Sunteți pe pagina 1din 10

Ind. Eng. Chem. Res.

2004, 43, 6861-6870 6861

Kinetics of Asphaltene Flocculation

Khashayar Rastegari, William Y. Svrcek, and Harvey W. Yarranton*


Department of Chemical and Petroleum Engineering, University of Calgary,
Calgary, Alberta, Canada T2N 1N4

The kinetics of asphaltene particle growth in model systems of asphaltenes, n-heptane, and
toluene were investigated using a particle-size analyzer. Individual precipitated asphaltene
particles were found to be on the order of 1 µm in diameter. The asphaltene particles formed
loose fractal-like flocs with volume mean diameters ranging up to 400 µm and a fractal dimension
of approximately 1.6. Asphaltene flocculation was reversible, and the flocculation rate increased
with increased asphaltene concentration, increased heptane content, and decreased shear rate.
A reversible kinetic flocculation model successfully fit the particle-size distributions and growth
in mean volume diameter over time.

Introduction showed that the fractal dimension for the three lower
molar mass fractions for five gel-permeation-chromato-
Heavy oil and bitumen resources are gaining in graphic fractions of Athabasca asphaltenes varied from
prominence as conventional oil supplies are depleted. 1.65 to 1.68 while the fractal dimension of the two
These heavier crude oils have a high asphaltene content higher molar mass fractions was approximately 2.0.
and are more prone to asphaltene deposition problems. These results indicate that the fractal dimension could
Chemical and mechanical methods have been used to depend on the molar mass of the asphaltenes.13 Both
clean up asphaltene deposits in oil wells, reservoir, and irreversible diffusion-limited and reaction-limited ki-
pipelines.1,2 These remedies are not always effective, netic models have been used to model asphaltene
and to better mitigate asphaltene deposition, a better flocculation in a mixture of n-heptane and toluene.4,6,14
understanding of asphaltene deposition is required. The fractal dimension of asphaltene flocs was treated
Deposition likely begins with asphaltene precipitation as a tuning parameter to correlate the experimentally
caused by the introduction of diluents (for viscosity observed rate of asphaltene particle growth. The fitted
reduction) or changes in temperature and pressure fractal dimensions varied from 1.49 to 1.77 for a
between the reservoir, well bore, and production facili- diffusion-limited flocculation model and approximately
ties. Once asphaltenes precipitate, the precipitated 2.0 for a reaction-limited model.
particles tend to flocculate. The size of the asphaltene The asphaltene floc size depends on the properties of
flocs is likely an important factor in asphaltene deposi- the medium and the flow conditions, including temper-
tion because larger flocs are more likely to settle and ature, pressure, and shear rate. In general, the less
form deposits. However, most research has focused on aromatic is the medium, the greater is the floc size. In
asphaltene precipitation, and there has been relatively mixtures of aromatics such as toluene and n-alkanes
little work on asphaltene flocculation. such as n-heptane, the asphaltene mean floc size
Asphaltene floc sizes from one to several hundred increases as the n-alkane content or asphaltene con-
microns have been reported. For dilute dispersions of centration in solution increases. Ferworn et al.9 and
asphaltenes in mixtures of n-heptane and toluene, the Mason and Lin14 showed that, for n-alkane-diluted oils,
mean diameter of asphaltene flocs ranged from 0.4 µm the floc size increased as the n-alkane content or
at the initial stage of flocculation to 12 µm after 30 h.3-6 paraffinic light oil content increased. The size of the
It is believed that asphaltene flocs in the size range of asphaltene flocs also increased as the number of carbon
1 µm are individual particles rather than flocs.3,7,8 In atoms in the n-alkane diluent decreased.7,10
n-alkane-diluted bitumen, the asphaltene flocs were The effect of temperature and pressure on asphaltene
found to be far from spherical, with diameters ranging flocculation is not well established. Funk7 reported that
from 80 to 600 µm.9,10 Ferworn et al. reported unimodal the mean floc size of asphaltenes precipitated from tar
size distributions of flocculated asphaltenes, but bimodal sand increased as the temperature increased. However,
size distributions have been observed for two low- Neilson et al.11 found that the mean asphaltene particle
viscosity oils.11 size in n-alkane (n-pentane and n-heptane)-diluted
Asphaltene flocs have a loose structure, and several bitumen decreased as the temperature increased. Neil-
researchers have attempted to determine a fractal son et al.11 also found that the floc size increased slightly
dimension for flocculated asphaltenes. Sheu et al.12 used with increased pressure.
a series of time-dependent viscosity measurements to The effect of shear on asphaltene flocculation is also
determine that the fractal dimension of asphaltene flocs not well established although flocculation is known to
varied from 1.49 to 1.8 depending on the types of be shear-sensitive. Flocculation was observed to be very
asphaltenes and the asphaltene concentrations. A low- unstable in toluene and n-heptane mixtures. With slight
temperature electron spin-lattice relaxation technique shaking, flocs disintegrated into what appeared to be
individual particles with a diameter on the order of 1
* To whom correspondence should be addressed. E-mail: µm.3 Ferworn et al.9 also observed that asphaltene flocs
hyarrant@ucalgary.ca. disintegrate upon agitation.
10.1021/ie049594v CCC: $27.50 © 2004 American Chemical Society
Published on Web 09/21/2004
6862 Ind. Eng. Chem. Res., Vol. 43, No. 21, 2004

One other factor that has received little attention is configured with one of two lenses for different ranges
the kinetics of asphaltene flocculation. Because oil of particle sizes: lens A, 0.5-150 µm; lens B, 5-600 µm.
production and processing take place in relatively short The particle-size measurements were performed on
time scales (minutes rather than hours), the kinetics of samples of approximately 3.5 cm3 placed in a 1 cm × 1
asphaltene flocculation can have a significant impact cm glass cuvette equipped with a magnetic bar for
on the potential for deposition. In this work, the kinetics stirring. Particles could be detected at asphaltene
of asphaltene flocculation are investigated in mixtures concentrations up to 0.1 kg/m3. At higher asphaltene
of n-heptane, toluene, and asphaltenes. This model concentrations in the model systems and for heptane-
system is employed because the fluid is sufficiently diluted bitumen, the fluid was too opaque to obtain a
transparent for particle-size analysis (at asphaltene measurement. At asphaltene concentrations below 0.05
concentrations below 0.1 kg/m3). Also, asphaltenes are kg/m3, there were too few particles to obtain an accurate
soluble in toluene and insoluble in n-heptane. Hence, analysis. Therefore, the experiments were limited to a
the amount of asphaltene precipitation can be experi- narrow range of asphaltene concentrations: 0.05-0.1
mentally controlled through both the heptane/toluene kg/m3.
ratio and the asphaltene concentration. The effects of To prepare a sample, asphaltenes were dissolved in
the asphaltene concentration, the heptane/toluene ratio, toluene and sonicated for 1 h at 23 °C to ensure that
and shear are considered, and a kinetic model capable all of the asphaltenes dissolved. Note that some of the
of fitting the data is developed. solutions were tested in the analyzer with lens A and
no measurable particles were detected; that is, no
Experimental Method asphaltene particles greater than 0.5 µm in diameter
were present. At time zero, heptane was added to the
Material. Toluene and n-heptane of greater than 99% solution of asphaltenes and toluene. The mixture was
purity were obtained from Aldrich Chemical Co. Hell- slightly shaken and then transferred to the cuvette.
manex cleaning concentrate and optical sampling cells Because each particle-size analysis requires 2-3 min,
were obtained from Hellma Cell Inc. Athabasca coker- the first measurement was obtained 5 min after the
feed bitumen was obtained from Syncrude Canada Ltd. heptane addition. Visual observation indicated that, in
The coker-feed bitumen has been treated to remove this first 5 min, the majority of the precipitated particles
solids and waters. The residual water and solids con- formed almost instantly and began to flocculate im-
tents are approximately 0 and 1 wt %, respectively. mediately.
Asphaltenes were precipitated from the bitumen upon At each specific condition, the experiments were
the addition of 40 cm3 of n-heptane/g of bitumen. The repeated for three separate samples, and three observa-
mixture was sonicated for 45 min at room temperature tions were obtained at each time interval. The variables
and left for 24 h, and then the supernatant was filtered and conditions investigated were as follows: heptane/
using Whatman’s No. 2 (8-µm pore size) filter paper. toluene volume ratio, 60:40 Heptol and 70:30 Heptol;
More n-heptane was added to the filter cake at a ratio asphaltene concentration, 0.05, 0.08, and 0.1 kg/m3;
of 4 cm3 of n-heptane/g of original bitumen. The mixture mixing speed, 110, 230, and 600 rpm; time, 5, 30, 45,
was sonicated for 30 min, left overnight, and again 60, 90, 120, 150, 180, 240, 300, and 360 min.
filtered using the same filter paper. The dried filter cake All experiments were performed at a temperature of
is termed the “unwashed C7 asphaltenes”. 23 °C. For convenience, a solution of heptane and
Unwashed C7 asphaltenes may contain resinous toluene is described as X:Y Heptol, where X and Y are
material and trapped maltenes. To wash out this the volume fractions of heptane and toluene, respec-
material and obtain “purer” asphaltenes, the unwashed tively. The mixing speeds were measured with a tacom-
C7 asphaltenes were placed in a Soxhlet apparatus and eter. Because the magnetic bar is free to move in all
refluxed with n-heptane for 3-4 days. The process was directions, an exact correlation for the shear rate is not
interrupted once a day to crush and mix the asphaltene available. If we assume that the magnetic bar is
sample to increase contact between the asphaltenes and equivalent to an impeller, an approximate shear rate
n-heptane. The residual dried asphaltenes are termed for laminar and transitional flow is given by17
“Soxhlet-washed C7 asphaltenes”.15
The Soxhlet-washed C7 asphaltenes may contain non-
asphaltenic solids that coprecipitate with the asphalt- γ̆ ) 13N(dimp/da)2 (1)
enes.16 To remove these solids, the Soxhlet-washed C7
asphaltenes were dissolved in toluene at a concentration where γ̆ is the shear rate (s-1), N is the rotational speed
of approximately 10 kg/m3. The mixture was centrifuged (s-1), dimp is the impeller diameter (cm), and da is the
at 4000 rpm for 5 min. The supernatant was decanted, diameter of the vessel (cm). The approximate shear
and the toluene was evaporated from the mixture in air rates corresponding to rotational speeds of 110, 230, and
at room temperature overnight and then in a vacuum 600 rpm are 9.2, 19.2, and 50 s-1, respectively. Given
oven in nitrogen at 70 °C. The dried residual asphalt- the uncertainty in the shear rate calculation, the
enes are referred to as “solids-free Soxhlet-washed C7 experimental results are reported in terms of the
asphaltenes”. All of the experiments presented here rotational speed.
were performed on solids-free Soxhlet-washed C7 as- The laser in the particle-size analyzer focuses on a
phaltenes, and they will be referred to simply as small fixed area within the cuvette. It is possible that
“asphaltenes” hereafter. the particles within this area are not representative of
Particle-Size Analysis. Asphaltene particle-size the overall particle-size distribution. Therefore, the
distributions were measured with a Brinkmann Instru- cuvette was shifted and some particle-size analyses were
ments 2010 particle-size analyzer. This apparatus uses repeated at five different locations. No significant dif-
the time-of-transition method; that is, the diameter of ference in the particle-size distribution was observed.
a particle is determined from the time a moving laser Also note that no particles were detected in the pure
beam is obscured by the particle. The apparatus can be toluene and pure heptane used in these experiments.
Ind. Eng. Chem. Res., Vol. 43, No. 21, 2004 6863

Previous asphaltene flocculation models have consid-


ered irreversible reaction-limited kinetics.4,6 However,
the experimental data presented later suggest that
asphaltene flocculation is a reversible process.3,9 In this
case, flocs are held together with dispersion forces and
can be separated if sheared. Reversible flocculation can
be described as follows:
Fi,j
Mi + Mj {\
D
} Mi+j (2)
i+j

where Fi,j and Di+j are the reaction rates for the
flocculation and disintegration processes, respectively,
and Mi and Mj indicate the masses of species i and j,
respectively. The Smoluchowski approach is modified
to obtain a population balance equation for the kth
species,18 given by

Figure 1. Kinetics of fractional precipitation of 0.5 and 1.0 kg/ dCk 1


m3 asphaltenes in 60:40 Heptol at 23 °C. Solid lines are a visual
aid only. dt
) ∑ (Fi,jCiCj - Di+jCi+j) -
2 i+j)k
N-k

Asphaltene Precipitation. To interpret some of the ∑ (Fk,jCkCj - Dk+jCk+j) (3)


results, it was necessary to determine the amount of j)1
precipitated asphaltenes over time. The kinetics of
asphaltene precipitation over 6 h were determined in k ) 1, 2, 3, ..., N
solutions of 60:40 Heptol at asphaltene concentrations
of 0.5 and 1 kg/m3 and at a temperature of 23 °C. where the indices i, j, k, and N indicate the number of
Asphaltenes were dissolved in toluene and sonicated for particles in the floc; Fi,j and Di+j are the number of
1 h. Then, n-heptane was added at a 60:40 volume ratio reactions per unit volume per unit time (or reaction
of n-heptane/toluene, and the mixture was slightly terms) that result in flocculation and disintegration,
shaken. After a given time, the mixture was centrifuged respectively; Ci and Ci are the concentrations of species
at 4000 rpm for 5 min, and the supernatant was i and j; and t is the time.
removed. The asphaltene deposits were first dried in air To apply the population balance, it is necessary to
at room temperature overnight and then in a vacuum evaluate the flocculation and disintegration reaction
oven in nitrogen at 70 °C. The fractional asphaltene rates. For flocs of approximately equal mass (Mi ≈ Mj),
precipitation is the mass of the dried precipitate over Ball et al.19 proposed that the flocculation reaction terms
the mass of asphaltenes in the initial solution. Figure depend on a boundary volume defined by the radii of
1 shows fractional asphaltene precipitation for 0.5 and the two flocs. For spheres, the reaction rate expression
1 kg/m3 asphaltenes in 60:40 Heptol. is given by

Asphaltene Flocculation Model Fi,j ∝ Vc ) π(di + dj)2w (4)

Two characteristic time scales can govern flocculation where di and dj are the diameters of the flocs and w is
kinetics: the diffusion time and the reaction time. The a separation distance within which flocculation can
diffusion time, τdiff, is the typical time that it takes for occur. For flocs of any geometry, the radius of the
two arbitrary particles to meet each other. The reaction particle can be related to the number of individual
time, τreact, is the typical time for two particles to particles in a floc using a fractal dimension as follows:
flocculate when they are held in mutual proximity.
When τdiff . τreact, the kinetics are diffusion-limited; that dfloc ) nf1/Df dp (5)
is, flocculation is a diffusion-controlled process. In this
case, particles most likely flocculate on their first
encounter with a neighbor, and hence the kinetics are where dfloc is the measured diameter, Df is the diameter
largely influenced by local fluctuations in the concentra- of gyration of the floc (approximately equivalent to the
tion of the particles. When τdiff , τreact, flocculation is measured spherical mean diameter), and nf is the
limited by the relatively large reaction time and is number of individual particles in the floc. Equation 4
known as a reaction-limited or reaction-controlled floc- can then be expressed in terms of the number of
culation. In this limit, particles may come within the individual particles in the flocs:
reaction range of each other numerous times before they
actually flocculate. In other words, a particle may Fi,j ) Kf(ni1/Df + nj1/Df)λ (6)
sample large volumes of its surrounding neighbors
before flocculating, and hence it effectively responds to where Kf is the flocculation reaction constant, ni and nj
the global concentration of other particles. Stirring is a are the number of individual particles in flocs i and j,
very efficient method of shortening the characteristic and λ is an adjustable parameter with a value between
diffusion time and smoothing out local concentration 2 and 3. The exponent does not necessarily equal 2
fluctuations. Hence, reaction-limited flocculation is because the effective surface area of a nonspherical floc
expected to be the dominant mechanism in the well- is not necessarily proportional to the square of the
mixed experiments examined here. diameter.
6864 Ind. Eng. Chem. Res., Vol. 43, No. 21, 2004

Preliminary modeling results using eq 6 could not


predict the change in the flocculation rate as the floc
size increased with time. Better predictions were ob-
tained when the reaction term was controlled by the size
of the smaller floc. Ball et al.19 argued that flocculation
between particles of different size was controlled by the
penetration of the smaller particle into the larger
particle. Such a process may be controlled by the size
of the penetrating particle. Therefore, the following
empirical expression for the overall reaction rate term
is proposed:

( nj ni
)
λ
Fi,j ) Kf ni1/Df + n 1/Df (7)
ni + nj ni + nj j

When ni ) nj, eq 7 reduces to eq 6. When ni . nj, eq 7


reduces to Fi,j ) Kfnjλ/Df; that is, the flocculation rate is
determined from the size of the smaller of the two
colliding particles.
Disintegration may occur through erosion or shatter-
ing. Erosion is the removal of individual particles from
flocs through grinding or shear. Shattering is the
fragmentation of flocs into smaller flocs of any size less
than the original floc. The experimental results dis-
cussed later suggest that shattering is the dominant
mechanism for the model systems considered here. The
shattering reaction rate is given by20

Di+j ) Si+j ) Kddi+jβ ) Kd(ni1/Df + nj1/Df)β (8)

where Kd is the shattering reaction constant and di is


the radius of the floc. The exponent β is an adjustable
parameter with a value between 0 and 3. Note that, if
required, the erosion reaction term has the same form
as the shattering reaction term except that one of the
particle numbers ni or nj must be unity.

Results and Discussion

Micrographs of asphaltene flocs in solutions of 60:40


Heptol and 70:30 Heptol are presented in parts a and b
of Figure 2, respectively. Figure 2a shows flocs of
relatively small size, and it can be seen that the
individual asphaltene particles are approximately 1-2
µm in diameter. Figure 2b shows larger flocs, and a
typical asphaltene floc is circled to illustrate its spheri-
cal diameter. The spherical diameter is similar to the
diameter measured with the particle-size analyzer.
Figure 2c shows the microscopic image of asphaltene
flocs in heptane-diluted bitumen at a 4:1 heptane/
bitumen volume ratio. The asphaltene flocs in a hep-
tane-diluted bitumen mixture resemble the flocs in the
Heptol solutions.
Typical number and volume floc size distributions for
asphaltenes in 60:40 Heptol are presented in parts a
and b of Figure 3, respectively. Figure 3a shows a large
population of particles in the 0.5-2-µm-diameter range.
A similar large population of small particles was
observed in all of the experiments in this study, and at
higher mixing rates (shear rate) or at low flocculation Figure 2. Precipitated asphaltenes 10 min after contact with
times, the bulk of the precipitated particles fell in this heptane: (a) 0.1 kg/m3 asphaltenes in 60:40 Heptol; (b) 0.1 kg/m3
range. Based on the micrographs in Figure 2, this asphaltenes in 70:30 Heptol; (c) 4:1 heptane/Athabasca bitumen
population can be interpreted to be individual asphalt- volume ratio.
ene particles. Note that, because the lower detection
limit of the particle-size analyzer is 0.5 µm, it is probable of the flocs in Figure 2 suggests that the majority of
that smaller individual particles make up part of the the individual asphaltene particles are greater than 0.5
particle-size distribution. However, a close examination µm in diameter.
Ind. Eng. Chem. Res., Vol. 43, No. 21, 2004 6865

Figure 5. Reversibility of asphaltene particle growth for 0.1 kg/


m3 asphaltenes in 60:40 Heptol.

Heptol at a mixing rate of 110 rpm. Three distinct


regions are shown: (1) a slow-growth region, (2) a rapid-
growth region, and (3) a steady-state plateau. The period
of slow initial growth followed by rapid growth suggests
that flocculation depends on the diameter to a higher
order than unity; for example, flocculation may depend
on the surface area, cross-sectional area, or volume of
the flocs. Therefore, as the flocs grow, the flocculation
rate increases. A physical interpretation is that larger
flocs have more potential bonding sites; i.e., they grow
faster than smaller flocs. This behavior is characteristic
of reaction-limited flocculation and will be discussed in
more detail in the kinetic modeling section.
Figure 3. Measured asphaltene particle distributions after 60 Note that the average particle size decreases in the
and 360 min for 0.1 kg/m3 asphaltenes in 60:40 Heptol at a mixing first 30 min. During this time, asphaltenes continue to
rate of 110 rpm: (a) number frequency; (b) volume frequency. precipitate, as was shown in Figure 1. Either new
particles are forming or existing particles are growing
to detectable sizes. In either case, small new particles
are joining the observed particle-size distribution. There-
fore, the mean volume diameter of asphaltene flocs
decreases as the population of the particles in the
diameter range of 0.5-2.0 µm increases. After 30 min,
little further precipitation occurs and no further de-
crease in the particle-size distribution is observed.
Reversibility of Asphaltene Flocculation. Figure
4 also shows that asphaltene flocs reach a steady-state
size distribution, indicating that flocculation does not
continue indefinitely. To achieve a steady-state condi-
tion, flocculation must eventually be balanced by dis-
integration. Hence, asphaltene flocculation appears to
be reversible.
To confirm that asphaltene precipitation is reversible,
asphaltenes were allowed to flocculate at low shear,
Figure 4. Kinetics of asphaltene particle growth for 0.1 kg/m3 then the shear rate was increased, and the change in
asphaltenes in 60:40 Heptol at a mixing rate of 110 rpm: (1) slow- the particle-size distribution was observed. Flocculation
growth region; (2) rapid-growth region; (3) steady-state region. was initiated in a solution of 0.1 kg/m3 asphaltenes in
60:40 Heptol at a mixing speed of 110 rpm. After 2 h,
While the particle number count is dominated by the mixing rate was increased to 600 rpm for 45 min
unflocculated asphaltene particles, the majority of the and then reduced back to 110 rpm. Figure 5 shows that,
asphaltenes are, in fact, flocculated as shown by the as the shear rate increased, the size of the asphaltene
volume-size distribution in Figure 3b. The “particles” flocs decreased sharply while the total number count
greater than 2 µm in diameter are assumed to be flocs. increased. When the shear rate decreased, asphaltene
The 360-min volume distribution in Figure 3b shows flocculation was observed again. Figure 5 demonstrates
that more than 90% of the mass of precipitated asphalt- that asphaltene flocculation is shear-sensitive and that
enes is distributed in flocs greater than 10 µm in asphaltene flocculation can be reversed at high shear
diameter. These flocs are similar to those in Figure 2b. rates.
Figure 4 shows the change in the volume mean Interestingly, the flocculation rate after the high
diameter with time for 0.1 kg/m3 asphaltenes in 60:40 shear period is slower than the initial flocculation rate.
6866 Ind. Eng. Chem. Res., Vol. 43, No. 21, 2004

Figure 6. Volume frequency distribution of asphaltene flocs Figure 7. Effect of the concentration on the asphaltene floc size
before (120 min) and after (165 min) 45 min of a high shear rate in 60:40 Heptol at a mixing rate of 110 rpm. Symbols are
of 600 rpm (0.1 kg/m3 asphaltenes in 60:40 Heptol). experimental data; lines are model predictions.

One plausible explanation is a restructuring effect


caused by the shear forces. The initial flocculation starts
from a nearly uniform particle-size distribution, takes
place at a low shear rate, and generates loose fractal-
like flocs. The higher shear rate breaks the fractal
structure of the flocs, and their size distribution ap-
proaches, but does not reach, the size distribution of the
individual particles (Figure 6). When the shear rate is
decreased, flocculation starts from a nonuniform size
distribution. It is likely that smaller flocs interpenetrate
larger flocs, increasing their fractal dimension but not
changing their diameter significantly. In other words,
the fractal dimension of the flocs increases rapidly while
their size grows slowly.
The post-shear size distribution of Figure 6 is also
characteristic of a shattering process. Erosion tends to
create a very large population of individual particles.
However, the volume fraction of individual particles in Figure 8. Effect of the heptane content on the asphaltene floc
size for 0.1 kg/m3 asphaltenes at mixing speeds of 110 and 230
Figure 6 is smaller than the volume fraction of floc- rpm.
culated particles. This distribution indicates that the
shattering mechanism is dominant in the asphaltene
disintegration process.20
Effects of the Concentration, Heptane Content,
and Shear Rate. The asphaltene concentration, hep-
tane content, and shear rate are expected to have a
significant effect on asphaltene flocculation. Figure 7
shows the effect of the asphaltene concentration on the
volume mean diameter in 60:40 Heptol. The volume
mean diameter increases as the asphaltene concentra-
tion increases. An increase in the asphaltene concentra-
tion is equivalent to an increase in the number of
individual particles. As the number of particles in-
creases, the flocculation rate also increases.
Figure 8 shows the effect of the n-heptane content on
asphaltene flocculation. Note that the volume mean
diameter of the asphaltene flocs reached the upper
detection limit of the analyzer in 70:30 Heptol at a
mixing rate of 110 rpm. Therefore, the measurement Figure 9. Effect of the mixing speed on the asphaltene floc size
for 0.1 kg/m 3 in 60:40 Heptol.
was terminated after 30 min. Figure 8 shows that the
flocculation rate increases dramatically as the heptane
content increases. The higher the n-heptane content Figures 9 and 10 show the effect of the mixing rate
precipitates, the higher the proportion of the asphalt- (shear rate) on asphaltene flocculation in solutions of
enes resulting in more particles. Also, relatively stronger 60:40 and 70:30 Heptol, respectively. The flocculation
interparticle interactions are expected because heptane rate decreases as the shear rate increases. In fact, the
is a poor solvent for asphaltenes. Both factors increase results indicate that flocculation can be reversed or even
the flocculation rate. stopped at higher shear rates. It appears that increasing
Ind. Eng. Chem. Res., Vol. 43, No. 21, 2004 6867

Figure 10. Effect of the mixing speed on the asphaltene floc size
for 0.1 kg/m 3 in 70:30 Heptol. Figure 11. Calculated fractional precipitation of 0.1 kg/m3
asphaltenes in 60:40 Heptol.

the shear rate increases the relative rate of disintegra-


tion versus flocculation. with a diameter of 1 µm. To illustrate the approach, the
Asphaltene Fractal Dimension. The fractal dimen- results for a concentration of 0.1 kg/m3 are discussed
sion of asphaltene flocs was determined indirectly from here. Figure 11 shows the calculated fractional precipi-
a calculation of the mass of precipitated asphaltenes tation, assuming fractal dimensions of 3.0, 2.5, 2.0, 1.75,
from the measured particle-size distribution. The frac- and 1.50.
tional precipitation of asphaltenes on a mass basis, FA, It was not possible to compare the calculated and
is given by measured fractional precipitation directly because the
particle-size distributions could not be measured at
FA ) VT FA/mA (9) concentrations above 0.1 kg/m3 while accurate gravi-
metric measurements could not be obtained at concen-
trations below 0.5 kg/m3. Also the total number count
where FA is the asphaltene density (1.12 kg/m3), mA is of the particle-size analyzer is not necessarily accurate,
the total mass of asphaltenes in the solution, and VT is although the relative number count is accurate. In other
the total volume of the flocculated asphaltenes. The total words, the absolute number counts may be inaccurate,
volume of a distribution of flocs is given by but the trends in the number count are expected to be
accurate. Hence, a qualitative comparison is still pos-
∑niVia
π sible.
VT ) (10)
6 The gravimetric measurements given in Figure 1
show that the fractional precipitation does not change
where nf is the number of particles in the floc and Via is significantly after 30 min. Figure 11 indicates that the
the actual volume of all of the particles in a floc. The fractal dimension must be below 2.0 to match this trend
actual volume, Via, of a floc is given by and indeed to avoid unrealistic calculations of the
fractional precipitation. It is not possible to determine
π di
() the fractal dimension more accurately, although the
Df
π
Via ) nd 3) dp3 (11) trends best match at the lower fractal dimensions such
6 i p 6 dp as 1.5.
Sheu12 measured the fractal dimension of asphaltenes
where di is the measured diameter, dp is the diameter in mixtures of 70:30 and 80:20 Heptol and found that
of the individual particle, and Df is the fractal dimen- it ranged from 1.49 to 1.8. Also, Raghunathan13 mea-
sion. Note that eq 5 was used to relate the number of sured the fractal dimension for three gel-permeation-
particles in a floc to the measured diameter, giving the chromatographic fractions of asphaltenes and obtained
final expression in eq 11. Equations 10 and 11 are values from 1.65 to 1.68. A visual inspection of the
substituted into eq 9 to obtain micrographs of asphaltene flocs (Figure 2a,b) is consis-
tent with a fractal dimension of less than 2 and closer

()
πFANf dp3 di Df to 1.5. Based on the calculated fractional precipitation
FA )
6mA ∑ fi
dp
(12) results, the microscopic examinations, and the values
reported in the literature, the fractal dimension was
assumed to be 1.6.
where Nf is the total number of flocs and fi is the number As a check, the asphaltene fractional precipitation
frequency of the ith floc. was calculated for 60:40 Heptol after 6 h of contact
The fractional precipitation of asphaltenes was cal- based on a fractal dimension of 1.6. The calculated
culated from the number frequency distributions and precipitation values are compared with direct gravimet-
the total number count over time for three systems: ric measurements in Figure 12. The calculated frac-
asphaltene concentrations of 0.05, 0.08, and 0.10 kg/m3 tional precipitations fall on the same trend as measured
all in 60:40 Heptol at a mixing rate of 110 rpm. The data, indicating that using a fractal dimension of 1.6
individual particles were assumed to be monodisperse gives plausible results.
6868 Ind. Eng. Chem. Res., Vol. 43, No. 21, 2004

Figure 12. Comparison of measured and calculated fractional


asphaltene precipitation in 60:40 Heptol after 360 min. The dotted
line is a curve fit provided as a visual aid.

Table 1. Concentrations of Individual Particles Input to


Flocculation Model
asphaltene fractional Ctot(actual) Ctot(model)
system concn (kg/m3) precipn (kg/m3) (kg/m3)
60:40 Heptol 0.05 0.036 0.0018 218
60:40 Heptol 0.08 0.041 0.0033 400
60:40 Heptol 0.10 0.081 0.0081 982

Note that the asphaltene flocs may not be fractals and


the fractal dimension may not remain constant at all
scales. In other words, the fractal dimension may
change as the size of the flocs changes with time,
asphaltene concentration, heptane content, and shear
rate. In this work, it was assumed that the fractal
dimension of the asphaltene flocs was constant. Figure 13. Flow diagram of the numerical routine for calculating
asphaltene flocculation.
Model Results. The inputs to the model are the
initial concentration of the individual particles, the
the fitted reaction terms should be scaled to the
diameter of the individual particles, the fractal dimen- experimental conditions as follows:
sion of the flocs, the reaction constants, and the reaction
exponents. The reaction constants and exponents are
Kf(corrected) ) Kf(model)[Ctot(model)/Ctot(actual)]2
employed as tuning parameters, and the fractal dimen-
sion of the asphaltene flocs is taken to be constant at
1.6 in the 60:40 Heptol system. The diameter of the
Kd(corrected) )
individual particles was found to range from 0.5 to 2 µm. Kd(model)[Ctot(model)/Ctot(actual)] (13)
For modeling purposes, the individual particles were
assumed to be monodisperse with a diameter of 1 µm. The computer flow diagram to calculate the asphaltene
The concentration of the individual particles was particle-size distribution is provided in Figure 13.
determined from the mass of precipitated asphaltenes Equations 3, 7, and 8 were solved explicitly with a time
at the given experimental conditions. Table 1 shows the increment of 0.5, and the kinetics of asphaltene particle
calculated concentrations of the individual particles for growth were calculated over 360 min. The time step was
concentrations of 0.05, 0.08, and 0.10 kg/m3 in 60:40 controlled so that the total number of individual par-
ticles remained constant. The material balance errors
Heptol. Because the fractal measurement and gravi-
were less than 3% in all cases. Note that the numerical
metric tests were performed only for the 60:40 Heptol
error from too large a time step always results in an
system, the 70:30 Heptol system was not modeled in error in the total number of particles. The output of the
this study. The fractional precipitation was assumed to program is the number of flocs with a given number of
be constant at its steady-state value (after 30 min). particles within each floc. When converted to a fre-
Hence, the low initial fractional precipitation and the quency distribution, the diameter increment is not
increase to steady-state values over the first 30 min uniform and does not match the increment of the
were not included in the model. To minimize the measured distributions. Therefore, the cumulative fre-
numerical error when the model is run, the total quency distribution was curve-fitted and subdivided into
concentration of the particles was scaled up, as indicated 0.5-µm increment to match the experimentally mea-
in Table 1. Hence, the fitted reaction constants do not sured distribution.
reflect the actual concentration of the particles in the The model was tuned with the inputs given in Table
system. Because the value of dCk/dt in eq 3 predicted 2. The modeled kinetics of particle growth for asphalt-
by the model at the scaled concentrations was matched enes in 60:40 Heptol at a mixing rate of 110 rpm are
to the experimental results at the actual concentrations, compared with the experimental data in Figure 7. A
Ind. Eng. Chem. Res., Vol. 43, No. 21, 2004 6869

Table 2. Model Inputs for 0.08 kg/m3 Asphaltenes in a


60:40 Heptol System
input modeled corrected
reaction-limited reaction constant, Kf 5.2 × 10-5 7.6 × 105
shattering reaction constant, Kd 5.0 × 10-5 6.1
flocculation reaction exponent, λ 2.6
shattering reaction exponent, β 1.5
size of the individual particles (µm) 1.0
fractal dimension of the 1.6
asphaltene flocs, Df

comparison of the measured and modeled particle-size


distributions is given in Figures 14-16 for asphaltene
concentrations of 0.05, 0.10, and 0.10 kg/m3, respec-
tively. The model fits the major trends in the experi-
mental data quite well with the following exceptions:
The model did not predict the reduction in the floc
size over the first 30 min. This discrepancy likely occurs
Figure 14. Measured and fitted asphaltene particle volume
because the kinetics of asphaltene precipitation were distribution after 60 and 360 min for 0.05 kg/m3 asphaltenes in
not included in the model and the initial total concen- 60:40 Heptol at a mixing rate of 110 rpm.
tration of the individual particles was considered to be
constant at the steady-state value. It is also possible
that large flocs formed in the first 2 min, before the
solutions were placed in the cuvette and the mixer
turned on. Some time may be required to eliminate the
large flocs at the mixer shear conditions.
The model underpredicts the volume mean diameter
at higher concentration. One possibility is that the
empirical expression for the flocculation reaction is not
correct. Another possibility is that the fractal dimen-
sions of asphaltene flocs are not true fractals; that is,
the number of particles per volume may not be the same
at all length scales. For example, a larger floc may have
a less compact structure than a small floc and therefore
have a lower fractal dimension. Hence, the apparent
fractal dimension may change with time, asphaltene
concentration, heptane content, and shear rate. In this
case, a correlation of the fractal dimension to the floc
size would be required. Figure 15. Measured and fitted asphaltene particle number and
Note that the model predicted the persistent popula- volume distribution after 60 and 360 min for 0.08 kg/m3 asphalt-
tion of individual particles that was observed in the enes in 60:40 Heptol at a mixing rate of 110 rpm.
experiments. The successful prediction of the persistent
individual particle-size distribution in the model simu-
lation indicates that the shattering mechanism is suf-
ficient to model this behavior and that the introduction
of surface erosion is not required.

Conclusions
Precipitated asphaltene particles in solutions of hep-
tane and toluene are approximately 0.5-2 µm in
diameter and tend to flocculate into loose fractal-like
flocs rather than dense, spherical aggregates. The
flocculated asphaltenes in solvent solutions are quali-
tatively similar to those in heptane-diluted bitumen.
The fractal dimension of flocculated asphaltenes in
solutions of heptane and toluene is approximately 1.6.
The asphaltene floc size increased with increased as-
phaltene concentration, increased heptane content, and Figure 16. Measured and fitted asphaltene particle volume
decreased mixing rate. distribution after 60 and 360 min for 0.1 kg/m3 asphaltenes in 60:
Asphaltene flocculation is reversible, indicating that 40 Heptol at a mixing rate of 110 rpm.
flocculation is opposed by some disintegration mecha-
nism. The shape of the sheared particle-size distribu- was observed in all cases. This behavior would indicate
tions indicates that shattering is the dominant disin- that reaction-limited flocculation is the dominant mech-
tegration mechanism. It was found that the flocculation anism in asphaltene flocculation.
rate increased as the volume mean diameter of asphalt- A reversible kinetic model was proposed based on
ene flocs increased. Also, a persistent population of reaction-limited flocculation and shattering mecha-
individual particles in the diameter range of 0.5-2.0 µm nisms. The model predictions correlate well the experi-
6870 Ind. Eng. Chem. Res., Vol. 43, No. 21, 2004

mental data with percent absolute average relative (9) Ferworn, K. A.; Svrcek, W. Y.; Mehrotra, A. K. Measurement
deviation of 17.8%, 14.8%, and 29.3% for concentrations of Asphaltene Particle Size Distribution in Crude Oils Diluted with
of 0.05, 0.08, and 0.1 kg/m3, respectively. The model is n-Heptane. Ind. Eng. Chem. Res. 1993, 32, 955.
(10) Ferworn, K. A.; Mehrotra, A. K.; Svrcek, W. Y. Measure-
suitable for relatively dilute systems with limited floc- ment of Asphaltene Agglomeration from Cold Lake Bitumen
culation. More data are required to extend the model Diluted with n-Alkanes. Can. J. Chem. Eng. 1993, 71, 699.
to higher asphaltene concentrations where larger flocs (11) Neilson, B. B.; Svrcek, W. Y.; Mehrotra, A. K. Effects of
are created and different flocculation and disintegration Temperature and Pressure on Asphaltene Particle Size Distribu-
mechanisms could dominate. tions in Crude Oils Diluted with n-Pentane. Ind. Eng. Chem. Res.
1994, 33, 1324.
(12) Sheu, E. Y. In Structures and Dynamics of Asphaltenes;
Acknowledgment Mullins, O. C., Sheu, E. Y., Eds.; Plenum Publishing Corp.: New
The authors thank NSERC and DBR Product Center- York, 1998.
Schlumberger for financial support and Syncrude Canada (13) Raghunathan, P. Evidence for Fractal Dimension in As-
phaltene Polymers from Electron-Spin-Relaxation Measurement.
Ltd. for supplying bitumen samples. Chem. Phys. Lett. 1991, 182 (3 and 4), 331.
(14) Mason, T. G.; Lin, M. Y. Time-Resolved Small Angle
Literature Cited Neutron Scattering Measurements of Asphaltene Nanoparticle
Aggregation in Incompatible Crude Oil Mixtures. J. Chem. Phys.
(1) Sheu, E. Y.; Mullins, O. C. Asphaltenes, Fundamental and 2003, 119 (1), 565.
Application; Plenum Press: New York, 1995; p 191 (15) Alboudwarej, H. Asphaltene Deposition in Flowing Sys-
(2) Minnsieux, L. Removal of Asphalt Deposits by Covalent tems. Ph.D. Dissertation, University of Calgary, Calgary, Alberta,
Squeeze: Mechanisms and Screening. SPE J. 2001, 6 (1), 289. Canada, 2003.
(3) Anisimov, M. A.; Yudin, I. K.; Nikitin, V.; Nikolaenko, G.;
(16) Speight, J. G. The Chemistry and Technology of Petroleum;
Chernoustan, A. Asphaltene Aggregation in Hydrocarbon Solutions
Marcel Dekker Inc.: New York, 1999.
Studied by Photon Correlation Spectroscopy. J. Phys. Chem. 1995,
99, 9576. (17) Bates, R. L.; Fondy, P. L.; Fenic, J. G. In Mixing: Theory
(4) Yudin, I. K.; Nikolaenko, G. L.; Gorodetskii, E. E.; Kosov, and Practice; Uhl, V. W., Gray, J. B., Eds.; Academic Press: New
V. I.; Melikyan, V. R.; Markhashov, E. L.; Frot, D.; Briolant, Y. York, 1966; Vol. I.
Mechanisms of Asphaltene Aggregation in Toluene/Heptane Mix- (18) Family, F.; Meakin, P.; Deutch, J. M. Kinetics of Coagula-
tures. J. Pet. Sci. Eng. 1998, 20, 297. tion with Fragmentation: Scaling Behavior and Fluctuations.
(5) Neves, G. B. M.; Sousa, M. D.; Travalloni-Louvisse, A. M.; Phys. Rev. Lett. 1986, 57 (6), 727.
Lucas, E. F.; Gonzalez, G. Characterization of Asphaltene Particles (19) Ball, R. C.; Weitz, D. A.; Witten, T. A.; Leyvraz, F.
by Light Scattering and Electrophoresis. Pet. Sci. Technol. 2001, Universal Kinetics in Reaction-limited Aggregation. Phys. Rev.
19 (1 & 2), 35. Lett. 1987, 58 (3), 274.
(6) Burya, Y. G.; Yudin, I. K.; Dechabo, V. A.; Kosov, V. I.; (20) Wei, J.; Lee, W.; Krambeck, F. Catalyst Attrition and
Anisimov, M. A. Light-scattering Study of Petroleum Asphaltene Deactivation in Fluid Catalytic Cracking System. Chem. Eng. Sci.
Aggregation. Appl. Opt. 2001, 40 (24), 4028. 1977, 32, 1211.
(7) Funk, E. W. Behavior of Tar Sand Bitumen with Paraffinic
Solvents and its Application to Separations for Athabasca Tar
Sands. Can. J. Chem. Eng. 1979, 57, 333. Received for review May 13, 2004
(8) Espinat, D.; Rosengerg, E.; Scarsella, M.; Barre, L.; Fenistein, Revised manuscript received July 31, 2004
D.; Broseta, D. In Structure and Dynamics of Asphaltenes; Mullins, Accepted August 6, 2004
O. C., Sheu, E. Y., Eds.; Plenum Publishing Corp.: New York,
1998. IE049594V

S-ar putea să vă placă și