ELSEVIER Fluid Phase Equilibria 99 (1994) 63-74

Prediction of vapor -liquid equilibria of aqueous systems in

the subcritical range by using the NRTL equation
Alessandro Vetere
Snamprogetti Research Laboratories, San Donato Mlanese, Italy
Received 23 September 1993; accepted in final form 17 March 1994

Abstract

A method is proposed to predict vapor-liquid equilibria (VLE) of aqueous mixtures by using the NRTL equation.
The algorithm closely follows that reported in previous works dealing with predictions of VLE of non-aqueous
mixtures. The main assumptions are:
- simple rules can be derived for families of mixtures characterized by similar chemico-physical properties;
- within each family, the energy interaction parameters of the NRTL equation can be calculated by knowing only
the well-known properties of the pure compounds, such as the Hildebrand solubility parameters.
The main subdivision adopted in this work is between soluble mixtures and mixtures with a solubility gap. Soluble
mixtures are subdivided, in turn, into four sub-classes:
- H,O/C, -Cd alcohols;
- H,O/amines;
- H,O/heterocyclic compounds or amides;
- H,O/other polar compounds.
The method is applied to several binary mixtures which represent difficult cases, as strongly associated or highly
asymmetric systems. For some systems a comparison is made with the group contribution method. New relationships
which improve the prediction of VLE of non-aqueous systems are also reported.

Keywords: Theory; Excess functions; Prediction; Vapor-liquid equilibria; Water; Polar compounds

1. Introduction

Modern equations for calculating the non-ideality of a mixture in VLE based on the
non-random distribution of molecules in a fluid, such as the Wilson and NRTL equations, are
useful for two main reasons. Firstly, they enable, in a simple way, the prediction of multicompo-
nent VLE from knowledge of binary systems only. Secondly, the binary adjusting parameters
required to describe the deviations from Raoults law according to relations of the type
gE =_@I, x2, (52, 32,)

037%3812/94/$07.00 0 1994 - Elsevier Science B.V. All rights reserved

SSDZO378-3812(94)02507-W
64 A. Vetere / Fluid Phase Equilibria 90 (1994) 63-74

are linked to parameters which arise from the interaction forces acting between molecules. To
illustrate, for the NRTL equation

G2 = exp( --a12ri2) G1 = exp( - QQ~) (2)

where
2*2 = 812 - g22 z2, =g21 -g11

RT RT
The physical significance of the energy parameters in Eq. (3) was elucidated soon after the work
of Wilson by Prausnitz et al. (1967). Further, an important step toward prediction was made by
Tassios ( 1971) and Bruin ( 1971) who showed how g 11 and g,, can be calculated from the
vaporization enthalpy of pure components. As a result, the remaining unknown, g12, can be
calculated from one experimental datum only. Following similar concepts, it was also demon-
strated that the cross-interaction term can be evaluated with fair accuracy in the context of the
Heil and Prausnitz equation for several polar systems if the Hildebrand solubility parameters of
the mixture components are known (Vetere, 1977). Furtheremore, a recent work indicates that
the NRTL equation is inherently predictive, since both T12 and z21 can be calculated through
generalized correlations which are applicable to families of mixtures similar for their chemico-
physical properties (Vetere, 1993). If the literature data for non-aqueous systems are grouped in
five families, a relation of the type

(812 - &,I + (g12 -g22) = A + B(4 - ~2)abs (4)

holds for each of them. By knowing A and B from the correlation of few experimental data,
the relation (4) gives the sum RT(r12 + z2,). The only other datum required to calculate the VLE
of a binary mixture is the asymmetry of the system, expressed as r12/221. This ratio does
not significantly affect the VLE calculations and can be evaluated with simple rules for each
family of mixtures (Vetere, 1993). As a result, the non-ideality of non-aqueous systems can
be calculated from the Hildebrand solubility parameters of the pure compounds as the only
input data. The reliability of the method, tested on the basis of nearly 90 literature systems, was
fair. However, the important case of aqueous systems was disregarded as in the MOSCED
method.
In this work an attempt is made to extend the described procedures to mixtures of polar
compounds with water.

2. Methods

The most obvious classification of the water/polar components mixtures in the literature is
into soluble mixtures and mixtures which exhibit a miscibility gap. However, soluble mixtures
encompass an exceeding variety of molecular structures which are responsible for the high
asymmetry (expressed as y T/y?) often found in systems belonging to this family. Since a single
procedure cannot take into account all types of non-ideality arising from relevant differences in
chemico-physical properties of components, the soluble mixtures have been subdivided into four
classes:
 A. Vetere / Fluid Phase Equilibria 90 (1994) 63-74 65

- mixtures with Cl -C4 alcohols;

- mixtures with amines;
- mixtures with heterocyclic compounds and amides;
- mixtures with other polar compounds.

each of them deserving a different approach.

Two predictive methods are proposed in this work, both linked to previous papers of the
author.

2.1. Method A

This relies on a procedure developed by Vetere ( 1977). Basically, one assumes that the pure
component interaction parameters g 11 and g,, which appear in Eq. (3) can be calculated
according to the relation proposed by Tassios ( 1971) and Bruin and Prausnitz ( 1971) :

g, = -(AH, - RT) (5)

The remaining parameter required to define the non-ideality of the binary mixtures, g12, can be
evaluated according to the following criteria. Within the same family of mixtures g,, deviates
from the geometric mean of the pure component energy parameters by a quantity AM which can
be related to the pure component properties.

g,, = AM - (g,,g,,).5 (6)

and

AM = A, + B,@: - &+,, (7)

Eq. (7) expresses the concept that the greater the differences between the pure component
properties, evaluated through the Hildebrand solubility parameter, the greater the deviations
from the geometric mean rule.
Another parameter useful to express the polar character of a compound can be defined as

W2 - 263
wk=- M

Eq. (8) gives a value of zero for n-paraffins. The highest value of W, pertains to water (Vetere,
1989).
Although ~5~and d2 can be calculated easily in most cases, for some not weII studied
compounds it is more advisable to apply Eq. (8). W, was used to characterize amines. It was
found that for amine/water systems AM is related to W, of amine as follows:

AM=A2+BzWp (9)
The empirical constants of Eqs. (7) and (9) can be calculated by regressing the experimental data
of at least three systems pertaining to the considered family of mixtures.
66 A. Vetere / Fluid Phase Equilibria 90 (1994) 63-74

2.2. Method B

This nearly coincides with the method proposed by Vetere (1993). The working equation is
(g,, -_g,,) + (g,, -gz) = A3 + B3@: - S&s (10)

which represents only a slight modification of Eq. (4). Water is the component always present
in all the systems considered in this work. Accordingly, if the temperature range investigated is
not too large, 6i is nearly constant and can be incorporated in B3. By slight rearrangement, Eq.
( 10) becomes
t&2 - g, 1) + (&2 - g22) = A4 + B462 (11)

Another alternative to Eq. (lo), exploiting W, of the non-aqueous component as a correlating

parameter, is
(g,, -g,,) + (812 - gz) = A5 + B5 wp (12)

which proved useful for water/alcohol mixtures. The asymmetry of the system, 2i2/r2i, is the
second datum required to calculate gE. This parameter is constant within each family of
mixtures and can be determined empirically by correlating some literature systems.

3. Methodology

The methodology followed to predict the NRTL constants is the same as in the previous
paper (Vetere, 1993): (i) three well-studied isothermal systems are chosen for each family of
compounds; (ii) the experimental data of these systems are regressed to derive the optimized
values of r12 and z21; (iii) according to these values, a rule is derived for the asymmetry of this
family of compounds; (iv) a relation is found which links the sum Zi RTz, with the difference
(Pi - Pi), where P stands for a property of the pure components, such as the Hildebrand
solubility parameter or similar ones quoted in the previous sections. The last two rules are
assumed as valid for all the systems pertaining to the same family.

4. Results

The calculation of the pure component properties, such as the Hildebrand solubility parame-
ters and the vapor pressures, was performed according to the procedures reported in a previous
work (Vetere, 1993). The VLE data were taken from two monographs of the DECHEMA series
(Gmehling and Onken, 1977; Gmehling et al., 1981).
Table 1 reports the equations used to calculate the three empirical constants of the NRTL
equation. To have a complete layout of the methods, Table 1 reports also the equations which
enable the VLE calculations of non-aqueous systems. In some cases, these latter represent an
improvement over the equations previously proposed (Vetere, 1993). Major modifications are:
- a less rough rule to calculate the asymmetry of polar/non-polar systems which exhibit a
miscibility gap;
 A. Vetere 1 Fluid Phase Equilibria 90 (1994) 63-74 67

Table 1
Equations for calculating the parameters of the NRTL equation

Aqueous mixtures a Non-aqueous mixtures

Mixtures with a c+r = 0.42 Mixtures with a a,, = 0.42; z = 0.72 + 0.0003 (g,, - g22)ab.
solubility gap r&1 = 2 miscibility gap
(index 1: the more polar compound)
RT(t,, + r2,) = 5060 - 2806,
RT(t,, + Q,)= -457 + 476.2(6, - SZ)sba
H,O/amines CZ,*= 0.1; A, = 630 - 5.375W,, Polar/polar mixtures tx,r = 0.2 + 0.063(6, -c&_

z = 2 (index 1: the more polar compound)

H,O/heterocyclics c+ = 0.0; AM = - 1414 + 3.93(6; - S:),, RT(r,,

+ r2,)
= 406 + 2.25(W,, - W,,),,
and amides

H,O/alcohols ~(,r = 0.3; g,, -g,, = - 100 cal mol- Nearly ideal mixtures a,* = 0.3; T,* = r*,
RT(r,, + r2,) = 2217 - 4.783W,, RT(r,, + To,) = 140 + 1.75(W,, - W,,).,
H,O/other polar = 200cal mol-
a,* = 0.3; g,, -g,, Mixtures of homologs CQ~= 0.3; T,* = r2,
compounds RT(r,,
+ T,*)
= 122 + 2.2(6: - a:),,, RT(r,, + r.2,) = -220 + 2.75(W,, - FVr&

a Index 1: water; index 2: other compound.

an equation to calculate aU for polar/polar mixtures which avoids the discontinuities of the
previous method;
a bifurcation for nearly ideal mixtures which are now subdivided into mixtures of ho-
mologous compounds (as, for example, the acetone/diethyl ketone system) and mixtures of
compounds having different functional groups.
Tables 2-4 report the results obtained by applying the methods proposed in this work along
with the empirical constants of the NRTL equation and the correlating parameter used for each
system. The reliability of the methods applied to predict the VLE of aqueous mixtures is here
briefly discussed.

4.1. Aqueous systems with a miscibility gap

This family of mixtures has been predicted by using Eq. ( 11) of Method B for calculating the
sum RT(rr, + rzl). The second datum required to apply the method is asymmetry, expressed by
r12/r2, = 2, where the index 1 represents water. The results obtained are only fair, for both
pressures and vapor composition. As a rule, the more difficult cases are represented by highly
asymmetric systems.

4.2. Miscible aqueous systems

This family encompasses four subclasses of mixtures, each of them deserving an ad hoc
prediction method as described below.

(a) Amine/water mixtures

Both amines and water are strongly associated compounds which in mixtures give rise to
significant deviation from Raoults law. The best results were obtained by using Eq. (9) of
68 A. Vetere 1 Fluid Phase Equilibria 90 (1994) 63-74

Table 2
Prediction of vapor-liquid equilibrium data for aqueous mixtures with a miscibility gap (index 1 = water)

Second component %j 82 wr*2 +tzl) g,z-g11 g21- g22 AT 1OOAY

of the mixture (Cal mol-) (cal mol-i) $mHg) (K)
-
Methyl acetate 0.42 9.1 2500 833 1667 27.3 - 2.50
Ethyl acetate (at WC) 8.5 2670 890 1780 3.3 - 0.80
Ethyl acetate (at 40C) 8.7 2600 867 1733 2.4 - 0.90
Ethyl acetate (at 760 mmHg) 7.9 2830 943 1887 - 0.89 8.90
Butyl acetate 8.3 2730 910 1820 13.1 _ 10.70
Isopentyl acetate 7.4 2970 990 1980 - 1.45 0.90
2-Butanone (at 73C) 8.6 2650 883 1767 6.9 - 0.90
2-Butanone (at 500 mmHg) 8.3 2730 910 1820 _ 0.48 1.40
Cyclopentanone (at 760 mmHg) 9.1 2500 833 1667 - 0.94 1.80
Cyclopentanone (at 9OC) 10.1 2230 743 1427 62.5 - 5.60
Acetophenone 9.7 2330 777 1553 _ 1.50 2.40
Methyl cyclopropyl ketone 9.0 2530 843 1687 - 1.30 5.60
Mesityl oxide 8.5 2670 890 1780 - 0.96 2.30
Acrylonitrile (at 40C) 11.5 1830 610 1220 54.6 - 19.60
Acrylonitrile (at 760 mmHg) 10.4 2140 713 1427 _ 1.95 9.60
Propiononitrile 10.1 2220 740 1480 - 1.21 6.40
/I-Methoxypropiononitrile 11.0 1960 653 1307 5.2 - 0.80
Diisopropyl ether 6.3 3220 1073 2147 _ 1.63 7.30
Isobutyraldehyde 8.0 2820 940 1880 - 1.20 4.50
Furfural (at 37.8C) 10.2 2200 733 1467 7.5 - 3.10
Furfural (at 50C) 11.6 1800 600 1200 3.4 2.50
Tetrahydrofuran 8.9 2550 850 1700 - 0.63 1.20
1-Butanol 10.3 2170 723 1447 - 1.45 4.91
2-Butanol 10.3 2170 733 1447 21.0 5.70
Phenol 12.1 1660 553 1107 5.7 - 0.60
Aniline 11.1 1950 650 1300 67.0 2.80
2-MethyUethylpyridine 8.7 2600 867 1733 - 1.63 4.92
Isopropylbenzene 7.7 2900 967 1933 _ 1.34 5.50
Triethylamine 6.8 3080 1027 2053 - 0.98 4.15
10.3 2170 723 1447 8.81 - 4.82
R 20.81 1.40 4.44

Method A and aI2 = 0.1. The calculated pressure and temperatures are acceptable, but the vapor
compositions are predicted with a considerable error.

(b) Heterocyclics or amides/water mixtures

These represent, perhaps, the most difficult case, since this family contains some perversely
asymmetric systems such as water/2_methylpyridine. The results reported in Table 3 have been
obtained according to Eq. (7) of Method A. Surprisingly, the best value of the randomness
parameter is ct12= 0, so that in this case the NRTL equation reduces to the two-suffix Margules
equation
gE/RT = Ax,x2 (13)
 A. Vetere 1 Fluid Phase Equilibria 90 (1994) 63-74 69

Table 3
Prediction of the vapor-liquid equilibrium data for miscible aqueous systems according to Method A (index
1 = water)

Second component NRTL constants Correlating parameters Results

of the mixture
uij g12-_gll g12 - g22 WP 6% - G)at,s AP AT 1OOAY
(Cal mol-) (cal mol-) (cal cm3) (mmHg) (R)

Ethylamine 0.1 2802 -2100 117.6 - 1.4 1.10

Propylamine 2333 - 1515 84.7 - 0.9 1.60
Butylamine 1704 -933 61.0 - 0.8 9.60
Isobutylamine 2154 -1013 40.1 - - 1.7 8.60
Diethylamine 2618 - 1210 28.4 - 32.1 - 3.90
Methyldiethylamine 2521 -878 -4.3 - 41.4 - 5.50
Triethylamine 2297 -904 13.6 - 2.4 - 5.80
Ethylbutylamine 804 341 11.1 - 4.5 - 5.70
Dimethylisopropylamine 2572 -933- -3.0 - 5.3 _ 6.50
Dimethylamine 3382 - 2069 96.0 - 9.8 - 0.50
n;i 15.9 1.20 4.88
1,CDioxane 0.0 1422.9 -518.4 - 462.1 6.9 _ 3.3
1,3,5Trioxane 686 - 376.6 _ 395 180 - 12.0
Dioxolane 1867.6 - 852.7 - 462 37.6 - 5.4
Pyridine 1328.7 - 254.6 - 488.1 6.3 - 6.9
2-Methylpyridine 683.2 - 236.8 - 413.5 168.6 - 10.8
2,6_Dimethylpyridine 354.5 706.1 - 494 3.6 - 4.2
N-Methylformamide -503.8 584.8 _ 366 12.2 - 0.6
N,N-dimethylformamide -275.1 264.3 - 357 8.2 0.7
N-Methylpyrrolidone - 1000.5 1198.0 - 370 15.4 - 0.3
Morpholine 350.3 - 120.2 _ 388.2 30.0 - 1.5
Quinoline - 1025.0 1782.6 - 433.5 71.6 - 0.4
Sulfolane -2849.1 3720.5 - 366.5 4.4 0.0
n;i 45.4 3.84

The strong asymmetry of many systems belonging to this family reflects, also on the reliability
of the method. The deviations from experimental values are often high for both pressures and
vapor compositions. However, the proposed procedure appears less sensitive to the proximity
effect than the group contribution method (see, for example, the data reported by Wu
and Sandler, 1989, for the systems l,Cdioxane/water and 1,3-dioxolane/water predicted with
UNIFAC).

(c) Alcohol/water mixtures

The prediction method is based on the application of F,q. (12). Further, for all the systems of
this family it can be set that g12 -g,, = - 100 cal mol- and al2 = 0.3 (where index 1 pertains to
water). The reliability of the method is good, taking into account the performances of the
group-contribution method for this class of mixtures.
70 A. Vetere 1 Fluid Phase Equilibria 90 (1994) 63-74

Table 4
Prediction of the vapor-liquid equilibrium data for miscible aqueous systems according to Method B (index
1 = water)

Second component NRTL parameters Correlating parameters Results

of the mixture
% Wr,*+r*,) g,,-g,, 812-&z (G-G),,, W, AP AT 1OOAY
(cal mol-I) (Cal mol-) (cal mol-) (mmHg) (R)

Methanol 0.3 150 -100 250 - 434.0 27.7 - 4.34

Ethanol 1020 -100 1120 - 256.0 6.0 - 1.68
1-Proponal (at 79.W) 1400 -100 1500 - 172.6 38.5 _ 2.94
I-Propanol (at 758 mmHg) 1400 -100 1500 - 172.6 - 1.78 5.18
2Propanol(65C) 1540 -100 1640 - 142.9 11.2 - 1.41
2-Propanol (at 760 mmHg) 1540 -100 1640 _ 142.9 - 0.35 1.14
1-Butanol 1620 -100 1720 - 124.6 24.1 - 5.99
tert-Butanol 1860 -100 1960 _ 66.2 - 0.88 1.46
Ally1 alcohol 1320 -100 1420 - 187.5 - 1.39 2.64
2-Methyl-3-butyl-2-01 1900 -100 2000 _ 59.4 - 0.68 2.22
1,2-Propanediol 1080 -100 1180 - 237.5 - 1.88 1.22
2,3-Butanediol 1570 -100 1670 - 134.5 - 1.68 5.50
Diethylene glycol 1360 -100 1460 - 178.4 - 1.68 0.18
Tetrahydrofurfuryl alcohol 1750 -100 1850 _ 96.3 - 0.68 1.09

h? 20.9 1.22 2.38

Acetaldehyde 0.3 1410 200 1210 489.3 - 24.3 _ 0.3

Acetone 1400 200 1200 484.0 - 23.2 - 1.6
Cyclohexanone 1290 200 1090 428.0 - 15.0 - 5.0
2-Chloroethanol 1150 200 950 371.5 - 20.4 - 2.4
Acetonitrile 1210 200 1010 403.0 - 103.4 _ 4.2
Propionic acid 1000 200 800 305.0 - 3.1 - 2.7
Butyric acid 1250 200 1050 418.0 _ 7.8 - 1.3
Acrylic acid 1170 200 970 383.8 - 2.8 - 1.2
Phenol 1180 200 980 389.7 - 6.0 - 0.5
Diacetone alcohol 1350 200 1150 465.0 _ 2.8 _ 0.9
3-Hydroxy-2-butanone 1180 200 980 385.5 _ 30.7 _ 1.2
2-Butoxyethanol (at 5C) 1500 200 1300 528.0 - 0.5 - 0.8
2-Butoxyethanol (at 85C) 1280 200 1080 434.0 - 22.9 - 0.9

w 20.2 1.8

(d) Water lo ther polar compounds

This class comprises mixtures of water with all the other polar compounds not considered in
previous sections, in particular molecules bearing two functional groups.
The system to be resolved is represented by Eq. (10) of Method B and the relation
g,, - g, 1 = 200 cal mol-, where the index 1 stands for water. Significant errors are found for the
mixture of water with the strongly associated compound acetonitrile.
 A. Vetere / Fluid Phase Equilibria 90 (1994) 63-74 71

5. Comparison with the group-contribution method (UNIFAC)

The binary systems studied in this work represent a relatively narrow slice of the wide corpus
of literature data. Therefore, in order to avoid unsound conclusions on the reliability of the
proposed method it was deemed useful to restrict the choice of the VLE data to difficult
systems only, showing strong deviations from ideal behavior or abnormally high asymmetries.
For this reason, it is not surprising that the results obtained are often only fair. However, it
would be incorrect to assume that some significant errors found in predicting the VLE data of
the systems listed in Tables 2-4 prove the unreliability of the method. As it results from the data
of Tables 5 and 6, the predictions based on the NRTL equation compare favorably with those

Table 5
Comparison with UNIFAC in predicting the VLE of of some binary systems

System T(K) This work UNIFAC =

1OOAY 1OOAY
:LHg)

Acetone/water a 298 23.2 - 1.6 18.0 f _ 4.8 f

Water/l ,Cdioxane a 308 6.9 - 3.3 8.4 f - 9.6 f
Water/diethylamine a 309 32.1 - 3.9 28.0 f - 2.8
Ethanol/water a 328 6.0 - 1.78 18.0 f - 3.8 f
Water/3-hydroxy-2-butanone a - - 0.44 2.08 - 2.4 2.9
2-butanollwater a _ - 1.90 5.70 _ 2.8 5.9
1,3-Dioxolane/water b 318 3.9 - 5.90 22.1 _ 15.9
1,3$Trioxane/water 373 179.9 - 12.8 136.1 - 10.5
2-Methylpyridinelwater d 373 168.6 - 10.8 107.7 _ 7.4

a Fredenslund et al. (1977). b Wu and Sandler (1991). Gmehling et al. (1981). d Gmehling and Onken (1977).
Hansen et al. (1991). f Optimized data.

Table 6
Comparison with UNIFAC (GCM) in predicting infinite dilution activity coefficients a

System T(K) YP YZm

Exp. NRTL GCM Exp. NRTL GCM

1,3-Dioxolane/ 318.2 22.02 4.98 2.28 10.39 4.98 0.94

water 343.2 17.62 4.43 2.57 7.99 4.43 1.oo
1,CDioxane/ 308.2 6.64 4.38 5.17 6.96 4.38 1.06
water 323.2 7.05 4.09 5.59 6.41 4.09 1.11
342.2 7.27 3.77 5.99 4.87 3.77 1.16
Tetrahydrofuranl 298.2 28.9 38.7 33.5 17.80 9.91 2.63
water 323.2 26.2 30.2 33.5 9.38 8.98 2.82
343.2 30.8 25.9 33.1 7.80 8.34 2.95

a Wu and Sandler (1989).

72 A. Vetere 1 Fluid Phase Equilibria 90 (1994) 63- 74

of the UNIFAC method for alcohols and ketones, and are only slightly worse for water in
mixtures with some polar nitrogen compounds, It must be stressed that the VLE literature data
for the water/2_methylpyridine system, for which both methods give very inaccurate predictions,
appear questionable, since the limiting activity coefficients reported for this mixture in the
DECHEMA monograph (Gmehling and Onken, 1977) are unrealistic, being higher than those
pertaining to the water/2,6-dimethylpyridine system. Further, on the basis of the Table 6 data
the new method seems less sensitive to promixity effects than UNIFAC.

6. Conclusions

The multiplicity of procedures described in this work is a clear sign that the panacea of a
universal method which can face whatever type of non-ideality in VLE is not realistic
at present. Partial failures are the common fate of all predictive methods. The group contribu-
tion method (GCM), which still remains the preferred way to calculate VLE of molecules
made of well-studied functional groups, proves inadequate in dealing with complex molecular
structures, typically molecules bearing more than one strong polar functional group. The
adopted remedy of assuming as a group the molecule as a whole is a clear exception to the
assumptions of this method. As a result, the advantage of using the GCM is partially lost. The
main drawback of using the NRTL equation as a predictive tool lies in the subdivision of the
binary systems into families not always well defined, since the lines separating these families
from each other are in some instances unclear. However, the method proposed in this work is
simple and more general in principle than literature methods. For these reasons, it can be used
to solve at least some engineering problems in fluid phase separations when the GCM appears
unapplicable.

List of symbols

A constant in Eq. (13)

Al constant in Eq. (7) (cal mol-)
A2 constant in Eq. (9) (cal mol- )
-43 constant in Eq. (10) (cal mol-)
A4 constant in Eq. ( 11) (cal mol-l)
A5 constant in Eq. ( 12) (cal mall)
4 constant in Eq. (7) (cm3)
B2 constant in Eq. (9) (cal mol- K-l.)
B3 constant in Eq. (10) (cm3)
& constant in Eq. ( 11) (cm1.5 Cal-/ mo1-.5)
4 constant in Eq. (12) (cal mol- K-.72)
gE excess Gibbs energy (cal mall)
gii energy parameter for the i-i interactions (cal mol-)
gij energy parameter for the i-j interactions (cal mol-)
Gj binary parameter in the NRTL equation
 A. Vetere 1Fluid Phase Equilibria 90 (1994) 63-74 73

AHi vaporization enthalpy of the pure i compound (Cal mol-)

M molecular weight
M mean value
P equilibrium pressure (mmHg)
AP absolute mean deviation between calculated and experimental values of pressures (mmHg)
R universal gas constant (cal cm3 mol- K-l)
T absolute temperature (K)
AT absolute mean deviation between calculated and experimental values of temperature (K)
Tb normal boiling temperature (K)
X, mole fraction of the compound i in liquid phase
Yi mole fraction of the compound i in vapor phase
AY absolute mean deviation between calculated and experimental values of vapor composi-
tions
w, polarity parameter defined by Eq. (8)

Greek letters

%j non-randomness constant in the NRTL equation

az activity coefficient of the compound i at infinite dilution
;; Hildebrand solubility parameter (caF5 cn~-.~)
zij NRTL coefficient defined by Eq. (3)

Subscripts

1 water
2 other component of the mixture

References

Bruin, S. and Prausnitz, J.M., 1971. One-parameter equation for excess Gibbs energy of strongly non-ideal liquid
mixtures. Ind. Eng. Chem., Process. Des. Dev., 10: 562-572.
Fredenslund, Aa., Gmehling, J. and Rasmussen, P., 1977. Vapor-Liquid Equilibria using UNIFAC. Elsevier,
Amsterdam.
Gmehling, J. and Onken, U., 1977. Vapor-Liquid Equilibrium Data Collection, Aqueous-Organic Systems, Vol. 1,
Part 1. DECHEMA Chem. Data Ser., DECHEMA, Frankfurt.
Gmehling, J., Onken, U. and Artl, W., 1981. Vapor-Liquid Equilibrium Data Collection, Aqueous-Organic Systems
(Supplement l), Vol. 1, Par la. DECHEMA Chem. Data Ser., DECHEMA, Frankfurt.
Hansen, H.K., Rasmussen, P., Fredenslund, Aa., Schiller, M. and Gmehling, J., 1991. Vapor-liquid equilibria by
UNIFAC group contribution. 5. Revision and extension. Ind. Eng. Chem. Res., 30: 2352-2355.
Prausnitz, J.M., Eckert, C.A., Orye, R.V. and OConnell, J.P., 1967. Computer Calculations for Multicomponent
Vapor-Liquid Equilibria. Prentice-Hall, Englewood Cliffs, NJ.
Tassios, D., 1971. A single-parameter equation for isothermal vapor-liquid equilibrium correlations. AICHE J., 17:
1367-1371.
Vetere, A., 1977. A modified Heil-Prausnitz equation for excess Gibbs Energy. Can. J. Chem. Eng., 55: 70-
77.
74 A. Vetere 1 Fluid Phase Equilibria 90 (1994) 63-74

Vetere, A., 1989. An empirical modification of the Pitzer-Curl equation. Chim. Ind. Milan, 71(3): 76-80.
Vetere, A., 1993. Prediction of vapor-liquid equilibria of non-aqueous systems in the sub-critical range by using the
NRTL equation. Fluid Phase Equilibria, 91: 265-280.
Wu, H.S. and Sandler, S.I., 1989. Proximity effects on the prediction of the UNIFAC model: I. Ethers. AIChE J., 35:
168-172.
Wu, H.S. and Sandler, S.I., 1991. Use of ab initio quantum mechanics calculations in group-contribution methods.
2. Test of new groups in UNIFAC. Ind. Eng. Chem. Res., 30: 889-897.