the complex nature behind
water-in-oilemulsion stability under real process conditions, a
thorough knowl-edge should involve the properties of the crude oil
components,their association tendencies and accumulation at w/o
interfaces,their solubilities and their association structures in
water/oil systems.It is also our intention to connect these
investigations with new ex-perimental techniques which were applied
on crude oil systems pre-viously [4]. Such conventional techniques
are SARA fractionation, NIR,electrocoalescers (determine critical
electric eld), Langmuir tech-nique, pedant drop technique, TG-QCM,
AFM.
2. Separation technology
The separation process normally involves a combination of
me-chanical impact and chemistry to break w/o emulsions.
Separationinlets and internals have been developed to maximize the
resolu-tion of water at a reasonable residence time. The whole
separationtechnology has been pushed forward by introducing
electrostaticdevices facilitating the droplet growth and
coalescence of the emul-sied systems. For gravity separators the
settling and coalescenceprocesses, as well as the obstructions to
coalescence describedabove are valid. If a background electric eld
is present, as in indus-trial electrocoalescers, the kinematics of
droplets and the coales-cence will be completely different. Several
mechanisms which arepresent in gravimetrically-induced
destabilization of emulsions areprobably still present, but their
relative level of inuence may becompletely different due to forces
induced by strong electrical elds[9]. The electrostatic effects
arise from the very different propertiesof oil and water, water
having dielectric permittivity and conductiv-ity values much higher
than those of oil and polarization effects inwater droplets.
Electrocoalescence is a process targeted to assist ap-proach,
contact and fusion of water droplets emulsied in a contin-uous of
low dielectric permittivity in order to increase their size,thus
accelerating their settling velocity and the total separationtime.
Various designs of these electrocoalescers have been intro-duced as
the Vessel Internal Electrostatic Coalescer (VIEC) [10,11],the High
Temperature VIEC (HTVIEC Fig. 1) and the upcomingLow Water Content
Coalescer (LOWACC Fig. 2) [12].
The patented Compact Electrostatic Coalescer (CEC) technolo-gy
is a small lightweight ow-through system that greatly enhancesthe
separation performance of existing downstream gravity separa-
tion equipment. It enhances separation performance by
coalescingemulsied water droplets entrained in the crude oil into
much larg-er droplets that readily settle in a downstream
separator. The coa-lescing action takes place very rapidly under
turbulent owconditions, as the emulsion is subjected to an intense
electrostaticeld inside the CEC unit. Fig. 3 presents typical CEC
unit con-sisting of a series of concentric circular electrodes with
a capacityof 130000 barrels per day [13,14]. The CEC is insensitive
to vesselmotions, and is not prone to plugging by solids in the
well uids.Voltages of up to several thousand volts can be applied
to the elec-trodes. By doing so, an intense electrical eld is
established in theannular channels, so that the coalescing process
is much fasterthan what is normally achieved in conventional grid
units. Waterdroplets merge several times within a matter of seconds
and in-crease their size around ten times in the coalescing
section. Thewater droplets and oil then enter a gravity separator
for separation,but one with much reduced dimensions compared to a
normal elec-trostatic coalescer. Ugelstad Laboratory has together
with industrialpartners built up a Electrostatic Separation Unit
(ESU) which is adown-scaled CEC, although with another ow prole
(laminar)than the commercial CEC (turbulent ow), (Fig. 4) [15]. The
lengthis one third of the commercial CEC and a ow channel which
alsoshould resemble the CEC with respect to dimensions and
electriceld strength (Fig. 5), but not to ow properties. Fossen et
al. [16]constructed a ow loop where water and oil are pumped
separatelyusing positive displacement pumps and then mixed in a tee
beforeowing through a choke valve where shear may be induced
inorder to form a dispersion which separates in the test
separator.The water and oil then are led (separately) back to the
feed separa-tor (Fig. 6). The droplet sizes were determined using
DVM on dilut-ed solutions. The droplet sizes determined were from 2
to 90 mapproximately which was conrmed by theoretical calculations
ofmaximum droplet sizes at the relevant pressure drops. The
meandiameter droplet size was between 4 and 12 m depending on
the
Fig. 1. Single HTVIEC module (left) and arrangement of 6 modules
in a large separator(right) [11].about the elasticity and viscosity
and for instance the presence of a
-
Fig. 2. VIEC and LOWACC installed in a separator [12].
108 I. Kralova et al. / Advances in Colloid and Interface
Science 169 (2011) 106127skin for the specic case of petroleum
crude oil at this interface. Inthis part we focused on our recent
major results about characteriza-tion of chemical composition and
physical properties of crude oil aswell as the interfacial rheology
properties at the water/oil interface.By their nature, emulsions
are physically unstable. The breakup ofemulsions consists of
several processes, which are mostly coupledand result in different
levels of instability, which could be dividedinto two parts:
creaming/sedimentation and occulation/coalescence.
Fig. 3. On the right the CEC unit with a capacity of 130000
Fig. 4. A water-in-oil (w/o) emulsion is produced by shear over
a choke valve and led to theelectric eld over the two spaced
plates. The oil and water exits through a opening approx3.1. Crude
oil components
Although the SARA fractionation method is a rough sorting of
thecrude oil constituents, it can provide an important classication
ofcrude oils. The SARA fractionation method has found great utility
incombination with high-performance liquid chromatography (HPLC).As
an example of the SARA fractionation method we will discuss
theprocedure from a study by Hannisdal and co-workers [17]. The
ability
barrels per day, on the left the CEC components [14].
inlet of the lab scale ESU. The ESU will work as small separator
even when there is noimately equal in area to the CEC [15].
-
of vibrational spectroscopy (IR and near-IR) to predict SARA
compo-nents in heavy and particle rich crude oil will be discussed
in the follow-ing section. In this study, 20 crude oil samples
(Table 1) at ambienttemperature and pressure were received from
exploration sites on theNorwegian Continental Shelf and sites
located in Brazil, France, theSouth China Sea, the Atlantic Ocean,
and the Gulf of Mexico. These oilswere quantitatively fractionated
into saturates, aromatics, resins, andasphaltenes (SARA) by
asphaltene precipitation in n-hexane and pre-parative HPLC. Here,
the focus will be on the HPLC procedure. Two col-umns were used:
one 21.2250 mm column packed with unbondedsilica 15 m and one
21.250 mm amino (10 m) column. Dichloro-methane (99.8%) and
n-hexane (95%) were used as mobile phases.
Twenty crude oils were analyzed especially for vibrational
featuresin the NIR and IR region. Parallel experiments showed good
reproduc-ibility for the spectroscopic measurements and the
fractionation meth-od. NIR spectra were used in combination with
partial least squares
Thirty crude oils were diluted in 30 vol.% toluene and
analyzedwith respect to the viscoelastic response to a sinusoidal
modulationin the frequency range from 0.01 to 1 Hz [20]. Diluted
and undilutedsystems were compared. Moreover, long-time dynamic
interfacialtension experiments of static drops of undiluted crude
oil in waterwere performed. As expected in the low frequency range
(0.011 Hz), molecular exchange from bulk strongly affected the
measureddilational parameters. For this reason the systems which
exhibitedparticularly low magnitude of the dilational modulus were
of theheaviest crude oils in the sample set, whereas the systems
with great-est dilational modulus were among the lightest crude
oils. The fre-quency dependence of the dilational modulus increased
with itsmagnitude as expected for diffusion-controlled relaxation
of solublelms. Overall, the undiluted crude oilwater interfaces had
similar re-laxation characteristics as the diluted samples except
for slightly re-duced magnitude of the dilational modulus.
The same samples of crude oils were analyzed with respect to
bulkand interfacial properties and the characteristics of their w/o
emul-sions to investigate the relative level of inuence that
individual pa-rameters have over the overall stability of w/o
emulsions. Thestability of emulsions was investigated by the
Ecritical cell [21]. Asexpected, a strong covariance between
several physicochemical prop-erties was identied. The comparison of
the experimental time for de-stabilization with the theoretical
time of droplet approach is showedin Fig. 10 (left) for a
water-in-heavy oil emulsion with droplet size of6 m. Given enough
time, the water-in-heavy oil emulsions could be
109I. Kralova et al. / Advances in Colloid and Interface Science
169 (2011) 106127(PLS) regression to predict SARA components from
the crude oil matrix.Regression models were built for each SARA
component from NIR datato predict the amount of SARA components.
NIR spectroscopy proved toperform well for the prediction of SARA
components with predictionvariances (RMSEP) of 2.82 wt.% (S), 1.47
wt.% (A), 1.46 wt.% (R), and0.44 wt.% (asphaltenes). As an example
the prediction vs. measuredamount of resins is presented
graphically in Fig. 7. The resin and satu-rate fractions were also
predicted in an excellent way from IR data.However adequate models
for the aromatic and asphaltene contentswere not obtained. It was
hypothesized that the similarity in the vibra-tional features from
aromatic carbon of asphaltenes and aromaticscaused this. These
models successfully tted the experimental datafrom NIR analyzes and
showed good predictive ability for the crudeoil composition.
Combination of spectroscopy and multivariable analysis could
alsopredict crude oil emulsion stability. Indeed Aske and
co-workers [18]studied correlation between emulsion stability
asmeasured by the crit-ical electric eld method (Ecritical) and
physicochemical properties asSARA composition determined by NIR
spectroscopy, density and TANvalue of test matrix consist of 18
crude oils. Analysis of the signicanceof the regression coefcients
of the model revealed that the NIR spec-tra, among some other
factors, were closely correlated to the measuredemulsion stability.
Based on this, an attempt was made to predictemulsion stability
exclusively from NIR spectra (Fig. 8). NIR spectracontain
information on both the aggregation state of asphaltenes andon
chemical composition. This may explain the good predictivepower for
the Ecritical values.
3.2. Crude oil emulsions
It was assumed that dilational relaxation properties
(surfacerheology) of surface active components can be probably of
greatimportance during droplet fragmentation and coalescence.
The
Fig. 5. The ow channel is formed by placing one coated steel
plate and one bare alu-
mina plate against each other spaced by a rubber gasket
[15].oscillating pendant drop method was used to study diluted
crudeoilwater interfaces and the effect of altering aromaticity of
the dil-uent and the concentration of crude oil [19]. The storage E
and lossE moduli determined by dilational rheology of one of the
crudeoil/water systems is presented in Fig. 9. At a perturbation
frequencyof 0.1 Hz, the equilibrium storage and loss moduli passed
throughdistinct maxima as a function of bulk concentration. The
apparentlylow viscoelasticity of the interfaces in systems with
high bulk con-centration was at least partly caused by high
diffusion ux of inter-facially active components from bulk. A
direct relation between themeasured interfacial relaxation
parameters and the overall emul-sion stability was not
identied.
Fig. 6. Flow scheme of the separator system with the main
process units. Arrows indi-cate the ow directions designed by
Fossen et al. to study oil/water separation [16].destabilized even
at very low electric eld magnitude (0.4 kV/cm).
-
Table 1Experimental compositions of crude oils separated into
SARA and water fractions [17].
Crude oil no. Origin Saturates (wt.%) Aromatics (wt.%) Resins
(wt.%) Asphaltenes (wt.%) Water (wt.%) Yield (wt.%)
Mean Sdev
Calibration set1 Brazil 33.5 46.5 18.8 1.2 0.1 0.8 100.82 Brazil
38.8 36.6 12.6 1.9 0.1 0.9 90.93 Brazil 30.5 43.2 20.9 5.0 0.3 0.2
99.94 China 36.0 30.8 22.1 3.1 0.1 9.6 101.55 Mexico 43.0 29.7 10.4
1.2 0.1 0.3 84.56 North Sea 26.5 38.8 20.5 2.8 0.1 12.6 101.37 West
Africa 46.9 37.9 14.0 1.7 0.0 0.1 100.68 Brazil 33.6 38.1 16.8 12.9
0.2 0.1 101.79 North Sea 45.0 32.9 9.8 1.2 0.1 0.2 89.210 North Sea
45.7 38.6 11.5 0.8 0.1 0.1 96.611 North Sea 43.6 32.9 7.2 0.6 0.0
0.4 84.812 North Sea 33.0 42.8 11.6 3.9 0.0 0.5 91.813 China 32.0
32.5 28.5 4.3 0.2 1.2 98.614 West Africa 53.1 30.9 8.2 1.0 0.0 0.2
93.315 Brazil 33.4 38.6 18.2 4.6 0.2 0.1 94.816 Brazil 37.3 42.6
14.1 3.8 0.4 0.3 98.117 North Sea 41.5 33.3 7.0 0.2 0.0 0.1 82.018
Brazil 26.0 41.1 21.9 10.2 0.3 0.1 99.419 North Sea 44.3 26.3 8.1
0.2 0.0 0.5 79.4
1.5
8.16.06.98.91.9
110 I. Kralova et al. / Advances in Colloid and Interface
Science 169 (2011) 106127The expression for the characteristic time
of droplet approach wasmodied for a constantly increasing eld
magnitude (Eq. (1)):
8 1=3 dE0 2=3 5=31
1=31
20 France 23.1 47.5 1
Replicates5 Mexico 43.3 29.58 Brazil 34.9 36.2 111 North Sea
44.4 30.215 Brazil 34.8 41.2 118 Brazil 26.1 40.7 2theo 5 dt 6W
where is viscosity and is permittivity of the continuous phase.w
is water cut (in this case 0.3) and dE0/dt is eld rate. The
primitivemodel in Eq. (1) describes to a certain extent, though a
simpliedforce balance, the dielectrophoretic effect and the
resulting dragforce on spherical water droplets. Even by assuming
monodispersityof the emulsion, perfect spherical and non-charged
droplets, no
Fig. 7. Correlation between modeled and measured amount of
resins from partial least squarand IR (right) region [17]. Stars
represent samples that were used to calibrate the model, andThe
solid line is a linear least-squares regression calculated from the
calibration set, and the1.46 wt.% with 6 PCs.contribution from
interfacial dynamics of surfactants, and insigni-cant contribution
of thermal or gravitational forces, Eq. 1 can onlymodel droplets
separated by a distance much greater than their radi-us. When
droplets approach each other this primitive model does notaccount
for the nal approach of droplets, that is, lm-thinning
5.4 0.1 4.2 91.7
1.2 0.1 0.3 82.412.3 0.2 0.1 99.50.6 0.0 0.4 82.54.6 0.2 0.1
99.6
10.6 0.3 0.1 99.3forces, neither concentrated and polydisperse
systems, captured thedifference between the 30 crude oils
reasonably well also at othereld rates (Fig. 10). It was proposed
that the destabilization of staticwater-in-heavy crude oil
emulsions in an electric eld was predomi-nantly retarded by the
viscosity of the oil phase [21]. When dropletsapproach each other
in an inhomogeneous electric eld the eldmagnitude increases
greatly. Strong dielectrophoretic forces disinte-grate the lms and
result in coalescence.
es calibration (stars) or prediction (circles) models of
spectral features in the NIR (left)open circles indicate unknown
samples that were predicted with full cross-validation.broken line
is the corresponding regression calculated from validation samples.
RMSEP,
-
It was investigated how different kinds of modications on
crudeoils, like deasphalting, dilution, and alkaline washing,
affect the emul-sion stability [22,23]. Generally, for the ve crude
oils, the emulsion sta-bility decreases as the viscosity is
decreased bydilution. However, as thecrude oils are diluted, the
concentration of surface active compoundslike resins and
asphaltenes is also decreasing, which will also affectthe emulsion
stability. Likewise, the interfacial tension and
interfacialelasticity will also change as a result of the dilution.
The most interest-ing information from Fig. 11 is what happens to
crude oils 3 and 4. Athigh viscosity their corresponding emulsions
show high stability, butas we increase the dilution (decreasing the
viscosity), the stabilitydrops, and levels off at a certain level,
independent of the viscosity. Inthis region, where the E-critical
value is fairly stable, the line up of thewater droplets cannot be
the limiting step for the break-up of theemulsion.
The E-critical also depends on water cut; increases at
lowerwater cut, due to the increased distances the droplets must
moveto form linear chains between the two electrodes [22]. The
samestudy showed in general, that the emulsion stability
decreaseswith increased dilution, because by the diluting the crude
oils, weare also decreasing the concentration of surface-active
compoundslike resins and asphaltenes, which will affect emulsion
stability.However, some systems show regions where the emulsion
stability
The type of emulsion obtained (w/o or o/w) depends on both
thecontact angle of particles and the water cut [28]. Emulsions
preparedwith Aerosil R7200 particles at different volume fraction
showed thatat 50% v/v or more of water cut, o/w emulsions were
formed. This isconsistent with the fact that Aerosil R7200 is
preferentially wettedby the aqueous phase rather than the oil
phase. The catastrophicphase inversion (inversion of the type of
emulsion by changing theoil water ratio) happens at water cut close
to 40% v/v. For lowerwater cuts, w/o emulsions are formed. There is
no catastrophicphase inversion for emulsions stabilized by Aerosil
R972 particles:w/o emulsions are formed for water cuts lower than
60% which isconsistent with the hydrophobic character of Aerosil
R972, whereasfor water cuts N60%, there is phase separation into a
water phaseand a w/o emulsion, reminiscent of a synaeresis
phenomenon.
Stabilizing properties of four different types of commercially
avail-able silica nanoparticles (Aerosil from Degussa, Table 2)
were stud-ied. Two of these products are hydrophobic; one is
extremelyhydrophilic, and one is expected to be wet to an
intermediate extentby both oil and water. These dry particles have
been modied withasphaltenes and resins and we have investigated the
performance ofthese solids as stabilizers in model oil/water
emulsion systems [29].Adsorption studies, using the QCM-D
technique, have shown theadsorbed amount from resin solution is
dramatically smaller thanwhen asphaltenes contribute to the
adsorption. The stabilization ef-ciency of the particles was
explained from a thorough characterization
111I. Kralova et al. / Advances in Colloid and Interface Science
169 (2011) 106127is independent of the dilution ratio or viscosity
of the crude oil. Theemulsion stability also shows temperature
dependence according toArrhenius law. Viscosity and emulsion
stabilities for water-in-oilemulsions were measured for the broad
crude oil matrix (27crude oils) and results were analyzed by
multivariate analysis. Ingeneral, there is an increase of the
emulsion stability as the viscos-ity increases. However, viscosity
also correlates with SARA data ofthe crude oils. This makes it
difcult to conclude whether the vis-cosity is the important
stabilization factor or if it is the heaviestcomponents in the
crude oil. For the high viscous crude oils, the an-swer is probably
both [22].
It was proved that deasphalting crude oils resulted in very
unsta-ble emulsions [23]. Emulsion stability is also strongly
affected by re-moving acidic compounds from the crude oils. The
polar resins playa very important role in stabilizing the
asphaltenes. By removingthe most polar resins we have disturbed the
interaction pattern be-tween the resins and the asphaltenes, and
the asphaltenes left inthe crude oil have a much higher propensity
to stabilize emulsions.This is in accordance with other study by
Spiecker [24].
Fig. 8. Predicted vs. measured plot from the PLS regression of
emulsion stability (Merit)based on the NIR spectra [18]. Stars
represent samples that were used to calibrate the
model. The solid line is a linear least-squares regression
calculated from this calibration set.3.2.1. Particle stabilized
emulsionsParticle-stabilized emulsions are an important topic in
petroleum
science. Indeed there are several types of particles present in
the pe-troleum uids such as reservoir particles (silica, clays),
mineral scales(CaCO3, BaCO3, SrSO4, CaSO4), corrosion products
(FeS, oxides) and soon. When exposed to crude oil, inorganic
particles may be modiedby the adsorption of heavy crude oil
components like resins andasphaltenes [25,26]. Consequently, you
can have formation of petro-leum emulsions stabilized by particles.
It has even been shown thatemulsions containing inorganic particles
can be more stable thanthose stabilized by asphaltenes only [27].
To mimic particle-stabilizedemulsions encountered in petroleum
production, we have studiedstabilizing properties of four different
types of commercially availablesilica nanoparticles (aerosil from
Degussa, Table 2).
Fig. 9.Near-equilibrium(2.5 h after preparation) of the storage
E and loss Emoduli deter-mined by dilational rheology at a diluted
crude oil/brine (3.5% NaCl solution) interface as afunction of the
bulk crude oil concentration and the aromaticity of the solvent.
The crudeoil was diluted to different extents in
heptanetoluenemixtures as indicated in the legend(vol.%H).
Themeasurementswere carried outwith a constant applied frequency
(0.1 Hz).The characteristics of the crude oil are reported in
Hannisdal et al. [19].of their surface properties, including
spectroscopy, contact angle
-
measurements () and zeta potential measurements. The
stabilizationefciency was greatly enhanced by adsorption of crude
oil componentsonto very hydrophilic or very hydrophobic silica.
Unmodied silica par-ticles Aerosil 200 (w/s/air=14) did not act as
good stabilizers due totheir very hydrophilic character whereas
silica particles coated withresins (w/s/air=73, Fig. 12, top) or
asphaltenes (w/s/air=84, lower)gave emulsions of large droplets
which were very stable to coalescence(induced by a high centrifugal
eld). However, when the particles weresuspended in the water phase
prior to emulsication, emulsion stabil-ity was signicantly reduced.
The stability of the emulsions decreasedprogressively with
increasing volume fraction of the disperse phase,in line with
increased drop size. Droplet size distributions of stableemulsions
revealed that the total interfacial area of a system was di-rectly
determined by the amount of particles present. Thus, the total
in-terface area remained constant when changing the volume fraction
ofthe disperse phase. Such observation is consistent with a
mixingmech-anism characterized by both droplet fragmentation and
coalescenceprocesses.
that the yield stress decreases when the mass fraction of
particle in-creases (for o/w emulsions, see Fig. 14). G and G
display the sametrends (they decreases with the adding of
hydrophobic particles)-
Fig. 11. Emulsion stability, measured by means of critical
electric eld technique, as a
n an electric d.c. eld. Destabilization of a water-in-heavy
crude oil emulsion under the in-left). Destabilization of 30
water-in-crude oil emulsions at a eld rate of 0.004 kV/cm s1
o is theoretical value of droplet approach. The experimental
time is given by the measured
112 I. Kralova et al. / Advances in Colloid and Interface
Science 169 (2011) 1061273.2.2. Rheology of particle-stabilized
emulsionsTwo different particles were used: Aerosil 7200
(hydrophilic) and
Aerosil 972 (hydrophobic). There were dispersed in solvents:
Aero-sil 7200 in a NaCl 3.5% solution and Aerosil 972 in decane.
The par-ticle concentration was kept constant at 25 g/L. The
rheologicalproperties of the emulsionswere determined by steady
shear and oscil-latory shear measurements using parallel plate
geometry (PhysicaMCR301 Rheometer) [30]. First, it was investigated
the rheologicalproperties of emulsions stabilized by a single type
of particles (eitherhydrophilic or hydrophobic) then by mixtures of
these particles. Dueto experimental requirements (emulsions must
not sediment/creamin the rheometer cell), only o/w emulsions
stabilized by Aerosil 7200particles and w/o stabilized by Aerosil
200 were characterized. Fig. 13presents a typical ow and
oscillation curve. This emulsion exhibits ayield stress, a
shear-thinning behavior and thixotropy. This system pre-sents the
following properties: G is higher than G (about ten times)and G and
G are constant over the whole investigated frequencyrange. All
these rheological features are specic for gels, according tothe
phenomenological denition proposed by Almdal et al. [31]. Thisgel
character is consistentwith the yield stress previously found and
in-dicates that there is formation of a network of connected
droplets. Theobserved shear thinning behavior could be due to
breaking of interac-tions between droplets and thixotropy to time
needed to reform inter-actions. We have also investigated the
properties of w/o emulsionsstabilized by R972 particles. They have
similar features as the o/w emul-sions stabilized by Aerosil R7200
particles that is, presence of a yieldstress and time-dependant
viscosity (either thixotropy or rheopexy).
Fig. 10. Time for destabilization compared to the theoretical
time for droplet approach iuence of a constant background electric
eld with magnitude from 0.4 to 4.0 kV/cm(where exp is
characteristic time of destabilization of emulsions in the d.c eld
and the
CEF value and dE0/dt whereas the theoretical time is predicted
according to Eq. 1 (right). [2Their viscosity also increases with
the dispersed phase volume fraction.To conclude, rheological
properties of w/o emulsions stabilized by R972particles are similar
to those of o/w emulsions stabilized by R7200 par-ticles. It also
seems that, in the investigated systems, rheological proper-ties of
particle-stabilized emulsions are related to the
rheologicalproperties of the dispersions of their particles.
So far we have only considered rheology of emulsions
stabilizedby one type of particles. In this part we have
investigated the prop-erties of emulsions stabilized by mixtures of
hydrophilic and hydro-phobic particles. Emulsions were prepared by
keeping the totalconcentration of particles constant but varying
the mass ratio be-tween hydrophobic particle (Aerosil 972) and
hydrophilic parti-cles (Aerosil 7200). There are formation of o/w
emulsions formass fractions of hydrophobic particles lower than 0.9
whereas w/o emulsions are formed at higher mass fractions (Particle
concen-tration=25 g.L1 and water cut=50% v/v). This transition is
calledtransitional phase inversion [32]. Emulsions prepared with
mixturesof particles present the same features as emulsions
stabilized by asingle type of particles. In particular their ow
curves exhibit yieldstress, shear-thinning behavior and thixotropy.
It was observed
function of the viscosity of the crude oil, for ve different
oils. The viscosity was changedby dilution of the different crude
oils by a heptanetoluene mixture (70:30 vol.%) [22].1].
-
results not shown. Finally, the stability of o/w emulsions
against co-alescence decreases when the mass fraction of
hydrophobic particleincreases. Fig. 15 presents a sketch of our
view of structures of o/w emulsions stabilized by only hydrophilic
particles and mixturesof hydrophilic and hydrophobic particles.
With only hydrophilicparticles (left), aggregated silica particles
are adsorbed at the liq-uid/liquid interface. There exists a
network of particles connectingthe droplets to each other. This
network explains the rheologicalproperties of these emulsions. When
hydrophobic particles areadded (right), as the concentration of
hydrophilic particles, the con-nections between droplets are looser
since the hydrophobic parti-cles are preferentially dispersed in
the oil phase. That explainswhy the rheological properties of these
emulsions are weaker thanwith only hydrophilic particles. However
we do not know if hydro-phobic particles displace the hydrophilic
particles from the interface(right bottom of Fig. 15) or not (right
top of Fig. 15).
3.3. Asphalthene stabilized emulsions
Asphalthenes are typically dened as the fraction of petroleum
in-soluble in n-alkanes (typically heptane, but also hexane or
pentane),but soluble in toluene, i.e. it is a solubility class. The
molecules arecomposed of small polyaromatic parts linked by
aliphaltic or naphte-nic moieties. They contain the major part of
the heteroatoms (Nitro-gen, Oxygen and Sulfur) and metal atoms
(Nickel, Vanadium)present in a crude oil. Moreover these molecules
can associate in so-lution (in crude oils or in model solvents) to
form aggregates with a(weight) average molar mass which can vary
between 10,000 and1,000,000 g.mol1 in model solvents (such as
toluene), dependingon thermodynamic conditions such as solvent
nature, temperatureor pressure. In the crude oil, presence of
asphaltenes induces the for-mation of very peculiar colloidal
suspensions. It is commonly knownthat the asphaltenes precipitates
when the crude oil is treated witha light aliphatic
hydrocarbon.
NIR and MIR spectra were correlated to the Hildebrand and
Hansensolubility parameters, using multivariate data analysis
[33,34]. Modelswere built from NIR and MIR spectra of different
solvents and solventmixtures. Table 3 shows the results from the
correlation of solubility pa-rameters to NIR spectra. Furthermore,
the solubility parameters of SARAfractions and crude oils were
predicted using the models developed
Table 2Chemistry and specication of silica particles [52].
Product id. After treated with Specic surface area (BET) (m2/g)
Tapped density (g/L) Primary particle size (nm)
Aerosil 200 Hydrophilic 20025 50 12Aerosil 7200 Hydrophilic
3-Methacryl-oxypropyl-trimethoxysilane 15025 230Aerosil 202
Hydrophobic Polydimethyl-siloxane 10025 60 14Aerosil 972
Hydrophobic Dimethyl-dichlorosilane 11025 50 16
113I. Kralova et al. / Advances in Colloid and Interface Science
169 (2011) 106127Fig. 12. Particle-stabilized emulsions of silica
(Aerosil 200) coated with resins (top)and asphaltenes (lower). The
ordinate axes show the amount of disperse phase re-solved after 10
min centrifugation at 1580g. The used solvents are water: NaCl
3.5%solution and oil: heptane/toluene 70/30 v/v [29].1 10 100
1000
sh
ear s
tress
/ Pa
1
10
100
Shear rate / s-11 10 100 1000
vis
cosi
ty /
Pa.s
0,010,1
110
100
.
shear rate / s-1
.
, G'' (
Pa)
1000
G'G''Frequency (Hertz)1010,1
G'
100
Fig. 13. Flow and oscillation curve of o/w emulsions stabilised
by Aerosil 7200 particles,water cut=50% v/v. Particle
concentration=25 g.L1.
-
depressurized in steps, and the resulting NIR spectra were
recordedat each pressure level and analyzed with multivariate
analysis. Theasphaltene precipitation onset pressure was identied
from increasedoptical density due to light scattering. The
reversibility of the asphal-tene aggregation was studied for the
crude oil and a model system byrepressurizing the systems stepwise
to the original pressure. The sys-tems were then left to
equilibrate for several hours. In this model sys-
1000100101
Sh
ear s
tress
/ Pa
1
10
100
m R972=0 o/w
m R972=0.2 o/w
m R972=0.4 o/w
m R972=0.5 o/w
m R972=1 w/o
.
Shear rate / s-1
Fig. 14. Flow curve (lled symbols: up curve, empty symbols: down
curve) of emul-
Table 3Results from PLS modeling and predictions based on the
correlation of solubility pa-rameters to unmodied NIR spectra.
Qualityparameters
Hildebrand(total)
Dispersiveparameter
Hydrogenbonding
Polarcontribution
Principalcomponents
4 4 2 4
Correlationvalidation
0.92 0.88 0.95 0.95
Correlationprediction
0.92 0.53 0.91 0.78
RMSEV (MPa1/2) 1.8 0.6 1.7 1.1RMSEP (MPa1/2) 2.6 1.1 3.5
3.8ExplainedY-variance (%)
87 81 90 89
ExplainedX-variance (%)
80 84 67 82
114 I. Kralova et al. / Advances in Colloid and Interface
Science 169 (2011) 106127(Table 4). This study proved that IR and
NIR spectra in general can becorrelated to Hansen solubility
parameters. Furthermore, IR and NIRspectra can be used to
distinguish between crude oils and crude oilcomponents.
Aliphatic solvent conditions and pressure reductions will
increaseasphaltene aggregation size. Under increasingly unfavorable
solventcondition, asphaltenes aggregate and eventually precipitate
as largeasphaltene occulates. The asphaltene precipitation onset
gives im-
sions stabilized by mixtures of Aerosil R7200 and R972 particles
(v/v water=50% v/v, total particle concentration=25 g.L1) tted by
the HerschelBulkley equation(solid line: up curve, dash line: down
curve).portant information about the solubility of the asphaltenes
in agiven hydrocarbon system. Aske and co-workers [35] have
investigat-ed the asphaltene aggregation behavior from crude oils
and modelsystems under high pressure (300 bar). The systems were
then
Hydrophilic particles Hydrophobic particles
water oil
Only hydrophilic particles
Fig. 15. Sketch of structure of o/w emulsions stabilized by only
hydrophilic partem the asphaltene aggregation was only partially
reversible.However, the possibility that the aggregation of the
model systemasphaltene is reversible if even more time is allowed
for equilibrationcannot be excluded.
In order to hinder asphaltene deposition, the petroleum
industryinjects large volumes of chemicals into reservoirs and
pipelines.These chemicals are supposed to imitate the indigenous
resin frac-tion, by dispersing the asphaltenes in the hydrocarbon
mixture as dis-cussed earlier in this chapter. In addition to the
direct problemsconcerning asphaltene deposition in process
equipment, the stabilityof water-in-oil emulsions will be strongly
dependent on the asphal-tene aggregation size and solubility state
[36,37]. The results showedthat additives, which are efcient in
replacing hydrogen bonds inasphaltene aggregates, possess
dispersive power and can serve as in-hibitors. Recently, in a study
concerning the potential for hydrateplugging, crude oils washed
with a strong alkaline solution (pH 14)showed much higher
water-in-oil emulsion stability than the original
water
Mixture hydrophobic/hydrophilic oil
or
ticles (left) and mixtures of hydrophilic and hydrophobic
particles (right).
-
crude oils [38]. The pH 14 wash extracts the most polar resins,
typi-cally napthenic acids and phenols. Removing them should cause
theasphaltenes to precipitate earlier when titrating the crude oil
withan n-alkane.
3.3.1. Asphaltenes precipitated by multi-step precipitation
procedureAsphaltenes were precipitated into two fractions using a
two-step
precipitation [39] procedure and later extended to separate
asphal-tenes in 4 fractions [40]. In the two/step precipitation
procedure,the rst fraction was obtained by mixing 3:1 volumes of
n-pentane/crude oil followed by ltration. In the following step the
second frac-
2, the 10:1 middle fraction exhibited the highest interfacial
activity,measured after 12 h. The measured interfacial activity at
the end ofthe experiments (equilibrium value) was plotted as a
function ofthe n-pentane-to-crude oil ratio (Fig. 17). It was
neither the leastnor the most soluble fractions, within the
fractions studied, whichwere the most interfacially active
compounds. This is in fact verywell in accordance with general
surfactant chemistry. For a com-pound to be interfacially active,
it has to have both water solubleand oil soluble parts. This
principle might be transferred to these re-sults by suggesting that
it is neither the least nor the most solublefraction that is the
more interfacial active, but a middle fraction. Thecommon 40:1
fraction may mask important features of asphaltenesresulting in
erroneous assumptions based on properties studies andthe structure
elucidation of the asphaltenes. This has large technical
115I. Kralova et al. / Advances in Colloid and Interface Science
169 (2011) 106127tion was precipitated out from the ltrate using
18:1 volumes of n-pentane/crude oil. Whole asphaltenes were also
precipitated using a40:1 n-pentane-to-crude oil ratio. The amount
of each fraction wasdetermined for comparison with the whole
precipitated asphaltenes.Three crude oils (named WA, NS-A and NS-B)
were used and theasphaltene fractions obtained were characterized
with regard toonset of precipitation, interfacial tension (o/w) and
radius of gyration(RG) of the aggregates. It was proved that
asphaltenes precipitated atthe 3:1 dilution ratio (rst fractions)
had a faster initial reduction ofthe o/w, while the second
fractions (18:1 dilution ratio) led to alower o/w over time. Whole
asphaltenes had a fast initial reductionlike the rst fractions and
reduced the value of the interfacial tensionmore, indicating that
there were also compounds which will not in-uence the value of o/w
after a longer period of time. The SANS mea-surements showed that
the aggregates of the rst fractions werelarger than the aggregates
of the second fractions (Table 5). Precipita-tion procedure used
was suitable for fractionating asphaltenes by adirect stepwise
precipitation from the crude oils without rst precip-itating the
whole fraction. It was also shown that the solvent proper-ties of
the two solubility fractions were quite different [39]. The
rstfraction that precipitated upon addition of a small amount of
n-pentane was less soluble in the precipitation onset
experiments,formed larger aggregates and had a different (lower)
interfacial activ-ity as compared to the second, more soluble
fraction. The assump-tions made, based on the results, were that
the rst fraction wouldcontain molecules with a higher molecular
weight, and be morepolar and more aromatic. In next study, Fossen
et al. [41] character-ized and quantied the functional groups,
aromaticity, polarity andsize of the asphaltene solubility
fractions precipitated before. Whatis important to keep in mind is
that asphaltenes are mixtures of thou-sands of different, yet
relatively similar compounds. It means thatvalues of these
measurements will result in average values of all thecompounds in
the asphaltene sample. It was shown that the relativeamounts of the
polar heteroatoms S, N and O were slightly higher inthe rst (least
soluble) fractions than in the second fractions andwhole
asphaltenes. LDI-MS showed that the average molecularweight is
higher for the rst fractions compared to the second frac-tions the
whole asphaltenes. There were no indications from the re-sults that
the alkyl chains substituted on the aromatic were longerfor the
second fraction.
The second fractions were also more substituted. All the
parame-ters obtained are an average of the molecular mixture in the
samples,thus there are no guarantee that the relative differences
commented
Table 4Results from prediction of the solubility parameters on
the SARA fractions presented asthe lowest and highest predicted
value.
Fraction Hildebrand Dispersive Hydrogen Polar
1/2 1/2 1/2 1/2
[MPa] [MPa] [MPa] [MPa]
Saturates 16.416.7 16.216.4 4.75.6 0.00.7Aromatics 16.716.9
16.416.6 4.95.4 0.71.0Resins 17.318.6 15.616.4 5.98.1
1.22.3Asphaltenes 18.019.1 15.015.8 7.610.3 1.23.2upon in this work
are the ones responsible for the properties deter-mined in the
previous work. For example comparing the aromaticityof the rst and
second fractions one nds, according to the assump-tions, that for
the WA and NS-B the less soluble fraction has a higheraromaticity,
while for the NS-A the aromaticities are equal for the
twosub-fractions [41]. The trends found that support the
assumptions arethat the less soluble fractions which formed larger
aggregates (SANS)were less interfacial active (pendant drop) were
more aromatic, morepolar (in the aromatic core) and had a higher
average molecularweight. The second fractions had alkyl groups that
were probablymore branched and contained a somewhat larger portion
of naphte-nic rings and had more of the hydroxyl and carboxylic
groups onthe aliphatic parts which could explain the higher
interfacial activity.
There is probably no single reason for the precipitation of
asphal-tene that can be explained by the molecular structure or
weight. This,and the previous results, indicates that it is of
great interest and im-portance to fractionate asphaltenes into less
and more soluble asphal-tenes. Furthermore, it is not yet
determined what the ultimatecharacter of the least soluble
asphaltene fraction is, and if this fractionis the most harmful
with regard to adsorbtion to surfaces and emul-sion stability.
Fossen et al. [40] extended the two-step precipitationto a
four-step precipitation procedure where it was shown that
theinterfacial activity was not linearly dependent on the order of
the sol-ubility fraction. The intention here was to further
investigate whichpart of the second fraction was more interfacially
active. The precipi-tation procedure was principally similar to
procedure before, onlyhere with four dilution and ltration steps.
Precipitation followedby inter-step ltration off of the
precipitated material was performedstep-wise after an addition of
3:1, 10:1, 15:1 and 20:1 n-pentane tocrude oil, named fraction 1,
2, 3 and 4 respectively. These fractionswere analyzed with regard
to precipitation onset in toluene/heptanemixtures and interfacial
tension (Fig. 16). Two oils were tested inthis study, the NS-A and
the NS-B crude oils. The results from the pre-cipitation onset
experiments verify that the solubility of the fractionsis in the
order of which they were precipitated.
Fig. 16 shows a great difference in the interfacial tensions for
thedifferent fractions from both crude oils. It is remarkable that
fraction
Table 5Calculated RG from the SANSmeasurements using the Guinier
approximation, whereWA,NS-A and NS-B are names of crude oils. The
uncertainty of the instrument is1 . N.D.(not determined) indicating
the values were outside the detection limit for the instru-ment.
This meant that these samples contained aggregates which were much
larger(N700 ) than the other samples.
Asphaltene fraction RG()
WA rst 30WA second 25NS-A rst N.D.NS-A second N.D.NS-B rst
26NS0-B second 21consequences within crude oil production and
processing.
-
3.4. Polyaromatic surfactants as model compounds for
asphaltenes
Asphaltenes as a solubility class comprise a very broad
distributionof chemical structures. Several studies have
investigated the asphalteneaverage structure and physico-chemical
properties to understand theirbehavior as a function of chemical
structures [39,40]. Oneway to do thisis to divide the asphaltenes
into several subfractions, but each subfrac-tion contains thousands
of different compounds, and a never-endingapproach is to keep on
dividing into numerous fractions to get morenarrow structure
distributions. Nordgrd et al. [42] used a different ap-proach. They
rst synthesized molecules with known structures andsmall structural
variations, and correlate the physico-chemical proper-ties obtained
directly to introduction of functional groups. Fig. 18shows these
synthesized compounds as polyaromatic surfactants,with size
andmolecular weights in the same region as the currently ac-cepted
average molecular weight as asphaltenes. The molecular designis one
xed part of themolecule and one part with varying hydrophilic-ity.
Also, three of themolecules incorporated an acidic
groupwhichwasexpected to increase the interfacial activity of such
compounds. Table 6presents interfacial tensions measured with the
pendant drop tech-nique ans shows that the acidic molecules were
highly interfacially ac-tive at low concentrations in toluene
towards a pH 9 aqueous solution.The absence of the acidic group
however resulted in a total absence ofinterfacial activity at
corresponding concentrations. Pressure-area iso-therms using the
Langmuir technique (Fig. 19) showed that the acidiccompounds formed
stable monolayer lms with high collapse pres-sures, and
themoleculeswere arrangedwith the aromatic cores normalto the
aqueous surface, yielding a sheet-like arrangement in the
lmresulting in highly favorable aromatic interactions. The
non-acidic com-pound did not show this arrangement and did not form
a stable mono-layer lm. The studies showed that the presence of an
acidic group insuch compounds were essential for their interfacial
and lm properties,
micrometer-size structure, and the nal large network gel.
Crystalli-zation of parafns also depends on physical factors
(cooling rate,shear force and so on). The length of the crystals is
dependent onthe temperature and cooling rate. The crystal
aggregation is alsovery sensitive to the shear rate. At static
condition, individual disksform a colloidal network. At high shear
rates, the aggregates becomemore spherical in shape and less
polydisperse.
Wax precipitation and deposition is a recurring challenge in
trans-portation of crude oil, and increased knowledge about the
behavior ofsuch systems is necessary. Microscopy, rheometry and
DifferentialScanning Calorimetry (DSC) were used to follow the
crystallizationof wax for two model systems. Chen et al. [44]
investigated the owand viscoelastic behavior around the wax
precipitation temperature,and the yield stress was determined both
after dynamic and staticcooling. They proved that long crystals
were formed during low cool-ing rate, resulting in the strongest
gel structures. The gels wereformed at very low amounts of solid
wax crystals (0.30.4%), andwax precipitation was promoted by
increased wax content dissolved
116 I. Kralova et al. / Advances in Colloid and Interface
Science 169 (2011) 106127Fig. 16. Interfacial tensions of 100 mg/L
solutions of asphaltene fractions from NS-Aand NS-B crude oils in
toluene. Water phase is pH 7 buffered water containing
3.5 wt.% NaCl [39].and the unusual interfacial activitywere
attributed to the formation of astable lmwhich were stabilized by
aromatic interactions due to an ar-rangement with the aromatic
cores normal to the surface plane.
3.5. Wax in crude oils
The presence of long-chain saturated alkanes in crude oil can
leadto severe problems associated with wax precipitation and
depositionin petroleum transport pipelines and processing
equipment. Indeed,the parafn deposition on the cold walls of
pipelines will restrictthe ow, and in the worst case entirely plug
the pipelines. Secondly,presence of solid waxes in continuous oil
systems may give rise tothe difculty of prediction and evaluation
of the ow propertieswhich is largely dependent on the waxy
constituents of the oils.Waxes are typically classied molecular
weight isoparafns and cy-cloparafns. Crude oils generally contain
not only n-parafns butalso considerable amounts of isoparafns and
cyclic compounds,which in fact constitute the largest fraction.
Visintin et al. [43] foundthat n-parafn dissolved in organic
solvents display a sharp transitionin gel strength at the pour
point, whereas by addition of isoparafns,the buildup in gel
strength as a function of temperature is much moregradual, because
increasing isoparafn fraction facilitates the forma-tion of
amorphous wax solids. At high temperatures, waxes are inthe molten
state, and crude oils normally behave like Newtonian liq-uids. When
the temperature drops below Wax Precipitation Temper-ature (WPT),
solid wax crystals precipitate out of oils and form a gel.In the
denition of wax crystallization at a more microscopic level,the
gelation mechanisms involves three different processes: forma-tion
of lamellar with thicknesses of ca. 1.53 nm, sheet-like
crystals,
Fig. 17. The plot shows the interfacial tension of each fraction
at the end of the exper-iment time. The second fraction (second
points on curves), that is the fraction precip-itated after the 3:1
fraction was taken out of the solution and when a 10:1 ratio
ofpentane-to-crude oil was added, was the most interfacially active
compound in theseexperiments [39].in the samples. Dynamic cooling
conditions decreased the gel
-
halt
117I. Kralova et al. / Advances in Colloid and Interface Science
169 (2011) 106127strengths considerably as the imposed shear forces
affect both crystalmorphology and crystalcrystal interactions. Fig.
20 shows the differ-ence between wax A and B is further underlined
by looking at the vis-cosity at different amounts of precipitated
wax. Wax A has steepincrease in viscosity from 0.25 to 0.35 vol.%,
as wax precipitates andform a gel. Around 0.2 vol.%, only a small
inection point is seen forwax B and the increase in viscosity is
very small. Clearly, there are
Fig. 18. Structures and molecular weights of four synthesized
model compounds for aspbeen varied inserting different amines
[42].no strong interactions between the wax crystals in this case,
andthe network formation is not signicant when the wax amount
isless than 1%. In this study it was also investigated the effects
ofwater cut, asphaltene amount and wax amount, and the results
inter-preted in terms of microstructure and aggregate state of wax
crystalsand water droplets.
Fig. 21 proved that the yield stress increases signicantly
withmore wax in the oil phase and also as the water cut increased.
Thevolume fraction of the dispersed phase is known to be a very
impor-tant variable in emulsion rheology, and the relative
viscosity of thesuspension with respect to the pure uid is often
well described bythe Krieger and Dougherty equation. Wax content
and water cut hasthe most pronounced effect upon the viscosity and
yield stress ofthe systems. The asphaltene content also result in
increased viscosityand yield stress values, but to a lesser degree
than the previousvariables.
A new analyticalmethodwas developed for investigatingwax
depo-sition in petroleum systems in our group. The technique is
based on aquartz crystal microbalance which has been modied to
accommodate
Table 6Interfacial tension of asphaltene model compounds at 50 M
after 500 s between tolu-ene and pH 9.
Compound (IFT)
[mN/m]C5 Pe 1PAP 4TP 4BisA 36a temperature gradient in the
direction normal to themeasuring surfaceaswell as a continuous ow
of uid above the crystal. The instrument iscalled a Thermal
Gradient Quartz Crystal Microbalance (TG-QCM), andprovides a very
sensitive probe of the mass and physical properties ofmaterial
which adheres to a deposition surface by recording changesin the
resonance frequency and energy dissipation of a quartz
crystalresonator. The TG-QCM method is demonstrated using the
model
enes. The right side of the aromatic core has been kept
constant, while the left side haswaxy oil which consists of a rened
parafnwax dissolved in dodecane.Wax deposition is performed on
quartz crystals coated with gold, stain-less steel, and
polyethylene. It is shown that rigid deposit formation andgel
deposition are separate and distinct phenomena, both of which
cancontribute to the formation of incipient wax deposits. Rigid
depositlayers grow slowly, and their formation depends on the
wettability ofthe solid substrate material. Gel deposits, on the
other hand, form rap-idly on all surfaces at high wax
concentrations, and contain a largeamount of occluded oil.
A special uid chamber was designed and built with the
capabilityto impose a temperature gradient in the uid regions above
the
Fig. 19. Pressure-area Langmuir isotherms of the model compounds
at room tempera-ture. The probe was a Wilhelmy plate and the
barrier speed was 5 mm/min. The sub-phase consisted of pure Milli-Q
water at pH 5.8 [42].
-
14 mm piezoelectric quartz crystal [45]. Fig. 22 shows a
conceptualcartoon of the thermal gradient QCM chamber. The uid
chamber ispositioned between a lower thermal block and an upper
thermal
block, such that an applied temperature difference between
theupper and lower blocks results in a temperature gradient in the
ver-tical direction. The uid chamber is constructed from Teon and
hasan upright cylindrical geometry, with a height of 10 mm and a
diam-eter of 20 mm. The measuring quartz crystal is positioned on
thelower block, such that the coated side of the quartz crystal is
in con-
Fig. 20. The viscosity of wax A and B (20% in decane) at low
fractions of precipitatedcrystals [44].
Fluid Chamber
Lower
Quartz Crystal
Upper
Fluid Outlet
Fluid Inlet
Fig. 22. Conceptual cartoon of the TG-QCM chamber [45].
118 I. Kralova et al. / Advances in Colloid and Interface
Science 169 (2011) 106127Fig. 21. a) Yield stress at 20 C and 30 C
for emulsions with different wax content. Thedotted lines represent
the t to the power law equation. b) Yield stress at 20 C
foremulsions with different water cuts [44].tact with the uid
chamber. The lower block is in thermal communi-cation with the
piezoelectric quartz crystal through a thermallyconducting crystal
support body. The temperature of the lowerblock is regulated via
software using a Peltier element which transfersthermal energy to a
radiator heat sink cooled by a fan. The upperblock contains a
heating element which is also in thermal communi-cation with the
uid chamber, and is controlled manually from an ex-ternal
temperature control unit. Fluid ow into the measurementchamber is
driven by an external peristaltic pump.
Before each experiment, the chamber and tubing are rinsed
withlarge amounts of ethanol and/or toluene, followed by drying.
Subse-quently, the measurement chamber is lled with dodecane
ormodel uid, and specied temperature values are set in the upperand
lower blocks. In each experiment, resonance frequencies and
dis-sipation factors are monitored for the 1st (fundamental), 3rd,
5th,7th, 9th, and 11th harmonic overtones (Figs. 23 and 24). After
stablebaselines are obtained, wax deposition is initiated on the
quartz crys-tal surface by pumping the waxy model oil through the
measurementchamber at a constant volumetric ow rate.
Rigid incipient wax layers are sufciently thin to be probed by
theacoustic shear wave propogating across the quartz crystal during
os-cillation [46], and are evidenced by scalable reductions in
resonancefrequencies measured at various harmonic overtones.
Surface wetta-bility is shown to have a large effect on the
formation of rigid waxlayers. Adsorbed water lms on stainless steel
completely preventthe formation of wax deposits at low
super-saturation conditions,due to interfacial hydrophobic forces.
Gel deposits, on the other
200
ft (10
-6 )
3rd Overtone5th Overtone7th Overtone9th Overtone0
100
0 50 100 150 200Time (min)
Dis
sipa
tion
Shi 11th Overtone
Fig. 23. Dissipation factor shifts measured during deposition of
the 10 wt.% wax modeluid unto a stainless steel surface. The lower
block temperature is 20 C, and the uidreservoir temperature is 40 C
[45].
-
hand, form on all surfaces at high wax contents and contain a
largeamount of occluded oil. The observed role of hydrophobic
forces in re-pelling the formation of rigid wax layers has
signicant implicationsfor the mechanism of incipient wax deposition
in standard stainlesssteel pipelines. Therefore, incipient wax
deposits must form by gela-tion instead of surface crystallization.
Additionally, depending onthe shear conditions present, wax
deposits may not always form onstainless steel surfaces at low
under-cooling conditions, affordingpipeline operators greater
latitude in avoiding wax deposit formation.
The wax deposits can be monitored and visualized by AFM.
Vari-ous modulation techniques can be used when AFM experiments
arecarried out and measure the attracting/repulsive forces
arisingwhen two bodies are close to each other. The force is
measured byusing a cantilever with a certain spring constant (0.01
to100 Nm1) that is pulled towards the investigated surface in
caseof attraction, or pushed up in case of repulsion from the
investigatedsurface. Force spectroscopy, that is moving the
cantilever in the z di-rection with no movement in the xy plane, is
used for measuring dif-ferent surface mechanical properties. There
are three main types of
randomly clustered structure which closely resembles the
substratepolyethylene material, which has possibly been modied by a
thinlayer of deposited cyclic or branched parafns. The
AFM-measuredwax layer thickness of 270 nm is in reasonable
agreement with the225 nm average layer thickness derived from
impedance analysismodeling of the TGQCM measurements performed
during thedeposition.
3.6. Naphthenic acids and phase diagrams
It is well known that oil-continuous emulsions can be stabilized
bymultiple layers of surfactant instead of only a monolayer [49].
In anequilibrium situation this corresponds to a sample location in
athree-phase area where two solution phases (L1 and L2) are in
equi-librium with a lamellar liquid crystalline phase (so-called D
phase).This situation is of relevance in crude oil systems with
high levels ofnaphthenic acids. In order to simulate the situation
in high asphalte-nic crude we also combined the D-phase
stabilization with asphaltene
Fig. 25. Shaded topography image obtained using TappingModeTM
for the dried waxlayer deposited from the 10% Sasolwax 5405
solution onto QCM crystal: a) gold-coated,b) polyethylene-coated.
The depth of the observed indentation toward the bottom ofthe image
is approximately 0.4 m.
119I. Kralova et al. / Advances in Colloid and Interface Science
169 (2011) 106127tip mode like contact, non-contact and tapping
mode [47].The technique has also been used for investigating the
effect of
resins and asphaltene inhibitors on the precipitation of
colloidalasphaltene [48]. In this study monolayers of asphaltenes
and resinswere transferred onto mica substrates using the L-B
technique, andthe topography analyzed by means of AFM. This work
shows thestructural change in the monolayer when the composition of
thelm was gradually changed from pure asphaltenes to pure resins.
Pic-tures of pure asphaltene show a closed-packed layer of round
disks orrod formed units. Addition of resins will change this rigid
structuretowards a more open network with regions completely
uncoveredby lm material. Pure resins build up a layer with an open
structure,i.e., more like a fractal pattern. Fig. 25 illustrates
the morphologicalimages of wax deposits and substrate materials
which were obtainin TappingModeTM probes. Lamellar contours are
clearly visible inthe image, and correspond to the growth steps of
the wax crystalsformed during the deposition process. However,
because the gellayer has likely collapsed during the drying period
in which the dode-cane solvent evaporated, the imaged deposit
morphology is unrepre-sentative of the initial gel structure.
Similar images were obtained atvarious positions on the crystal,
indicating homogenous surface cov-erage of the initial gel layer.
In case of wax deposition on the polyeth-ylene-coated crystal it
was observed two distinct recurring domains.The rst recurring
domain is a clearly observed rigid wax layer,appearing as
interlocking corrugated layers of solid wax, as depictedon the
right hand side of gure. The second recurring domain, asdepicted on
the left hand side of gure, is a less well dened
-200
-100
0
100
0 50 100 150 200Time (min)
delta
F (H
z)
3rd Overtone5th Overtone7th Overtone9th Overtone11th
Overtone
Fig. 24. Normalized resonance frequency shifts measured during
deposition of the 10wt.% wax model uid unto a stainless steel
surface. The lower block temperature is
20 C, and the uid reservoir temperature is 40 C [45]. particles.
Lowmolecular weight naphthenic acids have relatively high
-
water solubility compared to other crude oil components. High
mo-lecular weight naphthenic acids are interfacially active and
willform metal salts at the oilwater interface [50,51]. This leads
to pro-duction of acidic waste water, which represents a major
environmen-tal problem. It has been shown that a lamellar liquid
crystalline phase(LLC) might act as an efcient emulsier when
present in a three-phase system The LLC phase will cover the
emulsion droplets andlead to reduced interfacial mobility and
bending ability. Presence ofother organized structures in a
multiphase systemmight have the op-posite effect with regard to
emulsion stability. In contrast to lamellarliquid crystals,
microemulsions will reduce the emulsion stability.
Knowledge about phase behavior in an oilsurfactantwater systemis
of crucial importance when studying the mechanisms behind emul-sion
formation and stability in such a system. Hager and coworkers[52],
using model naphthenic acids, proved increasing pH decreasesthe
lipophilicity of the acid species and may induce a transition
fromw/o emulsion to an o/w emulsion. There were observed strong
compe-tition between salinity effect and pH effect especially in
case of systemswith anionic surfactant included fatty acids. Stable
w/o emulsions werefound at a pH close to 8. All emulsions were
shown to be stabilized by aliquid gel phase consisting of a
lamellar structure.
Many ternary systems based on a fatty acid (CnCOOH) and
long-chain alcohol (CnOH) have been thoroughly investigated in
past
pacity was obviously reduced. Hager et al. [55] also
investigated theeffect of the ratio between the undissociated
(RCOOH) and the disso-ciated (RCOO~) acid. It was detected the
solubilization limit of theacid around 10 wt.%. Increasing the
concentration of the surfactantleads to formation of the hexagonal
(E) phase. Addition of moreacid leads to a transition to a D phase
region above 40 wt.% surfactant.It means that replacing the
surfactant by the acid gave only normalmicellar and hexagonal
structures. The poor phase behavior of thesystem can be explained
in terms of a low ability of the acid to solu-bilisate into the
different self-assembly structures.
3.6.1. Langmuir lm propertiesA rigid interface on the emulsion
droplets prevents coalescence
while a highly compressible lm is more easily ruptured,
leavingthe droplets free to coalesce. By means of the Langmuir
technique,asphaltenes are found to build up close-packed rigid lms,
whichgive rise to quite high surface pressures. Resin lms, on the
otherhand, are considerably more compressible [56]. This may
explainthe experimental observations showing that asphaltenes are
able tostabilize crude oil based emulsions, while resins alone fail
to do so.The more hydrophilic resin fraction starts to dominate the
lm prop-erties due to the higher afnity towards the surface. Highly
compress-ible resin lms alone will not stabilize a crude oil
emulsion. Related tothis, demulsiers, which form lms of low
rigidity and high com-
120 I. Kralova et al. / Advances in Colloid and Interface
Science 169 (2011) 106127years. However, there are only a few
reports where the phase behaviorhas been studied in systems based
on polar aromatic acids or alcohols.Horvath-Szabo and coworkers
[53,54] have studied the phase equilib-ria in a sodium
naphthanatewater system and sodium naphthanatewaterhydrocarbon
systems. The inuence on phase diagrams ofmodel compounds for
naphthenic acids and phenols when mixedwith water were studied
based on the compounds are 5-phenylvalericacid, 5-phenylvalerate,
1-decanol, and 4-pentylphenol [55]. Fig. 26 il-lustrates an
extended isotropic o/wmicellar (L1) phase exists up to
ap-proximately 45 wt.% surfactant along the binary surfactantwater
axis.In the dilute aqueous regime only monomer exists. Addition of
morealcohol leads to a turbid lamellar liquid crystalline (D)
phase, whichwas identied by microscopic investigation (Fig. 27 (a)
and (b)). Thetypical texture of Maltese crosses and oily steaks is
an indication ofthe lamellar structure. The reason for this
transformation is that the al-cohol gives the aggregates a more
hydrophobic character and reducesthe charge density of the polar
surface layer, which will favor the for-mation of the lamellar
bilayer structure. This D phase has a very highswelling capacity,
i.e., it has a large extension towards the water
Fig. 26. Partial phase diagram based on weight fraction of the
ternary system based on
5-phenylvalerate, 4-pentylphenol, and water at 25 C.apex. The
swelling of a lamellar phase can be explained by a differencein
chemical potential of pure water and water incorporated into the
in-tervening layers of the lamellae. A lamellar phase in the
producedwater of a crude oil may have a large impact of the
elements that con-taminate the water stream. That the lamellar
phase is formed at lowconcentrations is important information when
treating productionwater. In the alcohol-rich comer of the phase
diagram the reverse w/o micellar (L2) phase exists. Also a reverse
hexagonal (F) structure isformed between the D and L2 phase. This
phase is found at the surfac-tant content around 20 wt.%.
Microscopic investigation identied theanisotropic phase on the
basis of the characteristic optical fan-like tex-ture (Fig. 27
(c)). The two-phase region between the D and F phasewas also
recorded by polarizing microscopy (Fig. 27 (d)).
The phase behavior was found to depend on salinity of the
system.The system in the presence of salt was shown to signicantly
changethe character of the lamellar phase. Its region of existence
was muchsmaller than that the system in the absence of salt, i.e.,
its swelling ca-
Fig. 27. Light microscopymicrographs showing the (a)Maltese
cross and (b) oily steak tex-tures of the lamellar liquid
crystalline (D) phase based on5-phenylvalerate, 4-pentylphenol,and
water at 25 C. The fan-like texture in (c) is an indication of the
hexagonal liquid crys-talline (E or F) phase. The two-phase region
(d) consisted of a lamellar and hexagonalphase; both Maltese cross
and fan-like textures were observed, respectively.pressibility,
should be most efcient [57]. When used as demulsiers,
-
the efciency depends on the ability of the chemicals to interact
withand modify the lm built up by asphaltene particles. Addition
ofdemulsiers of high molecular weight in the asphaltene lm gavethe
isotherms (Fig. 28). In this gure, Chemical G is highly
effectivewith respect to increased compressibility together with a
reduced ri-gidity. The efciency depends not only on the direct
inuence ofchemical additives within the lm, but also on the ability
of demulsi-ers to reach the w/o interface in an emulsion (diffusion
through theuid). This is a critical step regarding the effective
concentration ofdemulsiers at the interface. The results obtained
from the Langmuirinterfacial lm studies are important in explaining
why certain che-micals are more effective as inhibitors than as
demulsiers. Obviouslythe inhibitor/asphaltene interaction is so
strong in the bulk oil phasethat the interfacial structures being
gradually built up will no longerpossess properties required to
stabilize w/o emulsions.
Fig. 29 illustrates interfacial pressure isotherms of lms formed
be-
ship based on an appropriate lm drainage time, the natural
formationof the dense packed layer is determined. This model
represents a gen-eral approach to modeling batch gravity separation
based on rstprinciples and provides the methodology to incorporate
various ef-fects of the bulk uid properties, interfacial
properties, properties ofsurfactants, and the cumulative squeezing
force encountered in a sub-stantially thick layer of densely packed
droplets.
3.7.1. Model descriptionAs discussed above, the model is based
on a population balance
framework. This implies that the drop-size distribution is
representedby a continuous density function that represents the
average numberof droplets within an innitesimal droplet volume
element and inn-itesimal axial position element. The consequence of
this denition isthe ability to express the motion of droplets
through real space bytheir sedimentation velocity while the
coalescence process can berepresented as a birth-death process
where the droplet volume is
121I. Kralova et al. / Advances in Colloid and Interface Science
169 (2011) 106127tween water containing different types of
particles and an oil phase ofpure decane determined with the
Langmuir technique. The oil phaseconsists of pure decane while 1
wt.% particles are added to the waterphase. One of the particle
fractions (type I) forms a relatively rigid lmat the
interface,while theother (type II) ismore hydrophilic and
remainsdispersed in the aqueous phase. Studies of these particles
ability to stabi-lize emulsions have shown that type I particles
are highly efcient emul-siers, while type II particles will not
form stable emulsion [58].
3.7. Separation modeling
The separation of liquidliquid dispersions in the form of
crude-oil and water emulsions is a particularly crucial process in
the petro-leum industry, and the development of a fundamental
understandingand elucidation of this process is becoming more
important as exist-ing oil elds mature and newer elds with
increasingly heavier oilsare relied upon to meet global demands for
energy. Curiously, de-spite the long history of large scale
oilwater separation processes,modeling of the emulsion gravity
separation process has been limit-ed [5968] and rarely elucidates
the effect of the poly-dispersity ofthe emulsion drop-size
distribution [6568]. Additionally, Grimes etal. [69] have
demonstrated that surface active compounds can havenon-uniform
local concentrations in a separating emulsion whichcan dynamically
evolve with time; such a mechanism can have sig-nicant implications
with respect to the chemical additives as wellas indigenous
compounds of a crude oil, particularly in terms ofcompounds that
stabilize emulsions or cause scaling and foulingwhich are prevalent
in heavy crude oils. Consequently, there still isneed for an
emulsion gravity separation model that properly ac-counts for the
effect of the poly-dispersity of the emulsion, whileemploying
physically meaningful, mechanistic models for sedimenta-tion,
coalescence, and droplet collision that can be tied directly to
thelocal concentrations of key interfacially active molecules.
Fig. 28. A isotherms of mixed monolayers of asphaltenes and two
different demul-sifers on pure water. The mixed monolayers are
compared with the pure asphaltene
and the pure resin lm.In the following paragraphs, a new,
fundamental approach [70,71]for modeling the batch separation of
emulsions in terms of the popu-lation balance equation [70,72,73]
is discussed. Readers interested inthe detailed formulation of the
model and development of the systemequations are directed to Grimes
[70]. The formulation of this model[70] elucidates the methodology
of incorporating the physical proper-ties of the bulk liquids as
well as the physical properties of the phaseinterface by
formulating the coalescence closure relationships in termsof
coalescence times obtained from lm drainage models [7479]
thatappropriately describe the specic emulsion system as well as
colli-sion frequencies determined by the simultaneous incorporation
of dif-ferential sedimentation and Brownian motion; it should be
noted thatthe appropriate lm drainage model should be chosen to
best describethe specic oilwater system, e.g., one must consider
the nature of theadsorbed surfactant layer and determine if
interfacial mobility is im-portant and identify the physical
mechanisms that contribute to thedisjoining pressure while the
physical properties of the uids dictatewhether signicant dimpling
of the liquid lm occurs [7679], or ifthe liquid lm can be
satisfactorily modeled with a parallel disk lmdrainage model
[74,75]. This approach [70] can satisfactorily incorpo-rate both
the physical properties of the bulk uids as well as the phys-ical
properties of the phase interface to describe the binarycoalescence
phenomena. Additionally, interfacial coalescence isaccounted for in
terms of the coalescence time obtained from appro-priate lm
drainage expressions and, critically, a method is presentedthat
explicitly accounts for the deformation of the emulsion zone dueto
the dynamic growth of the volume of the resolved dispersed
phase.Furthermore, hydrodynamically hindered sedimentation is
consideredand, when combined with an interfacial coalescence
closure relation-
Fig. 29. Interfacial pressure isotherms of lms formed between
water containing differ-ent types of particles and an oil phase of
pure decane.conserved upon each coalescence event [72].
Furthermore, measureable
-
oil B and the demulsier additive will be referred to as chemical
7.Model simulations are compared to experimental data for the
separa-tion of crude oil B emulsions when the concentration of
chemical 7 is10 ppm and 50 ppm. Table 7 lists the model parameters
for the sepa-ration system having a concentration of chemical 7 at
10 ppm whileTable 8 lists the model parameters for the system
having a concentra-tion of chemical 7 at 50 ppm.
In Fig. 30, the initial volume and number density distribution
ofthe droplets is given. The volume density distribution was
obtaineddirectly from the NMR measurements and the number density
wasconverted directly from the measured volume density by
conserv-ing the fraction of the area under the curve at the mean
droplet ra-dius and median droplet radius. The average droplet size
by volumewas determined from the st moment of the volume density
and is11.9 m. The characteristic radius, dened as the value of the
drop-let radius where the integral of the initial volume density is
0.990,is 20.5 m. This distribution is composed of a fairly small
initialdroplet size distribution with a small to medium poly
dispersity rel-ative to the next case studied which is crude oil B
with 50 ppm ofchemical 7.
In Fig. 31 the iso-volume fraction proles of the dispersed
phaseare presented for the separation of crude oil B with 10 ppm of
chem-ical 7; the open symbols are the experimental data measured by
NMRand the solid lines are the model prediction. The results in
Fig. 31 in-dicate that the model prediction is very good for 1%
water (the dis-persed phase is water) and 100% water. The worst
model predictionis for the 10% water curve and, while the initial
prediction for the de-velopment of the dense packed layer is very
good, the simulation re-sults diverge from experiment as time
progresses. One important tonote is that both experiment and theory
indicate that a small popula-
122 I. Kralova et al. / Advances in Colloid and Interface
Science 169 (2011) 106127quantities such as the local droplet
volume fraction, local mean dropletvolume (radius), and standard
deviation of the droplet distribution canbe obtained directly from
the statistical moments of the droplet volumedensity function.
Finally, the coalescence process can then be describedin terms of
conformal integrals of the local droplet volume density func-tion
that employ a collision and efciency kernel that describes
theprobability of coalescence in terms of the bulk physical
properties ofthe uids and interfacial forces that are derived from
fundamental rstprinciples of physics.
In summary, the model is formulated to include the following
spe-cic attributes and physical mechanisms:
Dynamic evolution of a non-uniform (poly-disperse)
drop-sizedistributionHindered sedimentation and formation of a
dense packed layerDroplet collisions based on simultaneous
consideration of differ-ential sedimentation rates and Brownian
motionFilm drainage based on density difference, continuous phase
vis-cosity, interfacial tension, droplet radius, and interfacial
forces(expressed in terms of London-van der Waals dispersive
forces)Interfacial coalescence of drops with their homophase
expressedin terms of the lm drainage equation described
aboveTracking explicitly the moving interface of the separated
dis-persed phase (homophase) through the employment of
Eulerianboundary conditions
3.7.2. Notes on the lm drainage modelA complex model for lm
drainage based on the models devel-
oped by Slattery and coworkers [7478] has been constructed
andsolved during the course of this work. The numerical lm
drainagemodel is used to provide a detailed picture of the lm
drainage phe-nomenon based on the characteristic parameters of the
systems stud-ied. That is, based on the dispersed and continuous
phase densities,the continuous phase viscosity, the interfacial
tension, the interfacialdilational and shear viscosities, and the
interfacial forces involved,and appropriate model for the lm
drainage time can be selected tobest represent the system. Factors
to consider are, for example, thedeformation of the droplet
interface, the effect of surfactant mobility,and the effect of the
surface viscosity. The complete numerical lmdrainage model
considers the following:
London-van der Waals dispersion forces between
interfaciallayersInterfacial stress deformation expressed in terms
of interfacialtension as well as interfacial shear and dilational
viscositiesAxisymmetric deformation of the interfacial
lmInterfacial surfactant diffusion (GibbsMaragoni effect)
The complete details of the model formulation will not be
repeat-ed here; readers interested in the detailed formulation of
the lmdrainage model can consult Ref. [78].
Simulations of the numerical lm drainage model indicated thatfor
a viscous crude oil employed in the results section below
havingdrop radii less than 150 m, the interfacial lm could be
adequatelyrepresented by a parallel disk lm drainage model [74].
The represen-tation of the interfacial forces purely in terms of
the retardedHamaker constant is, strictly speaking, not entirely
correct as the in-terfacial forces due to steric repulsions should
also be considered[80,81]. However, in this case, the retarded
Hamaker constant canbe considered to be a lumping of these two
mechanisms and is re-ferred to as an interfacial force parameter.
The separation of theseto interfacial force mechanisms will be
addressed in later work.
3.7.3. Separation modeling resultsSimulations of the model were
applied to NMR data for the sepa-
ration of a heavy crude oil with various concentrations of a
chemical
demulsier additive. The heavy crude oil will be referred to as
crudetion of very small droplet eventually becomes isolated at the
top ofthe separator since earlier binary coalescence thins out the
popula-tions and decreases the collision rate of these droplets
signicantly.Thus, the very small droplets b5 m remain in an
effective suspendedstate since the high viscosity of the oil phase
and small droplet sizes
Table 7Model parameters for the separation of crude oil B with
10 ppm of chemical 7.
Parameter description Symbol Value
Initial water volume fraction 0 0.3767Absolute temperature T
306.15 K (33 C)Column height Hc 1.74 cmDispersed phase density d
995 kg/m3
Continuous phase density c 927 kg/m3
Continuous phase viscosity c 118 mPasEquilibrium interfacial
tension 0 14 mN/mInterfacial force parameter B0 4.051036 N/m2
(est.)Fitting parametersCollision rate parameter kcr 1.0Coalescence
efciency parameter kef 0.5Interfacial coalescence parameter kic
1.0
Table 8Model parameters for the separation of crude oil B with
50 ppm of chemical 7.
Parameter description Symbol Value
Initial water volume fraction 0 0.3773Absolute temperature T
306.15 K (33 C)Column height Hc 1.68 cmDispersed phase density d
995 kg/m3
Continuous phase density c 927 kg/m3
Continuous phase viscosity c 118 mPasEquilibrium interfacial
tension 0 10.8 mN/mInterfacial force parameter B0 1.651031 N/m2
(est.)Fitting parametersCollision rate parameter kcr 1.0Coalescence
efciency parameter kef 0.5Interfacial coalescence parameter kic
1.0
-
Fig. 30. Initial drop volume and number density distribution of
crude oil B with 10 ppm of chemical 7. The mean radius by volume is
11.9 m.
123I. Kralova et al. / Advances in Colloid and Interface Science
169 (2011) 106127mean any further separation will only occur after
extremely long pe-riods of time. This result may have important
implication in terms ofaccessing processing strategies for viscous
crude oils.
In Fig. 32, the axial volume fraction proles (separation proles)
atseveral different times are presented for the mode