ChcmicaI l?
ngiwer~
Sc&enc~. Vol. 41, No.
Printed in Gmt
Britain.
5. pp.
ooo9-2%9/&i
Peqamon
1303-1309.1986.
s3.00+0.00
Press Ltd.
VAN DER WAALS MIXING
RULES FOR CUBIC EQUATIONS
OF STATE, APPLICATIONS
FOR SUPERCRITICAL
FLUID
EXTRACTION
MODELLING
T. Y. KWAK and G. A. MANSOORI
Departmentof Chemical Engineering,Universityof Illinois, Box 4348, Chicago, IL 60680, U.S.A.
(Received11 January 1985)
Abstract-A new concept for the developmentof mixingrulesfor cubic equationsof state consistentwith
statistical-me&anicaltheory of the van der Wads mixing rules is [Link] of this concept is
illustratedby its applicationto the Redlich-Kwong (RK) and the Peng-Robinson (PR) equationsof state.
The resultingmixing rules for the Redlich-Kwong and the Peng-Robinson equationsof state are tested
throughpredictionof solubilityof heavy solids in supercritical [Link] is shown that the new mixing rules
can pralict supercriticalsolubilitiesmore accuratelythan the originalmixing rules of the Redlich-Kwong
and Peng-Robinson equations of state.
INTRODUCTION
There has been extensive progress made in recent years
in research towards the development of analytic
statistical-mechanical equations of state applicable for
process design calculations (Alem and Mansoori,
1984). However, cubicequations of state are still widely
used in chemical engineering practice for the calculation and prediction of properties of tluids and fluid
mixtures (Renon, 1983). These equations of state are
generally modifications of the van der Waais equation
of state,
PC-!!&;
(1)
which was proposed by van der Waals in 1873.
According to van der Waals, the extension of this
equation of state to mixtures requires replacement of (I
and 6 by the following composition-dependent
expressions:
(2)
(3)
Equations (2) and (3) are called the van der Waals
mixing rules. In these equations, Uij and 6ii (i = j) are
parameters corresponding to pure component (i) while
ai] and b, (i # j) are called the unlike-interaction
parameters. It is customary to relate the unlikeinteraction parameters to the pure-component parameters by the following expression:
Ut/ = (1
-kij)(U*iUjj)
b, = (hi + b,, )/2-
pE--
p,RT_
v-b
a(T)
T&&).
n(T)
v(u+b)+b(v--b)
= u(Tc){l
+~(l
-Tf))
a( T,) = 0.45724 Rfz T,z/P,
lc = 0.37464 + ls4226uJ - 0.26992402
(5)
(3.1)
RT
v-b
05)
This equation of state has found widespread applications in chemical engineering calculations. We
would also like to mention that there exist a large
number of more sophisticated empirical versions of the
van der Waals and Redlich-Kwong modifications. Of
this category of equations of state, the Peng-Robinson
equation of state (Peng and Robinson, 1976)
(4)
In eq. (4), kt, is a fitting parameter which is known as
the coupling parameter. With eq. (5) substituted in eq.
(3), the expression for b will reduce to the following
one-summation form:
b = $xtbti.
I
Extensive research on equations of state have indicated that the van der Waals equation of state is not
accurate enough for the prediction of properties of
compressed gases and liquids (Rowlinson
and
Swinton, 1982). This deficiency of the van der Waals
equation of state has initiated a great deal of research
on the development of other equations of state,
through the use of principles of statistical mechanics or
by empirical or semi-empirical means. The majority of
equations of state used in chemical engineering practice are of the second category and are mostly cubic in
the volume, like the van der Waals equation of state
itself, but contain other forms of the temperature and
volume dependencies. One such equation of state
which has a simple form is the Redlich-Kwong
equation of state (Redlich and Kwong, 1949):
b = 0.0778 RT,/P,
(7)
(8)
(8.1)
(9)
(10)
is
widely
used for
thermodynamic
property
calculations.
While there has been extensive activity in the
development of new and more accurate empirical cubic
equations of state, there has been little attention to the
fact that the van der Waals mixing rules are used
1303
T. Y. KWAX and G. A. -RI
1304
erroneously in these equations, eqs (2) and (3). In other
words, for the extension of applicability of a cubic
equation of state it is not generally correct to use eqs (2)
and (3.1) as the mixing rules without attention to the
algebraic form of the equation of state. In the present
paper we introduce a new concept for the development
of mixing rules for cubic equations of state in the spirit
of the van der Waals mixing rules. This concept is
based on statistical-mechanical arguments and the fact
that the van der Waals mixing rules are for constants of
an equation of state and not for any thermodynamic
state function which may appear in an equation
of state. The resulting mixing rules for the
Redlich-Kwong equation of state, which is a simple
modification of the van der Waals equation of state,
and for the Peng-Robinson equation of state, which is
an advanced modification of the van der Waals
equation of state, are tested through prediction of
solubility of heavy solids in supcrcritical fluids.
THEORY
OF THE
VAN
DER
WAAIS
MIXING
RULES
Leland and co-workers (1968a, b, 1969) were able to
re-derive the van der Waals mixing rules with the use of
statistical-mechanical theory of radial distribution
functions. According to these investigators, for a fluid
mixture with a pair intermolecular potential energy
function between molecules of the mixture in the form
aij (r) = sil f
(rlaiJ
(11)
the following mixing rules will be derived:
U3 = 2
I
II
&U3 = C
i
$JCixje$
(12)
i
n
C
XiXjQjU$.
(13)
In these equations, su is the interaction energy parameter between molecules i and j and eij is the
intermolecular interaction distance between the
two molecules. CoeEicients a and b of the van der
Waals equation of state are proportional to s and u
according to the following expressions:
a = 1_1250RT,v,
a No&u3
b = 0.3333 D, a Nou3.
(14)
(15)
We can see that eqs (12) and (13) are identical with
eqs [Z] and [3], respectively. Statistical-mechanical
arguments used in deriving eqs ( 12) and (13) dictate the
following guidelines in using the van der Waals mixing
rules:
(1) The van der Waals mixing rules are for constants of
an equation of state.
(2) Equation (12) is a mixing rule for the molecular
volume, and eq. (13) is a mixing rule for (molecular
volume) x (molecular energy). It happens that b
and a of the van der Waals equation of state are
proportional to molecular volume and (molecular
volume) x (molecular energy), respectively, and, as
a result, these mixing rules are used in the form
which was originally proposed by van der Waals.
(3) Knowing that uij (for i fj),
the unlike-interaction
diameter, for spherical molecules, is equal to
Uij
(Uii
This gives the following
spherical molecules:
b,
Ujj)/2-
(16)
expression for bij of
= [ (bi::3 + bjj3)/2]
(17)
Then for non-spherical molecules the expression
for b, will be
b, = (1 -&j)[(bii3+bf/3)]3.
(18)
With the use of these guidelines, we now derive the
van der Waals mixing rules for the two representative
equations of state. A similar procedure can be used for
deriving the van der Waals mixing rules for other
equations of state.
Mixing rules for the Redlich-Kwong
equation of state
The Redlich-Kwong equation of state, eq. (6), can
be written in the following form:
z_pv_
RT
v-b
a
(19)
RT.(v+b)
In this equation of state, b has the dimension of a molar
volume,
b = 0.26 11,a Nou3.
Then the mixing rule for b will be the same as that for
the first van der Waals mixing rule, eq. (3). However,
the mixing rule for a will be different from the second
van der Waals mixing rule. Parameter a appearing in
the Redlich-Kwong equation of state has the dimensions of R-l/
(molecular energy)3/2 (molecular
volume), that is,
a = 1.2828RTC~5v, CC NO(e/k).5u3_
As a result, the second van der Waals mixing rule, eq.
(13), cannot be used directly for the a parameter of the
Redlich-Kwong
equation of state. However, since
(R2ub/2)
has the dimensions of (molecular energy)
x (molecular volume), the second van der Waals
mixing rule, eq. (13), can bc written for this term.
Finally, the van der Waals mixing rules for the
Redlich-Kwong
equation of state will be in the
following form:
a
112
5 xixj&~3b,!/3
i
1
>
(20)
b = g 5 xixj b,
t i
These mixing rules, when combined with the
Redlich-Kwong equation of state, will constitute the
Red&h-Kwong
equation of state for mixtures, consistent with the statistical-mechanical basis of the van
der Waals mixing rules.
Mixing
rules for the Peng-Robinson
equation
of state
In order to separate thermodynamic variables from
Van der Waals mixing rules
constants of the Peng-Robinson equation of state, we
will insert eqs (8) and (9) into eq_ (7) and we will write it
in the following form:
a/RT+d
z=-Y--
u-b
- 2 ,/(ad/RT)
(u + 6) + (b/u)@
(21)
- 6)
where a = a( T,) (1 + rc)* and d = a( T,)h-*/RT,_
This form of the Peng-Robinson equation of state
suggests that there exist three independent constants in
this equation of state: a, b and d. Now, following the
prescribed guidelines for the van der Waals mixing
rules, the mixing rules for a, b and d of the
Peng-Robinson equation of state will be
a = 2
5 xixjaij
1 j
b = 2
In order to calculate solubility from eq. (29), we need
to choose an expression for the fugacity coefficient.
Generally, for calculation of a fugacity coefficient an
equation of state with appropriate mixing rules is used
in the following expression (Prausnitz, 1969):
.RTln&=
2 x,xjdij
(24)
(1 - kij)(aiiajj)
(25)
6,
= (1 - lij) { (b:i3 + bjj3)/2}
(261
d,
= (1 -n~~)((di~+d~_/~)/2}~.
(27)
These mixing rules, when joined
with the
Peng-Robinson
equation of state, eq. (7), will constitute the Peng-Robinson
equation of state of a
mixture, consistent with the statistical-mechanical
basis of the van der Waals mixing rules.
FOR
EXTRACTION
SUPERCRITICAL
(W*/P)
(l/+2MY
(PFt/P)
(1/4,)exp
(30)
{(-b+2Zx,b,)/(u-6)
exp
(@/RT)
{@*<P
- PF)/RT}.
(31)
In & = In (u/(u - 6)) + (2zxj 6, - b)/(u - 6) - In Z
+a((2Zxjbi,
dP
(29)
(In ((u + 6)/u)
- b)/(b*RT.))
- b/(u + 6)) -
((2Zx,a,,)/(bRT
))
x In ((u + 6)/u).
(32)
with the use of the correct version of the van
der Waals mixing rules, eqs (3) and (20), in the
Redlich-Kwong equation of state, the fugacity coefficient will assume the following form:
Now
6)) + (2Xx,
6, - b)/(u - 6) - In Z
+ (a(2Exj 6, - b)/(b*RT
where +I is the fugacity coefficient of the condensed
phase at the saturation pressure Pi, and C& is the
vapour phase fugacity at pressure P. Provided we
assume that uys* is independent of pressure, the above
expression is converted to the following form:
=
RTln
For the Redlich-Kwong equation of state, with eqs
(2) and (3) as its mixing rules (as customary in the
literature), the following expression for the fugacity
coefficient is obtained:
In&i = In (Y/(u
FLUID
(28)
Y,
C(dPldni)r.,.,,,~-_(RTIY)ldV
where 2 = Pu/RT.
In the case of the van der Waals equation of state, eq.
(l), with eqs (2) and (3) as the mixing rules, the
following expression for the fugacity coetlicient of a
solute in a supercritical gas will be derived:
MODELLING
A serious test of mixture equations of state is shown
to be their application for the prediction of solubility
of heavy solutes in supercritical fluids (Mansoori and
Ely, 1985). In the present paper, we apply the van der
Waals, the Redlich-Kwong and the Peng-Robinson
equations of state for supercritical fluid extraction of
solids and study the effect of choosing different mixing
rules for predicting the solubility of solids in supercritical fluids.
The solubility of a condensed phase, y,, in a vapour
phase at supercritical conditions can be defined as:
Y2
-22Zxjaii/uRT}.
with the following interaction parameters:
APPLICATION
& = RT/(u-b)(l/P)exp
k xixjbij
1 i
aij =
I
-
1 j
d = 5
1305
- b/(u + 6)) -
I.)) (In ((u + 6)/u)
((3~ 2(Zxjaf_,~3b~,~3)/bf2
- a2j3 (Zxj 6, )/b3/*)/(bRT
.5))
x In ((u +6)/u)
(33)
where
The fugacity coefficient based on the PengRobinson equation of state with eqs (2) and (3) as the
mixing rules (as usually derived) has the following
form (Kurnik et al., 1981):
In& = ((2Zxjbij-b)/b)(Z-1)sln(Z-B)
-
(A/2
J2B)
(2Xxjaii/a
x {ln((Z+2J2B)/(Z-2J2B))]
(2Xx1 6,
-6)/b)
(34)
* and B = bP/RT.
where A = aP/RT
With the use of the correct version of the van der
Waals mixing rules, eqs (22~(24),
in the PengRobinson equation of state, the following expression
T. Y. KWAK and G. A. MANSOOR~
1306
for the fugacity coefficient is derived:
In 4, = ((2Xx,&
- b)/b)(Z-
- (A/(2 ,/2B))
l)-ln(Z-B)
((2Xx,4,
+ 2RTZx,
d,,
-2,/(RT)(aEc,dij
+ dZxiuii)/J(ad))/~*
x (ln ((Z +
- (2Z~jb,
(1 + J2)B)l(Z
- b)/b)
(1 - J2)g)))
(35)
where
a* = a+RTd
-2J(adRT)
A = a*P/(RT)
B = bRT/(RT).
We now utilize the above expressions for fugacities
in order to predict the solubility of solids in supercritical gases at different temperatures and pressures,
and compare the results with the experimental data.
RESULTS
AND
DISCUSSION
In Fig. 1, the solubility of 2,3-dimethyl naphthalene
(DMN) in supercritical carbon dioxide is plotted vs.
pressure at 308 K along with predictions obtained
from the van der Waals equation of state. According to
this figure, predictions by the van der Waals equation
of state will improve when eq. (3), along with the
combining rule, eq. (17), is used as the mixing rule for b
instead of eq. (3.1) as customarily used. This comparison, and similar comparisons reported elsewhere
(Mansoori and Ely, 1985) for other solute-solvent
systems, establish the superiority of the doublesummation mixing rule, eq. (3), for b over the singlesummation expression, eq. (3.1).
In Fig. 2, the same experimental solubility data as in
Fig. I are compared with predictions using the
Redlich-Kwong equation of state. According to this
figure, the corrected van der Waals mixing rules for the
Red&h-Kwong
equation of state, eqs (3) and (20), are
clearly superior to the mixing rules which are customarily used, eqs (2) and (3.1), for this equation of
state. Similar observations are made for the prediction
of solubilities of other solids in supercritical fluids,
which are not reported here.
The Peng-Robinson
equation of state with its
customary mixing rules, eqs (2) and (3), is widely used
for predicting the solubility of heavy solutes in supercritical gases and for petroleum reservoir fluid-phase
equilibrium calculations (Firooxabadi et al., 1978;
Katz and Firoozabadi, 1978; Kurnik et al., 1981). In
Fig. 3 the same experimental solubility data as in Figs 1
and 2 are compared with the predictions using the
Peng-Robinson
equation of state with its original
mixing rules and with its corrected van der Waals
mixing rules. According to this figure, the corrected
k .=-0.3537
+
==_
kj-0
326
Co2 - 23
l
>
0
308K
k,-
_____
DMN
- 0.095
318 K ly =
-0.090
328K
-[Link]
Kt=
50
100
150
~~!%?ilJRE
200
250
(BAR >
Fig. 1. Solubility of 2, 3-dimethyl naphthaleue(DMN) in
supercriticalcarbon dioxide at 308 K vs. [Link] solid
dots arc the experimental dam (Kurnik et al., 1981). The
dashed lines are the results of the van der Waals equation of
state with eqs (2) and (3.1) as the mixing rules. The solids lines
are results of the van der Waals equation of state with eqs (2)
and (3) as the mixing rules and eqs (4) and (17) as the
combining rules.
Fig. 2. Solubility of 2, 3-dimethyl naphthalene (DMN) in
supercritical carbon dioxide at 308, 318 and 328 K vs.
pressure_ The solid dots are the experimental data (Kumik
1981). The dashed lines are the results of
et al..
the Redhch-Kwong equation of state with eqs (2) and (3.1) as
the mixing rules. The solid lines are the results of the
Redlich-Kwong equation of state with eqs (3) and (20) as
the mixing rules and eqs (4) and (17) as the combining rules.
Van der Wsals mixing ruks
1307
van der Waals mixing rules of the Peng-Robinson
equation of state apparently do not improve solubility
predictions over the original mixing rules. However,
variation of solubility vs. pressure for the new mixing
rules is more consistent with the experimental data
than the old mixing rules. Also considering the fact that
the new mixing rules for the Peng-Robinson equation
of state contain three adjustable parameters (ku, I, and
m,) whereas the old mixing rules contain only two
adjustable parameters (k,
and 1, ) makes the new
mixing rules more attractive. A demonstration of the
superiority of the new mixing rules for the
Peng-Robinson equation of state is shown here in Figs
4-9. According to these figures, when the unlikeinteraction adjustable parameters of the mixing rules
are fitted to the experimental data, the Peng-Robinson
equation of state with the corrected van der Waals
mixing rules can predict solubility of heavy solids in
supercritical fluid more accurately than with the
original mixing rules over different ranges of temperature and pressure and for different solutes and supercritical solvents.
A :318K
-:328
K
jc
0
50
loo
150
PRESSURE
200
250
(BAR)
Fig. 4. Volubility of 2, 3-dimethyl naphthalene (DMN)
in supercritical carbon dioxide as calculated by the
Peng-Robinson equationof state usingeqs (2) and (3) as the
mixing rules and compared with the experimentaldata.
CC523 DMN
-23DMN
Tm308K
50
loo
150
PRESURE
200
(BAR)
250
2 318K
: 328
3(
Fig. 3. Solubihty of 2, 3-dimethyl naphthalene(DMN) in
supercritical
carbondioxideat 308 K vs. [Link] dashed
linesare the resultsof the Peng-Robinson equation of state
witheqs (2) and (3) as the [Link] solid linesare the
resultsof the Peng-Robinson equation of state with eqs (3)
and (20) as the [Link] sets of linesare for k, = 0.
50
loo
150
200
250
3
PRESSURE
t BAR 1
2, 3-dimethyl naphthalene(DMN) in
Fig. 5. Solubigty of
carbondioxideas eakukted by the Peng-Robinson equation
of state with eqs (X$-o-(4) as the mixing ruks and compared
with the experimtntaldata.
1308
T. Y. KWAK and G. A. MANSOORI
328
CO2
9f4
23DMN
= 0-2
l-,sl,
50
100
150
.:308
4 -3t8
-:328
K
K
K
200
250
I ij x-02420
ii
Iii =-(X2482
0~308
~~318
I :328
2SDMN
kS = am
CI
PRESSURE
(SARI
Fig. 6. Solubihty of 2, 3-dimethyl naphthalene (DMN) in
supercritical ethylene as calculated by the Pcng-Robinson
equation of state with eqs (2) and (3) as the mixing rules and
compared with the experimental data
K
K
K
50
loo
150
PRESSURE
200
(BAR)
250
3(
Fig 8. Solubility of 2, 6-dimethyl naphthalene (DMN) in
carbon dioxide as calculated by the Peng-Robinson equation
of state with eqs (2) and (3) as the mixing rules and compared
with the experimental data.
.
c
328 K
328 K
w-23
DMN
k jj=o.l8~
Iii = -0.242
ij = 0.3117
I ijSO.2723
mu=06602
l t3D8K
43l8K
m:328K
.
A
loo
150
3
200
250
PRESSURE (BAR,
Fig. 7. Solubihty of 2,3-dimethyl naphthalene in supercritical
ethylene as calculated by the Peng-Robinson equation of
state with eqs (22)-(24) as the mixing rules and compared with
the experimental data.
50
C,
: 308X
: 318-K
: 328K
3(
150
200
250
loo
PRESSURE BAR)
Fig. 9. Solubility of 2. 6-dimethyl naphthalene (DMN) in
as calculated
by
the
dioxide
supercritical carbon
Peng-Robinson equation of state with eqs (22)-(24) as the
mixing rules and compared with the experimental data.
,0
50
1309
Van der Weals mixing rules
Acknowledgement-This
research was supported
National Science Foundation Grant CPE 8306808.
by
the
NOTATION
Variable
A, B
in the Peng-Robinson
equation
of
mii
state
parameter in the equation of state
parameter
in the Peng-Robinson
equation
of state
binary interaction Parameter
interaction
parameters
in
the
binary
Peng-Robinson
equation of state
interaction
parameters
in
the
binary
N0
P
Avogadros
number
pressure, Bar
R
T
universal
a, h
d
k,
lij
Peng-Robinson
letters
5
u
interaction energy
acentric factor
fugacity
parameter
coefficient
intermolecular
length
Subscripts
energy
molar volume, cm/mol
gas phase mole fraction
compressibility
factor
Y
z
i, i
2
of state
gas constant
temperature,
K
intermolecular
potential
Greek
&
equation
component
identifiers
solute
critical property
parameter
function
REFERENCES
Alem. A. H. and Mansoori, G. A., 1984, The VIM theory of
molecular thermodynamics. [Link].E. J. 30,46&480.
Firoomhsdi, A., Hekim, Y. and Katz, D. L., 1978, Reservoir
depletion calculations for gas condensates using extended
analysis in the Peng-Robinson equation of state. Can. J.
them. E?tgng 56,610-615.
Katz, D. L. and Firoozahadi, A., 1978. Predicting phase
behavior of condensate/crude-oil systems using methane
interaction coeffioients. J. Petrol. Tech. Nov. 1649-1655.
Kumik, R. T., Holla, S. J. and Reid, R. C., 1981, Solubility of
solids in supercritical carbon dioxide and ethylene. J. them.
Engng Data 26.47-51.
Leland. T. W. and Chsppelear, P. S., 1968a, Recent developments in the theory of fluid mixtures. Ind. Engng Chem. 60,
15-24.
Loland, T. W., Rowlinson, J. S. and Sather. G. A., 1968b,
Statistical thcrmodynamios of mixtures of molecules of
different sixes. Trans. Faraday Sot 64, 1447-1460.
Leland, T. W.. Rowlinson, J. S., Sather, G. A. and Watson,
I. D., 1969, Statistical thermodynamia of two-fluid models
of mixtures. Trans. Faraday Sot. 65,2034-2043.
Mansoori, G. A. and Ely, J. F., 1985. Density expansion
(DEX) mixing rules: thermodynamic modeling of supercritical extraction. J. Chem. Phys. 82, 406-413.
Peng. D. Y. and Robinson, D. B., 1976, A new two-constant
equation of state. Ind. Engng Chem. Fundam. 15, 59-64.
Prausnitz, J. M., 1969, Molecular Tltemwdynamies of Fluid
Phase Equilibria. Prentice-Hall, Englewood Cliffs, NJ.
Redlich, 0. and Kwong, J. N. S.. 1949, On the thermodynamics of solutions. V. An equation of state: fugacities of
gaseous solutions. Chem. Rev. 44, 232244.
Renon. H. (Ed.), 1983, Fluid properties and phase equilibrium
for chemical process design. Proc. 3rd Int. Conf., Callaway
Gardens, GA, 10-15 April, Fluid Phase Equilibria, Vol. 13,
pp. 15-203.
Rowlinson, J. S. and Swinton, F. L.. 1982, Liquids and Liquid
Mixtures, 3rd edn. Butterworths, Wolhom, MA.
Van der Wsals, 3. D., 1873, Doctoral dissertation, Loiden.
