Chemicd Engineering Science

Printed in Char Britain.

Vol. 39, No.

2, pp. 357-369,

1984

oce!%-2509/84

$3.00 + .cm
Ltd

Pergamon
Pms

IMPROVEMENT
OF THE VAN DER WAALS
EQUATION
OF STATE
G. SOAVE*
Special Process Dept., SNAMPROGETTI S.p.A., S. Donato
(Received 16 March

Milanesc, Italy

1983; accepred 14 June 1983)

Abstract-The
Van Der Waals equation of state has been modified in order to improve its accuracy. The
introduction of a temperature dependence of the attraction term allows reproducing accurately purecompound vapour pressures. The modification of the mixing rules, with the infinite-pressure excess-freeenergy term expressed by a NRTL-like model, allows reproducing accurately phase equilibria,of polar
and apolar systems. Finally, the introduction of a volume correction improves markedly the calculated
liquid densities, without changing the phase-equilibrium conditions. The improved equation is accurate
enough to be applied for design purposes.

INTRODUCTION

Among all types of equations of state, a most popular

class is formed by the so-called cubic equations of
state, so named because, although usually written in
pressure-explicit form:

In the crowd of equations proposed, the one which

perhaps received most attention in the last years is
that by Redlich and Kwong[2]:

p--T

v-b

P = P(T,

v-b

(2)

V)

when solved with respect to volume they give rise to a

cubic equation, which can be solved by fast and accurate methods, with no ambiguity in the choice of the
roots, so avoiding one of the main problems posed by
other, more complex equations of state.
The oldest, simplest and best known cubic equation
of state is that conceived by Van der Waals in the 19th
century [l]:

p=--_ RT

_c
v(v+b)

02'

In this equation, the first term on second member

expresses (in a somewhat simplified form) the effect of
repulsion between molecules, assimilated to hard
spheres, while the second term (theoretically sounder)
expresses the attraction forces due to the Van der
Waals forces. In spite of its simplicity, the VdW equation gives a correct description (at least qualitatively)
of the P-V-T
behaviour of substances, in the liquid
and gaseous states. Yet it is not accurate enough to be
suitable for design purposes.
With the development of electronic computers, the
equation-of-state route for the calculation of physical
properties and phase equilibria proved a powerful
tool, thus great impulse was given to the development
of new, accurate equations of state.
*Present address: ASSORENI, S. Donato Milanese,
Italy.

which can be considered a modification of the VdW

equation.
Curiously the change lies only in the form of its
attraction term (theoretically sounder), while the
inaccurate repulsive term has been maintained. Despite that, the R-K equation represents a great
improvement over the VdW one and has found wide
application for the treatment of gaseous fluids, while
still unapplicable to the Liquid state and phase equilibria.
Trying to develop an equation of state applicable
to either physical states, without neglecting the advantages
of simplicity,
many authors [3-61 started
from the R-K
quation,
usually assuming
a diff-

erent law of dependence on temperature of the

attraction term and forcing the agreement between
calculated and experimental pure-compound vapour
pressures. The improved forms of the R-K equation
can treat either physical states and some of them have
found wide application for design purposes. The real
reason why the R-K equation was so often chosen as
a starting point is not well clear; such equation has
less theoretical justification than, for instance, the
VdW equation and, although more accurate in its
original formulation, still has to be improved to
become acceptably accurate.
This paper wants then to give an example of how
a simple equation of state can be improved in order
to be applicable to both physical states (liquid and
gaseous), polar and apolar compounds and, by the
use of proper mixing rules, can describe accurately
357

358

G.

SAVE

the phase equilibria of binary and multicomponent

systems, with a minimum of empirical parameters.
The equation selected is that by Van der Waals; it is
the simplest form of a cubic equation of state and has
at least the same theoretical support as the more
celebrated R-K equation, so it can be a good starting
point.
PURE

COMPOUNDS

The first modification is to assume that the attraction parameter a is temperature dependent:
RT
p=---.
v-b

calculated on such basis is very similar to that obtained from the R-K equation [S] and, for apolar
compounds, can be expressed as well by an equation
of the type:
a(T,J

= (1 + m(1 - ,/T,&*

where m is typical of the compound being examined

and can be expressed as a function of the acentric
factor w:
m = 0.4998 + 1.5928~ - 0.19563~ * + 0.025~ 3

a(T)
vz

(for 0 < w 5 1).

Imposing, as usual, that the first and second derivatives of pressure on volume be zero in the critical
point, we obtain:
a, = a(Tc) = 918 RT,V< = 27f64R2Tc2/Pc
6 = l/3 V, = l/8 RTJP,

(4)

For polar compounds, the trend of a(T) i.e. a(T,)

deviates from that expressed by eqn (6); a more general
expression is required, containing at least two adaptable coefficients. Following the same approach proposed for the R-K equation[7] the expression proposed is then:

(5)
a(T,)=l+m(l-TT,)+n

whence:
Z, = P,v,/RT, = 318.
The values of a(T) at the other temperatures can be
derived from experimental vapour pressures, by imposing the equality of the fugacities calculated for the
liquid and gaseous states.
The trend with respect to the reduced temperature
of the correction coefficient:
a(T,)

= GVa(T,)

f-l
R

o-1 -

1 -

O-

1 -

ammonia

reduced

temperature

j
.3
Fig.

.4
1. Comparison

.5

.6

>

(7)

where m, n are empirical constants, to be derived by

correlation of experimental vapour pressures (or of
the corresponding values of a). Equation (7) gives a
good correlation of the function a(T,) for most compounds, polar and apolar (Fig. 1 shows the errors
between vapour pressures of ethane, water and ammoma, calculated with optimized values of m, n, and the
experimental values [8,9].
In particular if, in addition to the critical constants
T,, PC, the boiling temperatures at 10 and 760 mmHg

1 -

.l

(6)

.7

.a

.9

between calculated and experimental vapour pressures.

1.f

Improvement of the Van der Waals equation of state

are known, the corresponding values of m and n can be
calculated directly using (instead of the more complicated general procedure) the shortcut procedure described in Appendix A.
The critical constants T,, PC and the constants ml n
define completely the P-V-T
behaviour of a pure
compound. At given temperature T and pressure P,
posing:
Z = PVJRT

A = aP/RT

= craP{R2T2 = 27/64~PdT,~

(8)

B = bP/RT = 118 PJT,

(10)

eqn 3 can be written in the more practical form:

23-Zzyl

+B)+A(Z-B)=O.

where
a, = a,

a, = a,
and the cross terms a, (which are symmetrical, i.e.
a,- = ajJ are.obtained empirically.
Equation (15) is equivalent to eqn (13) in the case
where a, = (~q)*,~. For such reason. the cross terms a0
are usually written in the form:
a9 = (1 - 1y;,)(~p~)~~

(9)

(where a is calculated from eqn 7) and:

(Ill

Solution of eqn (I 1) can give one or three roots; in the

latter case, the valid root is the highest or the lowest
one, for a gaseous and a liquid phase, respectively.
From the calculated 2, one can calculate directly
(i.e. without iterative procedures) the density, the
pure-compound fugacity coefficient:

359

(16)

where Kg (usually called binary interaction constant) is zero for regular solutions. It is usually considered independent on temperature, so that only one
value is needed for each binary couple i -j.
With such implementation, the mixing rules can be
extended to systems containing slightly polar compounds_
In presence of strongly polar compounds such rules
are no more adequate. A higher degree of flexibility
must be given, for instance by an extension of the
linear law expressed by eqn (14) to a quadratic rule, and
the introduction of a second empirical binary constant:

kn(f/P) = Pu/RT - 1 - in(P(u - b)/RT)j - a/RTu

=Z-I-ln(Z-B)-_4/Z

(12)

and other physical properties, using the expressions

obtainable from eqn (3) by the usual thermodynamic
procedures.

where
b, = bi
b,- = b,
b, = b,X1 - h, Mb,+ b,)/2

MIXTURES

In order to extend the application of an equation of

state to mixtures, proper mixing rules (that is, eXpSsions which connect the mixture parameters to purecomponent parameters and mixture composition)
must be stated.
The mixing rules originally proposed for the VdW
and R-K equations, derived from theoretical considerations, are:
a.s = C xp,

(13)

b = c x,&

(14)

(where a, b are referred to the mixture and a,, bi to the

components). Such rules are adequate for regular
solutions, such as hydrocarbon mixtures. In presence
of polar compounds they must be improved by introducing empirical correction terms.
A most popular way is to modify eq. (13) by expressing the mixture parameter o with a quadratic rule:
a =Cx;rjag

(1%

The above rules, together with eqns (15) and (16), have
been shown [q to apply well to binary systems of polar
compounds. Unfortunately, as shown by several authors (for instance in [IO]), such rules, although theoretically well supported and completely adequate for
binary systems, yet fail when applied to multicomponent mixtures.
It is likely that more complicated rules, involving
ternary and higher order terms have to be considered,
but it is an impractical route, owing to the extremely
large number of terms and long computation times
involved. It is then necessary to adopt mixing rules of
a new type, which make use of binary terms only,
avoiding the introduction of higher order terms, and
can apply both to binary and to multicomponent systems.
A new very primising approach was suggested in a
recent paper by Vidal et al. [l 11.
Starting from the genera1 expression of the excess
Gibbs energy:

360

G.

?aOAVE

for the particular case of the VdW equation we have

(see eqn 12):
lnrp,=g-

1 -ln{~]-&

whence:

-cln((t&)}-(&

-xx,&).

accurate mixing rules). More generally, g: can be

expressed through one of the models already published.
The expression proposed here is derived from the
NRTL equation; it was selected for several reasons:
its relative simplicity; good behaviour for polar systems, even in presence of miscibility gaps; only binary
terms are considered;
and extension to multicomponent systems, using only binary constants,
derived from binary systems.

where

At infinite pressure:

Gji = exp( - agrjji).

limu =b

(21)

In the original NRTL method, tii and fj, are expressed

in the form:

lim uj = !+

gji - gii _
Zji=RT
-

whence:

s
RT

(23)

where the Ag terms are treated as constants.

In this work, the original expression for r was
modified, so that only (a,/RTb,) terms are involved in
eqn (19):
(17)
The condition

(24)

for a finite limit of gz is:

b =cxJ$

(25)
(18)

whence:

--$~Xi$&.$

(19)

This equation sets a connection between a/RTb and

the excess free energy at infinite pressure.
Once an expression for &RT
is assigned, the
mixing rules are completely defined.
The simplest case is that of an ideal solution at
infinite pressure:

where C, and C, are dimensionless constants.

With eqns (24) and (25) rii and tii are no more
inversely proportional to temperature, like in eqns
(22) and (23) due to the temperature dependence of
a; however, the difference is negligible and seems not
to affect the accuracy of results.
Like before, it is:
r# # rji
rii = Q = 0
hence:

g, E=Q

c~ *

whence:

(this is a fair approximation for hydrocarbon mixtures and a good starting point in the search for more

c,=

Equations (18H21),
rules proposed. For
the binary constants
It is now possible
fugacity coefficients

cji

cii=o.

(24) and (25) define the

each binary couple, they
C,, C,.
to derive the expression
of single compounds in

mixing
require
for the
a mix-

Improvement of the Van der Waals equation of state

ture,

starting from the general expression:

Such coefficients are used for the calculation of phase

equilibrium constants:

From eqn (3) and the mixing rules proposed above

we have:

361

gether with the results of correlations with Margules,

Van Laar, Wilson, NRTL, UNIQUAC equations.
First, the coefficients m, n (eqn. 7) of each component were derived by correlation of the vapour
pressures reported by [ 131, using the critical constants
taken from [14]:

acetone
chloroform
methanol

T&R)

P, @ia)

914.70
965.40
922.64

682
794
1174

0.86368
0.71142
1.34442

0.19672
0.18812
0.19739

The agreement between calculated and experimental

pressures is within 0.5%
Second, for each binary system a selection of the
experimental data reported by [13] was made, retaining only the data at a constant pressure of
760 mmHg.
The selected data were correlated, obtaining the
best values of the binary constants C,, C, (eqns.
24, 25) (the constants txiiwere assigned a fixed value
0.3, as their influence was found to be secondary):

In

(26)
acetone-

chloroform
aceton*
methanol
chlorofornmethanol

B=bP
RT
Now, according to eq. (19):

(27)
And according to eqn 20:

xkTk,Gkj
2.

j.

source

0.02739

0.3

[13, 151

0.02438

0.03537

0.3

[13, 161

0.14000

0.00403

0.3

[13, 171

(28)

As a particular case, for a pure compound, In r? = 0

and eqn (26) reduces down to eqn (12).
As an example, the binary and ternary mixtures
formed by acetone, chloroform and methanol were
considered.
The experimental data are reported in [13], to-

-0.07039

Figures 2-4 show the curves of the boiling tcmperatures at 1 atm and relative volatilities, towards
liquid compo&tion.
The agreement between calculated and experimental values is very -good, comparable to the other
methods used by [ 131.
Third, the pure-compound and binary constants
obtained were used to calculate the boiling temperatures at 1 atm of ternary mixtures, and the results
compared with the experimental data [ 181 reported by
[131.

C,

Gj

having assumed, as for pure compounds:

Surprisingly, the agreement was poorer than with

the binary systems.
A more accurate examination of the results led to
the following considerations:
-the experimental data look dispersed and little
accurate
-the values calculated with the NRTL method,
reported by [13], show lower errors, although still
dispersed. However, the reported results were calculated using parameters which are different from those
obtained from the binary systems, and clearly fitted
to the ternary data
-the NRTL parameters reported for the ternary
systems, when applied to the binaries given poor
results.

362
2

G.

SOAVE

rel. volatility
YP
KP
1

._------

.6 _

6-

0
-

ex~~runental
calculated

(13)(15)

64 -

63

62

61 .
60.
19 L
66 _

Fig. 4. Chloroform/methanol

.5
Fig. 2. Acetone/chloroform

-the
NRTL

1
at 760 mm Hg

.5

67.

at 760 mm Hg.

racy to the classical NRTL method, for the calcuof VLE equilibria.
Such approach is probably applicable to any equation of state, on condition that it can reproduce
lation

calculations
of ternary equilibria, using the
parameters
obtained from the binary systems,

show the same degree of inaccuracy of the proposed

method and can be almost exactly superimposed to
them.
Table 1 and Fig. 5 show the results for the ternary
mixtures with 10% mol. fr. of acetones in the liquid,
for the various methods.
The conclusion then is that the modified Van der
Waals EOS, making use of a NRTL-type free energy
excess term, can be substituted without loss of aceu-

correctly

the pure-compound

phase

equilibrium

mol. fr. ct~loroform in the liquid

mol. fr. acPtone

In the liquid

.5
Fig. 3. Acetone/methanol

I
1

at 760mmHg.

p+%___j

be-

haviour.
This makes it possible to exploit the main advantage of the EOS route over the activity-coefficient
route, that is the possibility of treating systems
containing supercritical compounds.
With the classical methods, this can be done only

.5

X2

,
1.

Fig. 5. Boiling temperatures of ternary mixtures of acetone,

chloroform, methanol, with 10% acetone in the liquid: 0,
experimental [13, 181; l , experimental [13, 15, 161; -, calculated, NRTL and modif. VdW with constants from binary
data; --, calculated, NRTL with binary constants adapted to
ternary data [13].

fiotes:

59.8

56.5

56.0

56.0

56.0

57.0

.770

.I40

.I

(3) calculated

(1) calculated

.090

.8

-.C611
.1002

.7089
.2402

.0323

-.0391

.osog

56.72

55.41

55.13

1.43

-3.08

-1.03

-0.87

.2436

.7053

.0511

.0330

.1036

-.0647

-.0389

.0417

-.a128
i3167

-.0289
-6322

.0183

.3583
.0511

.OlSti

-5956

using

values

of the binary

parameters

adjusted

to

the

ternary data (13)(18)

iri (1);

1.81

0.24

2.52

2.46

(2) like

60.04

59.02

58.46

using the vaIues of the binary constants derived far two - component systems.

l.l.6

-3.07

.3!53

. 275

.2

.I

-.0106

.6344
.C403

-.0296

.0504

-1.15

.0189

.3589

.645

.0160

-.0349

.5G60

.0551

.OdO

.340

.3

-0.96

.ci9q

.OG41

.0015

-.0056

-.C2GE

.018R

.OOlB

.0030

.OC56

-.@087

see (:3)(16)(17).

.1441

.7715

.0844

.2544

.6638

.C81a

43430

.5756

.0813

.034R

.4178
.0130

.3960

-.0339

~0308
.4992

.0561

-.co40
.063C

.0092

.5392

2.24

-.0222

.0648

-3636 -.0234
.5380 .0080
.3984 .0154

.7

.570

-0.85

.l

.090

,383

.4

.6

,530

.5

.l

.C87

.I

58.24

.4845

.0079

.4399

.0117

.4437

.432

.5
-0.76

.C525

.4204

55.24

.0092
-.06!6

.0872

-.0071

2.31

-.0007

-0.34
.4829

55.66

58.31

. C688

.0773

-.ooq7

-.0020

.4803

.0760

.490

-0.41

m.0653
.5488

SO160

.4960

.0206

.5OOG

.078

.6

1.65

-.0006

58.65

.OS4rA

-.0153

.4097

.3567

-*0007

-0943

-0.57

.6183

-.0196

.5704

-.0188

.0293

.2753

-.OOL8

.W64
..0347

.1064

-0022

.7057

(1682

.I261

.0175

0.63

dt

.3122

.1175

.4062

56.43

51.64

-.0443

.0932

.4

.480

.3

-0.62

.1

.425

.5749

.510

.7

.095

-.0151

.3081,

.l

.0167
-.0016

.I167

.3lG

.2

-1.10

.lOD

.1

-1.06

.8

50.7

-6707

-.0410

-6740

.715

.OlOS

.1805

.170

.0338

.1488

.l

59.36

.0075

-1.69

-1.64

!)
Y

NRTL
dt

.0336

.1775

dy

.I486

.115

dt

-r

.l

modified V1W (II)

61.0

ii-

t(c)

r experiment.

Table 1. Bubble point calculations for ternary mixtures at 760 mmHg (x, = 0.1) (1 = acetone,
2 = chloroform, 3 = methanol)

Improvement of the Van der Waals equation of state

365

With the R-K equation, the initial slope of

BPcIRTc = -0.3408, much better, and the agreement with the experimental curve is good up to
almost the critical point, where the error between
calculated and experimental Z moves upwards and
ramains on the high side by a small, constant value.
A great improvement over the VdW equation is
obtained by introducing the volume correction described above. With a value cPc/RTc = l/24 =
0.04167 corresponding to a critical compressibility
Z,= l/3 (and much lower than that obtained from
the critical density) we have:
-an
initial
BPc/RTc = - 19/44slope
cPc/RTc = -0.3385, comparable
to the experimental value
-a good fit (comparable to that given by the R-K
equation) up to almost the critical pressure
-Z values slightly in excess at supercritical pressures (like with the R-K equation).
Finally,
using
the
value
cPcfRTc = 318Zc = 0.0928 obtained from the critical density, the

The values of (cPc/RTc) so obtained, when plotted

against
reduced temperature show similar trends,

starting with almost constant values at the lowest

temperatures, then raising more and more steeply, up
to the critical value (3/S - Zc). (Figure 6 shows the
curves obtained for ethane, toluene and water).
However, the critical value of (cPc/RTc) so obtamed, although corresponding to the critical density,
is not adequate to reproduce the whole critical isotherm.
Figure 7 shows the Z vs PR curves for ethane, both
from [S] and calculated with the R-K equation and
the VdW equation (both original and modified).
The original VdW equation (regardless of which
law a(T) is assumed) gives too high values of Z over
all the pressure range; in particular the initial slope,
given by the value of the reduced second virial
coefficient BPc/RTc, is - 19164 = - 0.2969 (in comparison with the experimental value -0.3470)
and
the critical compressibility is 3/8 = 0.375 (experimental 0.2822).

-10

I/

.05

ethane
reduced

.3

.4

Fig. 6. Volume corrections

.6

.7

temperature

.8

from saturated liquid densities.

- experimental (8)
--R-K
equation
VdW equation

.3 .2

i-educed pressure
1

Fig. 7. Critical isotherm of ethane.

.9

366

G.S~AVE

temperature,

K
I

1
200

100

3ou

Fig. 8. Density of saturated liquid ethane.

mole
1..

50

60

70

96 Ha
*

100

Fig. 9. Hydrogen+rbon
monoxide at 70%
a, experimental[l9]; -,
calculated; C,, = 0.15809,
C,, = 0.04739, a = 0.3; ----, calculated by SRK method using true critical constants and acentric factor
for hydrogen and C,2 = 0.

Improvement of the Van der Waals equation of state

resulting corrections over the VdW equation look
clearly excessive: over all the pressure range (excepting the critical point) the Z curve runs on the low side
(even in the immediate vicinity of the critical point,
where all other curves are inaccurate by excess) with
errors of magnitude comparable to those from the
original equation.
In conclusion, the value of rPc/RTc
obtained
from the critical density is too high and must be
substituted with a lower value, in order to give a good
reproduction of the critical isotherm.
The best value obtained in such a way is very
similar to the values (almost constant) obtained from
the liquid densities at low temperatures. It looks then
logical to assume a constant value of c over all the
temperature range, tolerating some degree of error at
the temperatures nearest to the critical one.
One reference point could be the atmospheric
boiling temperature. The value of cPcfRT at such
temperature can be calculated directly according to
the procedure outlined in Appendix B and used at all
temperatures.
Figure 8 shows the saturated liquid density curve
of ethane so calculated in comparison with the experimental values. The agreement looks very good up to
a reduced temperature of about 0.7.
No comparison is shown for water, due to the
anomalous trend of its density towards temperature.
(b) Mixtures
Following the same approach described before, for
the modified VdW equation (eqn 29) the expression
for the excess Gibbs energy at infinite pressure is now:

The conditions

for a finite limit to g, are then:

367

where A, B, C are defined by eqns (1 l), (12) and (34)

and the term within square brackets, by eqns (27) and
(28); Z is obtained by solution of eqn (33).
For the particular ease of a pure compound, eqn
(38) is reduced to eqn (35). As shown before the value
of (Z + C) from eqn (33) is not influenced by the
value of C (i.e. it is equal to the value of Z obtained
from the original VdW equation). Because of the last
term in eqn (38), the calculated fugacity coefficients
depend on the assumed cls.
In the calculation of V-L-E
K values such
differences cancel each other. Thus, for phase equilibrium calculations (not for the calculation of singlephase fugacity coefficients), the K values can be
derived assuming zero values for the component
volume corrections, that is using eqns (13) and (26).
Only for density calculations have the volume
corrections to be considered. In such cases, the Z
values obtained by eqn (13) will be treated as
(Z + C); then subtracting C (eqns 34 and 37) the
correct Z is obtained.
CONCLUSION

The results of calculations show that even a simple

equation of state like the one by Van der Waals can be
modified in order to become applicable for design
purposes. In particular:
-with the introduction of proper law of variation
of the attractive term with temperature, it can reproduce accurately the vapour pressures of pure compounds, polar and apolar
-with a modification of the original mixing rules,
according to Vidal (1 l), it can reproduce accurately
the phase equilibria of mixtures
-with
the introduction of a volume correction,
according to the approach proposed by Peneloux
[12]), it can reproduce liquid densities (limitedly to the
lowest temperature range) without changing the equilibrium conditions.
The suggested approach can be applied to other
similar equations of state and to different models for
the free-energy term in the mixing rules.

b =Xx&+
NOTATION

c =

1 x,ci

(37)

whence again:

(19)
Assuming, as before, for gfJRT a NRTL-like model
expressed by eqns (22), (23) and (25) the new expression for the fugacity coefficients is:
In j$
(

= -ln(Z+C+B)+:(Z+C-1)

-($

TES 01. 39. No. 2-K

+~~)z~,~&-~C

(38)

pressure, bar
absolute temperature, K
molal specific volume, m3/Kg mole
V
R
universal gas constant, 0.0831434 bar m3/kg
mole/K
a, b equation of state parameters
Z compressibility factor PV/RT, adimensional
0: correction factor n (T)/a (Tc ), adimensional
% binary constant (eqn 21)
m, n coefficients in the equations for (eqns 6 and 7),
adimensional
A
aP/R2T2, adimensional
B bP/RT, adimensional
fugacity, bar
f
mote fraction in the liquid phase, adimenxi
sional
P
T

G.

%AVE

mole fraction in the gaseous phase, adimensional

no. of kgmoles of ith component
equilibrium constant Y/X
binary constant (eqn 16)
excess free energy, kJ/kgmole
fugacity coefficient ffP (pure compoundr)
or
jJPX,
(mixture components), adimensional
terms in the NRTL equation, adimensional
binary constant for talc. of zii (eqn 24),
adimensional
infinite-pressure activity coefficient, adimensional

APPENDIX

n from the boiling temperatures

Shortcut calculation of m,
10 and 760 mm Hg

at

For any given couple of values T, = T/T,, PR = P/P, (P

being the vapour pressure at the temperature T), a value of

a = a(T)/a(T,)
can be derived, which satisfies the equilibrium condttrons.
The values of a can then be correlated towards TR, P,; the
expression obtained for P = 10 and 76OmmHg are:

qM

E,T&,W:,

+ 1

i==*j=*

where:
Subscripts

.
1,.I
R

00

component index
reduced value (i.e. divided by the critical value)
at infinite pressure

critical

value

vapour

phase

liquid phase

w,O=;log(~)/(l-&)-

w_=;log(7~~Hg)/(l

-+-)-

1.

The values of W are very close to the values of the acentric

factor and A,, 4 are given in Table Al.
The above equations am applicable for the ranges
W=Otl,

PC=

100+600psia.

Once the values of (x,,,, c+, have been calculated, the

constants m, n can be obtained by solving the system:
[l]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[IO]
[I l]
[I21

[13]

[ 141
[15]
[16]
[17j
[18]
[19]

REFERENCES
Van der Waals, Doctorat Dissertation, Leiden 1873.
Redlich and Kwong, Chem. Rar. 44, 233 (1949).
Horvath, Chem. Engng Sci. 1974 29 1334.
Wilson, A&an. Criogenic Engng 1964 9 168; 1966 11
392.
Soave G., Cheer. Engng Sci. 1972 27 1197.
Barnes, Ph.D. Thesis, University of California (1973).
Soave G., I. C&m. E. Symp. Ser. 1979 561.2/l.
Nar. Bur. Std., Tech. Note 684 (1976).
Keenan J. H. er al., Steam Tables. Wiley, New York
1977.
Leland T. W., 2nd Innt. Conf. on Phase EugiIIbria and
FluidProperties
in the Chem. Ind., Paper 5. West Berlin,
1980.
Huron M. J. and Vidal H., Fluid Phase Equil. 1979 3
255.
Peneloux A. and Rauzy E., Fluid Phase Equil. 1982 8 7;
Rauzy E., Peneloux A. and Vidal J.; 7th Intern. Congr.
Chem. Engts., Paper Dl.29. CHISA, Prague, Sept.
1981.
Gmehling J. and Gnken U., VLE Data Coil. DECHEMA, Chem. Data Series. Vol. 1, Parts 3 + 4, page
92 (acetone-chloroform);
Vol. 1, Parts Za, page 85
(acetone-methanol);
Vol.
1, Parts 2a, page 19
(chloroform-methanol);
Vol. 1, Parts 2a, page 609 (ternary system).
Kudchadker el al., The Critical Constants of Organic
Substances (1968).
Kojima K. et al., k&&u
Kogaku 1968 32 337.
Uchida S. et al., Kagaku Kikai Chem. Engng 1953 17
191.
Nagata I., Chem. Engng Darn 1962 7 367.
Shishkin K. and Kotsyuba A. A., Tr. Dnepropetr. Khim.
Tekhn. In.w 1955 4 18.
Tsang C. Y. and Street W. B., Fluid Pha.w Equil. 1981
6 261.

M(t - TR.,o) + n(J/Ta,, - 1) = n,,, - l

m(l - TR.760)+ n(llTa76O - 1) = Iz,a - 1.
Table Al.
1

.60%3743

0.7090731

1.7469772

1.586825

.125414

.099942

. 529270

.a793134

-l.l48**

2
2

Bi i

*i i

- I.548328
.217587

0.28652

-.

.1558999

.I38438
43896
.I7537

-.

.203435
119137

APPENDIX B
the atmospheric boiling
temperature
For any given couple of values Te = T/T,, Pa = PIP, (T
being the atmospheric boiling point and P the atmospheric
pressure) a value of GL= a(T)/a(T,)
can be derived, which
satisfies the equilibrium condition of equal fugacities in
liquid and vapour phases:
Such value of (r (which can be calculated directly according to Appendix A) corresponds to defined values of
(Z,., + C) which can be correlated towards I,, Pa. In
partrcular, in this case the value of (2, + C) T,IP, (which
is independent on the value of C = cPc/RTc)
can be

Short+ut calculation of cP,/RTJrom

Improvement

369

of the Van der Waals equation of state

Table Bl.

expressed through the equation:

TR = TJT,
1

and the coefficients A, are listed in Table Bl.

Finally, from the experimental density of the saturated
liquid:

.1681523

.1110119

.I638897

-.1435373

-.3788939

-.3485x5

.1923412

.2561318

.191974