Fuel 215 (2018) 766–777

Contents lists available at ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

Full Length Article

Regular solution based approach to modeling asphaltene precipitation from T

native and reacted oils: Part 2, molecular weight, density, and solubility
parameter of saturates, aromatics, and resins
⁎
H.W. Yarrantona, , D.P. Powersa, J.C. Okafora, F.G.A van den Bergb
a
Department of Chemical and Petroleum Engineering, University of Calgary, 2500 University Dr. NW, Calgary, Alberta T2N 1N4, Canada
b
Black Oil Solutions, Park Ypenburg 2, 2396 CS Koudekerk aan den Rijn, The Netherlands

A R T I C L E I N F O A B S T R A C T

Keywords: The molecular weight, density, and solubility parameter of the crude oil medium are inputs to regular solution
Saturates based models for asphaltene precipitation. For native oils, these properties can be determined from a SARA
Aromatics (saturate, aromatic, resin, asphaltene) assay and existing correlations for each SARA component. However,
Resins thermo- and hydrocracking alter these properties and these changes must be accounted for when modeling the
Molecular weight
solubility of reacted oils. The effect of cracking on asphaltene properties was considered in a previous study. In
Density
the current study, the molecular weight and density for saturate, aromatic, and resin fractions from 15 native
Solubility parameter
Asphaltene precipitation and reacted fluid samples are measured. The solubility parameters of saturates and aromatics are determined by
Regular solution model fitting asphaltene precipitation yield data in these solvents with a modified regular solution model. The solu-
Thermocracking bility parameter of resins is calculated using correlations previously developed for the asphaltenes. An elemental
Hydrocracking analysis of each fraction is also performed. The effect of thermo- and hydrocracking on the SAR fraction
properties is discussed. Average properties and correlations are developed to predict the SAR fraction properties
in the absence of direct measurements. These properties and correlations are a key step in extending regular
solution models for asphaltene precipitation to refinery streams.

1. Introduction Crude oils are typically characterized into pseudo-components for

Asphaltene precipitation can occur during the depressurization of
light crude oils [1,2], when diluting heavy oils for recovery or transport
processes [3], and when different streams are blended to obtain desired
feed or product properties in refinery processes [4,5]. Asphaltene pre-
cipitation in the oilfield can lead to deposition and fouling in reservoirs,
wellbores, and surface facilities. In refineries, it can cause equipment
fouling, catalyst deactivation, and process control problems [6]. Hence,
a reliable method to predict asphaltene precipitation at different con-
ditions is required to design and operate both upstream and down-
stream processes.
One approach to model asphaltene precipitation is Scatchard-
Hildebrand regular solution theory. Hirschberg et al. added a Florry-
Huggins entropic contribution to apply the theory to asphaltene pre-
cipitation from crude oils [7]. This modified regular solution model
(RSM) has been extended to predict asphaltene precipitation from crude
oils from different sources, blends, and live oils [8–14]. The modified
regular solution model (RSM) requires the molecular weight, density,
and solubility parameter distributions within the crude oil.
phase behavior modeling. For refinery applications, this characteriza-
tion is based on a distillation assay. However, this assay is inadequate
for heavy oils and residues because only a small fraction of the fluid is
distillable in the former case and none is distillable in the latter case.
Therefore, the residue is also characterized based on a SARA (saturates,
aromatics, resins, asphaltenes) assay. The version of the RSM used in
this work uses SARA as the input to the model [12–14].
Akbarzadeh et al. reported typical properties of the SARA fractions
from native heavy oils [13]. However, the SARA properties of down-
stream fluids are unknown because these fluids contain reacted mate-
rials with altered chemistry compared with the native oils. Thermal or
hydro cracking processes not only changes the amount of each SARA
fraction in an oil but also alters the physical and chemical properties of
each SARA fraction [15–21]. Mild thermal cracking (visbreaking) and
severe thermal cracking of heavy residues removes alkyl side chains
and reduces the alkyl side chain length of the SAR fractions [21,22].
The cracked products can move from the resin to the aromatics and the
cracked side chains may report to the saturate or the gas phase. Under
thermal conversion conditions, the recombination of radicals may even

⁎
Corresponding author.
E-mail address: hyarrant@ucalgary.ca (H.W. Yarranton).

https://doi.org/10.1016/j.fuel.2017.11.071
Received 6 August 2017; Received in revised form 14 November 2017; Accepted 19 November 2017
Available online 22 December 2017
0016-2361/ © 2017 Elsevier Ltd. All rights reserved.
H.W. Yarranton et al. Fuel 215 (2018) 766–777

Nomenclature ρ density
τ fitting parameter in density equation
A1, A2 non-ideality constants in VPO calibration equation
C concentration Subscripts
ΔV voltage difference in VPO
Kihl equilibrium constant of component i between light and A asphaltenes
heavy liquid phases i component
K VPO calibration constant F feed
ṁ 524 °C + mass flow rate of the 524 °C + boiling cut mix mixture
M molecular weight min minimum
R universal gas constant max maximum
T temperature res resin
v molar volume s solvent
w mass fraction
x mole faction Superscripts
X fractional conversion
h heavy phase
Greek symbols l light phase
n shape factor for solubility parameter correlation
δ solubility parameter

create larger molecules; for example, two resin molecules forming an 2.1. Chemicals and materials
asphaltene. Thermal cracking decreases the average molecular weight
of the asphaltenes but also increases their solubility parameter, ulti- Samples from three native oils, two vacuum bottoms, and ten
mately leading to the formation of unsoluble compounds and even coke
[16,22]. The loss of molecules which contribute to the solubilization of Table 1
asphaltenes also contributes to coke formation. Catalytic hydro- Distillables and SARA composition of distillation residue for native oils, vacuum residues,
in situ converted oils, thermocracked oil, and hydrocracked oils. Asphaltenes are C5-as-
conversion is more complex: in addition to the thermal cracking reac-
phaltenes. IS indicates in-situ converted samples and the number indicates the pressure in
tions, radical reactions are suppressed, more labile hetero-atoms such as the reservoir in MPa when the sample was obtained. TC and HC indicate thermocracked
sulfur and oxygen are removed and metals are deposited on the catalyst and hydrocracked samples, respectively, and the number indicates conversion in mass%.
as well [23,24]. Hydro-processed asphaltenes also show a decrease in C indicates a commercial sample obtained from a refinery process. All other samples were
molecular weight and an increase in aromaticity corresponding to obtained from pilot plant processes. S-IC, F-TC and F-HC denote the source and feeds for
the in situ, thermocracked, and hydrocracked samples, respectively.
changes in chemical structure [25,26].
For both catalytic and thermal cracking, some of the reacted as- Sample Distillables Saturates Aromatics Resins Asphaltenes TI***
phaltenes will report to other SAR fractions changing their properties. wt% of oil
Similarly, some of the components in the SAR fractions will react al- wt% of residue
tering the chemistry of that fraction whether they remain in that frac-
Native Crudes*
tion or report to another. Hence, the proportion and properties of each WC-B-B2 (S- <5 17.0 44.0 19.0 19.8 0.16
SARA fraction depends on the severity of the reaction conditions. The IC)
alteration of the SARA properties with upgrading, particularly their WC-DB-A2 20 21.3 45.0 23.8 10.0 0.02
WC-B-C1 15 15.3 44.7 21.2 18.7 0.07
solubility parameters has not been systematically investigated. In Part I
of this work, the properties distributions of the asphaltenes were de- Vacuum
termined [27]. It remains to determine the properties of the rest of the Bottoms**
WC-VB-B2 (F- 0 5.1 37.4 20.2 36.4 0.99
crude oil. TC)
The main objective of this work was to determine the molecular WC-SR-A3 (F- 0 7.9 37.6 26.7 27.6 0.10
weight, density, and solubility parameter of the non-distillable SAR HC)
fractions of native and reacted oils. Molecular weights and densities In Situ
were measured directly and average properties for different types of WC-B-IS10 61 28.9 39.5 21.1 10.5 0.02
reaction were established. Solubility parameters were indirectly de- WC-B-IS3 71 21.4 53.6 14.3 10.7 0.01
WC-B-IS2 34 17.9 46.3 20.9 14.9 0.04
termined from modeling asphaltene solubility data. The elemental
analysis of each fraction was also measured to help interpret the effect Thermocracked
of cracking on the SAR properties. The property data are used to de- WC-VB-TC17 0 5.0 33.7 18.8 42.0 0.60
WC-VB-TC31 0 4.0 31.3 13.1 49.2 2.34
velop guidelines and correlations towards a regular solution based ap- WC-VB-TC51 0 4.0 24.0 11.0 56.2 4.84
proach to predicting the stability of upgraded oils. The properties of the
Hydrocracked
distillables and modeling of whole oils are to be addressed in future
WC-SR-HC54 0 33.0 49.0 15.0 3.0 0.00
work. WC-SR-HC70 0 20.8 47.5 17.8 13.9 0.02
WC-SR-HC83 0 15.0 45.0 15.0 24.7 0.26
WC-SR- 0 20.0 47.0 18.0 14.9 0.06
HC79C
2. Experimental methods
* Samples were pretreated removing the distillable fraction with a standard simple
Elemental analysis (ASTM D5291, D5453, and D5792) was per- batch atmospheric distillation (ADC metrology apparatus from NIST, described elsewhere
formed by the Calgary Shell Technology Center. All other measure- [27,28].
ments were taken at the University of Calgary as described below. ** Samples were received as the residue after vacuum distillation performed at Shell
labs.
*** TI: Toluene insolubles.

767
H.W. Yarranton et al.
reacted oils were examined in this study and are listed in Table 1. All of
the samples except for WC-B-A1 were supplied by Shell Global Solu-
tions. Sample WC-B-A1 was provided by Syncrude. More details on the
source of each sample are provided elsewhere [27]. Note, the numbers
for the thermocracked (TC##) and hydrocracked (HC##) samples in-
dicate the extent of conversion of each sample in mass percent where
conversion is defined as follows:
[(ṁ 524 °C +)FEED−(m524 °C +)PRODUCT ]
Conversion =
(ṁ 524 °C +)FEED (1)
where ṁ 524 °C + is the mass flow rate of the 524 °C+ boiling cut. The
conversion of the in situ samples is unknown.
Technical grade- n-heptane, n-pentane, toluene and acetone pur-
chased from VWR International were used for SARA fractionation.
Molecular weight, density, and refractive index measurements, and
solubility experiments were carried out using analytical solvents
(Omnisolv high purity toluene and heptane) purchased from VWR.
Sucrose octaacetate (98%), octacosane (99%) and polystyrene standard
(99%) were obtained from Sigma-Aldrich Chemical Company.
2.2. Sample preparation – removal of volatiles
Many of the samples in this study contained volatile components
which cannot be captured in a SARA assay. Therefore, these samples
were distilled to remove the volatiles and the SARA assays were per-
formed on the atmospheric residue from the distillation. The distilla-
tions were performed using the Advanced Distillation Curve (ADC)
apparatus designed by NIST as in improved version of the ASTM D86
method [28]. The ADC procedure is described in detail elsewhere [29]
and is summarized in the supporting information. The typical un-
certainty in temperature measurements with the ADC apparatus is less
than ± 0.5 °C. The amount of distillables removed from each oil is re-
ported for the native oils, vacuum residues, in situ oils, thermocracked
short residues, and hydrocracked short residues in Table 1.
2.3. SARA fractionation
SARA fractionation was performed using a modified ASTM D4124
procedure as described elsewhere [14] and summarized in the sup-
porting information. The SARA compositions of the samples char-
acterized for the native oils, vacuum residues, in situ oils, thermo-
cracked short residues, and hydrocracked short residues are provided in
Table 1. The repeatability of the SARA contents are ± 1.9, 3.0, 2.8,
0.15, and 0.04 wt% for the saturates, aromatics, resins, asphaltenes, and
toluene insolubles (TI), respectively.
2.4. Molecular weight measurement
The molecular weights of the SAR fractions were measured using a
Jupiter Model 833 Vapor Pressure Osmometer (VPO). All the mea-
surements were performed with toluene at 50 °C. The voltage difference
output in the VPO, is related to the molecular weight of the solute, M2
as follows [30]:
ΔV 1 Asphaltene precipitation (solubility) measurements were performed
= K⎛ ⎜+ A1 C2 + A2 C22 + …⎞ ⎟
C2 ⎝ M2 ⎠ (2)
where ΔV is the voltage difference between the thermistors, C2 is the
solute concentration, K is the proportionality constant, and A1 and A2
are coefficients arising from the non-ideal behavior of the solution. In
most cases, at low concentrations, most of the higher order terms be-
come negligible, and Eq. (2) reduces to:
ΔV 1 asphaltenes were washed and dried in an oven under vacuum at 60 °C
= K⎛ ⎜+ A1 C2⎞ ⎟
C2 ⎝ M2 ⎠ (3)
For an ideal system, the second term in Eq. (3) is zero and ΔV/C2 is
768
Fuel 215 (2018) 766–777
constant. In this case, the molecular weight is determined from the
average ΔV/C2 as follows,
K
M2 =
(ΔV / C2) (4)
For the calibration, the molecular weight of the solute is known, and
the proportionality constant, K, is calculated by extrapolation of a plot
of ΔV C2 versus C2 to zero concentration as per Eq. (3). For a non-ideal
solution with an unknown solute, the molecular weight is calculated
from the intercept of a plot of ΔV C2 versus C2 this time solving for M2.
For an ideal solution with an unknown solute, the molecular weight is
calculated at each concentration from Eq. (4) and then averaged.
The instrument was calibrated with sucrose octaacetate (679 g/mol)
as solute and octacosane (395 g/mol) was used to check the calibration.
The measured molecular weight of octacosane was within 3% of the
correct value. The repeatability of the molecular weight measurements
for the SAR fractions was approximately ± 12% for all the samples.
The saturate and aromatic fractions exhibited slightly non-ideal
behavior with positive slopes (A1) in the plot of ΔV C2 versus C2 (see
supporting information). The average slopes for native saturates and
aromatics were found to be 0.13 mV/(g/L)2 and 0.09 mV/(g/L)2 re-
spectively. The slopes for reacted (thermocracked and hydrocracked)
saturates and aromatics were systematically lower than those of the
native fractions with average values of 0.04 mV/(g/L)2 and 0.02 mV/
(g/L)2, respectively. The molecular weights for these fractions were
determined by extrapolation using the average slopes and Eq. (3). The
VPO response for the resin fractions did not exhibit any consistent trend
(see supporting information). Given the relatively small slope for these
fractions, A1 was set to zero and the resin molecular weights were de-
termined from the average VPO response, Eq. (4).
2.5. Density measurements
Densities were measured at 21 °C and atmospheric pressure with an
Anton Paar DMA 4500 M density meter. The instrument precision
was ± 0.01 kg/m3 with an accuracy of ± 0.05 kg/m3. The saturate and
aromatic densities were measured directly. The repeatability of the
direct density measurements were ± 0.83 kg/m3 and ± 0.94 kg/m3 for
saturates and aromatics respectively. Resins densities could not be
measured directly because they were too viscous to handle or were solid
at ambient temperature. Instead, their densities were determined in-
directly from the measured densities of solutions of resins in toluene or
o-dichlorobenzene at different concentrations (from 2 to 160 g/L).
Density values were determined from a mixing rule. As noted by
Yarranton et al. [31], a mixture of resins and these solvents appear to
form regular solutions and therefore the resin densities are given by:
wres 1 w
= − s
ρres ρmix ρs (5)
where w is mass fraction, ρ is density, and subscripts mix, res, and s
indicate the mixture, resin, and solvent, respectively.
2.6. Asphaltene solubility measurements
at 21 °C and atmospheric pressure with mixtures of 10 g/L of asphal-
tenes in: a) saturates/toluene and b) aromatics/n-heptane. To measure
the mass of precipitate, the asphaltenes were dissolved in a specified
mass of the aromatic solvent by sonicating for 20 min. The non-aro-
matic solvent was added and the mixture sonicated for 45 min and left
to settle for 24 h. The mixture was then centrifuged at 4000 rpm for
6 min. The supernatant was decanted and the recovered/precipitated
until the mass was constant. Asphaltene precipitation yields were cal-
culated as the mass of precipitated asphaltenes divided by the initial
mass of asphaltenes. The data are reported as an asphaltene solubility
H.W. Yarranton et al.
curve, a plot of the yield of precipitated asphaltenes versus the mass
fraction of the poor solvent.
Typically in our laboratory, asphaltene precipitation measurements
are conducted with at least 10 cm3 of solvent. However, it was very
time consuming to obtain the saturates and aromatics required for these
experiments. Therefore, the volume of solvent was decreased to less
than 2 cm3 to minimize the consumption of the saturates and aromatics.
The repeatability for the yields from this procedure was ± 8 wt%
compared with less than 1 wt% error in a typical yield measurement.
Note, only WC-B-B2 C7-asphaltenes were used in this study.
3. Regular solution model
Details of the regular solution model (RSM) are provided elsewhere
[13] and a brief summary is given here. In this model, a liquid-liquid
equilibrium is assumed to exist between the heavy liquid phase (as-
phaltene-rich phase including only asphaltenes and resins) and the light
liquid phase (solvent-rich phase including all components). Since only
asphaltenes and resins are in the heavy phase, the activity coefficients
of the components in the heavy phase are set to unity. The equilibrium
constant, K, is then given by:
x ih vl vl vl
K ihl = l
= exp ⎧ln ⎜⎛ il ⎟⎞ ⎫− il + i (δil−δ l )2
xi ⎨ ⎝ v ⎠⎬ v RT (6)
⎩ ⎭
where x is mole fraction, R is the universal gas constant, T is tem-
perature, v is molar volume, δ is solubility parameter, subscript i de-
notes the ith component, and superscript l denotes the light liquid
phase. Once the equilibrium ratios are known, the phase equilibrium is
determined using standard techniques [32].
To use this model, the mole fraction (compositions), molar volume,
and solubility parameter of each component (or pseudo-component) in
the mixture must be specified. The properties of the solvents used in
this study, toluene and n-heptane, are known and are provided else-
where [14]. The mass fraction of each pseudo-component in the heavy
oil or bitumen is determined from the SARA assay. Saturates, aromatics,
and resins are each treated as a single uniform pseudo-component.
Their properties are discussed later in the results and discussion sec-
tions.
Asphaltenes are considered to consist of a distribution of nano-ag-
gregates with a range of molecular weights starting from the monomer
molecular weight. They are divided into mass fractions of different
molecular weight using the gamma distribution. The density of each
asphaltene pseudo-component is given by Powers et al. [27]:
ρA = ρmin + (ρmax −ρmin ) ∗ (1−exp (−wA ∗τ )) (7)
where wA is the cumulative mass fraction, ρA, ρmin and ρmax are the
asphaltene densities in kg/m3 at wA, wA = 0, and wA = 1, respectively,
and τ is a fitting parameter. The solubility parameter for each asphal-
tene (and resin) pseudo-component is determined as follows:
δ = δmin + Δδ∗ (wA)n (8)
where δ is the solubility parameter, wA is the cumulative mass fraction
of asphaltenes, the δmin and Δδ parameters set the minimum and range
of the solubility parameter, respectively, and the exponent n determines
the shape of the distribution [27].
The parameters required to specify the asphaltenes are the average
nano-aggregate molecular weight in the crude oil, the monomer mo-
lecular weight, the shape factor in the Gamma distribution (β), ρA, ρmin,
ρmax, τ, δmin, Δδ, and n. Typical values for the monomer molecular
weight, β, and density and solubility parameter equation coefficients
were provided Powers et al. [27]. Native asphaltenes precipitated from
WC-B-B2 bitumen were used for all of the tests in this study. Their
properties were determined by Powers et al. and are summarized in
Table 2 [27]. The average density of these asphaltenes was 1170 kg/m3
at 21 °C.
769
Fuel 215 (2018) 766–777
Note that the asphaltene molecular weight distribution was assumed
to remain constant even though the composition of the mixture may
change. It is possible that asphaltenes self-associate differently at dif-
ferent compositions. This possible effect was neglected because there
are few data with which to quantify the effect and, more importantly,
the model is not very sensitive to the asphaltene molecular weight and
is far more sensitive to the asphaltene solubility parameter.
4. Results
The molecular weight, density, and elemental analysis of the satu-
rates, aromatics, and resins for each oil sample were measured as de-
scribed in Section 3. However, the solubility parameters of the saturate
and aromatics could not be measured directly but rather were de-
termined from the asphaltene solubility data. The methodology used to
determine the solubility parameters is shown in Fig. 1 and described
below. Then all of the data for each fraction are summarized.
4.1. Determination of solubility parameters
The saturates and aromatics solubility parameters were determined
by fitting asphaltene precipitation data from toluene-saturates and
aromatics-heptane mixtures using the regular solution model summar-
ized earlier. The solubility parameter of the resins could not be de-
termined because they self-associate with the asphaltenes [33] and alter
the asphaltene property distribution required to model the yield data.
Recall the equilibrium constant in the modified regular solution
model (Eq. (6)) requires the molar volumes (densities and molecular
weights) and solubility parameters (δi) of each of the components (or
pseudo-components) making up the mixture. The properties of the
solvents used, toluene and n-heptane, are known from literature and the
properties of the asphaltenes used, WC-B-B2, were summarized in
Table 2. Hence, the only unknown is the solubility parameter of satu-
rates or aromatics in the mixture.
Fig. 2 shows the yield data for WC-B-B2 asphaltenes in solutions of
WC-B-B2 saturates and toluene (Fig. 2a) and WC-B-B2 aromatics and
heptane (Fig. 2b). The fitted model is also shown (solid line) and the
fitted solubility parameters were 16.0 and 20.5 MPa0.5, respectively.
For the solutions with saturates, the model did not match the shape of
the yield curve near the onset of precipitation. The reason for this
discrepancy is unknown. The model was fit to the higher yield data
because these data points were based on a larger mass of precipitate
and therefore are expected to be more accurate. The scatter in the yield
data from the solutions of aromatics is relatively high because the
aromatics were viscous and difficult to wash consistently from the as-
phaltenes when determining the asphaltene yield.
The solubility parameter could be varied by ± 0.3 MPa0.5 and still
fit the data within the scatter of the measurements (dotted lines). The
sensitivity of the model to the uncertainty of the input asphaltene
molecular weight and density was also examined. A molecular weight
error of 12% is equivalent to a 0.1 MPa0.5 shift in the fitted solubility
parameter. A density error of 1 wt% had negligible effect. Hence, the
precision of the fitted solubility parameter is approximately 0.3 MPa0.5.
The fitted yield data for all of the samples in this study are provided in
the supporting information except for the in situ sample data presented
later.
Table 2
Coefficients for asphaltene density and solubility parameter equations [27].
Parameter Value Parameter Value Parameter Value
MWmono, g.mol 800 ρmin, kg/m 3
1050 δmin, MPa0.5
20.15
MWavg, g.mol 4500 ρmax, kg/m3 1200 Δδ, MPa0.5 1.15
β 1 τ 5 n 4
H.W. Yarranton et al. Fuel 215 (2018) 766–777

Fig. 1. Implementation of modified regular solution model

Composition:
(RSM) to determine unknown solubility parameter of a
Mass solvent saturate or aromatic fraction. The unknown solubility
Mass saturate or aromatic parameter is adjusted to match the model yields to the
Mass asphaltenes measured yields.

Solvent:
MWs, ρs, δs xs
(known properties)

Saturate or Aromatic:
MWx, ρx, δx xx
(known MW, ρ)
RSM Yield
Asphaltenes MW1, ρ1, δ1 x1
(properties from [27]) .. ..
. .

MWm, MWavg, β MWi, ρi, δi xi

ρmin, ρmax, τ .. ..
. .
δmin, Δδ, n
MWn, ρn, δn xn

4.2. Properties of SAR fractions
The molecular weight, density, and solubility parameter of the sa-
turate and aromatic fractions from all of the samples are reported in
Tables 3 and 4, respectively. The molecular weight and density of the
resin fractions from all of the samples are reported in Table 5. The
atomic ratios for the saturates, aromatics, and resins for all of the
thermocracked and hydrocracked samples and their feeds are reported
in Tables 6–8, respectively. Note, the WC-VR_B2 vacuum bottoms
contained too little saturates to perform solubility experiments and
therefore its solubility parameter was not determined.
5. Discussion
5.1. Native oils
Figs. 3 and 4 compare the molecular weights and densities, re-
spectively, of SAR fractions from native oils. Both the molecular
weights and densities fall within the range of previous data reported by
Akbarzadeh et al. for saturates and aromatics from seven different na-
tive oils from around the globe [13]. Note, the resins from the WC-B-C1
bitumen had a higher molecular weight than the other samples. The
WC-B-C1 asphaltenes had to be more extensively washed (three times
for the sonicated-washing stage) than any other oil due to the “gummy”
nature of these asphaltenes, as explained elsewhere [27]. This extensive
washing likely transferred some low molecular weight (more soluble)
asphaltenes into the maltene (deasphalted oil) fraction. During SARA
fractionation, these washed-through asphaltenes would report to the
resin fraction increasing the molecular weight of these resins.
Fig. 5 shows that the solubility parameters of the native saturate
fractions are within 0.2 MPa0.5 of each other. The average saturate
solubility parameter of 16.1 MPa0.5 is within the measurement error of
the value of 15.9 MPa0.5 from Akbarzadeh et al. [13]. The solubility
parameters of the native aromatic fractions are only within 0.4 MPa0.5
of each other. The average aromatic solubility parameter of 20.1 MPa is
slightly lower than the value of 20.3 from Akbarzadeh et al. [13].
The properties of the native oils all fall within relatively narrow
ranges with maximum differences of 20 g/mol (10%), 62 kg/m3 (7%),
and 0.4 MPa0.5 (0.5%) for molecular weight, density, and solubility
parameter, respectively. Therefore, it is recommended to use the
average values presented in Table 9 when estimating these properties
for native oils. The average values for the molecular weight and density
were based on the combined data from Akbarzadeh et al. and this study
(excluding the WC-B-C1 resins) [13]. The 90% confidence intervals
based on both datasets are ± 25% and ± 8% for molecular weight and
density, respectively. The methodology to determine solubility para-
meters was updated in this study and therefore the average values are
based on the data from this study only. Note, since the resin solubility

100 100 Fig. 2. Asphaltene yields for 10 g/L WC-B-B2 as-

data data phaltenes in 21 °C solutions of: a) WC-B-B2 saturates
model model and toluene; b) heptane and WC-B-B2 aromatics.

80 sensitivity 80 sensitivity
Asphaltene Yield, wt%

Asphaltene Yield, wt%

60 60

40 40

20 20

(a) (b)
0 0
20 40 60 80 100 20 40 60 80 100
Saturate Content, wt% Heptane Content, wt%

770
H.W. Yarranton et al. Fuel 215 (2018) 766–777

Table 3
Molecular weight, density, and solubility parameter for saturates from native, vacuum
bottom, in-situ converted, thermocracked, and hydrocracked oils at 21 °C and atmospheric
pressure. S-IC, F-TC and F-HC denote the source and feeds for the in situ, thermocracked,
and hydrocracked samples, respectively.
Table 5
Molecular weight and density for resins from native, vacuum bottom, in-situ converted,
thermocracked, and hydrocracked oils at 21 °C and atmospheric pressure. S-IC, F-TC and
F-HC denote the source and feeds for the in situ, thermocracked, and hydrocracked
samples, respectively.

Sample Molecular Density Molar Solubility Parameter Sample Molecular Density Molar
Weight kg/m3 Volume MPa0.5 Weight kg/m3 Volume
g/mol cm3/mol g/mol cm3/mol

Native Native
WC-B-B2 (S-IC) 370 887.1 417 16.2 WC-B-B2 (S-IC) 990 1074.4 921
WC-DB-A2 440 888.2 495 16.2 ME-CO-A1 1000 1066.1 938
WC-B-C1 400 877.4 456 16.0 WC-DB-A2 1050 1066.5 985
WC-B-C1 1280 1064.2 1203
Vacuum Bottom
WC-VB-B2 (F-TC) 760 860 884 – Vacuum Bottom
WC-SR-A3 (F-HC) 710 870 816 – WC-VB-B2 (F-TC) 1400 1059.7 1321
WC-SR-A3 (F-HC) 1320 1045.0 1263
In Situ
WC-B-IS10 330 841.5 392 15.5 In Situ
WC-B-IS3 320 847.8 377 15.9 WC-B-IS10 830 1081.6 767
WC-B-IS2 360 860.6 418 16.0 WC-B-IS3 860 1073.1 801
WC-B-IS2 880 1081.7 814
Thermocracked
Thermocracked
WC-VB-TC17 770 896.5 859 14.5
WC-VB-TC17 1430 1072.0 1334
WC-VB-TC31 785 896.4 876 –
WC-VB-TC31 1080 1082.0 998
WC-VB-TC51 800 896.2 893 14.2
WC-VB-TC51 1030 1075.0 958
Hydrocracked
Hydrocracked
WC-SR-HC54 710 879.0 808 14.6
WC-SR-HC54 1000 1050.0 952
WC-SR-HC70 710 876.6 810 14.5
WC-SR-HC70 780 1086.5 718
WC-SR-HC83 560 875.6 640 14.6
WC-SR-HC83 700 1098.8 637
WC-SR-HC79C 560 876.9 639 15.0
WC-SR-HC79C 670 1098.0 610

Table 4
Table 6
Molecular weight, density, and solubility parameter for aromatics from native, vacuum
Carbon content and atomic ratios for saturates from vacuum bottom, thermocracked, and
bottom, in-situ converted, thermocracked, and hydrocracked oils at 21 °C and atmospheric
hydrocracked oils. S-IC, F-TC and F-HC denote the source and feeds for the in situ, ther-
pressure. S-IC, F-TC and F-HC denote the source and feeds for the in situ, thermocracked,
mocracked, and hydrocracked samples, respectively.
and hydrocracked samples, respectively.
Sample wt% C H/C N/C S/C
Sample Molecular Density Molar Solubility Parameter
Weight kg/m3 Volume MPa0.5
Vacuum Bottom
g/mol cm3/mol
WC-VB-B2 (F-TC) – – – –
WC-SR-A3 (F-HC) 86.19 1.86 0.0000 0.0226
Native
WC-B-B2 (S-IC) 440 1005.9 437 20.2 Thermocracked
WC-DB-A2 470 1002.9 469 19.9 WC-VB-TC17 85.45 1.80 0.0000 0.0023
WC-B-C1 480 1001.6 479 20.3 WC-VB-TC31 85.30 1.81 0.0000 –
WC-VB-TC51 85.34 1.85 0.0000 0.0020
Vacuum Bottom
WC-VB-B2 (F-TC) 840 1016.5 826 20.6 Hydrocracked
WC-SR-A3 (F-HC) 740 1005.0 736 20.2 WC-SR-HC54 85.83 1.87 0.0000 –
WC-SR-HC70 85.79 1.89 0.0000 0.0000
In Situ
WC-SR-HC83 85.83 1.90 0.0000 0.0000
WC-B-IS10 340 1008.3 337 20.2
WC-SR-HC79C 85.48 1.85 0.0000 0.0003
WC-B-IS3 300 1028.1 292 20.0
WC-B-IS2 380 1008.7 377 20.0

Thermocracked
WC-VB-TC17 780 1023.0 762 21.4
WC-VB-TC31 750 1042.0 720 – Table 7
WC-VB-TC51 690 1040.0 663 20.5 Carbon content and atomic ratios for aromatics from vacuum bottom, thermocracked, and
Hydrocracked hydrocracked oils. S-IC, F-TC and F-HC denote the source and feeds for the in situ, ther-
WC-SR-HC54 640 977.0 655 20.5 mocracked, and hydrocracked samples, respectively.
WC-SR-HC70 550 1026 536 20.4
WC-SR-HC83 470 1054 446 20.6 Sample wt% C H/C N/C S/C
WC-SR-HC79C 460 1034 445 20.6
Vacuum Bottom
WC-VB-B2 (F-TC) 81.86 1.39 0.0013 0.0275
WC-SR-A3 (F-HC) 83.53 1.46 0.0013 0.0008
parameter could not be determined experimentally, it is recommended
to calculate this solubility parameter as if the resins were a small as- Thermocracked
WC-VB-TC17 82.14 1.34 0.0015 0.0286
phaltene; that is, to use the minimum asphaltene solubility parameter of
WC-VB-TC31 82.49 1.18 0.0017 0.0270
20.15 MPa0.5. WC-VB-TC51 82.80 1.21 0.0018 0.0280

Hydrocracked
5.2. Vacuum residues WC-SR-HC54 88.12 1.44 0.0017 0.0026
WC-SR-HC70 86.71 1.29 0.0027 0.0107
WC-SR-HC83 87.59 1.17 0.0033 0.0112
Fig. 3 shows that the molecular weights of the SAR fractions from
WC-SR-HC79C 86.93 1.14 0.0034 0.0095
the vacuum residues (WC-VB-B2 and WC-SR-A3) were significantly
higher than the corresponding average molecular weights of the native

771
H.W. Yarranton et al. Fuel 215 (2018) 766–777

Table 8 22

Solubility Parameter, MPa 0.5

Carbon content and atomic ratios for resins from vacuum bottom, thermocracked, and WC-B-B2
hydrocracked oils. S-IC, F-TC and F-HC denote the source and feeds for the in situ, ther- 21 WC-DB-A2
mocracked, and hydrocracked samples, respectively. WC-B-C1
20
WC-VB-B2
Sample wt% C H/C N/C S/C
19 WC-SR-A3
Vacuum Bottom
WC-VB-B2 (F-TC) 80.08 1.30 0.0082 0.0247 18
WC-SR-A3 (F-HC) 80.98 1.42 0.0082 0.0302
17
Thermocracked
WC-VB-TC17 80.83 1.28 0.0098 0.0275
16
WC-VB-TC31 81.14 1.20 0.0097 0.0255
WC-VB-TC51 81.39 1.30 0.0092 0.0208
15
Hydrocracked Saturates Aromatics
WC-SR-HC54 85.84 1.34 0.0107 0.0039
Fig. 5. Comparison of solubility parameters of saturates and aromatics from native oils
WC-SR-HC70 84.42 1.26 0.0160 0.0108
and vacuum residues at 21 °C and atmospheric pressure.
WC-SR-HC83 84.27 1.17 0.0165 0.0105
WC-SR-HC79C 84.61 1.13 0.0170 0.0091
Table 9
Recommended properties for the SAR fractions for each type of oil.
1600
WC-B-B2 Fraction Molecular Weight Density at 21 °C Solubility Parameter
1400 WC-DB-A2 kg/cm3 MPa0.5
Molecular Weight, g/mol

g/mol
WC-B-C1
1200 WC-VB-B2 Native Oils
WC-SR-A3
1000 literature
Saturates 440 880 16.1
Aromatics 500 990 20.1
800 Resins 1050 1060 20.15

600 Vacuum Residues

Saturates 740 880 16.4
400 Aromatics 790 990 20.4
Resins 1360 1060 20.15
200
Thermocracked
0 Saturates same as feed same as feed 14.5
Saturates Aromatics Resins Aromatics MFi(1–0.0037X) ρFi(1 + 0.049X) same as feed
Resins MFi(1–0.0037X) ρFi(1 + 0.049X) 20.15 + 0.20(1-exp
Fig. 3. Comparison of molecular weights of saturates, aromatics, and resins from native
(−0.037X))**
oils and vacuum residues. The black bars for the literature data show the range of the
reported values for native oils [13]. Hydrocracked
Saturates same as feed* same as feed 14.5
Aromatics MFi(1–0.0059X) ρFi(1 + 0.049X) same as feed
1100 Resins MFi(1–0.0059X) ρFi(1 + 0.049X) 20.15 + 0.85(1-exp
WC-B-B2 (−0.050X))**
1050 WC-DB-A2
* valid up to conversion of approximately 65%;
Density, kg/m³

WC-B-C1
1000 WC-VB-B2
WC-SR-A3
950
literature
900
850
800
Saturates Aromatics Resins
Fig. 4. Comparison of densities of saturates, aromatics and resins from native oils and
vacuum residues at 21 °C and atmospheric pressure. The black bars for the literature data
show the range of the reported values for native oils [13]. Note the error in the density
measurements is less than 1 kg/m3 and therefore is not shown with error bars.
oil fractions. A higher fraction of low molecular weight compounds are
removed during the vacuum distillation process leaving behind a higher
molecular weight residue. These lower molecular weight compounds
come primarily from the saturate and aromatic fractions and may also
include some lower molecular weight resins; the effective cutpoint of
the distillation was approximately 520 °C and therefore any resins with
a boiling point between 500 and 600 °C can report to the vacuum gasoil
fraction. Therefore, the molecular weight of the SAR fractions from
vacuum residues is expected to be higher than the same fractions from
native oils. Note that the resins are more difficult to separate from
vacuum residues than from crude oils because the residues are more
** minimum asphaltene solubility parameter [27].
viscous. During the precipitation procedure, some light asphaltenes
may remain entrained with the resins also increasing their molecular
weight.
Fig. 4 shows that, although their molecular weights differed, the
density of the SAR fractions from the vacuum residues were almost the
same as the average densities of the native oil fractions. Density is more
sensitive to the chemical family than the size of the molecules, parti-
cularly for larger molecules. For example, the specific gravity of n-
hexadecane is 0.77 (MW = 226 g/mol) compared with 0.73 for n-de-
cane (142 g/mol). The specific gravity of toluene (MW = 110 g/mol) is
0.87 compared with 0.68 for n-heptane (MW = 100g/mol). Hence,
density varies significantly from saturates to aromatics but relatively
little for a saturate or aromatic fraction from one oil versus another oil.
The solubility parameter of the saturates from the vacuum residues
could not be measured because their saturate content was low and there
was too little of the fraction available. The solubility parameters of the
aromatics from the two residues were within 0.4 MPa0.5 of each other,
Fig. 5. The average value of 20.4 MPa0.5 is higher but within the ex-
perimental error of the value of 20.1 MPa0.5 for the native oil aromatics.
Like density, solubility parameters are more sensitive to chemical fa-
mily than molecular weight but a slightly higher solubility parameter is
consistent with a higher molecular weight. For example, the solubility
parameter of n-hexadecane is 16.3 MPa0.5 compared with 15.3 MPa0.5
for n-heptane and 18.3 MPa0.5 for toluene.

772
H.W. Yarranton et al.
In cases where the properties of vacuum residue fractions are un-
known, the average properties provided in Table 9 are recommended.
The average values for the molecular weight and the solubility para-
meter of the aromatics were based on the data from this study. The
densities of all fractions and the solubility parameter of the resins were
assumed to be the same as those of the native oil fractions. The solu-
bility parameter of the saturates was assumed to be 0.3 higher than that
of the native oils; that is, the same as the difference between the so-
lubility parameters of the vacuum residue aromatics versus the native
aromatics.
5.3. Thermocracked oils
The main reaction in thermocracking is the cleavage of alkyl chains
from naphthenic and aromatic structures. This process converts some
asphaltenes to resins and some resins to aromatics and produces some
smaller saturated and aromatic components which report to the gas
phase or the saturate and aromatic fractions. Some of the converted
aromatic molecules can undergo condensation reactions to form resins.
Similarly, resins can convert to asphaltenes, and asphaltenes to coke.
The SARA assays of the thermocracked samples versus the feed,
Table 1, show a decrease in the mass fraction of saturates, aromatics,
and resins with an increase in the asphaltene and toluene insoluble
content. These changes are consistent with a loss of converted small
molecules to the gas phase and a shift towards coke in the asphaltene
fraction. This shift was also observed in the molecular weight and
density distribution of the reacted asphaltenes from these samples [27].
Fig. 6 shows that the atomic H/C ratio of the saturates was virtually
unchanged while it decreased for the aromatic fractions as the extent of
reaction increased, again consistent with a loss of alkyl side chains and/
or cracking of naphthenic rings. There was little change in H/C ratios of
the resin fraction. It appears that the effect of losing side chains is
compensated for by other mechanisms such as the addition of less
aromatic fragments from the cracked asphaltenes and conversion of the
more aromatic resins to the asphaltenes. As expected, the heteroatom
content changed little with thermocracking, Tables 6–8.
5.3.1. Molecular weight
Fig. 7 shows the normalized molecular weights of the saturates,
aromatics, and resins versus conversion. The molecular weights were
normalized to the molecular weight of the given fraction in the feed in
order to compare with the hydrocracked data later on. The molecular
weight of the saturates from the thermocracked oils did not change with
conversion beyond experimental error although it may decrease at
conversions beyond the measured range (> 65% conversion). Pre-
sumably, most of the side chains cracked from the aromatic, resins, and
asphaltene fractions reported to the gas phase and therefore did not
impact the saturate fraction. For cases where the molecular weight of
the thermocracked saturates is unknown, it is recommended to set the
saturate molecular weight equal to the saturate molecular weight in the
feed.
The molecular weight of the aromatics and resins decreased with
conversion, consistent with the loss of side chains and/or cracking of
naphthenic rings. The normalized molecular weights for the thermo-
cracked and hydrocracked samples were fit simultaneously to obtain
the following correlations:
M 0.8
Aromatics: = 1−0.0037X
MFi (9)
M
Resins: = 1−0.0059X
MFi (10)
where MFi is the molecular weight of the given fraction i in the feed and
X is the conversion as defined in Eq. (1). The average deviation of the
thermally cracked aromatic and resins molecular weights from the
correlation was 6%. Note that the correlation is based on a small dataset
773
Fuel 215 (2018) 766–777
and can only be considered as a preliminary approximation.
5.3.2. Density and molar volume
Table 3 shows that the densities of the saturate fractions from
thermocracked oils were higher than the feed density but did not
change at higher conversion. Density is mainly sensitive to chemical
family rather than molecular weight. For example, the densities of n-
alkanes with molecular weights greater than 170 g/mol do not sig-
nificantly change with increasing carbon number (up to 310 g/mol).
Hence, the chemistry of the cracked products that report to the satu-
rates must differ from the feed saturate chemistry but does not appear
to change with conversion. Unfortunately, there are too few data to
develop a generalized correlation for the reacted saturates densities
based on feed properties. Therefore, in cases where the density of the
thermocracked saturates is unknown, it is recommended to set the
cracked saturate density equal to the saturate density in the feed.
Fig. 8 shows that aromatic and resin densities increases as expected
with the loss of side chains leaving more aromatic and denser frag-
ments. Note, the densities were normalized to the density of the given
fraction in the feed in order to compare with the hydrocracked data
later on. The aromatic and resins normalized molecular weights for the
hydrocracked and thermocracked samples were fit simultaneously to
obtain the following correlation for their density:
ρ
= 1 + 0.049X
ρFi (11)
where ρFi is the molecular weight of the given fraction i in the feed and
X is the conversion as defined in Eq. 1. The average deviation of the
thermally cracked aromatic and resins densities from the correlation,
excluding the outlier, was 6%. The correlation is based on a small da-
taset and can only be considered as a preliminary approximation.
Recall that molar volume is the required input parameter for the
regular solution model. Since the molecular weight and density of the
saturates were found to be constant, a constant molar volume is also
recommended, Fig. 9a. For aromatics and resins, simpler correlations
were found for the density and molecular weight data than were pos-
sible for the molar volume data. Therefore, it is recommended to cal-
culate the normalized molar volumes from the normalized density and
molecular weight correlations as shown in Fig. 9b.
5.3.3. Solubility parameter
The solubility parameters of the reacted thermocracked and hy-
drocracked saturates were 14.5 ± 0.3 MPa0.5, Table 3, with one ex-
ception which is discussed in the hydrocracked sample section. The feed
for the thermcracked samples had negligible saturates and therefore
this solubility parameter arises mainly from the reaction products. The
value is surprisingly low; for example, lower than the solubility para-
meter of n-pentane (14.9 MPa0.5). It appears that the regular solution
2.0
WC-VB-B2
1.8 WC-VB-TC17
WC-VB-TC31*
1.6
WC-VB-TC51
Atomic H/C Ratio
1.4
1.2
1.0
0.6
0.4
0.2
0.0
Saturates Aromatics Resins
Fig. 6. Comparison of atomic H/C ratios for saturates, aromatics, and resins from feed
(WC-VB-B2) and thermocracked products. Note that, after the property measurements,
there were insufficient saturates in the WC-VB-B2 feed to obtain an elemental analysis.
 H.W. Yarranton et al. Fuel 215 (2018) 766–777

1.3 1.2 Fig. 7. Effect of conversion on the normalized mole-

(a) (b) cular weights of: a) saturates; b) aromatics, and re-
1.2 1.1 sins. The diamonds are thermal cracked product from
Normalized Molecular Weight

Normalized Molecular Weight

the WC-VB-B2 feed. The squares are hydrocracked
1.1 products from the WC-SR-A3 feed.
1.0
1.0
0.9
0.9
0.8
0.8
0.7
0.7 TC aromatics
0.6 TC resins
0.6 TC saturates HC aromatics
HC saturates HC resins
0.5 0.5 corr. aromatics
constant value
corr. resins
0.4 0.4
0 20 40 60 80 100 0 20 40 60 80 100
Percent Conversion Percent Conversion

1.06 recommended to use a value of 14.5 MPa0.5 for reacted saturates to

obtain an accurate solubility prediction even though the value is lower
1.04 than expected for this chemical family.
The solubility parameter of the reacted aromatics range from 20.4
1.02 to 21.4 MPa0.5 with an average of 20.6 MPa. The increase in aromaticity
Normalized Density

is consistent with the removal of side chains. However, the increase

does not follow a clear trend with conversion. Therefore, it is re-
1.00
commended to set the solubility parameter of the reacted aromatics to
the feed value. Note, since the resin solubility parameter could not be
0.98 measured, it is recommended to set it equal to the value of the pre-
viously determined lowest asphaltene solubility parameter [27].
0.96
TC aromatics
0.94 TC resins 5.4. In situ converted oils
HC aromatics
HC resins The in situ converted oils are thermocracked fluids where the re-
0.92
correlation action conditions are not as well controlled as a surface process and the
product is mixed with an unknown amount of simultaneously produced
0.90
0 20 40 60 80 100 unreacted reservoir fluid. Hence, the properties of the in situ samples
Percent Conversion are expected to fall between those of the native oil and the thermo-
cracked samples. Since the conversions for each sample are not known,
Fig. 8. Effect of conversion on the normalized densities of aromatics, and resins. The the data are only used to check the consistency of the property changes
diamonds are thermal cracked product from the WC-VB-B2 feed. The squares are hy-
reported for the thermocracked oils.
drocracked products from the WC-SR-A3 feed.
Table 1 shows that the in situ samples have a substantially lower
asphaltene content that the original oil in the reservoir (WC-B-B2). The
model does not accurately capture the solubility of saturate fraction and content of the other fractions is correspondingly higher, although the
the error is compensated by a low solubility parameter. The same issue data are scattered. It appears that the in situ process left a considerable
was previously observed with cyclohexane, suggesting that the saturate fraction of the asphaltenes in the reservoir.
reaction products may contain naphthenic species [34]. It is Fig. 10 shows that, as expected with thermocracking, the molecular

1.2 1.1 Fig. 9. Effect of conversion on the normalized molar

(a) (b) volumes of saturates (a) and aromatics and resins (b).
The diamonds are thermal cracked product from the
1.0
1.1 WC-VB-B2 feed. The squares are hydrocracked pro-
Normalized Molar Volume

Normalized Molar Volume

ducts from the WC-SR-A3 feed. The lines are the

0.9 normalized molar volumes calculated from the nor-
1.0 malized molecular weight and density correlations.

0.8
0.9
0.7

0.8 TC aromatics
0.6 HC aromatics
TC saturates TC resins
0.7 HC resins
HC saturates 0.5
corr. aromatics
constant value
corr. resins
0.6 0.4
0 20 40 60 80 100 0 20 40 60 80 100
Percent Conversion Percent Conversion

774
H.W. Yarranton et al. Fuel 215 (2018) 766–777

1200 aromatics with n-heptane. As expected, the solubility data fall between
WC-B-B2
the trends observed for native and thermally cracked samples. The so-
Molecular Weight, g/mol

1000 WC-B-IS10
lubility parameters of the in situ converted saturates ranged from 15.5
WC-B-IS3
to 16.0 MPa0.5 slightly below the native oil average of 16.1 MPa0.5. The
800 WC-B-IS2
solubility parameters of the in situ converted aromatics were uniformly
20.1 ± 0.1 MPa0.5, identical to the native oil average. The data are
600
closer to the values for the native oils suggesting that the in situ samples
were not significantly cracked even though some asphaltenes were left
400
in the reservoir. All of the trends, except for the saturate densities, are
consistent with mildly thermocracked material.
200

0
Saturates Aromatics Resins 5.5. Hydrocracked oils
Fig. 10. Comparison of molecular weights of saturates, aromatics, and resins from in situ
converted oils. Hydrocracking not only cleaves alkyl side chains but converts aro-
matics to naphthenes and eventually olefins and paraffins. Some of the
converted aromatic molecules can undergo condensation reactions to
1100
WC-B-B2 form resins. Similarly, resins can covert to asphaltenes, and asphaltenes
1050 WC-B-IS10 to coke. Hydrocracking also converts sulfur and nitrogen species into
WC-B-IS3 hydrogen disulfide, ammonia, and water.
Density, kg/m³

1000 WC-B-IS2
950
900
850
800
Saturates Aromatics
Fig. 11. Comparison of densities of saturates, aromatics, and resins from in situ converted
oils at 21 °C and atmospheric pressure.
weights of the in situ converted saturates, aromatics, and resins are all
lower than the corresponding molecular weights in the original oil.
Fig. 11 shows that, also as expected, the density of the in situ converted
aromatics and resins is higher than in the original oil. However, the
density of the saturates is lower than in the original oil, contrary to the
trend observed for the thermocracked samples. The saturate cut, and
particularly its density, may be more sensitive to variations in the
procedures used to separate volatiles from the samples. In this case, the
original oil sample was received as a dead oil with little volatile content
while the in situ converted samples had a high volatile content and were
distilled at the University of Calgary lab.
Fig. 12 shows the asphaltene yield curves from solutions prepared
from the in situ converted saturates with toluene and the in situ
The SARA assays of the hydrocracked samples versus the feed show
an increase in the mass fraction of saturates and aromatics, a decrease
in the resin and asphaltene content, and a small amount of toluene
insolubles, Table 1. The shift in the SARA contents are consistent with
an accumulation of cracked fragments and a small shift towards coke in
the asphaltene fraction. This shift was also observed in the molecular
weight and density distribution of the reacted asphaltenes from these
samples [27]. Note that after a sharp initial decrease in the asphaltene
content at low 524 °C+ residue conversion, the asphaltene content
Resins
increased at higher conversions. In these pilot tests, the catalyst gra-
dually deactivated during the run and, as a result, the asphaltene
conversion decreased even though the 524 °C+ residue conversion re-
mained constant.
Fig. 13 shows that the atomic H/C ratio of the saturates increased
slightly with increasing extent of reaction while it decreased for the
aromatic and resin fractions, consistent with a loss of alkyl side chains
from the aromatic and resin species. The significant decrease in the
aromatic and resin H/C ratios suggest that there was relatively little
hydrogenation of the aromatic species. Another possibility is that the
structures that were saturated and fragmented may report to saturates
or distillables after reaction leaving the aromatics and resins enriched
in aromatic structures. Finally, some of the aromatics and resins may
form coke precursors [19,35]. Coke and coke precursors are high
carbon content material and therefore contribute to a lower H/C ratio
especially for the resins fractions. The increased content of toluene
insolubles in the hydrocracked samples is consistent with some coke
formation.

100 100 Fig. 12. Asphaltene yields at 21 °C from: a) in situ

WC-B-IS10 WC-B-IS10 converted saturates with toluene; b) in situ converted
WC-B-IS3 WC-B-IS3 aromatics with n-heptane. Solid and dotted lines are
WC-B-IS2 WC-B-IS2
the average solubility curves for the native and
model - native model - native
Asphaltene Yield, wt%

80 80 thermally cracked oils, respectively.

Asphaltene Yield, wt%

model - TC model - TC

60 60

40 40

20 20

(a) (b)
0 0
20 40 60 80 100 20 40 60 80 100
Saturate Content, wt% Heptane Content, wt%

775
H.W. Yarranton et al.
2.0 WC-SR-A3
WC-SR-HC54
1.8 WC-SR-HC70
1.6 WC-SR-HC83
Atomic H/C Ratio
WC-SR-HC79C
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
Saturates Aromatics Resins
Fig. 13. Comparison of atomic H/C ratios for saturates, aromatics, and resins from feed
(WC-SR-A3) and hydrocracked products.
Surprisingly, hydrocracking had little effect on the S/C ratio of the
aromatics and resins, Tables 7 and 8. C-S and S-S bonds are cracked
more easily than sulfur in aromatic rings due to their lower bonding
energy [22,35]. It appears that there were relatively few C-S and S-S
bonds in this sample. Most of the sulfur atoms were likely attached to
aromatic rings and did not hydrogenate in this process. Instead, the S/C
ratio increased slightly with conversion as the residual sulfur content
became more concentrated as more hydrocarbon material was con-
verted to gas and left the residue.
The N/C ratios of the hydrocracked products increased relative to
the feed, indicating that the nitrogen species were not hydrogenated
and simply became more concentrated as lighter components were re-
moved during the process. Nitrogen is present in residue as both non-
basic and basic heterocyclic structures. The latter are less active as the
nitrogen atom is embedded in an aromatic ring. The increase in ni-
trogen content suggests that nitrogen is present mostly in a basic form.
Kekalainen et al. also reported that the number of nitrogen species was
almost unaltered after hydroconversion [36].
5.5.1. Molecular weight
Fig. 7a shows that the molecular weight of the hydrocracked satu-
rates is constant until the conversion reaches approximately 65% after
which the molecular weight decreases sharply. The reason for the
change in molecular weight at high conversion is unknown. It is pos-
sible that smaller side chains or bridging chains are cleaved at the
higher conversions bringing down the average molecular weight of the
saturates. There are too few data to justify a correlation. Therefore, in
cases where the saturate molecular weight is unknown it is re-
commended to set it to the molecular weight of the saturates in the
feed.
Fig. 7b shows that the molecular weight of the hydrocracked aro-
matics and resins follows the same trend with conversion as the ther-
mocracked samples, as expected with the loss of side chains. As noted
previously, the thermocracked and hydrocracked data were fit si-
multaneously to obtain Eqs. (9) and (10). Eqs. (9) and (10) are re-
commended in cases where the molecular weights of the hydrocracked
aromatics and resins are unknown.
5.5.2. Density and molar volume
Table 3 shows that the densities of the saturate fractions from hy-
drocracked oils increased slightly with conversion. As with the ther-
mocracked samples, the product saturate density was higher than the
original saturates (although only 9 kg/m3 in this case) and there was
little change in density with additional conversion. Therefore, in cases
where the density of the hydrocracked saturates is unknown, it is re-
commended to set the cracked saturate density equal to the saturate
density in the feed.
Fig. 8 shows that aromatic and resin densities increases as expected
with the loss of side chains leaving more aromatic and denser
776
Fuel 215 (2018) 766–777
fragments. Their densities follow a similar but more scattered trend
with conversion as the thermocracked samples. There is a significant
aromatic outlier at a conversion of 56% with a lower density than the
feed. The reason for the outlier is unknown and it was not included in
the data fitting. The normalized molar volumes for the hydrocracked
aromatics and resins were calculated from the correlations for the
normalized density and molecular weight as discussed for the thermo-
cracked samples and shown in Fig. 9.
5.5.3. Solubility parameter
Like the thermocracked saturates, the solubility parameters of the
reacted saturates were narrowly spread around a value of 14.5 MPa0.5
with one exception, Table 3. The exception was the saturate fraction
from the commercial reactor with a value of 15.0 MPa0.5. The reason for
this difference is unknown. Without any other data to go on, in cases
where the hydrocracked saturate solubility parameter is not known, it is
recommended to set the hydrocracked saturate solubility parameter to
the saturate feed value.
The solubility parameters of the reacted aromatics ranged from 20.4
to 20.6 MPa0.5 with an average of 20.5 MPa. There was no clear trend
with conversion. The increase in solubility parameter expected with a
loss of side chains may have been compensated for by the conversion of
some aromatic rings to cyclic structures and the hydrogenation of some
sulfur and nitrogen species. Naphthenes have lower solubility para-
meters than similar aromatics (e.g. 16.8 MPa MPa0.5 for cyclohexane
versus 18.6 MPa0.5 for benzene). Hence, conversion of aromatics to
naphthenes will reduce the average solubility parameter if these species
remain in the aromatic fraction; for example, if the rest of the molecule
contains unconverted aromatic rings. Heteroatomic species have higher
solubility parameters than similar hydrocarbons (e.g. 22.2 MPa0.5 for
nitrobenzene versus 18.3 MPa0.5 for toluene). Hence, the hydrogenation
of heteroatoms will rend to reduce the average solubility parameter.
Recall that there was little evidence for significant hydrogenation of
sulfur or nitrogen so this effect is likely small for this sample. For the
example in this study, the trends counterbalance each other and there is
little change in the aromatic solubility parameters with conversion.
Therefore, it is recommended to set the solubility parameter of the
reacted aromatics to the feed value. As before, the resin solubility
parameter is set equal to the value of the previously determined lowest
asphaltene solubility parameter [27].
6. Conclusions
In Part I of this series of papers, the molecular weight, density, and
solubility parameter distributions of the asphaltenes from a wide range
of native and reacted petroleum samples were determined. In this
contribution, the same properties were established for the SAR frac-
tions.
The molecular weight, density, and solubility parameters of each
fraction from native oils and vacuum residues varied relatively little
from source to source. Therefore, the average values for each property
are recommended for use when data are not available.
For reacted saturates, it is recommended to set the density and
molecular weight equal to the feedstock values. If the feedstock prop-
erties are not available, the average values determined from native oils
and vacuum residues can be used. The recommended value for the
solubility parameter is 13.8 MPa0.5.
For reacted aromatics and resins, the density and molecular weight
were correlated to the feedstock value and the conversion. It is re-
commended to set the aromatic solubility parameter to the feedstock
value. The resin solubility parameter could not be measured and was
instead set equal to the previously determined lowest asphaltene solu-
bility parameter.
The average values and correlations for the SAR fractions and the
previously determined asphaltene properties are sufficient to char-
acterize the distillation residue of an oil for phase stability modeling.
H.W. Yarranton et al.
The final step will be to characterize the distillable fraction and de-
monstrate that the combined distillate + SARA characterization can be
used to predict the stability of both native and reacted fluids.
Acknowledgement
The authors thank Shell Global Solutions for financial support and
Zhongxin Huo for his input.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in the
online version, at http://dx.doi.org/10.1016/j.fuel.2017.11.071.
References
[1] Leontaritis KJ. Asphaltene deposition: a comprehensive description of problem
manifestations and modeling approaches. In SPE Prod. Operations Symp.,
Oklahoma City; 1989. p. SPE-18892-MS.
[2] Hammami A, Phelps CH, Monger-McClure T, Little TM. Asphaltene precipitation
from live oils: an experimental investigation of onset conditions and reversibility.
Energy Fuels 2000;14(1):14–8.
[3] Anderson SI. Flocculation onset titration of petroleum asphaltenes. Energy Fuels
1999;13(2):315–22.
[4] Wiehe IA, Kennedy RJ. The oil compatibility model and crude oil incompatibility.
Energy Fuels 2000;14(1):56–9.
[5] Macchietto S, Hewitt GF, Coletti F, Crittenden BD, Dugwell DR, Galindo A, et al.
Fouling in crude oil preheat trains: a systematic solution to an old problem. Heat
Transfer Eng 2011;32(3):197–215.
[6] Wattana P, Wojciechowski DJ, Bolanos G, Fogler HS. Study of asphaltene pre-
cipitation using Refractive Index measurement. Pet Sci Technol
2003;21(3–4):591–613.
[7] Hirschberg A, deJong LNJ, Schipper BA, Meijer JG. Influence of temperature and
pressure on asphaltene flocculation. Soc Petrol Eng J 1984;24(3):283–93.
[8] Anderson SI, Speight JG. Thermodynamic models for asphaltene solubility and
precipitation. J Petrol Sci Eng 1999;22(1–3):53–66.
[9] Yarranton HW, Masliyah JH. Molar mass distribution and solubility modeling of
asphaltenes. AIChE J 1996;42(12):3533–43.
[10] Kawanaka S, Park SJ, Mansoorl GA. Organic deposition from reservoir fluids: a
thermodynamic predictive technique. SPE Reservoir Eng 1991;6(2):185–92.
[11] Wang JX, Buckley JS. A two-component solubility model of the onset of asphaltene
flocculation in crude oils. Energy Fuels 2001;15(5):1004–12.
[12] Alboudwarej H, Akbarzadeh K, Beck J, Svrcek WY, Yarranton HW. Regular Solution
Model of asphaltene precipitation from bitumen. AIChE J 2003;49(11):2948–56.
[13] Akbarzadeh K, Alboudwarej H, Svrcek W, Yarranton HW. A generalized regular
solution model for asphaltene precipitation from n-alkane diluted heavy oils and
bitumens. Fluid Phase Equilib 2005;232:159–70.
[14] Tharanivasan AK, Yarranton HW, Taylor SD. Application of regular solution based
models to asphaltene precipitation from live oils. Energy Fuels 2011;25(2):528–38.
[15] Rogel E. Theoretical approach to the stability of visbroken residues. Energy Fuels
777
Fuel 215 (2018) 766–777
1998;12(5):875–80.
[16] Joshi JB, Pandit AB, Kataria KL, Kulkarni RP, Sawarkar AN, Tandon D, et al.
Petroleum residue upgradation via visbreaking: a review. Ind Eng Chem Res
2008;47(23):8960–88.
[17] Hauser A, AlHumaidan FS, Al-Rabiah H. NMR investigations on products from
thermal decomposition of Kuwaiti vacuum residues. Fuel 2013;113:506–15.
[18] Quignard A, Kressman S, Heavy crude oils: from geology to upgrading: an overview.
Chapter 16: Visbreaking, A. Huc; 2010. Editions TECHNIP, Paris.
[19] Xu Z, van den Berg FGA, Sun X, Xu C, Zhao S. Detailed characterization of virgin
heavy oil resid and its thermally cracked resid. Energy Fuels 2014;28(3):1664–73.
[20] Schabron JF, Pauli AT, Rovani Jr. JF. Molecular weight polarity map for residua
pyrolysis. Fuel 2001;80(4):529–37.
[21] Lababidi HMS, Sabti HM, AlHumaidan FS. Changes in asphaltenes during thermal
cracking of residual oils. Fuel 2014;114(A):59–67.
[22] Zhao S, Kotlyar LS, Woods JR, Sparks BD, Hardacre K, Chung KH. Molecular
transformation of Athabasca bitumen end-cuts during coking and hydrocracking.
Fuel 2001;80(8):1155–63.
[23] Wiehe IA. Process chemistry of petroleum macromolecules. Florida, U.S.A.: CRC
Press; 2008.
[24] Gray MR. Upgrading oil sands bitumen and heavy oil. Edmonton, Canada:
University of Alberta; 2015.
[25] Buch L, Groenzin H, Buenrostro-Gonzalez E, Anderson SI, Lira-Galeana C, Mullins
OC. Molecular size of asphaltene fractions obtained from residuum hydrotreatment.
Fuel 2003;82(9):1075–84.
[26] Groenzin H, Mullins OC. Asphaltene molecular size and weight by time-resolved
fluorescence depolarization. In: Mullins OC, editor. Asphaltenes, heavy oils, and
petroleomics. New York, U.S.A.: Springer; 2007.
[27] Powers DP, Sadeghi H, Yarranton HW, van den Berg FGA. Regular solution based
approach to modeling asphaltene precipitation from native and reacted oils: part 1,
molecular weight, density, and solubility parameter distributions of asphaltenes.
Fuel 2016;178:218–33.
[28] Bruno TJ, Ott LS, Lovestead TM, Huber ML. Relating complex fluid composition and
thermophysical properties with the advanced distillation curve approach. Chem
Eng Technol 2010;33:363–76.
[29] Ortiz DP, Satyro MA, Yarranton HW. Thermodynamics and fluid characterization
using trajectory optimization. Fluid Phase Equilib 2013;351:34–42.
[30] Peramanu S, Pruden BB. Molecular weight and specific gravity distributions for
Athabasca and Cold Lake bitumens and their saturate, aromatic, resin, and as-
phaltene fractions. Ind Eng Chem Res 1999;38(8):3121–30.
[31] Yarranton HW, Okafor JC, Ortiz DP, van den Berg FGA. Density and refractive index
of petroleum, cuts, and mixtures. Energy Fuels 2015;29(9):5723–36.
[32] Rijkers MPW, Heidemann RA. Convergence behavior of single-stage flash calcula-
tions. In: Chao KC, Robinson RL, editors. Equation of state: theories and applica-
tions. U.S.A.: American Chemical Society; 1986.
[33] Yarranton HW, Fox WA, Svrcek WY. Effect of resins on asphaltene self-association
and solubility. The Canadian Journal of Chemical Engineering 2007;85(5):635–42.
[34] Mannistu KD, Yarranton HW, Masliyah JH. Solubility modeling of asphaltenes in
organic solvents. Energy Fuels 1997;11(3):615–22.
[35] Raseev S. Thermal and catalytic processes in petroleum refining. New York, U.S.A.:
Marcel Dekker; 2003.
[36] Kekäläinen T, Pakarinen JMH, Wickström K, Lobodin VV, McKenna AM, Jänis J.
Compositional analysis of oil residues by ultra high-resolution Fourier Transform
ion cyclotron resonance mass spectrometry. Energy Fuels 2013;27(4):2002–9.