Journal of Food Engineering 65 (2004) 577–583

www.elsevier.com/locate/jfoodeng

Phase equilibrium modeling in binary mixtures found

in wine and must distillation
undez a, Jos
e O. Valderrama
Claudio A. Fa
b,c,*

a
Fac. de Ciencias, Univ. de La Serena, Casilla 554, La Serena, Chile
b
Fac. de Ingenier
ıa, Univ. de La Serena, Casilla 554, La Serena, Chile
c
Centro de Informaci
on Tecnol
ogica (CIT), Casilla 724, La Serena, Chile
Received 25 September 2002; received in revised form 13 August 2003; accepted 16 February 2004

Abstract
Phase equilibrium in binary mixtures found in wine and must distillation processes have been analyzed using equations of state
and models for activity coefficients. In wine and must distillation the presence of polar substances found in the mixture to be distilled
and the very low concentration of many components (different from ethanol and water), called congeners, make it difficult to model
these mixtures, for simulation and design purposes. In this study, two liquid-phase models, NRTL and UNIFAC and a predictive
equation of state, the PSRK equation, have been used. Nine binary ethanol + congener mixtures and nine binary water + congener
mixtures have been considered for analysis. The congeners used in this study are considered as legal compounds by the Chilean
legislation for the production of a spirit called Pisco. Comparison with available literature data is done. Conclusions on the accuracy
of the models used in the study are drawn and recommendations on which models should be considered in these situations are
given.
Ó 2004 Elsevier Ltd. All rights reserved.

Keywords: Distillation; Wine; Must; Thermodynamic; Phase equilibrium

1. Introduction called Pisco (Alonso, 1985; Baquerizo, 1999). However,

In alcoholic distillation the great amount of sub-
stances found in the mixture to be distilled, called must
or wine, and the very low concentration of many other
components (different from ethanol and water), called
congeners, makes it difficult to correlate and predict the
distilled concentration, considered to be the most
important variable in the spirit to be distilled. Several of
the congener compounds are essential part of the aroma
of the distilled products and therefore their concentra-
tions are important enological parameters (Lora, Iborra,
P
erez, & Carbonell, 1992). These congener substances
are chemically complex and are usually present in con-
centrations of part per million, 10
6 to 10
4 mg/L
(Hikari & Kubo, 1975). More than one hundred cong-
eners have been found in a wine to produce a spirit
of all these compounds only few of them are controlled
by legislation, meaning that their concentration cannot
be higher than that established limit. Table 1 shows a list
of substances present in a typical must that in the
Chilean legislation are considered as legal compounds for
the production of Pisco.
The operating pressure in alcoholic distillation to
produce Pisco is of the order of the atmospheric pres-
sure, the temperature varies between 20 and 110 °C, and
most of the substances involved are highly polar. As
known, phase equilibrium conditions need to be deter-
mined to model the distillation process (Billet, 1979;
Valderrama, Rojas, Garc
es, & Rosell
o, 1998). This pa-
per deals with modeling vapor–liquid equilibrium data
of binary mixtures to be used in distillation simulation.

2. Modeling vapor–liquid equilibrium

*
Corresponding author. The problem of phase equilibrium consists of the
E-mail addresses: cfaundez@userena.cl (C.A. Fa

undez), jvald- calculation of some variables of the set ðT ; P ; x; yÞ, if
err@userena.cl (J.O. Valderrama). some of them are known. For a given mixture, the

0260-8774/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jfoodeng.2004.02.023
578 C.A. Fa
undez, J.O. Valderrama / Journal of Food Engineering 65 (2004) 577–583

Nomenclature

Symbols xi , xj experimental mole fraction of congener in the

A1 constant in the PSRK EoS liquid phase
ac , b parameter in the PSRK EoS W objective function
aij , bij interaction parameters in the mixing rules
Abbreviations
Aij , Aji , Bij , Bji parameters in the NRTL model
cal calculated
c1 , c2 , c3 parameters in the PSRK EoS
EoS equation of state
g0E excess Gibbs free energy
exp experimental
P pressure
NRTL non-random two liquid model
r, q UNIFAC parameters
UNIFAC universal functional activity coefficient
R ideal gas constant
model
T temperature
PSRK predictive Soave–Redlich–Kwong equation
Tc critical temperature
of state
Tb normal boiling point temperature
SRK Soave–Redlich–Kwong equation of state
Tr reduced temperature ðTr ¼ T =Tc Þ
%D percent deviation
V volume
Vc critical volume Greek letters
y1 mole fraction of congener in the vapor phase a temperature function in the PSRK EoS
(component 1) d empirical parameter in the NRTL model
ycalc calculated mole fraction of a congener in the u fugacity coefficient
vapor phase c activity coefficient
yexp experimental mole fraction of a congener in
the vapor phase

Table 1
Properties for all substances involved in this study
Components M Tb (°C) Tc (°C) Pc (MPa) Vc (m3 /kmol) x
Acetic acid 60.05 117.9 318.8 5.79 0.180 0.4665
Acetaldehyde 44.05 20.9 192.9 5.50 0.154 0.2907
Ethyl acetate 88.11 77.1 250.2 3.88 0.286 0.3664
Furfural 96.09 161.7 397.0 5.66 0.252 0.3678
Methanol 32.04 64.7 239.5 8.10 0.118 0.5640
3-Methylbutanol 88.15 131.2 306.3 3.88 0.327 0.5558
2-Methyl-1-propanol 74.12 107.7 274.6 4.30 0.273 0.5848
1-Pentanol 88.15 137.8 313.0 3.88 0.326 0.5938
1-Propanol 60.10 97.2 263.6 5.18 0.219 0.6218
Water 18.02 100.0 374.2 22.12 0.064 0.3480
Ethanol 46.07 78.3 240.8 6.15 0.167 0.6452

number of variables F to be established, so the system The fugacity of a component in the vapor phase is
is completely defined, is determined by the Gibbs phase V
usually expressed through the fugacity coefficient / i
rule: F ¼ c
p þ 2, being ‘‘c’’ the number of compo-
 VP
fiV ¼ yi / ð2Þ
nents in the mixture and ‘‘p’’ the number of phases. i

For a vapor–liquid mixture in thermodynamic equilib- while the fugacity of a component in the liquid phase is
rium, the temperature and the pressure are the same expressed through either the fugacity coefficient /  L or
i
in both phases, and the remaining variables are defined the activity coefficient ci :
by the material balance and the ‘‘fundamental equa-
tion of phase equilibrium’’. This fundamental equa-  LP
fiL ¼ xi / ð3Þ
i
tion can be expressed as the equality of fugacities of
each component in the mixture in both phases (Walas, fiL ¼ xi cifi0 ð4Þ
1985).
At low pressures the standard state fugacity fi0 can be
replaced by the vapor pressure at the temperature of the
fiL ¼ fiV ð1Þ system, Pisat . Thus,
 C.A. Fa
undez, J.O. Valderrama / Journal of Food Engineering 65 (2004) 577–583 579

fiL ¼ xi ci Pisat ð5Þ state + gE model’’, applies to both phases and does not
require empirical parameters.
In these equations yi is the mole fraction of component For the NRTL model (Non-Random Two Liquids),
‘‘i’’ in the vapor phase, xi is the mole fraction of com- three adjustable empirical parameters were obtained by
ponent ‘‘i’’ in the liquid phase, and P is the pressure. The regression analysis of experimental data. The UNIFAC
fugacity is related to the temperature, the pressure, the model (Universal Functional Activity Coefficient), is a
volume and the concentration though a standard ther- predictive tool which considers two contributions: one
modynamic relation (Walas, 1985). If the fugacity to account for non-ideality of a mixture arising from
coefficient is used in both phases, the method of solution differences in size and shape of the components in the
of the phase equilibrium problem is known as ‘‘the mixture (combinatorial contribution) and a residual part
equation of state method’’. If the fugacity coefficient is that considers the difference between intermolecular and
used for the vapor phase and the activity coefficient intramolecular interaction energies (residual contribu-
is used for the liquid phase the equilibrium problem is tion).
known as ‘‘the gamma–phi method’’. Modern methods
for phase equilibrium correlation include an excess 3.1. NRTL equation
Gibbs free energy model gE in the mixing rules of an
equation of state, giving origin to the so-called ‘‘equa- The NRTL model for the activity coefficient at a gi-
tion of state + gE model’’. The present study includes ven absolute temperature T in Kelvin, has the following
one of these models. form (Prausnitz, Lichtenthaler, & Gomes de Azevedo,
Commonly, at the conditions at which must and wine 1999):
distillation processes take place, models for the activity PN " PN #
j sji Gji xi
X
N
coefficients in the ‘‘gamma–phi method’’ are used. This xj Gij k ski xk Gkj
ln ci ¼ PN þ PN sij
PN
means that an activity coefficient model ðcÞ is used to
k Gki xk j k Gkj xk k Gkj xk
describe the complex liquid phase, and a fugacity coef-
ficient model ð/Þ is used to describe the low-pressure ð6Þ
vapor phase (Gmehling, Onken, & Arlt, 1982). How- with
ever, the EoS method ‘‘equation of state + gE model’’
can be used (Orbey & Sandler, 1998). sji ¼ Aji þ Bji =T Gji ¼ expð
dji  sji Þ dij ¼ dji ð7Þ
Most models available in the literature are of the The terms Aji , Aij , Bij , Bji and dij represent adjustable
correlating type, meaning that experimental data are parameters to be calculated from experimental vapor-
needed to calculate certain empirical parameters, al- equilibrium data. The NRTL equation may be used
though some predictive models are also available. The either as a three-parameter (Bij , Bji and dij only) or a
models known as van Laar, Margules, Redlich–Kister, five-parameter (Aji , Aij , Bij , Bji and dij equation). Com-
NRTL and Wilson, among others, are correlating monly, the three-parameter model is used.
models while UNIFAC and the PSRK equation of state
are predictive models (Chemstations, 2001). In corre- 3.2. UNIFAC model
lating models such as those mentioned above, the
parameters of the models are determined using experi- In the UNIFAC model, the liquid phase activity
mental vapor–liquid equilibrium data for binary systems coefficients for each species are calculated from the
(Valderrama, Pizarro, & Rojas, 2001; Valderrama, UNIFAC group contribution method. The UNIFAC
Rojas, & Pizarro, 2000). equation has the following form (Fredenslund, Gmeh-
ling, & Rasmussen, 1977):

3. The models studied ln ci ¼ ln cci þ ln cR

i ð8Þ
with the combinatorial part ln cci and residual part ln cR
i
In this paper, nine binary mixtures water + congener given by:
and nine binary mixtures ethanol + congener are ana-
lyzed using the predictive models UNIFAC and PSRK / z hi / X
ln cci ¼ ln þ qi ln þ li
i x j lj ð9Þ
and the correlating model NRTL. Available experi- xi 2 /i xi j
mental data has been used and the necessary binary " ! #
interaction parameters for the NRTL model, three for X X hi sij
R
ln ci ¼ qi 1
ln hj sji
P ð10Þ
each mixture, have been calculated by regression ana- k hk skj
j j
lysis. When the UNIFAC and the NRTL models are
used for the liquid phase, the Soave–Redlich–Kwong Details on how to calculate the different contributions
equation of state (Soave, 1972) is used for the gas phase. and parameters in the UNIFAC model are given in
The predictive PSRK model is of the type ‘‘equation of Table 3.
580 C.A. Fa
undez, J.O. Valderrama / Journal of Food Engineering 65 (2004) 577–583

3.3. The PSRK model Table 3

UNIFAC group data and contribution parameters r and q
The PSRK (Predictive Soave–Redlich–Kwong) Components Sub- Contribution r q
model is an equation of state based on the group con- group
tribution method of Mollerup (1981) that considers the Acetic acid 1 1 2.2024 2.0720
Soave–Redlich–Kwong equation of state (see Table 2) 43 1
and the UNIFAC model for the excess free energy and Acetaldehyde 1 1 1.8991 1.7960
the activity coefficients. The model was first proposed by 21 1
Holderbaun and Gmehling (1991), as follows: Ethyl acetate 1 2 3.4786 3.1160
RT a 23 1
P¼
ð11Þ
v
b vðv þ bÞ Furfural 62 1 3.1680 2.4810
2
R Tci2 Methanol 16 1 1.4311 1.4320
a1 ¼ 0:42748 aðT Þ ð12Þ
Pci 3-Methylbutanol 1 2 4.5979 4.2040
h
2 2 2
aðT Þ ¼ 1 þ c1 ð1
Tr0:5 Þ þ c2 ð1
Tr0:5 Þ
3 1
i2 15 1
þ c3 ð1
Tr0:5 Þ3 Tr < 1 ð13Þ
2-Methyl-1-propanol 1 2 3.9235 3.6640

2 2 1
aðT Þ ¼ 1 þ c1 ð1
Tr0:5 Þ Tr > 1 ð14Þ 3 1
In these equations, Tc is the critical temperature, 15 1
Tr ¼ T =Tc is the reduced temperature and c1 , c2 and c3 1-Pentanol 1 1 4.5987 4.2080
are empirical parameters, available in the literature. 2 4
The mixing rules, which arise from combining the 15 1
equation of state and a model for the excess Gibbs free 1-Propanol 1 1 3.2499 3.1280
energy (Huron & Vidal, 1978; Mollerup, 1981), are: 2 2
 E X  15 1
g ai RT X b
a¼b 0 þ x1 þ xi ln Ethanol 1 1 2.5755 2.5880
A1 b i Ai bi
X 2 1
b¼ x i bi ð15Þ 15 1
i¼1
Water 17 1 0.9200 1.4000
In these equations, ai and bi are the pure component
equation of state constants, A1 is a constant A1 ¼

0:64663, and g0E is the excess Gibbs free energy calcu- model. In the case of the predictive models the results of
lated using UNIFAC. Table 3 shows the UNIFAC binary mixtures provide insight about the behavior of the
group and contribution parameters r and q used for the models in mixtures containing highly polar components.
PSRK model for all substances involved in this study.
Although this papers deals with binary mixtures, the
models employed can be directly applied to multicom- 4. Mixtures studied
ponent mixtures. However, the information provided by
this study on binary mixtures is first needed to obtain The mixtures studied considered nine binary etha-
binary interaction parameters in the case of the NRTL nol + congener mixtures and nine water + congener

Table 2
Parameters for the UNIFAC model and SRK EoS
UNIFAC model SRK EoS
PN
li ¼ 2z ðri
qi Þ
ri
1, z ¼ 10 b¼ i xi bi
l
l  P ðiÞ PN
hi ¼ Pqi xri x , /i ¼ Pri xri x , sji ¼ exp
jiRT ij , ri ¼ k vk RK b ¼ i 0:08664 RT
Pci
ci
j j j j
j j
P ðiÞ PN PN
qi ¼ Vwk
k vk Qk , Rk ¼ 15:17 a¼ i j xi xj ðai aj Þ0:5 ð1
kij Þ
Awk R
P ðiÞ ðiÞ
Qk ¼ 2:510 9 , ln ci ¼ k vk ðln 1k
ln 1k Þ ai ¼ aci ai
2
All groupsh aci ¼ 0:42748 ðRTPcici Þ
P  P hm wkm i
ln sk ¼ Qk 1
ln m hm wmk
m wnm a0:5
i ¼ 1 þ mi ð1
Tri0:5 Þ
hm ¼ PQmQXmX mi ¼ 0:48 þ 1:574xi
0:176x2i
n n
n

Unm
 
wmn ¼ exp
UmnRT ¼ exp
aTmn
 C.A. Fa
undez, J.O. Valderrama / Journal of Food Engineering 65 (2004) 577–583 581

Table 4
Ranges of temperature, liquid phase mole fraction and vapor phase mole fraction for the 18 binary mixtures studied
Major component Congener DT (°C) x1 range y1 range
Water Acetic acid 100–118 0–1 0–1
Acetaldehyde 33–100 0–0.2 0–0.93
Ethyl acetate 71–76 0.124–0.988 0.693–0.917
Furfural 98–135 0.004–0.949 0.021–0.428
Methanol 65–100 0.0004–0.987 0.003–0.978
3-Methylbutanol 95–130 0.001–0.988 0.015–0.939
2-Methyl-1-propanol 90–107 0.005–0.984 0.150–0.931
1-Pentanol 96–120 0.01–0.90 0.144–0.441
1-Propanol 88–98 0.003–0.955 0.071–0.878

Ethanol Acetic acid 77–116 0–1 0–1

Acetaldehyde 20–78 0–1 0–1
Ethyl acetate 72–77 0.095–0.95 0.137–0.899
Furfural 78–134 0.020–0.98 0.005–0.350
Methanol 65–78 0–1 0–1
3-Methylbutanol 78–131 0–1 0–1
2-Methyl-1-propanol 79–106 0.034–0.961 0.010–0.891
1-Pentanol 80–132 0.083–0.985 0.012–0.875
1-Propanol 78–97 0–1 0–1
The pressure is 0.1013 MPa for all systems, except ethanol + acetaldehyde for which the pressure is 0.093 MPa.

mixtures. The congeners considered are acetic acid,
acetaldehyde, ethyl acetate, furfural, methanol, 3-
methylbutanol, 2-methyl-1-propanol, 1-pentanol and 1-
propanol. These are the substances considered as legal
compounds by the Chilean legislation for the production
of Pisco (Valderrama, Huerta, & Alarc
on, 2002).
Table 1 shows the pure component properties for all
the substances involved in this study. In the table, M is
the molecular weight, Tc is the critical temperature, Tb
is the normal boiling point temperature, Pc is the critical
pressure, Vc is the critical volume and x is the acentric
factor. The data were obtained from Chemstations
(2001) and Daubert, Danner, Sibul, and Stebbins (1996).
The experimental data needed to obtain the interaction
parameters in the NRTL model were taken from the
2
literature (Chu, Getty, Brennecke, & Paul, 1950;
Gmehling et al., 1982). Table 4 gives some details on the
experimental data used in the study. In this table, DT
(°C) is the temperature range in the data set, Dx1 is the
liquid mole fraction range for the congener in the data
set, and Dy1 is the vapor mole fraction range for the
congener in the data set.
5. Results and discussion
Table 5 shows the binary parameters in the NRTL
model for the 18 binary mixtures studied in this work.
These parameters were obtained by regression analysis
cular chosen property. Then, the differences between
predicted and experimental values of that property,
according to a specified objective function, are mini-
mized. The value of the parameters that minimize the
objective function corresponds to the optimum values of
those parameters, regarded as empirical since experi-
mental data were used to determine their values. The
objective function is arbitrarily but conveniently defined
and several criteria have been presented in the literature.
In our study, the difference between the experimental
and calculated values of the congener mole fraction in
the liquid phase is included in the objective function W ,
as follows:
X
N
W ¼ xexp
i
xcalc
i ð16Þ
l¼1
Tables 6 and 7 shows the results for the saturation
temperature and the vapor mole fraction of the cong-
eners for the 18 mixtures studied, nine ethanol + con-
gener mixtures (Table 6) and nine water + congener
mixtures (Table 7). In these tables, the relative devia-
tions and absolute deviations for the saturation tem-
perature, DT % and jDT %j, and for the congener
concentration in the vapor phase, Dy1 % and jDy1 %j, for
the three models studied NRTL, UNIFAC and PSRK,
are given. The relative and absolute deviations for a set
of N data points, are defined as:
X
DT % ¼ ð100=N Þ ½ðTcalc
Texp Þ=Texp

of experimental data published in the literature using a X 

jDT %j ¼ ð100=N Þ ½ðTcalc
Texp Þ=Texp

commercial software (Chemstations, 2001). In the X ð17Þ
regression analysis employed to determine the empirical Dy1 % ¼ ð100=N Þ ½ðycalc
yexp Þ=yexp
i
parameters, the thermodynamic models NRTL and the X 
jDy1 %j ¼ ð100=N Þ ½ðycalc
yexp Þ=yexp
i 
SRK equation of state is applied to calculate a parti-
582 C.A. Fa
undez, J.O. Valderrama / Journal of Food Engineering 65 (2004) 577–583

Table 5
Binary parameters in the NRTL model for the 18 mixtures studieda
System DT (° C) b12 b21 d
Acetic acid (1) + water (2) 100–119 )47.032 183.82 0.39688
Acetaldehyde (1) + water (2) 33–100 116.42 528.32 0.28877
Ethyl acetate (1) + water (2) 71–76 727.01 1881.15 0.39304
Furfural (1) + water (2) 98–135 120.92 1256.1 0.32651
Methanol (1) + water (2) 65–100 )208.12 573.90 0.29426
3-Methylbutanol (1) + water (2) 95–130 )469.02 2129.4 0.20335
2-Methyl-1-propanol (1) + water (2) 90–107 192.68 1179.2 0.39935
1-Pentanol (1) + water (2) 96–120 18.227 1668.6 0.31332
1-Propanol (1) + water (2) 88–95 )12.175 1094.4 0.30282

Acetic acid (1) + ethanol (2) 77–116 )186.30 12.901 0.34205

Acetaldehyde (1) + ethanol (2) 20–78 )145.60 2500 0.40000
Ethyl acetate (1) + ethanol (2) 72–77 268.74 51.620 0.22242
Furfural (1) + ethanol (2) 78–134 646.03 )34.853 0.33320
Methanol (1) + ethanol (2) 65–78 )164.07 179.87 0.29880
3-Methylbutanol (1) + ethanol (2) 78–131 86.300 )47.489 0.29222
2-Methyl-1-propanol (1) + ethanol (2) 79–106 461.17 )331.28 0.29953
1-Pentanol (1) + ethanol (2) 80–132 197.25 )146.45 0.24174
1-Propanol (1) + ethanol (2) 78–97 30.384 )18.127 0.29902
a
Experimental data are from Gmehling et al. (1982) except for ethanol + acetaldehyde, which are from Chu et al. (1950).

Table 6
Percent deviations for the temperature and congener vapor phase mole fractions for systems congener (1) + water (2)
System NRTL UNIFAC PSRK
%DT j%DT j %Dy1 j%Dy1 j %DT j%DT j %Dy1 j%Dy1 j %DT j%DT j %Dy1 j%Dy1 j
Acetic acid (1) + water (2) )0.4 0.5 )1.3 2.8 2.0 2.0 )12.7 21.3 0.8 1.1 )5.3 15.4
Acetaldehyde (1) + water (2) )2.1 2.3 4.5 7.9 )0.6 1.5 1.5 5.4 )1.3 2.3 2.2 5.9
Ethyl acetate (1) + water (2) 1.5 1.5 )0.7 1.7 3.2 3.3 )2.3 4.0 2.1 2.2 )5.3 5.3
Furfural (1) + water (2) )0.2 0.9 0.5 8.8 )0.4 0.9 )0.1 8.8 )1.8 1.8 10.8 26.9
Methanol (1) + water (2) 0.2 0.4 0.4 2.4 0.3 0.3 2.0 3.7 0.0 0.2 5.0 6.3
3-Methylbutanol (1) + water (2) 0.0 1.1 )2.2 2.8 3.1 3.1 )21.0 35.4 2.3 2.8 )8.8 16.4
2-Methyl-1-propanol (1) + water (2) 0.2 0.7 4.1 6.0 )0.1 0.6 4.0 5.8 )1.6 1.6 8.4 12.7
1-Pentanol (1) + water (2) 0.2 0.3 )0.4 7.5 2.0 2.0 3.2 4.9 0.2 1.2 )6.0 14.5
1-Propanol (1) + water (2) 0.1 0.6 3.9 5.9 )0.9 0.9 7.1 8.6 )1.7 1.7 10.7 14.3

Table 7
Percent deviations for the temperature and congener vapor phase mole fractions for systems congener (1) + ethanol (2)
System NRTL UNIFAC PSRK
%DT j%DT j %Dy1 j%Dy1 j %DT j%DT j %Dy1 j%Dy1 j %DT j%DT j %Dy1 j%Dy1 j
Acetic acid (1) + ethanol (2) )2.0 3.0 1.0 6.6 0.9 1.7 )3.0 20.9 0.6 1.6 3.8 19.3
Acetaldehyde (1) + ethanol (2) 3.6 3.6 6.4 6.4 7.3 7.3 1.3 2.4 4.4 7.1 2.5 2.9
Ethyl acetate (1) + ethanol (2) 0.1 0.5 )2.5 3.1 1.2 1.3 )4.6 5.8 1.3 1.4 )2.5 3.8
Furfural (1) + ethanol (2) 0.1 1.4 0.2 20.8 )1.4 1.5 )5.7 23.1 )1.3 1.5 )3.0 26.1
Methanol (1) + ethanol (2) )0.7 0.7 )1.5 3.5 0.8 1.1 )5.4 6.8 1.0 1.3 )3.1 5.4
3-Methylbutanol (1) + ethanol (2) 0.1 0.2 )2.2 2.3 0.0 0.1 )2.1 2.2 0.9 0.9 )7.6 7.6
2-Methyl-1-propanol (1) + ethanol (2) 0.4 0.4 )0.6 1.4 0.7 0.7 )1.0 1.8 1.0 1.0 )0.2 1.7
1-Pentanol (1) + ethanol (2) )0.7 0.7 7.1 8.5 )0.6 0.6 6.8 8.4 )0.2 0.9 2.2 4.9
1-Propanol (1) + ethanol (2) 0.1 0.4 6.2 7.5 )0.1 0.4 6.2 7.6 0.5 0.5 6.4 7.6

As observed in Table 6, for the systems congener
(1) + water (2), the predictive models PSRK and UNI-
FAC give results similar to those given by the correlat-
ing NRTL model for the saturation temperature. The
UNIFAC model gives results with deviations of the
order of 9% except for acetic acid (1) + water (2) and for
3-methylbutanol (1) + water (2). For the systems con-
gener (1) + ethanol (2) shown in Table 7, the three
models give similar results for the saturation tempera-
ture. The UNIFAC and the PSRK models give results
with deviations lower than 9% except for acetic acid
(1) + ethanol (2) and furfural (1) + ethanol (2). The only
 C.A. Fa
undez, J.O. Valderrama / Journal of Food Engineering 65 (2004) 577–583
explanation we have for this is that experimental data
could be inaccurate due to the instability of furfural and
acetic acid in water and in ethanol (Internet-1, 2003a,
2003b).
In the overall, the NRTL model give the best results,
so predictive models should be used only if experimental
data are impossible to obtain. If models are needed for
simulation and design, we recommend to obtain binary
VLE data from the laboratory.
6. Conclusions
Vapor–liquid equilibrium in binary water + congener
and ethanol + congener systems has been modeled using
a correlating model (NRTL) and two predictive models
UNIFAC and PSRK. Based on the results, the follow-
ing conclusions can be drawn: (i) the NRTL model
represents a good tool for correlating vapor–liquid
equilibrium data of water + congener and etha-
nol + congener mixtures; (ii) the predictive UNIFAC
model can be used for cases in which no-experimental
data is available, although care must be taken and at
least one data could be useful for checking the results;
(iii) the study done here represents a necessary and
useful study to extend and apply the models to multi-
component mixtures.
Acknowledgements
The authors thank the support of the Directorate of
Research of the University of La Serena-Chile (grants
220-2-05 and 220-2-21) and of the National Council for
Scientific and Technological Research (CONICYT),
through the research grant FONDECYT 3020020.
References
Alonso, P. (1985). Study on the distillation of the main components of
wines employed to obtain Pisco. B.Sc. Thesis, School of Engineer-
ing, Catholic University of Chile, Santiago-Chile.
Baquerizo, G. (1999). Wine distillation to produce Pisco. M.Sc. Thesis,
School of Engineering, Catholic University of Chile, Santiago-
Chile.
Billet, R. (1979). Distillation engineering. New York, NY: Chemical
Publishing Corporation.
Chemstations (2001). CC-Batch; en ChemCAD III, Process Flowsheet
Simulator, Operating Manual. Chemstations Inc., Houston, USA.
Chu, J. C., Getty, R. J., Brennecke, L. F., & Paul, R. (1950).
Distillation equilibrium data. New York, USA: Reinhold Publishing
Corporation.
Daubert, T. E., Danner, R. P., Sibul, H. M., & Stebbins, C. C. (1996).
Physical and thermo-dynamic properties of pure chemicals. Data
compilation. London, UK: Taylor & Francis.
Fredenslund, A., Gmehling, J., & Rasmussen, P. (1977). Vapor–liquid
equilibria using UNIFAC. Amsterdam, Holland: Elsevier.
583
Gmehling, J., Onken, U., & Arlt, W. (1982). Vapor–liquid equilibrium
data collection. Alemania: Dechema.
Hikari, A., & Kubo, R. (1975). Behavior of various impurities in
simple distillation of aqueous solutions of ethanol. Journal of
Chemical Engineering of Japan, 8(4), 294–299.
Holderbaun, T., & Gmehling, J. (1991). PRSK: A group contribution
equation of state based on UNIFAC. Fluid Phase Equilibria, 70,
251–265.
Huron, M. I., & Vidal, J. (1978). New mixing rules in simple equations
of state for representing vapor–liquid equilibria of strongly non-
ideal solutions. Fluid Phase Equilibria, 3, 255–271.
Internet, http://www.indunor.com/furfura5.htm (2003a).
Internet, http://www.winza.cl/winza101.htm (2003b).
Lora, J., Iborra, M. I., P
erez, R., & Carbonell, I. (1992). Simulation of
distillation processes for the concentration of wine aromas. Revista
Espa~nola de Ciencia y Tecnolog
ıa de Alimentos, 32(6), 621–633.
Mollerup, J. (1981). A note on the excess Gibbs energy models,
equations of state and local composition concept. Fluid Phase
Equilibria, 7, 121–138.
Orbey, H., & Sandler, S. I. (1998). Modeling vapor–liquid equilibria.
Cubic equations of state and their mixing rules. USA: Cambridge
University Press.
Prausnitz, J. M., Lichtenthaler, R. N., & Gomes de Azevedo, E.
(1999). Molecular thermodynamics of fluid-phase equilibria. New
Jersey: Prentice Hall International Series.
Soave, G. (1972). Equilibrium constants from a modified Redlich–
Kwong equation of state. Chemical Engineering Science, 27, 1197–
1203.
Valderrama, J. O., Huerta, R., & Alarc
on, R. (2002). Base de datos de
propiedades de sustancias para procesos de destilaci
on de vinos.
Information Technology, 13(4), 157–166.
Valderrama, J. O., Pizarro, C., & Rojas, R. (2001). Vapor–liquid
equilibrium in complex mixtures for the simulation of must and
wine distillation. Alimentaria, 39(334), 151–156.
Valderrama, J. O., Rojas, A. S., Garc
es, I., & Rosell
o, A. (1998).
Application of commercial process simulators to industrial pro-
cesses. Revista Investigaci
on y Desarrollo ULS (3), 1–10.
Valderrama, J. O., Rojas, R., & Pizarro, C. (2000). Comparative study
of thermodynamic models to describe complex mixtures present in
wine distillation. Information Technology, 11(6), 189–192.
Walas, S. M. (1985). Phase equilibria in chemical engineering. Store-
ham, USA: Butterworth.
Dr. Jos
e O. Valderrama is a Full Professor of the Mechanical Engi-
neering Department of the University of La Serena––Chile, where he
has been for the last 14 years. He is also the Director of the Center for
Technological Information (CIT). The Center edits and publishes the
international refereed journal ‘‘Informaci
on Tecnol
ogica’’ (ISSN:
0716-8756). The CIT, founded by Dr. Valderrama is a private insti-
tution dedicated to the diffusion and dissemination of scientific and
technological information in Iberoamerica. Besides his interest on
editorial matters, his main research interests are on the modeling of
properties of pure substances and mixtures, phase equilibria, and
simulation of physico-chemical processes. On all these subjects he has
publications, congress presentations and invited conferences world-
wide.
Dr. Claudio A. Fa
undez is a Professor of the Physics Department of the
University of La Serena––Chile, where he studied and has been
teaching for the last 10 years. He is a postdoctoral researcher of
Fondecyt, the research program of the National Council for Scientific
and Technological Research of the Chilean Government. Being a
physicist and a teacher by formation, he became interested, since his
undergraduate studies, on physics of fluids, subject on which he has
done research, publications and congress presentations. Presently, his
main activity is centered on a project on modeling physical and phase
equilibrium properties of mixtures involved in wine distillation.
However he is still interested in molecular aspects of fluid behavior and
any application of physics to engineering problems.