Вы находитесь на странице: 1из 16

CHERD-1708; No.

of Pages 16
ARTICLE IN PRESS
chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx

Contents lists available at ScienceDirect

Chemical Engineering Research and Design

journal homepage: www.elsevier.com/locate/cherd

Conceptual design of unit operations to separate


aromatic hydrocarbons from naphtha using ionic
liquids. COSMO-based process simulations with
multi-component “real” mixture feed

V.R. Ferro ∗ , J. de Riva, D. Sanchez, E. Ruiz, J. Palomar


Sección Ingeniería Química, Universidad Autónoma de Madrid, 28049 Madrid, Spain

a b s t r a c t

COSMO-based process simulations using Aspen Plus and Aspen HYSYS were systematically applied to the conceptual
design of the two main unit operations commonly proposed to separate aromatic and aliphatic hydrocarbons with
ionic liquids (ILs) as extracting solvents; the extraction itself and the vacuum distillation for regenerating the ionic liq-
uid. By the first time, multi-component (up to 28 components) “real” mixture feeds were taken into account to model
the naphtha in the process design. Binary model (n-hexane + benzene, n-heptane + toluene, n-octane + ethylbenzene,
n-octane + m-xylene) mixture feeds were also considered to validate the computational procedure. Nine differ-
ent ionic liquids and mixtures of them, (IL–IL) mixtures, were selected as extracting solvents. Ionic liquids were
introduced in the process simulations as (pseudo)components and the COSMOSAC property model was used for
estimating the activity coefficients of the individual components in the mixtures. The information needed to both
create the non-data bank ionic liquid (pseudo)components and to specify the COSMOSAC property model was gath-
ered from COSMO-RS calculations. COSMO-based models exhibited a reasonably good predictability of both the
thermo-physical properties of the pure (hydrocarbons and ionic liquids) components and the LL and VL equilibria of
their mixtures. The performances of extraction and regeneration individual operations were analyzed at different
operating conditions, including the nature of the IL-based extracting agent, the solvent-to-feed ratio, and the hydro-
carbon mixture composition. The present results suggest that COSMO-supported process simulations are capable
of confidently dealing with complex multicomponent mixtures of hydrocarbons and ionic liquids. This opens new
perspectives to improved developments of this process based on ionic liquids and their mixtures.
© 2014 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

Keywords: Aromatic–aliphatic separation; Ionic liquids; Process simulation; Aspen Plus; COSMO models

1. Introduction aromatics content being the extraction suitable for the range
of 20–65 wt% (Meindersma and de Haan, 2008; Meindersma
Separation of aromatics and aliphatic compounds from et al., 2010). Several polar organic solvents and mixtures of
petrochemical feedstocks, having a relevant economical them have been evaluated as possible extracting agents in
and practical interest (Meindersma and de Haan, 2008; this process: sulfolane, N-methyl pyrrolidone, N-formyl mor-
Meindersma et al., 2010), is probably one of the most exten- pholine, ethylene glycols, propylene carbonate, furfural and
sively studied separation processes using ionic liquids (ILs) as others (Gaile et al., 2007; Hamid and Ali, 1996; Kumar and
an alternative to organic solvents. The conventional processes Mohan, 2013) but sulfolane, as a rule, exhibits higher per-
used to remove the aromatics from naphtha depend on the formances than the remaining organic solvents evaluated.


Corresponding author. Tel.: +34 91 497 7607.
E-mail address: victor.ferro@uam.es (V.R. Ferro).
Received 31 May 2014; Received in revised form 17 August 2014; Accepted 5 October 2014
http://dx.doi.org/10.1016/j.cherd.2014.10.001
0263-8762/© 2014 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

Please cite this article in press as: Ferro, V.R., et al., Conceptual design of unit operations to separate aromatic hydrocarbons from
naphtha using ionic liquids. COSMO-based process simulations with multi-component “real” mixture feed. Chem. Eng. Res. Des. (2014),
http://dx.doi.org/10.1016/j.cherd.2014.10.001
CHERD-1708; No. of Pages 16
ARTICLE IN PRESS
2 chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx

Fig. 1 – Simplified conceptual diagram of the process to separate aromatic hydrocarbons from naphtha using ionic liquids
as extracting solvents.

Hence, UOP has developed an industrial process to separate their mixtures with the aromatic hydrocarbons (to be recy-
aromatics from naphtha using this solvent (Asselin, 1984). cled at the beginning of the process) by vacuum distillation.
Nevertheless, extraction with conventional organic solvents Stripping with air has also been proposed as an alternative to
(including sulfolane) seems not to be a definitive option since IL regeneration (Meindersma, 2005; Meindersma et al., 2010).
additional separation operations are required to separate the Other schemes developed for recovering styrene monomer
extraction solvent from both the extract and the raffinate from 4-mebupyBF4 IL (Jongmans et al., 2013) could also be
phases and to purify the solvent, resulting in additional invest- appropriate to regenerate the ILs in processes devoted to sep-
ments and energy consumption (Meindersma and de Haan, arate aromatic hydrocarbons from naphtha.
2008; Meindersma et al., 2010). Application of ILs as solvents Most of the conceptual designs cited above are supported
for extraction operations in the current and other indus- by process simulations which, in general, reveal significant
trial processes is widely accepted as promising since, due computational difficulties. In these and other related works
to their negligible vapor pressure, it is expected to require either the conceptual design of the processes is merely out-
fewer process steps and less energy consumption than extrac- lined and/or simulations of the individual operations (usually
tion with conventional solvents. In particular, solvent recovery LL-extraction) are carried out. Complex flowsheets have been,
seems to be relatively easy (Meindersma and de Haan, 2008; as a rule, not calculated. In particular, the IL regeneration after
Meindersma et al., 2010). Several ILs have been investigated as the primary separation with ILs is not simulated or simply
extracting solvents in this process: imidazolium, pyridinium, mass balances are performed specifying the IL recovery but
pyrrolidinium, etc. A complete review of the ILs investigated without solving the phase equilibrium for the corresponding
in this role over the past 10 years can be found in Meindersma VL operation. Only a few works simulate the IL regeneration
et al. (2010). More recently, low-viscous ILs based on the with one-stage VL separators, shortcut or rigorous distillation
cyano-anions have been proposed for the aromatic extraction columns (Arce et al., 2007c; Jongmans et al., 2013; Pereiro and
(Hansmeier et al., 2010; Meindersma and de Haan, 2012). In Rodriguez, 2009, 2010; Seiler et al., 2004; Shiflett et al., 2011;
addition, mixtures of mutually miscible (Garcia et al., 2012b; Shiflett and Yokozeki, 2006). Regarding the conceptual designs
Potdar et al., 2012) and even immiscible (Arce et al., 2008a) of the aromatic separation from naphtha and the correspond-
ILs have also been studied in order to improve the extracting ing process simulations, another distinctive feature should be
properties in respect to the individual ILs. emphasized: they are habitually limited to the use of binary
Separation of aromatic compounds from naphtha with ILs (aromatic + aliphatic) mixtures to model the real feed. Multi-
has been the subject not only of basic thermodynamic studies component more “real” mixtures are not considered. The most
devoted to elucidate the equilibrium conditions of the process frequent difficulties found in process simulations with ILs in
(see for example references Arce et al., 2007a,b, 2008a,b, 2009; general and in this process in particular are related to severe
Calvar et al., 2011; Corderi et al., 2012; Garcia et al., 2011a,b, errors occurring in the flowsheet calculations (Meindersma,
2012a; Gonzalez et al., 2010; Meindersma et al., 2005, 2006a,c, 2005) and also to the absence of reliable information on the
2010; Pereiro and Rodriguez, 2010 and references therein), but VL equilibrium, the heat capacities of the ILs and their mix-
also of process developments (Meindersma, 2005; Meindersma tures with conventional organic solvents (Arce et al., 2007c),
and de Haan, 2008; Meindersma et al., 2006b) including pilot etc. This second issue hinders the complete specification of
plant scaled experiments (Meindersma, 2005; Meindersma the simulation. Generally speaking, two conflicting problems
et al., 2012). In all the conceptual designs (in spite of their com- are recognizable in the simulations referred here:
plexities) developed so far for this and other similar processes (i) Since ionic liquids are not present in the process simula-
(Arce et al., 2007c; Meindersma, 2005; Meindersma and de tors’ databanks (frequently Aspen Plus and/or Aspen HYSYS),
Haan, 2008; Meindersma et al., 2006b; Pereiro and Rodriguez, the values of parameters for several physical properties of
2009, 2010; Seiler et al., 2004; Shiflett et al., 2011; Shiflett and the ionic liquids are estimated, guessed or simply left out
Yokozeki, 2006) two main operations are considered (Fig. 1): (Meindersma, 2005). To include ILs in process simulations,
(i) aromatic hydrocarbons are separated from their mixtures users are forced to “create” them as non-databank compo-
with aliphatic ones by liquid–liquid extraction using ILs as nents. This is especially complex because the models and
extracting solvents and (ii) further, ILs are regenerated from computational routes implemented in the process simulators

Please cite this article in press as: Ferro, V.R., et al., Conceptual design of unit operations to separate aromatic hydrocarbons from
naphtha using ionic liquids. COSMO-based process simulations with multi-component “real” mixture feed. Chem. Eng. Res. Des. (2014),
http://dx.doi.org/10.1016/j.cherd.2014.10.001
CHERD-1708; No. of Pages 16
ARTICLE IN PRESS
chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx 3

to estimate the properties of pure components have not been properties of pure ILs and their mixtures with conventional
developed for this particular class of compounds. Conse- compounds. Nevertheless, the elevated number and/or the
quently, the simulation crashes if the non-supply or non- system-specific character of the parameters required limit
consistently estimated information of the pure components is their application with a priori and extensive character in the
demanded in any critical calculation step. For “creating” non- search and design of ILs with optimized properties for a given
databank IL components two alternatives have been explored application. Moreover, they are not implemented (or not con-
using Aspen Plus and/or Aspen HYSYS: ILs are introduced veniently implemented) in commercial process simulators to
(a) as conventional components (see a good example in Ref. model processes with ILs. This restricts their use in tasks nor-
Shiflett et al., 2010), or (b) as pseudo-components (Bedia et al., mally addressed by the conceptual design of new processes
2013; Ferro et al., 2012a; Ruiz et al., 2014). In the first one, as a using ILs.
rule, a huge amount of data should be supplied to the program COSMO-RS model (Klamt, 2005; Klamt et al., 2010), needing
because the property methods and models for estimating the only few general (no system-dependent) adjustable param-
unknown properties of the new components fail when work- eters, has shown to yield good qualitative and satisfying
ing with ILs. In most cases these data are not available or they quantitative predictions for the activity coefficients of neu-
are hardly obtainable. The second way allows working with tral compounds in ionic liquids and for binary mixtures of
much less experimental information (in the current situation, ionic liquids and neutral solvents (Diedenhofen and Klamt,
only Andrade’s equation parameters to model IL viscosity) 2010). Therefore, COSMO-RS has become a widely used effi-
since it uses the a priori COSMO-based methods to estimate cient tool for the screening/selection/design of ILs with desired
the required pure solvent and mixture properties (see below). properties for specific applications (Ab Manan et al., 2009;
(ii) NRTL and UNIQUAC activity models are often used Gutierrez et al., 2012; Palomar et al., 2008, 2009). COSMO-RS
to calculate the activity coefficients of the components in has been applied with reasonable success to predict densi-
the mixtures of ILs and organic compounds, their binary ties (Palomar et al., 2007), vapor pressures and vaporization
interaction parameters being correlated from experimental enthalpies (Diedenhofen et al., 2007; Schröder and Coutinho,
equilibrium data (see, for example, Refs. Arce et al., 2007c; 2014) of pure ILs. It has also turned out well in predicting:
Garcia et al., 2012b; Kato and Gmehling, 2005; Pereiro and (i) the activity coefficients at infinite dilutions (Banerjee and
Rodriguez, 2010; Seiler et al., 2004; Shiflett and Yokozeki, Khanna, 2006; Diedenhofen et al., 2003; Gonzalez-Miquel et al.,
2006 and other related ones). However, these models exhibit 2012; Liu et al., 2014) in mixtures of hydrocarbons, alcohols and
a strong system- and equilibrium-dependent character, i.e. polar organics in several ILs, (ii) gas solubilities in ILs and the
the interaction parameters obtained by correlation of the corresponding Henry’s law constants (Ab Manan et al., 2009;
LLE data for a specific system could hardly be used to solve Bedia et al., 2013; Fallanza et al., 2013; Gonzalez-Miquel et al.,
either the VL equilibrium of the same system or the LL 2011, 2012; Liu et al., 2014; Palomar et al., 2011a,b), (iii) the
equilibrium of other mixtures. Furthermore, VLL equilib- LL equilibria of binary, ternary and quaternary mixtures of
rium data are extremely scarce in multi-component mixtures organic compounds and ILs (Banerjee et al., 2008; Domanska
including ILs. Additionally, it should be taken into account et al., 2006; Ferreira et al., 2011, 2012; Freire et al., 2007, 2008;
that creating new IL components and NRTL/UNIQUAC prop- Potdar et al., 2012) and, (iv) the VL equilibria of binary mix-
erty packages for multi-component (aliphatic + aromatic + ILs) tures of conventional organic solvents and ILs (Banerjee et al.,
process simulations by using experimental data repre- 2006, 2008; Ferro et al., 2012a; Freire et al., 2007, 2008; Kato and
sents a considerable amount of previous work to be done Gmehling, 2005; Ruiz et al., 2014). Additionally, COSMO-based
because the corresponding binary interaction parameters models (Klamt, 1995; Lin and Sandler, 2002; Mathias et al.,
for all the component pairs considered have to be deter- 2002) have been wholly implemented in Aspen Plus since its
mined. This is almost experimentally and economically version 12.1 under the general denomination of COSMOSAC
unaffordable. Hence, as it was previously mentioned, pro- property model (Aspen Technology, 2003). Taking advantage
cess simulations on the aromatic and aliphatic hydrocarbon from this development, we have fruitfully used the COSMO-
separation with ILs model the “real” feed mixtures as based calculations with the Aspen Technology programs for
simply binaries (n-hexane + benzene), (n-heptane + toluene), carrying out different tasks directly related to the concep-
(n-octane + ethylbenzene), (n-octane + m-xylene), etc. tual design of several processes involving organic compounds
Recently, the most significant achievements on molecular (Ferro et al., 2012b) and ILs (Bedia et al., 2013; Ferro et al., 2012b;
thermodynamics of fluid phase equilibria as well as several Ruiz et al., 2014). Based on the technical, energy and eco-
other theoretical and empirical approaches have been suc- nomic analysis derived from these results the selection/design
cessfully applied to reproduce the experimental data on the of optimized ILs for specific uses is improved. Certainly, the
thermodynamic properties of pure ILs and mixtures that con- integration of the COSMO-based results into Aspen Plus has
tain them (Paduszynski and Domanska, 2012; Vega et al., 2010). been developed by other authors (Cadoret et al., 2009; Tian
For example, the electrolyte-NRTL (eNRTL) (Simoni et al., 2008) et al., 2010) avoiding the direct use of the COSMOSAC property
and the NRTL-SAC (Chen et al., 2008) models have been used model already implemented in the process simulator. How-
to correlate values of infinite-dilution activity coefficients for ever, the direct integration seems to be more efficient and less
organic compounds in ionic liquids and further to predict time-consuming.
the LL and VL phase behavior of the corresponding mixtures. In this work, the conceptual design of the individual unit
Likewise, the statistical associating fluid theory (SAFT)-based operations (Fig. 1) used to remove aromatic hydrocarbons from
models have been strongly recommended due to their phys- naphtha was performed with the support of Aspen Plus and
ical background (Ji et al., 2012; Paduszynski and Domanska, Aspen HYSYS process simulations by using the COSMOSAC
2012) and used to model SL, LL, VL and GL equilibria in systems property model. The operating conditions at both the extrac-
containing ILs (Ji et al., 2012; Kroon et al., 2006; Paduszynski tion and the flash distillation process units were explored as a
and Domanska, 2012; Rahmati-Rostami et al., 2011). These function of the IL and hydrocarbons nature. The hydrocarbon
models provide a realistic description of the thermodynamic mixture composition was also taken into account. Here,

Please cite this article in press as: Ferro, V.R., et al., Conceptual design of unit operations to separate aromatic hydrocarbons from
naphtha using ionic liquids. COSMO-based process simulations with multi-component “real” mixture feed. Chem. Eng. Res. Des. (2014),
http://dx.doi.org/10.1016/j.cherd.2014.10.001
CHERD-1708; No. of Pages 16
ARTICLE IN PRESS
4 chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx

for the first time, multi-component (aliphatic + aromatic) of the fluid. The TURBOMOLE suit (TmoleX version 3.1) of
mixtures (up to 28 components) were used to model the “real” programs (Schäfer et al., 2000) and the program package
naphtha fed to the process. Multi-component hydrocarbon COSMOthermX, version C30 1201 (Diedenhofen et al., 2012)
mixtures were composed by C4 –C10 aliphatic hydrocarbons, were used, respectively, in all the quantum chemical and
cycloalkanes, and by benzene, toluene, ethylbenzene and COSMO-RS calculations. The COSMOthermX calculations use
xylenes. Binary (aromatic + aliphatic) mixtures were also the implicit BP TZVP C30 1201 parametrization.
considered in order to validate the results obtained for
multi-component (aromatic + aliphatic) mixtures with those 2.2. Creating the property package
obtained in previous conceptual designs and thermodynamic
studies carried out on simpler binary mixtures. Ongoing from IL (non-databank) components were introduced in the Aspen
binary to multi-component (aromatic + aliphatic) mixtures Plus process simulations as pseudo-components. Following
had the primary intention to give a more real description of the procedure implemented in the program, their molecu-
the process, and consequently, a more realistic conceptual lar weights, normal boiling temperatures and densities were
design. On the other hand, nine individual ILs with different specified by the user. The remaining unknown properties of
cations and anions as well as from binary to quaternary the pseudo-components were estimated by the methods and
(IL–IL) systems were used as extracting solvent in the pro- models used implicitly in Aspen Plus (Aspen Technology, 2011)
cess. The ILs selected to the present study were composed for this class of components by performing a Property Estima-
by cations: 1-alkyl-3-methylimidazolium (Cn mim+ , being tion calculation (at the basis environment of the simulator).
C1 ≡ methyl(m), C2 ≡ ethyl(e), C4 ≡ butyl(b), C6 ≡ hexyl(h) Additionally, the coefficients of the Andrade’s equation for the
and C8 ≡ octyl(o)) and anions: triflates (CF3 SO3 − , shortly viscosity-to-temperature dependency were also specified by
TfO− ) and bistriflamides (N(CF3 SO2 )2 − , shortly NTf2 − ). 3- the user. These coefficients were obtained by regression of vis-
Methyl-N-butylpyridinium dicyanamide (3-mebypyDCN) and cosity vs. temperature experimental data (de Riva et al., 2014).
4-methyl-N-butylpyridinium tetrafluoroborate (4-mebupyBF4 ) The aromatic and aliphatic hydrocarbons were selected as
were also considered. conventional components from the Aspen Plus database. The
The paper is organized into two main parts: Firstly, COSMO-SAC (Lin and Sandler, 2002) equation as implemented
the information required to generate non-databank IL by default in Aspen Plus (under the general denomination
(pseudo)components and to specify the COSMOSAC property COSMOSAC) was selected to estimate the activity coefficients
model in the Aspen Plus basis environment was derived with of the components in the mixtures. Original parametrization
the COSMO-RS method. Then, it was evaluated and validated of the COSMO-SAC model (Lin and Sandler, 2002) was used.
the capacity of the COSMOSAC property model (as specified The molecular volumes and ␴-profiles needed to specify the
in this work) to predict the properties of the individual ILs COSMOSAC property model for the ILs were provided by the
selected, the component activity coefficients in binary mix- user, whereas those corresponding to the organic solvents
tures of organic components and ILs and, finally, the LL and were directly taken from the database of Aspen Plus. The infor-
VL equilibria in Aspen Plus and Aspen HYSYS. Secondly, the mation used for specifying both the IL pseudo-components
individual extraction and distillation (Fig. 1) unit operations and the COSMOSAC property model were attained from the
were conceptually designed for the cases of both binary and previous COSMO-RS calculations. The property package cre-
multi-component (aromatic + aliphatic) mixtures feed. The ated in Aspen Plus was imported from Aspen HYSYS which
performances of the individual operations were evaluated in supports some of the process simulations performed in this
terms of parameters commonly used in Process Engineer- work. The version 8.4 of the Aspen ONE program package was
ing (see Section 2). The results presentation evolves stepwise used.
from the simplest separation of binary (aromatic + aliphatic)
mixtures with single IL to multi-component hydrocarbon mix- 2.3. Validating the property system created
tures with binary to quaternary (IL–IL) mixtures. At the end of
this paper, an example of how the results of the conceptual In order to evaluate the predictive capacity of the current
design of the individual unit operations enrich the selection computational procedure, several simulations (flowsheet
of the best IL-based extracting agent was discussed. run type at the simulation environment of the simula-
tor) focused calculating some selected properties of the
pure ILs (density, viscosity, heat capacity and heat of
2. Computational details vaporization) and their mixtures with aromatic and/or
aliphatic hydrocarbons (component activity coefficients at
2.1. Quantum chemical calculations infinite solutions, VL and LL equilibria) were performed.
LLE diagrams of ternary (aromatic + aliphatic + IL) and
In COSMO and COSMO-RS calculations (Klamt, 2005; Klamt quaternary (aromatic + aliphatic + IL1 + IL2 ) mixtures were
et al., 2010) the molecular structure of the selected ILs was calculated. Ternary mixtures consist of (n-hexane + benzene),
described by the ion-paired (molecular-like) model. These (n-heptane + toluene), (n-octane + ethylbenzene) and (n-
structures were obtained by geometry optimization at BP- octane + m-xylene) with 3-mebupyDCN, 4-mebupyBF4 ,
TZVP computational level considering solvent effects through C2,4 mimNTf2 and emimTfO ILs. The LL equilib-
the COSMO continuum solvation model (Andzelm et al., 1995; rium of a single quaternary mixture composed by
Klamt, 1995). Vibrational frequency calculations to confirm the (n-hexane + benzene + emimMeSO4 + emimEtSO4 ) was cal-
presence of an energy minimum were performed. After opti- culated. VLE diagrams of (benzene + C1,2,4 mimNTf2 ILs) and
mization, a file with extension.cosmo, containing the ideal (toluene + C1,2,4,6,8 mimTfO) were also predicted. The obtained
screening charges on the molecular surface computed by the results, when possible, were compared to the experimental
COSMO methodology was generated. It was used in further data published elsewhere (see below for references). Two
COSMO-RS calculations of the thermo-physical properties statistics were employed for quantifying the goodness of

Please cite this article in press as: Ferro, V.R., et al., Conceptual design of unit operations to separate aromatic hydrocarbons from
naphtha using ionic liquids. COSMO-based process simulations with multi-component “real” mixture feed. Chem. Eng. Res. Des. (2014),
http://dx.doi.org/10.1016/j.cherd.2014.10.001
CHERD-1708; No. of Pages 16
ARTICLE IN PRESS
chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx 5

the predictions in respect to the experimental values: (i) the


Table 1 – Composition of the multi-component
relative absolute average deviation (RAAD) (also known as (aliphatic + aromatic) mixtures used in this work for
average absolute relative deviation) and (ii) the root mean modeling the “real” naphtha feed to the process to
square deviation (RMSD). They are defined as follows: remove aromatic hydrocarbons.

100  |xi
Calc
− xi
Exp
| Component wt%
RAAD (%) = ·
n Exp
xi Mixture 1a
i
Benzene 1.8
where x is any variable considered and n is the number of Toluene 3.3
values to be compared. Ethylbenzene 2.0
m-Xylene 2.9
⎛  Exp 2
⎞1/2 n-Hexane 43.2
Calc
(xilm − xilm )
⎜ ⎟ n-Heptane 15.8
RMSD (%) = 100 · ⎜ ⎟
i l m
n-Octane 31.0
⎝ 6·k ⎠
Mixture 2b
n-Butane 1.5
where x is the mole fraction and the subscripts i, l and m desig- i-Pentane 4.2
nate, respectively, the components, phases and tie lines. The n-Pentane 10.3
Cyclopentane 1.5
value k designates the number of tie lines considered and 6
2,3-Dimethylbutane 0.8
refers to the number of compositional variables per tie line. 2-Methylpentane 6.0
3-Methylpentane 4.0
2.4. Process simulations n-Hexane 8.6
Methylcyclopentane 4.1
300 ton/h of an (aliphatic + aromatic) mixture containing c.a. Benzene 1.8
Cyclohexane 2.8
10 wt% of aromatic(s) hydrocarbon(s) was always fed to the
2-Methylhexane 2.8
extraction column. Calculation of the extractive unit operation
3-Methylhexane 3.8
with binary (aliphatic + aromatic) mixture feeds were carried n-Heptane 4.4
out only with the aim to: (i) validate the computational pro- Methylcyclohexane 4.8
cedure, since the majority of the experimental information Toluene 3.0
on the current process has been obtained for this kind of 2-Methylheptane 2.4
systems and (ii) gain support to further interpret the results 1,3-Dimethylcyclohexane 7.0
n-Octane 5.4
obtained with real multi-component mixtures. Indeed, binary
Ethylcyclohexane 2.0
mixtures do not have practical significance in potential real 2,6-Dimethylheptane 1.9
processes. Two multi-component mixtures (1 and 2) with the Ethylbenzene 2.0
compositions given in Table 1 were used in this work as fresh p-Xylene 1.9
feed to the separation process. Multi-component mixture 1 (7 3-Methyloctane 2.9
components) is composed by some representative aromatic o-Xylene 1.0
n-Nonane 2.6
and aliphatic hydrocarbons which allows more realistic mod-
n-Decane 3.0
eling of the process than the binary (aromatic + aliphatic)
i-Decanes 4.0
ones while avoiding several complexities related to the pres-
ence of the lighter (C4 - and C5 -) hydrocarbons (see further). a
Meindersma (2005).
b
Multi-component mixture 2 (28 components) has a FCC-like Meindersma and de Haan (2008).
composition.
The extractor was simulated by the rigorous model of
the counter-current liquid–liquid extraction (Extract) in Aspen (aliphatic + aromatic) mixtures using individual ILs as extract-
Plus. Operating temperature and pressure were always set to ing solvents, (ii) multi-component (aliphatic + aromatic) mix-
40 ◦ C and 1 atm. In all the simulations the extractor was calcu- tures (1 and 2) with individual ILs and finally, (iii) multi-
lated considering a 98.0% of aromatic hydrocarbon(s) recovery component (aliphatic + aromatic) mixtures (1 and 2) with
through the extract (design specification). In order to primar- binary to quaternary (IL–IL) mixtures as extracting agents.
ily evaluate the extractive unit operation, the (S/F) molar ratios The two-outlet VL flash model in Aspen Plus (Flash2)
needed to reach the target recovery were computed for differ- was used for modeling the IL regenerator. The operating
ent numbers of equilibrium stages (N). Based on these data, temperature at the regenerator was selected in order to
the (S/F) molar ratio for N → ∞ ((S/F)min. ) was determined for avoid the thermal decomposition of the regenerated IL(s). 3-
the systems assessed. Afterwards, the operation performance mebupyDCN and 4-mebupyBF4 begin to decompose at c.a.
was detailed evaluated for the most promising studied ILs in 230–250 ◦ C, whereas NTf2 and TfO ILs decompose at 320–330 ◦ C
a 12 equilibrium stages column. This allows the direct com- (Crosthwaite et al., 2005; Fredlake et al., 2004; MacFarlane
parison of the present results with those published elsewhere et al., 2002; Meindersma and de Haan, 2012). In all the process
(Meindersma, 2005; Meindersma and de Haan, 2008). Perfor- simulations of the IL regenerator, the following design spec-
mances of the extractive unit operation with different ILs were ifications were set: (i) the IL recovered was 99.0 mol% pure,
evaluated in terms of solvent consumption, (S/F), and prod- (ii) the mixture to be separated was fed at the bubble point
uct recoveries and/or purities (aromatics in the extract and temperature for the corresponding operating pressure, and (iii)
aliphatics in the raffinate). Recoveries or purities were used operating temperature was fixed to 230 ◦ C for 3-mebupyDCN
casuistically depending on the relationship studied. Analysis and 4-mebupyBF4 ILs or 320 ◦ C for TfO- and NTf2 -ILs. The next
of the extractive process operation was conducted sequen- two process variables were used to evaluate the IL regenera-
tially throughout the following systems: (i) binary model tion process operation: (i) the operating (equilibrium) pressure

Please cite this article in press as: Ferro, V.R., et al., Conceptual design of unit operations to separate aromatic hydrocarbons from
naphtha using ionic liquids. COSMO-based process simulations with multi-component “real” mixture feed. Chem. Eng. Res. Des. (2014),
http://dx.doi.org/10.1016/j.cherd.2014.10.001
CHERD-1708; No. of Pages 16
ARTICLE IN PRESS
6 chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx

at which the target separation was reached for the specified


temperature and (ii) the calculated heat duty (in kJ per kg Heat vaporiz. n=6
of IL regenerated) of the operation. This treatment is simi-
lar to the one employed in our previous study (Ferro et al.,
2012a). Four different situations were studied in relation to Heat capac. n=9
the IL regeneration: (i) individual ILs were regenerated from
their binary mixtures with different aromatic hydrocarbons.
This is a hypothetical circumstance because it essentially Viscosity n = 11
assumes a perfect separation in the extractor. However, it
can give a general outlook about the influence of the IL and
aromatic nature on the operating conditions and energetic Density n=8
requirements of the IL regeneration. 15 wt% aromatic mixtures
were supposed every time, (ii) individual ILs were regener- 0 2 4 6 8 10
ated from their ternary mixtures with (aromatic + aliphatic) RAAD, %
hydrocarbons. The presence of aliphatic hydrocarbons in the
extract is not desirable not only because it reduces their Fig. 2 – Relative absolute average deviation (RAAD) of some
recovery through the raffinate and the aromatic purity in the selected thermo-physical and transport properties for the
extract, but also because it represents an extra energetic load pure ILs considered in the present work. The number of
in the IL regeneration (see below). However, it is unavoidable pure ILs taken for the statistical analysis is shown inside
when multi-component “real” (aromatic + aliphatic) mixtures the bars.
are fed to the process because the previous extraction does
not completely separate aromatic and aliphatic hydrocar- here were similar to those obtained by Banerjee et al.
bons. Hence, the consequence of the aliphatic hydrocarbon (2008) with COSMO-RS model for ternary and quater-
presence in the mixture feed to the IL regeneration on both nary systems containing ethanol, hexane, hexene and
the operating conditions and the energetic needs of the IL hxmimBF4 and/or emimEtSO4 ILs. They were also similar
recovery were analyzed previously to discuss the IL regener- to those obtained by Shah and Yadav (2012) for (aro-
ation from multi-component (aromatic + aliphatic) mixtures. matic + aliphatic + IL) ternary mixtures using the COSMO-SAC
In these studies, 2 wt% of n-hexane, n-heptane or n-octane model and Potdar et al. (2012) for ternary and quaternary (aro-
was always added to mixtures of different ILs and aromatic matic + aliphatic + IL(s)) mixtures. Predicted LLE diagrams for
hydrocarbons (benzene, toluene, ethylbenzene and m-xylene). the ternary (hexane + benzene + C(2,4,6,8,10) mimNTf2 ILs) suc-
Aliphatic addition was made in such a way that the flows of the cessfully reproduced the increase, experimentally observed,
remaining components of the mixture (aromatic and IL) were of the mutual solubility of both hydrocarbons and the ILs
kept constant; (iii) individual ILs were regenerated from the as the length of the alkyl chain becomes longer (Arce et al.,
extract resulting from the extraction of multi-component real 2007b), however, the region of mutual solubility was some-
mixtures 1 and 2, and (iv) (IL–IL) mixtures were regenerated what overestimated in respect to the experiments. This result
from the real extracts (as in the previous situation). can be understood from the previous ones reported here on
the estimation of the activity coefficients of the individual
components in binary mixtures with ILs. At the same time,
3. Results and discussion they were responsible for certain inaccuracies obtained (see
next paragraphs) in the estimation of the aromatic purities
3.1. Property prediction of pure ILs and their mixtures
with the organic compounds
6
Aliphatic
Densities, viscosities, heat capacities and heats of vaporiza-
5 Aromatic
tion predicted by Aspen Plus and Aspen HYSYS simulations
Cyclic aliphatic
Experimental ln(γ)

(with the COSMOSAC property model) for the pure ILs used
4
in this work agree reasonably well (Fig. 2) with experimental
reported data. Computed RAADs (in %) for all the ILs included
3
in this work are 2.4, 5.0, 9.8 and 8.7, respectively, for the
mentioned properties. These accuracies are similar to those
2
obtained by us in other related works where a similar com-
putational procedure was used (Bedia et al., 2013; Ferro et al.,
1
2012a; Ruiz et al., 2014).
Current calculations also predicted reasonably well (Fig. 3) 0
the component activity coefficients at infinite dilution
in binary (aliphatic or aromatic hydrocarbons + ILs) mix- 0 1 2 3 4 5 6
tures as derived by comparison with experimental data Calculated ln(γ)
(Domanska and Marciniak, 2008). Nevertheless, activity coef-
ficients of the aromatic and aliphatic hydrocarbons are Fig. 3 – Experimental vs. calculated activity coefficients at
slightly under- and over-estimated, respectively. Relatively infinite dilution for the organic component in binary
small errors were also obtained in the prediction of the (organic + IL) mixtures. Results are given in logarithmic
LL equilibria in ternary (aromatic + aliphatic + IL) and quater- scale. T = 308 K.
nary (aromatic + aliphatic + IL1 + IL2 ) mixtures (Figs. 4 and 5 Experimental data were taken from Domanska and
and Table 2). The accuracies of the predictions reached Marciniak (2008).

Please cite this article in press as: Ferro, V.R., et al., Conceptual design of unit operations to separate aromatic hydrocarbons from
naphtha using ionic liquids. COSMO-based process simulations with multi-component “real” mixture feed. Chem. Eng. Res. Des. (2014),
http://dx.doi.org/10.1016/j.cherd.2014.10.001
CHERD-1708; No. of Pages 16
ARTICLE IN PRESS
chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx 7

0.00 0.00
1.00 1.00
A Experimental
Calculated
0.25 0.25
0.75 0.75

e
an

Be
e

ex
n

To

nz
pta

n-H
lu

en
0.50
e

0.50

en
0.50
n-H

0.50

e
e
0.75
0.25 0.75
0.25

1.00
0.00 1.00
0.00 0.25 0.50 0.75 1.00 0.00
0.00 0.25 0.50 0.75 1.00
emimTfO
(emimMeSO4+emimEtSO4)

Fig. 5 – Experimental (solid line) and Aspen Plus (dotted


0.00 line) calculated (with COSMOSAC property model) LL
1.00 mmim
B emim equilibria (wt%) of the quaternary mixtures
(n-hexane + benzene + emimMeSO4 + emimEtSO4 ).
bmim
hxmim Experimental data were taken from Ref. Potdar et al. (2012).
e

0.25
IL mixture contains 80 mol% emimEtSO4 . T = 298.15 K.
t an

0.75 omim
To
ep

lue
n-H

ne

containing aromatic hydrocarbons and ILs, the main compo-


0.50 nents of the mixtures to be treated in the IL-regenerator, were
0.50
satisfactorily well described for the characteristic composi-
tions considered in this work. However, for certain individual
0.75 (hydrocarbon + IL) systems and hydrocarbon molar fractions
0.25
above 0.5 some disagreements with experimental data have
been observed (Ferro et al., 2012a).

1.00
0.00 3.2. Simulation of the extracting unit operation
0.00 0.25 0.50 0.75 1.00
CnmimTfO Separation of binary (aromatic + aliphatic) mixtures with indi-
vidual ILs
Fig. 4 – (A) Experimental and Aspen Plus calculated (with In order to recover 98.0% of the aromatic hydrocarbon from
COSMOSAC property model) LL equilibria (wt%) for the the binary (aromatic + aliphatic) mixtures with different indi-
ternary mixture (n-heptane + toluene + emimTfO IL). vidual ILs (Fig. 7, Table 3), (S/F)min. in the interval from 0.6 to
Experimental data were taken from Garcia et al. (2011a). (B) 4.2 were required. A reasonable low number of 4–6 equilib-
Predicted LLE of ternary mixtures rium stages also guarantied the target separation but using
(n-heptane + toluene + C(1,2,4,6,8)mimTfO ILs). T = 313.15 K. considerably higher solvent flows (Fig. 7). Taking into consider-
ation the elevated prices of the ILs (Meindersma and de Haan,
in the extract. The increase of the miscibility region with the 2008), the (S/F) factor has noticeable significance in the selec-
length of the alkyl chain of the cation was also predicted for tion of the extracting solvent being desirable to operate as
Cn mimTfO ILs (Fig. 4B) while preserving a high immiscible near as possible to (S/F)min. condition. This is viable, for exam-
region even for the ILs with the longest alkyl chains. ple, in a 12-stage column which represent, simultaneously,
Predicted VLE for binary mixtures (ben- a reasonably small equipment. Under any condition where
zene + C(1,2,4) mimNTf2 ILs) also matched reasonably well the aromatic target recovery was achieved, aliphatic hydro-
with the experimental ones (Fig. 6A) for all the composi- carbons were always separated throughout the raffinate with
tion interval. The computed MRDs between calculated and purities over 99.7–99.8 wt% and recoveries higher than 80.0%
experimental values were, respectively, 4.8%, 5.4% and 5.4%. (Table 3). Moreover, the aromatic (free-of-solvent) purity in
Fig. 6B shows the predicted VLE of the binary mixtures the extract was about 35–65 wt% for all the systems consid-
(toluene + C(1,2,4,6,8) mimTfO ILs) in the region of toluene molar ered. Loses of IL through the raffinate were always lower than
fractions below 0.25. It is interesting to note that, in both the 0.001%. For example, in toluene separation from its mixtures
cases shown in Fig. 6, the vapor pressure of the mixtures with n-heptane using 4-mebupyBF4 as extracting agent in a 12-
for any composition decreased as the length of the cation stage LL extractor, an (S/F)min. molar ratio of 2.64 guarantees a
alkyl chain becomes longer. The accuracies reached in these toluene recovery of 98.0% in the extract and a heptane purity of
predictions are similar to those obtained previously (Banerjee 99.8 wt% in the raffinate. This results are very similar to those
et al., 2006, 2008; Ferro et al., 2012a; Freire et al., 2007, 2008; obtained by Meindersma (2005) who modeled the extraction
Kato and Gmehling, 2005; Ruiz et al., 2014). From the present with NRTL activity model the binary interaction parame-
results it could be concluded that VLE behavior of mixtures ters being regressed from experimental LL equilibrium data.

Please cite this article in press as: Ferro, V.R., et al., Conceptual design of unit operations to separate aromatic hydrocarbons from
naphtha using ionic liquids. COSMO-based process simulations with multi-component “real” mixture feed. Chem. Eng. Res. Des. (2014),
http://dx.doi.org/10.1016/j.cherd.2014.10.001
CHERD-1708; No. of Pages 16
ARTICLE IN PRESS
8 chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx

Table 2 – RMSD between experimental and Aspen Plus/Aspen HYSYS (with COSMOSAC property model) calculated LLE
diagrams of some ternary and quaternary (aliphatic + aromatic + IL(s)) mixtures studied in the present work.
Hydrocarbon mixture IL T (K) No. tie lines RMSD (%)

313.2 8 3.86
4-mebupyBF4 a
333.2 8 3.44
298.15 7 6.53
(n-Hexane + benzene) emimNTf2 b
333.15 7 6.79
bmimNTf2 c 298.15 7 8.20
emimMeSO4 + emimEtSO4 d 298.15 8 4.80

3-mebupyDCNe 313.15 6 2.88


f
313.2 14 4.09
4-mebupyBF4
(n-Heptane + toluene) 348.2 13 2.62
emimTfOg 313.2 11 1.22
emimNTf2 h 298.15 7 26.6

4-mebupyBF4 a 313.2 8 1.95


(n-Octane + ethylbenzene)
4-mebupyBF4 a 348.2 8 1.34

313.2 10 2.49
(n-Octane + m-xylene) 4-mebupyBF4 a
348.2 10 2.55

Experimental data for comparison are taken from references:


a
Meindersma et al. (2006a).
b
Arce et al. (2007a).
c
Arce et al. (2007b).
d
Potdar et al. (2012). (IL–IL) mixture has 80 mol% of emimEtSO4 .
e
Hansmeier et al. (2010).
f
Meindersma et al. (2006c).
g
Garcia et al. (2011a).
h
Arce et al. (2008b).

However, the toluene purity obtained here (49.5 wt%) was mixtures (Table 3) as reflected in the following evidences: (i)
lower than that one reported by Meindersma (82.3 wt%). This the (S/F)min. ratio decreased in the sequence: mmimTfO > 4-
later could be explained by the quality of the activity coeffi- mebupyBF4 > mmimNTf2 ≈ 3-mebupyDCN (Table 4) similarly
cient predictions discussed in the previous paragraph. to the individual binary mixtures (Table 3), (ii) for a spe-
(S/F)min. molar ratios calculated in this paper showed cific extracting solvent, 4-mebupyBF4 for instance, aliphatic
an inverse trend respect to the partition coefficients hydrocarbons in the raffinate yielded 98.8 wt% purity and
obtained in laboratory experimental determinations of the 85.8% recovery, whereas the overall aromatic free-of-solvent
LLE (Arce et al., 2007b; Garcia et al., 2011a; Meindersma purity was 72.9% when aromatics were separated from multi-
et al., 2006a) for series of both the IL and the (aro- component mixture 1 (Table 4). This result was also similar to
matic + aliphatic) mixtures. Thus, recovering 98.0% of the results obtained for binary mixtures (n-heptane + toluene),
the aromatic from its mixtures with the aliphatic hydro- for example, using the same IL flow (Table 3), (iii) individual
carbons needed an increase in solvent flows in the aromatic recoveries (in the extract) from multi-component
sequence (n-hexane + benzene) < (n-heptane + toluene) < (n- mixture 1, using a (S4-mebupyBF4 /F) molar ratio 3.5 in a 12-
octane + ethylbenzene) ≈ (n-octane + m-xylene) for any IL stage column follow the sequence: benzene (100%) > toluene
used as extracting solvent (Table 3). On the other hand, (99.9%) > ethylbenzene (97.7%) > m-xylene (93.2%), which is
(S/F)min. molar ratio decreases as the length of the alkyl chain comparable to that one obtained for binary mixtures con-
bonded to the imidazolium cation increases. It correlated taining the same aromatics. When multi-component mixture
well with the partition coefficients obtained in experimental 2 was fed to the extraction column, calculated (S/F)min.
determinations of the LLE for the Cn mimNTf2 (n = 2, 4, 6, 8) IL molar ratios and aliphatic recoveries were very similar to
series (Arce et al., 2007b) (Table 3). Furthermore, the aliphatic those obtained for multi-component mixture 1. However, the
recovery in the raffinate (and correspondingly, the aromatic free-of-solvent purity of the aromatic fraction was signifi-
free-of-solvent purity in the extract) decreased as the length cantly reduced (Table 4) for all the ILs evaluated because the
of the alkyl chain increased. Similar results were obtained for relatively high contents in light aliphatic hydrocarbons of the
(n-octane + ethylbenzene) and (n-octane + m-xylene) binary feed respect to mixture 1.
mixtures with different ILs.

3.4. Separation of (aromatic + aliphatic)


3.3. Separation of multi-component multi-component mixtures with (IL–IL) mixtures
(aromatic + aliphatic) mixtures with individual ILs
Separation of aliphatic and aromatic hydrocarbons from
Results (Table 4) of the aromatic and aliphatic separation the multi-component mixture 1 with (IL–IL) mixtures of
with some individual ILs when multi-component mixtures the ILs 4-mebupyBF4 , mmimTfO, and mmimNTf2 showed
1 and 2 were fed to a 12-stage extraction column (fixing different behaviors depending on the ILs combined (Fig. 8).
overall 98.0% recovery for the aromatics) were consistent Binary (IL–IL) mixture (mmimTfO(1) + mmimNTf2 (2)) and
with the results obtained for the corresponding binary reciprocal (IL–IL) binary (4-mebupyBF4 (1) + mmimTfO(2))

Please cite this article in press as: Ferro, V.R., et al., Conceptual design of unit operations to separate aromatic hydrocarbons from
naphtha using ionic liquids. COSMO-based process simulations with multi-component “real” mixture feed. Chem. Eng. Res. Des. (2014),
http://dx.doi.org/10.1016/j.cherd.2014.10.001
CHERD-1708; No. of Pages 16
ARTICLE IN PRESS
chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx 9

Table 3 – Results of the Aspen Plus simulations (with COSMOSAC property model) in the conceptual design of the
extraction unit operation for (aliphatic + aromatic) binary mixtures with ionic liquids.a Design specification in the all
simulations: to recover 98.0% of the aromatic component in the extract. Feed: 10 wt%-aromatic mixture. Extractor
operating conditions: P = 1 atm, T = 40 ◦ C.
Hydrocarbon mixture IL (S/F)min. (M)b Aliphatic recoveryc,d (wt%)

3-mebupyDCN 0.88 87.0


4-mebupyBF4 1.33 90.0
mmimNTf2 0.72 88.7
emimNTf2 0.63 86.4
n-Hexane benzene
mmimTfO 1.37 92.2
emimTfO 1.23 91.2
bmimTfO 0.85 88.1
hxmimTfO 0.73 84.8
omimTfO 0.55 81.3

3-mebupyDCN 1.49 86.5


4-mebupyBF4 2.41 88.9
mmimNTf2 1.26 88.7
emimNTf2 1.05 86.5
n-Heptane toluene mmimTfO 3.58 93.2
emimTfO 2.65 91.5
bmimTfO 1.51 87.4
hxmimTfO 1.23 84.2
omimTfO 0.90 81.0

3-mebupyDCN 2.17 87.4


4-mebupyBF4 3.79 89.0
mmimNTf2 1.93 89.0
emimNTf2 1.58 87.5
n-Octane ethylbenzene mmimTfO 6.20 93.4
emimTfO 4.37 91.7
bmimTfO 2.34 87.9
hxmimTfO 1.84 85.0
omimTfO 1.32 81.8

3-mebupyDCN 2.35 86.4


4-mebupyBF4 4.18 87.9
mmimNTf2 2.12 88.7
emimNTf2 1.73 86.5
n-Octane m-xylene mmimTfO 6.91 92.8
emimTfO 4.85 90.9
bmimTfO 2.56 86.9
hxmimTfO 2.00 83.7
omimTfO 1.43 80.6

a
In all the cases studied the aliphatic purity in the raffinate is 99.7–99.8 wt%, the aromatic (free-of-solvent) purity in the extract lies in the
interval 35.0–65.0 wt% and the IL loss through the raffinate is lower than 0.001%.
b
For N → ∞, see Fig. 9.
c
Aliphatic recovery in the raffinate.
d
In a 12-staged column.

mixture disclosed quasi-linear performance ((S/F)min. and with (S/F)min. and aliphatic recovery values larger than
aliphatic recovery in the raffinate, in this case) of the extrac- those obtained with the respective pure ILs (Fig. 8). A
tion with the IL2 mass percentage in the (IL–IL) mixture quaternary (4-mebupyBF4 + mmimTfO + mmimNTf2 ) mix-
(Fig. 8). In contrast, the (4-mebupyBF4 (1) + mmimNTf2 (2)) ture having equi-mass fraction composition gave molar
mixture showed a non-ideal behavior for the extraction, (S/F)min. ratio, aliphatic hydrocarbons purity (wt%) and

Table 4 – Results of the Aspen Plus simulations (with COSMOSAC property model) in the conceptual design of the unit
operation to remove aromatic hydrocarbons from (aromatic + aliphatic) mixtures by extraction with ionic liquids.
Multi-component mixtures 1 and 2 were fed to the extractor. 12-Staged extraction column operating at atmospheric
pressure and T = 40.0 ◦ C was used. Design specification: an overall 98.0% aromatic recovery in the extract. Aliphatics were
separated in the raffinate.
Ionic liquid Multi-component mixture 1 Multi-component mixture 2

Molar (S/F) Aliphatic Aliphatic Molar (S/F) Aliphatic Aliphatic


purity (wt%) recovery (%) purity (wt%) recovery (%)

3-mebupyDCN 1.9 99.7 81.0 1.8 99.7 74.9


4-mebupyBF4 3.9 99.7 80.7 3.8 99.8 72.5
mmimNTf2 1.8 99.7 82.1 1.8 99.8 75.2
mmimTfO 5.9 99.7 85.9 5.7 99.7 79.6

Please cite this article in press as: Ferro, V.R., et al., Conceptual design of unit operations to separate aromatic hydrocarbons from
naphtha using ionic liquids. COSMO-based process simulations with multi-component “real” mixture feed. Chem. Eng. Res. Des. (2014),
http://dx.doi.org/10.1016/j.cherd.2014.10.001
CHERD-1708; No. of Pages 16
ARTICLE IN PRESS
10 chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx

mmimTfO
binary (15 wt% aromatic + IL) mixtures with some selected aromatics. Mixtures are feed to the regenerator at the corresponding bubble point temperature. Regenerator operating

362
334
311
308
Table 5 – Results of the Aspen Plus simulations (with COSMOSAC property model) in the conceptual design of the individual IL regeneration by vacuum distillation from their

mmimNTf2
QIL-Regener. (kJ/kg-IL-regenerated)

325
303
285
282
4-mebupyBF4

453
418
389
385
3-mebupyDCN

459
424
395
391
mmimTfO

55.1
44.5
36.5
36.7
Fig. 6 – (A) Experimental (solid line) and Aspen Plus
calculated (dashed line) VL equilibria (using COSMOSAC
property package) for the binary mixtures of benzene with
mmim (), emim () and, bmim () NTf2 ILs. Insert shows
mmimNTf2

the (benzene + mmimNTf2 ) VLE for all composition


28.1
20.1
14.9
14.9

intervals. Experimental data were taken from Ref. Kato


Equilibrium pressure (kPa)
temperature was 230.0 ◦ C. Purity of the IL regenerated was 99.0 mol%.

et al. (2004). (B) Predicted VLEs of the binary mixtures


(toluene + C(1,2,4,6,8) mimTfO ILs). T = 353.15 K.
4-mebupyBF4

aliphatic hydrocarbons recovery (%) 4.2, 99.7 and 84.0 when


the multi-component (aromatic + aliphatic) mixture 1 was
51.7
40.6
32.5
32.9

fed to a 12-stage extractor. The multinary equi-mass (3-


mebupyDCN + 4-mebupyBF4 + mmimTfO + mmimNTf2 ) (IL–IL)
mixture ILs gave values 3.4, 99.7 and 83.3 for the same
variables.

4. Simulation of the IL regeneration


3-mebupyDCN

4.1. Regeneration of individual ILs from (aromatic + IL)


33.3
23.8
17.6
17.6

binary mixtures

IL regeneration from their binary (15 wt% aromatic + IL) mix-


tures needed reduced pressures in the range 15–60 kPa when
an operating temperature of 230 ◦ C was set (Table 5). Vac-
Ethylbenzene

uum requirements of the process increased in the sequence:


Aromatic

benzene < toluene < ethylbenzene ≈ m-xylene for all the ILs
m-Xylene
Benzene
Toluene

evaluated in this work (Table 5). Moreover, the heat duties


(QIL-Regener. ) required to regenerate the ILs from their mix-
tures with aromatics under the same conditions are in the

Please cite this article in press as: Ferro, V.R., et al., Conceptual design of unit operations to separate aromatic hydrocarbons from
naphtha using ionic liquids. COSMO-based process simulations with multi-component “real” mixture feed. Chem. Eng. Res. Des. (2014),
http://dx.doi.org/10.1016/j.cherd.2014.10.001
CHERD-1708; No. of Pages 16
ARTICLE IN PRESS
chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx 11

A A
6
6 (n-Hexane + Benzene)
(n-Heptane + Toluene)
5

(S/F), molar
4
(S/F)

3
2

0 5 10 15 20 0 20 40 60 80 100
Number of stages IL(2), wt%
B
10
B
86
(n-Octane + Ethylbenzene)
(n-Octane + m-Xylene) 85

Aliphatic recovery, %
8
(S/F)

84

6 83

82

4
81

5 10 15 20
0 20 40 60 80 100
Number of stages
IL(2), wt%
Fig. 7 – (S/F) vs. number of theoretical separation stages for
Fig. 8 – Results of the aromatic separation from
LL extraction of the aromatics from their binary mixtures
(aromatic + aliphatic) mixtures as a function of the
with aliphatic hydrocarbons using 4-mebupyBF4 IL as
composition of the (IL–IL) mixture used as extracting
extracting solvent. T = 313.15 K. (A) Benzene and toluene
solvent. (A) (S/F) molar ratio and (B) aliphatic recovery in the
from their respective mixtures with n-hexane and
raffinate. () 4-mebupyBF4 (1) + mmimNTf2 (2), ()
n-heptane. (B) Ethylbenzene and m-xylene from their
mmimTfO(1) + mmimNTf2 (2), ()
mixtures with n-octane. Profiles shown here were
4-mebupyBF4 (1) + mmimTfO(2). Multi-component mixture
analogous to those obtained for all the other ILs studied in
model 1 was fed to the extractor. A 12-staged extractor was
this work with the same binary (aromatic + aliphatic)
used operating at 40 ◦ C and atmospheric pressure. Similar
mixtures.
profiles were obtained for other variables like aliphatic
purities and recoveries.

range 300–450 kJ/kg-IL-regenerated (Table 5), following the 4.2. Regeneration of individual ILs from ternary
opposite tendency in respect to the operating pressures for (aliphatic + aromatic + IL) mixtures
the aromatic hydrocarbons. Similar results were obtained
when heat duties were given in kJ/kg-aromatic-removed. An extra (2 wt%) aliphatic (n-hexane, n-heptane or n-octane)
For NTf2 and TfO ILs, setting the operating temperature load in the (aromatic + IL) mixtures increased the operat-
at the IL regenerator to 320 ◦ C (see Section 2), allowed the ing pressure making the IL regeneration possible under
process to operate at higher pressures even near to the atmo- weaker vacuum conditions. However, simultaneously the
spheric one when TfO ILs were used as extracting solvents heat requirements of the regeneration increased (Fig. 10).
(Fig. 9A). Increases of the equilibrium pressure computed for Average increments (at 230 ◦ C) were approx. 25% for the
(15 wt% aromatic + mmimNTf2 ) mixtures under temperature equilibrium pressure and approx. 20% for the heat duties
growths from 230 to 320 ◦ C were 51.9, 40.4, 31.8 and 33.1 for the mmimTfO regeneration, respect to the values shown
with benzene, toluene, ethylbenzene and m-xylene, respec- in Table 5. These increases were moderated regenerating
tively. Nevertheless, the increase of the operating temperature mmimNTf2 (average growths: 10.6% and 13.1% for pressure
simultaneously had a negative energetic effect since the and QIL-Regener. ) and 4-mebupyBF4 (average growths: 10.5%
heating requirements of the aromatic removal from their and 14.3% for pressure and QIL-Regener. ) ILs but, they were more
binary mixtures with ILs increased (Fig. 9B) approximately 1.4 acute when 3-mebupyDCN was regenerated (average growths:
times (115–125 kJ/kg-IL-regenerated) for all the studied (15 wt% 235.5% and 57.3% for pressure and QIL-Regener. ). On the other
aromatics + mmimTfO or mmimNTf2 ) mixtures in the temper- hand, additional loads of n-hexane in the IL-regenerator
ature interval 230–320 ◦ C. feed gave the highest growths (respect to the other aliphatic

Please cite this article in press as: Ferro, V.R., et al., Conceptual design of unit operations to separate aromatic hydrocarbons from
naphtha using ionic liquids. COSMO-based process simulations with multi-component “real” mixture feed. Chem. Eng. Res. Des. (2014),
http://dx.doi.org/10.1016/j.cherd.2014.10.001
CHERD-1708; No. of Pages 16
ARTICLE IN PRESS
12 chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx

A A
150 60
Benzene
Toluene
Ethylbenzene 56
120 Benzene
m-Xylene
Toluene

Pressure, kPa
Pressure, kPa

52

90
48

Ethylbenzene
60 44 m-Xylene

40
30 n-Hexane n-Heptane n-Octane
225 250 275 300 325
Aliphatic
Temperature, ºC

B 405 B
440 Benzene

QIL-Regener., kJ/kg-IL-regenerated
QIL-Regener., kJ/kg-IL-regenerated

Toluene
Ethylbenzene 390
400 m-Xylene

375
360
Benzene
360 Toluene
320 Ethylbenzene
m-Xylene
345
280
n-Hexane n-Heptane n-Octane
225 250 275 300 325 Apliphatic
Temperature, ºC
Fig. 10 – Behavior of (A) the operating pressure and (B) the
Fig. 9 – (A) Operating pressure and (B) heat duties of the IL heat duties of the IL regeneration for binary mixtures
regeneration as a function of the operating temperature for (aromatic + mmimTfO) when an extra 2 wt% load of
binary mixture (15 wt% aromatic + mmimTfO). The purity of aliphatic hydrocarbons was added to the feed (see text for
the IL regenerated was 99.0 mol%. more details). See Table 5 for comparison with binary
(aromatic + IL) mixtures. T = 230.0 ◦ C. The purity of the IL
regenerated was 99.0 mol%.
studied in this work) on both the equilibrium pressures (at
230.0 ◦ C) and heat duties for all the ILs evaluated here.
were 259.4 and 111.8 kPa; 361 and 369 kJ/kg-IL-regenerated.
4.3. Regeneration of individual ILs from These results also suggest that effects of the considered vari-
multi-component (aromatics + aliphatics + IL) mixtures ables are additionally interrelated.

The main tendencies depicted in the two previous paragraphs 4.4. Regeneration of (IL–IL) mixtures from
were revealed at once when multi-component mixtures 1 multi-component (aromatics + aliphatics + ILs) mixtures
and 2 were fed to the extractor and the individual ILs were
recovered from the corresponding extracts (Tables 5 and 6). The regeneration of (IL–IL) mixtures essentially behaved like
The following conclusions were attained: (i) equilibrium pres- the regeneration of individual ILs for the studied systems,
sures and QIL-Regener. exhibited similar behaviors as a function obtaining in this case an almost linear dependence of the oper-
of the IL nature, (ii) computed equilibrium pressures and ating conditions and heat duties with the composition of the
QIL-Regener. (Table 6) were always higher than those obtained IL mixture (Fig. 11) for the multi-component mixture 1. This
for any binary (aromatic + IL) mixture (Table 5) due to the pres- seems to be a differentiating feature in respect to the oper-
ence of aliphatics in the extract, (iii) they were even higher ation at the extractor where non-ideal behaviors have been
in multi-component mixture 2 than in mixture 1 (Table 6), previously described.
corresponding with the higher concentration of aliphatic
hydrocarbons in mixture 2 respect to mixture 1 (Table 1), (iv) an 5. General outlook
increase of the regenerator operating temperature caused an
increase of both the equilibrium pressures and the QIL-Regener Examining of the altogether results related to both the extrac-
In fact, recovering mmimTfO and mmimNTf2 ILs at 320.0 ◦ C tion and the IL regeneration (Fig. 12) suggests that selection
from the extract corresponding to mixture 1 required equilib- of the IL-based extracting agent for (aromatic + aliphatic) sep-
rium pressures over 246 and 107 kPa and heat duties of 297 and aration processes is not a straightforward task. This was
324 kJ/kg-IL-regenerated. When the extract resulted from the previously discussed (Ferro et al., 2012a) from the point of
mixture 2, the equilibrium pressures and heat duties needed view of the IL regeneration. Two main conclusions can be

Please cite this article in press as: Ferro, V.R., et al., Conceptual design of unit operations to separate aromatic hydrocarbons from
naphtha using ionic liquids. COSMO-based process simulations with multi-component “real” mixture feed. Chem. Eng. Res. Des. (2014),
http://dx.doi.org/10.1016/j.cherd.2014.10.001
CHERD-1708; No. of Pages 16
ARTICLE IN PRESS
chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx 13

Table 6 – Results of the Aspen Plus simulations (with COSMOSAC property model) in the conceptual design of the
individual IL regeneration by vacuum distillation from the multi-component extract (in aromatic + aliphatic separation
process). Mixtures are feed to the regenerator at the corresponding bubble point temperature. Regenerator operating
temperature was 230.0 ◦ C. Purity of the IL regenerated was 99.0 mol%. Composition of the mixtures fed to the regenerator
is also provided.
Variable/ionic liquid 3-mebupyDCN 4-mebupyBF4 mmimNTf2 mmimTfO

Real model mixture 1


Aromatics in extract (wt%) 3.7 1.1 2.1 0.70
Equilibrium pressure (kPa) 47.2 88.6 37.3 94.8
QIL-Regener. (kJ/kg-IL-regenerated) 579 413 226 209

Real model mixture 2


Aromatics in extract (wt%) 4.9 2.4 3.0 1.2
Equilibrium pressure (kPa) 53.2 104.2 40.3 103.2
QIL-Regener. (kJ/kg-IL-regenerated) 439 363 262 255

A → A 86

90 60

Aliphatic recovery, %
(S/F) x 10, molar
84
75 45
Pressure, kPa

60 82
30

45
15 80

30 CN yBF4 NTf2 mTfO re (1) re (2) re (3) re (4)


yD p i u
0 20 40 60 80 100 bup mebu mmi
m
mm mixt ixtu ixtu ixtu
3-m
e
4- IL IL m IL m IL m
IL(2), wt%
B
450
B 60
QIL-Regener., kJ/kg-IL-Regenerated

QIL-Regener.x0.1,kJ/kg-IL-regenerated
90
400
Equilibrium pressure, kPa

50
350 75

40
300 60

30
250 45

200 20
30

CN yBF4 NTf2 mTfO re (1) re (2) re (3) re (4)


0 20 40 60 80 100 yD p i tu
bup m ebu m m i
m
m m mi x i x tu i x tu ixtu
3 -m
e
4- IL IL m IL m IL m
IL(2), wt%

Fig. 11 – (A) Operating pressures and (B) heat duties of the Fig. 12 – Result summary for the (A) extracting operation
regenerator when (IL–IL) mixtures were used as extracting and (B) solvent regeneration for the ILs (and their mixtures)
solvents. ILs were regenerated from the extracts studied in this work. Multi-component mixture 1 was fed
corresponding to the multi-component mixture 1. to the extractor. (IL–IL) mixtures (1), (2), (3) and (4) are,
Operating temperature at the regenerator was 230.0 ◦ C. () respectively, equi-mass mixtures of
(4-mebupyBF4 (1) + mmimNTf2 (2)), () (4-mebupyBF4 + mmimNTf2 ), (mmimTfO + mmimNTf2 ),
(mmimTfO(1) + mmimNTf2 (2)), () (4-mebupyBF4 + mmimTfO) and
(4-mebupyBF4 (1) + mmimTfO(2)). (4-mebupyBF4 + mmimNTf2 + mmimTfO).

derived from the present results: (i) none of the IL-based mixtures exhibit better performances than the individual
extracting agents evaluated here have simultaneously the ILs.
best performances for all the response variables analyzed.
This makes process optimizations on complete flow dia- 5.1. Concluding remarks
grams necessary to select IL-based extracting solvents with
optimized properties for separating aromatics from their mix- Separation of near-to-real multi-component (aro-
tures with aliphatic hydrocarbons. (ii) as a rule, the (IL–IL) matic + aliphatic) mixtures using ionic liquids substantially

Please cite this article in press as: Ferro, V.R., et al., Conceptual design of unit operations to separate aromatic hydrocarbons from
naphtha using ionic liquids. COSMO-based process simulations with multi-component “real” mixture feed. Chem. Eng. Res. Des. (2014),
http://dx.doi.org/10.1016/j.cherd.2014.10.001
CHERD-1708; No. of Pages 16
ARTICLE IN PRESS
14 chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx

differed from the scheme consolidated from the binary liquids using COSMOthermX. J. Chem. Eng. Data 54,
model (aromatic + aliphatic) mixtures ordinarily employed to 2005–2022.
study this process. A more realistic design of the individual Andzelm, J., Kölmel, C., Klamt, A., 1995. Incorporation of solvent
effects into density functional calculations of molecular
operations involved was achieved in this work by using
energies and geometries. J. Chem. Phys., 9312–9320.
COSMO-based process simulations with mixtures of complex Arce, A., Earle, M.J., Katdare, S.P., Rodriguez, H., Seddon, K.R.,
composition. This allowed new insights in both the technolog- 2008a. Application of mutually immiscible ionic liquids to the
ical and the economical features of this process. Thus, feeding separation of aromatic and aliphatic hydrocarbons by liquid
multi-component (aromatic + aliphatic) mixtures directly to extraction: a preliminary approach. Phys. Chem. Chem. Phys.
the extraction column reduces the purity of the aromatics 10, 2538–2542.
in the extract due to the presence of aliphatic hydrocarbons Arce, A., Earle, M.J., Rodriguez, H., Seddon, K.R., 2007a. Separation
of aromatic hydrocarbons from alkanes using the ionic liquid
which additionally represent an extra energetic load to the IL
1-ethyl-3-methylimidazolium bis{(trifluoromethyl) sulfonyl}
regeneration. This demands new design solutions respect to amide. Green Chem. 9, 70–74.
the simplified flow diagram shown in Fig. 1 of this paper. Arce, A., Earle, M.J., Rodriguez, H., Seddon, K.R., 2007b. Separation
(IL–IL) mixtures seemed to be good extracting solvents of benzene and hexane by solvent extraction with
for separating aromatic and aliphatic hydrocarbons. The 1-alkyl-3-methylimidazolium
present results suggest that process performance could be bis{(trifluoromethyl)sulfonyl}amide ionic liquids: effect of the
alkyl-substituent length. J. Phys. Chem. B 111, 4732–4736.
conveniently tailored by modifying the composition of the IL-
Arce, A., Earle, M.J., Rodriguez, H., Seddon, K.R., Soto, A., 2008b.
mixture. This is a way to improve the extracting capacity of the
1-Ethyl-3-methylimidazolium
solvent and, consequently, to reduce the solvent consumption. bis{(trifluoromethyl)sulfonyl}amide as solvent for the
All in all, current results validated the integration of separation of aromatic and aliphatic hydrocarbons by liquid
the COSMO-based models to Aspen Plus and Aspen HYSYS extraction. Extension to C7- and C8-fractions. Green Chem.
process simulators for accomplishing conceptual designs 10, 1294–1300.
of the operations involved in the process to separate Arce, A., Earle, M.J., Rodriguez, H., Seddon, K.R., Soto, A., 2009.
Bis{(trifluoromethyl)sulfonyl}amide ionic liquids as solvents
(aromatic + aliphatic) hydrocarbon mixtures using ILs as
for the extraction of aromatic hydrocarbons from their
extracting solvents. They particularly revealed the capac- mixtures with alkanes: effect of the nature of the cation.
ity of COSMO-supported process simulations for confidently Green Chem. 11, 365–372.
dealing with complex multi-component mixtures of hydrocar- Arce, A., Rodriguez, H., Soto, A., 2007c. Use of a green and cheap
bons and ILs which is absolutely indispensable for extending ionic liquid to purify gasoline octane boosters. Green Chem. 9,
the use of computational tools in the creation of new pro- 247–253.
cesses for removing aromatics from naphtha using ILs as Aspen Technology, 2003. What’s new in Aspen Engineering
SuiteTM 12.1. Aspen Technology Inc.
extracting solvents (including the selection of IL-based sys-
Aspen Technology, 2011. Aspen Physical Property System.
tems with optimized properties to this application). Even, the Physical Property Models. Aspen ONE v7.3 Documentation.
contradictions addressed in this paper respect to both the Aspen Technology Inc., MA, USA.
experimental results and the NRTL-based process simulations Asselin, G.F., 1984. Recovery of aromatic hydrocarbons and a
(i.e. the under estimated purity of toluene in extraction with non-aromatic raffinate stream from a hydrocarbon charge
4-mebupyBF4 ) of the aromatic extraction with ILs define a pro- stock, EP 0122341A1.
Banerjee, T., Khanna, A., 2006. Infinite dilution activity
ductive contribution of the COSMO-based process simulations
coefficients for trihexyltetradecyl phosphonium ionic liquids:
to the creation of new processes focused to the separation
measurements and COSMO-RS prediction. J. Chem. Eng. Data
of (aromatic + aliphatic) mixtures with ILs. In fact, the pre- 51, 2170–2177.
dictive capacity of this methodology could be exploited for Banerjee, T., Singh, M.K., Khanna, A., 2006. Prediction of binary
both proposing and evaluating (including the screening of IL- VLE for imidazolium based ionic liquid systems using
based systems with extracting improved properties) as most COSMO-RS. Ind. Eng. Chem. Res. 45, 3207–3219.
as possible design alternatives. Finally, on the basis of these Banerjee, T., Verma, K.K., Khanna, A., 2008. Liquid–liquid
equilibrium for ionic liquid systems using COSMO-RS: effect
results, the most promising alternatives could be selected
of cation and anion dissociation. AIChE J. 54, 1874–1885.
and submitted to the further engineering developments (Basic Bedia, J., Ruiz, E., de Riva, J., Ferro, V.R., Palomar, J., Rodriguez, J.J.,
Engineering, for example) where more accurate predictions 2013. Optimized ionic liquids for toluene absorption. AIChE J.
are needed. This approach ensure an economical procedure 59, 1648–1656.
as the experimental data needed to regress the adjustable Cadoret, L., Yu, C.-C., Huang, H.-P., Lee, M.-J., 2009. Effects of
parameters of the NRTL model should be generated only for physical properties estimation on process design: a case study
the most promising systems. using AspenPlus. Asia Pac. J. Chem. Eng. 4, 729–734.
Calvar, N., Dominguez, I., Gomez, E., Dominguez, A., 2011.
Separation of binary mixtures aromatic plus aliphatic using
Acknowledgments ionic liquids: influence of the structure of the ionic liquid,
aromatic and aliphatic. Chem. Eng. J. 175, 213–221.
Corderi, S., Gonzalez, E.J., Calvar, N., Dominguez, A., 2012.
The authors are grateful to Comunidad de Madrid (project
Application of hmimNTf2, hmimTfO and bmimTfO ionic
S2013-MAE-2800) and to Ministerio de Ciencia e Innovación liquids on the extraction of toluene from alkanes: effect of the
of Spain (Project CTQ2011-26758) for financial support. We are anion and the alkyl chain length of the cation on the LLE. J.
very grateful to “Centro de Computación Científica de la Uni- Chem. Thermodyn. 53, 60–66.
versidad Autónoma de Madrid” for computational facilities. Crosthwaite, J.M., Muldoon, M.J., Dixon, J.K., Anderson, J.L.,
Brennecke, J.F., 2005. Phase transition and decomposition
temperatures, heat capacities and viscosities of pyridinium
References ionic liquids. J. Chem. Thermodyn. 37, 559–568.
Chen, C.-C., Simoni, L.D., Brennecke, J.F., Stadtherr, M.A., 2008.
Correlation and prediction of phase behavior of organic
Ab Manan, N., Hardacre, C., Jacquemin, J., Rooney, D.W., Youngs,
compounds in ionic liquids using the nonrandom two-liquid
T.G.A., 2009. Evaluation of gas solubility prediction in ionic

Please cite this article in press as: Ferro, V.R., et al., Conceptual design of unit operations to separate aromatic hydrocarbons from
naphtha using ionic liquids. COSMO-based process simulations with multi-component “real” mixture feed. Chem. Eng. Res. Des. (2014),
http://dx.doi.org/10.1016/j.cherd.2014.10.001
CHERD-1708; No. of Pages 16
ARTICLE IN PRESS
chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx 15

segment activity coefficient model. Ind. Eng. Chem. Res. 47, Garcia, S., Larriba, M., Garcia, J., Torrecilla, J.S., Rodriguez, F.,
7081–7093. 2011b. Liquid–liquid extraction of toluene from heptane using
de Riva, J., Ferro, V.R., del Olmo, L., Ruiz, E., Lopez, R., Palomar, J., 1-alkyl-3-methylimidazolium
2014. Statistical refinement and fitting of experimental bis(trifluoromethylsulfonyl)imide ionic liquids. J. Chem. Eng.
viscosity-to-temperature data in ionic liquids. Ind. Eng. Chem. Data 56, 113–118.
Res. 53, 10475–10484. Garcia, S., Larriba, M., Garcia, J., Torrecilla, J.S., Rodriguez, F.,
Diedenhofen, M., Eckert, F., Hellweg, A., Huniar, U., Klamt, A., 2012a. Alkylsulfate-based ionic liquids in the liquid–liquid
Loschen, C., Reinisch, J., Schroer, A., Steffen, C., Thomas, C., extraction of aromatic hydrocarbons. J. Chem. Thermodyn. 45,
Wichmann, K., Ikeda, H., 2012. COSMOThermX, Version 68–74.
C30 1201. COSMOlogic GmbH & Co KG, Leverkusen, Germany. Garcia, S., Larriba, M., Garcia, J., Torrecilla, J.S., Rodriguez, F.,
Diedenhofen, M., Eckert, F., Klamt, A., 2003. Prediction of infinite 2012b. Separation of toluene from n-heptane by liquid–liquid
dilution activity coefficients of organic compounds in ionic extraction using binary mixtures of bpyBF4 and 4bmpyTf2N
liquids using COSMO-RS. J. Chem. Eng. Data 48, ionic liquids as solvent. J. Chem. Thermodyn. 53, 119–124.
475–479. Gonzalez-Miquel, M., Palomar, J., Omar, S., Rodriguez, F., 2011.
Diedenhofen, M., Klamt, A., 2010. COSMO-RS as a tool for CO2 /N2 selectivity prediction in supported ionic liquid
property prediction of IL mixtures: a review. Fluid Phase membranes (SILMs) by COSMO-RS. Ind. Eng. Chem. Res. 50,
Equilib. 294, 31–38. 5739–5748.
Diedenhofen, M., Klamt, A., Marsh, K., Schaefer, A., 2007. Gonzalez-Miquel, M., Palomar, J., Rodriguez, F., 2012. Selection of
Prediction of the vapor pressure and vaporization enthalpy of ionic liquids for enhancing the gas solubility of volatile
1-n-alkyl-3-methylimidazolium-bis- organic compounds. J. Phys. Chem. B 117, 296–306.
(trifluoromethanesulfonyl) amide ionic liquids. Phys. Chem. Gonzalez, E.J., Calvar, N., Gonzalez, B., Dominguez, A., 2010.
Chem. Phys. 9, 4653–4656. Measurement and correlation of liquid–liquid equilibria for
Domanska, U., Marciniak, A., 2008. Activity coefficients at infinite ternary systems {cyclooctane plus aromatic
dilution measurements for organic solutes and water in the hydrocarbon + 1-ethyl-3-methylpyridinium ethylsulfate} at
ionic liquid 1-butyl-3-methylimidazolium T = 298.15 K and atmospheric pressure. Fluid Phase Equilib.
trifluoromethanesulfonate. J. Phys. Chem. B 112, 11100–11105. 291, 59–65.
Domanska, U., Pobudkowska, A., Eckert, F., 2006. Liquid–liquid Gutierrez, J.P., Meindersma, G.W., de Haan, A.B., 2012.
equilibria in the binary systems (1,3-dimethylimidazolium, or COSMO-RS-based ionic-liquid selection for extractive
1-butyl-3-methylimidazolium methylsulfate plus distillation processes. Ind. Eng. Chem. Res. 51, 11518–11529.
hydrocarbons). Green Chem. 8, 268–276. Hamid, S.H., Ali, M.A., 1996. Comparative study of solvents for
Fallanza, M., Gonzalez-Miquel, M., Ruiz, E., Ortiz, A., Gorri, D., the extraction of aromatics from naphtha. Energy Sources 18,
Palomar, J., Ortiz, I., 2013. Screening of RTILs for 65–84.
propane/propylene separation using COSMO-RS methodology. Hansmeier, A.R., Ruiz, M.M., Meindersma, G.W., de Haan, A.B.,
Chem. Eng. J. 220, 284–293. 2010. Liquid–liquid equilibria for the three ternary systems
Ferreira, A.R., Freire, M.G., Ribeiro, J.C., Lopes, F.M., Crespo, J.G., (3-methyl-N-butylpyridinium dicyanamide plus toluene plus
Coutinho, J.A.P., 2011. An overview of the liquid–liquid heptane), (1-butyl-3-methylimidazolium dicyanamide plus
equilibria of (ionic liquid plus hydrocarbon) binary systems toluene plus heptane) and (1-butyl-3-methylimidazolium
and their modeling by the conductor-like screening model for thiocyanate plus toluene plus heptane) at T = (313.15 and
real solvents. Ind. Eng. Chem. Res. 50, 5279–5294. 348.15) K and p = 0.1 MPa. J. Chem. Eng. Data 55, 708–713.
Ferreira, A.R., Freire, M.G., Ribeiro, J.C., Lopes, F.M., Crespo, J.G., Ji, X., Held, C., Sadowski, G., 2012. Modeling imidazolium-based
Coutinho, J.A.P., 2012. Overview of the liquid–liquid equilibria ionic liquids with ePC-SAFT. Fluid Phase Equilib. 335, 64–73.
of ternary systems composed of ionic liquid and aromatic and Jongmans, M.T.G., Trampé, J., Schuur, B., de Haan, A.B., 2013.
aliphatic hydrocarbons, and their modeling by COSMO-RS. Solute recovery from ionic liquids: a conceptual design study
Ind. Eng. Chem. Res. 51, 3483–3507. for recovery of styrene monomer from [4-mebupy][BF4].
Ferro, V.R., Ruiz, E., de Riva, J., Palomar, J., 2012a. Introducing Chem. Eng. Process. Process Intensif. 70, 148–161.
process simulation in ionic liquids design/selection for Kato, R., Gmehling, J., 2005. Systems with ionic liquids:
separation processes based on operational and economic measurement of VLE and ␥∞ data and prediction of their
criteria through the example of their regeneration. Sep. Purif. thermodynamic behavior using original UNIFAC, mod.
Technol. 97, 195–204. UNIFAC(Do) and COSMO-RS(Ol). J. Chem. Thermodyn. 37,
Ferro, V.R., Ruiz, E., Tobajas, M., Palomar, J.F., 2012b. Integration of 603–619.
COSMO-based methodologies into commercial process Kato, R., Krummen, M., Gmehling, J., 2004. Measurement and
simulators: separation and purification of reuterin. AIChE J. correlation of vapor-liquid equilibria and excess enthalpies of
58, 3404–3415. binary systems containing ionic liquids and hydrocarbons.
Fredlake, C.P., Crosthwaite, J.M., Hert, D.G., Aki, S., Brennecke, J.F., Fluid Phase Equilib. 224, 47–54.
2004. Thermophysical properties of imidazolium-based ionic Klamt, A., 1995. Conductor-like screening model for real solvents:
liquids. J. Chem. Eng. Data 49, 954–964. a new approach to the quantitative calculation of solvation
Freire, M.G., Santos, L.M.N.B.F., Marrucho, I.M., Coutinho, J.A.P., phenomena. J. Phys. Chem. 99, 2224–2235.
2007. Evaluation of COSMO-RS for the prediction of LLE and Klamt, A., 2005. COSMO-RS: From Quantum Chemistry to Fluid
VLE of alcohols plus ionic liquids. Fluid Phase Equilib. 255, Phase Thermodynamics and Drug Design, first ed. Elsevier,
167–178. Amsterdam.
Freire, M.G., Ventura, S.P.M., Santos, L.M.N.B.F., Marrucho, I.M., Klamt, A., Eckert, F., Arlt, W., 2010. COSMO-RS: an alternative to
Coutinho, J.A.P., 2008. Evaluation of COSMO-RS for the simulation for calculating thermodynamic properties of liquid
prediction of LLE and VLE of water and ionic liquids binary mixtures. Annu. Rev. Chem. Biomol. Eng. 1, 101–122.
systems. Fluid Phase Equilib. 268, 74–84. Kroon, M.C., Karakatsani, E.K., Economou, I.G., Witkamp, G.J.,
Gaile, A.A., Zalishchevskii, G.D., Erzhenkov, A.S., Kayfadzhyan, Peters, C.J., 2006. Modeling of the carbon dioxide solubility in
E.A., Koldobskaya, L.L., 2007. Extraction of aromatic imidazolium-based ionic liquids with the tPC-PSAFT equation
hydrocarbons from reformates with mixtures of triethylene of state. J. Phys. Chem. B 110, 9262–9269.
glycol and sulfolane. Russ. J. Appl. Chem. 80, Kumar, U.K.A., Mohan, R., 2013. Quinary and eight-component
591–594. liquid–liquid equilibria of mixtures of alkanes, aromatics, and
Garcia, S., Garcia, J., Larriba, M., Torrecilla, J.S., Rodriguez, F., solvent (furfural). J. Chem. Eng. Data 58, 2194–2201.
2011a. Sulfonate-based ionic liquids in the liquid–liquid Lin, S.T., Sandler, S.I., 2002. A priori phase equilibrium prediction
extraction of aromatic hydrocarbons. J. Chem. Eng. Data 56, from a segment contribution solvation model. Ind. Eng. Chem.
3188–3193. Res., 899–913.

Please cite this article in press as: Ferro, V.R., et al., Conceptual design of unit operations to separate aromatic hydrocarbons from
naphtha using ionic liquids. COSMO-based process simulations with multi-component “real” mixture feed. Chem. Eng. Res. Des. (2014),
http://dx.doi.org/10.1016/j.cherd.2014.10.001
CHERD-1708; No. of Pages 16
ARTICLE IN PRESS
16 chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx

Liu, X., Ruiz, E., Afzal, W., Ferro, V., Palomar, J., Prausnitz, J.M., Integration of a novel COSMO-RS molecular descriptor on
2014. High solubilities for methane, ethane, ethylene, and neural networks. Ind. Eng. Chem. Res. 47, 4523–4532.
propane in trimethyloctylphosphonium Palomar, J., Torrecilla, J.S., Ferro, V.R., Rodriguez, F., 2009.
bis(2,4,4-trimethylpentyl) phosphinate (P8111TMPP). Ind. Eng. Development of an a priori ionic liquid design tool 2. Ionic
Chem. Res. 53, 363–368. liquid selection through the prediction of COSMO-RS
MacFarlane, D.R., Forsyth, S.A., Golding, J., Deacon, G.B., 2002. molecular descriptor by inverse neural network. Ind. Eng.
Ionic liquids based on imidazolium, ammonium and Chem. Res. 48, 2257–2265.
pyrrolidinium salts of the dicyanamide anion. Green Chem. 4, Pereiro, A.B., Rodriguez, A., 2009. Purification of hexane with
444–448. effective extraction using ionic liquid as solvent. Green Chem.
Mathias, P.M., Lin, S.T., Song, Y., Chen, C.C., Sandler, S.I., 11, 346.
November 2002. Phase-equilibrium predictions for Pereiro, A.B., Rodriguez, A., 2010. An ionic liquid proposed as
hydrogen-bonding systems from a new expression for COSMO solvent in aromatic hydrocarbon separation by liquid
solvation models. In: AIChE Annual Meeting, Indianapolis, pp. extraction. AIChE J. 56, 381–386.
3–8. Potdar, S., Anantharaj, R., Baneriee, T., 2012. Aromatic extraction
Meindersma, G.W., 2005. Extraction of aromatics from naphtha using mixed ionic liquids: experiments and COSMO-RS
with ionic liquids: from solvent development to pilot RDC predictions. J. Chem. Eng. Data 57, 1026–1035.
evaluation, Enschede., pp. 164. Rahmati-Rostami, M., Behzadi, B., Ghotbi, C., 2011.
Meindersma, G.W., de Haan, A.B., 2008. Conceptual process Thermodynamic modeling of hydrogen sulfide solubility in
design for aromatic/aliphatic separation with ionic liquids. ionic liquids using modified SAFT-VR and PC-SAFT equations
Chem. Eng. Res. Des. 86, 745–752. of state. Fluid Phase Equilib. 309, 179–189.
Meindersma, G.W., de Haan, A.B., 2012. Cyano-containing ionic Ruiz, E., Ferro, V.R., de Riva, J., Moreno, D., Palomar, J., 2014.
liquids for the extraction of aromatic hydrocarbons from an Evaluation of ionic liquids as absorbents for ammonia
aromatic/aliphatic mixture. Sci. China Chem. 55, 1488–1499. absorption refrigeration cycles using COSMO-based process
Meindersma, G.W., Hansmeier, A.R., de Haan, A.B., 2010. Ionic simulations. Appl. Energy 123, 281–291.
liquids for aromatics extraction. Present status and future Schäfer, A., Klamt, A., Sattel, D., Lohrenz, J.C.W., Eckert, F., 2000.
outlook. Ind. Eng. Chem. Res. 49, 7530–7540. COSMO implementation in TURBOMOLE: extension of an
Meindersma, G.W., Podt, A., de Haan, A.B., 2006a. Ternary efficient quantum chemical code towards liquid systems.
liquid–liquid equilibria for mixtures of an aromatic plus an Phys. Chem. Chem. Phys. 2, 2187–2193.
aliphatic hydrocarbon + 4-methyl-N-butylpyridinium Schröder, B., Coutinho, J.A.P., 2014. Predicting enthalpies of
tetrafluoroborate. J. Chem. Eng. Data 51, 1814–1819. vaporization of aprotic ionic liquids with COSMO-RS. Fluid
Meindersma, G.W., Podt, A., Klaren, M., de Haan, A., 2006b. Phase Equilib. 370, 24–33.
Separation of aromatic and aliphatic hydrocarbons with ionic Seiler, M., Jork, C., Kavarnou, A., Arlt, W., Hirsch, R., 2004.
liquids. Chem. Eng. Commun. 193, 1384–1396. Separation of azeotropic mixtures using hyperbranched
Meindersma, G.W., Podt, A.J.G., de Haan, A.B., 2005. Selection of polymers or ionic liquids. AIChE J. 50, 2439–2454.
ionic liquids for the extraction of aromatic hydrocarbons from Shah, M.R., Yadav, G.D., 2012. Prediction of liquid–liquid equilibria
aromatic/aliphatic mixtures. Fuel Process. Technol. 87, 59–70. of (aromatic + aliphatic + ionic liquid) systems using the
Meindersma, G.W., Podt, A.J.G., de Haan, A.B., 2006c. Ternary COSMO-SAC model. J. Chem. Thermodyn. 49, 62–69.
liquid–liquid equilibria for mixtures of toluene plus n-heptane Shiflett, M.B., Drew, D.W., Cantini, R.A., Yokozeki, A., 2010. Carbon
plus an ionic liquid. Fluid Phase Equilib. 247, 158–168. dioxide capture using ionic liquid
Meindersma, W.G.W., Onink, F.S.A.F., Hansmeier, A.R., de Haan, 1-butyl-3-methylimidazolium acetate. Energy Fuels 24,
A.B., 2012. Long term pilot plant experience on aromatics 5781–5789.
extraction with ionic liquids. Sep. Sci. Technol. 47, 337–345. Shiflett, M.B., Shiflett, A.D., Yokozeki, A., 2011. Separation of
Paduszynski, K., Domanska, U., 2012. Thermodynamic modeling tetrafluoroethylene and carbon dioxide using ionic liquids.
of ionic liquid systems: development and detailed overview of Sep. Purif. Technol. 79, 357–364.
novel methodology based on the PC-SAFT. J. Phys. Chem. B Shiflett, M.B., Yokozeki, A., 2006. Separation of difluoromethane
116, 5002–5018. and pentafluoroethane by extractive distillation using ionic
Palomar, J., Ferro, V.R., Torrecilla, J.S., Rodriguez, F., 2007. Density liquid. Chim. Oggi Chem. Today 24, 28–30.
and molar volume predictions using COSMO-RS for ionic Simoni, L.D., Lin, Y., Brennecke, J.F., Stadtherr, M.A., 2008.
liquids. An approach to solvent design. Ind. Eng. Chem. Res. Modeling liquid–liquid equilibrium of ionic liquid systems
46, 6041–6048. with NRTL electrolyte-NRTL, and UNIQUAC. Ind. Eng. Chem.
Palomar, J., Gonzalez-Miquel, M., Bedia, J., Rodriguez, F., Res. 47, 256–272.
Rodriguez, J.J., 2011a. Task-specific ionic liquids for efficient Tian, X., Zhang, X., Wei, L., Zeng, S., Huang, L., Zhang, S., 2010.
ammonia absorption. Sep. Purif. Technol. 82, 43–52. Multi-scale simulation of the 1,3-butadiene extraction
Palomar, J., Gonzalez-Miquel, M., Polo, A., Rodriguez, F., 2011b. separation process with an ionic liquid additive. Green Chem.
Understanding the physical absorption of CO2 in ionic liquids 12, 1263–1273.
using the COSMO-RS method. Ind. Eng. Chem. Res. 50, Vega, L.F., Vilaseca, O., Llovell, F., Andreu, J.S., 2010. Modeling
3452–3463. ionic liquids and the solubility of gases in them: recent
Palomar, J., Torrecilla, J.S., Ferro, V.R., Rodriguez, F., 2008. advances and perspectives. Fluid Phase Equilib. 294, 15–30.
Development of an a priori ionic liquid design tool 1.

Please cite this article in press as: Ferro, V.R., et al., Conceptual design of unit operations to separate aromatic hydrocarbons from
naphtha using ionic liquids. COSMO-based process simulations with multi-component “real” mixture feed. Chem. Eng. Res. Des. (2014),
http://dx.doi.org/10.1016/j.cherd.2014.10.001

Вам также может понравиться