A Method For Characterization of Bitumen
A Method For Characterization of Bitumen
A Method For Characterization of Bitumen
Fuel
journal homepage: www.elsevier.com/locate/fuel
h i g h l i g h t s
a r t i c l e i n f o a b s t r a c t
Article history: Characterization of bitumen is a necessary step to perform phase equilibrium computations involved in
Received 5 January 2015 bitumen production and processing. This study presents a methodology for bitumen characterization
Received in revised form 28 February 2015 using a residue curve map. A fugacity–activity coefficient approach is applied to model the thermody-
Accepted 2 March 2015
namic equilibrium of the species in the gas and liquid phases. The Peng–Robinson equation of state
Available online 13 March 2015
(PR-EoS) and NRTL activity model are utilized to calculate the fugacity and activity coefficients in the
gas and liquid phases, respectively. The proposed model was evaluated using experimental simulated dis-
Keywords:
tillation (SD) data as well as experimental solubility data of light hydrocarbon (CH4, C2H6) and non-hy-
Heavy oil
Bitumen characterization
drocarbon gases (CO2, N2) in bitumen. The molecular weight, specific gravity, and SD curves were
Simulated distillation represented using defined pseudo-components. The tuned EoS was able to regenerate the solubility data
Solubility with an acceptable accuracy (AARD less than 5.6% and 4.5% for hydrocarbon and non-hydrocarbon sol-
NRTL model vents, respectively). The proposed method can be applied for bitumen characterization and to predict
the solubility of the light gases in bitumen.
Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction Bitumen and extra heavy oil assays should be obtained under
deep vacuum distillation to avoid possible thermal cracking at
Increase of worldwide energy demand besides the decline of temperatures higher than 523 K. Standard tests such as D5236
conventional oil reserves resulted more attention toward bitumen ASTM test [3] and based on gas chromatography (GC) have been
and heavy oil recourses. Currently, thermal oil recovery is the lead- developed to characterize high boiling point hydrocarbon mix-
ing processes for bitumen and heavy oil extraction in Alberta. tures. Simulated distillation is fast, repeatable, and inexpensive.
Disadvantages of thermal recovery methods such as high energy Additionally, a small amount of bitumen/heavy oil sample (less
and water consumption, water pollution and greenhouse gas emis- than 1 ml) is required for the analysis [1].
sion have raised the interests to find alternative methods such as Saajanlehto and Alopaeus [4] and Saajanlehto et al. [5] have
solvent injection. used distillation curve data to develop a characterization method
Characterization of bitumen is the initial step for all thermody- for Perturbed-Chain Statistical Associating Fluid Theory (PC-
namic and phase behavior calculations. Crude oil assay obtained SAFT). Their model has been verified to represent density, sat-
from distillation experiments is the common tool to characterize uration pressures for propane-Athabasca bitumen and CO2–pro-
hydrocarbon mixtures and petroleum fractions. The underlying pane–Athabasca bitumen systems and solubility of hydrogen in
idea is that hydrocarbon molecules, which boil at a range of tem- heavy oil. They proposed a single carbon number type character-
perature, are lumped into a pseudo-component and assigned an ization method for heavy oils and bitumen. In their model, each
average boiling temperature [1,2]. cut of crude oil was split into saturate, aromatic and polyaromatic
(SAP) fractions based on its boiling point. Therefore, distribution of
SAP in crude oils is assumed universal.
⇑ Corresponding author. Doherty and Perkins [6–8] developed the residue curve map to
E-mail address: [email protected] (J. Abedi). model the simple distillation assay. This model calculates the loci
http://dx.doi.org/10.1016/j.fuel.2015.03.005
0016-2361/Ó 2015 Elsevier Ltd. All rights reserved.
B. Azinfar et al. / Fuel 153 (2015) 240–248 241
of saturation temperatures of the mixture versus the composition introduction of additional binary interaction energy parameters.
of the liquid phase in the flask during the distillation. To apply this While some methods have been presented in literature to find
model there is no need to know the exact condition and procedure these parameters, their application to complex mixture such as
of the distillation process. Only pressure at which the test has been bitumen has not been confirmed. To overcome this problem, we
carried out and the resulting experimental distillation curve are determine the interaction energy parameters between the
required. This is the remarkable advantage of the model developed pseudo-components of bitumen using the residue curve map
by Doherty and Perkins [6–8]. method.
Residue curve map has been used by Satyro and Yarranton [2] After characterization of the bitumen, we follow the u–c
to characterize several oil samples using distillation curve data. approach to calculate the solubility of light hydrocarbon and
They assumed ideal gas and liquid phases in their thermodynamic non-hydrocarbon gases (solvents) in bitumen. Using this approach,
equilibrium calculations. While this assumption is quite well the energy interaction parameters between molecules of the sol-
applicable for the gas phase, it may not be valid for the liquid phase vent and the pseudo-components are the only required parameters
especially for heavy oil and bitumen. Ortiz et al. [9] used the resi- to be tuned for solubility calculations.
due curve map upon well-defined mixtures to tune equation of In summary, we propose a general approach with a solid
state parameters and then calculated the phase behavior data. thermodynamic background to characterize bitumen. The devel-
One of the important parameters in heavy oil and bitumen oped method can be used to calculate the required interaction
applications is the solubility of light components in bitumen or parameters and solubility of light solvents in bitumen.
heavy oil. Prediction of solubility of gaseous solvents in bitumen Application of this methodology can be an alternative for expen-
is key for reservoir simulation and engineering studies of bitumen sive laboratory PVT tests. Therefore, it will decrease the costs of
production, upgrading, fractionation and refining. Once the solubil- bitumen characterization and find many applications in phase
ity of solvent in bitumen is defined, density, viscosity and other behavior studies related to bitumen and extra heavy oil production
physical properties can be calculated using the established meth- and refining.
ods. In previous solubility models bitumen has been considered The rest of this paper is organized as follow: first, we present
as single pseudo-component (such as Zirrahi et al. [10]; Fu et al. our methodology to characterize bitumen. Then, the solubility
[11]) or multicomponents (such as Kariznovi et al. [12]; Eastick model is presented. Next, accuracy of the approach is evaluated
et al. [13]; Huang and Radosz [14]; Mehrotra and Svrcek [15]). using experimental data and discussed followed by the summary
After definition of the pseudo-components, a proper equation of and conclusion.
state should be tuned to find the binary interaction parameters
(kij). The tuned equation of state then can be used to find
2. Bitumen characterization procedure
thermophysical properties required for engineering calculations.
Tuning of equation of state for conventional oils is typically per-
We use the results obtained from simulated distillation (SD)
formed using the data obtained from PVT experiments such as con-
assay to characterize the bitumen. The SD test results in mass per-
stant composition expansion (CCE), differential liberation and
cent of the distilled (% off) versus the temperature. Residue curve
separator tests. For bitumen, aforementioned laboratory tests are
map, which is the trajectory of bubble point temperature of the
not common. Therefore, there are no independent calculations to
mixture from the initial to the end state of batch distillation, is
find and tune the interaction parameters between pseudo-compo-
applied to represent the simulated distillation results. As described
nents. Some studies estimated the interaction parameters using
by Doherty and Perkins [6–8] liquid composition changes during
correlations while some ignored these interaction parameters
the simple batch distillation can be represented by:
[14,15].
Kariznovi et al. [12] defined the pseudo-components and their dxðnÞ
¼ xðnÞ yðxðnÞ; TÞ ¼ Fðn; xÞ ð1Þ
properties such as molecular weight, specific gravity to match dn
the average molecular weight and specific gravity of bitumen.
where x = [x1, x2, xs, . . .., xN] and y = [y1, y2, ys, . . ., yN] are the liquid
They tuned the normal boiling point, acentric factor, and EoS
and gas mole fraction arrays, respectively, N is the number of
parameters (XA and XB) of pseudo-components and the binary
pseudo components, n is the dimensionless time called warped time
interaction parameters between solvent-pseudo-component (ks-
varying from 0 to +1 where n = 0 corresponds to the start of the dis-
pc) to match the experimental solubility data.
tillation (t = 0) and n = +1 corresponds to the time when the dis-
Eastick et al. [13] and Huang and Radosz [14] characterized Cold
tillation flask becomes empty. Liquid hold-up in the flask is
Lake bitumen using five pseudo-components to calculate solubility
calculated by:
of CO2 in the bitumen and its pseudo-components. They applied
Peng–Robinson equation of state (PR-EoS) and statistical associat- H ¼ H0 expðnÞ ð2Þ
ing fluid theory (SAFT), respectively. Mehrotra and Svrcek [15] where H is the liquid hold-up and subscript 0 denotes the initial
divided Cold Lake bitumen into three pseudo-components, includ-
hold-up [6]. At each warped time, the liquid phase is at saturation
ing asphaltene, distillable and undistillable maltene. They used PR- condition (T = Tsat) and the gas phase composition can be obtained
EoS to model the gas and liquid phases and ignored the interaction
using the saturation temperature equilibrium calculations. The set
between pseudo-components. of Eqs. (1) and (2) was solved using fourth-order Runge–Kutta
Previously, we used the Krichevsky–Ilinskaya’s equation [10] to
method with the warped time interval Dn as:
correlate solubility of solvents in bitumen where bitumen was
assumed as a single component without characterizing the bitu- 1
men. In this work we propose an approach using the simulated dis-
xnþ1 ¼ xn þ Dnðk1 þ 2k2 þ 2k3 þ k4 Þ ð3Þ
6
tillation (SD) data to characterize the bitumen by applying the
where
residue curve map. We apply the u–c (u is the fugacity coefficient
of a component in the gas phase and c is the activity coefficient of k1 ¼ f ðnn ; xn Þ ð4Þ
the component in the liquid phase) approach to model the
thermodynamic equilibrium of gas and liquid phases. We use the 1 1
NRTL activity model to capture the non-ideal behavior of bitumen k2 ¼ f nn þ Dn; xn þ Dnk1 ð5Þ
2 2
molecules. The major difficulty to apply NRTL model is the
242 B. Azinfar et al. / Fuel 153 (2015) 240–248
1 1 We used the central finite difference to find the right hand of Eq.
k3 ¼ f nn þ Dn; xn þ Dnk2 ð6Þ
2 2 (22) with temperature difference of 1 K (DT = 1). Other properties
of bitumen and heavy oil that should be considered for character-
k4 ¼ f ðnn þ Dn; xn þ Dnk3 Þ ð7Þ ization are the average molecular weight (M) and specific gravity
(SG). We also used the following mixing rules to find the average
T sat ¼ f ðxn ; PÞ ð8Þ molecular weight and specific gravity of bitumen [1].
X
sat xi M i ¼ M a v e ð23Þ
yn ¼ KðT ; xn Þxn ð9Þ
X xi M i M av e
nn ¼ nn1 þ Dn ð10Þ ¼ ð24Þ
SGi SGav e
Hn ¼ H0 expðnn Þ ð11Þ Correlation of Rao and Bardon [21] was applied to relate the
boiling point of each pseudo-component to its molecular weight
H 0 M Hn M n as given by:
%off ¼ 100 ð12Þ
H0 M
T bi
where M is the molecular weight, n is the step index and Dn is the ln M i ¼ ð1:27 þ 0:071K w Þ ln ð25Þ
22:31 þ 1:68K w
integration step of the warped time. This parameter (Dn) was set to
0.005 as the optimum integration step for the sake of accuracy and where Tb is the boiling temperature and Kw is the Watson character-
low run time for calculations. Tsat and K are the saturation tempera- ization factor. Then, we used SØreide [22] correlation to relate the
ture at the distillation pressure and equilibrium constant at Tsat and molecular weight and specific gravity of each pseudo-component.
P, respectively. To find the equilibrium constant, ideal gas phase This correlation has been developed using 843 samples obtained
was assumed (u = 1). Therefore, thermodynamic equilibrium can from 68 reservoir fluids. This correlation is given by:
be written as:
SGi ¼ 0:2855 þ C f ðM i 66Þ0:13 ð26Þ
yi P ¼ xi i Psat
c i ð13Þ
where Cf obtained to satisfy the Eq. (26) for representation of
where Psat and ci are the saturation pressure and the activity coeffi- experimental specific gravity of reservoir fluid using its molecular
cient of component i in the liquid phase, respectively. To find the weight. This parameter is typically between 0.27 and 0.31 [22].
saturation pressure, Lee–Kesler [16] equation was used as given by: To perform characterization, first, 28 pseudo-components were
defined with boiling point range from 300 to 1110 K with interval
ln Psat 0
r ¼ h þxh
1
ð14Þ
of 30 K. The lower bound (300 K) is between boiling point of C4H10
and C5H12, which are the lightest possible hydrocarbon in bitumen.
6:09648
h0 ¼ 5:92714 1:28862 ln T r þ 0:169347T 6r ð15Þ The lightest hydrocarbon component in bitumen is usually C7H16
Tr to C10H22. The upper limit that simulated distillation can capture
is C120H242 with a boiling point of 1023 K. Therefore, the range of
15:6875
h1 ¼ 15:2518 13:4721 ln T r þ 0:43577T 6r ð16Þ 300 to 1110 K can safely capture the hydrocarbon components pre-
Tr
sented in bitumen. Since the simulated distillation apparatus is
where Tr and Pr are the reduced temperature (Tr = T/Tc) and pressure calibrated with normal paraffin (n-alkanes) components, we used
(Pr = P/Pc), respectively, and x is the acentric factor. Furthermore, Twu [23] correlations to calculate the molecular weight, critical
NRTL model was applied to calculate the activity coefficient of com- temperature and pressure for each pseudo-component using its
ponents in the mixture as given by [17]: boiling point. Also, acentric factor was obtained using Lee–Kesler
P P
j xj sji Gji
X xj Gjj xs G
equation [1], which has been presented specifically for petroleum
lnðci Þ ¼ P þ P sij Pl l lj lj ð17Þ fractions [24]. Next, mole fraction of each pseudo-component
k xk Gki j k xk Gkj k xk Gkj was optimized by solving Eqs. ((1)-(12)) to find the best match of
the SD assay results.
where Gij is a characteristic of interaction energies between mole-
After finding the mole fraction distribution of the pseudo-com-
cules of i and j given by [17]:
ponents present in bitumen, the average Watson characterization
Gij ¼ expðaij sij Þ ð18Þ factor is obtained to find the average molecular weight of the bitu-
men using Eqs. (23) and (25). Then, SØreide correlation (Eq. (26)) is
g ij g jj tuned with Cf = 0.308 to match the experimental specific gravity of
sij ¼ ð19Þ
RT bitumen or heavy oil using Eqs. (24) and (26). Characterization
where g is the energy interaction parameter and a specifies the ran- procedure is shown in a diagram shown in Fig. 1.
domness in the mixture. In this work a = 0.3 was assumed for all
pseudo-components of bitumen [18]. Energy interaction parameter 3. Solubility modeling
between molecules i and j (gij) was calculated assuming the geo-
metric mean proposed by Vetere [19] as: The output of characterization scheme is a multicomponent
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi system, which is not suitable for computationally demanding ther-
g ij ¼ g jj g ii ð20Þ
mal reservoir simulators. Reservoir fluids are usually lumped into
Tassios [20] proposed defining the energy parameter of mole- manageable number of pseudo-components to decrease the run
cule i by the heat of vaporization (DHvap) as follows: time required for compositional and thermal reservoir simulations
[1,24]. Molecular weight, Mi, which separates the groups is
g ij ¼ DU ¼ ðDHv ap RTÞ ð21Þ obtained as given by [24]:
Applying Clausius–Clapeyron equation we reach to: Mi ¼ M 1 ðMn =M 1 Þi=NH ð27Þ
v ap sat
DH d ln P where NH is the number of groups (pseudocomponents), i = 1, . . .,
¼ ð22Þ
RT dT NH, M1 is the molecular weight of the lightest component and Mn
B. Azinfar et al. / Fuel 153 (2015) 240–248 243
is the molecular weight of the heaviest component. Components After grouping of the components using the scheme described
with molecular weight falling between Mi1 and Mi are lumped in above, we use the u–c approach to calculate the solubility of gases
group i. Next, mixing rules proposed by Lee and Kesler [16] were in bitumen for the sake the consistency with the characterization
applied to find the critical properties and acentric factor of each procedure described earlier. Therefore, the thermodynamic
group as follows: equilibrium of a component in gas and liquid phases can be written
h PP 3
i as:
1=3
1
8 i j xi xj ðV ci þ V 1=3
cj Þ L
V cg ¼ P 2 ð28Þ yi ui P ¼ xi ci f i ð33Þ
i xi
where u denotes the fugacity coefficient of specie i in the gas phase.
3
fL is the fugacity in hypothetical liquid state given by Prausnitz et al.
PP 0:5
1
8V cg i j xi xj ðT ci T cj Þ V ci1=3 þ V cj1=3 as [18]:
T cg ¼ P 2 ð29Þ !
i xi L 0 V Li ðP 1:013Þ
f i ¼ f i P c exp ð34Þ
P RT
xx
xg ¼ Pi i i ð30Þ
i xi and f0 is the reduced hypothetical liquid fugacity at atmospheric
pressure. VLi is the partial molar volume of component i in the liquid
Z cg ¼ 0:2905 0:085xg ð31Þ phase.
Prausnitz et al. [18] reported f0 in a graphical format, which was
Z cg RT cg later presented by Riazi and Vera [25] using the correlation given
Pcg ¼ ð32Þ
V cg by:
where Tc, Pc, Vc and Zc are critical temperature, pressure, volume and 0 8:19643
f i ¼ exp 7:902 3:08 lnðT r Þ ð35Þ
compressibility factor of the group, respectively. R is the universal Tr
gas constant and x is the acentric factor.
244 B. Azinfar et al. / Fuel 153 (2015) 240–248
Table 2 0.4
Mole fraction and properties of lumped pseudo-components.
299.95 K, Mehrotra and Svrcek [29]
319.15 K, Mehrotra and Svrcek [29]
Pseudo Mole Molecular Critical Critical Acentric
376.55 K, Mehrotra and Svrcek [29]
components fraction weight temperature Pressure factor
Mole fraction of CH 4
373.15 K, Fu et al. [11]
(K) (MPa) 0.3 Proposed Model
PC1 0.1268 153.4 570.85 2.517 0.3965
PC2 0.3922 306.8 700.12 1.572 0.6750
PC3 0.2905 551.2 820.60 0.909 1.0563
PC4 0.1163 1005.9 958.19 0.453 1.5650 0.2
PC5 0.0742 1485.4 1061.0 0.274 1.9135
temperature. The linear function successfully captured the interac- 0.1 (a)
tion between the pseudo-components and the solvent in the mix-
ture. Table 3 lists the parameters required to calculate binary
2 4 6 8 10
interaction parameters for light hydrocarbon solvents (CH4 and
Pressure (MPa)
C2H6) and non-hydrocarbons (CO2 and N2).
The experimental data for solvent solubility in bitumen is very 0.4
scarce in the open literature. Here, we apply the same reference
data sets [11,29] used in our previous study [10] to validate the 343.15 K, Fu et al. [11]
350.55 K, Mehrotra and Svrcek [29]
developed characterization procedure as well as the solubility 423.15 K, Fu et al. [11]
Table 3
Parameters for implementation of Eq. (35) to determine the binary interaction energy parameter of the solvents and pseudo-components.
Solvent A B C D
CH4 1.593 8.8262 104 2.9303 103 1.0354 103
C2H6 9.1211 102 8.6345 104 9.9354 105 2.0314 103
CO2(supercritical) 0.8329 2.7713 103 1.9092 106 0.2809
CO2(subcritical) 5.9255 2.2605 102 4.0392 103 10.801
N2 4.7267 1.1687 103 3.6593 103 7.129 102
246 B. Azinfar et al. / Fuel 153 (2015) 240–248
0.8 0.7
(a)
0.6
0.6 0.5
0.4
0.4
297.35 K, Mehrotra and Svrcek [29] 0.3
325.25 K, Mehrotra and Svrcek [29] 288.15 K, Mehrotra and Svrcek [29]
349.35 K, Mehrotra and Svrcek [29] 299.15 K, Mehrotra and Svrcek [29]
343.15 K, Fu et al. [11] 0.2 350.05 K, Mehrotra and Svrcek [29]
Proposed Model Proposed Model
0.2
0.1
(a)
2 4 6 8 10 2 4 6 8 10 12
Pressure (MPa) Pressure (MPa)
0.8
0.8 325.65 K, Mehrotra and Svrcek [29]
(b) 323.3 K, Yu et al. [28]
371.05 K, Mehrotra and Svrcek [29]
373.3 K, Yu et al. [28]
0.6 473.7 K, Yu et al. [28]
0.4
0.4
Fig. 4. Comparisons between the results of the proposed model and experimental Fig. 5. Comparisons of the results of the proposed model and the experimental CO2
C2H6 solubility data in bitumen. Experimental data was obtained from Fu et al. [11] solubility data in bitumen. Experimental data was obtained from Yu et al. [28] and
and Mehrotra and Svrcek [29]. Mehrotra and Svrcek [29].
Table 4
AARDs of the experimental solubility data of light solvents in bitumen and the results of the proposed model and other predictive models.
Exp. data Predictive models AARD (%) Temperature range (K) Pressure range (MPa)
CH4
Mehrotra and Svrcek [29] Proposed model 3.4 299.85–376.95 2.57–10.07
Mehrotra and Svrcek [15] 2.7
Kariznovi et al. [12] 3.74
Fu et al. [11] Proposed model 6.2 343.15–423.15 2.17–11.83
Mehrotra and Svrcek [15] 4.5
Fu et al. [11] 4.82
C2H6
Mehrotra and Svrcek [29] Proposed model 5.7 296.05–375.85 1.02–10.07
Mehrotra and Svrcek [15] 5.7
Kariznovi et al. [12] 7.64
Fu et al. [11] Proposed model 7.2 343.15–423.15 2.62–10.96
Mehrotra and Svrcek [15] 6.5
Fu et al. [11] 4.72
CO2
Mehrotra and Svrcek [29] Proposed model 2.0 287.85–371.35 2.14–10.95
Kariznovi et al. [12] 8.89
Eastick et al. [13] 6.8
Mehrotra and Svrcek [15] 1.9
Yu et al. [28] Proposed model 5.8 323.15–523.1 4–16.04
Huang and Radosz [14] 5.67
Huang and Radosz [30] 5.44
N2
Mehrotra and Svrcek [29] Proposed model 7.6 303.85–371.35 2.46–10.66
Kariznovi et al. [12] 5.62
Mehrotra and Svrcek [15] 8.1
P x
xexperimental
AARD (Absolute Average Relative Deviation) = 1n calculated
xexperimental .
B. Azinfar et al. / Fuel 153 (2015) 240–248 247
0.16 accuracy of the proposed model was also compared with that of
some of the predictive models reported in the literature and
0.14 acceptable accuracy was found.
304.15 K, Mehrotra and Svrcek [29]
329.95 K, Mehrotra and Svrcek [29] This model can be applied to characterize bitumen and extra
0.12 370.95 K, Mehrotra and Svrcek [29] heavy oils. Application of the proposed model for bitumen
Mole fraction of N 2
Acknowledgments
0.06
0.04
The authors would like to thank the reviewers for their con-
structive comments. The authors also wish to express their
0.02 appreciation for the financial support of all member companies
2 4 6 8 10 of the SHARP Research Consortium: Alberta Innovates Energy and
Pressure (MPa) Environment Solutions, Athabasca Oil Sands, BP Canada Energy
Group ULC, Brion Energy, Chevron Energy Technology Co.,
Fig. 6. Comparisons of the results of the proposed model and the experimental N2 Computer Modelling Group Ltd., ConocoPhillips Canada, Devon
solubility data in bitumen. Experimental data was obtained from Mehrotra and
Canada Co, Foundation CMG, Husky Energy, Japan Canada Oil
Svrcek [29].
Sands Limited, Nexen Energy ULC, Laricina Energy Ltd., Natural
Sciences and Engineering Research Council of Canada (NSERC),
of the solubility of N2 in bitumen were compared with the Osum Oil Sands Co., PennWest Energy, Statoil Canada Ltd.,
Mehrotra and Svrcek [29] experimental N2 solubility data, result- Suncor Energy and Total E&P Canada. The support of the
ing in an AARD of 7.6%. Fig. 6 shows the comparison between Department of Chemical and Petroleum Engineering and the
results of the proposed model and experimental data of N2 solubil- Schulich School of Engineering at the University of Calgary is also
ity in bitumen. This experimental data does not show consistent acknowledged.
trend with temperature while increasing N2 solubility by increas-
ing pressure can be found. References
We also compared the predictions of the proposed model with
those of the models reported in the literature (Table 4). Our pro- [1] Riazi MR. Characterization and properties of petroleum fractions (1st edition).
ASTM Manual Series. Philadelphia, PA: American Society of Testing and
posed model proved to be more accurate in some cases. Table 4 Materials; 2005.
highlights the applicability and accuracy of the proposed model [2] Satyro MA, Yarranton H. Oil characterization from simulation of experimental
for calculating the solubility of light hydrocarbon and non-hydro- distillation data. Energy Fuels 2009;23:3960–70.
[3] ASTM D5236-13. Standard Test Method for Distillation of Heavy Hydrocarbon
carbon solvents in bitumen over a wide range of temperatures Mixtures (Vacuum Potstill Method); 2007.
and pressures. This method can be used for bitumen character- [4] Saajanlehto M, Alopaeus V. Heavy oil characterization method for PC-SAFT.
ization and solubility calculations involved in bitumen recovery Fuel 2014;133:216–23.
[5] Saajanlehto M, Uusi-Kyyny P, Alopaeus V. Hydrogen solubility in heavy oil
and processing. systems: experiments and modeling. Fuel 2014;137:393–404.
[6] Doherty MF, Perkins JD. On the dynamics of distillation processes I: the simple
distillation of multicomponent non-reacting, homogeneous liquid mixtures.
5. Conclusions Chem Eng Sci 1978;33:281–301.
[7] Doherty MF, Perkins JD. On the dynamics of distillation processes II: the simple
distillation of model solutions. Chem Eng Sci 1978;33:569–78.
The ability to accurately characterize bitumen and predict the [8] Doherty MF, Perkins JD. On the dynamics of distillation processes III: the
solubility of solvents in bitumen are essential to engineering topological structure of ternary residue curve maps. Chem Eng Sci
calculations for bitumen production, upgrading, fractionation, 1979;34:1401–14.
[9] Ortiz DP, Satyro MA, Yarranton HW. Thermodynamics and fluid
and refining. Due to the high molecular weight and boiling point
characterization using trajectory optimization. Fluid Phase Equilib
of the hydrocarbon molecules present in bitumen, conventional 2013;351:34–42.
characterization methods are not applicable or are not [10] Zirrahi M, Hassanzadeh H, Abedi J, Moshfeghian M. Prediction of solubility of
straightforward. CH4, C2H6, CO2, N2 and CO in bitumen. Can J Chem Eng 2014;92:563–72.
[11] Fu CT, Puttagunta R, Vilcsak G. Vapour-liquid equilibrium properties for gas-
In this work, we proposed a fast, accurate, and inexpensive Cold Lake bitumen. In: The 37th Annual Technical Meeting of the Petroleum
methodology based on solid theoretical background to characterize Society of CIM, Calgary, June 1986.
bitumen and tune an equation of state to predict the solubility of [12] Kariznovi M, Nourozieh H, Abedi J. Bitumen characterization and
pseudocomponents determination for equation of state modeling. Energy
light solvents in bitumen. The residue curve map method was Fuels 2010;24:624–33.
applied to characterize the bitumen using simulated distillation [13] Eastick RR, Svrcek WY, Mehrotra AK. Phase behaviour of CO2-bitumen
(SD) data. The Peng–Robinson equation of state (PR-EoS) and the fractions. Can J Chem Eng 1992;70(1):159–64.
[14] Huang SH, Radosz M. Phase behavior of reservoir fluids V: SAFT model of CO2
NRTL activity model were applied to capture the departure from and bitumen systems. Fluid Phase Equilib 1991;70:33–54.
the ideal state in the gas and liquid phases. The proposed method [15] Mehrotra AK, Svrcek WY. Correlation and prediction of gas solubility in Cold
successfully characterized the bitumen and was applied to esti- Lake bitumen. Can J Chem Eng 1988;66(4):666–70.
[16] Lee BI, Kesler MG. A generalized thermodynamic correlation based on three-
mate the interaction parameters of the pseudo-components. parameter corresponding states. AIChE J 1975;21(3):510–27.
The results of the characterization compared favorably with the [17] Wong DSH, Sandler SI. A theoretically correct mixing rule for cubic equations
experimental molecular weight, specific gravity, and simulated of state. AIChE J 1992;38(5):671–80.
[18] Prausnitz JM, Lichtenthaller RN, Azedevo EGD. Molecular thermodynamics of
distillation (SD) curve of the bitumen. We verified the applicability
fluid phase equilibria. New York: Prentice Hall; 1986.
of the proposed model for calculation of the solubility of light sol- [19] Vetere A. The NRTL equation as a predictive tool for vapor–liquid equilibria.
vents in bitumen by comparison with the experimental data. The Fluid Phase Equilib 2004;218:33–9.
calculated and the experimental solubility data of CH4, C2H6, CO2 [20] Tassios D. A single-parameter equation for isothermal vapor-liquid
equilibrium correlations. AIChE J 1971;17(6):1367–71.
and N2 were in good agreement (AARD less than 5.6 and 4.5% for [21] Rao VK, Bardon MF. Estimating the molecular weight of petroleum fractions.
hydrocarbon and non-hydrocarbon solvents, respectively). The Ind Eng Chem Process Des Dev 1985;24(2):498–500.
248 B. Azinfar et al. / Fuel 153 (2015) 240–248
[22] Søreide I. Improved phase behavior predictions of petroleum reservoir fluids [26] Peng DY, Robinson DB. A new two-constant equation of state. Ind Eng Chem
from a cubic equation of state. PhD dissertation, Norwegian Inst. of Fundam 1976;15:59–64.
Technology, Trondheim, Norway, 1989. [27] Kontogeorgis GM, Folas GK. Thermodynamic models for industrial
[23] Twu CH. An internally consistent correlation for predicting the critical applications: From classical and advanced mixing rules to association
properties and molecular weights of petroleum and coal-tar liquids. Fluid theories. 1st ed. NewYork: JohnWiley & Sons, Ltd.; 2010.
Phase Equilib 1984;16(2):137–50. [28] Yu JM, Huang SH, Radosz M. Phase behavior of reservoir fluids: supercritical
[24] Whitson CH, Brule MR. Phase Behavior, Monograph, vol. 20. Richardson, carbon dioxide and Cold Lake bitumen. Fluid Phase Equilib 1989;53:429–38.
Texas: SPE, Henry L. Doherty Series; 2000. [29] Mehrotra AK, Svrcek WY. Properties of Cold Lake bitumen saturated with pure
[25] Riazi MR, Vera JH. Method to calculate the solubilities of light gases in gases and gas mixtures. Can J Chem Eng 1988;66(4):656–65.
petroleum and coal liquid fractions on the basis of their P/N/A composition. [30] Huang SH, Radosz M. Phase behavior of reservoir fluids II: supercritical carbon
Ind Eng Chem Res 2005;44:186–92. dioxide and bitumen fractions. Fluid Phase Equilib 1990;60:81–98.