Journal of Petroleum Science and Engineering 190 (2020) 106782

Contents lists available at ScienceDirect

Journal of Petroleum Science and Engineering

journal homepage: http://www.elsevier.com/locate/petrol

New correlations and deposition envelopes for predicting asphaltene

stability in crude oils
Aliyu Adebayo Sulaimon a, *, John Kevin Mendoza De Castro b, Suyash Vatsa c
a
Petroleum Engineering Department, Universiti Teknologi PETRONAS, Malaysia
b
Petroleum Engineering, Batangas State University, Philippines
c
Indian Institute of Technology (Indian School of Mines), Dhanbad, India

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

Keywords: The precipitation and deposition of asphaltene are among the most pressing issues in the petroleum industry.
Asphaltene stability This flow assurance issue may cause formation damage in porous media, the plugging and coking of wellbore and
SARA analysis production pipelines and difficulties in the refining process. Such events lead to additional operating costs for the
Colloidal instability index
company. Therefore, knowing the conditions at which asphaltene may precipitate and can be stable may enable
Asphaltene/resin ratio
its management. Existing models of asphaltene stability are dependent on saturate, aromatic, resin and
asphaltene (SARA) analysis which can be very expensive and time-consuming. Therefore, using regression
analysis as well as MATLAB curve fitting and optimization procedures, a new set of correlations and asphaltene
stability envelops (ASEs) based on the newly developed Density-Based Asphaltene/Resin ratio (DBAR), Density-
Based Saturate/Aromatic ratio (DBSAr) and the Density-Based Colloidal Instability Index (DBCII) all as functions
of oil density have been developed to consistently predict asphaltene stability. The new criteria were developed
for the correlations by considering the boundary between stable and unstable regions in their respective plots.
The new envelopes showed 83% reliability by accurately predicting asphaltene stability in 24 (eight stable and
16 unstable samples) out of 29 (13 stable and 16 unstable) crude oil samples. The correlation coefficient (R2),
average absolute relative error (AARE) and the root-mean-square error (RMSE) for the new DBAR correlation are
0.9456, 0.8357 and 0.1219 respectively. The DBSAr correlation resulted in R2 of 0.9437, AARE of 0.1860 and
RMSE of 0.2450. For the DBCII, the statistical indicator results were 0.9639 for the R2, 0.1376 for the RMSE and
0.1038 for the AARE.

1. Introduction significant reduction of formation productivity/injectivity, formation

damage in porous media, the plugging and coking of wellbore and
1.1. Asphaltene precipitation and its challenges production pipelines and difficulties in the refining process. Such oc­
currences tend to lead to additional operating costs for the company
Asphaltene is the fraction of crude oil precipitating in light alkanes, (Leontaritis, 1995; Akbarzadeh et al., 2007; Hammami and Ratulowski,
such as pentane, hexane or heptane, belonging to saturate fractions (Fan 2007).
et al., 2002; Creek et al., 2009; Dufour et al., 2010). This precipitate is Crude oils contain asphaltenes and resins in the disperse phase,
soluble in aromatic solvents, such as toluene and benzene, and contains constituting a colloidal system. The flocculation of asphaltenes depends
the largest percentage of heteroatoms and organometallic constituents on the colloidal stability of these systems and varies due to the different
(Mansoori, 1994). The precipitation of asphaltene can block the flow factors that influence the stability of crude oils (Ahmadi et al., 2017;
path for oil, leading to low productivity. Asphaltene precipitation is Wang and Buckley, 2003). Composition only plays the main role. Pre­
considered one of the most pertinent issues in the petroleum industry vious researches have indicated that the nature and content of asphal­
since it causes many problems due to the complex composition of tenes and the dispersion medium are the main factors determining the
asphaltene and many parameters affecting its precipitation (Alimo­ relative stability of crude oils and related materials (Leon et al., 2000;
hammadi et al., 2019). If not properly addressed, it may result in a Ahmadi et al., 2017). Although the exact nature of the original state of

* Corresponding author.
E-mail address: aliyu.adebayor@utp.edu.my (A.A. Sulaimon).

https://doi.org/10.1016/j.petrol.2019.106782
Received 28 August 2019; Received in revised form 8 November 2019; Accepted 3 December 2019
Available online 10 December 2019
0920-4105/© 2019 Elsevier B.V. All rights reserved.
A.A. Sulaimon et al. Journal of Petroleum Science and Engineering 190 (2020) 106782

archipelago-type asphaltenes could potentially reduce the deposition

tendency for island asphaltene dominated oil.
According to de Boer et al. (1995), the tendency of asphaltene to
flocculate is dependent on the extent to which the crude oil is saturated
with asphaltenes, its saturation with gas under the downhole condition
and the density of crude oil in reservoir conditions (Wattana and Fogler,
2005). The problems in asphaltene did not primarily result from large
amounts of asphaltene in the oil but from high saturate fractions
(Ashoori et al., 2017). According to Eutamene et al. (2014), the ten­
dency of asphaltenes to precipitate and deposit can be caused by changes
in pressure, temperature and composition.
Monitoring asphaltene stability in crude oil systems is challenging
and has formed the subject of many studies (Rogel et al., 2001; Jama­
luddin et al., 2001; Rogel et al., 2001; Carrier et al., 2000; Wang et al.,
2006; Vargas et al., 2014; Dufour et al., 2010; Moura et al., 2010; Rogel
et al., 2015; Likhatsky and Syunyaev, 2010; Gonzalez et al., 2012;
Sulaimon and Govindasamy, 2015; Punase et al., 2016; Guzman et al.,
2017; Ashoori et al., 2017; Prakoso et al., 2018; Hemmati-Sarapardeh
et al., 2018). According to Hammami et al. (2000), asphaltene stability
Fig. 1. Deposit of asphaltene in the tubing. depends on several factors, including the composition of the fluids
(Source: Goual, 2012) involved in the reservoir and their pressure and temperature. The sol­
ubility of the asphaltene fraction is highly dependent on the composition
equilibrium of asphaltenes in petroleum fluids is still under investiga­ of the crude, less affected by pressure and indirectly influenced by
tion, one characteristic is the tendency of asphaltenes to form aggregates temperature (Buenrostro-Gonzalez et al., 2004). Thus, monitoring or
in hydrocarbon solutions. These aggregates are called micelles (Man­ predicting the stability of asphaltene in crude oil is required to minimise
soori, 1997; Mansoori et al., 2001). Micelles and the hydrocarbon me­ its effects.
dium together form a colloidal system. One commonly held view is that
the colloids are stabilised by resins adsorbed on their surfaces and the 1.2. Oil composition and asphaltene stability
dispersion of colloids in the fluid form a two-phase system. Many re­
searches have been conducted on asphaltene deposition; however, the In the production of oil and gas from the subsurface, fluid properties
main cause is yet to be fully understood. Fig. 1 depicts the asphaltene and composition are considered to properly characterise the fluid and
deposition in tubing. predict its phase behaviour under certain conditions. Petroleum engi­
Chaisoontornyotin et al. (2016) studied the asphaltene aggregation neers and analysts are responsible for monitoring the production phase
and deposition under various precipitants and different concentrations. and ensuring that asphaltene precipitation is avoided; however, the
They concluded that for a fixed volume fraction of precipitant, the prediction of asphaltene stability is difficult due to the many factors
collision efficiency decreases with increasing precipitant carbon num­ affecting its physicochemical behaviour. Crude oil composition is among
ber, thereby leading to a reduced rate of aggregation. Also, their findings the most important factors that must be considered and monitored to
revealed that the differences in collision efficiency for bulk aggregation mitigate oil recovery problems and flow assurance issues. Saturate, ar­
do not influence deposition behaviour. F� avero et al. (2016) investigated omatic, resin and asphaltene (SARA) analysis is a method of expressing
the dependency of the asphaltene deposition rate on the concentration the composition of crude oil according to its polarizability and polarity
of unstable asphaltenes and on fluid flow velocity. They found out that (Fan and Buckley, 2002; Wattana and Fogler, 2005; Likhatsky and
the asphaltene deposition rate increases linearly with the concentration Syunyaev, 2010); Punase et al. (2016); Guzman et al. (2017); Ashoori
of unstable asphaltenes and that the asphaltene deposition process can et al., (2017). SARA analysis began with the work of Jewell et al. (1972)
be explained by a diffusion-limited deposition of nanoparticles. A study and it is a useful screening technique in the production phase of the oil
of the dynamics of asphaltene deposition in porous media using and gas industry since it can be used to characterise crude oil, specif­
microfluidic devices was conducted by Lin et al. (2016). They demon­ ically its asphaltene component (Alboudwarej et al., 2002); (Zen­
strated that the deposition dynamics varied significantly with different dehboudi et al., 2014).
concentrations of heptane and that once the deposit was formed, the The saturate fraction consists of nonpolar materials, including linear,
deposits tended to grow against the flow direction. branched and cyclic saturated hydrocarbons. Aromatics, which contain
Recently, the chemical structure of asphaltenes as subfractions has one or more aromatic rings, are more polarisable. The remaining two
been investigated by various groups. Discussion of asphaltene structure fractions, resins and asphaltenes, have polar substituents (Allen and
has shifted from molecular weight to asphaltene architecture. The island Roberts, 1989). The distinction between the two is that asphaltenes are
model portrays asphaltenes as solitary core aromatic molecules with insoluble in an excess of heptane (or pentane), whereas resins are
peripheral alkyl side chains, while the least assented archipelago model, miscible with heptane (or pentane). Resins are structurally similar to
incorporates multiple aromatic cores that are alkyl-bridged with mul­ asphaltenes but have smaller molecular weights (<1000 g/mol). The
tiple polar functionalities. Using the Fourier transform ion cyclotron proportions of each of the SARA fractions in crude oil are related to the
resonance mass spectrometry, Chacon-Patin et al. (2017) showed that stability of asphaltenes in that oil. Resins have a strong tendency to
island and archipelago motifs coexist in petroleum asphaltenes and that associate with asphaltenes. Such association largely determines their
asphaltene purification is required to detect and characterise archipel­ solubility in crude oil wherein resin molecules react to the addition of
ago species by mass spectrometry. Lin et al. (2019) investigated the light paraffin components by desorbing from asphaltenes to re-establish
deposition dynamics of various asphaltene subfractions in a porous thermodynamic equilibrium, thus increasing the probability of asphal­
media microfluidic model. They concluded that the deposition rate is tene self-aggregation (Chamkalani, 2012).
Consistent with the amount of precipitated asphaltenes and that One of the requirements in achieving the goals of a trouble-free pe­
island-type asphaltenes could be the inhibitors or stabilizers for archi­ troleum production operation is the prediction of asphaltene precipita­
pelago asphaltene dominated oils. However, increased content of tion potential of crude oil at the early stage of reservoir exploration by
proactively monitoring asphaltene stability. However, the use of SARA

2
A.A. Sulaimon et al. Journal of Petroleum Science and Engineering 190 (2020) 106782

1.3. Asphaltene stability parameters and screening criteria

Alkafeef, Al-Medhadi, and AL-Shammari (2003) defined the stability

of a colloid dispersion as its resistance to flocculation, and its degree of
‘resistance’ is used as a measure of the dispersion stability. They noted
that the destabilisation (i.e. flocculation) of colloidal asphaltenes in
oil-production flowing systems depends principally on breaking the
balances of attraction forces between the absorbed resin molecules and
asphaltene particles. Therefore, asphaltene stability is a function of the
ratio of asphaltenes to stabilizing factors in the crude such as aromatics
and resins. In the following sections, we review some of the existing
techniques that have been developed to assess the potential for asphal­
tene precipitation from crude oils.
De Boer et al. (1995) proposed a plot that can be used as a first
screening tool to identify the potential for the oil to exhibit solid for­
mation problems in terms of the difference between the reservoir pres­
sure and bubble point pressure against the fluid density (Fig. 2). The
Fig. 2. Simplified De Boer plot. upper left region of the plot which is mostly characterized by low density
and high-pressure difference is the unstable region while the right
portion describing denser fluid and low-pressure difference is regarded
as the unstable region.

1.3.1. Asphaltene stability index

Considering the peptizing activity of resin on asphaltene, the
asphaltene/resins (A/R) ratio has been used to evaluate the stability of
asphaltene in crude oils. Jamaluddin et al. (1996) evaluated the
asphaltene-solvating power of various nonconventional solvents as well
as deasphalted oil. The gas oils produced by vacuum distillation and
having a saturate content below 35 wt% were found to be desirable
solvents for asphaltene dissolution. Thus, due to its native resin and
aromatic contents, the deasphalted oil is a strong asphaltene solvent.
Therefore, the higher the A/R ratio, the more likely will the asphaltene
precipitation occur. Specifically, if the A/R ratio is less than 0.35, the
crude oil is considered stable (Jamaluddin et al., 2001; Mousavi-Deh­
ghani et al., 2004).

1.3.2. Colloidal instability index (CII)

Fig. 3. Stankiewicz et al. cross plot. The colloidal instability index (CII) is another screening criterion
(Watkinson, 2005; Yen et al., 2001; Sulaimon and Govindasamy, 2015)
analysis may be a disadvantage largely due to the extensive time which can be used to identify crude oil systems with problems of
required and its cost implications. Furthermore, the accuracy of the asphaltene deposition. The CII uses SARA fractions as inputs for deter­
SARA analysis depends on the adopted technique of measurement: mining crude oil stability and it is expressed as the ratio of the sum of
latroscan TLC-FID (IP-469), IP-143, IP-368, and ASTM D-2007 (Rogel asphaltenes and saturates to the sum of aromatics and resins:
et al., 2001; Fan et al., 2002; Wang and Buckley, 2003; ASTM, 2011). ðAsphaltenes in wt%Þ þ ðSaturates in wt%Þ
The time constraint can be a factor that would hinder early mitigation of CII ¼ (1)
ðResins in wt%Þ þ ðAromatics in wt%Þ
asphaltene precipitation. Hence, early prediction of asphaltene stability
by a cost-effective method can be beneficial for efficient and The lower the CII, the higher is the asphaltene stability. However,
problem-free production. Fan, Wang and Buckley (2002) investigated such stability is considered relative when evaluating whether a certain
the irreproducibility of similar SARA analysis by three different crude oil is more stable than another. If the CII �0.9, the asphaltene
analytical techniques: ASTM column (gravity-driven chromatographic) fraction tends to be unstable, but when CII �0.7, asphaltene stability is
separation, thin-layer chromatography (TLC), and high-performance certain. However, for saturates-rich oils, the asphaltene stability largely
liquid chromatography (HPLC). They concluded that a considerable depends on solubility or CII while the stability is dependent on polarity
number of volatile components is lost in the TLC analysis while results for degraded oils (Hascakir, 2017).
from the HPLC compared favourably well with the ASTM analysis.
In this study, we sought to develop an understanding of asphaltene 1.3.3. Colloidal stability index (CSI)
stability as a function of an easily measurable property of the oil such as Likhatsky and Syunyaev (2010) presented a new colloidal stability
crude oil density. We have developed correlations to eliminate the cost index (CSI) for crude oils based on the composition of the crude oil and
and time constraints inherent in SARA analysis that can hinder the early the polarity of its components. The CSI which considers the quantitative
prediction of asphaltene precipitation. We have also validated these orientation interactions between the crude oil components is expressed
correlations using published data from experiments and real case studies as:
from field observation. These correlations will be beneficial to the oil
ðεasph ÞðAsphaltene in wt%Þ þ ðεsat ÞðSaturate in wt%Þ
and gas industry, allowing cost-effective, simple but effective prevention CSI ¼ (2)
ðεres ÞðResins in wt%Þ þ ðεarom ÞðAromatics in wt%Þ
and mitigation of asphaltene precipitation.
where ϵasph, ϵsat, ϵres, and ϵarom, are the dielectric constant of asphaltenes,
saturates, resins and aromatics, respectively. Specifically, for the

3
A.A. Sulaimon et al. Journal of Petroleum Science and Engineering 190 (2020) 106782

Fig. 4. R/A ratio for the Brazilian crude oil (Guzman et al., 2017).

Fig. 5. Stability crossplot (Guzman et al., 2017). It is claimed that the SCP offers a better confidence level of 92% compared with Stankiewicz plot of 86% and 72%
for the CII.

4
A.A. Sulaimon et al. Journal of Petroleum Science and Engineering 190 (2020) 106782

Table 1 method. Therefore, it is necessary to see the behaviour of each of the

Data sources. samples in the four graphs to obtain a single result that characterizes the
Author/s Title of the Paper No. Of oil state of the asphaltenes. Using the STA GRAPHIC software, the bound­
samples aries of stable and unstable regions can be determined as shown in
Fan et al. (2002). Evaluating Crude Oils by SARA Analysis 6 Fig. 5.
Guzman et al. (2017). Methods of Determining Asphaltene 29
Stability in Crude Oils 1.3.7. Heithaus parameter (P)
Moradi et al. (2012). Investigation of Asphaltene Precipitation 3 A method developed by Heithaus (1962) predicts the stability of
in Miscible Gas Injection Processes:
Experimental Study and Modelling
crude oils and asphalt by titration (Guzman et al., 2017). In his exper­
Zendehboudi et al. Asphaltene Precipitation and Deposition 2 iment, a solution of crude oil or bitumen in an aromatic solvent, such as
(2014). in Oil Reservoirs–Technical aspect, toluene or α-methyl naphthalene, is titrated with a paraffinic solvent
experimental and hybrid neural network (n-heptane, isooctane or n-hexadecane) until asphaltene precipitation
predictive tools
occurs. The Heithaus parameter (P) which represents the stability of
Buenrostro-Gonzalez Asphaltene Precipitation in Crude Oils: 2
et al., 2004 Theory and Experiments bitumen is expressed as:
Wang and Buckley Asphaltene Stability in Crude Oil and 11
P0
(2003). Aromatic Solvents–The Influence of Oil P¼ (3)
Composition ð1 Pa Þ
Hassanvand et al. (2012) Study of Temperature Effect on 2
Asphaltene Precipitation by Visual and where Po is the solubility of maltenes, and Pa is the solubility of
Quantitative Methods asphaltene in bitumen. When P < 1, bitumen or crude oil are unstable
Jokuty et al. (1995) Hydrocarbon Groups and Their 61
(low colloidal stability), and the sample is prone to asphaltene precipi­
Relationship to Oil Properties and
Behaviour
tation. For higher values of P, bitumen will be more stable, and there will
Creek et al. (2009) Verification of Asphaltene-Instability- 1 be increased asphaltene peptization, as described by Guzman et al.
Trend Predictions for Low-Molecular- (2017).
Weight Alkanes
Jamaluddin et al. An Investigation of Asphaltene Instability 1
1.3.8. Toluene equivalence (TE)
(2002). Under Nitrogen Injection
Toluene equivalence (TE) is the solubilisation capacity and is
TOTAL 118
required to keep asphaltenes dissolved in petroleum. Anderson and
Reynolds (1989) developed rapid simple tests which utilities can use to
unstable crude oils, εasph ¼ 18.4 and εres ¼ 3.8 while εasph ¼ 5.5 and εres ¼ anticipate and prevent problems from instability/incompatibility. TE is
4.7 for stable crude oils. Generally, εsat ¼ 1.921 and εarom ¼ 2.379 for all calculated (Moura et al., 2010) as
crude oils. Moreover, if the CSI >0.95, the crude oil is unstable and VT
susceptible to asphaltenes precipitation; otherwise, the oil is stable with TE ¼ � 100% (4)
VT þ VH
no risk of asphaltene precipitation.
where VT is the toluene volume in the mixture (solvent), and VH is the n-
1.3.4. Stankiewicz plot heptane volume (titrant). TE ranges from zero to 100; the lower the
Stankiewicz et al. (2002) developed a cross plot of two ratios: (Sat­ value, the higher the stability.
urates/Aromatics) versus (Asphaltenes/Resins) to determine the ten­ Irrespective of their relative performance over one another, the
dency of asphaltene precipitation. This plot provides a quick evaluation heavy reliance of all the current stability parameters and criteria on
of the potential asphaltene precipitation by separating the stable and SARA analysis coupled with the associated experimental cost and the
unstable regions. Two regions are marked as stable and unstable zones. time constraints constitute the inevitable downsides of these methods.
Fig. 3 illustrates the cross plot which shows that the crude oil is stable Therefore, some new, timely, robust and cost-effective techniques for
when A/R ratio is less than 0.5 irrespective of the S/Ar ratio. determining asphaltene stability and/or conditions are required and this
is the main objective of this study.
1.3.5. Qualitative-quantitative analysis (QQA)
Using extensive SARA analysis data from 72 samples of pure oils with 2. Methodology
or without asphaltene deposition from Brazilian, Venezuelan, Colum­
bian and Mexican crudes, Gaona et al. (2010) developed two techniques 2.1. Data gathering
based on relationships between physicochemical parameters. The first
technique is the qualitative and quantitative analysis (QQA) which We collected data on fluid properties and SARA fractions for 118
� � � � � � � � � � � �
crude oil samples as shown in Table 1. The compositional data quality
involved the following six ratios: AR ; AS ; AAr ; R*Ar
S*A ;
S*Ar Ar
A ; and S*A . was evaluated and the sum of the fractions for each sample equate unity
They divided the relationship into three zones: the stable part, which is while the density values were converted at 15 � C for uniformity. Sub­
assigned a numerical value of 10, the metastable or uncertain zone a sequently, the data were used to develop correlations by using regres­
value of 5, and the third zone indicating instability a value of one. Fig. 4 sion analysis and the MATLAB Curve Fitting Toolbox. Fig. 6 illustrates
shows the quantitative analysis for the Brazilian crude oils. the flowchart of the plan of study. Another unique field data sets that are
strictly based on direct asphaltene deposition experiences from 29 oil­
1.3.6. Stability Cross Plot (SCP) fields (Rogel et al., 2001; Rogel and Carbognani, 2003; Wattana and
Gaona et al. (2010) also built a series of graphs based on the physi­ Fogler, 2005; Moura et al., 2010) were used for the development of new
cochemical analysis of the SARA fractions and the QQA method stability plots and validation purposes.
described in section 1.3.6. Using the same procedure, the curves SCP-1 The developed correlations from both tools were compared using the
� � � �
statistical parameters such as correlation coefficient (R2), root mean
which is a plot of R*A Ar
S*A vs A , SCP-2 which is a graph of
r
square error (RMSE) and the average absolute relative error (AARE) to
� � � � � � � � � � � �
R*Ar R S R R*Ar Ar
achieve the best correlation to represent the relationship between SARA
S*A vs A , SCP-3 of Ar vs A ; and SCP-4 for S*A vs S*A equally fractions and density.
demonstrate relationships that constitute the Stability Cross Plot (SCP)

5
A.A. Sulaimon et al. Journal of Petroleum Science and Engineering 190 (2020) 106782

Fig. 6. Research paradigm of the study.

2.2. Crude oil density 2.4. Data refinement

The oil density data used for the correlation development were ob­ Data refinement is the process of managing different data in the same
tained from various sources (Table 1) and at different temperatures. condition to increase data understanding, remove variability and
Therefore, it is imperative to determine the equivalent values of the redundancy and develop a concrete, integrated data resource. It was
densities at a unique temperature for consistency. Therefore, the density necessary to refine the data due to noise in order to obtain better ac­
is corrected and adjusted at a reference temperature of 15 � C (approx. curacy and reliability. The erratic data points contributed a high vari­
60 � F) by using a simple curve-fit equation for the density-temperature ance to the correlation. A more reliable and defined relationship
data from API Publication 421 (Wright, 1994): between the dependent and independent variables were subsequently
� � achieved after the refinement.
SG15� C ¼ SGT þ 5:93x10 4 ½Tð� CÞ 15� C� (5)

where SG is the specific gravity of the oil sample. 2.5. Data correlation using MATLAB (curve fitting toolbox)

2.3. Correlation development Data from regression analysis were imported into the workspace of
the Curve Fitting Toolbox application in MATLAB. In the regression
Three existing asphaltene stability models were considered in this analysis, a scatter plot was generated, with the only difference being the
study. These are stability index (A/R ratio), A/R vs S/Ar plots and the necessary commands typed into the command window. The command
CII. The values of each model were calculated and correlated with crude ‘cftool’ creates an interface where 2D or 3D graphs can be plotted. In this
oil density. Using regression analysis, exponential, linear, logarithmic or work, 2D graphs (x and y variables) were chosen since there is only one
polynomial functions were selected to establish befitting relationships variable for both x and y. Available options for the equation, having best
between the stability parameters and density. The curve fitting pro­ fitting relationships are, power, polynomial, linear fitting, interpolant,
cedure in MATLAB was also adopted to establish correlations between rational, smoothing spline, the sum of sine and Weibull. On the left side
the stability criteria and density. Furthermore, to improve on the ac­ of the plot, the coefficients of the equation are shown with their corre­
curacy and reliability of the correlations, the stability parameters were sponding R2 and RMSE. These values were recorded and used to predict
also correlated with the square function of the density. This was to the values of each of the stability models.
determine the best possible correlation and whether this could improve Data refinement in the Curve Fitting toolbox was conducted using
the outcome. the ‘exclude outliers’ function, which removes any data in the dataset
that are at variance with the well-established trend. As data are being
removed, the R2 and RMSE automatically adjust, inevitably improving
the correlation. Determining the coordinates of the removed data points

6
A.A. Sulaimon et al. Journal of Petroleum Science and Engineering 190 (2020) 106782

Table 2
SARA fractions of 29 crude samples with their stability based on operational experience.
Crude Sample Saturates wt% Aromatic wt% Resins wt% Asphaltenes wt% API gravity Stability

A 32.3 42.2 19.8 5.8 19.8 Stable

B 37.9 38.3 14.9 8.9 20.0 Stable
LM1 32.3 42.2 19.8 5.8 18.3 Stable
Mars-Pink 38.4 29.8 25.8 4.8 16.5 Stable
P01 48.2 35.2 16 0.6 18.8 Stable
S4 49.41 38.23 10.44 1.92 27.2 Stable
S20 32.45 41.5 21.12 4.93 16.3 Stable
S22 54.73 30.41 12.78 2.08 25.1 Stable
S23 51.79 29.93 15.83 2.46 26.7 Stable
S24S 49.53 41.33 2.54 6.15 26.1 Stable
S27 42.94 36.35 12.96 7.74 24.1 Stable
S277 26.13 45.3 22.57 6.01 14.0 Stable
S29 39.62 38.71 16.83 4.93 18.3 Stable
D 36.9 37.9 19.4 5.8 24.6 Unstable
LM2 37.9 38.3 14.9 8.9 24.1 Unstable
Louisiana 63.5 28.2 7.2 1.1 31.0 Unstable
NM1 36.9 37.9 19.4 5.8 22.4 Unstable
NM2 52.4 36.3 10.5 0.8 28.0 Unstable
P20 80.8 12.4 6.8 0.5 40.3 Unstable
Reservoir A 57.4 30.8 10.4 1.4 32.0 Unstable
U100 54.54 35.35 8.23 1.87 25.0 Unstable
U103 55.41 36.77 6.89 0.94 26.3 Unstable
U16 45.31 33.29 17.55 3.85 22.2 Unstable
U16L 49.68 37.38 9.28 3.34 26.8 Unstable
U2 43.21 35.3 16.68 4.99 22.4 Unstable
U91 48.61 34.35 11.65 5.38 22.8 Unstable
U94 49.43 37.62 8.61 4.43 25.7 Unstable
U98 48.83 37.4 9.51 4.26 25.6 Unstable
West Texas 52.4 42.5 4.8 0.3 39.6 Unstable

Table 3
a Refined correlations of A/R vs density using regression analysis. B Refined
correlations of A/R vs density2 using regression analysis.
Equation Correlation Equation R2 AARE RMSE
number

6 AR=
¼ 0:0051e3:9868ρ 0.0611 0.7288 0.5116
7 AR=
¼ 4:5042ρ 3:7493 0.1667 1.6288 0.4771
8 AR=
¼ 3:8036lnðρÞ þ 0:7083 0.1468 1.5933 0.4828
9 AR=
¼ 95:218 ρ2 167:58ρ þ 0.5878 1.5191 0.3356
73:775
10 AR=
¼ 949:43ρ3 2522:1ρ2 þ 0.8958 0.9479 0.1692
2229:2ρ 655:36
11 AR=
¼ 5371:3ρ4 18830ρ3 þ 0.9418 2.4296 0.2832
24704ρ2 14375ρ þ 3130:7
12 AR=
¼ 32677ρ5 144084ρ4 þ 0.9472 2.6682 0.3149
253905ρ3 223526ρ2 þ 98311ρ
17282
13 AR=
¼ 0:2663ρ3:4415 0.0562 0.7284 0.5140
14 AR=
¼ 0:0285e2:301ρ
2
0.0666 0.7302 0.5085
15 AR=
¼ 2:6524ρ2 1:8477 0.1892 1.6624 0.4706
16 AR=
¼ 1:9018lnðρ2 Þ þ 0:7083 0.1468 1.5933 0.4828
17 AR ¼ 28:965 ρ4 45:297ρ2 þ 0.6515 1.4335 0.3085
Fig. 7. Regression Plot for Unrefined A/R vs Density.
=

17:755
18 AR=
¼ 140:93ρ6 335:63ρ4 þ 0.9171 0.8538 0.1505
in the plot was achieved using the ‘data cursor’. After refining and 264:55ρ2 68:808
choosing the best correlation, the correlation between each asphaltene 19 AR=
¼ 387:56ρ8 1197:4ρ6 þ 0.9446 0.91 0.1246
stability model and the square of the density was also determined. 1375:5ρ4 695:93ρ2 þ 131
Comparisons between the chosen correlations using the density and 20 AR=
¼ 1250:2ρ10 4930:8ρ8 þ 0.9475 1.0332 0.1272
square of the density in terms of R2, RMSE and AARE were conducted to 7762:7ρ6 6099:2ρ4 þ 2392:4ρ2
achieve the best possible outcome. 374:69
21 AR=
¼ 0:2663ρ2ð1:7208Þ 0.0562 0.7284 0.5140
3. Results and discussion

3.1. Data gathering evaluated during data gathering to ensure the validity of all data.
Compositional and fluid properties data from a total of 118 oil samples
Data from field observations and previous studies were used and were used for the correlation development while data sets based on real

7
A.A. Sulaimon et al. Journal of Petroleum Science and Engineering 190 (2020) 106782

correlation between the A/R ratio and the oil density. Correlating den­
sity with asphaltene stability parameters and the square of density with
the asphaltene stability parameters was undertaken. Unrefined data
were imported in both tools, and the correlations were recorded
together with their corresponding R2 and RMSE. Fig. 7 shows the un­
refined scatter plot of the chosen correlation for A/R.
Data refinement was carried out to remove noise to facilitate reliable
analysis. Tables 3a and b shows the correlations from normal regression
for ‘R/A and density (ρ)’, as well as ‘R/A and square of density (ρ2)’. The
highlighted correlations were selected based on the values of the sta­
tistical indicators R2, AARE and RMSE. Predicted values for the best-fit
and weak functions were considered as evident in the R2 values for the
linear and logarithmic functions. The results show that Eqs (10) and (19)
that are both polynomial functions, gave better statistical indicators
than other functions. Specifically, the values of R2, AARE and RMSE for
the density correlation given by Eq. (10) are 0.8958, 0.9479, and 0.1692
respectively. For the squared density correlation (Eq. 19), these values
are respectively 0.9446, 0.91 and 0.1246. Although Eq 19 which is an
eight-degree polynomial gives the best statistical parameters, we chose
Eq (10) as a good correlation for R/A for its simplicity. Fig. 8 illustrates
the scatter plot when the square of the density is used as the independent
Fig. 8. Refined scatter plot for A/R vs density.2.
variable.
Using MATLAB, the linear and logarithmic functions are excluded
Table 4
from further consideration due to their weak correlation coefficients.
a Refined correlations of A/R vs Density using the MATLAB Curve Fitting Tool. B Tables 4a and b shows the performance of different functions generated
Refined correlations of A/R vs Density2 using the MATLAB Curve Fitting Tool. from MATLAB curve fitting toolbox.
Analysis reveals that both the density and squared-density approach
Equation Correlation Equation R2 AARE RMSE
number yielded the same power function correlation with a high R2 of 0.9456,
minimum AARE of 0.8357 and RMSE of 0.1219. Comparatively, the
22 AR ¼ 4:504ρ 3:749 0.1667 1.6290 0.4771
MATLAB curve fitting tool generated Eq. 28 or Eq. 35 is more accurate
=

23 AR ¼ 95:22 ρ2 167:6ρ þ 0.5878 1.5136 0.3357

than those obtained from the normal regression analysis due to the
=

73:78
ability of MATLAB to optimise data. Correlating either A/R with the
24 AR ¼ 949:4ρ3 2522ρ2 þ 0.8958 0.8778 0.2245
density (ρ) or squared-density (ρ2) does not affect the results for the
=

2229ρ 655:4
25 AR 0.9418 53.6610 7.3081
power function since the same data were used; the only difference is in
¼ 5371ρ4 18830ρ3 þ
the correlation between A/R and ρ2, which affects only power but not
=

24700ρ2 14380ρ þ 3131

26 AR 0.9472 75.2626 10.0035
the base of the function.
=
¼ 32680ρ5 144100ρ4 þ
253900ρ3 223500ρ2 þ Analysis shows that all polynomial function’s results for both the
98310ρ 17280 normal regression analysis and MATLAB curve fitting are almost the
27 AR
=
¼ 0:9809ρ26:76 0.8448 0.9390 0.2059 same; however, for the power function, the case is different. This is
28 AR
=
¼ 0:2329ρ50:29 þ 0:2025 0.9456 0.8357 0.1219 because MATLAB performs a linear fit on the log-log transform of the
29 AR
=
¼ 2:652ρ2 1:848 0.1892 1.6607 0.4706 data which optimizes the results compared to regression analysis. This is
30 AR
=
¼ 28:97 ρ4 45:3ρ2 þ 0.6515 1.4326 0.3086
evident in the R2 results of best-fit power function for both tools used
17:75 which are far more different from each other. Based on actual field
31 AR
=
¼ 140:9ρ6 335:6ρ4 þ 0.9171 1.0630 0.1562 observation of asphaltene deposition and by using the method of clas­
264:6ρ2 68:81 sification, we have developed a new stability envelop for A/R ratio as
32 AR
=
¼ 387:6ρ8 1197ρ6 þ 0.9446 0.5919 0.1344 shown in Fig. 9. The area located within the right portion of the
1375ρ4 695:9ρ2 þ 131 boundary line denotes the stable region while the left side represents the
33 AR
=
¼ 1250ρ10 4931ρ8 þ 0.9475 0.6940 0.2031 unstable region. In the correlation developed by Fan et al. (2002), the
7763ρ6 6099ρ4 þ 2392ρ2 term for asphaltenes and resins is in the form of axb with a and b having
374:7 low values. From this analysis, we have shown that both resin and
34 AR
=
¼ 0:9809ρ2ð13:38Þ 0.8448 0.9390 0.2059 asphaltene fractions contribute to the relationship of density with SARA
35 AR
=
¼ 0:2329ρ2ð25:15Þ þ 0:2025 0.9456 0.8357 0.1219 fractions, but not to the high degree of saturates and aromatics.
By comparing the predicted A/R values obtained from equation 28 or
equation 35 with Asomaning and Watkinson (2000) criterion, analysis
field observations from 29 unique crude oils (Rogel et al., 2001; Rogel shows that their criterion for the classification of stable and unstable
and Carbognani, 2003; Wattana and Fogler, 2005; Moura et al., 2010), crude oils based on a ceiling value of 0.35 would not be applicable in this
were used to develop new asphaltene stability envelops (ASEs). case. A scatter plot of the predicted density-based A/R ratio (DBRA) and
Asphaltene stability was determined according to operational experi­ density for the 29-validation data generates a clear boundary that can be
ence and experimental observations. Table 2 shows the SARA fractions, used as the criterion to categorise stable and unstable crudes. The new
densities at 15 � C and the status of the asphaltenes based on operational correlation that defines this asphaltene stability envelope is thus given
experience. as

DBAR ¼ 2 � 10 9 e19:48ρ (6)

3.2. Correlation for the asphaltene-resin ratio
where DBAR ¼ density-based asphaltene/resin ratio and ρ ¼ oil density.
Using regression analysis and the curve fitting toolbox in MATLAB,
As a result, by using the new correlation for DBAR, it was possible to
118 datasets (SARA fractions and density) were used to develop the

8
A.A. Sulaimon et al. Journal of Petroleum Science and Engineering 190 (2020) 106782

Fig. 9. Asphaltene stability plot for density-based Asphaltene-Resin ratio (DBAR).

Fig. 10. Regression plot for unrefined S/Ar versus density.

satisfactorily predict the asphaltene precipitation tendency of 24 (eight S/Ar to separate stable and unstable regions. To present the same plot
stable and 16 unstable) crudes from the 29 validating oil samples that without the need for SARA analysis, S/Ar is also correlated with density
composed of 13 stable and 16 unstable crude oils. This represents and squared density. Using S/Ar as the dependent variable, a similar
approximately 83% success rate. Crude oils S4, S22, S23, S24S and S27 procedure to the A/R correlation is adopted. Fig. 10 shows the unrefined
are in the unstable region, although they have been categorised as stable scatter plot of the chosen correlation for S/Ar vs density.
based on operational experience. In this work, we have opted not to Similarly, the data is refined by removing noise until a better, reli­
define any section of the envelope as a metastable or uncertain region able and uniform trend is achieved. Tables 5a and 5b shows the corre­
similar to the De Boer plot (De Boer et al., 1995). This is because we do lations obtained from the normal regression analysis using the refined
not ascribe any comparative benefit to uncertainty over instability in the data while Tables 6a and 6b shows the corresponding equations from the
unstable region regarding decision making on effective flow assurance curve fitting toolbox in MATLAB. The polynomial equations 42 and 51
management with respect to asphaltene deposition. are the mathematical expressions with the best statistical indicators.
However, equation 51 is selected for its lower errors, RMSE of 0.2314,
AARE of 0.1817, and a higher correlation coefficient of 0.9501.
3.3. Correlation for saturate-aromatic ratio (S/Ar) The results from the MATLAB toolbox after refinement show that the
power function equation 59 or equation (6) best fits the plotted points
Stankiewicz et al. (2002) developed a scattered plot of A/R against

9
A.A. Sulaimon et al. Journal of Petroleum Science and Engineering 190 (2020) 106782

Table 5 with minimum RMSE of 0.245 and AARE of 0.1860. The refined scatter
a Refined correlations of S/Ar vs density using Regression analysis. B Refined plot for the S/Ar ratio and the squared density is shown in Fig. 11. When
correlations of S/Ar vs density2 using Regression analysis. the performance of the correlations from the toolbox is compared with
Equation Correlation Equation R2 AARE RMSE the selected equation from the normal regression, equation 51 yielded
number better statistical performance indicators. However, it appears that
37 s= ¼ 21865e 10:68ρ 0.8286 0.1752 0.4284 equation 51 could have resulted from overfitting of the data, a charac­
Ar
38 s= ¼ 17:724ρ þ 17:677 0.6127 0.3735 0.6423 teristic observation for the squared density correlations. Therefore,
Ar
39 s= ¼ 16:63 lnðρÞ 0:0385 0.6459 0.3640 0.6142 equation 59 is adopted as the new correlation for S/Ar ratio as a function
Ar
40 s= ¼ 135:55 ρ2 266:56ρ þ 0.8426 0.2552 0.4096 of density. Fig. 11 shows the refined scatter plot of the chosen correla­
Ar tion for S/Ar. Using the 29-real data from field observation, we devel­
131:59
41 0.9367 0.2258 0.2626
oped a new asphaltene stability envelope based on the S/Ar vs A/R
s= ¼ 994:35ρ3 þ 2890:2ρ2
Ar technique proposed by Gaona et al. (2010). However, both ratios are
2801:7ρ þ 906:71
density-based and are therefore denoted respectively as DBSAr and
42 s= ¼ 5071:8ρ4 19717ρ3 þ 0.949 0.1820 0.2346
Ar DBAR (Fig. 12). The new correlation is given by equation (7):
28727ρ2 18598ρ þ 4516:8
43 s= ¼ 25435ρ5 111976ρ4 þ 0.9502 0.3211 0.4300 DBSAr ¼ 605770ðDBARÞ5 þ 615105ðDBARÞ4
Ar
195192ρ3 168072ρ2 þ 71283ρ 248067ðDBARÞ3 þ 49656ðDBARÞ2 4934:3ðDBARÞ þ 195:76
11862
(7)
44 s= ¼ 0:5141ρ 9:805 0.8358 0.1687 0.3749
Ar
45 S= ¼ 158:93e 5:78ρ
2
0.8188 0.1802 0.4828 where DB denotes density-based parameter.
Ar
46 S= ¼
Ar
2
9:3794ρ þ 9:3305 0.5793 0.3818 0.6694 Analysis shows that the DBSAr vs DBRA correlation also correctly
47 S= ¼
Ar
8:316 lnðρ2 Þ 0:0385 0.6459 0.3641 0.6142 predicted 24 (eight stable and 16 unstable) of the 29 crude oils’
48 S= ¼ 39:615 ρ4 77:062ρ2 þ 0.8119 0.2717 0.4476 asphaltene stability status which is similar to the performance of the
Ar
37:955 DBAR correlation (equation (6)). Crude oils S4, S22, S23, S24S, and S27
49 S= ¼ 175:12ρ6 þ 497:2ρ4 0.9264 0.2238 0.2799 that were operationally confirmed stable were predicted unstable by the
Ar
2
470:27ρ þ 149:09 new correlation.
50 S= ¼ 622:86ρ8 2338:9ρ6 þ 0.95 0.1895 0.2354
Ar
3281:4ρ4 2042:9ρ2 þ 478:09 3.4. Correlation for colloidal instability index (CII)
51 S= ¼ 416:98ρ 10
1175:7ρ þ 8 0.9501 0.1817 0.2314
Ar CII values were calculated using SARA fractions and plotted against
734:09ρ6 681:78ρ4 þ 954:18ρ2
297:52 density and square density in a scatter plot to find the correlation that
52 S= ¼ 0:5141ρ2ð 4:903Þ 0.8358 0.1687 0.3747 best fits the plotted data points. The stability criterion was developed by
Ar normal regression analysis and the MATLAB curve fitting toolbox.
Fig. 13 shows the scatter plot of CII vs density using the unrefined data.
Similar plots were generated for the CII with the square of density to
Table 6 consider more scattered data points and to determine whether doing so
a Refined correlations of S/Ar vs density using the curve fitting toolbox.. B
would improve the results of the regression (Fig. 14). Tables 7a and b
Refined correlations of S/Ar vs density2 using the curve fitting toolbox.
shows the performance analysis of various mathematical expressions
Equation Correlation Equation R2 AARE RMSE considered in developing reliable correlations for the CII as a function of
number
density and squared density of crude oil respectively. Of the 14 equa­
53 s= ¼
Ar
17:72ρ þ 17:68 0.6127 0.3742 0.6423 tions in both tables, the quadratic expressions given by eqs 71 and 80
54 s= ¼ 135:5 ρ2 266:6ρ þ
Ar
0.8426 0.2552 0.4161 yielded the most acceptable statistical evaluating parameters. Never­
131:6 theless, for its simplicity, equation 71 with the R2 of 0.9639, AARE of
55 s= ¼ 994:4ρ3 þ 2890ρ2 0.9367 0.3953 0.5229 0.1038 and RMSE value of 0.1376 is selected.
Ar
2802ρ þ 906:7 The MATLAB curve fitting toolbox was equally utilised to construct
56 s= ¼ 5072ρ4 19720ρ3 þ 0.949 1.2090 1.2980 an appropriate mathematical relationship between the CII and the
Ar
28730ρ2 18600ρ þ 4517 square of oil density. Tables 8a and 8b give the performance analysis of
57 s= ¼ 25440ρ5 112000ρ4 þ 0.9502 159,768 156,214 the correlations based on density and squared density respectively.
Ar
Obviously, the fourth-order polynomial eq. 92 represents the best fit
195200ρ3 168100ρ2 þ
71280ρ 11860 correlation with a high R2, lowest AARE and RMSE of 0.9636, 0.1049
58 s= ¼ 0:4056ρ 11:9 0.9247 0.1776 0.2832 and 0.1382 respectively. Comparatively, eq 71 is chosen in preference to
Ar
59 s= ¼ 0:1717ρ 15:44 þ 0:5051 0.9437 0.1860 0.245 eq 92 based on its simplicity and better performance attributes.
Ar Asomaning and Watkinson (2000) stability condition states that
60 S= ¼ 9:379ρ2 þ 9:331 0.5793 0.3819 0.6694
Ar stable crudes have CII values less than 0.7 while unstable crudes have
61 S= ¼ 39:61 ρ4 77:06ρ2 þ 0.8119 0.2719 0.4476
Ar values greater than 0.9. However, analyses have shown that this con­
37:96
dition is not always true since all the calculated CII are greater than 0.9
62 S= ¼ 175:1ρ6 þ 497:2ρ4 0.9264 0.2232 0.2799
Ar which is indicative of unstable crude oils. However, eight oils are known
470:3ρ2 þ 149:1
to be stable while the remaining oils are unstable. Therefore, we
63 S= ¼ 622:9ρ8 2339ρ6 þ 0.95 0.3238 0.4285
Ar developed a new asphaltene stability plot with two regions which can be
3281ρ4 2043ρ2 þ 478:1 used to differentiate between stable and unstable crude oils. Fig. 15
64 S= ¼ 417ρ10 1176ρ8 þ
Ar
0.9501 0.1910 0.2783 shows that seven oils are stable while the remaining nine are in the
734:1ρ6 681:8ρ4 þ 954:2ρ2 unstable region. Thus, as observed for the R/A correlation as well as
297:5 S/Ar vs A/R technique, the CII correlation has also classified oils S4, S22,
65 s= ¼ 0:4056ρ
Ar
11:9 0.9247 0.1776 0.2832 S23, S24S, and S27 as unstable, the outcomes that are different from the
66 s= ¼ 0:1717ρ
Ar
15:44
þ 0:5051 0.9437 0.1860 0.245 field observation. Based on field observation and the method of classi­
fication, we developed Eq (8) which defines the boundary between the

10
A.A. Sulaimon et al. Journal of Petroleum Science and Engineering 190 (2020) 106782

Fig. 11. Refined scatter plot for S/Ar vs squared density.

Fig. 12. Asphaltene stability plot for density-based S/Ar and A/R technique.

two regions shown in Fig. 15.

DBCII ¼ C1 � ρC2 (11)
DBCII ¼ 2:25ρ 9:3
(8)
where the coefficients are as provided in Table 9.

3.5. Summary of results

3.6. Comparative analysis with other methods
In summary, the new asphaltene stability models that are correlated
with density can be written in general terms as shown in equations (9)– Table 10 shows the comparison of the results from the new density-
(11): based asphaltene stability envelop (DBASE) with predictions from the
existing CII, CSI, SP, QQA, and SCP. Analysis shows that the simple
DBAR ¼ C1 � e C2 � ρ (9) DBASE provided similar performance when compared to the CSI and the
SCP. The three techniques accurately predicted asphaltene stability for
DBSAr ¼ C1 ðDBARÞ5 þ C2 ðDBARÞ4 þ C3 ðDBARÞ3 þ C4 ðDBARÞ2 þ C5 ðDBARÞ 11 of the 16 crude oil samples albeit with varied results. However, the
þ C6 CSI requires the complete SARA fractions and the dielectric constants
(10) data while the SCP method also strictly rely on full SARA analysis as well
as four stability plots for the SCP criteria. The results also show that the

11
A.A. Sulaimon et al. Journal of Petroleum Science and Engineering 190 (2020) 106782

Table 7
a: Correlations of CII with density using regression analysis. B Correlations of CII
with density2 using regression analysis.
Equation Correlation Equation R2 AARE RMSE
number

68 CII ¼ 11620e 9:993ρ 0.9266 0.1085 0.1455

69 CII ¼ 14:365ρ þ 14:551 0.9029 0.1646 0.2256
70 CII ¼ 13:11 lnðρÞ þ 0:2244 0.9133 0.1565 0.2131
71 CII ¼ 88:793 ρ2 175:88ρ þ 0.9639 0.1038 0.1376
87:792
72 CII ¼ 36:285ρ3 þ 187:63ρ2 0.9639 0.1037 0.1380
265:46ρ þ 114:82
73 CII ¼ 6293:9ρ4 þ 22788ρ3 0.9648 0.3451 0.4180
30812ρ2 þ 18423ρ 4105
74 CII ¼ 171889ρ5 þ 773767ρ4 0.9661 92,297 95,140
1 � 106 ρ3 þ 1 � 106 ρ2 þ 561616ρ
100799
75 CII ¼ 0:5475ρ 9:083 0.9291 0.1069 0.1424
76 CII ¼ 123:85e 5:482ρ2 0.9232 0.1109 0.1519
77 CII ¼ 7:8455ρ2 þ 7:994 0.8916 0.1738 0.2384

Fig. 13. Scatter plot for unrefined CII vs density. 78 CII ¼ 6:557 lnðρ2 Þ þ 0:2244 0.9133 0.1565 0.2131
79 CII ¼ 29:29 ρ4 56:591ρ2 þ 0.9636 0.1049 0.1382
28:052
80 CII ¼ 26:529ρ6 þ 95:389ρ4 0.964 0.1033 0.1374
111:13ρ2 þ 42:953
81 CII ¼ 619:16ρ8 þ 2022:7ρ6 0.965 0.1042 0.1398
2435:5ρ4 þ 1271:2ρ2 238:72
82 CII ¼ 7850:7ρ10 þ 31975ρ8 0.966 0.1257 0.1536
51910ρ6 þ 42021ρ4 16983ρ2 þ
2748
83 CII ¼ 0:5868ρ2ð 4:528Þ 0.7136 0.1069 0.1425

Table 8
a Refined correlations of CII vs density using the MATLAB curve fitting toolbox.
B Refined correlations of CII vs density2 using the MATLAB curve fitting toolbox.
Equation Correlation Equation R2 AARE RMSE
number

84 CII ¼ 14:36ρ þ 14:55 0.9029 0.1650 0.2256

85 CII ¼ 88:79 ρ2 175:9 þ 87:79 0.9639 0.1051 0.1400
86 CII ¼ 36:29ρ3 þ 187:6ρ2 0.9639 0.1108 0.1566
265:5ρ þ 114:8
87 CII ¼ 6294ρ4 þ 22790ρ3 0.9648 0.2194 0.2754
30810ρ2 þ 18420ρ 4105
88 CII ¼ 171900ρ5 þ 773800ρ4 0.9661 81.5082 87.8936
Fig. 14. Refined scatter plot for CII vs squared density.
1392000ρ3 þ 1251000ρ2
561600ρ 100800
new DBASE and the CSI are capable of reliably identifying all the un­ 89 CII ¼ 0:5472ρ 9:154 0.9617 0.1070 0.1416
stable crude oils, a feat which the SCP method appears not to reproduce. 90 CII ¼ 0:5196ρ 9:377
þ 0:03614 0.9618 0.1067 0.1418
Therefore, the new DBASE is the simplest, quickest and cheapest
91 CII ¼ 7:845ρ2 þ 7:994 0.8916 0.1738 0.2383
approach for identifying and evaluating the asphaltene deposition ten­
92 CII ¼ 29:29 ρ4 56:59ρ2 þ 0.9636 0.1049 0.1382
dency without SARA analysis. 28:05
93 CII ¼ 26:53ρ6 þ 95:39ρ4 0.964 0.1054 0.1395
4. Conclusions 111:1ρ2 þ 42:95
94 CII ¼ 619:2ρ8 þ 2023ρ6 0.965 0.2679 0.3291
The existing correlations for determining asphaltene precipitation 2436ρ4 þ 1271ρ2 238:7
tendency rely exclusively on SARA analysis which can be time- 95 CII ¼ 7851ρ10 þ 31970ρ8 0.9660 1.2099 1.3125
consuming and cost ineffective. In this study, we have developed a 51910ρ6 þ 42020ρ4 16980ρ2
new set of correlations with data from 118 crude oil samples. New 2748
mathematical expressions that relate the stability parameters such as the 96 CII ¼ 0:5472ρ 9:154 0.9617 0.1070 0.1416
density-based asphaltene/resin (DBAR) ratio, density-based saturates/ 97 CII ¼ 0:5196ρ 9:377
þ 0:03614 0.9618 0.1067 0.1418
aromatic (DBSAr) ratio, and density-based colloidal instability index
(DBCII) as functions of density have been developed. Crude oil density
can be readily measured without the long duration or time constraints
and high cost associated with the SARA fractions-dependent techniques.
The generated correlations have been utilised to develop a new set of

12
A.A. Sulaimon et al. Journal of Petroleum Science and Engineering 190 (2020) 106782

Fig. 15. Asphaltene stability plot for CII.

Table 9
Coefficients of the new asphaltene stability correlations.
Stability parameter Coefficients

C1 C2 C3 C4 C5 C6
9
DBAR 2 � 10 19.48 – – – –
DBSAr 605770 615,105 248067 49,656 4934.3 195.76
DBCII 2.25 9.3 – – – –

Acknowledgement
Table 10
Comparison of asphaltene stability prediction techniques. The authors acknowledge the Malaysian Ministry of Higher Educa­
SAMPLE CII CSI SP QQA SCP ASE ACTUAL tion (MOHE) for providing the Fundamental Research Grant Scheme
S4 0 0 0 1 1 0 Stable
(FRGS) with Cost Centre 015MA0-082 which was partly utilized to fund
S20 1 1 0 0 1 1 Stable this research. We also wish to express our deep appreciations to the
S22 0 0 0 1 1 0 Stable management of the Universiti Teknologi PETRONAS (UTP) in Malaysia,
S23 0 0 0 1 1 0 Stable Batangas State University in the Philippines, and the Indian Institute of
S24S 0 0 0 0 0 0 Stable
Technology (Indian School of Mines) India for supporting the research
S27 0 0 0 0 0 0 Stable
S277 1 1 0 0 1 1 Stable collaboration and providing the requisite human resources for the
S29 0 1 0 0 1 1 Stable project. We also acknowledge useful discussions and contributions from
U100 1 1 1 0 0 1 Unstable Dr Jalal Foroozesh, Dr Mohammed Abdalla Ayoub Mohammed, and Dr
U103 1 1 0 0 0 1 Unstable Berihun Mamo Negash, all from the Department of Petroleum Engi­
U16 1 1 1 0 0 1 Unstable
U16L 1 1 1 0 1 1 Unstable
neering, UTP Malaysia.
U2 1 1 1 0 1 1 Unstable
U91 1 1 1 1 1 1 Unstable References
U94 1 1 1 1 1 1 Unstable
U98 1 1 1 0 1 1 Unstable Ahmadi, S., Amiribakhtiar, M.S., Gholami, A., Bahrami, N., 2017. Upgrading fuzzy logic
by GA-PS to determine asphaltene stability in crude oil. Egypt. Petrol. J. 26,
‘1’ denotes correct prediction while ‘0’ denotes an incorrect prediction. 505–510.
Akbarzadeh, K., Hammami, A., Kharrat, A., Zhang, D., Allenson, S., Creek, J., Kabir, S.,
Jamaluddin, A., Marshall, A.G., Rodgers, R.P., Mullins, O.C., Solbakken, T., 2007.
density-based asphaltene stability envelops (DBASEs) that can be used to
“Asphaltenes – problematic but rich in potential”, schlumberger oilfield review.
evaluate asphaltene deposition tendency with appreciable reliability. Summer 22–43. https://pdfs.semanticscholar.org/660d/16ee85f906111cefeb85
The new models can provide a dependable, quick and accurate evalua­ e7c57f59f6d95aee.pdf?_ga¼2.258948170.1196265177.1565075005-1548042456
tion of asphaltene precipitation potential without SARA analysis. .1565075005.
Alboudwarej, H., Beck, J., Svrcek, W.Y., Yarranton, H.W., 2002. Sensitivity of asphaltene
properties to separation techniques. Energy Fuels 16, 462–469.
Declaration of competing interest Alimohammadi, S., Zendehboudi, S., James, L., Amin, J.S., 2019, June 3. Investigation of
asphaltene precipitation; an experimental and CPA EOS approaches. In: Conference
and Exhibition, Production and Management EOR II. https://doi.org/10.3997/2214-
The authors wish to declare no interest. 4609.201901314.

13
A.A. Sulaimon et al. Journal of Petroleum Science and Engineering 190 (2020) 106782

Alkafeef, S.F., Al-Medhadi, F., AL-Shammari, A.D., 2003, January 1. Asphaltene Jamaluddin, A.K.M., Nazarko, T.W., Sills, S., Fuhr, B.J., 1996. August 1). Deasphalted Oil
Remedial Technology Using Advanced Deasphalted Oil. Society of Petroleum - A Natural Asphaltene Solvent. Society of Petroleum Engineers. https://doi.org/
Engineers. https://doi.org/10.2118/81570-MS. 10.2118/28994-PA.
Allen, T.O., Roberts, A.P., 1989. Production Operations: Paraffins and Asphaltenes, vol. Jamaluddin, A.K.M., Creek, J., Kabir, C.S., McFadden, J.D., D’Cruz, D., Manakalathil, J.,
2. Oil & Gas Consultants International Inc., pp. 1–16 Joshi, N., Ross, B., 2001. January 1). A Comparison of Various Laboratory
Anderson, R.P., Reynolds, J.W., 1989. Methods for Assessing the Stability and Techniques to Measure Thermodynamic Asphaltene Instability. Society of Petroleum
Compatibility of Residual Fuel Oils. National Institute for Petroleum and Energy Engineers. https://doi.org/10.2118/72154-MS. SPE 72154.
Research. Project 2527-1; Technical report published by the Electric Power Research Jamaluddin, A.K.M., Joshi, N., Iwere, F., Gurpinar, O., 2002, January 1. An Investigation
Institute (EPRI-GS-6570), United States. of Asphaltene Instability under Nitrogen Injection. Society of Petroleum Engineers.
Ashoori, S., Sharifi, M., Masoumi, M., Salehi, M.M., 2017. The relationship between https://doi.org/10.2118/74393-MS.
SARA fractions and crude oil stability. Egypt. Petrol. J. 26, 209. Jewell, D.M., Weber, J.H., Bunger, J.W., Plancher, H., Latham, D.R., 1972. Ion-exchange,
Asomaning, S., Watkinson, A.P., 2000. Petroleum stability and heteroatom species effects coordination, and adsorption chromatographic separation of heavy-end petroleum
in fouling of heat exchangers by asphaltenes. Heat Transf. Eng. 21 (3), 10–16. distillates. Anal. Chem. 44 (8), 1391–1395.
https://doi.org/10.1080/014576300270852. Jokuty, P., Whiticar, S., Fingas, M., Meyer, E., Knobel, C., 1995. “Hydrocarbon Groups
ASTM, 2011. D2007-11: Standard Test Method for Characteristic Groups in Rubber and Their Relationships to Oil Properties and Behaviour”, 18th Arctic and Marine Oil
Extender and Processing Oils and Other Petroleum-Derived Oils by the Clay-Gel Spill Program Technical Seminar; 1995; Edmonton; Canada. https://www.boem.go
Absorption Chromatographic Method. v/uploadedFiles/BOEM/Oil_and_Gas_Energy_Program/Plans/jokuty_et_al_1995.pdf.
Buenrostro-Gonzalez, E., Lira-Galeana, C., Gil-Villegas, A., Wu, J., 2004. Asphaltene Leon, O., Rogel, E., Espidel, J., Torres, G., 2000. Asphaltenes: structural characterization,
Precipitation in Crude Oils: Theory and Experiments. AIChE Journal 50 (10), self-association, and stability behaviour. Energy Fuels 14, 6–10.
2552–2570. https://doi.org/10.1002/aic.10243. Leontaritis, K.J., 1995. The asphaltene and wax deposition envelopes. The symposium on
Carrier, H., Plantier, F., Daridon, J.-L., Lagourette, B., Lu, Z., 2000. Acoustic method for thermodynamics of heavy oils and asphaltenes, area 16C of fuels and petrochemical
measuring asphaltene flocculation in crude oils. J. Pet. Sci. Eng. 27 (3), 111–117. division. In: AIChE Spring National Meeting and Petroleum Exposition, Houston,
Chacon-Patin, M.L., Rowland, S.M., Rodgers, R.P., 2017. Advances in asphaltene Texas, March 19-23.
petroleomics. Part 1: asphaltenes are composed of abundant island and archipelago Likhatsky, V.V., Syunyaev, R.Z., 2010. New colloidal stability index for crude oils based
structural motifs. Energy Fuels 31, 13509–13518. https://doi.org/10.1021/acs. on polarity of crude oil components. Energy Fuels 24, 6483–6488. https://doi.org/
energyfuels.7b02873, 2017. 10.1021/ef101033p, 2010.
Chaisoontornyotin, W., Haji-Akbari, N., Fogler, H.S., Hoepfner, M.P., 2016. Combined Lin, Y.-J., He, P., Tavakkoli, M., Mathew, N.T., Fatt, Y.Y., Chai, J.C., Goharzadeh, A.,
asphaltene aggregation and deposition investigation. Energy Fuels 30, 1979–1986. Vargas, F.M., Biswal, S.L., 2016. Examining asphaltene solubility on deposition in
https://doi.org/10.1021/acs.energyfuels.5b02427. model porous media. Langmuir 32, 8729–8734. https://doi.org/10.1021/acs.
Chamkalani, A., 2012. Correlations between SARA fractions, Density, and RI to langmuir.6b02376.
Investigate the stability of Asphaltene. ISRN Anal. Chem. 2012, 6. https://doi.org/ Lin, Y.-J., Cao, Y., Chacon-Patin, M.L., Rowland, S.M., Rodgers, R.P., Yen, A., Biswal, S.
10.5402/2012/219276. Article ID 219276. L., 2019. Microfluidic study of the deposition dynamics of asphaltene subfractions
Creek, J.L., Wang, J., Buckley, J.S., 2009, May 1. Verification of Asphaltene-Instability- enriched with island and archipelago motifs. Energy Fuels 33 (3), 1882–1891.
Trend (ASIST) Predictions for Low-Molecular-Weight Alkanes. Society of Petroleum https://doi.org/10.1021/acs.energyfuels.8b03835, 2019.
Engineers. https://doi.org/10.2118/125203-PA. Mansoori, G.A., 1994. Modelling and prevention of asphaltene and other heavy organic
De Boer, R.B., Leerlooyer, K., Eigner, M.R.P., van Bergen, A.R.D., 1995, February 1. deposition in oil wells. Argentina vol. 27, 29. https://doi.org/10.2118/27070-MS.
Screening of Crude Oils for Asphalt Precipitation: Theory, Practice, and the Selection April. 1325-1335.
of Inhibitors. Society of Petroleum Engineers. https://doi.org/10.2118/24987-PA. Mansoori, G.A., 1997. Asphaltene deposition and its control (the role of asphaltene in
Dufour, J., Calles, J.A., Marug�an, J., Gim� enez-Aguirre, R., Pe~na, J.L., Merino-García, D., petroleum flow arterial blockage). In: Proceedings of Controlling Hydrates, Waxes
2010. Influence of hydrocarbon distribution in crude oil and residues on asphaltene and Asphaltenes in Oil Production, Transportation and Processing IBC Conference
stability. Energy Fuels 24 (4), 2281–2286. https://doi.org/10.1021/ef900934t, and Workshop, pp. 28–30. April 1997. https://trl.lab.uic.edu/1.OnlineMaterials/HO
2010. https://pubs.acs.org/doi/pdf/10.1021/ef900934t?rand¼4xdr0m4c. D/97.ASPHALTENE%20DEPOSITION%20AND%20ITS%20CONTROL.pdf.
Eutamene, M., Mouloudj, N., Meniai, A.H., Djema, Hakim, Khodja, M., 2014. Asphaltene Mansoori, G.A., Branco, V.A.M., Xavier, L.C. De A., Park, S.J., Manafi, H., 2001.
precipitation in Algerian Oilfields: experimental investigation and field mitigation. Asphaltene flocculation and collapse from petroleum fluids. J. Pet. Sci. Eng. 32
Chem. Eng. Trans. 39, 1663–1668. (2001), 217–230. http://citeseerx.ist.psu.edu/viewdoc/download?doi¼10.1
Fan, T., Buckley, J.S., 2002. Rapid and accurate SARA analysis of medium gravity crude .1.564.2005&rep¼rep1&type¼pdf.
oils. Energy Fuels 16 (6), 1571–1575. https://doi.org/10.1021/ef0201228. Moradi, S., Dabiri, M., Dabir, B., Rashtchian, D., Emadi, M.A., 2012. Investigation of
Fan, T., Wang, J., Buckley, J.S., 2002. Evaluating crude oils by SARA analysis. In: Paper asphaltene precipitation in miscible gas injection processes: experimental study and
SPE 75228 Presented at the SPE/DOE Improved Oil Recovery Symposium Held in modelling. Braz. J. Chem. Eng. 29 (3), 665–676.
Tulsa, Oklahoma, 13-17 April. Moura, L.G.M., Santos, M.F.P., Zilio, E.L., Rolemberg, M.P., Ramos, A.C.S., 2010.
F�
avero, C.V.B., Hanpan, A., Phichphimok, P., Binabdullah, K., Fogler, H.S., 2016. Evaluation of indices and of models applied to the prediction of the stability of crude
Mechanistic investigation of asphaltene deposition. Energy Fuels 30 (11), oils. Elsevier J. of Pet. Sc. and Eng. 74 (1–2), 77–87. https://doi.org/10.1016/j.
8915–8921. https://doi.org/10.1021/acs.energyfuels.6b01289, 2016. petrol.2010.08.011. October 2010.
Gaona, J.A.S., Manrique, J.P.B., Maj� e, Y.M., 2010. “Stability Prediction for Asphaltenes Mousavi-Dehghani, S.A., Riazi, M.R., Vafaic-Scfti, M., Mansoori, G.A., 2004. An analysis
Using SARA Analysis for Pure Petroleum”, Revista Ingeniería Y Regi� on No. 7 A~ no of methods for determination of onsets of asphaltene phase separations. J. Pet. Sci.
2010. Facultad de Ingeniería Universidad Surcolombiana. Eng. 42, 145–156. https://doi.org/10.1016/j.petrol.2003.12.007.
Gonzalez, D., Frank, L., Nikhil, J., Wilson, C.-M., Raul, O., 2012, April 3. Prediction of gas Prakoso, A., Punase, A., Rogel, E., Ovalles, C., Hascakir, B., 2018. Effect of Asphaltene
injection effect on asphaltenes precipitation using different thermodynamic models. Characteristics on its Solubility and Overall Stability. Energy & Fuels. https://doi.
In: AIChE Annual Meeting, Conference Proceedings. org/10.1021/acs.energyfuels.8b00324.
Goual, L., March 2nd, 2012. Petroleum Asphaltenes, Crude Oil Emulsions - Composition Punase, A., Prakoso, A., Hascakir, B., 2016, May 23. The Polarity of Crude Oil Fractions
Stability and Characterization. Manar El-Sayed Abdel-Raouf, IntechOpen. https:// Affects the Asphaltenes Stability. Society of Petroleum Engineers. https://doi.org/
doi.org/10.5772/35875. Retrieved from: https://www.intechopen.com/books/cr 10.2118/180423-MS.
ude-oil-emulsions-composition-stability-and-characterization/petroleum-asphalte Rogel, E., Carbognani, L., 2003. Density estimation of asphaltenes using molecular
nes. dynamics simulations. Energy Fuels 17, 378–386, 2003.
Guzman, R., Ancheyta, J., Trejo, F., Rodriguez, S., 2017. Methods for determining Rogel, E., Leon, O., Espidel, Y., Gonzalez, Y., 2001, May 1. Asphaltene Stability in Crude
asphaltene stability in crude oils. Fuel 188, 530–543. Oils. Society of Petroleum Engineers. https://doi.org/10.2118/72050-PA.
Hammami, A., Ratulowski, J., 2007. Precipitation and deposition of asphaltenes in Rogel, E., Ovalles, C., Moir, M., Vien, J., Morazan, H., 2015, April 13. Solubility
production systems: a flow assurance overview. In: Mullins, O.C., Sheu, E.Y., Characterization of Asphaltene Deposits. Society of Petroleum Engineers. https://
Hammami, A., Marshall, A.G. (Eds.), Asphaltenes, Heavy Oils, and Petroleomics. doi.org/10.2118/173757-MS.
Springer, New York, NY. Stankiewicz, A., Flannery, M.D., Fuex, N.A., Broze, J.G., Coach, J.L., Dubey, S.T.,
Hammami, A., Phelps, C.H., Monger-McClure, T., Little, T.M., 2000. Asphaltene Leitko, A.D., Nimmons, J.F., Iyer, S.D., Ratulowski, J., 2002. “Prediction of
precipitation from live Oils: an experimental investigation of onset conditions and Asphaltene Deposition Risk in E&P Operations”, Proceedings of the International
reversibility. Energy Fuels 14 (1), 14–18. https://doi.org/10.1021/ef990104z. Conference on Petroleum Phase Behaviour and Fouling. AIChE, ISBN 0816997748,
Hascakir, B., 2017, October 9. A New Approach to Determine Asphaltenes Stability. pp. 410–416.
Society of Petroleum Engineers. https://doi.org/10.2118/187278-MS. Sulaimon, A.A., Govindasamy, K., 2015, October 20. New Correlation for Predicting
Hassanvand, M., Shahsavani, B., Anooshe, A., 2012, February. Study of temperature Asphaltene Deposition. Society of Petroleum Engineers. https://doi.org/10.2118/
effect on asphaltene precipitation by visual and quantitative methods. J. of Pet. 176436-MS.
Tech. and Altern. Fuels 3 (2), 8–18. Available online at: http://www. Vargas, F.M., Garcia-Bermudes, M., Boggara, M., Punnapala, S., Abutaqiya, M.,
academicjournals.org/JPTAF DOI: 10.5897/JPTAF11.035. Mathew, N., Al Asafen, H., 2014, May 5. On the development of an enhanced method
Heithaus, J.J., 1962. Measurement and significance of asphaltene peptization. J. Inst. to predict asphaltene precipitation. In: Offshore Technology Conference. https://doi.
Pet. 48, 45, 1962. org/10.4043/25294-MS.
Hemmati-Sarapardeh, A., Ameli, F., Dabir, B., Ahmadi, M., Mohammadi, A., 2018. On Wang, J., Buckley, J.S., 2003. Asphaltene stability in crude oil and aromatic Solvent-The
the evaluation of asphaltene precipitation titration data: modeling and data influence of oil composition. Energy Fuels 17, 1445–1451.
assessment. Fluid Phase Equilib. 415, 88–100. Wang, J., Creek, J.L., Buckley, J.S., 2006, January 1. Screening for Potential Asphaltene
Problems. Society of Petroleum Engineers. https://doi.org/10.2118/103137-MS.

14
A.A. Sulaimon et al. Journal of Petroleum Science and Engineering 190 (2020) 106782

Watkinson, A.P., 2005. “Deposition from crude oils in heat exchangers”, ECI Symposium Yen, A., Yin, Y.R., Asomaning, S., 2001. Evaluating Asphaltene Inhibitors: Laboratory
Series, Volume RP2. In: Proceedings of 6th International Conference on Heat Tests and Field Studies. Society of Petroleum Engineers. https://doi.org/10.2118/
Exchanger Fouling and Cleaning–Challenges and Opportunities, Germany. 65376-MS. SPE-65376-MS. https://www-onepetro-org.eu1.proxy.openathens.net/d
Wattana, P., Fogler, H.S., 2005. Characterization of polarity-based asphaltenes ownload/conference-paper/SPE-65376-MS?id¼conference-paper%2FSPE
subfractions. Energy Fuels 29, 101–110. -65376-MS.
Wright, W., 1994. Simple Equations to Approximate Changes to the Properties of Crude Zendehboudi, S., Shafiei, A., Bahadori, A., James, L.A., Elkamel, A., Lohi, A., 2014.
Oil with Changing Temperature. Retrieved from. https://www.petroskills.com/b Asphaltene precipitation and deposition in oil Reservoirs-Technical aspects,
log/entry/crude-oil-and-changing-temperature#.XbvYEm5uKM8. experimental and hybrid neural network predictive tools. Chem. Eng. Res. Des. 92,
857–875.

15