1580

Ind. Eng. Chem. Res. 1998, 37, 1580-1585

Comparison of the Peng-Robinson and Soave-Redlich-Kwong

Equations of State Using a New Zero-Pressure-Based Mixing Rule
for the Prediction of High-Pressure and High-Temperature Phase
Equilibria
Chorng H. Twu,* John E. Coon, and David Bluck
Simulation Sciences Inc., 601 Valencia Avenue, Brea, California 92823

The Peng-Robinson (PR) and Soave-Redlich-Kwong (SRK) equations of state are probably
the most widely used cubic equations of state in the refinery and gas-processing industries for
the prediction of vapor-liquid equilibria for systems containing nonpolar components. The new
mixing rules which have recently been developed that combine liquid activity models with the
equations of state, however, have extended the application of such equations to highly nonideal
systems. A new zero-pressure-based mixing rule is presented here that reproduces, with
extremely high accuracy, the excess Gibbs free energy as well as the liquid activity coefficients
of any activity model without requiring any additional binary interaction parameters. We
examine the performance of the Peng-Robinson and Soave-Redlich-Kwong equations of state
using the NRTL liquid activity model with binary parameters determined at low temperatures
in this new mixing rule, MHV1, and Wong-Sandler for the prediction of high-pressure and
high-temperature phase equilibria.
Introduction
Among many equations of state proposed for predicting the phase behavior of nonpolar systems, cubic
equations of state have been widely used because of
their simplicity and accuracy. Among the cubic equations of state, two equations which have enjoyed widespread acceptance in the refinery and gas-processing
industries are the Soave (1972) and the Peng-Robinson
(1976) equations of state. This paper, therefore, will
focus on these two equations of state. One of the most
frequent questions asked by engineers is which equation
of state, PR or SRK, should be selected for phase
equilibrium prediction. To answer this question, we will
compare the performance of extended PR and SRK
equations on the prediction of phase equilibria for
nonideal mixtures.
The application of the PR or SRK equation of state
to systems containing highly nonideal components
requires an appropriate mixing rule for the equation of
state parameter a. Huron and Vidal (1979) pioneered
linking the equation of state parameter a to the excess
Gibbs free energy at infinite pressure. However, their
mixing rule has not become widely used because the
available excess Gibbs energy parameters at low pressures cannot be used in their mixing rule.
In view of that, several authors have proposed different approaches to directly use the existing liquid
activity model parameters in equations of state. Among
them, two models have been quite successful. One is a
zero-pressure model by Michelsen (1990a,b), Dahl and
Michelsen (1990), and Heidemann and Kokal (1990) to
equate the excess Helmholtz free energy (AE) at zero
pressure from an equation of state to that from an
activity model, and the other is an infinite-pressure
* Corresponding author. E-mail: ctwu@simsci.com. Telephone: (714) 985-5298. Fax: (714) 579-0113.

model by Wong and Sandler (1992) to force the AE at

infinite pressure derived from a cubic equation of state
to be equal to that from a liquid activity coefficient
mode. Both models can directly use the available
activity coefficient model parameters from low-pressure
data in their mixing rules for predicting phase equilibria
at high temperatures and pressures quite successfully.
However, neither the zero-pressure models such as
MHV1 or MHV2 nor the infinite-pressure approaches
such as Huron-Vidal or Wong-Sandler can accurately
reproduce the GE model with which it is combined
(Coutsikos et al., 1995; Kalospiro et al., 1995).
A methodology has been developed to extend the
infinite-pressure Twu-Coon (1996) mixing rule to correctly reproduce the incorporated GE model without
introducing any additional binary interaction parameters. With this capability, the available activity coefficient models at low pressures can be used directly in
this new mixing rule. We will insert the NRTL excess
Gibbs free-energy model into the PR or SRK equations.
The comparison of the PR or SRK equations will be
made on the same exact basis (i.e., the same GE model
and the same binary interaction parameters reported
in the DECHEMA Chemistry Data Series for the GE
model, without using any other regressed binary interaction parameters, are used in both the PR and the SRK
equations of state). The mixing rules included in this
comparison are our new mixing rule and the MHV1 and
Wong-Sandler mixing rules.
Extended Twu-Coon Mixing Rule
A two-parameter cubic equation of state is considered
here:

P)

RT
a
v - b (v + ub)(v + wb)

(1)

where P is the pressure, T is the absolute temperature,

S0888-5885(97)00642-8 CCC: $15.00 1998 American Chemical Society

Published on Web 03/13/1998

Ind. Eng. Chem. Res., Vol. 37, No. 5, 1998 1581

Table 1. L, M, and N Parameters of the

Temperature-Dependent r Function Given by Equation 3
for Pure Components Used with the PR Cubic Equation
of State
component
n-pentane
n-hexane
n-heptane
cyclohexane
benzene
acetone
methanol
ethanol
water

Tc, K

Pc, bar

469.70 33.70
507.85 30.31
540.16 27.36
553.58 40.73
562.16 48.98
508.20 47.01
512.64 80.97
513.92 61.48
647.13 220.55

0.829 974
0.863 991
0.823 369
0.844 513
0.877 078
0.882 439
0.873 183
4.822 35
0.870 460

1.658 79
3.379 34
1.458 72
2.481 53
3.593 57
1.275 92
2.145 38
0.138 949
1.960 14

Table 2. L, M, and N Parameters of the

Temperature-Dependent r Function Given by Equation 3
for Pure Components Used with the SRK Cubic Equation
of State
component

T c, K

Pc, bar

n-pentane
n-hexane
n-heptane
cyclohexane
benzene
acetone
methanol
ethanol
water

469.70
507.85
540.16
553.58
562.16
508.20
512.64
513.92
647.13

33.70
30.31
27.36
40.73
48.98
47.01
80.97
61.48
220.55

0.379 229
0.158 080
0.340 339
0.245 880
0.163 664
0.479 844
0.690 551
1.076 46
0.413 297

0.841 706
0.872 819
0.844 963
0.845 046
0.860 016
0.870 627
0.911 298
0.964 661
0.874 988

1.823 31
3.844 18
2.383 32
2.258 95
2.984 98
1.790 10
1.969 41
1.353 69
2.194 35

and v is the molar volume. The constants u and w are

equation of state dependent (for the PR equation, u )
-0.4142, w ) 2.4141; for the SRK equation, u ) 0, w )
1). The parameter a in eq 1 is a function of temperature, and the parameter b is assumed to be a constant
for pure components. The value of a(T) at temperatures
other than the critical temperature, ac, can be calculated
from

(2)

The R function, R(T) in eq 2, is a function only of the

reduced temperature, Tr ) T/Tc. Since the prediction
of pure-component vapor pressure must be of high
accuracy for accurate vapor-liquid calculations, we
have chosen to use the R correlation of Twu et al. (1991):

R ) TrN(M-1)eL(1-Tr

NM)

(3)

Equation 3 has three parameters, L, M, and N, which

are unique to each component and are determined from
the regression of pure-component vapor pressure data.
The values for the components used in this study are
given in Tables 1 and 2 for the PR and SRK equations,
respectively. Consequently, the PR and SRK equations
will not suffer in accuracy by comparison if the R
correlation can reproduce the same accuracy for purecomponent vapor pressure from the triple point to the
critical point.
Twu and Coon (1996) have related the excess Helmholtz free energy, AE, with respect to a van der Waals
fluid to the Helmholtz free-energy departure function,
A, by the following:

AE - AEvdw ) A - Avdw

(5)

with C1, avdw, and bvdw being

L
0.331 853
0.132 963
0.507 729
0.140 885
0.081 0842
0.598 248
0.516 917
2.665 93
0.386 676

a(T) ) R(T)ac

E
AE Avdw
a* a*
vdw
) C1
RT
RT
b* b*
vdw

(4)

Equation 4 was used by Twu and Coon (1996) to derive

the following mixing rule for the cubic equation of state
mixture a and b parameters at infinite pressure:

C1 ) -

1
1+w
ln
w-u
1+u

(6)

i j xixjxaiaj

(7)

avdw )
bvdw )

i j xixj 2(bi + bj)

(8)

E
AE and Avdw
in eq 5 are the excess Helmholtz energy
at infinite pressure evaluated from a cubic equation of
state using the complete mixing rules for its a and b
parameters and using the van der Waals mixing rules
for its a and b parameters (avdw and bvdw), respectively.
If eq 4 is applied at zero pressure, instead of infinite
pressure, an equation containing liquid volume is
obtained:

[(

)( )]
)
(

v*
AE0vdw
AE0
0vdw - 1 bvdw
) ln
RT
RT
v*
b
0 - 1
a*
v*
v*
0 + w
vdw
0vdw + w
1 a*
ln
ln
w - u b*
v*
+
u
b*
v*
0
vdw
0vdw + u

[ (

)]

(9)

AE0 and v*0 ) v0/b are the excess Helmholtz energy and
reduced liquid volume at zero pressure. As mentioned,
the subscript vdw denotes that the properties are
evaluated from the cubic equation of state using the van
der Waals mixing rule for its a and b parameters. The
zero-pressure volume is obtained from eq 1 by setting
pressure equal to zero and selecting the smallest root:

v*0 )

{(

1 a*
a* 2
-u-w - u+w2 b*
b*

) [(

4 uw +

)] } (10)

a*
b*

1/2

Equation 10 has a root as long as

a*
g (2 + u + w) + 2x(u + 1)(w + 1)
b*

(11)

Equations 9 and 10 represent an exact model for a new

mixing rule. However, because the equation of state
parameter a*/b* and the zero-pressure liquid volume
are interrelated by eqs 9 and 10, the exact model does
not permit explicit solution of eq 9 for a*/b* and an
iterative technique is required for the solution. If eqs
9 and 10 are used to solve for a*/b*, the resulting new
mixing rule will give an exact match between the excess
Helmholtz free energy of the equation of state at zero
pressure and that of the incorporated excess Gibbs freeenergy model. Nevertheless, the nonexplicit nature of
the expression for the mixing rule becomes cumbersome
in the evaluation of thermodynamic properties such as
fugacity coefficients from the equation of state. This
paper presents a methodology to overcome this obstacle
to obtain an explicit expression for this new mixing rule.
A variety of alternatives have been proposed to
simplify the exact model so that the equation of state
parameters, a and b, can be explicitly expressed (Michelsen, 1990b; Dahl and Michelsen, 1990). However,

1582 Ind. Eng. Chem. Res., Vol. 37, No. 5, 1998

these modifications of the exact model sacrifice to some

extent the quality of the match between the equation
of state and the GE model. For example, the MHV1
model developed by Michelsen (1990b) can be alternately derived from our new mixing rule by assuming
v*0 is a constant (1.235 47 for SRK and 1.227 56 for PR),
instead of solving for it from eq 10. This means that
the MHV1 model assumes that the ratio of the zeropressure liquid volume to the close packing parameter,
b, is the same for the mixture and for all pure components. This is the main reason why the MHV1 model
performs poorly in reproducing the behavior of the GE
model when applied to systems either with components
that are different in size or where the value of v*0 of the
system is not close to the fixed values given above.
In this paper, v*
0 will not be assumed to be constant.
We propose instead that the ratio of the zero-pressure
liquid volume to b for the system, v*
0, be assumed to be
the same as that of the van der Waals fluid, v*0vdw.
Using eqs 7 and 8 for a and b in eq 10, v*0vdw can be
readily calculated from the equation. Equation 10 is
used to calculate v*
0vdw for both the mixture and the
pure components. Equation 9 can then be simplified
to

( ) [

AE0
bvdw
AE0vdw
a* a*
vdw
+ Cv0
) ln
RT
RT
b
b* b*
vdw
with the density function Cv0 being

Cv0 ) -

(13)
vdw

( )]

E
bvdw
C1 AE0
AE0vdw
AE Avdw
)
- ln
RT
RT
Cv0 RT
RT
b

(14)

Substituting eq 14 into the mixing rule proposed by Twu

and Coon (1996) results in a new and explicit mixing
rule in terms of AE0 at zero pressure:

[
[

b*
vdw - a*
vdw

(
(

( ))]
( ))]

E
AE0vdw
a*
bvdw
1 A0
vdw
- ln
1+
b*
C
RT
RT
b
vdw
v0

a* ) b*

E
a*
bvdw
AE0vdw
1 A0
vdw
+
- ln
b*vdw Cv0 RT
RT
b

AE0vdw

Since AE0 in eqs 15 and 16 is at zero pressure, its

value is identical to the excess Gibbs free energy at zero
pressure, GE. Therefore, any activity model such as the
NRTL equation can be used directly for the excess
Helmholtz free-energy expression, AE0 , in these two
equations.
For a solution of n components, NRTL equation is
n

GE
RT

j xjjiGji

i xi

(17)

k xkGki

with ij and Gij defined as

ji )

(12)

Since the equation of state parameters, a and b, are

pressure independent, these two parameters can be
canceled out from eqs 5 and 12 to give the interrelationship of AE between infinite pressure and zero
pressure as:

b* )

Incorporation of the NRTL Activity Model into

the New Mixing Rule

Aji
T

(18)

Gji ) exp(-Rjiji)

v*
1
0 + w
ln
w-u
v*
0 + u

the second virial coefficient boundary condition. The

most important aspect is that the mixing rule is density
dependent in an explicit form, which allows the mixing
rule to reproduce accurately the incorporated GE model.

(15)

(16)

As mentioned above,
in eqs 15 and 16 is the
excess Helmholtz energy at zero pressure evaluated
from a cubic equation of state using the mixing rules
for its avdw and bvdw parameters, as given in eqs 7 and
8. The zero-pressure volume v*0vdw ) v0vdw/b is obtained from eq 10 by substituting avdw and bvdw for the
a and b parameters.
There are some nice features of this new mixing rule.
The new mixing rule reduces to the van der Waals
mixing rule when AE0 is equal to AE0vdw. The mixing
rule satisfies the quadratic composition dependence of

(19)

The NRTL parameters, Aij, Aji, and Rij, obtained from

the lowest temperature of each binary reported in the
DECHEMA Chemistry Data Series, are used directly
in the mixing rule models. These values of Aij, Aji, and
Rij are then used in the mixing rules at all temperatures,
where temperature T is in Kelvin. The values of the
NRTL binary interaction parameters for the binaries
used in this study are given in Table 3.
In this work, we have considered 10 highly nonideal
binary mixtures which are traditionally described by
liquid activity models. They are listed in Table 3. In
order to obtain liquidlike values for v*0 at zero pressure
from eq 10, we have limited our analysis to systems with
components and temperatures such that a*/b* is larger
than the limiting value of 5.828 43 for the SRK and
6.828 43 for the PR equations of state. By doing this,
we have eliminated the need to include an extrapolation
methodology into our comparisons.
The mixing rule for b, as given by eq 15, forces the
mixing rule to satisfy the quadratic composition dependence of the second virial coefficient. Alternatively, the
conventional linear mixing rule could be chosen for b
(i.e., ignoring the second virial coefficient boundary
condition):

b)

i j xixj 2(bi + bj)

(20)

We will examine the capability of this mixing rule for

phase equilibrium prediction with and without the
second virial coefficient constraint on the b parameter.
We will also compare our new mixing rule with two of
the most successful and widely used mixing rules,
MHV1 (Michelsen, 1990b) and the mixing rule proposed
by Wong and Sandler (1992). Recently, Orbey and
Sandler (1997) have made a comparison of a number of
Huron-Vidal-type mixing rules, including MHV1 and

Ind. Eng. Chem. Res., Vol. 37, No. 5, 1998 1583

Table 3. NRTL Interaction Parameters and Results of the Prediction in Terms of Average Absolute Deviation
Percentage (AAD%) in Activity Coefficients, Bubble-Point Pressure, and K Values with Respect to PR (First Value) and
SRK (Second Value) Equations of State, Respectively
mixing rule

1, %

2, %

P, %

K1, %

K2, %

WS
MHV1
TCB
TCB(0)

Ethanol(1)/n-Heptane(2) from 30.12 to 70.02 C;a

I/2e/377, -379, I/2c/457, -458; A12 ) 521.746, A21 ) 727.003, R12 ) 0.4598 at 30.12 C
24.72/24.87
24.64/24.77
16.93/16.88
13.21/13.36
4.25/3.46
4.52/3.78
2.21/2.02
1.78/1.49
0.10/0.11
0.06/0.06
1.23/1.27
1.37/1.27
0.08/0.08
0.04/0.05
1.21/1.25
1.39/1.28

26.11/26.25
2.89/2.62
2.33/2.13
2.35/2.15

WS
MHV1
TCB
TCB(0)

Ethanol(1)/Water(2) from 24.99 to 120 C;a

I/1b/93, -106, -107, -108; A12 ) 13.3878, A21 ) 437.683, R12 ) 0.2945 at 24.99 C
18.13/18.04
16.37/16.29
13.24/13.26
15.99/15.87
6.24/5.70
4.55/4.14
4.01/4.15
6.02/6.17
0.26/0.25
0.17/0.17
4.19/4.33
7.17/7.19
0.21/0.20
0.14/0.13
4.05/4.19
7.04/7.06

14.91/14.84
4.17/4.23
5.30/5.36
5.16/5.22

WS
MHV1
TCB
TCB(0)

Methanol(1)/Cyclohexane(2) from 25 to 55 C;a

I/2a/242, I/2c/208, -209; A12 ) 644.886, A21 ) 784.966, R12 ) 0.4231 at 25 C
25.16/25.34
25.02/25.19
19.00/19.27
15.84/16.01
4.07/3.25
4.49/3.68
2.18/1.56
2.62/2.43
0.20/0.21
0.15/0.16
1.29/1.15
1.86/1.86
0.15/0.16
0.12/0.12
1.29/1.15
1.84/1.85

24.28/24.53
4.09/3.76
2.85/2.87
2.82/2.84

WS
MHV1
TCB
TCB(0)

Methanol(1)/Benzene(2) from 25 to 90 C;a

I/2c/188, I/2a/207, -210, -216, -217, -228; A12 ) 441.228, A21 ) 738.702, R12 ) 0.5139 at 25 C
20.69/20.87
19.77/19.93
13.63/14.01
13.87/14.10
3.71/3.03
3.45/2.86
4.17/3.49
7.02/6.62
0.17/0.17
0.13/0.13
3.20/3.00
5.89/5.88
0.13/0.13
0.10/0.10
3.16/2.96
5.89/5.87

15.44/15.64
4.87/4.57
3.99/4.00
3.96/3.97

WS
MHV1
TCB
TCB(0)

Acetone(1)/Benzene(2) from 25 to 45 C;
I/3+4/194, -203, -208; A12 ) -35.4443, A21 ) 193.289, R12 ) 0.3029 at 25 C
4.25/4.32
4.08/4.15
2.39/2.55
2.15/2.18
0.85/0.73
0.69/0.58
0.92/0.84
1.11/1.08
0.01/0.01
0.01/0.01
0.82/0.79
1.03/1.03
0.01/0.01
0.01/0.01
0.81/0.79
1.03/1.03

4.34/4.37
1.79/1.74
1.74/1.66
1.75/1.66

WS
MHV1
TCB
TCB(0)

Acetone(1)/Ethanol(2) from 32 to 48 C;a

I/2a/323, -324, -325; A12 ) 24.3880, A21 ) 224.395, R12 ) 0.3007 at 32 C
4.98/5.02
4.93/4.98
3.36/3.49
3.04/3.11
1.95/1.75
1.55/1.37
1.46/1.39
2.14/1.98
0.01/0.01
0.01/0.01
1.20/1.22
0.98/0.97
0.01/0.01
0.01/0.01
1.20/1.22
0.98/0.97

2.58/2.63
2.92/2.79
1.83/1.84
1.81/1.82

WS
MHV1
TCB
TCB(0)

Acetone(1)/Methanol(2) from 45 to 55 C;a

I/2a/75, -80, -81; A12 ) 31.5237, A21 ) 180.554, R12 ) 0.3004 at 45 C
7.95/7.93
7.71/7.71
4.20/4.47
5.21/5.13
0.92/0.74
0.62/0.45
0.51/0.59
1.00/0.97
0.00/0.00
0.00/0.45
0.49/0.62
0.92/0.87
0.00/0.00
0.00/0.00
0.49/0.63
0.92/0.87

6.13/6.20
0.82/0.86
0.84/0.86
0.84/0.86

WS
MHV1
TCB
TCB(0)

Ethanol(1)/Benzene(2) from 25 to 55 C;a

I/2a/398, -407, -415, -417, -418, -421, -422; A12 ) 115.954, A21 ) 584.473, R12 ) 0.2904 at 25 C
14.84/14.80
12.92/13.06
10.20/10.41
12.17/12.32
3.25/2.73
3.13/2.69
1.39/1.24
3.34/3.29
0.15/0.15
0.12/0.13
1.05/1.11
3.18/3.19
0.12/0.12
0.10/0.10
1.06/1.12
3.20/3.21

9.99/10.13
2.41/2.38
2.51/2.53
2.53/2.54

WS
MHV1
TCB
TCB(0)

Methanol(1)/Water(2) from 24.99 to 100 C;a

I/1b/29, I/1/41, -49, -72, -73; A12 ) -23.1150, A21 ) 188.147, R12 ) 0.3022 at 24.99 C
6.16/6.00
5.51/5.37
5.13/5.21
7.06/6.94
3.58/3.37
3.03/2.85
2.23/2.32
3.56/3.67
0.16/0.16
0.13/0.13
2.81/2.97
4.77/4.78
0.13/0.13
0.11/0.10
2.78/2.97
4.72/4.73

8.75/8.66
3.56/3.66
5.02/5.05
4.91/4.93

WS
MHV1
TCB
TCB(0)

Methanol(1)/n-Hexane(2) from 25 to 45 C;a

I/2c/219, I/2a/252; A12 ) 823.172, A21 ) 848.519, R12 ) 0.4388 at 25 C
27.63/27.73
27.69/27.77
25.54/25.52
33.88/33.94
3.33/2.47
5.06/4.12
3.62/3.13
2.41/1.99
0.12/0.12
0.05/0.05
1.73/1.82
1.26/1.23
0.09/0.10
0.04/0.04
1.67/1.76
1.28/1.23

31.51/31.54
2.19/1.85
1.13/1.13
1.16/1.13

Data taken from DECHEMA Chemistry Data Series by Gmehling, Onken, and Arlt; numbers corresponding to volume/part/page.

MHV2, for mixtures of molecules with large differences

in size. They conclude while there are differences
between the models, none was clearly superior to the
others for all mixtures, though MHV2 was the least
accurate. Since MHV2 is the least accurate model,
MHV1, instead of MHV2, is selected for the comparison
in this work.

Wong and Sandler assumed that the excess Helmholtz free energy at infinite pressure can be approximated by the excess Gibbs free energy at low pressures:

AE(T, x, P ) ) ) AE(T, x, P ) low) )

GE(T, x, P ) low) (21)

1584 Ind. Eng. Chem. Res., Vol. 37, No. 5, 1998

The Wong-Sandler approximation will be tested in

this comparison to see how well the assumption in eq
21 stands. Another objective in this paper is to test the
ability of different mixing rules to reproduce the incorporated GE model. In this work, we have performed
rigorous tests of the capability of reproducing the GE
model using the PR and SRK equations of state combined with our new mixing rules. We use WS to refer
to the Wong-Sandler mixing rules. TCB is used to
represent the mixing rule developed by us in this work
(eqs 15 and 16) and TCB(0) to eqs 16 and 20. The
zero in parentheses in TCB means no second virial
coefficient constraint. The accuracy of reproducing the
activity coefficients of component i, i, in terms of
average absolute deviation percentage (AAD%), from the
incorporated GE model using these different mixing
rules in the SRK and PR equations is given in Table 3.
Similarly, the accuracy of the VLE prediction from the
different mixing rules, expressed in terms of AAD% in
bubble-point pressure and K values of component 1 and
2, is also presented in Table 3 for the PR and SRK
equations.
The activity coefficient of component i in a mixture
can be derived from the fugacity coefficients of the
equation of state:

i )

i
0i

(22)

where i is the activity coefficient of component i at the

equilibrium temperature and pressure, i is the fugacity
coefficient of component i in the mixture, and 0i is the
fugacity coefficient of pure component i at the same
temperature and pressure.
Examining the accuracy of reproducing the activity
coefficients, as given in Table 3, the PR and SRK
equations of state produce almost identical results for
all four different mixing rules. Table 3 also compares
the VLE predictions of the cubic equations of state. The
comparisons show that there is little difference in the
accuracy of the predictions with these two methods.
Based on these results, it seems to indicate that both
equations of state are equivalent to each other and
neither one has an advantage over the other in phase
equilibrium calculations as long as the R correlation of
Twu et al. (1991) is used.
For the mixing rule comparison, the Wong-Sandler
mixing rule gives the largest deviation for all the
systems in this comparison. The inability to match the
GE derived from the equation of state with that from
the incorporated GE model invalidates the basic assumption behind the Wong-Sandler mixing rule. The
predictions from the Wong-Sandler mixing rule without
using any additional binary interaction parameters are
unacceptable. Table 3 contains the results for the
systems acetone-benzene and acetone-methanol. The
zero-pressure model, MHV1, reproduces closely the GE
model it is combined with for systems where the liquid
volume of the components is close to the fixed value
chosen by Michelsen (1990b). It is not surprising that
if the liquid volume of the system is not close to the fixed
constant, the GE model is not reproduced by the equation of state using the MHV1 mixing rule. These results
are in agreement with a statement by Kalospiro et al.
(1995), although we have a different explanation for it.
The results shown in Table 3 illustrate that our new
mixing rule reproduces the GE model almost exactly.

The errors in the reproduction of the activity coefficients

for these systems are minimal from our mixing rule.
For the VLE predictions, our new mixing rule gives
consistent results and, in general, provides good agreement between the experimental data and the predictions over a wide range of temperatures and pressures
using only the information in the GE model. It was
somewhat surprising that good agreement was also
obtained from MHV1, although it cannot reproduce
accurately the incorporated liquid activity coefficients.
This unexpected result from MHV1 might come from
some mutual cancellation of errors, but we do not know
at this time. Again, the worst predictions are obtained
from the Wong-Sandler mixing rule because of its
inability to match the GE model.
As we mentioned, we want to investigate the impact
on phase equilibrium prediction of the mixing rule with
and without the second virial coefficient condition
constraint. By reviewing the results shown in Table 3,
they show that our mixing rule yields almost identical
results either with or without the second virial coefficient condition constraint. This indicates that the
second virial coefficient constraint has no effect on the
phase equilibrium prediction. Theoretically, it would
be nice to have the mixing rule satisfy the quadratic
composition dependence of the second virial coefficient
boundary condition. Practically, it is simpler just to use
the conventional linear mixing rule for the b parameter.
The same quality of phase behavior will be predicted
from both cases.
Conclusion
We have successfully extended the Twu-Coon mixing
rule from infinite pressure to zero pressure. Incorporating a CEOS/AE mixing rule into either the PR or SRK
equation of state results in similar accuracy in the phase
equilibrium prediction. Therefore, as long as the same
CEOS/AE mixing rule is used, engineers can use either
PR or SRK in design calculations without worrying
about the discrepancies between the equation of state
methods. We also demonstrate that CEOS/AE models
such as the Wong-Sandler mixing rule do not reproduce
the GE models with which they are associated. We show
why the zero-pressure models do not reproduce exactly
the GE models at low pressure and reveal that approximate reproduction is feasible for MHV1 only for
systems with liquid volumes close to the assumed
constant volume. On the other hand, the new model
we developed in this work accurately reproduces the
activity coefficients of the GE model.
Nomenclature
a, b ) cubic equation of state parameters
a*, b* ) reduced parameters of a and b
A ) Helmholtz free energy
Aij, Aji ) NRTL binary interaction parameters, K
C1 ) infinite pressure function defined in eq 6
Cv0 ) zero pressure function defined in eq 13
G ) Gibbs free energy
Ki ) K value of component i defined as yi/xi
L, M, N ) parameters in the R function defined in eq 3
P ) pressure
R ) gas constant
T ) temperature
u, w ) cubic equation of state constants
v ) molar volume
v*0 ) reduced zero-pressure liquid volume

Ind. Eng. Chem. Res., Vol. 37, No. 5, 1998 1585

xi ) mole fraction of component i
Z ) compressibility factor
Greek Letters
R ) cubic equation of state R function defined in eq 2
i ) fugacity coefficient of component i in the mixture
i ) activity coefficient of component i
) departure function
Subscripts
0 ) zero pressure
) infinite pressure
c ) critical property
i, j ) property of component i, j
ij ) interaction property between components i and j
vdw ) van der Waals
Superscripts
0 ) pure component
* ) reduced property
E ) excess property

Literature Cited
Coutsikos, P.; Kalospiros, N. S.; Tassios, D. P. Capabilities and
Limitations of the Wong-Sandler Mixing Rules. Fluid Phase
Equilib. 1995, 108, 59-78.
Dahl, S.; Michelsen, M. L. High-pressure Vapor-liquid Equilibrium with a UNIFAC-based Equation of State. AIChE J. 1990,
36, 1829-1836.
Heidemann, R. A.; Kokal, S. L. Combined Excess Gibbs Energy
Models and Equations of State. Fluid Phase Equilib. 1990, 56,
17-37.

Huron, M. J.; Vidal, J. New Mixing Rules in Simple Equations of

State for Representing Vapor-Liquid Equilibria of Strongly
Nonideal Mixtures. Fluid Phase Equilib. 1979, 3, 255-271.
Kalospiros, N. S.; Tzouvaras, N.; Coutsikos, P.; Tassios, D. P.
Analysis of Zero-reference-pressure EOS/GE Models. AIChE J.
1995, 41, 928-937.
Michelsen, M. L. A Method for Incorporating Excess Gibbs Energy
Models in Equation of State. Fluid Phase Equilib. 1990a, 60,
47-58.
Michelsen, M. L. A Modified Huron-Vidal Mixing Rule for Cubic
Equations of State. Fluid Phase Equilibria 1990b, 60, 213219.
Orbey, H.; Sandler, S. L. A Comparison of Huron-Vidal type
Mixing Rules of Mixtures of Compounds with Large Size
Differences, and A New Mixing Rule. Fluid Phase Equilib. 1997,
132, 1-14.
Peng, D.-Y.; Robinson, D. B. A New Two-constant Equation of
State. Ind. Eng. Chem. Fundam. 1976, 15, 59-64.
Soave, G. Equilibrium Constants from a Modified Redlich-Kwong
Equation of State. Chem. Eng. Sci. 1972, 27, 1197-1203.
Twu, C. H.; Coon, J. E. CEOS/AE Mixing Rules Constrained by
the vdW Mixing Rule and the Second Virial Coefficient. AIChE
J. 1996, 42, 3212-3222.
Twu, C. H.; Bluck, D.; Cunningham, J. R.; Coon, J. E. A Cubic
Equation of State with a New Alpha Function and a New Mixing
Rule. Fluid Phase Equilib. 1991, 69, 33-50.
Wong, S. H.; Sandler, S. I. A Theoretically Correct Mixing Rule
for Cubic Equations of State. AIChE J. 1992, 38, 671-680.

Received for review September 10, 1997

Revised manuscript received November 12, 1997
Accepted November 12, 1997
IE9706424