Journal Pre-proof

Simulation of the deasphalting process of crude oils: Models development and

extraction conditions analysis

Igor De Las Heras, Javier Dufour, Coto Baudilio

PII: S0920-4105(21)01250-X
DOI: https://doi.org/10.1016/j.petrol.2021.109615
Reference: PETROL 109615

To appear in: Journal of Petroleum Science and Engineering

Received Date: 12 May 2021

Revised Date: 8 September 2021
Accepted Date: 4 October 2021

Please cite this article as: De Las Heras, I., Dufour, J., Baudilio, C., Simulation of the deasphalting
process of crude oils: Models development and extraction conditions analysis, Journal of Petroleum
Science and Engineering (2021), doi: https://doi.org/10.1016/j.petrol.2021.109615.

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.

© 2021 Published by Elsevier B.V.

CRediT author statement

Igor de las Heras López: Conceptualization, Methodology, Validation, Investigation,

Formal analysis, Data curation, Writing – Original Draft, Writing – review and editing;
Javier Dufour Andía: Formal analysis, Writing – review and editing, Visualization,
Supervision; Baudilio Coto García: Conceptualization, Validation, Formal analysis,
Writing – review and editing, Supervision.

f
r oo
-p
re
lP
na
ur
Jo
 1 Simulation of the deasphalting process of crude oils: models development and extraction

2 conditions analysis

3
4 Igor De Las Herasa, Javier Dufoura,b, Baudilio Cotoa*
a
5 Chemical, Energy and Mechanical Technology Department, Universidad Rey Juan Carlos,
6 Madrid, Spain
b
7 Systems Analysis Unit, IMDEA Energy Institute, Madrid, Spain
8

9 *To whom correspondence should be addressed

f
oo
10 e-mail: baudilio.coto@urjc.es

r
11 Phone: +34 914887089
12
-p
re
lP
na
ur
Jo
13 Keywords

14 Asphaltenes Elimination; Liquid-Liquid Equilibrium; Solvent Extraction; Solvent

15 Deasphalting; Modeling; Crude Oil Model

16 Abstract

17 Heavy crude oils are under constant investigation due to light oil exhaustion. So, its use is

18 compulsory to meet the world energy demand. Heavy crude oils present several drawbacks as

19 high content of metals, heteroatoms, and asphaltenes, and, therefore, upgrading processes are

20 becoming essential. A deasphalting process, conventionally used for vacuum residue

f
oo
21 upgrading, can be an interesting pretreatment to eliminate the heavier phases and ease the crude

r
22 processing in conventional refining facilities.

23
-p
Most of the models developed for deasphalting simulation show problems to be implemented
re
24 in simulation software packages. In this work, a versatile deasphalting simulation method is
lP

25 developed, composed by the definition of a synthetic crude oil model, formed by discrete
na

26 pseudo components, and a deasphalting model approaching the process to a liquid-liquid

27 extraction based on the thermodynamic model modified UNIFAC (Dortmund). Both models
ur

28 were satisfactorily validated. The deasphalting method was used to explore the solvent/crude
Jo

29 ratio, extraction temperatures, and alternative solvents in substitution of conventional light

30 paraffins. Higher extraction temperatures did not enhance remarkably the deasphalting process

31 and the optimum solvent to crude ratio depends on the solvent type. Alcohols were found as

32 promising compounds as their use is recommended by some studies and they showed great

33 deasphalting results.

34 Introduction

35 The reduction of conventional crude oil reservoirs has led to an increase in the exploitation of

36 heavy crude oils. So, the study of the effects of higher asphaltene content and the process to

37 remove it has become very important (Mohammed et al., 2021). Within the refining process,
38 deasphalting operation after the vacuum distillation leads to a heavy cut free of resins and

39 asphaltenes, with reduced sulfur and metal content, and ready for catalytic cracking or

40 hydrocracking, among other processes (Speight, 2020). The typical deasphalting process

41 consists of a heavy feedstock upgrading step by the injection of light paraffins as propane,

42 butane, or pentane into the vacuum residue. A heavy fraction formed essentially by asphaltenes

43 is precipitated, eliminating contaminants as well as heteroatoms, metals, coke precursors, etc.,

44 and hence, upgrading the deasphalted oil (DAO) obtained that can be sent to further processes

45 as Fluid Catalytic Cracking (FCC) or hydrocracking.

f
oo
46 The deasphalting process is a complete mature technology included in most refineries, where

r
47 four variations of the process can be found: Residuum Oil Supercritical Extraction (ROSE)

48
-p
licensed by Kerr-McGee, the Demex process by UOP, the Solvahl process by Axens, and the
re
49 Institut Français Du Petrol, and the Advanced Supercritical Deasphalting Process (PASD) by
lP

50 Petrobras. The differences of these deasphalting processes are found in the separation scheme,
na

51 the solvent employed, temperature, and pressure, and their characteristics are wide detailed in

52 the bibliography among with more variations with some improvements from the conventional
ur

53 processes (Ramirez-Corredores, 2017; Speight, 2020).

Jo

54 In the future, due to the increasing quality requirements of petroleum products, the deasphalting

55 will become the main upgrading process for various feedstocks in refinery processes. In this

56 respect, the use of deasphalting as a process before the atmospheric distillation for heavy crude

57 oils can be an interesting approach. This would eliminate the heaviest phases, along with many

58 contaminants, making easier the processing of the crude in the refining processes, such as

59 catalytic ones. Furthermore, this approach would allow testing different conditions of the

60 process like the temperature, solvent to crude ratio, or the solvent used.

61 Concerning the solvent used, some alternative ones have been considered for deasphalting in

62 the refinery, as superheated water (Brons et al., 1993), N-methyl-2-pyrrolidone (Bushnell et al.,
63 1977), dimethyl carbonate with CO2 at overpressure conditions (Cesar Savastano and Cimino,

64 1993) and with sulfuric acid (Lerda et al., 1990). Furthermore, inside the well, more possible

65 solvents have been studied for deasphalting as pretreatment before conventional refining. Ilyin

66 et al. (Ilyin et al., 2016) studied the amount of precipitated asphaltene and the final viscosity in

67 relation to the solvent (paraffins, ethers, or HMDSO). Recent studies were focused on the use

68 of propane or butane with additives within the reservoir (Rogel and Ovalles, 2013), the use of

69 conventional solvents with solid particles of SiO2 (Guzmán et al., 2017), acidic compounds that

70 lead to the polymerization of asphaltenes (Sun et al., 2017), or using supercritical solvents as

f
oo
71 CO2 (Nascimento et al., 2021). Their objectives are the improvement of the standard process

r
72 by increasing the sulfur and metal reduction (Guzmán et al., 2017), solvent recovery, and

73
-p
reducing energy consumption (Ahn et al., 2016). Nevertheless, none of these works have
re
74 studied the use of less hazardous solvents. Some studies (Byrne et al., 2016; Prat et al., 2015,
lP

75 2014) have developed technical criteria to recommend the use of alternative solvents in
na

76 chemical industry. Alcohols, esters and ketones appear in these recommendations since health,

77 safety and environmental concerns are reduced compared to conventional ones. The use of
ur

78 alternative and recommended solvents in deasphalting appears as a new field to study.

Jo

79 The study of the deasphalting with experimental work is necessary but laborious and time-

80 consuming. The addition of simulation experiments appears as a useful alternative to reduce

81 the time and resources of experimentation. The simulation of such a process implies, first, the

82 need of model crude oils, whose composition and components distribution are complex and

83 variable for every crude oil type. Secondly, the deasphalting separation is based on molecular

84 types differences between the precipitated and the remaining phases of the DAO, unlike most

85 of the separations in the refinery, based on the boiling point gradient. These aspects hinder the

86 simulation of the deasphalting process in simulation software packages.

 87 For a long time, several models have been applied to describe the asphaltene solubility and

88 precipitation, but only a qualitative agreement was found, where models incorporate molecular

89 weight distribution and distributions of solubility parameters of solutes as the most important

90 data (Andersen and Speight, 1999). Some authors use semi-empirical models where these

91 parameters are considered (Fahim, 2007; Nunes et al., 2019; Painter et al., 2015) and recent

92 studies are found that support the use of empirical models over other ones (Santos et al., 2019).

93 In contrast, other authors use thermodynamic models with specific modifications for asphaltene

94 solubility. For example, models with a modification of the Flory-Huggins polymer theory

f
oo
95 (Wang and Buckley, 2001) and the Kikic modification of the Flory-Huggins theory (Santos et

r
96 al., 2017) are found. The Peng-Robinson equation of state for the simulation of several

97
-p
upgrading units, assuming liquid-liquid equilibrium (LLE) for the deasphalting unit is also used
re
98 (Gholami et al., 2021). As well, the regular solution theory and the assumption of LLE between
lP

99 the asphaltenic and non-asphaltenic phases is assumed to simulate the process (Tharanivasan et
na

100 al., 2009). A recent work (Daryasafar et al., 2020) found better modeling results usinga more

101 complex thermodynamic model as perturbed-chain statistical associating fluid theory (PC-
ur

102 SAFT).
Jo

103 Most of the above models describe quite accurately the asphaltene behavior. However, they are

104 limited to specific crude oils and conditions, and their accurateness is directly related to their

105 complexity. Therefore, their implementation in simulation software packages and within the

106 rest of the refinery models is a gap to fill. To our knowledge, there is not described a

107 straightforward and versatile method to use in simulation software packages that work

108 correctly.

109 This work focuses on the option to carry on a deasphalting process during the refining process

110 before the atmospheric distillation and the optimization of such process using simulation. The

111 aims of this work are, first, to develop a tunable model (synthetic crude oil model -SCO-), to
112 define any crude oil with specific pseudo components, allowing its use with group contribution

113 thermodynamic models, and validate it with experimental data of real crude oils from literature.

114 Second, to develop a straightforward and versatile model based on the LLE fundamentals to

115 simulate the deasphalting process implemented in Aspen Plus®, and validate it by two steps:

116 with experimental data literature, on the one hand, and experimental results from this work, on

117 the other hand. Finally, the simulation method, combining the SCO and the deasphalting

118 models, is used to test different conditions of the deasphalting process, combining both models:

119 the use of alternative solvents, solvent to crude ratio, and temperature of deasphalting.

f
oo
120 Synthetic crude oil model

r
121 2.1 Crude oil model development
-p
re
122 The tunable synthetic crude oil model (SCO) was developed emulating a medium-heavy crude
lP

123 oil, whose properties were compared with experimental data of Maya (heavy) and Istmo (light)
na

124 oils (Espinosa-Peña et al., 2004). The elemental and structural compositions were also

125 calculated and compared with the bibliography (Iwase et al., 2018).
ur

126 The use of the SCO for deasphalting modeling implies the separation by molecular type
Jo

127 (asphaltenic/non-asphaltenic). For this purpose, the saturates, aromatics, resins, and asphaltenes

128 (SARA) families distribution was employed and it was set according to the literature (Fulem

129 et al., 2008; Zhao and Shaw, 2007): 30 % saturates, 50 % aromatics, 5 % resins and 15 %

130 asphaltenes, in mass basis. The SCO was formed then by discrete pseudo components where

131 thirty molecules were included: nine saturates, ten aromatics, five resins, and six asphaltenes.

132 Mass fractions of individual molecules were equally divided within their family. Fig. 1 shows

133 one example of each SARA molecule. All the molecules included can be consulted in

134 supplementary materials.

135

f
oo
136 Figure 1. Examples of some molecules of the SCO. a) saturate; b) aromatic; c) resin; d)

r
137 asphaltene.

138
-p
Table 1 shows the boiling point, calculated by the Gani equation (Constantinou and Gani,
re
139 1994), and molecular weight of the molecules estimated, sorted by SARA families, and ordered
lP

140 by boiling point. As can be seen, the SCO includes molecules covering a wide range of
na

141 properties including asphaltenes molecules that are big enough to represent the heavy fraction,
ur

142 whose shape of all of them can be consulted in supplementary materials.

143 Heteroatoms (Nitrogen, Sulphur, metals) have not been considered since their presence does
Jo

144 not affect the process.

145

146 Table 1. Properties of the pseudo components of the synthetic crude oil.

Family Molecules Boiling point (ºC) Molecular weight (g·mol-1)

1 133.5 114.2
2 164.8 128.3
3 167.8 140.0
4 236.9 212.0
Saturates 5 277.6 252.0
6 278.9 254.0
7 321.8 282.6
8 369.4 381.0
9 404.9 422.8
Aromatics 10 138.2 106.2
 11 222.0 176.0
12 234.8 174.3
13 297.9 212.3
14 308.4 256.4
15 315.7 270.4
16 319.0 240.4
17 326.6 238.4
18 347.2 298.0
19 350.9 314.0
20 354.9 312.0
21 407.6 400.6
Resins 22 427.0 454.7
23 447.0 490.0
24 465.4 504.8

f
25 502.8 588.9

oo
26 516.5 564.8
27 521.3 638.9
Asphaltenes
28 524.8 626.0

r
147
29
30
527.6
550.5
-p 607.0
719.0
re
lP

148 2.1 Characterization and validation of the SCO model

149 Once created the SCO, its TBP curve, and API gravity were estimated by Aspen Plus®, among
na

150 the rest of the properties, and validated with experimental data from literature [23,24]. The
ur

151 validation of the SCO model was first made comparing the TBP curve of the SCO, calculated
Jo

152 by Aspen Plus® software, and the literature values of the Istmo and Maya crude oil samples

153 (Espinosa-Peña et al., 2004), showed in Fig. 2. The calculated TBP is similar to the

154 experimental ones. The literature does not show more than 80 % distilled, but the final tendency

155 of the SCO with a slope increase is the usual one because of the asphaltenic fraction.
 700 Istmo Crude Oil
Maya Crude Oil
600 SCO

500
TBP (ºC)
400

300

200

100

f
0 20 40 60 80 100

oo
% Distilled
156

r
157 Figure 2. Total Boiling Point distillation curves.

158
-p
The API gravity of the SCO resulted in 28 ºAPI. If compared with the experimental API of the
re
159 Maya (< 20ºAPI) and Istmo oils (<30ºAPI) (Espinosa-Peña et al., 2004), the SCO model is
lP

160 similar to the Istmo crude oil, and, concerning this property, classified as intermediate crude oil
na

161 model.

162 Table 2 lists the resulting elemental and structural composition of the SCO model and the
ur

163 comparison with bibliographic data. Bibliography (Iwase et al., 2018) considered different
Jo

164 compound families (saturates, lost components, aromatics, and resins) and next correspondence

165 between SARA and such bibliographic families was established: SARA saturates as saturates,

166 SARA aromatics as lost components, SARA resins as aromatics, and SARA asphaltenes as

167 resins. A final comparison of the SCO values with the bibliographic ones was carried out

168 considering the sum of saturates and aromatics compounds as the light fraction and the sum of

169 resins and asphaltenes as the heavy fraction. The validation of these parameters can be

170 considered more realistic for deasphalting modeling.

171 Table 2. Elemental and structural composition of SARA families of this work and bibliographic

172 data.

Saturates + Aromatics Resins + Asphaltenes

SCO
Bibl.
(Iwas
Mode Bibl. (Iwase et al., Mode Bibl. (Iwase et al., Mode
e et
l 2018) l 2018) l
al.,
2018)
Mole fraction (%) 90,6 88,8 9,4 11,2
Mn (g/mol) 234 265 576 480 302 306
Ratio C/H (%) 87,9 88,0 91,1 88,8 88,5 88,1
Total Aromaticity Content (%) 27,1 21,6 60,6 55,8 30,2 25,4
Total Aliphatic Content (%) 72,9 78,4 39,4 44,2 69,8 74,6

f
oo
Other aliphatic (%) 25,4 28,3 22,3 10,8 25,1 26,3
Chain CH2 + Aliphatic CH (%) 25,6 23,9 8,5 17,2 23,9 23,1
Total α-CH3 + β-CH3 + γ-CH3

r
16,0 18,6 7,4 12,8 15,2 17,9
(%)
Naphthenic CH2 (%) 5,9
-p
7,7 1,3 3,1 5,5 7,2
173
re
174 The comparison of the values serves to check a correct composition of the SCO as an average
lP

175 intermediate crude oil. In main terms, SCO and references fit adequately, which means that the
na

176 molecular weight and composition of the SCO molecules are correctly estimated. However,
ur

177 since the SCO was not developed to be a duplicate of the literature values (Iwase et al., 2018),
Jo

178 some deviations appear.

179 The aromatic content of the SCO is higher than that of the literature. Given that the SCO is

180 heavier in terms of asphaltenic proportion, its aromaticity is increased. In the heavy fraction,

181 the aliphatic content of the SCO is mainly due to small branches rather than large chains,

182 methyl-branched or naphthenic carbons, as the values indicate. Concerning the structural

183 composition in the aliphatic content of the SCO, the values are lower, but proportionally, they

184 fit with the bibliographic values. The rest of the values do not show large deviations and for

185 individual structural groups, the standard deviation (σ) between calculated and bibliographic

186 values is 6.6 %. The standard deviation was calculated as the square root of the average of the

187 squared differences of bibliographic and calculated values.

188 From the results shown above, the SCO model is characterized and validated. Moreover, the

189 SCO can be modified to represent other crude oils according to the modeling requirements by

190 adjusting the composition of the SARA groups, the individual compositions or even adding

191 new molecules, which gives versatility to the model. Nonetheless, we remark that the

192 simplification of the SCO by not including heteroatoms in the molecules should not have a

193 strong effect on the thermodynamic model selected, however, for further works, the

194 incorporation of heteroatoms in the molecules, whenever the binary parameters are available,

195 would serve to confirm this fact and manage a more realistic model.

f
oo
196 Deasphalting model

r
197 3.1 Thermodynamic model
-p
re
198 In this work, the deasphalting process was approached to an LLE. The model uses the LLE
199 Eqs. (1-2):
lP

β
200 fiα = fi Eq.1
na

201 (xi γi )α = (xi γi )β Eq.2

202 where xi, fi, and γi are the molar fraction, fugacity, and activity coefficient of component i, and
ur

203 superscripts α and β represent the two liquid phases.

Jo

204 A predictive thermodynamic model would be the best option to calculate the activity

205 coefficients due to the variable nature of the molecules. In this sense, the group contribution

206 methods are very powerful models that can be applied to any molecule. The most used ones for

207 activity coefficient calculation are the UNIFAC (Fredenslund et al., 1975) and its variations:

208 mod. UNIFAC (Dortmund) (Constantinescu and Gmehling, 2016), mod. UNIFAC (Lingby)

209 (Larsen et al., 1987), etc. The mod. UNIFAC (Dortmund) has been demonstrated to give better

210 results than the rest of the models (Gmehling et al., 2015), mainly due to its improvement in

211 the temperature dependence of the group interaction parameter. Furthermore, its parameters

212 matrix data have been updated regularly during the last years according to new experimental
213 information in the Dortmund Data Bank. Therefore, the mod. UNIFAC (Dortmund) was the

214 thermodynamic model selected.

215 The deasphalting process was simulated in a one-stage LLE extractor using Aspen Plus®

216 software, as schemed in Fig. 3. The SCO and the solvent are the inlet currents to the LLE

217 extractor. Two equilibrium phases are obtained as outlet currents, an asphaltenes rich phase

218 (ARP) that contains the highest quantity of resins and asphaltenes, and a solvent rich phase

219 named deasphalted oil (DAO), keeping most of the light fraction (saturates and aromatics).

220

f
r oo
-p
re
221
lP

222 Figure 3. LLE extractor simulation scheme

na

223 The comparison of the deasphalting results was carried out according to two criteria:
ur

224  The asphaltenes reduction in the DAO and the extraction yield.
Jo

225  The partition coefficients and the selectivity.

226 The first two parameters are commonly used as parameters of the deasphalting processes. The

227 asphaltene reduction parameter stands for the mass quantity of the heavy fraction that is

228 eliminated from the SCO forming the ARP phase, and then, when it is higher, the quality of the

229 DAO improves. The extraction yield represents the efficacy of the DAO obtention, and its

230 optimal is the sum of the content of the SCO light fraction (when asphaltene reduction is 100

231 %).

232 The second criteria of parameters are widely used in liquid-liquid extraction processes. A

233 partition coefficient indicates how one or some substances are distributed between the two

234 liquid equilibrium phases. Finally, the selectivity helps to know if a solvent extracts specifically
235 a target substance, or a group of substances, of the multicomponent mixture, or if it extracts

236 several substances non-selectively. With both parameters, a solvent screening can be carried

237 out to select one for an extraction process design. Such magnitudes are calculated by Eqs. (3-

238 7).

(𝑊𝑅+𝐴 )𝐴𝑅𝑃
239 𝐴𝑅(%) = (𝑊𝑅+𝐴 )𝑆𝐶𝑂
· 100 Eq.3

240 where AR is the asphaltenes reduction (%), (WR+A) is the mass flow of resins and asphaltenes

241 and ARP and SCO superscripts refer to asphaltene rich phase and synthetic crude oil currents,

f
242 respectively.

oo
𝑊 𝐷𝐴𝑂
243 𝑦𝐸 (%) = · 100 Eq.4
𝑊 𝑆𝐶𝑂

r
244
-p
where yE is the extraction yield (%), W is a mass flow current (solvent not counted) and DAO
re
245 superscript refers to deasphalted oil current.
lP

𝐾𝐻
246 𝑆𝐻 = Eq.5
𝐾𝐷𝐴𝑂
na

247 where SH is the selectivity, and KH and KDAO are the partition coefficients of the SCO heavy

248 fraction and DAO light fraction, respectively, as determined by the following equations:
ur

(𝑥 )𝐴𝑅𝑃
𝐾𝐻 = (𝑥 𝑅+𝐴)𝐷𝐴𝑂 Eq.6
Jo

249
𝑅+𝐴

250 where KH is the partition coefficient of the SCO heavy fraction, (xR+A) is the molar composition

251 of the resins and asphaltenes.

(𝑥 )𝐴𝑅𝑃
252 𝐾𝐷𝐴𝑂 = (𝑥 𝑆+𝐴𝑟)𝐷𝐴𝑂 Eq.7
𝑆+𝐴𝑟

253 where KDAO is the partition coefficient of the SCO light fraction, (x S+Ar) is the molar

254 composition of the saturates and aromatics.

255 3.2 Validation of the deasphalting model

256 The validation of the deasphalting model was carried out in two steps. The first one consisted

257 of a benchmarking between the simulated and experimental data of LLE extractions of
258 aromatics, from mixtures of saturates and aromatics compounds. The experimental data of the

259 LLE extractions were obtained from the literature (Al-Jimaz et al., 2005; Alkhaldi et al., 2009;

260 Fandary et al., 2006). Table 3 lists the five systems selected for the validation, including all the

261 compounds involved in each system, the solvent, the extraction temperatures, and the

262 bibliographic references.

263 Table 3. Data of the materials and their experimental works selected for validation of the

264 model.

f
Compounds Experimental conditions

oo
System Saturates Aromatics Solvent Temperatures (ºC) Reference
1 n-Dodecane and n- Mesitylene and n- NMP 20, 30 (Alkhaldi et
hexadecane butylbenzene al., 2009)

r
2 n-Tetradecane Pentylbenzene NMP 24.85; 34.85; 44.85; 54.85 (Fandary et al.,

4
n-Dodecane

n-Tetradecane
Propylbenzene

Propylbenzene
-p
NMP

NMP
24.85; 34.85; 44.85; 54.85

24.85; 34.85; 44.85; 54.85

2006)
(Al-Jimaz et
al., 2005)
(Al-Jimaz et
re
al., 2005)
5 n-Heptadecane Propylbenzene NMP 24.85; 34.85; 44.85; 54.85 (Al-Jimaz et
al., 2005)
lP

265

266 The parameters analyzed were the compositions (mole fraction) of the saturates, aromatics, and
na

267 solvent at each equilibrium phase, selectivity SH (Eq. 5), and the partition coefficient KH (Eq.
ur

268 6). Fig. 4 shows the comparison between the experimental and the simulated results of some of
Jo

269 these parameters for systems 3, 4, and 5 (Table 3). Since these systems carry out the extraction

270 of aromatics at low concentrations, there is a solvent-rich phase that contains the largest content

271 of aromatics (ARP), and the saturates remain mostly in a saturates rich phase (DAO). In Fig. 4,

272 it is observed that the simulated and experimental results mostly agree within the point clouds.

273 Nonetheless, the deviations appreciated are mostly due to the compositions of the solvent and

274 the saturates, since their compositions are noticeably higher than the aromatics ones, leading to

275 higher deviations in absolute terms, which does not mean that the relative errors are higher.

276 Furthermore, some tendencies of the deviations of the saturate and the solvent are visualized in

277 both DAO and ARP phases, which are caused by the error committed by the mod. UNIFAC
278 when the LLE is calculated and leading to a displacement of the phase coexistence region,

279 which drags the tendency of the deviation in the selectivity as well.

1,0 1,0
Saturates DAO Phase Saturates ARP Phase
Simulated compositions (mol frac.)

Simulated compositions (mol frac.)

Aromatics Aromatics
Solvent Solvent
0,8 0,8

0,6 0,6

0,4 0,4

0,2 0,2

0,0 0,0
0,0 0,2 0,4 0,6 0,8 1,0 0,0 0,2 0,4 0,6 0,8 1,0

f
Experimental compositions (mol frac.) Experimental compositions (mol frac.)

oo
30 SH

r
-p
Simulated SH

20
re
10
lP

0
0 10 20 30
na

Experimental SH
280
281 Figure 4. Experimental (Al-Jimaz et al., 2005) vs simulated mole fractions for ARP and DAO
ur

282 phases and selectivities (SH) of the reference systems 3, 4, and 5 of Table 3.
Jo

283 Table 4 shows the quantitative analysis of this validation, wherein the standard deviation and

284 the maximum values observed of the compositions (of each family and extraction phase),

285 partition coefficients (KH), and selectivities (SH) are presented. The maximum values observed

286 give an idea of the relative significance of the standard deviation of each case.

287 Table 4. Full model validation: standard deviations (σ) and respective maximum values

288 observed (Max. value obs.) for each SARA family and extraction phase (mole fraction),

289 partition coefficient, and selectivity.

σ Max. value obs.

XSATURATES 0.112 0.935
Deasphalted
XAROMATICS 0.031 0.228
Oil
XSOLVENT 0.093 0.527
 XSATURATES 0.061 0.318
Asphaltene
XAROMATICS 0.019 0.187
Rich Phase
XSOLVENT 0.078 0.988
KH 0.245 1.288
SH 4.638 33.52

290

291 Small deviations appear between the model and the experimental data, so the deasphalting

292 model predicts correctly the LLE extraction of the aromatics of the systems. The standard

293 deviations are not remarkable respect to the maximum values observed, which indicates that

294 the model results are correct. Moreover, if the temperature effect on the deviations was

f
oo
295 calculated, a tendency of increasing errors when extraction temperatures get higher would be

296 observed, being the smaller phase coexistence region at higher temperatures as the possible

r
297 reason behind this fact.
-p
re
298 The second validation step was performed in our laboratory and it consisted on the
lP

299 benchmarking with an experimental deasphalting, carried out through asphaltene precipitation
na

300 with isopropanol of a Maya crude oil following the standard test method IP 143 (Energy

301 Institute, 2016). This standard allows precipitating C7-asphaltenes of a crude oil, isolating as
ur

302 well the saturates, aromatics, and resins by means of a multistage solvent extraction with n-
Jo

303 heptane and toluene. In this work, n-heptane was replaced by isopropanol to validate the model

304 with a solvent that we are going to analyze and the literature endorses its use as a solvent (Byrne

305 et al., 2016; Prat et al., 2015, 2014). Maya crude oil sample was supplied by Repsol S.A. and

306 isopropanol (100 %) purity by VWR.

307 The extraction yield was calculated for experimental deasphalting and the corresponding

308 simulation at different R-S/C ratios (Fig. 5). The Maya crude oil is viscous enough to heavily

309 hinder the experiments at R-S/C lower than 5, so that, trials at R-S/C 2 and 3 did not show

310 satisfactory results due to massive sample mass loss. Therefore, the onset region was not

311 determined completely. However, the rest of the experiments ran correctly at R-S/C 5, 10, 20

312 and 30, by triplicate. From R-S/C 5 onwards, the experimental yE sets constant, despite a low
313 descent that we attribute it to experimental uncertainties. Secondly, the simulated yE does not

314 stabilize until around R-S/C 30, and a slight overestimation of the yE by the model is detected

315 for R-S/C over 5. However the standard deviation between all the experimental and simulated

316 yE was 4.44 % and consequently, the deasphalting results from the simulation are considered

317 acceptable, and the model validated.

318 Nevertheless, this model only contemplates the LLE as the separation mechanism of the

319 deasphalting, however, the deasphalting is far more complex, including solid-liquid equilibria

320 and the destabilization of the asphaltenes when the solvent is added, producing flocculation and

f
oo
321 the final precipitation of these molecules. The validation confirms that the LLE is a satisfactory

r
322 estimation for the solvents tested in this work, nevertheless, we suggest that future research

323
-p
should be made concerning the incorporation of these precipitation mechanisms into the
re
324 simulation software package without worsening the versatility of the model.
lP

100
na

80
ur

60
Jo
yE (%)

Experimental yE
40 Simulated yE

20

0
0 10 20 30

325 R-S/C

326 Figure 5. Validation of deasphalting model through experimental deasphalting.

327 Analysis of the deasphalting conditions

328 Once validated the SCO and the deasphalting models are combined to study different
329 deasphalting conditions. The variables analyzed were: solvent to crude oil ratios (values ranging
330 from 1-30), extraction temperature (set at 20 and 50 ºC), and some alternative solvents. Related
331 to the solvent selection, we considered technical criteria (risk, health, environment, quality,
332 industrial constraints, and cost) (Prat et al., 2014), and ten solvents were selected: four aprotic
333 polar ones, four alcohols, and two esters, covering a wide range of solvent types. Table 5
334 summarizes their boiling points and use recommendations.

335 Table 5. Properties of the solvents used: boiling point (BP, ºC), solvent type, and

336 recommendation from literature (Prat et al., 2014).

Solvent BP (ºC) Solvent type Recommendation BP reference

DMF (N,N- 152.95±1.0
dimethyl-

f
formamide) Aprotic polar Hazardous (Joshi et al., 1990)

oo
165.05 (Robert C. Weast,
DMA (N,N- Jeanette G.
dimethylacetamide) Aprotic polar Hazardous Grasselli, 1989)
202.05 (Robert C. Weast,

r
NMP (1-methyl-2- Jeanette G.
pyrrolidinone))
Sulfolane
287.85
-p
Aprotic polar
Aprotic polar
Hazardous
Recommended
Grasselli, 1989)
(Lee, 1986)
re
64.65±0.3 Recommended (Blanco and
Methanol Alcohol Ortega, 1998)
82.35±0.4 (Hiaki et al.,
lP

Isopropanol Alcohol Recommended 1995)

82.35±0.7 Recommended (Loras et al.,
Tert-butanol Alcohol 1999)
na

197.35±0.5 (Robert C. Weast,

Jeanette G.
Ethylene glycol Alcohol Recommended Grasselli, 1989)
77.05±0.2 (Blanco and
ur

Ethyl acetate Ester Recommended Ortega, 1998)

101.55±0.8 (Robert C. Weast,
Jeanette G.
Jo

n-Propyl acetate Ester Recommended Grasselli, 1989)

337

338 4.1 Solvents selection

339 The analysis of the solvents selected was carried out, wherein the yE, AR, SH, and KH served to

340 rapidly find the most interesting solvents in a preliminary selection. Fig. 6 plots the best results

341 for each solvent for all the R-S/C (20 ºC), where the most suitable ones are found at the top-

342 right side of both plots.

  (DMA)  (NMP)  (Sulfolane)  (DMF) (Ethyl acetate)  (Propyl acetate)  (Iso-propanol) (Tert-butanol)  (Ethylene glycol)  (Methanol)
100
100

80

60

yE (%)
SH

10
40

20

a) b)
1 0
1 10 100 0 20 40 60 80 100
KH AR (%)
343

f
344 Figure 6. Results for each solvent used. a) partition coefficient (KH) vs. selectivity (SH); b)

oo
345 asphaltene reduction (AR) vs. extraction yield (yE).

r
346 -p
Fig. 6 a) shows a conventional solvent screening for solvent extraction (KH and SH). The
re
347 solvents line up as seen, being the methanol optimal, followed by the rest of the alcohols, then,
lP

348 the polar aprotic solvents, and finally, the esters. Fig. 6 b) shows the solvents' results in terms

349 of extraction yields and asphaltene reduction. Three main zones can be found, the first in which
na

350 some alcohols proposed as the best solvents (methanol, isopropanol, and tert-butanol) with an
ur

351 average yE of 78 % and AR of 84 %. Methanol reaches a better position in Fig. 6, but it needs
Jo

352 an R-S/C 30. Tert-butanol and isopropanol produce worse AR but higher yE in most cases and

353 for an optimum R-S/C of 4-5. Thus, they need 7 times less amount than methanol at their

354 optimum point, so this lower R-S/C would allow a lower use of solvent. Further

355 experimentation and a techno-economic assessment are needed to confirm the goodness of

356 these solvents for deasphalting use and to discard potential problems.

357 The second zone corresponds to three of the polar aprotic solvents (DMF, NMP, and DMA)

358 with yE of 67 % and AR of 44 %. These solvents have particular behaviors, and their high

359 boiling points represent an interesting advantage. Moreover, they serve usually as aromatic

360 selective solvents, so we think they need a deeper study to check their potential use. Eventually,

361 their great disadvantage is that they are not recommended because of their dangerousness (Prat
362 et al., 2014). The third zone is formed by ethylene glycol and sulfolane, with a yE of 19 % and

363 AR of 94 %, so they would not be favorable for this process with these conditions. The esters

364 stay apart of these zones, being discarded for deasphalting due to their poor results.

365 Some differences between the KH-SH and yE-AR criteria are observed. They lead to consider

366 that KH-SH calculation for solvent screening for the deasphalting process is not adequate. The

367 reason is that the optimal solvents selected with this criterion are not eliminating the asphaltenic

368 fraction significantly. Therefore, we think that yE-AR criterion is better to evaluate the

369 deasphalting.

f
oo
370 Finally, the typical industrial extraction yields with light paraffins are stated at 56±9 % and also

r
371 total AR (Speight, 2020). Besides, the experimental yE of the Maya crude oil with isopropanol

372
-p
in this laboratory resulted, at its best, in 79.4±0.8 % and total AR, as shown in Fig. 5. This
re
373 demonstrates that the first zone (alcohols) of Fig. 6 would equal or even enhance the industrial
lP

374 yields. Therefore, these are promising results since these alcohols are also recommended
na

375 solvents by scientific criteria.

ur

376 4.2 Solvent to crude ratio discussion

Jo

377 Different results for each solvent were observed when the deasphalting model was simulated.

378 Fig. 7 shows the yE (a)), AR (b)) and SH (c)) for six solvents. Note that the selectivity axis is

379 logarithmic.
 100 100

90 90

80 80

70 70

60 60

AR (%)
Methanol
yE (%)

Isopropanol
50 Ethylene glycol 50
Propyl Acetate
40 40
DMF Methanol
30 Sulfolane 30 Isopropanol
Ethylene glycol
20 20 Propyl Acetate
DMF
10 10 Sulfolane
a) b)
0 0
5 10 15 20 25 30 5 10 15 20 25 30
R-S/C R-S/C

100

f
Methanol

oo
Isopropanol
Ethylene glycol
Propyl Acetate
SH

DMF
Sulfolane
10

r
-p
re
c)
1
5 10 15 20 25 30
R-S/C
lP

380
381 Figure 7. Extraction yield (yE), Asphaltene reduction (AR), and selectivity (SH) in the
na

382 function of the solvent to crude ratio (R-S/C) for six solvents.
ur

383 Generally, when increasing the R-S/C, the yE rises and the AR reduces. The SH reduces in most

384 cases, except for methanol and ethylene glycol. According to these tendencies, higher R-S/C
Jo

385 implies a carryover of all the SCO families to the DAO, including a higher amount of the heavy

386 fraction, and thus, lower quality of the DAO. Therefore, the R-S/C selected will depend on the

387 yield and asphaltenes and sulfur and metals content required for the DAO. Since the DAO will

388 be sent to the refining process, we consider that higher yields (yE) are preferred, whenever AR

389 is satisfactory, e.g., if the DAO is going to be sent to catalytic conversion units, the quality

390 required for the DAO is higher and then, a higher AR is needed despite the lower yE.

391 4.3 Temperature effect

392 Deasphalting is a medium temperature process, which proceeds between 40-150 ºC depending

393 on the solvent (Speight, 2020). The extraction temperature has to be high enough to get the
394 crude oil fluid and to obtain a satisfactory yield but without overpassing the thermal degradation

395 temperature of the crude. The use of alternative solvents could widen the range of available

396 temperatures while yielding sufficiently. Fig. 8 shows the comparison of SH, KH, yE, and AR

397 obtained for the best results of each solvent between both temperatures (20 & 50 ºC). The

398 analysis of the different temperatures showed multiple behaviors of the solvents. The increase

399 of the temperature in alcohols mostly showed an improvement of the deasphalting, with an

400 average rise of 24 % of the selectivities, but it led to a slight decline of the AR due to a carryover

401 effect in the heavy fraction. The methanol yE decreased a 45 %. The iso-propanol was an

f
oo
402 exception with an extraction enhancement without side effects.

r
100
20 ºC 100 20 ºC
50 ºC -p 50 ºC
Asphaltene Reduction (%)

80 80
Yield Extraction (%)

re
60 60
lP

40
40

20
na

20

0
ne

ne

l
A

P
e

ol

ol

l
F

l
no

no
co

co
no

no
at

at
M

M
M

M
an

an
la

la
ly

ly
ha

ha
ta

ta
et

et
D

D
D

N
lfo

lfo
op

op
G

G
bu

bu
Ac

Ac
et

et
ur
Su

Su
Pr

Pr
ne

ne
M

M
rt-

rt-
yl

yl
o-

o-
le

le
Te

Te
op

op
is

is
hy

hy
Pr

Pr
Et

Et

20 ºC 20 ºC
100
50 ºC 50 ºC
Jo

100
SH

KH

10 10

1 1
ne

ne

l
A

P
e

ol

ol

l
F

l
no

no
co

co
no

no
at

at
M

M
M

M
an

an
la

la
ly

ly
ha

ha
ta

ta
et

et
D

D
D

N
lfo

lfo
op

op
G

G
bu

bu
Ac

Ac
et

et
Su

Su
Pr

Pr
ne

ne
M

M
rt-

rt-
yl

yl
o-

o-
le

le
Te

Te
op

op
is

is
hy

hy
Pr

Pr

403
Et

Et

404 Figure 8. Comparison between extraction temperatures of the selectivity, partition coefficient

405 and extraction yield (yE), asphaltene reduction (AR), selectivity (SH), and partition coefficient

406 (KH).
407 The aprotic polar solvents presented different results. The DMF showed an improvement of the

408 AR of 26 % with higher temperatures and a decrease of the SH of 28 %. Moreover, when the R-

409 S/C is raised from 14 to 30, the yE descends a 26 %. The sulfolane obtained worse separation,

410 lowering the SH a 20 % and the yE an 11 %. The NMP did not show remarkable changes in its

411 results. The DMA showed a major LLE and the SH raised an 18 % and the yE augmented a 33

412 %, but with a worsening of the AR, descending a 37 %. Finally, the ethyl acetate did not produce

413 LLE at 50 ºC. The propyl acetate enhanced its AR to 13 %, but still, this result is very poor.

414 In conclusion, the increase in temperature produced slightly better results in the AR, SH, and KH

f
oo
415 with an average increase of 37 %, 8 %, and 9 % respectively. However, an average negative

r
416 effect was observed in the yE in most cases of 7 %. The temperature effect did not show an

417
-p
important influence on the process and the poor improvement of the results may not justify a
re
418 higher temperature and its corresponding costs.
lP

419 Conclusions
na

420 In this work, the necessity of a deasphalting model that can be applied and modified for different
ur

421 feedstocks and conditions has been satisfied. The deasphalting model is formed by the
Jo

422 combination of a discrete pseudo-component-based crude oil model and a deasphalting unit

423 with an approach of an LLE with the predictive group contribution thermodynamic model mod.

424 UNIFAC (Dortmund) implemented in Aspen Plus®. The validation of the crude oil and

425 deasphalting models has been appropriately performed in two ways with different experimental

426 and literature data, and it showed its good reliability. The combination of both models (synthetic

427 crude oil and deasphalting unit) allowed modeling the elimination of the asphaltenes fraction

428 from crude oil by phase equilibrium effortlessly.

429 The model described in this work allows screening new solvents and potential mixtures could

430 be proposed. We found that solvent screening using partition coefficients and selectivity as a

431 criterion is not useful to obtain the best asphaltenes elimination. However, extraction yield and
432 asphaltenes reduction are more accurate variables to optimize the selection and analysis.

433 Among all the solvents used, the alcohols appear as an interesting alternative to the

434 conventional solvents, showing high deasphalted oil flows besides large asphaltenes reduction,

435 whose results were corroborated by experimental work. In most solvents, the higher solvent to

436 crude ratio implied a higher extraction yield of the deasphalted oil, but its quality is reduced

437 due to an increase in the asphaltenic content. The temperature did not affect remarkably in the

438 results, concluding that its increase does not improve notably the deasphalting.

439

f
oo
440 Bibliography

r
441 Ahn, S., Shin, S., Im, S.I., Lee, K.B., Nho, N.S., 2016. Solvent recovery in solvent deasphalting

442
-p
process for economical vacuum residue upgrading. Korean J. Chem. Eng. 33, 265–270.
re
443 https://doi.org/10.1007/s11814-015-0146-3
lP

444 Al-Jimaz, A.S., Fandary, M.S., Al-Kandary, J.A., Fahim, M.A., 2005. Liquid-liquid equilibria
na

445 for n-alkanes (C12, C14, C17) + propylbenzene + NMP mixtures at temperatures between

446 298 and 328 K. Fluid Phase Equilib. 231, 163–170.

ur

447 https://doi.org/10.1016/j.fluid.2005.01.012
Jo

448 Alkhaldi, K.H.A.E., Fandary, M.S., Al-Jimaz, A.S., Al-Tuwaim, M.S., Fahim, M.A., 2009.

449 Liquid-liquid equilibria of aromatics removal from middle distillate using NMP. Fluid

450 Phase Equilib. 286, 190–195. https://doi.org/10.1016/j.fluid.2009.09.002

451 Andersen, S.I., Speight, J.G., 1999. Thermodynamic models for asphaltene solubility and

452 precipitation. J. Pet. Sci. Eng. 22, 53–66. https://doi.org/10.1016/S0920-4105(98)00057-6

453 Blanco, A.M., Ortega, J., 1998. Densities and vapor-liquid equilibrium values for binary

454 mixtures composed of methanol + an ethyl ester at 141.3 kPa with application of an

455 extended correlation equation for isobaric VLE data. J. Chem. Eng. Data 43, 638–645.

456 https://doi.org/10.1021/je980012o
457 Brons, G.B., Siskin, M., Wrzeszczynski, K.O., 1993. Upgrading of bitumen asphaltenes by hot

458 water treatment. US5316659A.

459 Bushnell, J.D., Glivicky, A.P., Leighton, M.D., Sankey, B.M., 1977. Simultaneous

460 deasphalting-extraction process. US4125458A.

461 Byrne, F.P., Jin, S., Paggiola, G., Petchey, T.H.M., Clark, J.H., Farmer, T.J., Hunt, A.J., Robert

462 McElroy, C., Sherwood, J., 2016. Tools and techniques for solvent selection: green solvent

463 selection guides. Sustain. Chem. Process. 4, 1–24. https://doi.org/10.1186/s40508-016-

464 0051-z

f
oo
465 Cesar Savastano, Cimino, R., 1993. Process for deasphalting and demetallizing petroleum

r
466 residues. EP0612829A1.

467
-p
Constantinescu, D., Gmehling, J., 2016. Further development of modified UNIFAC
re
468 (Dortmund): Revision and extension 6. J. Chem. Eng. Data 61, 2738–2748.
lP

469 https://doi.org/10.1021/acs.jced.6b00136
na

470 Constantinou, L., Gani, R., 1994. New group contribution method for estimating properties of

471 pure compounds. AIChE J. 40, 1697–1710. https://doi.org/10.1002/aic.690401011

ur

472 Daryasafar, A., Masoudi, M., Kord, S., Madani, M., 2020. Evaluation of different
Jo

473 thermodynamic models in predicting asphaltene precipitation: A comparative study. Fluid

474 Phase Equilib. 514, 112557. https://doi.org/10.1016/j.fluid.2020.112557

475 Energy Institute, 2016. IP 143: Determination of asphaltenes (heptane insolubles) in crude

476 petroleum and petroleum products.

477 Espinosa-Peña, M., Figueroa-Gómez, Y., Jiménez-Cruz, F., 2004. Simulated Distillation Yield

478 Curves in Heavy Crude Oils: A Comparison of Precision between ASTM D-5307 and

479 ASTM D-2892 Physical Distillation. Energy and Fuels 18, 1832–1840.

480 https://doi.org/10.1021/ef049919k

481 Fahim, M.A., 2007. Prediction of asphaltene precipitation from empirical models. Pet. Sci.
482 Technol. 25, 1605–1612. https://doi.org/10.1080/10916460600695504

483 Fandary, M.S., Al-Jimaz, A.S., Al-Kandary, J.A., Fahim, M.A., 2006. Extraction of

484 pentylbenzene from high molar mass alkanes (C14 and C17) by N-methyl-2-pyrrolidone.

485 J. Chem. Thermodyn. 38, 455–460. https://doi.org/10.1016/j.jct.2005.06.012

486 Fredenslund, A., Jones, R.L., Prausnitz, J.M., 1975. Group‐ contribution estimation of activity

487 coefficients in nonideal liquid mixtures. AIChE J. 21, 1086–1099.

488 https://doi.org/10.1002/aic.690210607

489 Fulem, M., Becerra, M., Hasan, M.D.A., Zhao, B., Shaw, J.M., 2008. Phase behaviour of Maya

f
oo
490 crude oil based on calorimetry and rheometry. Fluid Phase Equilib. 272, 32–41.

r
491 https://doi.org/10.1016/j.fluid.2008.06.005

492
-p
Gholami, R., Alvarez-Majmutov, A., Ali, M., Chen, J., 2021. Understanding bitumen partial
re
493 upgrading through process modelling and simulation. Can. J. Chem. Eng. 99, 222–234.
lP

494 https://doi.org/10.1002/cjce.23850
na

495 Gmehling, J., Constantinescu, D., Schmid, B., 2015. Group Contribution Methods for Phase

496 Equilibrium Calculations. Annu. Rev. Chem. Biomol. Eng. 6, 267–292.

ur

497 https://doi.org/10.1146/annurev-chembioeng-061114-123424
Jo

498 Guzmán, J.D., Franco, C.A., Cortés, F.B., 2017. An Enhanced-Solvent Deasphalting Process:

499 Effect of Inclusion of SiO2 Nanoparticles in the Quality of Deasphalted Oil. J. Nanomater.

500 2017. https://doi.org/10.1155/2017/9470230

501 Hiaki, T., Takahashi, K., Tsuji, T., Hongo, M., Kojima, K., 1995. Vapor-Liquid Equilibria of

502 1-Propanol or 2-Propanol with Octane at 101.3 kPa. J. Chem. Eng. Data 40, 274–276.

503 https://doi.org/10.1021/je00017a060

504 Ilyin, S., Arinina, M., Polyakova, M., Bondarenko, G., Konstantinov, I., Kulichikhin, V.,

505 Malkin, A., 2016. Asphaltenes in heavy crude oil: Designation, precipitation, solutions,

506 and effects on viscosity. J. Pet. Sci. Eng. 147, 211–217.

507 https://doi.org/10.1016/j.petrol.2016.06.020

508 Iwase, M., Sugiyama, S., Liang, Y., Masuda, Y., Morimoto, M., Matsuoka, T., Boek, E.S.,

509 Ueda, R., Nakagawa, K., 2018. Development of Digital Oil for Heavy Crude Oil:

510 Molecular Model and Molecular Dynamics Simulations. Energy and Fuels 32, 2781–2792.

511 https://doi.org/10.1021/acs.energyfuels.7b02881

512 Joshi, S.S., Aminabhavi, T.M., Balundgi, R.H., Shukla, S.S., 1990. Densities and Viscosities

513 of Binary Liquid Mixtures of Nitrobenzene with Cyclohexane and N,N-

514 Dimethylformamide. J. Chem. Eng. Data 35, 185–187.

f
oo
515 https://doi.org/10.1021/je00060a027

r
516 Larsen, B.L., Rasmussen, P., Fredenslund, A., 1987. A Modified UNIFAC Group-Contribution

517
-p
Model for Prediction of Phase Equilibria and Heats of Mixing. Ind. Eng. Chem. Res. 26,
re
518 2274–2286. https://doi.org/10.1021/ie00071a018
lP

519 Lee, F., 1986. Use of Organic Sulfones as the Extractive Distillation Solvent for Aromatics
na

520 Recovery. Ind. Eng. Chem. Process Des. Dev. 25, 949–957.

521 https://doi.org/10.1021/i200035a019
ur

522 Lerda, D., Colombani, G., Prevedello, A., 1990. Process for the removal of heavy metals and
Jo

523 asphaltenes from crude petroleum or heavy fractions thereof.

524 Loras, S., Aucejo, A., Muñoz, R., Ordoñez, L.M., 1999. Phase equilibria in the systems 2-

525 methyl-2-propanol + methyl 1,1-dimethylpropyl ether and 2-methylpentane + 2-methyl-

526 2-propanol + methyl 1,1-dimethylpropyl ether. J. Chem. Eng. Data 44, 1163–1168.

527 https://doi.org/10.1021/je990080q

528 Mohammed, I., Mahmoud, M., Al Shehri, D., El-Husseiny, A., Alade, O., 2021. Asphaltene

529 precipitation and deposition: A critical review. J. Pet. Sci. Eng. 197, 107956.

530 https://doi.org/10.1016/j.petrol.2020.107956

531 Nascimento, F.P., Pereira, V.J., Souza, R.P., Lima, R.C.A., Costa, G.M.N., Rosa, P.T.V., Forca,
532 A.F., Vieira de Melo, S.A.B., 2021. An experimental and theoretical investigation of

533 asphaltene precipitation in a crude oil from the Brazilian pre-salt layer under CO2

534 injection. Fuel 284, 118968. https://doi.org/10.1016/j.fuel.2020.118968

535 Nunes, R.C.P., Valle, M.R.T., Reis, W.R.D., Aversa, T.M., Filipakis, S.D., Lucas, E.F., 2019.

536 Model molecules for evaluating asphaltene precipitation onset of crude oils. J. Braz.

537 Chem. Soc. 30, 1241–1251. https://doi.org/10.21577/0103-5053.20190019

538 Painter, P., Veytsman, B., Youtcheff, J., 2015. Guide to asphaltene solubility. Energy and Fuels

539 29, 2951–2961. https://doi.org/10.1021/ef502918t

f
oo
540 Prat, D., Hayler, J., Wells, A., 2014. A survey of solvent selection guides. Green Chem. 16,

r
541 4546–4551. https://doi.org/10.1039/c4gc01149j

542
-p
Prat, D., Wells, A., Hayler, J., Sneddon, H., McElroy, C.R., Abou-Shehada, S., Dunn, P.J.,
re
543 2015. CHEM21 selection guide of classical- and less classical-solvents. Green Chem. 18,
lP

544 288–296. https://doi.org/10.1039/c5gc01008j

na

545 Ramirez-Corredores, M.M., 2017. Bottom of the Barrel Upgrading Technologies, The Science

546 and Technology of Unconventional Oils. https://doi.org/10.1016/b978-0-12-801225-

ur

547 3.00005-x
Jo

548 Robert C. Weast, Jeanette G. Grasselli, D.R.L., 1989. CRC Handbook of Data on Organic

549 Compounds, 2nd Editon, 2nd Editon. ed. CRC Press.

550 Rogel, C., Ovalles, E., 2013. Process for in situ upgrading of a heavy hydrocarbon using

551 asphaltene precipitant additives. US9670760B2.

552 Santos, D.C., Filipakis, S.D., Lima, E.R.A., Paredes, M.L.L., 2019. Solubility parameter of oils

553 by several models and experimental oil compatibility data: implications for asphaltene

554 stability. Pet. Sci. Technol. 37, 1596–1602.

555 https://doi.org/10.1080/10916466.2019.1594288

556 Santos, D.C., Filipakis, S.D., Rolemberg, M.P., Lima, E.R.A., Paredes, M.L.L., 2017.
557 Asphaltene flocculation parameter in Brazilian crude oils and synthetic polar and nonpolar

558 mixtures: Experimental and modeling. Fuel 199, 606–615.

559 https://doi.org/10.1016/j.fuel.2017.03.024

560 Speight, J.G., 2020. Deasphalting and dewaxing, in: The Refinery of the Future. pp. 227–255.

561 https://doi.org/10.1016/b978-0-12-816994-0.00007-5

562 Sun, M., Melibari, F.M., Al-Daous, M.A., 2017. Non-solvent asphaltene removal from crude

563 oil using solid heteropoly compounds. US10800980B2.

564 Tharanivasan, A.K., Svrcek, W.Y., Yarranton, H.W., Taylor, S.D., Merino-Garcia, D., Rahimi,

f
oo
565 P.M., 2009. Measurement and modeling of asphaltene precipitation from crude oil blends.

r
566 Energy and Fuels 23, 3971–3980. https://doi.org/10.1021/ef900150p

567
-p
Wang, J.X., Buckley, J.S., 2001. A two-component solubility model of the onset of asphaltene
re
568 flocculation in crude oils. Energy and Fuels 15, 1004–1012.
lP

569 https://doi.org/10.1021/ef010012l
na

570 Zhao, B., Shaw, J.M., 2007. Composition and size distribution of coherent nanostructures in

571 Athabasca bitumen and Maya crude oil[1] Zhao B, Shaw JM. Composition and size
ur

572 distribution of coherent nanostructures in Athabasca bitumen and Maya crude oil. Energy
Jo

573 and Fuels 2007;21:2795–80. Energy and Fuels 21, 2795–2804.

574

575
576 SUPPLEMENTARY MATERIALS

577 Table 6. Molecules designed for discrete pseudo component-based crude oil model.

f
r oo
-p
re
lP
na
ur
Jo
SARA Molecule Molecule representation
family number
Saturates 1

Saturates 2

Saturates 3

Saturates 4

f
r oo
Saturates 5 -p
re
lP

Saturates 6
na
ur

Saturates 7
Jo

Saturates 8

Saturates 9

Aromatics 10
Aromatics 11

Aromatics 12

Aromatics 13

f
oor
Aromatics 14 -p
re
Aromatics 15
lP
na
ur

Aromatics 16
Jo

Aromatics 17

Aromatics 18

Aromatics 19
Resins 20

Resins 21

Resins 22

f
oor
-p
re
lP

Resins 23
na
ur
Jo

Resins 24
Asphaltenes 25

Asphaltenes 26

f
oor
-p
re
lP
na

Asphaltenes 27
ur
Jo

Asphaltenes 28
 Asphaltenes 29

Asphaltenes 30

f
oor
-p
re
lP

578
na

579
ur
Jo
Highlights

 A discrete pseudocomponent-based crude oil model is developed.

 A deasphalting model is proposed using a liquid-liquid equilibrium extraction and the

modified UNIFAC (Dortmund) model.

 The extraction yield and the asphaltene reduction appear as better parameters for the

solvent selection over the selectivity and partition coefficients.

 Alcohols are found as alternative sustainable solvents for deasphalting use.

f
r oo
-p
re
lP
na
ur
Jo
Declaration of interests

☒ The authors declare that they have no known competing financial interests or personal relationships
that could have appeared to influence the work reported in this paper.

☐The authors declare the following financial interests/personal relationships which may be considered
as potential competing interests:

of
ro
-p
re
lP
na
ur
Jo