Fluid Phase Equilibria 425 (2016) 98e107

Contents lists available at ScienceDirect

Fluid Phase Equilibria

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / fl u i d

A new UNIFAC parameterization for the prediction of liquid-liquid

equilibrium of biodiesel systems
Larissa C.B.A. Bessa a, Marcela C. Ferreira a, Charlles R.A. Abreu b, Eduardo A.C. Batista a,
Antonio J.A. Meirelles a, *
a
Laboratory of Extraction, Applied Thermodynamics and Equilibrium, Department of Food Engineering, Faculty of Food Engineering, University of
Campinas, Campinas, Sa ~o Paulo, 13083e862, Brazil
b
School of Chemistry, Federal University of Rio de Janeiro (UFRJ), 21941-909, Rio de Janeiro, RJ, Brazil

a r t i c l e i n f o a b s t r a c t

Article history: The environmental adversities and the global concern about the conservation of non-renewable natural
Received 24 March 2016 resources have stimulated a search for environmentally friendly energy sources. In this context, biodiesel
Received in revised form has emerged as an important alternative to replace fossil fuels, due to its renewability, non-toxicity and
11 May 2016
biodegradability. Modeling, simulation and design of unit operations involved in the production of edible
Accepted 14 May 2016
Available online 17 May 2016
oils and biodiesel require knowledge of phase equilibrium. Several versions of the UNIFAC model are
frequently used for process design when experimental determination of phase equilibrium data is
difficult or time-consuming. In this work, the original UNIFAC model parameters are first checked for
Keywords:
Liquid-liquid equilibrium
their predictive capability and then modified in terms of new readjusted binary interaction parameters. It
Biodiesel was noted that the UNIFAC model without any changes in its parameters results in inadequate pre-
Modeling dictions. Thus, in order to obtain a good predictive tool, a comprehensive liquid-liquid equilibrium data
UNIFAC bank of systems present in biodiesel production was organized and new UNIFAC interaction parameters
Proximity effects were adjusted. At first, the molecules were divided into UNIFAC traditional structural groups. However,
this first approach resulted in poor prediction, probably as a consequence of the strongly polar hydroxyl
groups bonded to the consecutive carbon atoms of glycerol and acylglycerol molecules. Thus, a new
group (‘OHgly’) was introduced and two matrices of parameters were adjusted. In general, satisfactory
predictions were obtained and a significant improvement in the performance of this group contribution
model has been achieved.
© 2016 Elsevier B.V. All rights reserved.

1. Introduction
Increasing search for alternatives to petroleum-based fuels has
led to the development of fuels from various sources, including
renewable feedstocks such as fats and oils. Several types of fuels can
be derived from these triacylglycerol-containing feedstocks. One of
them is biodiesel, which is defined as the mono-alkyl esters of
vegetable oils or animal fats [1]. Considering its well-known envi-
ronmental and economic benefits, biodiesel is expected to be a
good alternative to petroleum-based fuels. It can be used to address
the limitations associated with fossil fuels, as the continued rise in
* Corresponding author.
E-mail addresses: larissacbabessa@gmail.com (L.C.B.A. Bessa), marcelacravo@
gmail.com (M.C. Ferreira), abreu@eq.ufrj.br (C.R.A. Abreu), eacbat@unicamp.br
(E.A.C. Batista), tomze@unicamp.br (A.J.A. Meirelles).
crude oil prices, scarce fossil energy resources and environmental
concerns [2].
So far, biodiesel has mainly been produced by transesterification
of triacylglycerols (TAGs) and/or esterification of free fatty acids
(FFAs) using homogeneous basic or acid catalysts [3,4]. The trans-
esterification reaction is a three-stage reaction, which produces
two intermediate products (diacylglycerols and mono-
acylglycerols). Stoichiometrically, the reaction requires a molar
ratio alcohol:oil of 3:1, but excess alcohol (methanol being the most
commonly used alcohol) is usually added to shift the reaction to-
wards the products [4,5]. In Brazil, however, the use of ethanol is
advantageous because of its large scale production, apart from
allowing obtaining biodiesel through a totally renewable process.
On the other hand, one disadvantage related to the use of ethanol is
that the phase separation can be more difficult when compared
with methanol [6]. In this sense, a better understanding and pre-
diction of phase equilibrium of the biodiesel related systems are

http://dx.doi.org/10.1016/j.fluid.2016.05.020
0378-3812/© 2016 Elsevier B.V. All rights reserved.
 L.C.B.A. Bessa et al. / Fluid Phase Equilibria 425 (2016) 98e107 99

required for the proper design, operation and optimization of the improvement of industrial investment both in equipment design as
reactor and separation units [4]. in process optimization.
Modeling of the reaction and separation processes required to
produce higher purity biodiesel involves determining the liquid-
2. Thermodynamic modeling
eliquid equilibrium and thus, a reliable thermodynamic model is
essential [7]. However, because of the size of the molecules and the
In the context of the ever increasing importance of the liquid-
strong molecular interactions involved in the transesterification,
liquid equilibrium in the processing of oils and fats, especially
the thermodynamic modeling is particularly challenging [8].
during the oil deacidification and biodiesel production, the read-
Considering that the biodiesel production process basically in-
justment of a new set of interaction parameters of interest for these
volves fatty acid esters, alcohol and glycerol, and that the various
types of systems is necessary to improve the predictive capacity of
kinds of esters have many physical-chemical similarities among
the UNIFAC method when applied to fatty systems. In the present
them, such systems are generally treated as a pseudoternary one,
work, the adjustment of new interaction parameters was based on
consisting of an equivalent alkyl ester of fatty
experimental data for real fatty systems already available in the
acid þ alcohol þ glycerol. Based on this assumption, equilibrium
literature. In fact, experimental equilibrium data for systems con-
data for a wide variety of alkyl esters may be correlated using
taining pure fatty compounds are very rare and do not provide a
models such as UNIQUAC (Universal Quasi Chemical) and NRTL
data basis sufficiently comprehensive for readjusting parameters of
(Non-random, two liquid) with very good results [7,9e12]. An
existing groups.
alternative to the usually applied activity coefficient models to
In order to obtain a suitable predictive tool, it is required to work
predict systems with polar compounds with strong associative in-
with an experimental liquid-liquid equilibrium database as
teractions found at the biodiesel production and purification pro-
comprehensive as possible. It has been collected, from the litera-
cesses is the use of association equations of state. Among these
ture, 218 systems containing biodiesel/esters, partial acylglycerols
associating equations are the statistical associating fluid theory
(mono and diacylglycerol), alcohol (methanol or ethanol), glycerol
(SAFT) [13] and the Cubic-Plus-Association (CPA) equations of state
and water, summing up a total of 1145 tie lines, at temperatures
[14]. Recently, both equations have been used to describe these
ranging from T/K ¼ 293.15 to 393.15. Table 1 shows a summary of
systems [15e20], presenting encouraging results, although their
the equilibrium systems used in the readjustment procedure. For
predictive character is still limited to simpler fatty compounds.
each group of data, Table 1 gives the original oil, the number of
Nevertheless, this approach does not take into account the
systems, the number of tie lines, the temperature range and the
different behaviors of triacylglycerols, fatty acids and specific esters
corresponding reference. It should be considered that a larger set of
and does not permit reliable prediction of equilibrium data for
liquid-liquid equilibrium data involving compounds of interest for
different types of biodiesel/oil systems (different compositions in
this study is available in the literature, apart from the data shown in
terms of esters/triacylglycerols) with the parameters adjusted us-
Table 1. However, all those data whose error in the sum of mass
ing those models. Furthermore, the number of components
fractions at each phase was greater than or equal to 0.001 were not
involved is usually large and the experimental information is
considered in the parameter readjustment procedure, since these
scarce, so that it is particularly interesting to use a group contri-
data could incorporate errors in the deviations values not derived
bution method.
from the equilibrium calculations.
The UNIFAC group contribution method [21] has proven to be a
In addition, the data collected by Hirata et al. [31], which involve
fast and in many cases a reliable tool for the prediction of liquid-
systems composed of (vegetable oil þ fatty
phase activity coefficients. Extensive tables with revised and
acids þ ethanol þ water), were updated and also considered,
updated interaction parameters have been published [22e27].
applying the same tolerance in the mass balance deviation adopted
However, the current methods and corresponding set of available
for biodiesel systems. For this type of systems, the final database
parameters provide improperly description of the liquid-liquid
comprises 105 systems summing up a total of 937 tie lines. Thus, it
equilibrium (LLE) of systems containing vegetable oils, partial
was intended to obtain a matrix of parameters that is appropriate
acylglycerols, free fatty acids, alcohol and/or biodiesel
for the representation of systems from the oil deacidification up to
[9e11,28e30].
the biodiesel production.
For this reason, the purpose of this article is to present a UNIFAC
The software used for adjusting the interaction parameters was
parameter matrix especially suited for the prediction of LLE of
developed in Fortran and was formulated in a previous work [31]. It
biodiesel systems. The original UNIFAC model was first checked
is worth mentioning that the parameters obtained by Hirata et al.
using two different sets of parameters available in the literature to
[31] presented a great improvement in the prediction of the sys-
assess their predictive capability and then improved in terms of
tems studied by the authors. However, these parameters do not
new regressed binary interaction parameters. It was used an
represent precisely systems containing biodiesel [10,11].
approach to readjust the interaction parameters for UNIFAC model
The modeling developed in this work is based on the original
based on experimental data for real multicomponent systems. A
UNIFAC model [21]. The parameters adjustment is based on the
similar analysis has been previously performed by the research
minimization of the composition objective function, given by Eq.
group for the adjustment of UNIFAC interaction parameters for
(1), using the simplex method [64].
systems present in the deacidification of vegetable oils, i.e., systems
composed of (vegetable oil þ fatty acids þ ethanol þ water) [31]. X
D X X
N P 1    
FI;calc 2 FII; calc 2
The database collected in that previous work has been updated and S¼ wFI;exp
inm
 winm
þ wFII;exp
inm
 winm
was also used in the present adjustment procedure. Thus, it was m¼1 n¼1 i¼1
intended to obtain a matrix of parameters appropriate for the (1)
representation of fatty systems from the oil deacidification up to
the biodiesel production. where D is the total number of data banks, N is the total number of
With the correct representation of the phase equilibrium tie lines, P is the total number of pseudocomponents in each data
involved in the entire sequence of biodiesel production, the cor- bank; i, n, and m stand for component, tie line and data group,
responding process can be investigated and optimized with confi- respectively; FI and FII refer to phases I and II, respectively; exp and
dence. A good-quality predictive tool contributes to an calc stand for experimental and calculated mass fractions (w),
100 L.C.B.A. Bessa et al. / Fluid Phase Equilibria 425 (2016) 98e107

Table 1
Summary of the data bank.

Original oil/ester T/K Systems Tie lines Reference

Brazil nut 303.15e323.15 2 10 [32]

Canola 293.15e313.15 7 31 [33e35]
Canola þ partial acylglycerols 303.15e318.15 4 22 [11]
Castor 298.15e333.15 28 136 [36e38]
Coconut 293.15e323.15 4 23 [39,40]
Commercial biodiesel 293.15 1 7 [41]
Corn 293.15e313.15 5 27 [40,42]
Corn þ partial acylglycerols 303.15e318.15 4 22 [11]
Cottonseed þ partial acylglycerols 303.15e318.15 2 11 [12]
Cottonseed 293.15e333.15 6 36 [43,44]
Crambe 298.15e338.15 4 24 [29,45]
Ethyl laurate 298.15e353.15 5 30 [18,46]
Ethyl linoleate 313.15e353.15 3 20 [17,18]
Ethyl myristate 298.15e333.15 3 19 [46]
Ethyl oleate 298.15e353.15 3 19 [17,18]
Ethyl palmitate 298.15e353.15 8 51 [17,18,47]
Ethyl stearate 313.15e323.15 2 19 [6]
Fodder radish 298.15 1 6 [29]
Frying oil 293.15e313.15 3 15 [42]
High-oleic sunflower þ partial acylglycerols 303.15e318.15 4 23 [10]
Jatropha curcas 298.15e333.15 23 102 [48]
Macauba 298.15 2 11 [9,29]
Methyl linoleate 298.15e318.15 8 50 [49,50]
Methyl oleate 293.15e393.15 13 83 [49,51e53]
Methyl palmitate 333.15 1 7 [54]
Palm 293.15e313.15 7 44 [55,56]
Palm þ partial acylglycerols 318.15 1 4 [28]
Rice bran þ partial acylglycerols 303.15e318.15 2 11 [12]
Soybean 293.15e343.15 50 220 [4,37,40,56e62]
Soybean þ partial acylglycerols 303.15e318.15 4 23 [12,63]
Soybean þ ethyl oleate 303.15e318.15 2 12 [63]
Sunflower 293.15e323.15 6 27 [34,35,57]

respectively. Table 2
Each system was analyzed considering all its main components UNIFAC group volume (Rk) and area (Qk) parameters.a

(i.e., several esters, tri-, di- and monoacylglycerols) in phase equi- Main group Sub group Rk Qk
librium calculations, so that the system can be considered in all its CH2 CH3 0.9011 0.848
complexity. However, to make it possible to compare the calculated CH2 0.6744 0.540
values and the experimental data, in the objective function (Eq. (1)), CH 0.4469 0.228
which evaluates the difference between the calculated values and C]C CH]CH 1.1167 0.867
OH OH 1.0000 1.200
the experimental compositions, the different classes of compo-
H2O H2O 0.9200 1.400
nents, such as esters, tri-, di- and monoacylglycerols, are grouped COOH COOH 1.3013 1.224
into pseudocomponents, in the way the experimental data are COOC CH2COO 1.6764 1.420
usually presented in the literature. a
Magnussen et al. [27].
The deviations between experimental and calculated composi-
tions in both phases were calculated using the root mean square
deviation (Dw/%), which is given by the following equation: presented by Magnussen et al. [27], showed in Table 2.
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi Then, the adjustment of new parameters of interaction between
u   
uPN PP  FI;exp FI;calc 2
 
FII;exp FII;calc 2 groups was carried out in order to achieve a better description of
u n¼1 i¼1 wi;n wi;n þ wi;n wi;n
t the experimental data. For the parameters adjustment, both
Dw ¼ 100 UNIFAC-LLE and UNIFAC-HIR parameters were tested as initial es-
2NP
timate for the first attempt to minimize the objective function. The
(2)
adjustment strategy was to use the resulting parameters of a prior
Unfortunately, in the literature, the biodiesel used in the phase minimization to the next one, until the parameters are not altered
equilibrium was not always expressed in terms of its ester anymore. The compounds used in this study were divided, in a first
composition. On the other hand, the composition of the oil that approach, into structural groups characteristic of the original
originated that biodiesel, in terms of fatty acids, was always avail- UNIFAC (‘‘CH3’’, ‘‘CH2’’, ‘‘CH’’, ‘‘CH]CH’’, ‘‘CH2COO’’, “COOH” e
able in the selected references. Thus, for the references that do not ‘‘OH’’).
contain that first information, the esters composition was assumed In order to validate the results of the proposed modeling pro-
to be the same as that of its original oil, in terms of fatty acids. cedure, the systems composed by sunflower oil þ commercial
First, the liquid-liquid equilibrium of systems present in the mixture of mono- and diacylglycerols (þethyl linoleate) þ ethanol,
database was calculated through the original UNIFAC model [21] at T/K ¼ 303.15 and 318.15, presented in a previous work [10] were
using two sets of interaction parameters already available in the used. They are representative systems of the transesterification
literature: (i) UNIFAC-LLE proposed by Magnussen et al. [27] and (ii) step in the biodiesel production and they were chosen in order to
UNIFAC-HIR, adjusted by Hirata et al. [31]. The UNIFAC area and reveal the great improvement in the behavior description of the
volume parameters were assumed to be the same as those minor components on phase equilibrium, since that was one of the
 L.C.B.A. Bessa et al. / Fluid Phase Equilibria 425 (2016) 98e107 101

main difficulties presented when using the parameters available in

the literature. It is important to mention that these systems were
not considered in the database for the parameters adjustment.

3. Results and discussion

Initially, an attempt was made to describe the molecules of the

compounds studied with the elementary groups originally pro-
posed by Fredenslund et al. [21]. However, poor predictions were
obtained, leading to unacceptable deviations from experimental
data. In addition, regarding the partial acylglycerols distribution,
the slopes of the calculated tie lines exhibit an inverse behavior in
comparison to the experimental ones, estimating the preferred
phase of diacylglycerols wrongly, as can be seen in Fig. 1.
Furthermore, the use of the UNIFAC model without any changes
in their functional groups underestimates the mass fraction of
monoacylglycerols in the oil phase and overestimates the mass
fraction of these components in the alcoholic phase. This is
observed by the more accentuated slope of the calculated tie lines
Fig. 2. Liquid-liquid equilibrium for the system high-oleic sunflower
in relation to the experimental ones, as shown in Fig. 2. This
oil(1) þ diacylglycerol(2) þ monoacylglycerol(3) þ ethanol(4) at 318.15 K, MAG dis-
behavior was observed for whatever the matrix of parameters used. tribution: C, experimental data [10]; - - -, calculated values using UNIFAC-LLE; ,
In the UNIFAC model, polar groups, whether bonded to calculated values using UNIFAC-HIR; , calculated values using the first attempt of
consecutive atoms or not, are treated the same way. Such a simple UNIFAC parameters readjustment.
parameterization requires a small number of structural groups, but
cannot represent interactions between adjacent atoms in a mole-
bonded to consecutive carbon atoms in the glycerol and acylgly-
cule [65]. The UNIFAC model does not take into account, for
cerol molecules. This proximity effect may force this ’OH’ group to
example, proximity effects that occur when two or more strongly
present specific interactions, different from those of the usual
polar groups are bonded to the same or adjacent carbon atoms [66].
alcohol group. Proximity effects on predictions of UNIFAC model
One way to overcome this limitation is to define specific functional
have been discussed previously in the literature [67,68]. The
groups for certain classes of compounds.
adoption of specific functional groups for certain classes of com-
Since the predictions using this first attempt of parameters
pounds due to proximity effects has already been proposed by Peng
adjustment were not good, and taking advantage of the fact that the
et al. [69] for dicarboxylic and hydroxycarboxylic acids, by Ninni
UNIFAC model is quite flexible, allowing the addition of new groups
et al. [70] for poly(ethylene glycol), by Ozdemir and Sadikoglu [71]
to better describe a system, the ‘OH’ group attached to the carbon
and Marcolli and Peter [65] for the polyol/water systems, among
chain of the glycerol (present, thus, in the mono- and diac-
others. In addition, modified parameterizations for other UNIFAC
ylglycerols molecules, besides glycerol) was considered as a
model [72] have also been used for systems containing polyols [73]
different functional group (‘OHgly’). It was considered that this
and sugars [74].
group behaves differently than the traditional ‘OH’ group of alcohol
For this second approach, values for group volume and surface
molecules due to the effect of strongly polar hydroxyl groups
area parameters Rk and Qk for the new group were considered equal
to the alcohol group ‘OH’ (Table 2). For the adjustment of the pa-
rameters of the ‘OHgly’ group, the interaction parameters for the
traditional ‘OH’ group were used as initial estimate. In this case, it
was first used a reduced database, containing only the systems that
comprise the components that contain the ‘OHgly’ group (i.e.,
glycerol, mono- and diacylglycerols) and interaction parameters
were readjusted only for this group. Then, the other systems were
added to the database and all the parameters were readjusted.
Again, the matrix of parameters of a resulting minimization was
used to the next until the parameters are not altered anymore.
Table 3 shows the resulting matrix of parameters of this adjustment
procedure.
This new set of parameters was used to predict the LLE of each

Table 3
Matrix of readjusted UNIFAC interaction parameters.

CH C]C OH H2O COOH COOC OHgly

CH 0.00 239.32 714.56 2045.36 14.88 560.42 1092.77

C]C 470.13 0.00 954.33 328.14 4709.13 49.03 69.93
OH 73.52 212.36 0.00 19.65 417.57 172.62 0.00
Fig. 1. Liquid-liquid equilibrium for the system high-oleic sunflower
H2O 230.67 207.51 90.08 0.00 30.79 11.22 23.26
oil(1) þ diacylglycerol(2) þ monoacylglycerol(3) þ ethanol(4) at 318.15 K, DAG dis-
COOH 445.72 508.21 24.00 63.41 0.00 6289.17 4.39
tribution: C, experimental data [10]; - - -, calculated values using UNIFAC-LLE; ,
COOC 462.69 1040.51 243.03 616.17 306.77 0.00 25.02
calculated values using UNIFAC-HIR; , calculated values using the first attempt of
OHgly 414.99 1274.10 0.00 57.25 952.41 146.14 0.00
UNIFAC parameters readjustment.
102 L.C.B.A. Bessa et al. / Fluid Phase Equilibria 425 (2016) 98e107
Table 4
Overall average deviations.
System Dw/%
UNIFAC-LLE UNIFAC-HIR
Deacidification 5.72 1.84
Biodiesel 4.27 6.50
Partial acylglycerols 7.31 2.64
Global 5.03 4.32
system present in the database. Table 4 shows the overall average
deviations for each type of system using the three matrices of pa-
rameters studied, and the overall average deviation considering all
systems. The average deviations for all systems can be found in the
Supplementary material.
It can be seen that for the deacidification systems, the UNIFAC-
HIR matrix of parameters resulted in the lowest value of average
deviation. This result was, in a certain way, expected, since the
authors who proposed those parameters studied specifically the
systems containing (vegetable oil þ fatty acids þ solvent), charac-
teristic of the deacidification process of vegetable oils. However,
values of deviations obtained with the parameters readjusted in
this study were also low, especially when compared to the original
UNIFAC-LLE parameters. Fig. 3 shows that for the deacidification
system, the UNIFAC-HIR parameters result in better prediction, but
the parameters readjusted in this work also yielded values close to
the experimental data. One can also observe that the slopes of the
tie lines obtained using the original UNIFAC-LLE parameters exhibit
an inverse behavior in relation to the experimental ones, estimating
partition coefficients erroneously.
In contrast, for the other systems, it was noted that, in general,
the prediction of equilibrium data was significantly improved when
readjusted parameters were used. Regarding the systems present in
the transesterification of vegetable oils, with special emphasis on
partial acylglycerols, an analysis can be performed by the validation
system. It is worth noting that the low global average deviation
value for systems containing partial acylglycerols using the
UNIFAC-HIR parameters (2.64%) can cause an improper perception
that the use of these parameters would be better for predicting the
LLE of this type of system. However, more important than evaluate
Fig. 3. Liquid-liquid equilibrium for the system soybean oil(1) þ commercial linoleic
acid(2) þ anhydrous ethanol(3) at 298.15 K: C, experimental data [75]; - - -, calculated
values using UNIFAC-LLE; , calculated values using UNIFAC-HIR;
values using the new readjusted parameters.
only the numerical value of the deviation, it is essential to analyze
the behavior of the components in the system. Figs. 4e9 show the
equilibrium diagrams for the validation system, including both
This work
experimental and calculated data, using the UNIFAC-LLE parame-
ters, the UNIFAC-HIR and the readjusted ones. It is worth remem-
2.59
bering that these data were not included in the database for the
3.66
1.81 readjustment procedure. In Figs. 6 and 9, in order to have a better
3.13 interpretation of the five-component phase equilibrium data, the
experimental and calculated results were represented in a simpli-
fied form, showing only the ethyl linoleate and ethanol composi-
tions in a explicit way and grouping the acylglycerols as a third
pseudocomponent, which consists of TAG-DAG-MAG.
From Figs. 4e9, it can be seen that the equilibrium data pre-
dictions were significantly improved by using the parameters
readjusted in this work when compared to the use of the original
UNIFAC-LLE and the UNIFAC-HIR parameters. This can also be
proven by the deviation values between experimental and calcu-
lated data, presented in Table 5, and by the average distribution
coefficients (ki, given by the ratio between the content of the
component i in the alcoholic phase and the content of this
component in the oil phase), shown in Fig. 10 and whose values are
of interest in order to observe the behavior of the minor compo-
nents in the system. One can see that the average distribution co-
efficients calculated using the parameters readjusted in this work
are quite similar to the experimental ones, different from the other
parameters matrices from literature.
For the system at 303.15 K, one can see in Table 5 that a sig-
nificant improvement of the phase equilibrium predictions is ach-
ieved, both numerically and graphically. As for the system at
318.15 K, graphically, it can be noted very satisfactory predictions of
the LLE. However, numerically, it is observed that when using the
UNIFAC-HIR parameters, the calculated deviations values are the
lowest. Again, this low deviation value may lead to an improper
conclusion that, by using these parameters, the liquid-liquid equi-
librium involved is properly predicted. This may be true when
analyzing the esters partition in the phases, as shown in Figs. 6 and
9. Through these figures, it is observed that both the UNIFAC-HIR
parameters and the new parameters adjusted in this work have
correctly predicted the equilibrium data, both much better than the
prediction using the original UNIFAC-LLE parameters, and
Fig. 4. Liquid-liquid equilibrium for the system sunflower
oil(1) þ diacylglycerol(2) þ monoacylglycerol(3) þ ethanol(5) at 303.15 K, DAG dis-
tribution: C, experimental data [10]; - - -, calculated values using UNIFAC-LLE; ,
, calculated calculated values using UNIFAC-HIR; , calculated values using the new readjusted
parameters.
 L.C.B.A. Bessa et al. / Fluid Phase Equilibria 425 (2016) 98e107 103

Fig. 5. Liquid-liquid equilibrium for the system sunflower Fig. 7. Liquid-liquid equilibrium for the system sunflower
oil(1) þ diacylglycerol(2) þ monoacylglycerol(3) þ ethanol(5) at 303.15 K, MAG dis- oil(1) þ diacylglycerol(2) þ monoacylglycerol(3) þ ethanol(5) at 318.15 K, DAG dis-
tribution: C, experimental data [10]; - - -, calculated values using UNIFAC-LLE; , tribution: C, experimental data [10]; - - -, calculated values using UNIFAC-LLE; ,
calculated values using UNIFAC-HIR; , calculated values using the new readjusted calculated values using UNIFAC-HIR; , calculated values using the new readjusted
parameters. parameters.

presenting between them only a small difference in the biphasic
region size.
Nevertheless, when observing the partition of partial acylgly-
cerols (DAGs and MAGs), it is found that the predictions using both
the original UNIFAC-LLE and the UNIFAC-HIR parameters were not
adequate. In the case of diacylglycerols (Figs. 4 and 7), it is observed
that using the parameters from literature the slopes of the calcu-
lated tie lines were inverted when compared to experimental ones,
which reinforces the poor quality of the systems composition
predictions, estimating a diacylglycerol preference for the alcoholic
phase, which is not consistent with the experimental observation.
Fig. 10 also shows the error in the prediction of the distribution of
DAG when using the UNIFAC-HIR parameters. The experimental
distribution coefficient (ki) of DAGs is smaller than unity, indicating
their preference for the oil phase. On the other hand, the average
distribution coefficients calculated using UNIFAC-HIR are greater
than unity, indicating an opposite distribution.
In relation to the monoacylglycerols (Figs. 5 and 8), it can be
seen that by using both sets of interaction parameters from litera-
ture, the slopes of the calculated tie lines are more accentuated
than those of the experimental ones, indicating that the mass
fraction of MAG is underestimated in the oil phase and over-
estimated in the alcoholic phase. This effect is greatest when the
UNIFAC-LLE parameters were used, justifying the larger deviation
values. When the newly readjusted parameters are used, there is a
very significant improvement in the description of tie lines. The
slope of the tie lines for both DAGs and MAGs are in agreement with
the experimental ones and much closer to the experimental data,

Fig. 6. Liquid-liquid equilibrium for the system sunflower Fig. 8. Liquid-liquid equilibrium for the system sunflower
oil(1) þ diacylglycerol(2) þ monoacylglycerol(3) þ ethyl linoleate (4) þ ethanol(5) at oil(1) þ diacylglycerol(2) þ monoacylglycerol(3) þ ethanol(5) at 318.15 K, MAG dis-
303.15 K: C, experimental data [10]; - - -, calculated values using UNIFAC-LLE; , tribution: C, experimental data [10]; - - -, calculated values using UNIFAC-LLE; ,
calculated values using UNIFAC-HIR; , calculated values using the new readjusted calculated values using UNIFAC-HIR; , calculated values using the new readjusted
parameters. parameters.
104 L.C.B.A. Bessa et al. / Fluid Phase Equilibria 425 (2016) 98e107

model for mixtures containing components with multiple hydroxyl

groups attached to adjacent carbon atoms shows insufficient pre-
dictive capabilities. According to the authors, the main reason for
such behavior is the reduction in hydrogen bonding interaction due
to steric hindrances.
In addition to satisfactorily describe ternary systems, it is
important that the obtained UNIFAC parameters are also able to
predict the phase behavior of the corresponding binary sub-
systems. Thus, some binary mixtures reported in the literature
were also used to validate the parameters. Table 6 gives the system
analyzed, the temperature range, the corresponding reference and
the calculated deviations (Dw).
Turning back to Table 4, the average deviation value obtained for
biodiesel systems when using the new parameters readjusted in
this study indicates that there was an improvement in the equi-
librium prediction for this type of system, presenting a reduction in
the deviation value of, approximately, 14.3% compared to UNIFAC-
LLE parameters and 43.7% when compared to the UNIFAC-HIR pa-
rameters. However, the obtained deviation value (3.66%) is still
Fig. 9. Liquid-liquid equilibrium for the system sunflower relatively high, indicating a difficulty in predicting the equilibrium
oil(1) þ diacylglycerol(2) þ monoacylglycerol(3) þ ethyl linoleate (4) þ ethanol(5) at
318.15 K: C, experimental data [10]; - - -, calculated values using UNIFAC-LLE; ,
of this type of system using the UNIFAC model. This can be
calculated values using UNIFAC-HIR; , calculated values using the new readjusted demonstrated by the analysis of Fig. 11. This figure shows that there
parameters. is indeed an improvement in the equilibrium prediction, especially
regarding biodiesel-rich phase. However, the new parameters
Table 5 yielded results that are not yet fully adequate.
Deviation values between experimental and calculated data for the validation sys- According to Kang et al. [76], certain strong interactions, present
tem (sunflower oil þ commercial mixture of mono- and diacylglycerols þ ethyl
linoleate þ ethanol).
in diols or glycerol, for example, cannot be described by the UNIFAC
additivity concept. According to them, it is unlikely that defining
Temperature/K Dw/% new subgroups can resolve this problem. In fact, other approaches
UNIFAC-LLE UNIFAC-HIR This work have been adopted in this work as attempts to further improve the
303.15 9.07 4.90 2.26 LLE prediction of these systems, such as to consider the entire
318.15 7.40 2.30 2.45 carbon chain of the glycerol molecule as an independent structural
group, to consider the whole methanol molecule as a single group,
to differentiate the groups ‘OH primary’ from ‘OH secondary’ and to
showing the improvement in the prediction of equilibrium data. differentiate the ester group (‘CH2COO’) of acylglycerol molecules
Thus, more important than assessing only the numerical value of from those of alkyl esters. However, none of these strategies
the calculated deviations, the analysis of the components behavior generated better results, indicating that the deviations observed in
in the system is essential to ensure the quality of the equilibrium this work must be associated with a limitation of the UNIFAC model
prediction. itself, and not only the interaction parameters used.
In general, it can be concluded that when proximity effects are As a last alternative to improve the prediction of liquid-liquid
considered, the newly defined functional groups are clearly supe- equilibrium involving compounds of the biodiesel production, it
rior to the groups currently used. The adoption of a new group was opted to adjust new parameters considering only these sys-
‘OHgly’ led to an improvement in the agreement with the experi- tems in the database. For this, the interaction parameters shown in
mental data without losing the simplicity of the group contribution Table 3 were used as initial estimate for the readjustment of new
approach. Kang et al. [76] also observed that using the UNIFAC parameters. It is worth noting that in this case, the structural group
‘COOH’, specific for acid molecules, was ignored since it does not
appear in any molecule of this new reduced database. Again, the
matrix of parameters of a resulting minimization was used for the
next until there were no further changes in the parameters. The
new matrix of parameters resulting from this readjustment pro-
cedure is presented in Table 7.
In general, the new readjustment procedure showed a

Table 6
Deviation values between experimental and calculated data for binary sub-systems.

System Temperature (K) Dw/% Reference

Glycerol þ methyl oleate 323.15e383.15 1.76 [20]

Triolein þ glycerol 323.15e383.15 3.46 [20]
FAME_castora þ water 303.15e333.15 1.41 [38]
FAME_castora þ glycerol 303.15e333.15 1.94 [38]
FAEE_castorb þ water 303.15e333.15 1.07 [38]
FAEE_castorb þ glycerol 303.15e333.15 1.66 [38]
Fig. 10. Average distribution coefficient (ki): ◊, experimental data [10]; , calculated
values using UNIFAC-LLE; D, calculated values using UNIFAC-HIR; B, calculated values a
Fatty acid methyl ester from castor oil.
b
using the new readjusted parameters. Fatty acid ethyl ester from castor oil.
 L.C.B.A. Bessa et al. / Fluid Phase Equilibria 425 (2016) 98e107 105

data is due mainly to the use of a specific database. Furthermore,

Fig. 11. Liquid-liquid equilibrium for the system ethylic biodiesel of palm
oil(1) þ ethanol(2) þ glycerol(3) at 323.15 K: C, experimental data [55]; - - -, calcu-
lated values using UNIFAC-LLE; , calculated values using UNIFAC-HIR;
lated values using the new readjusted parameters.
the removal of the group ‘COOH’ reduced the number of interaction
parameters to be readjusted (from 42 to 30), which facilitates the
computational work. It is worth noting that this new matrix should
be used only to systems present in the purification steps of bio-
diesel, i.e. systems composed of biodiesel/esters, ethanol or meth-
anol, water and/or glycerol. For systems containing fatty acids and
acylglycerols, these parameters are not recommended. Indeed, the
equilibrium data for the deacidification and transesterification
systems were calculated, using the parameters listed in Table 7, and
the overall average deviations obtained were 4.48 and 3.31%,
respectively.
Thus, one can conclude that when it is of interest, or if it is
necessary, using a single matrix of parameters from the oil
deacidification up to the biodiesel production, the use of the matrix
shown in Table 3 is suggested. However, when the system to be
studied comprises exclusively compounds involved in the biodiesel
purification, the matrix of Table 7 is the most recommended. As for
the case where only compounds present in vegetable oils deacidi-
fication are involved, both the parameters of Table 3, and those
presented by Hirata et al. [31] can be used with confidence.
, calcu-
4. Conclusions

Table 7 The results presented in this study contribute to a more precise

UNIFAC interaction parameters readjusted specifically for biodiesel systems. description of the actual behavior of the transesterification systems
CH C]C OH H2O COOC OHgly
involved in the biodiesel production. In addition, the results
confirm that the UNIFAC model without any changes in their pa-
CH 0.000 380.420 394.210 2577.210 1127.950 1170.100
rameters may give unsatisfactory predictions. Therefore, it should
C]C 2037.300 0.000 1192.650 74.650 12.050 91.750
OH 134.050 266.610 0.000 14.680 202.160 0.000 be used with caution or modified to obtain better estimations. A
H2O 129.970 351.070 78.610 0.000 7.200 85.320 first attempt was made to describe the molecules of the compounds
COOC 446.430 4.763 555.530 326.490 0.000 21.650 studied with the traditional groups originally proposed by the
OHgly 375.350 1720.100 0.000 8.415 23.250 0.000 model, but poor predictions were obtained, generating improper
deviations from experimental data. A new group ‘OHgly’ was then
introduced due to proximity effects, and two matrices of new pa-
significant improvement in the prediction of the equilibrium data, rameters have been readjusted. In general, adequate predictions
which is evidenced by the low overall mean deviation value were obtained and significant improvement of the performance of
(2.69%). This can also be seen from Fig. 12, where one can observe this group contribution method was achieved in comparison with
the great improvement in the system description. the results using UNIFAC parameters available in the literature. This
The significant improvement in the prediction of equilibrium modified UNIFAC model resulted in an important improvement in
the agreement between calculated and experimental data without
losing the simplicity of the group contribution approach.

Acknowledgements

The authors wish to acknowledge CAPES for the scholarship and

FAPESP (08/56258-8 and 09/54137-1), FAEPEX/UNICAMP and CNPq
(483340/2012-0, 406856/2013-3, 305870/2014-9 and 309780/
2014-4) for their financial support.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://

dx.doi.org/10.1016/j.fluid.2016.05.020.

References

[1] G. Knothe, Biodiesel and renewable diesel: a comparison, Prog. Energy

Combust. Sci. 36 (2010) 364e373.
[2] R. Coelho, P.G. dos Santos, M.R. Mafra, L. Cardozo-Filho, M.L. Corazza, (Vapor þ
liquid) equilibrium for the binary systems {water þ glycerol} and {ethanol þ
glycerol, ethyl stearate, and ethyl palmitate} at low pressures, J. Chem.
Thermodyn. 43 (2011) 1870e1876.
Fig. 12. Liquid-liquid equilibrium for the system ethylic biodiesel of palm [3] V.B. Veljkovic, I.B. Bankovic-Ili
c, O.S. Stamenkovi
c, Purification of crude bio-
oil(1) þ ethanol(2) þ glycerol(3) at 298.15 K: C, experimental data [55]; - - -, calcu- diesel obtained by heterogeneously-catalyzed transesterification, Renew.
lated values using UNIFAC-LLE; , calculated values using UNIFAC-HIR; , calcu- Sustain. Energy Rev. 49 (2015) 500e516.
lated values using the parameters readjusted specifically for biodiesel systems. [4] M.A. Mazutti, F.A.P. Voll, L. Cardozo-Filho, M.L. Corazza, M. Lanza, W.L. Priamo,
106 L.C.B.A. Bessa et al. / Fluid Phase Equilibria 425 (2016) 98e107
J.V. Oliveira, Thermophysical properties of biodiesel and related systems:
(Liquid þ liquid) equilibrium data for soybean biodiesel, J. Chem. Thermodyn.
58 (2013) 83e94.
[5] A.H. West, D. Posarac, N. Ellis, Assessment of four biodiesel production pro-
cesses using HYSYS. Plant, Bioresour. Technol. 99 (2008) 6587e6601.
[6] E.F. Andrade, L. Igarashi-Mafra, M.R. Mafra, M.L. Corazza, (Liquid þ liquid)
equilibrium for the system {ethyl stearate(1) þethanol(2) þ glycerol(3)},
J. Chem. Thermodyn. 47 (2012) 213e218.
[7] F.R. do Carmo, N.S. Evangelista, R.S. de Santiago-Aguiar, F.A.N. Fernandes,
H.B. de Sant’Ana, Evaluation of optimal activity coefficient models for
modeling and simulation of liquideliquid equilibrium of biodiesel þ
glycerol þ alcohol systems, Fuel 125 (2014) 57e65.
[8] S. Zerpa, M. Satyro, M.A. Clarke, A short note on the use of an equation of state
(EOS) based approach to modelling the thermodynamics of biodiesel systems,
Fuel Process. Technol. 121 (2014) 70e75.
[9] R.C. Basso, C.A.S.D. Silva, C.D.O. Sousa, A.J.D.A. Meirelles, E.A.C. Batista, LLE
experimental data, thermodynamic modeling and sensitivity analysis in the
ethyl biodiesel from macauba pulp oil settling step, Bioresour. Technol. 131
(2013) 468e475.
[10] L.C.B.A. Bessa, M.C. Ferreira, S. Shiozawa, E.A.C. Batista, A.J.A. Meirelles,
(Liquid þ liquid) equilibrium of systems involved in the stepwise ethanolysis
of vegetable oils, J. Chem. Thermodyn. 89 (2015) 148e158.
[11] M.C. Ferreira, L.C.B.A. Bessa, S. Shiozawa, A.J.A. Meirelles, E.A.C. Batista, Liq-
uideliquid equilibrium of systems containing triacylglycerols (canola and
corn oils), diacylglycerols, monoacylglycerols, fatty acids, ester and ethanol at
T/K ¼ 303.15 and 318.15, Fluid Phase Equilibria 404 (2015) 32e41.
[12] S. Shiozawa, L.C.B.A. Bessa, M.C. Ferreira, A.J.A. Meirelles, E.A.C. Batista, Liq-
uideLiquid equilibrium data for fatty systems containing monoacylglycerols
and diacylglycerols, J. Chem. Eng. Data 60 (2015) 2371e2379.
[13] W.G. Chapman, K.E. Gubbins, G. Jackson, M. Radosz, New reference equation
of state for associating liquids, Ind. Eng. Chem. Process Des. Dev. 29 (1990)
1709e1721.
[14] G.M. Kontogeorgis, E.C. Voutsas, I.V. Yakoumis, D.P. Tassios, An equation of
state for associating fluids, Ind. Eng. Chem. Process Des. Dev. 35 (1996)
4310e4318.
[15] M.B. Oliveira, A.J. Queimada, J.A.P. Coutinho, Modeling of biodiesel multi-
component systems with the cubic-plus-association (CPA) equation of state,
Ind. Eng. Chem. Process Des. Dev. 49 (2010) 1419e1427.
[16] M.B. Oliveira, A.R.R. Teles, A.J. Queimada, J.A.P. Coutinho, Phase equilibria of
glycerol containing systems and their description with the Cubic-Plus-
Association (CPA) Equation of State, Fluid Phase Equilibria 280 (2009) 22e29.
[17] L.A. Follegatti-Romero, M.B. Oliveira, E.A.C. Batista, J.A.P. Coutinho,
A.J.A. Meirelles, Liquideliquid equilibria for ethyl esters þ ethanol þ water
systems: experimental measurements and CPA EoS modeling, Fuel 96 (2012)
327e334.
[18] L.A. Follegatti-Romero, M.B. Oliveira, F.R.M. Batista, E.A.C. Batista,
J.A.P. Coutinho, A.J.A. Meirelles, Liquideliquid equilibria for ternary systems
containing ethyl esters, ethanol and glycerol at 323.15 and 353.15 K, Fuel 94
(2012) 386e394.
[19] M.L. Corazza, W.A. Fouad, W.G. Chapman, PC-SAFT predictions of VLE and LLE
of systems related to biodiesel production, Fluid Phase Equilibria 416 (2016)
130e137.
[20] C. Silva, L. Soh, A. Barberio, J. Zimmerman, W.D. Seider, Phase equilibria of
triolein to biodiesel reactor systems, Fluid Phase Equilibria 409 (2016)
171e192.
[21] A. Fredenslund, R.L. Jones, J.M. Prausnitz, Group-contribution estimation of
activity coefficients in nonideal liquid mixtures, Am. Inst. Chem. Eng. 21
(1975) 1086e1099.
[22] S. Skjold-Jorgensen, B. Kolbe, J. Gmehling, P. Rasmussen, Vapor-liquid equi-
libria by UNIFAC group contribution. Revision and extension, Ind. Eng. Chem.
Process Des. Dev. 18 (1979) 714e722.
[23] J. Gmehling, P. Rasmussen, A. Fredenslund, Vapor-liquid equilibriums by
UNIFAC group contribution. Revision and extension. 2, Ind. Eng. Chem. Pro-
cess Des. Dev. 21 (1982) 118e127.
[24] E.A. Macedo, U. Weidlich, J. Gmehling, P. Rasmussen, Vapor-liquid equilib-
riums by UNIFAC group contribution. Revision and extension. 3, Ind. Eng.
Chem. Process Des. Dev. 22 (1983) 676e678.
[25] D. Tiegs, P. Rasmussen, J. Gmehling, A. Fredenslund, Vapor-liquid equilibria by
UNIFAC group contribution. 4. Revision and extension, Ind. Eng. Chem. Res. 26
(1987) 159e161.
[26] H.K. Hansen, P. Rasmussen, A. Fredenslund, M. Schiller, J. Gmehling, Vapor-
liquid equilibria by UNIFAC group contribution. 5. Revision and extension, Ind.
Eng. Chem. Res. 30 (1991) 2352e2355.
[27] T. Magnussen, P. Rasmussen, A. Fredenslund, UNIFAC parameter table for
prediction of liquid-liquid equilibria, Industrial Eng. Chem. Process Des. Dev.
20 (1981) 331e339.
[28] F.A.P. Voll, L.R.S. Kanda, L.C. Filho, M.L. Corazza, (Liquid þ liquid) equilibrium
for the system (hydrolyzed palm oil þ ethanol þ water) for diacylglycerol
enrichment, J. Chem. Thermodyn. 58 (2013) 1e7.
[29] R.C. Basso, F.H. Miyake, A.J. de Almeida Meirelles, E.A.C. Batista, Liquideliquid
equilibrium data and thermodynamic modeling, at T/K ¼ 298.2, in the
washing step of ethyl biodiesel production from crambe, fodder radish and
macauba pulp oils, Fuel 117 (Part A) (2014) 590e597.
[30] R.S. Pinheiro, A.M.M. Bessa, B.A. de Queiroz, A.M.S.F. Duarte, H.B. de Sant’Ana,
R.S. De Santiago-Aguiar, Optimization of the methylic biodiesel purification
process by intermediate of liquid-liquid equilibrium data for ternary systems
containing methanol þ water þ (soybean, corn or brown shell of coconut)
biodiesel, Fluid Phase Equilibria 361 (2014) 30e36.
[31] G.F. Hirata, C.R.A. Abreu, L.C.B.A. Bessa, M.C. Ferreira, E.A.C. Batista,
A.J.A. Meirelles, Liquid-liquid equilibrium of fatty systems: a new approach for
adjusting UNIFAC interaction parameters, Fluid Phase Equilibria 360 (2013)
379e391.
[32] J.D. Gonçalves, M. Aznar, G.R. Santos, Liquid-liquid equilibrium data for sys-
tems containing Brazil nut biodiesel þ methanol þ glycerin at 303.15 K and
323.15 K, Fuel 133 (2014) 292e298.
[33] M.B. Oliveira, S. Barbedo, J.I. Soletti, S.H.V. Carvalho, A.J. Queimada,
J.A.P. Coutinho, Liquideliquid equilibria for the canola oil biodiesel þ
ethanol þ glycerol system, Fuel 90 (2011) 2738e2745.
[34] M. Rostami, S. Raeissi, M. Mahmoodi, M. Nowroozi, Liquid-liquid equilibria in
biodiesel production, J. Amer Oil Chem. Soc. 90 (2013) 147e154.
[35] M. Hakim, H. Abedini Najafabadi, G. Pazuki, M. Vossoughi, Novel approach for
liquideliquid phase equilibrium of biodiesel (canola and sunflower) þ
glycerol þ methanol, Ind. Eng. Chem. Process Des. Dev. 53 (2013) 855e864.
[36] A.B. Machado, Y.C. Ardila, L.H. de Oliveira, M.N. Aznar, M.R. Wolf Maciel,
LiquidLiquid equilibrium study in ternary Castor oil biodiesel þ ethanol þ
glycerol and quaternary Castor oil biodiesel þ ethanol þ glycerol þ NaOH
systems at (298.2 and 333.2) K, J. Chem. Eng. Data 56 (2011) 2196e2201.
[37] Y.C. Ardila, A.B. Machado, G.M.F. Pinto, R.M. Filho, M.R.W. Maciel, Liquid-
eLiquid equilibrium in ternary systems present in biodiesel purification from
soybean oil and Castor oil at (298.2 and 333.2) K, J. Chem. Eng. Data 58 (2013)
605e610.
[38] M.A. Mazutti, F.A.P. Voll, L. Cardozo-Filho, M.L. Corazza, M. Lanza, W.L. Priamo,
J.V. Oliveira, Thermophysical properties of biodiesel and related systems:
(Liquid þ liquid) equilibrium data for castor oil biodiesel, J. Chem. Thermodyn.
62 (2013) 17e26.
[39] F.M.R. Mesquita, N.S. Evangelista, H.B. de Sant’Ana, R.S. de Santiago-Aguiar,
Liquideliquid equilibrium for the glycerol þ alcohol þ coconut biodiesel
system at different temperatures and atmospheric pressure, J. Chem. Eng.
Data 57 (2012) 3557e3562.
[40] R.S. Pinheiro, A.M.M. Bessa, B.A. de Queiroz, A.M.S.F. Duarte, H.B. de Sant’Ana,
R.S. De Santiago-Aguiar, Optimization of the methylic biodiesel purification
process by intermediate of liquid-liquid equilibrium data for ternary systems
containing methanolþwaterþ(soybean, corn or brown shell of coconut) bio-
diesel, Fluid Phase Equilib. 361 (2014) 30e36.
[41] S.N. Csernica, J.T. Hsu, Reverse-Phase Ternary Phase Diagram, Tie Lines,
Plait Point, For commercial BiodieselGlycerolMethanol, Ind. Eng. Chem.
Process Des. Dev. 50 (2010) 1012e1016.
[42] M. Rostami, S. Raeissi, M. Ranjbaran, M. Mahmoodi, M. Nowroozi, Experi-
mental investigation and modeling of liquideliquid equilibria in two systems
of concern in biodiesel production, Fluid Phase Equilibria 353 (2013) 31e37.
[43] F.M.R. Mesquita, A.M.M. Bessa, D.D. de Lima, H.B. de Sant’Ana, R.S. de San-
tiago-Aguiar, Liquideliquid equilibria of systems containing cottonseed
biodiesel þ glycerol þ ethanol at 293.15, 313.15 and 333.15 K, Fluid Phase
Equilibria 318 (2012) 51e55.
[44] A.M.M. Bessa, R.S. Pinheiro, N.S. Evangelista, F.R.D. Carmo, H.B.D. Sant’Ana,
R.S.D. Santiago-Aguiar, Liquideliquid equilibrium for cottonseed biodiesel þ
water þ alcohol (Methanol/Ethanol) systems at (293.15 and 313.15) K:
experimental data and thermodynamic modeling, J. Chem. Eng. Data 60
(2015) 707e713.
[45] R.C. Basso, A.J. de Almeida Meirelles, E.A.C. Batista, Liquid-liquid equilibrium
of pseudoternary systems containing glycerolþethanolþethylic biodiesel
from crambe oil (Crambe abyssinica) at T/K¼(298.2, 318.2, 338.2) and ther-
modynamic modeling, Fluid Phase Equilib. 333 (2012) 55e62.
[46] L.A. Follegatti-Romero, M. Lanza, F.R.M. Batista, E.A.C. Batista, M.B. Oliveira,
J.o.A.P. Coutinho, A.J.A. Meirelles, Liquidliquid equilibrium for ternary sys-
tems containing ethyl esters, anhydrous ethanol and water at 298.15, 313.15,
and 333.15 K, Ind. Eng. Chem. Process Des. Dev. 49 (2010) 12613e12619.
[47] L.R. Kanda, F.A.P. Voll, M.L. Corazza, LLE for the systems ethyl palmitate
(palmitic acid)(1)þethanol(2)þglycerol (water)(3), Fluid Phase Equilibria 354
(2013) 147e155.
[48] J.R.F. Silva, M.A. Mazutti, F.A.P. Voll, L. Cardozo-Filho, M.L. Corazza, M. Lanza,
W.L. Priamo, J. Vladimir Oliveira, Thermophysical properties of biodiesel and
related systems: (Liquid þ liquid) equilibrium data for Jatropha curcas bio-
diesel, J. Chem. Thermodyn. 58 (2013) 467e475.
[49] M.-J. Lee, Y.-C. Lo, H.-M. Lin, Liquideliquid equilibria for mixtures containing
water, methanol, fatty acid methyl esters, and glycerol, Fluid Phase Equilibria
299 (2010) 180e190.
[50] S.C. Beneti, W.L. Priamo, M. Lanza, M.H. Kunita, M.A. Mazutti, L. Cardozo Filho,
J.V. Oliveira, Determination of liquid-liquid equilibrium data for biodiesel
containing ternary systems using near infrared technique, Braz. J. Chem. Eng.
30 (2013) 23e31.
[51] A.E. Andreatta, Liquid-liquid equilibria in ternary mixtures of methyl oleate þ
ethanol þ glycerol at atmospheric pressure, Industrial Eng. Chem. Res. 51
(2012) 9642e9651.
[52] R. Tizvar, D.D. McLean, M. Kates, M.A. Dube , Optimal separation of glycerol
and methyl oleate via liquid-liquid extraction, J. Chem. Eng. Data 54 (2009)
1541e1550.
[53] A.E. Andreatta, L.M. Cas as, P. Hegel, S.B. Bottini, E.A. Brignole, Phase equilibria
in ternary mixtures of methyl oleate, glycerol, and methanol, Ind. Eng. Chem.
Res. 47 (2008) 5157e5164.
 L.C.B.A. Bessa et al. / Fluid Phase Equilibria 425 (2016) 98e107
[54] J.C. Bell, R.A. Messerly, R. Gee, A. Harrison, R.L. Rowley, W.V. Wilding, Ternary
liquideliquid equilibrium of biodiesel compounds for systems consisting of a
methyl ester þ glycerin þ water, J. Chem. Eng. Data 58 (2013) 1001e1004.
[55] E.G.d.A. Rocha, L.A. Follegatti-Romero, S. Duvoisin Jr., M. Aznar, Liquideliquid
equilibria for ternary systems containing ethylic palm oil biodiesel þ
ethanol þ glycerol/water: experimental data at 298.15 and 323.15 K and
thermodynamic modeling, Fuel 128 (2014) 356e365.
[56] M. Rostami, S. Raeissi, M. Mahmoudi, M. Nowroozi, LiquideLiquid phase
equilibria of systems of palm and soya biodiesels: experimental and modeling,
Ind. Eng. Chem. Process Des. Dev. 51 (2012) 8302e8307.
[57] F.M.R. Mesquita, F.X. Feitosa, N.E. Sombra, R.S. de Santiago-Aguiar, H.B. de
Sant’Ana, Liquideliquid equilibrium for ternary mixtures of biodiesel (soy-
bean or sunflower) þ glycerol þ ethanol at different temperatures, J. Chem.
Eng. Data 56 (2011) 4061e4067.
[58] A.B. Machado, Y.C. Ardila, L.H. de Oliveira, M. Aznar, M.R. Wolf Maciel, Liq-
uideLiquid equilibria in ternary and quaternary systems present in biodiesel
production from soybean oil at (298.2 and 333.2) K, J. Chem. Eng. Data 57
(2012) 1417e1422.
[59] B.B. França, H.G.D.A. Villardi, F.L.P. Pessoa, A.M.C. Uller, Liquideliquid equi-
librium for compounds of soybean ethyl biodiesel production by trans-
esterification, J. Chem. Eng. Data 58 (2013) 1927e1933.
[60] Y.C. Ardila, in: Liquid-liquid Extraction Systems for Biodiesel Purification
Processes, School of Chemical Engineering, University of Campinas, Campinas,
2009.
[61] X. Liu, X. Piao, Y. Wang, S. Zhu, Liquideliquid equilibrium for systems of (fatty
acid ethyl esters þ ethanol þ soybean oil and fatty acid ethyl esters þ
ethanol þ glycerol), J. Chem. Eng. Data 53 (2008) 359e362.
[62] S.C. Beneti, M. Lanza, M.A. Mazutti, M.H. Kunita, L. Cardozo-Filho, J. Vladimir
Oliveira, Experimental (liquid þ liquid) equilibrium data for ternary and
quaternary mixtures of fatty acid methyl and ethyl esters (FAME/FAEE) from
soybean oil, J. Chem. Thermodyn. 68 (2014) 60e70.
[63] L.C.B.A. Bessa, E.A.C. Batista, A.J.A. Meirelles, Liquid-liquid equilibrium of
reactional systems of soybean oil ethanolysis, in: VIII Brazilian Congress of
Applied Thermodynamics, Aracaju, Brazil, 2015 (in Portuguese).
107
[64] J.A. Nelder, R. Mead, A simplex-method for function minimization, Comput. J.
7 (1965) 308e313.
[65] C. Marcolli, T. Peter, Water activity in polyol/water systems: new UNIFAC
parameterization, Atmos. Chem. Phys. 5 (2005) 1545e1555.
[66] S.I. Sandler, Models for Thermodynamic and Phase Equilibria Calculations,
Taylor & Francis, 1993.
[67] H.S. Wu, S.I. Sandler, Proximity effects on the predictions of the UNIFAC
model: I. Ethers, Am. Inst. Chem. Eng. 35 (1989) 168e172.
[68] H.S. Wu, S.I. Sandler, Use of ab initio quantum mechanics calculations in group
contribution methods. 1. Theory and the basis for group identifications, Ind.
Eng. Chem. Process Des. Dev. 30 (1991) 881e889.
[69] C. Peng, M.N. Chan, C.K. Chan, The hygroscopic properties of dicarboxylic and
multifunctional Acids: measurements and UNIFAC predictions, Environ. Sci.
Technol. 35 (2001) 4495e4501.
[70] L. Ninni, M.S. Camargo, A.J.A. Meirelles, Water activity in poly(ethylene glycol)
aqueous solutions, Thermochim. Acta 328 (1999) 169e176.
[71] M. Ozdemir, H. Sadikoglu, Use of UNIFAC in the prediction of water activity
values of aqueous polyol solutions, J. Food Process. Preserv. 23 (1999)
109e120.
[72] B.L. Larsen, P. Rasmussen, A. Fredenslund, A modified UNIFAC group-
contribution model for prediction of phase equilibria and heats of mixing,
Ind. Eng. Chem. Process Des. Dev. 26 (1987) 2274e2286.
[73] L. Ninni, M.S. Camargo, A.J.A. Meirelles, Water activity in polyol systems,
J. Chem. Eng. Data 45 (2000) 654e660.
[74] A.M. Peres, E.A. Macedo, A modified UNIFAC model for the calculation of
thermodynamic properties of aqueous and non-aqueous solutions containing
sugars, Fluid Phase Equilib. 139 (1997) 47e74.
[75] M. Ansolin, R.C. Basso, A.J.A. Meirelles, E.A.C. Batista, Experimental data for
liquideliquid equilibrium of fatty systems with emphasis on the distribution
of tocopherols and tocotrienols, Fluid Phase Equilib. 338 (2013) 78e86.
[76] J.W. Kang, V. Diky, M. Frenkel, New modified UNIFAC parameters using crit-
ically evaluated phase equilibrium data, Fluid Phase Equilib. 388 (2015)
128e141.