Beruflich Dokumente
Kultur Dokumente
3 (1979) 255-271
@ Elsevier Scientific Publishing Company,
255
Amsterdam
Franpzis
(Received
du Pktrole,
December
1 et 4, Avenue
de Bois-Pr&u,
92502
Rueil-Malmaison
(France)
ABSTRACT
Huron, M.-J. and Vidal, J., 1979. New mixing rules in simple equations of state for representing vapourliquid
equilibria of strongly non-ideal mixtures. Fluid Phase Equilibria,
3: 255-271.
Good correlations of vapour-liquid
equilibria can be achieved by applying the same
two-parameter cubic equation of state to both phases. The results primarily depend on the
method used for calculating parameters and, for mixtures, on the mixing rule. True parameters are the covolume b and the energy parameter a/b. For this latter one, deviations from
a linear weighting rule are closely connected to the excess free energy at infinite pressure.
Thus any mixing rule gives a model for the excess free energy, or any accepted models for
this property can be used as mixing rules.
From the above, an empirical polynomial mixing rule is used for data smoothing and
evaluation, while for practical work a local composition
model is used. The mixing rule
thus obtained can be reduced to the classical quadratic rule for some easily predicted
values of the interaction energies. For highly polar systems, it includes three adjustable
parameters. Using literature data, the new mixing rule is applied, in the low and high pressure range, to binary mixtures with one or two polar compounds,
giving good data correlation and sometimes avoiding false liquid-liquid
immiscibility.
INTRODUCTION
When applied to both vapour and liquid phases, cubic equations of state
can be used to calculate phase equilibria. Pure-component parameters are
evaluated from critical conditions and vapour pressure data. To obtain the
mixture parameters, classical mixing rules are used, and good results are obtained for non-polar gases (Soave, 1972; Peng et al., 1976; Robinson et al.,
1977). The binary parameters which appear in the mixing rules can be predicted by various combining laws (Hicks et al., 1975) or fitted to experimental data. In this way, the application range can be extended (Huron et al.,
256
1977; Asselineau et al., 1978a; Graboski et al., 1978) to slightly polar mixtures.
However, classical mixing rules imply a quasi-regular behaviour (Vidal,
1978) and, unless a thorough revision is made, the method cannot be applied
to polar mixtures. It has been proposed that the value of the pure-component
parameters be modified (Wenzel et al., 1971; Deiters et al., 1976; Wenzel et al.,
1978). However, the close relation existing between excess Gibbs energy and
mixing rules led us to the problem of evaluating mixture parameters, and this
paper is devoted to mixing rules.
EXCESS GIBBS ENERGY
OF STATE
As pointed out by Smith and Van Ness (1975), there are three fugacity
coefficients, one for a pure component, cp;, one for the solution, cp, and one
for a component in solution, Cpi.They are related to Gibbs energy and chemical potential by their definition equations:
RTln&=+f-gi*=!
(~r-y)dP
0
RTln~=g-g=~p(v-~)dP
0
RTlnIpi=~i-&=pi-pj=J
(ci-y)dP
0
g = iz Xi&j* + RT In
(4)
xi)
(5)
(6)
i=l
267
state that is used and is the same for cpr and cp:
~pf= F(T, P, u;, aii, bii, e-e)
cp = F(T, P, u, a, b)
So the various expressions we shall deduce from eqn. (6) will depend on the
equation of state, but not on the mixing rules.
We shall first apply eqn. (6) to the modified Redlich-Kwong
equations of
state :
(u -
b) = RT
(8)
and from eqn. (6) we obtain:
(9)
RELATION
The relation between the excess Gibbs energy gE and the mixing rules
implicit in the a and b values is not obvious. At this step, we consider the
limiting value of gE at infinite pressure. If at infinite pressure the excess
volume is not zero, the limit of gE is infinite. Thus, as a necessary condition
to obtain a finite limit, we apply the common linear mixing rule for volume
parameter b :
b = 2
biixi
i=l
1) that the
(11)
We see that the gz value and the excess value of the energy parameter a/b
258
are proportional
. So the value of the excess Gibbs energy gz will correspond
to the choice of mixing rule. As an example (Vidal, 1978) we give the classical
quadratic mixing rule:
2 2 QijXiXj
a =
i=l
associated
CZij
(12)
j=l
rule:
(13)
(a$Zjj)1'2(1hij)
and with the linear law (10) for parameter b leads to a quasi-regular
For a binary system, for example, we obtain:
&
= (In 2) b&@st(%
6~)~ + 2
mixture.
(14)
b&,6,1
with
4i
Xibii
=b
EIowever,
ai=
and
eqn. (11)
v--~aiilbii
bit
(15)
a mixing
a
(17)
from the chosen excess Gibbs energy at infinite pressure. We do this in this
paper.
When dealing with equilibrium problems, the fugacity coefficients
for the
components in solution must be evaluated. By applying classical thermodynamics and mixing rules (10) and (17), we obtain:
Inv,=-_ln(P(ii*))
+$(%-I)
= &
[ g
coefficients
n a&
+,zz(Aij--Xj)
j
--(k-s)
at infinite
In(F)
(18)
by:
The pure-component
parameters will be determined from stability
and vapour pressure values, as explained in Appendix
2.
NEW MIXING
(19)
conditions
RULES
In vapour-liquid-liquid
equilibrium investigations, we meet two kinds of
problems: (1) data smoothing, data and model evaluation; and (2) in the
common practice of the chemical engineer, equilibrium predictions from a
minimum amount of data. We now consider briefly the first of these.
259
Polynomial
mixing
rule
For data smoothing and evaluation, we look for great flexibility in the
mixing rule. In the low pressure range, a Redlich-Kister
polynomial expansion
of the excess Gibbs energy is often applied to binary systems:
gE = RTxlxZ
mfO
A,(xl
(20)
~2)~
Applying the same equation to the excess Gibbs energy at infinite pressure,
from eqns. (10) and (17) we obtain:
a = (z,b,,
+ x,b,,)
x1 p
+x2
11
u22
b22
-__
RT
In 2
x1x2
A,n(x,
m=O
-x2Y
(21)
Such a mixing rule is applied, as an example, to the carbon dioxide(l)ethane(2) system (Ohgaki et al., 1977) at 283.15 K. The pressure range is
20-50 atm. The polynomial parameters, A,,,, will be determined by minimizing the objective function:
(uP,ex)2
+
[(E),%J2+[(E),UTJ2
:
(22)
ur = 0.01 K,
u, = 0.002
260
A P-Pexp (atm.)
45.
l*=
'.
m
.
.- .
.
-x1
-95
t
4 P-P,,(atm.)
0.5 -
are small and random, and because of the great flexibility of the model they
can be considered as an evaluation of experimental data scattering. For example, after fitting to the same data (Ohgaki et al., 1977) the interaction
parameter kij involved in classical mixing and combining rules (12) and (13),
the results obtained:
ZP = 0.48 atm
have to be considered with reference to the limit previously obtained (0.1 atm).
Furthermore, as shown by Fig. 1, the deviations in pressure are not random,
because of the wrong skewness of the excess Gibbs energy associated with the
classical mixing rule.
Local-composition
mixing rule
261
we look for a new mixing rule which, when only non-polar compounds are
present, can be reduced to the classical one by a straightforward choice of
parameters, but which can provide more flexibility if necessary. The localcomposition concept (Wilson, 1964) has led to many powerful models of the
liquid phase (Vetere, 1977). The Scatchard-Hildebrand
theory can be established, among other hypotheses, by considering only the interactions between
first-neighbour molecules (interactions between more than two molecules and
influence of long-range molecules are neglected), and by assuming that the
first-neighbour interaction occurrences are proportional to the overall volume
fractions (Moelwyn-Hughes, 1961). Therefore we deduce local compositions
tji and Eii, which give the distribution of molecules j and i about the central
molecule i, from overall volume fractions $j and @i :
tji _ @j exP(--cYjigji/RV
Eji
(23)
@i exP(-ajigii/RV
The volume fractions introduced here, and already defined by eqn. (15),
must be evaluated at infinite pressure since they will be applied to the g!
formulation. They are weighted by Boltzman factors, including interaction
energies between unlike (gji) and like (gii) molecules and, as in the NRTL
model (Renon, 1968), a non-randomness parameter eji. The value of the excess Gibbs energy at infinite pressure is then calculated in exactly the same
way as in the NRTL equation, and the resulting equation is the same:
?JXjGjiCji
g2 =
gxij=l
(24)
i=l
2
k=l
XhGhi
with
(25)
exp(-aji&)
(26)
The only difference from the classical NRTL model is the definition of the
local composition as corrected volume fractions, which leads to the introduction of the volume parameter bj in the calculation of Gji. From the choice of
such a model for the excess Gibbs energy at infinite pressure, the mixing rule
is deduced by applying eqn. (17):
n
==b
xx1
s-1
bii
i=l
Xj
GjiCji
j=l
(27)
In 2
~
k=l
XkGki
262
An exact reduction to the classical mixing and combining rules (12) and
(13) can easily be obtained by the following calculation of the coefficients:
Lyij= 0
gii = -(In
2)-i
ii
(29)
where kij is the interaction parameter involved in the combining rule (13).
Owing to the zero value of the non-randomness parameter, local compositions are equal to volume fractions and the excess Gibbs energy at infinite
pressure is quasi-regular. The iriteractions energies can then be calculated
from the parameters of the equation of state. So the new mixing rule (27) contains, as an optional possibility, the classical rtfle. When applying the mixing
rule to vapour-liquid equilibria calculations, we shall consider three possibilities. In the first, the parameter 01is zero and we apply eqns. (24) to (29). It is
the classical mixing rule, and the interaction parameter will be adjusted to
experimental data. In the second, eqns. (24) to (29) will still be applied, but
both kij and aij will be determined from experimental data. We found by
experience that data correlation can be greatly improved in this way. Finally,
we can apply eqn. (27), in which the three adjustable parameters will be Cij,
Cji and ii for a binary mixture.
As in the classical NRTL model, the skewness of the excess Gibbs energy
versus composition curve mainly depends on the C, values, and flatness is
increased by high values of OL,thus providing great flexibility in the equilibrium correlation and preventing false liquid-phase splitting.
APPLICATION
TO VAPOUR-LIQUID
EQUILIBRIUM
CORRELATION
We applied the local-composition mixing rule (27) in the low and high pressure ranges. A phase equilibrium algorithm, constructed to avoid the trivial
solution occurring near the critical point (Asselineau et al., 1978b), was used.
Parameters were adjusted by minimizing the function:
NT
SQ F 5
N=l
(P-P,,);
+ c
N=l
(yr -y;,,,)&
=SQP+SQY
(39)
263
it can be seen that data correlation can be greatly improved by the local-composition mixing rule, and adjusting only the interaction parameter and the
non-randomness parameter (second option in the mixing rule) gives quite
good results for most systems. Furthermore, it must be emphasized that the
classical mixing rule has a strong tendency towards false phase-splitting predictions. For example, as shown by Fig. 2, the behaviour of an acetone-water
P&m
425
0.50
Q75
Xl,Yl
I
0
425
w5
Xl,Yl
P/atm
P/atm
t = 25oT
0,25
950
0*75
J
WY1
Ref.
Berro,
1974
Tasic,
1978
Griswold,
1952
System
Methanol-1,2dichloroethane
Acetonecyclohexane
Acetonewater
Temperature
l-4
5-12
15-30
39-67
150
209
250
0.15-0.35
25
100
0.31-0.65
Pressure
range
(atm)
50
(G)
NT
Fitted
parameters
13
25
17
22
23
Cl29 Cal* a
k12
&?,a
c12, c21, a
kg
k12>o
k12
kl2, a
Cl% c21, a
h2
kl2r a
Cl% c21, a
c12, c21, Q
k12
+z, o
kl,? a
Cl,, c21, @
30 h2
NT
LY = {[ c (Yl -Yl,ex)~l/NT~o*5
N=l
TABLE 1
-0.087
+0.001
-0.128
-0.022
-0.165
-0.063
-0.202
-0.108
0.112
0.093
0
0.269
0.429
0
0.277
0.416
0
0.293
0.422
0
0.281
0.461
0
-0.445
-0.835
0.570
0.471
0.227
CI
0.084
k12
73011
71261
67446
61798
24568
58849
39743
38961
34579
14768
60768
(atm
cm3/mo1e)
(atm
cm3 /mole)
17833
c21
Cl2
0.36
0.06
0.05
0.97
0.10
0.04
2.17
0.21
0.16
2.80
0.44
0.24
0.016
0.004
0.001
0.014
0.005
0.049
(atm)
LJ
0.08
0.01
0.01
0.05
0.01
0.01
0.05
0.01
0.01
0.04
0.01
0.01
0.049
0.012
0.004
0.07
0.02
0.01
AY
Propaneethanol
GomezNieto,
1978
25
Ohgaki,
1976a
co2
25
Ohgaki,
1976b
Ethanemethyl
acetate
Methanol-
11
12
6-41
6-47
13-56
101.85
126.85
151.85
11
12
3-27
76.85
16
10
16
l-16
5-80
7-59
4-38
4-39
30-50
51.85
40
25
Ohgaki,
1976h
Ethaneacetone
10
Ohgaki,
1977
COs-ethane
42
kl2,a
kl2va
Cl29 c21,a
h2
kl2.a
Cl29 CZlt~
kl2
kl2,a
Cl29 c21>a
h2
kl2,a
c12> c21,a
h2
km
Cl27 c21,a
kl2
Cl, C21,Q
kl2
kl2,a
Cl% C21ra
h2
kl2ra
Cl% c21, a
Cl23 G?l%~
h
km Q!
0.016
0.039
0.028
0.060
0.027
0.063
0.026
0.056
0.035
0.070
0.062
0.051
0.129
0.153
0.142
0.158
0.139
0.2
0.102
0.124
0.731
0.117
1.112
0.078
0.972
0.070
0.740
0
0.043
0.437
0
0.730
0
0.652
0
0.025
0.476
0
0.673
0.638
26525
53826
41604
57644
31884
48457
47118
47933
18257
6929
27879
24938
28717
14120
11940
15192
21130
19910
0.65
0.48
0.14
0.66
0.48
0.21
1.18
0.97
0.41
0.76
0.14
0.12
0.90
0.76
1.23
0.59
0.51
2.68
0.50
2.33
0.71
1.02
0.66
0.52
0.34
0.16
0.15
0.01
0.02
0.01
0.01
0.02
0.01
0.01
0.01
0.01
0.03
0.02
0.02
0.04
0.04
0.001
0.001
0.003
0.001
0.004
0.002
0.002
0.001
0.002
0.001
0.005
0.006
0.005
266
system cannot be correlated by the quadratic mixing rule since a large miscibility gap would be predicted. On the contrary, the local-composition
model can
be applied with good results, up to the critical range.
DISCUSSION
AND CONCLUSION
Calculating vapour-liquid
equilibria by applying the same equation of
state to both phases was discussed by Van der Waals. When using such a method, choosing the equation of state, the choice of the pure-component
parameters and the choice of the mixing rules are decisive steps. The more constants
the equation of state has, the more mixing rules are required, and the more
data are required for evaluating the pure-component
parameters. Thus twoparameter cubic equations of state are often preferred. Proposed by Berthelot,
adopted by Van der Waals and now extensively used, the quadratic mixing
rule, as applied to the attraction parameter, is strongly supported by the
:
results obtained in phase equilibria predictions for non-polar or slightly polar
systems. Furthermore, the existing relation between a, b and the second virial
coefficient:
B = 2 5 BijXiXj = b i=l j=l
(a/RT)
and the commonly accepted quadratic law for B imply a quadratic law for a.
When applying the mixing rule (27) for parameter a, the quadratic law for the
second virial coefficient B will not be obeyed. However, in the high density
range and for polar mixtures, the experimental behaviour of excess Gibbs
energy does not obey the law implied by the classical mixing rule. Our method for deducing a mixing rule from a model of excess Gibbs energy is highly
empirical, but is consistent with the vapourliquid
equilibrium calculations.
The Soave equation of state is used here with new mixing rules for parameter a. The close connection found between the energy parameter (a/b) and
the excess Gibbs energy g: can be extended to other cubic equations of state
having the form:
p=
- RT
u-
a(T)
-
(31)
b -9(u)
FyLgxi%a]^
i=l
+ &AZ)
gz=-
to parameter
iI
(32)
constants, then the equation
for g? is:
(33)
267
with
1
Iz = AZ - A1 In
if h2 # hi
A = -&ln(fs));
A =
But eqn. (33) does not apply to the three-parameter equations of state of
Clausius, Fuller (1976), Horvath and Lin (1977) and Usdin and McAuliffe
(1976), for which the roots of Q(u) are not proportional to b. For these last
equations, we need to set mixing rules for two parameters in order to be
able to calculate the third one from gz .
Like the choice of a mixing rule, the choice of a model for the excess Gibbs
energy is difficult, as shown by several attempts in the litterature. We used
two expressions, depending on the problem.
It would be attractive to apply the polynomial mixing rule to a consistency
test by comparing calculated and experimental vapour compositions. In the
low pressure range, the influence of the polynomial mixing rule on the density
of the vapour phase is slight, and we obtained the same results by using the
standard method (Redlich-Kister
excess Gibbs energy, reference fugacities in
the liquid phase, virial equation of state for the vapour phase) or by applying
the straightforward technique of the polynomial mixing rule (21). In a higher
pressure range, however, such a method would be questionable since the
calculated vapour phase compositions are sensitive to the vapour density
evaluation (Won et al., 1973). On the whole, our results did not show any
systematic discrepancy between experimental and calculated vapour phase
composition, but we shall not use them as a consistency test and we shall
reserve the polynomial mixing rule for evaluating data scattering.
Application of a local-composition
(NRTL) model to the excess Gibbs
energy at infinite pressure was first considered by Chaudron et al. (1973).
However, the proposed method used reference fugacities and, together with
the NRTL model, a mixing rule was used for the equation of state, which was
applied to the calculation of the fugacities in the vapour phase, for the
Poyinting correction and for the reference fugacities. Although thermodynamically more sound, this method requires more parameters and does
268
not use the same model for calculating vapour phase fugacities and liquid
phase fugacities.
If we decide to apply the same model to both phases, the implicit relation
between mixing rule and excess Gibbs energy has to be considered and can be
used.
ACKNOWLEDGEMENTS
a, b
B
C
g
k
;P
NT
P
R
OF SYMBOLS
Redlich-Kister coefficient
parameters in the equation of state
virial coefficient
parameter in the mixing law
Gibbs free energy or interaction energy parameter
interaction parameter associated with a
number of components in the mixture
number of parameters
number of experimental points
pressure (atm) 1 atm = 1.01325 X 10 Pascals
perfect gas constant = 82.0562 atm cm3 K-l mol-
mole1 K-l
function to be minimized
temperature (K)
molar volume ( cm3/mole)
molar fraction
non randomness parameter
activity coefficient
solubility parameter
Kronecker delta
deviation in pressure
deviation in mole fraction
numerical constants
local composition
weighting factor
fugacity coefficient
volume fraction
volume function
Superscripts or subscripts
C
critical property
excess value
or 8.3143
269
experimental value
component identification
experimental point
ideal state
pure component
at infinite pressure
an overbar denotes a partial molar property
ex
i, j
N
*
00
-
APPENDIX
1. LIMIT
OF EXCESS
GIBBS
ENERGY
AT INFINITE
PRESSURE
gE=$+&+g3E
where
when P + 00, u + b, vr + bs and from the equation of state (7) we see that:
I
P(v-Wcl_
4v-b)
RT
RTu(u + b)
--& _P(u-bb)
RT
RT
+
RT
or
SZ _P(V-bb)_~x_P(uf-bi,)+
-I
RT
RT
RT
i=l
Pb
&,Pb,
RT
i=l
if
b = 2
i=l
xibii
b)]/RT,
RT
we
270
So gg has the same limit as gt , the value of which is given by eqn. (11).
For the other equations of state mentioned in discussion, the limit is established in the same way.
APPENDIX
2. DETERMINATION
OF PURE-COMPONENT
PARAMETERS
The values of U,+ and bCi are determined from critical conditions:
1
=ci = 9(21/a
1)
R2Zi
pci
bi* = bci =
213~1%
pci
271
Griswold, .J. and Wang, S.Y., 1952. Phase equilibria of the acetone-methanol
water system from 1OOC into the critical region. Chem. Eng. Prog. Sympos. Series, 48, NO. 3:
18-34.
Harmens, A., 1977. A cubic equation of state for the prediction of N,-Ar-Og
phase
equilibrium. Cryogenics, 17: 519-522.
Hicks, C.P. and Young, CL., 1975. The gas-liquid critical properties of binary mixtures.
Chem. Rev., 75: 119-175.
Horwath, C. and Lin, H.J., 1977. A simple three-parameter
equation of state with critical
compressibility-factor
correlation.
Can. J. Chem. Eng., 55: 450-456.
Huron, M.J., Dufour, G.N. and Vidal, J., 1978. Vapour-liquid
equilibrium and critical
locus curve calculations with the Soave equation for hydrocarbon
systems with carbon
dioxide and hydrogen sulfide. Fluid Phase Equilibria, 1: 247-265.
Moelwyn-Hughes,
E.A., 1961. Physical Chemistry. Pergamon, London, second revised
edition, p. 821.
Ohgaki, K. and Katayama, T., 1976a. Isothermal vapour--liquid equilibrium data at high
pressures: methanol-carbon
dioxide, n-hexane--carbon
dioxide and benzene-carbon
dioxide systems. J. Chem. Eng. Data, 21: 53-55.
Ohgaki, K., Sane, F. and Katayama, T., 197613. Isothermal vapour-liquid
equilibrium
data for binary systems containing ethane at high pressures. J. Chem. Eng. Data, 21:
55-58.
Ohgaki, K. and Katayama, T., 1977. Isothermal vapour-liquid
equilibrium data for the
ethane-carbon
dioxide system at high pressures. Fluid Phase Equilibria, 1: 27-32.
Peneloux, A., Deyrieux, R. and Neau, E., 1975. Reduction des don&es sur les equilibres
liquide-vapeur
binaires isothermes. Criteres de precision et de coherence. Analyse de
linformation.
J. Chim. Phys., 72: 1107-1117.
Peng, D.Y. and Robinson, D.B., 1976. A new two-constant
equation of state. Ind. Eng.
Chem. Fundam., 15: 59-64.
Renon, H. and Prausnitz, J.M., 1968. Local composition in thermodynamic
excess functions for liquid mixtures. AIChE J., 14: 135-144.
Robinson, D.B., Peng, D.Y. and Ng, H.J., 1977. Some applications of the Peng-Robinson
equation of state to fluid property calculations. Proc. 56th Ann. Conv. Gas. Process.
Assoc., 56: 11-17.
Smith, J.M. and Van Ness, H.C., 1975. Introduction
to chemical engineering thermodynamics. McGraw Hill, New York, Chemical Engineering Series, 3rd edition, section
D2, p. 605.
Soave, G., 1972. Equilibrium constants from a modified Redlich-Kwong
equation of
state. Chem. Eng. Sci., 27: 1197-1203.
Tasic, A., DjordjeviF, B., Grozdanic, D., Afgan, N. and Malir, D., 1978. Vapourliquid
equilibria of the systems acetone-benzene,
benzene-cyclohexane
and acetone-cyclohexane at 25C. Chem. Eng. Sci., 33: 189-197.
Usdin, E. and McAuliffe, J.C., 1976. A one parameter family of equations of state. Chem.
Eng. Sci., 31: 1077-1084.
Vetere, A., 1977. A modified Heil-Prausnitz
equation for excess Gibbs energy. Can. J.
Chem. Eng., 55: 70-77.
Vidal, J., 1978. Mixing rules and excess properties in cubic equations of state. Chem. Eng.
Sci., 33: 787-791.
Wenzel, H. and Peter, S., 1971. Berechnung von bin&en phasengleichgewichten
mit einer
iiberkritischen komponente.
Chemie Ing. Techn., 43: 856-861.
Wenzel, H. and Rupp, W., 1978. Calculation of phase equilibria in systems containing
water and supercritical components.
Chem. Eng. Sci., 33: 683-687.
Wilson, G.M., 1964. Vapour-liquid
equilibrium. XI. A new expression for the excess free
energy of mixing. J. Am. Chem. Sot., 86: 127-130.
Won, K.W. and Prausnitz, J.M., 1973. High pressure vapour-liquid
equilibria. Calculation
from total pressure data. Thermodynamic
consistency.
Ind. -Eng. Chem. Fundam., 12:
459-463.