Documente Academic
Documente Profesional
Documente Cultură
1021/ef9014263
Published on Web 04/05/2010
peptized on their surface. Resins are considered to be the 2. Modeling and Theory
compounds chemically intermediate between all of the other In this work, asphaltene precipitation in live oils is modeled as
species in the oil and asphaltenes. Asphaltene precipitation is the traditional liquid-liquid or gas-liquid-liquid phase separa-
due to the resin desorption from the surface of asphaltene tion. The live oils are complex fluid mixtures of thousands of
particles. components. We characterize them by considering the pure
These two types of models have been used in correlating and components (N2, CO2, H2S, C1, C2, C3, iC4, nC4, iC5, and nC5),
predicting the onsets and amount of asphaltene precipitation the pseudo-hydrocarbon components (C6, C7, C8, etc.), and the
in various petroleum fluids, with some limitations that may hydrocarbon residue (Cnþ). The pseudo-hydrocarbon compo-
restrict their applications in the petroleum industry. First, in nents are defined by lumping a number of hydrocarbon compo-
some of the theories, the reservoir fluid is represented by either nents within a certain normal boiling point range.38 For instance,
asphaltene þ resin or asphaltene þ solvent. The two-compo- C6 includes all of the components with the normal boiling point
between those of nC5 and nC6. Full characterization can avoid the
nent model is an oversimplification, and it may not be of value
extra parameter adjustment because the required EOS para-
for use in compositional reservoir simulators. Second, most of meters are either standard for the pure components or correlated
models cannot describe the gas, oil, and asphaltene phases for the pseudo-hydrocarbon components. Thus, our model
within a unified framework; i.e., they cannot work indepen- requires not only the asphaltene precipitation information but
dently. Third, some of the methods contain many complex also the complete fluid analysis. The hydrocarbon residue is
equations and/or adjustable parameters, which may compli- further divided into the “heavy” component and asphaltene.
cate the implementation in the simulation modeling. Finally, The “heavy” component contains the heavy alkanes, heavy
some approaches do not account for the polar-polar and aromatics, and all resins. Here, the heavy alkanes and heavy
polar-induced polar interactions pertinent to asphaltenes and aromatics are those not included in pure components and pseudo-
other components (e.g., resins and aromatics), which could hydrocarbon components. The “heavy” component represents
also be important for the proper description of the asphaltene the secondary most polar component in petroleum fluids. The
higher the molecular weight, the higher the aromatic content
phase behavior.
and the higher the polarity. Although we coarse-grain the
In a recent paper, we have proposed a cubic-plus-asso- heavy alkanes, heavy aromatics, and resins as only one pseudo-
ciation equation of state (CPA-EOS) to study asphaltene component, the necessary physical features are essentially pre-
precipitation in model and real heavy oils from the addition sented and well-described in our work. The advantages of our
of n-alkanes.37 We have examined the effects of tempera- method have been demonstrated in detail in our recent work.37
ture, pressure, and n-alkanes on asphaltene precipitation Specifically, for the asphaltene precipitation in live oils, the
and demonstrated that our model can reproduce the experi- additional advantage of our model is that it is based on the
ments. This is the first time that CPA-EOS is applied to existing fluid characterization, which can be readily implemented
study the asphaltene problem. Our approach is, in princi- in the compositional reservoir simulators.
ple, a lyophilic model; i.e., the asphaltene-containing fluid In the framework of CPA-EOS, the excess Helmholtz free
is regarded as a true solution. Differently, besides the energy Aex consists of the physical part, which describes the non-
associating molecular interactions, such as short-range repulsions
molecular size and dispersion attractions, the polar-polar
and dispersion attractions, and the association part, which
interactions relevant to asphaltenes and other components describes the polar-polar interactions (self-association and
are taken into consideration. However, different from the cross-association) of asphaltene and “heavy” molecules.
lyophobic model, we do not specifically require the species The physical contribution is represented by the Peng-Robinson
that are polar (e.g., resins) and easy to be polarized (e.g., (PR)-EOS39
aromatics) to stabilize asphaltenes in petroleum fluids. pffiffiffi !
They may either inhibit or promote asphaltene precipita- Aex
ph a 1 þ ð1 þ 2ÞbF
¼ - lnð1 - bFÞ - pffiffiffi ln pffiffiffi ð1Þ
tion depending upon the composition of the petroleum nRT 2 2bRT 1 þ ð1 - 2ÞbF
fluid. Our approach includes the possible intermolecular
interactions relevant to asphaltene problems and can par- where R is the universal gas constant, T is the absolute tempera-
tially and in some cases completely overcome the current ture, n is the total number of moles, and F is the molar density of
limitations. the mixture. a and b are the energy and volume parameters of the
In this work, we apply our CPA-EOS to investigate asphal- mixture. They can Pbe estimatedP by applying the van der Waals
tene precipitation in several live oils from a pressure decrease mixing rules: a = i,jxixjaij, b = ixibi, and aij = (aiaj)1/2(1 - kij),
where xi, ai, and bi stand for the mole fraction, energy para-
and mixing with CO2 at high temperature and pressure.
meter, and volume parameter of component i, respectively,
Within a unified theoretical framework, we successfully and kij is the binary interaction coefficient (BIC) between
capture the bubble point pressure, asphaltene precipitation components i and j (kij = 0 if i = j). ai and bi can be determined
amount and onset pressures, and gas-oil-asphaltene three- from the critical properties and acentric factor of an individual
phase coexistence. More importantly, our model can be read- component
ily implemented in compositional reservoir simulators. The R2 Tci2 pffiffiffiffiffiffi 2 RTci
remainder of this paper is organized as follows: Section 2 ai ¼ 0:45724 1þci ð1 - Tri Þ , bi ¼ 0:0778 ð2Þ
Pci Pci
describes the modeling and the formulation of our CPA-EOS
for asphaltene systems in live oils. Section 3 compares calcula- with ci ¼ 0:37464 þ 1:54226ωi - 0:26992ωi 2 , ωi < 0:5 ,
0:3796 þ 1:485ωi - 0:1644ωi 2 þ 0:01667ωi 3 , ωi > 0:5
tions to experiments for asphaltene precipitation in seven live
where Tri, Tci, Pci, and ωi denote the reduced temperature, critical
oils induced by a pressure decrease and CO2 mixing. In section temperature, critical pressure, and acentric factor of component i,
4, the main results and conclusions are summarized. respectively.
The contribution to the excess Helmholtz energy because of cross-associations between asphaltene and pseudo-hydrocarbon
association is derived from the thermodynamic perturbation components are assumed negligible, but they can be included
theory.40-43 It is the same as that used in the SAFT.44-47 We within the same theoretical framework with the introduction of
assume that each asphaltene molecule has NA identical associa- more adjustable parameters.
tion sites and each “heavy” molecule has NR identical association
sites 3. Results and Discussion
Aex 1 - χA 1 - χR
ass
¼ NA xA ln χA þ þ NR xR ln χR þ In this work, we study the asphaltene precipitation in seven
nRT 2 2
live oils. Table 1 provides the concentration of the pure
ð3Þ components and pseudo-hydrocarbon components for these
live oils. It also gives the mole fraction, molecular weight
The subscripts “A” and “R” represent asphaltenes and the
“heavy” component, respectively. χA and χR are the mole frac- (MW), and specific gravity (SG) of the hydrocarbon residue.
tions of asphaltene and “heavy” molecules not bonded at one of The asphaltene weight fractions in stock tank oils (STOs) are
the association sites, respectively. Association bonding occurs also included. The physical parameters TC, PC, ω, and MW of
between two sites, with one on asphaltene molecule and the other the pure components, pseudo-hydrocarbon components, and
on either asphaltene or a “heavy” molecule. “Heavy” molecules asphaltene are either directly available or obtained in our
are assumed not to associate with themselves. As a result, χA and recent work.37,48 They are repeated in Table 2 for the sake of
χR are given by completeness. For the reservoir fluids A,49,50 C1,36 Y3,36 X1,49
1 X2,22 and X3,23 precipitation is from the pressure decrease.
χA ¼ ,
1 þ FNA xA χA Δ þ FNR xR χR ΔAR
AA For the Weyburn oil, precipitation is induced by CO2 mix-
ing.51 The common association parameters NA = NR = 4,
1
χR ¼ ð4Þ κAA = κAR = 0.01, and εAA/kB = 2000 K are also provided in
1 þ FNA xA χA ΔAR our recent work for asphaltene precipitation in heavy oils.37
where Δij = gκijbij[exp(εij/kBT) - 1] (i = A and j = A or R) The physical parameters of the “heavy” component are
characterizes the “association strength” with bij = (bi þ bj)/2, kB obtained in two steps. First, for the hydrocarbon residue, on
is the Boltzmann constant, g is the contact value of the radial the basis of its molecular weight and specific gravity, Cavett’s
distribution function of the hard-sphere mixture, and κij and εij correlation is used to provide the rough estimation of TC
are the association volume and energy parameters, respectively. and PC52 and the normal boiling point TB is estimated using
To simplify the calculations, g is approximated by that of the pure the interpolation of the generalized properties of petroleum
hard-sphere fluid as g ≈ (1 - 0.5η)/(1 - η)3, with η = bF/4. The hexane-plus groups.38 Because the uncertainty in TC estima-
tion is less than that in PC,38 we only adjust PC of the
(40) Wertheim, M. S. J. Stat. Phys. 1984, 35, 19. hydrocarbon residue to match the bubble point pressure
(41) Wertheim, M. S. J. Stat. Phys. 1984, 35, 35.
(42) Wertheim, M. S. J. Stat. Phys. 1986, 42, 459.
(43) Wertheim, M. S. J. Stat. Phys. 1986, 42, 477. (48) Database of RERI Flash Computation Software Package.
(44) Chapman, W. G.; Jackson, G.; Gubbins, K. E. Mol. Phys. 1988, (49) Fahim, M. A. Pet. Sci. Technol. 2007, 25, 949.
65, 1057. (50) Jamaluddin, A. K. M.; Joshi, N.; Iwere, F. Society of Petroleum
(45) Chapman, W. G.; Gubbins, K. E.; Jackson, G.; Radosz, M. Fluid Engineers (SPE) International Petroleum Conference and Exhibition,
Phase Equilib. 1989, 52, 31. Villahermosa, Mexico, Feb 10-12, 2002; SPE 74393.
(46) Chapman, W. G.; Gubbins, K. E.; Jackson, G.; Radosz, M. Ind. (51) Srivastava, R. K.; Huang, S. S.; Dyer, S. B.; Mourits, F. M.
Eng. Chem. Res. 1990, 29, 1709. J. Can. Pet. Technol. 1995, 34, 31.
(47) Jackson, G.; Chapman, W. G.; Gubbins, K. E. Mol. Phys. 1988, (52) Cavett, R. H. Proceedings of the American Petroleum Institute
65, 1. (API) 27th Mid-year Meeting, San Francisco, CA, 1962; Vol. 52, p 351.
2958
Energy Fuels 2010, 24, 2956–2963 : DOI:10.1021/ef9014263 Li and Firoozabadi
Table 2. Common Physical Parameters of Pure Components, Pseudo- Table 4. Non-zero Binary Interaction Coefficients for the Seven Live
hydrocarbon Components, and Asphaltene37,48 Oils38,53
TC (K) PC (bar) ω MW C1-HC N2-HC CO2-HC H2S-HC
N2 126.21 33.90 0.039 28.0 fluid A
CO2 304.14 73.75 0.239 44.0 fluid C1
H2S 373.20 89.40 0.081 34.1 fluid Y3
C1 190.56 45.99 0.011 16.0 fluid X1 0.0289 þ 1.633 0.15
0.1 0.1
C2 305.32 48.72 0.099 30.1 fluid X2 10-4MW
C3 369.83 42.48 0.153 44.1 fluid X3
iC4 407.80 36.04 0.183 58.1 Weyburn oil 0.071
nC4 425.12 37.96 0.199 58.1
iC5 460.40 33.80 0.227 72.2 Table 5. Cross-association Energy between Asphaltene and “Heavy”
nC5 469.70 33.70 0.251 72.2 Molecules εAR for the Seven Live Oils
C6 507.40 30.12 0.296 86.2
fluid A εAR/kB (K) = 665.82 þ 1.0145T
C7 556.48 26.75 0.294 100.0
fluid C1 εAR/kB (K) = 1258.85 - 1.2462T
C8 574.76 25.24 0.418 114.0
fluid Y3 εAR/kB (K) = 422.54 þ 2.0125T
C9 593.07 23.30 0.491 128.0
fluid X1 εAR/kB (K) = 1311.49 - 0.8835T
C10 617.07 21.55 0.534 142.0
fluid X2 εAR/kB (K) = 1219 (303 K)
C11 638.24 19.74 0.566 156.0
fluid X3 εAR/kB (K) = 1500 (303 K), 1420 (353 K), 1400 (403 K)
asphaltene 1474 6.34 2 1800
Weyburn oil εAR/kB (K) = 500 (332 K)
well with the experiments for both bubble point and upper
onset pressures.
Our work predicts that the plot of upper onset pressure
tends to be parallel to or away from that of the saturation
pressure at high temperatures. This is in agreement with the
theories in the literature.23-26,28 However, some of the the-
ories predict that these two plots may eventually meet.18,56
The difference is possibly due to different reservoir oils. After
a careful literature search, we do not notice any direct experi-
mental evidence to support either of the predictions.
For the reservoir fluids X2 and X3, we study the effect of the
pressure on the amount of asphaltene precipitation.22,23 The
asphaltene content is determined using nC7 titration. The
comparison between measurements and calculations is dis-
played in Figures 5 and 6. The bubble point and upper and
Figure 5. (a) Fraction of precipitated asphaltenes and (b) relative lower onset pressures can be readily seen in the plots. As the
volume of the mixture (bubble point is the reference) as function of
the pressure for the reservoir fluid X2 at 303 K. Symbols are
pressure drops, asphaltene precipitation first increases when
experiments from ref 22, and lines represents calculations. In panel the pressure is between the upper onset and bubble point and
a, the calculated bubble point and upper and lower onsets are then decreases when the pressure is between the bubble point
marked by thin lines. and lower onset. The maximum precipitation appears at the
bubble point. This behavior can be interpreted from the
Our calculation indicates that, in the oil-asphaltene two-phase solubility parameter, which will be presented later. Our model
region when the temperature is above 400 K, the asphaltene correctly captures the effect of the pressure on the asphaltene
precipitation is strengthened as the temperature increases.
This phenomenon also weakly appears for the reservoir fluids
(56) Lopez-Chavez, E.; Pacheco-Sanchez, J. H.; Martinez-Magadan,
A and C1 possibly because of the counter effect of the J. M.; Castillo-Alvarado, F. D.; Soto-Figueroa, C.; Garcia-Cruz, I. Pet.
temperature. For the fluids C1 and Y3, our calculations agree Sci. Technol. 2007, 25, 19.
2961
Energy Fuels 2010, 24, 2956–2963 : DOI:10.1021/ef9014263 Li and Firoozabadi
precipitation amount particularly in the oil-asphaltene two- where vs is the molar volume of the solvent. PR-EOS can
phase region. However, for the reservoir fluid X3, our method provide accurate predictions for the solubility parameters (we
predicts that the asphaltene precipitation disappears too fast have tested the solubility parameters of several pure sub-
when the pressure is lower than the bubble point; i.e., the lower stances, including nC5, nC6, nC8, benzene, and toluene, at
onset pressure is overestimated, which is similar to Figure 2. 298 K). From Figure 8, as expected, the solvent solubility
This is possibly one limitation of our model. Besides, we also parameter decreases when the pressure drops from the upper
present the comparison between calculations and experiments onset to the bubble point because of the decrease of the oil
for the relative volumes (the ratio between the total volume phase density and then increases when the pressure drops
and the bubble point volume) during depressurization. There further from the bubble point to the lower onset because of
is a slope change at the bubble point because the vapor phase evaporation of light components (which causes the increase of
vanishes when the pressure is higher than the bubble point. the oil phase density). The minimum of both the solvent
Again, the agreement is excellent. solubility parameter and oil phase density is at the bubble
For the Weyburn oil, we investigate the asphaltene pre- point where the asphaltene precipitation reaches the maximum.
cipitation from mixing with CO2.51 Saturates, aromatics, As a representative, in Figure 9, we show the calculated
resins, and asphaltenes (SARA) analysis of STO was con- composition of both asphaltene and oil phases for the reser-
ducted using a modified Syncrude method, but the precipitant voir fluid X2 at 303 K and 98.1 bar (bubble point). While some
is not provided. The asphaltene content of the flashed CO2- of the existing models assume that the precipitated phase is
saturated oils is determined by a photometric technique. A pure asphaltenes and/or the oil phase is pure oil (with no
change of 0.1 wt % in asphaltene content in the oil can be asphaltenes), our method describes the heterogeneity of both
estimated with confidence. Weyburn oil contains a very small asphaltene-rich and asphaltene-lean phases. Even though
amount of light components. CO2 injection can potentially the asphaltene precipitation reaches the maximum at the
improve the oil recovery but may lead to a large amount of bubble point, the precipitated phase still contains a non-
asphaltene precipitation. In Figure 7a, the bubble point negligible amount of other species (about 12 wt %) and, at
pressure at 332 K is presented for different CO2/oil mixtures. the same time, there is also a small amount of asphaltenes
At a very high CO2 concentration, asphaltene precipitation remaining in the oil phase (about 1.2 wt %). It can be expected
occurs at the bubble point. In Figure 7b, the fraction of that, at other precipitation conditions, both the asphaltene
precipitated asphaltenes at 332 K and 160 bar is shown as a and oil phases should be even more heterogeneous.
2962
Energy Fuels 2010, 24, 2956–2963 : DOI:10.1021/ef9014263 Li and Firoozabadi
2963