Fluid Phase Equilibria, 47 (1989) 133-152 133

Elsevier Science Publishers B.V.. Amsterdam -Printed in The Netherlands

PREDICTION OF GAS SOLUBILITY AND VAPOR-LIQUID

EQUILIBRIA BY GROUP CONTRIBUTION

R. GANI, N. TZOUVARAS, P. RASMUSSEN and AA. FREDENSLUND

Instituttet for Kemiteknik, The Technical University of Denmark,

DK-2800 Lyngby, Denmark

(Received November 1, 1988; accepted in final form April 25, 1989)

ABSTRACT
The UNIWAALS group contribution equation of state as developed by Gupte et
al. (1986) has been revised and extended to applications involving mixtures with
non-condensable gases and applications to broader ranges of temperature and
pressure. This paper presents numerical and computational aspects related to
phase equilibrium predictions and examples of applications.

INTRODUCTION
Several group contribution models for the prediction of the solubilit of gases in
polar and non-polar liquids have been presented in the past, see e.g. 8 ander et al.
(1983) and Skjold-Jergensen (1984, 1988). The model of Sander et al. is an
extension of the UNIFAC model (Fredenslund et al. 1977) and uses the so-called
unsymmetric activity coefficient a preach to describe phase equilibria (Prausnitz
et al. 1986). Hence the model of 8 ander et al. is limited to pressures below ap-
proximately 15 bar. Both of the models mentioned above require cumbersome
estimation of solvent-solvent, gas-solvent and gas-gas group-interaction para-
meters.
The purpose of this work is to directly extend an existing model for predicting
activity coefficients - in this case the modified UNIFAC model by Larsen et al.
(1987) - to high pressures and to cover gas solubility at low and high pressures.
The aim is to avoid the tremendous task of re-estimating solvent-solvent group-
interaction parameters thus leaving only gas-solvent and gas-gas group-inter-
action parameters to be determined.
This work is a further development of the UNIWAALS equation of state of Gupte
et al. (1986).
In developing the UNIWAALS equation of state, Gupte et al. combined the van
der Waals equation of state with the modified UNIFAC method (Larsen et al.,
1987). The modified UNIFAC model was chosen (rather than old UNIFAC by
Fredenslund et al. (1977)) due to its improved temperature dependence. The
equation of state formulation rendered the large matrix of group-interaction
parameters for modified UNIFAC based on low-pressure VLE data, applicable
also tn high pressures. This approach ensured that the equation of state reprodu-
ced exactly the results of the activity coefficient model at low pressures and gave
excellent predictions at high pressures.
Gupte et al. (1986): however, did not include the possibility of application of the
UNIWAALS equation of state to mixtures containing non-condensables (suck? as
CO, CO,, N,, and CH,). Also, if the system temperature approached the crltlcal

0378-3812/89/$03.50 0 1989 Elsevier Science Publishers B.V.

134

temperature of any of the constituent components of the mixture, a special and

cumbersome extrapolation procedure was needed.
As formulated by Gupte et al. (1986), for VLE predictions, the UNIWAALS
equation of state requires a liquid-like root of the pure components at the
temperature and pressure of the system and of the mixture at the temperature,
pressure and composition of the phase (vapor or liquid) in question! As the system
temperature approaches the critical temperature of a component or the mixture,
the existence of liquid-like roots becomes uncertain. A procedure to obtain the
liquid-like roots is therefore necessary if the UNIWAALS equation of state is to
be applied under these conditions.
This paper describes a new formulation of the UNIWAALS equation of state that
retains the great advantages of the formulation by Gupte et al. (19861 and in
addition allows its annlication to mixtures containing non-condensables and to
conditions of temperatures and pressures approaching-the mixture critical point.
The paper also clarifies some inconsistencies in the previous formulation and
shows how the UNIWAALS equation of state can be applied to phase equilibrium
predictions at low and high pressures.

THEOKEIICALHACKGKOUND
Vidal (1978) and Huron and Vidal (1979) have proposed a method for deriving
mixing rules for eauations of state from excess Gibbs enerav models. Thev derived
the myxing rules by combining the Redlich-Kwong equation of state with the
NRTL liquid phase activity coefficient model (Renon and Prausnitz, 1968) on the
basis of the following assumptions:

i) gEIEos
= gElg-model at infinitepressure

ii) u=b at infinitepressure

iii) E = 0

In the above, g is the excess Gibbs energy. The subscript EOS indicates that gE is
evaluated from the equation of state, and the subscript g-model indicates that g
is evaluated from an excess Gibbs energy (or activity coefficient1 model. b is the
covolume parameter in the equation of state, u is the molar volume and us is the
excess molar volume.
A modified form of the mixing rule of Huron and Vidal has been proposed by
Mollerup (1986). Mollerups derivation is based on the zero pressure limit and
gives expressions for mixing rules for different equations of state (van der Waals,
Redlich-Kwong) combined with different liquid phase activity coefficient models
(Wilson, NRTL and UNIQUAC).
Soave (1986) has extended the approach pro osed by Mollerup to the deter-
mination of fugacity coefficients and K-values Por vapor-liquid equilibrium (VLE)
predictions.
All the above-mentioned approaches require: unlike that of Gupte et al. and that
of the present work, re-estimation of all the Interaction parameters in the excess
Gibbs energy model used.
Gupte et al. (19861, in developing the UNIWAALS equation of state, derived the
necessary mixing rule by combining the van der Waals equation of state and the
modified UNIFAC method (Larsen et al. 1987). The model avoids assumptions il,
ri), and iii) above, but introduces a new assumption:

= gElg-model
at all pressures.
 135

Mixing Rules
Essentially, all the different formulations of the equations of state employ the
same expression for the b parameter and derive different expressions for the a/b
parameter. Gupte et al. (1986) proposed the following mixing rules:

b = -&,b,

(1)

a-
-=- Cl
KTb f
(2)
where in this paper a is the same as the a-parameter of Gupte et al.,

and

(3)
According to assumption iv), Eq. (2) holds at each pressure level.
The expression derived by Mollerup (1986) and Soave (1986) can be obtained by
setting ti = 0 in Eq. (3). By setting C, = 0 and f = t; = 1, the mixing rule of
Huron and Vidal(1979) is obtained.
The mixing rule given by equations (2) and (3) was obtained by Gupte et al. (1986)
using assumption iu) and the van der Waals equation of state,

(4)
(The superscript prime of the u-parameter is a reminder that in UNIWAALS,
the u-parameter is volume-dependent.)

VLE Prediction: Expression for Fugacitv Coefficient

For VLE predictions employing equations of state, expressions for fugacity
coefficients are necessary. The expression for fugacity coeffkient, defined as

(5)
(where V is the total volume, y1 the number of moles of component i, nT the total
number of moles) can be obtained by combining P in Eq. (5) with Eq. (4) and
integrating the result. Gupte et al. (1986) derived the following expression for the
fugacity coefficient,

L
(6)
136

The above expression is not mathematically consistent since during its deri-
vation, the parameter a ib was considered independent of volume for an inte
gration with respect to volume, even though equations (2) and (3) indicate other-
wise. The reason why Gupte et al. (1986) obtained excellent results is that the
dependence of the a lb parameter on pressure is significant only at very hi h
pressures. It should also be noted that if C, as defined in Eq. (3) is negligib Be,
there is no inconsistency.

VLE Prediction: Calculation Procedure

For any VLE prediction problem solved by an iterative method, during each
iteration, values of the temperature, T, the pressure P and the phase composition,
5 are either specified or guessed. Fugacity coefficients are then determined
through equations of the type given by Eq. (6) for each phase. Examination of Eq.
(6) shows that in addition to T, P and r_, the value of the volume u (or com-
pressibility factor, Z = PuIRT) is necessary for each phase. The value of u (or Z) is
obtained by solving simultaneously equations (1) through (4).
Gupte et al. (1986) and Soave (1986) state that for the evaluation of the a lb
parameter (needed for determining Z or u), a liquid-like root for Z (or for volume
u), must be used. Once the a lb and b parameter values have been obtained, the
equation of state can be used in the usual way (i.e. using the actual, relevant
liquid or vapor roots) to obtain the phase Z values and fugacity coefficients.
A major problem arises if, in the evaluation of the a ib parameter, the liquid-
like root for Z does not exist either for the mixture or for the pure component.
This usually happens when A./B. (for the pure component) or A/B (for the
mixture) becomes less than 3.75. Gupte et al. (1986) proposed an extrapolation
procedure when T, = TT,, > 0.8 (approximately when AI/B,< 3.75). In the
above, A = a *P/R2T2 and B = bPRT.

VLE Prediction: Problem Definition for New Formulation

One alternative to overcome the consistency problem arising from neglecting the
volume-dependence of a lb is to set C, = 0 in Eq. (2). However, if we want to take
advantage of the existing parameter tables in a g-model at high pressures, this
alternative is not feasible.
A better alternative is therefore to reformulate the equation of state avoiding the
mathematical inconsistency and employing a procedure to supply hypothetical li-
quid-like values for Z when real values do not exist. A number less than (1 + B)/3
is regarded as a liquid-like value. Note that the hypothetical liquid-like value is
only used for determining the parameter a:
The g-model parameters will have to be determined for those not nresent in the
exist&g parameter table, i.e. those for gas-solvent and gas-gas group interaction.
This formulation will ensure identical nredictions (comnared to the e-model) at
low pressures and acceptable predictions at high pressures and temperatures
(even when the dependence of the a lb parameter on pressure becomes signi-
ficant) for non-ideal mixtures including those containingnoncondensables.

NEW FORMULATION OF THE UNIWAALS EQUATION OF STATE

Like GUD~ et al. (19861. the new formulation of the UNIWAALS eouation of state
is developed by combining the van der Waals equation of state and the modified
UNIFAC LOUD contribution method of Larsen et al. (19871. The van der Waals
equation gf state, in terms of pressure is given by Eq: (4). An expression for the
excess Gibbs energy may be obtained from the van der Waals equation of state by
using the relationship:
 137

In Eq. (7), + is the fugacity coefficient of the mixture and Q, is the pure
component fugacity coefficient at the temperature and pressure of the system.
Expressions for I$ and a, are derived using equations (4) and (5). The resulting
expression for g is:

-gE =In(l A)_ &

RT " I RT
(8)
Let us now define a new, volume-dependent parameter a by the relationship

-1 ma,
RT I -du=-&
vu2

(9)
For pure component i,

a =a

since both are independent ofvolume (see Eq: (12)).

Then., replacing the first integral in Eq. (8) by Eq. (9) yields the following exact
relationship for gL:

E
g

HT=-

The mixing rule for b is given by:

b = x x,b,
(11)
a, and b, of equations (10) and (11) are defined as follows:

(12)

ac= (27/64)R2T2c,lPc
1
(13)

(14)
a(T,.) in Eq. (12) can be any suitable function. The function proposed by Mathias
and bopeman (1983) is used in this work. It contains three parameters which are
fitted to pure component vapor pressure data. The expression is:

l+I?,(l-Tr +)+&'&I -Trt)+EB(l

1 1
(15)
138

Values of the parameters E,, E, and E, are for selected compounds, given in Table
1. Parameter values have been obtained for more than 200 compounds; they may
be obtained from the authors.The mixing rule for b is given by:

Table 1. Values of the pure component parameters for the function of

Mathias and Copeman (1983).

At this point both gE and a in Eq. (10) and v in Eq. (4) are unknown. b, b, and at are
obtained from equations (11) through (15). Thus, if a is known, equations (10) and
(4) can be used to determine g and v. The procedure for the evaluation of a is
given in the section JXvaluation of the parameters a and D.
The difference between the a-parameter of Eq. (4) and the a-parameter of Eq.
(10) should be noted. A relationship between the two parameters can be obtained
by differentiating Eq. (9) on both sides with respect to volume, i.e.,

The new formulation of the UNIWAALS equation of state then becomes:

(17)
where D = daldv, and both a and D are functions of (T, P, v and 2).
 139

Expression for the Fugacitv Coefficient

The vapor-liquid equilibrium criterion is given by

,
Q(X, = qr, i = 1,2,...N
(18)
where @ is the fugacity coefficient, x and y are compositions of component i in the
different phases and the L and V superscripts indicate the liquid and vapor pha-
ses, respectively.
Using equations (5) and (17) together with the definitions of a, D, etc., the
following expression for the fugacity coefficient is derived:

7
L
lnQL= $Z- & +
(19)
where

Z= PuIRT ; A =aPIRT ; B = bPlRT

ZL= (atilan,),
p n ; A, = (anAlan),,, p n
,=, ,T
(21)

Evaluation of the Parameters a and D

The parameters a and D = daldu must be determined before any prediction of
thermodynamic properties can be made. The evaluation of a and daldu, however,
is not so straight forward. Examination of Eq. (10) shows that for a given com-
position, temperature and pressure, the u-parameter can be determined if the
values of gEIRT and v are known. For the evaluation of the a-parameter, there-
fore, we propose the following

and, gIRT = gEIRTIg_mode,

(22)
in Eq. (10). The g-model in this work is the modified UNIFAC group contribution
method (Larsen et al. 1987).
Because of the complexities involved in obtaining an analytical expression for
D = daldu, a simplified form has been derived. This is given in appendix 2. The
condition given by Eq. (22) is also used for the evaluation ofD.
The implication of Eq. (22) is that, for the vapor-phase, the functions a and D are
obtained by considering a hypothetical liquid phase with the same composition
and temperature as the vapor phase.
It may happen at a specified temperature, pressure and composition that the
equation of state does not have a liquid-like root. For such conditions, a hype-
thetical, liquid-like value of u or Z must be used. Appendix 1 gives details of how
u or Z are determined for a hypothetical liquid, where a real liquid-like root does
not exist.
The procedure described in appendix 1 replaces the cumbersome extrapolation
procedure by Gupte et al. (1986), where gE in Eq. (22) was fitted to a NRTL-
140

expression as a function of temperature at conditions where a liquid-like root

exists, and this Iit was then used to extrapolate gE to temperatures where liquid-
like roots do not exist. The new procedure is more consistent and may be used for
much larger extrapolations than that of Gupte et al.
It should be emphasized that it is onlv in the evaluation of the a and D parameters
liquid-like (hypothetical or real) roots are used in vapor-phase computations.
Once a and D are evaluated, the actual vapor-phase properties are obtained from
Eq. (17) using the vapor-like root. This is an iterative process, which is further
described below.

COMPIJIATIONAI, DRTAILS
Prediction of phase equilibria requires evaluation of the fugacity coefficient as
shown in Eq. (18). The expression for the fugacity coefficient is given by Eq. (19).
From this it may be seen that the following terms must be evaluated for each
phase
A,,B,,Z,A,B~D,~, andii

a) Determination of the pure component properties and (A,, BL) and B.

a and b are determined by using equations (12) through (15). Aa and B, are
obtained from Eq. (20) and B is obtained from

b) Evaluation of A, D and A,.

RearrangingEq.(lO) and introducing condition of Eq. (22) yields the mixing
rule for A.

(23)
where

cl z z - 1 x,z, - [h(z -B) - z xi In(ZI - B,) 1

(24)
Eq. (17) can be rewritten in terms of 2 as follows

z3-z2(B+1 + -$+,,A+ g,-AH=0

(25)
where
da
D = - = /(I, I, ,I)
du
The problem in this section therefore can be defined thus:
* Given, x_,T, P.
* Determine A, At,
Equations (231, (24) and (25) are solved simultaneously for A at a correspon-
ding liquid-like Z. At the solution, AL (see Eq. (21)) is determined from:
 141

A Z-B Z-B Ai
Ai=-(ZI-Z) ~- +Z(Z - Z i + In( Z - B )+ 1 Z-B + ~i -lnYilg-mdel)
(26)
The Z, in Eq. (26) is obtained by finite difference corresponding to 'liquid-
like' Z v a l u e s .
From numerical experience of using the simplified expression for d a / d v of
appendix 2, it was observed that at low pressures and for the liquid phase,
the value of D is small enough to be neglected. Thus, under these conditions,

~'~a
and Eq. (25) reduces to

Z3-Z2(B+ 1)ZA-AB=0
(27)
At h i g h e r pressures, the value of D becomes significant ( > 10 -4) and hence
can no longer be neglected. Including the effect of D means adding correc-
tions to the coefficients of the Z 2 and Z terms in Eq. (25).
Using the approximate expression for D in appendix 2, it has been found
t h a t D becomes significant near the critical region of the mixture when a
'liquid-like' root for Z for the different phases cannot be obtained. The tem-
p e r a t u r e and pressure at which D becomes significant is for some m i x t u r e s
given in Table 2. It can be noted that the T and P at which D becomes
significant, move further away from the critical point, with the presence of a
lighter component.

T a b l e 2. Values of P, T, x(1), y(1) at which D becomes s i g n i f i c a n t

( > 0.0001).

System T(K) P greater x(1) mole y(1) mole

than (atm) fraction fraction
methanol(1)-water(2) 508.15 68.0 0.870 0.880
523.15 82.0 0.732 0.756
ethanol(1)-water(2) 523.15 71.0 0.737 0.740
548.15 97.0 0.455 0.486
pentane(1)-hexane(2) 479.60 27.7 0.500 0.549
butane(1)-pentane(2) 449.11 34.9 0.460 0.500
propane(1)-pentane(2) 384.00 22.2 0.500 0.708
ethane(1)-octane(2) 290.20 11.3 0.500 0.993

Note: All the alkane-alkane predictions were made by representing the

compounds with the sub-groups of the "CH2" main group of modified
UNIFAC (Larsen et al., 1987).

A simple rule can thus be proposed to determine when D m u s t be evaluated

and used in Eq. (25). The rule is as follows: Obtain the smallest value for Z
from Eq. (25) by setting D = 0. If this value of Z is not 'liquid-like', D cannot
be ignored and must be evaluated.
142

Note that the 2, 2 and Z used in the evaluation of A, D and A all

correspond to liquid-like volumes which are not used for any other c&u
lations.
cl Determination of In +,
In 4 is determined by using Eq. (19). From steps a) and b) we have values for
AI, b , A, B, D and A In order to evaluate In 4 , we need values for Z and Z
Z is obtained from the equation of state (Eq. 17). In this case Z is the real
root ofEq. (17). When the phase type is known, 2 corresponds to the smallest
real root for the liquid phase, and the largest real root for the vapor phase.
If the phase identity is unknown, Z will correspond to the real root with the
minimum G (Gibbs energy). Zz (see Eq. 21) is obtained by finite difference.

SinPle-Model or Two-Model Formulation

At low pressures and/or temperatures (far from the critical point), the solution for
Z usually has two roots (a liquid-like and a vapor-like) and the liquid phase
fugacities obtained from Eq. (19) become exactly equal to those obtained from the
g-model. Thus, for these conditions of temperature and pressure exactly the same
phase diagrams result whether the UNIWAALS equation of state or the modified
UNlFAC activity coefficient model has been used. This is one of the strenghts of
the present approach.
At higher pressures and/or temperatures, a condition is reached when the lightest
component in the mixture does not have a liquid-like compressibility factor.
From this pressure and temperature, the liquid phase fugacity obtained from Eq.
(19) is no longer equal to that from the g-model. At these conditions of tern-
perature and pressure, Eq. (19) is used to determine +, for each phase, employing
therefore one model for each phase.
At even higher pressures and/or temperatures, a point is reached when, for a
specified phase, the mixture 2 also does not have a liquid-like value. At these
conditions, D becomes significant and Eq. (19) must again be used for each phase.
Thus, it can be seen that since the same equation (model) is used for either phase,
and if D can be correctly determined, the present formulation is a single-model
approach. Thus it should be possible to predict the critical point of a mixture and
to determine the VLE phase envelope from normal pressure to the critical point.

The numerical results from applications of the UNIWAALS equation of state are
grouped into three parts. In part (a) the results for non-ideal mixtures with
condensable components at different temperatures and pressures are given. For
part (b), the results of non-condensable gas solubility correlations are given. Part
(c) shows the results of a phase envelope (including critical point) prediction.

Non-Ideal Mixtures
Figures (11, (2) and (3) compare the VLE predictions obtained by the new
formulation of the UNIWAALS eauation of state with exoerimental values for
the binary mixtures, ethanol-water, methanol-water, benzene-phenol. These
results confirm that at low pressures and temperatures, there is no loss of
accuracy of prediction, compared with using modified UNIFAC. For example, the
ethanol-water azeotrope is predicted using UNIWAALS with very good accuracy.
Also at high pressures and temperatures, the predictions using the same
parameters as at low pressure are within acceptable limits of error. It should be
noted that at conditions of high pressure and temperature, unlike Gupte et al.
(19861, the proposed methodology does not need any extrapolation procedure.
That is, the same parameters have been used for all the predictions. For the case
of methanol-water, the modified UNIFAC parameters determined by Gupte et al.
(1986) were used.
 143

70
A
I

50 data (523.15)

30 I I I L I I I I

0.0 0.2 0.4 0.6 0.8 1.0

mole fraction, x ,' Y, ---_>

Figure 1. Comparison of VLE prediction with UNlWAALS and experimental
data for the system methanol(l)-water(2) at 523.15 K. (Gmehling, J. and Onken,
U., Vapor-Liquid Equilibrium Data Collection, DECHEMA Chemistry Data
Series, Frankfurt, 1977).
1001

l experimental data
--- UNIWAALS'
UNIWAALS2

300
0.0 0.4 0.8

mole fraction, x ____>

1' Yl

Figure 2. Comparison of VLE prediction with UNIWAALS and experimental

data for the system ethanol(l)-water(2) at 523.15 K and 548.15 K (Gmehling, J.,
and Onken, U., Vapor-Liquid Equilibrium Data Collection, DECHEMA Che-
mistry Data Series, Frankfurt, 1977).
Note: UNIWAALS (results from Gupte et al. (1986))
UNIWMLS (results from present version)
 r

0.5 1.0
mole fraction, x, ---->

Figure 3. Comparison of VLE predictions with UNIWAALS and modified

UNIFAC with experimental data for the system benzene(l)-phenol(2) at low
pressures and at 343.15 K and 353.15 K (Schmelzer, J., SEP 8809, Instituttet for
Kemiteknik, DTH, Lyngby, Denmark, 1988).

Gas Solubility
The modified UNIFAC method parameter tables do not include any gases. Thus,
for gas solubility predictions, first the missing gas-solvent group interaction
parameters were estimated. This was done by fitting experimental gas solubility
data to predictions from the UNIWAALS equation of state with D = 0. In all
cases, P, T, 4, y data were used, and P and y were calculated , given T and x. Since
new parameters are estimated, the error made by neglecting D will be compen-
sated by the estimated parameters.
The results from using UNIWAALS equation of state for the description of the
binary systems, CO,-methanol, N. -benzene, butane-phenol and methane-alkanes
(C, to C,) are given in Table 3. Tk e estimated group interaction parameters are
 145

given in Table 4. In all cases, the parameters were obtained by minimizing the
objective function,

where superscripts talc indicate calculated values, and expt indicate experimen-
tal values and, subscript i indicates data point i.
Figures 4, 5 and 6 compare the experimental values with those obtained with
UNIWAALS. As can be seen from these figures, excellent representation of the
experimental data has been possible.

Table 3. Results of gas solubility calculations.

Bubble Point (o/oRMS)

System T (K) P(atm) I
P (rel.1 y(l) cabs.1

nitrogen(l)-benzene(2) 348.15 60-303 1.27 0.20

373.15 63-155 1.45 0.25
carbon dioxide(l)- 29815 8-48 2.87 0.15
methanol(2)
butane(l)-phenol(2) 298.15 l-4 2.99 0.04
methane(l)-alkanes(2) 248.15 - l-120 5.70 0.51
(C5 - C9) 348.15

Reference of experimental data:

butane-phenol system: Obst, S.O., Ph.D. Thesis, Karl Marx University, Leipzig,
Sektion Chemie, 1988.
all other systems: Knapp H., Doring, R., Oellrich, L., Plocker, V., and
Prausn~tz, J. M. Vapor-Liquid Equilibria for Mixtures of
Low Boiling Substances, DECHEMA Chemistry Data
Series, Vol. VI, 1982.

Phase Envelope and Critical Point

Prediction of the phase envelope and critical point for the system butane-pentane
is shown in Table 5. The predicted results are compared with predictions by SRK.
These results illustrate that the UNIWAALS equation of state can predict a cri-
tical point.
146

Table 4. Binary group interaction parameters corresponding to results of

table 3.

System Interaction Parameter value

nitrogen(l)-benzene(2) N2-ACH 152.8080

2.0311
0.0000
ACH-N2 205.0500
-2.5920
0.0000
carbon dioxide(l)-methanol(2) C02-CH30H -110.7220
0.0000
0.0000
CH30H-CO2 1746.9300
0.0000
0.0000
butane(l)-phenol(2) ACH-CH2 20.7320 -
0.0000
0.0000
CH2 -ACH 569.1590
0.0000
0.0000
OH-CH2 -124.2980
0.0000
0.0000
CH2 -OH 303.2670
0.0000
0.0000
methanol(l)-alkanes(2) (C5 - C9) CH4-CH2 18.9700
-0.0096
20.9600
CH2-CH4 99.4500
-0.9580
-35.9900

Note: CH2 main group as belonging to a gas (in this case butane).

A mathematically consistent formulation of the UNIWAALS equation of state

has been presented. The evaluation of the attractive energy parameter of the
equation of state, a, requires liquid-like volumes, even though the equation of
state may not have a root corresponding to a such volume. A procedure for
generating hypothetical liquid-like root has been developed. The use of
hypothetical, liquid-like roots is limited to the evaluation of the u-parameter.
Once a is evaluated, the equation of state is used in an entirely consistent
manner.
At high pressures, however, additional empiricism is necessary where we, as in
this work, wish to make use of the already existing modified UNIFAC group-
interaction parameter table, based on low pressure VLE and He data. A great
advantage of this approach is that of a g-model, intended for low-pressure predic-
tions, may be extended to high pressures without adding new parameters or loss
of accuracy in the low pressure range. If the temperature-dependence of the g-
model is adequate, also the high-pressure VLE predictions will be accurate.
 147

348 K

l experimental data
---- UNIWAALS ( 373 K )
- UNIWAALS ( 348 K )
.

373 K

I I I I
0
0.0 0.2 0.4 0.6 0.8 1.0

mole fraction, x,. y, ---->

Figure 4. Correlation of VLE experimental data with UNIWAALS for the system
nitrogen(l)-benzene(2) at 348 K and 373 K.

0 I I I I ! t I

0.0 0.2 0.4 0.6 0.8 1.

mote fraction, x,, y, ---->
Figure 5. Correlation of VLE experimental data with UNIWAALS for the system
carbon dioxide(l)-methanol(2) at 298 K.
148

d
0 experimental data
- UNIWAALS

313.15 K

/--

0.0 0.5 1 .o
mole fraction, x,

Figure 6. Correlation of VLE experimental data with LJNIWAALS for the system
butane(I)-phenol(2) at 313.15 K, 343.15 K and 373.15 K (Schmelzer, J., SEP
8809, Instituttet for Kemiteknik, DTH, Lyngby, Denmark, 1988).

With the new formulation, it has been possible to extend the UNIWAALS
equation of state to gas solubility calculations over wide ranges of temperature
and pressure. The numerical results show excellent matching of experimental
data. These results also confirm the feasibility of application of UNIWAALS with
solvent-solvent group-interaction parameters from Larsen et al. (1987) to gas
solubility predictions.
To render UNIWAALS generally applicable to multicomponent gas solubility
calculations, gas-gas and gas-solvent modified UNIFAC group-interaction para-
meters are needed. Such a parameter table is under development. When this is
completed, UNIWAALS will be a versatile, important tool for high pressure pha-
se equilibrium calculations whenever polar components and super critical gases
are present.
 149

Table 5. Prediction of phase envelope by UNIWAALS for the system butane(l)-

pentane(2) (equimolar mixture).

Pressure (atm)

27.18 27.18
30.00
34.90 34.90
36.01 36.01
36.29 36.31
36.58
36.87
37.15
36.84 36.37
36.42 36.28
36.74
34.90 34.12
27.18 27.88
16.49 16.20
10.00 12.47

Note: K in In(K) signifies the equilibrium constant.

REFERENCES
Fredenslund,. Aa., Gmehling, J. and Rasmussen, P., 1977. Vapor-Liquid
Equilibria Using UNIFAC, Elsevier, Amsterdam.
Gundersen, T., 1982. Numerical Aspects of the Implementation of Cubic
Equations of State in Flash Calculation Routines. Computers and Chemical
Engineering, 6: 245255.
Gupte, PA., Rasmussen, P. and Fredenslund, Aa., 1986. A New Group-
Contribution Equation of State for Vapor-Liquid Equilibria. Ind. Eng. Chem.
Fundam., 25: 636-645.
150

Huron, M.-J. and Vidal, J., 1979. New Mixing Rules in Simple Equations of State
for Representing Vapor-Liquid Equilibria of Strongly Non-ideal Mixtures. Fluid
Phase Equilibria, 3: 255.271.
Jovanovic, S. and Paunovic, R., 1984. Generating Appropriate Density Values
from a Cubic State Equation to Avoid False Unit K Values. Application to
Distillation Problems. Ind. Eng. Chem. Process Des.Dev., 23: 801-805.
Larsen, B.L.,.Rasmussen, P. and Fredenslund, Aa., 1987. A Modified UNIFAC
Group-Contribution Model for Prediction of Phase Equilibria and Heats of
Mixing. Ind. Eng. Chem. Res., 26: 2274-2286.
Mathias, P.M. and Copeman, T.W., 1983. Extension of the Peng-Robinson
Equation of State to Complex Mixtures: Evaluation of the Various Forms of the
Local Composition Concept. Fluid Phase Equilibria, 13: 91-108.
Mollerup, J., 1986. A Note on the Derivation of Mixing Rules from Excess Gibbs
Energy Models. Fluid Phase Equilibria, 25: 323-327.
Prausnitz, J.M., Lichtenthaler, R.N. and de Azevedo, E.G., 1986. Molecular
Thermodynamics of Fluid Phase Equilibria. 2nd Edition, Prentice-Hall.
Renon, H. and Prausnitz: J.M., 1968. Local Compositions in Thermodynamic
Excess Functions for Liquid Mixtures. AIChE Journal, 14: 135-144.
Sander, B., Skjold-Jorgensen, S. and Rasmussen, P., 1983. Gas Solubility
Calculations. I. UNIFAC. Fluid Phase Equilibria, 11: 105-126.
Skjold-Jorgensen, S., 1984. Gas Solubility Calculations. II. Application of a New
Group-Contribution Equation of State. Fluid Phase Equilibria, 16: 317-351.
Skjold-Jargensen, S., 1988. Group Contribution Equation of State (GC-EOS): A
Predictive Method for Phase Equilibrium Computations over Wide Ranges of
Temperature and Pressures up to 30 MPa. Ind. Eng. Chem. Res., 27: 110-118.
Soave, G., 1986. Infinite-Pressure Excess Functions and VLE K values from
Liquid-Phase Activity Coefficients. Fluid Phase Equilibria, 31: 147-152.
Vidal, J., 1978. Mixing Rules and Excess Properties in Cubic Equations of State.
Chem.Eng.Sci., 33: 787-791.

ACKNOWLEDGEMENT
One of the authors (N.T.) gratefully acknowledges a scholarship from the
Technical University of Denmark that supported his visit at Instituttet for
Kemiteknik. All of the authors thank wholeheartedly dr. G. Soave, professor R.
A. Heidemann, dr. P. Gupte and others, who have contributed to this work.

APPENDIX 1: Determination of Hypothetical Roots

Being a cubic equation, the van der Waals equation of state, defines a third-
degree polynomial with respect to the compressibility factor 2. Such a function
may have five possible shapes (Jovanovic and Paunovic, 19841. When a liquid-
type root is needed but is unavailable, it is possible to determine a hypothetical
one as follows:
Let f(Z) = Z9-Z2 (1 + B) + AZ - AB
For f(Z) curves that show local extrema, the Z-value that results in a local
maximum may be used as a hypothetical liquid-type root, when such does not
actually exist. This corresponds to curve II (Jovanovic and PaunoviC) and Z =
Z,.. For f(Z) curves that have no local extrema, the Z-value that corresponds to
the inflection point can be used. This corresponds to curve I (JovanoviC and
Pauovic) and Z = Z,.. Z,,,l.is the point of local maximum and Z,,is the inflection
point.
The above choices can be considered as extensions of the ideas presented by
Gundersen (1982). There are two details that need be considered:
 151

Firstly, the choice of a hypothetical liquid-type root may lead to a Z-value that
causes problems to terms of the form In (Z-B) such as those encountered in the
mixing rule for a and the expressions for the fugacity coefficients. Thus a correc-
tion is necessary, that applies to the hypothetical liquid-type root determined as
above and ensures that such problems are avoided. The corrections follow the
guidelines presented by Gundersen.
The basic assumption is that f(Z) vanishes by correcting parameter B only. If Bc
stands for the corrected parameter B, then the hypothetical root is multiplied by
B/BC = [l + RZ)I(B(A + Z))]- where Z is the hypothetical root.
Secondly, the hypothetical root, determined as above, may correspond to molar
volumes that are larger than the volume corresponding to the critical compressi-
bility factor related to the cubic EOS. In such cases the critical compressibility
factor is used as the hypothetical liquid-type root.

APPENDIX 2: Approximate Expression for D

We start with Eq. (25). Differentiating with respect to Z and neglecting the dB/dZ
and dD/dZ terms, yields,

Therefore,

D
-=((3Z2-2Z(l+B)+A+
RT
(A.2)

The dA/dZ term in Eq. (A.2) is obtained by differentiatin Eq. (23) with respect to
Z. Alternatively, dA/dZ can be determined by finite di% erence. Neglecting the
dB/dZ term, the analytical expression for dAJdZ is given by,

dA
Z-ZI(Z-B)+ C,+ =&,A,Z,- gEIRTlg_modo,
z=
(A.3)

NOMENCLATURE

a energy size parameter

A EOS parameter (see Eq. 20)

b covolume parameter

B EOS parameter (see Eq. 20)

Cl term defined by equation (24)

D EOS parameter

E,, E,, E, parameters used in Eq. (15)

152

gE excess Gibbs energy, caUmol

n n u m b e r of moles

P p r e s s u r e , arm.

R gas c o n s t a n t

T temperature

U m o l a r volume, m3/mol

excess m o l a r volume, m3/mol

V total volume, m a

X mole fraction

Z c o m p r e s s i b i l i t y factor

Greek Symbols

C1 t e r m defined by e q u a t i o n (15)

fugac[ty coefficient

g a c t i v i t y coefficient

Superscripts

E excess p r o p e r t y

o pure c o m p o n e n t p r o p e r t y

Subscripts

C critical p r o p e r t y

EOS p e r t a i n i n g to E O S

g- m o d e l p e r t a i n i n g to g~-model

i,j c o m p o n e n t i, c o m p o n e n t j

r reduced property

T total n u m b e r