Professional Documents
Culture Documents
Equations of State For The Calculation of Fluid-Phase Equilibria
Equations of State For The Calculation of Fluid-Phase Equilibria
JOURNAL REVIEW
Introduction
Equations of state play an important role in chemical engi- 1979; Gubbins, 1983; Tsonopoulos and Heidman, 1985; Han
neering design, and they have assumed an expanding role in et al., 1988; Anderko, 1990; Sandler, 1994; Economou and
the study of the phase equilibria of fluids and fluid mixtures. Donohue, 1996..
Originally, equations of state were used mainly for pure com- The van der Waals equation of state was the first equation
ponents. When first applied to mixtures, they were used only to predict vapor-liquid coexistence. Later, the Redlich-Kwong
for mixtures of nonpolar ŽSoave, 1972; Peng and Robinson, equation of state ŽRedlich and Kwong, 1949. improved the
1976. and slightly polar compounds ŽHuron et al., 1978; Asse- accuracy of the van der Waals equation by introducing tem-
lineau et al., 1978; Graboski and Daubert, 1978.. Subse- perature-dependence for the attractive term. Soave Ž1972.
quently, equations of state have developed rapidly for the and Peng and Robinson Ž1976. proposed additional modifica-
calculation of phase equilibria in nonpolar and polar mix- tions to more accurately predict the vapor pressure, liquid
tures. There are many advantages in using equations of state density, and equilibria ratios. Carnahan and Starling Ž1969.,
for phase equilibria calculations. Equations of state can be Guggenheim Ž1965., and Boublik Ž1981. modified the repul-
used typically over wide ranges of temperature and pressure, sive term of the van der Waals equation of state to obtain
and they can be applied to mixtures of diverse components, accurate expressions for hard body repulsion. Other workers
ranging from the light gases to heavy liquids. They can be ŽChen and Kreglewski, 1977; Christoforakos and Franck,
used to calculate vapor-liquid, liquid-liquid, and supercritical 1986; Heilig and Franck, 1989. modified both the attractive
fluid-phase equilibria without any conceptual difficulties. and repulsive terms of the van der Waals equation of state.
The calculation of phase equilibria has been discussed exten- In addition to modeling small molecules, considerable em-
sively elsewhere ŽSadus, 1992a, 1994; Sandler, 1994; Dohrn, phasis has been placed recently on modeling chain-like
¨
1994; Raal and Muhlbauer, 1998. and earlier reviews of vari- molecules. Based on the theories of Prigogine Ž1957. and
ous aspects of equations of state are also available ŽMartin, Flory Ž1965., other workers ŽBeret and Prausnitz, 1975;
Donohue and Prausnitz, 1978. developed a perturbed hard-
Correspondence concerning this article should be addressed to R. J. Sadus. chain theory ŽPHCT. equation of state for chain molecules.
The other equation of state parameters are calculated from Modification of the repulsi©e term
The other way to modify the van der Waals equation is to
bs V b Ž b .
RTc ¶ examine the repulsive term of a hard-sphere fluid. Many ac-
pc curate representations have been developed for the repulsive
interactions of hard spheres and incorporated into an equa-
Ž 1y b .Ž 2q c b . y Ž 1q c b . tion of state. Some alternative hard-sphere terms are summa-
V bŽ b . s b
Ž 2q c b .Ž 1y b . 2 rized in Table 2. Perhaps the most widely used alternative to
the van der Waals hard-sphere term is the equation proposed
V a Ž b . R 2 Tc a Ž T .
aŽ T . s by Carnahan and Starling Ž1969. who obtained an expression
pc • Ž 11. for the compressibility factor of hard-sphere fluids that com-
2
Ž 1q c b . V b Ž b . pares very well ŽFigure 1. with molecular-dynamics data ŽAl-
V aŽ b . s 2
der and Wainwright, 1960.. The form of the Carnahan-Star-
b Ž 1y b . Ž 2q c b . ling equation is
a 1r2 Ž T . s1q q Ž b . Ž 1yTr1r2 .
q Ž b . s Ž br0.26 .
1r4
m 1qh qh 2 yh 3
ms 0.480q1.5740 v y0.176 v 2
ß Zhss
Ž 1yh . 3
Ž 13.
1q Ž 3 a y2 . h q Ž 3 a 2 y3 a q1 . h 2 y a 2h 3 a
Zs 3
y Ž 16.
Ž 1yh . RTV
Sadus et al. Ž1988. and Christou et al. Ž1991. have used Eq.
16 for the calculation of the vapor-liquid critical properties of
binary mixtures containing nonspherical molecules. The re-
sults obtained were slightly better than could be obtained
from similar calculations using the Guggenheim equation of
state. Sadus Ž1993. proposed an alternative procedure for ob-
taining the equation of state parameters. Equation 16 in con-
junction with this modified procedure can be used to predict
Type III critical equilibria of nonpolar binary mixtures with a
Figure 1. Hard-sphere compressibility factors from dif- good degree of accuracy.
ferent equations of state with molecular simu- Sadus Ž1994. compared the compressibility factors pre-
lation data. dicted by the van der Waals, Guggenheim and Carnahan-
^, Alder and Wainwright Ž1960 .; `, Barker and Hender- Starling hard-sphere contributions with molecular simulation
son Ž1971 . for hard spheres. data ŽAlder and Wainwright, 1960; Barker and Henderson,
Z rep s
4h y2h 2 ¶
Equation 28 accounts for the effect of rotational and vibra- 3
tional degrees of freedom on both the repulsive and attrac- Ž 1yh .
tive forces. The pure component and binary parameters can TU • Ž 30.
be obtained by fitting experimental data. Donohue and
Y sexp ž /
2T
y1
Prausnitz Ž1978. reported that the PHCT equation of state h st V UrV
can represent the properties of most mixtures commonly en-
T U s e qrck
ß
countered in petroleum refining and natural-gas processing
even when the components differ greatly in size, shape, or
potential energy. For pure substances, the SPHCT equation of state can be
Subsequently, many workers ŽKaul et al., 1980; Liu and written as
Prausnitz, 1979a,b, 1980; Ohzone et al., 1984. have applied
the PHCT equation to predict thermodynamic properties of 4h y2h 2 Zm cV U Y
numerous and varied types of systems of industrial interest. Zs1q c y Ž 31.
Kaul et al. Ž1980. predicted Henry’s constants using the PHCT Ž 1yh . 3 V qV U Y
m h Ž 3.498y0.24h y0.414h 2 . The 1q Z rep term is evaluated in an identical way to the cor-
y ž 2
y1 /ž 1q
Ž 1yh .Ž 2yh .Ž 0.534q0.414h . / Ž 39.
responding of the PHCT equation of state Žsee Eq. 27..
By extending the perturbation expansion of Barker and
Henderson Ž1967. for spherical molecules to chainlike
Using the TPT-D approach ŽEq. 36., Sadus Ž1995. pro-
molecules, the attractive Lennard-Jones isotropic interac-
posed that, in general
tions are calculated as
g hd s g h s Ž ah q c . Ž 40 . 2
A 2L J A 2L J
where a and c are the constants for a straight line and the
values can be obtained by fitting the molecular simulation
Z iso s Z1L J q Z2L J y2 Z1L J
ž / žA1L J
1y
A1L J / Ž 43.
Aseg
sm
Aseg
0
Ž 51. Z hss m
4h y2h 2 ¶
NkT NkT 3
Ž 1yh .
where m is the number of segments per chain, and Aseg Žper u i h j
0 Z disp s m Ý Ý jDi j
mole of segments. is the residual Helmholtz function of the kT t
nonassociated spherical segments. It has two contributions:
i j
• Ž 58.
the hard-sphere and dispersion Ž 5r2 . h yh 2
Z chain s Ž 1y m .
Ž 1yh . w 1y Ž 1r2 . h x
Aseg
0 Ah0 s Adisp
0 1 1 X A
s q Ž 52. Z assoc s r Ý y ß
NkT NkT NkT A X A
2 r
Ah s 6 z s3 3z 1 z 2 z 23 Ah s
5 h Ž 34y33h q4h 2 .
y z 0 ln Ž 1y z 3 . q s ln Ž 1yh . q Ž 66 .
NkT
s
pr ž z 32 / Ž 1y z 3 .
q
z 3 Ž 1y z 3 .
2 NkT 3 6 Ž 1yh .
2
D B2, h B H s Ý
0
Ci T ir2
¶ where g sws is the site-site correlation function at contact of
sws
i sy7 square-well spheres, and Zref is the compressibility factor of
r U s mbrVm square-well spheres which can be calculated from Barker-
p • Ž 68.
Henderson perturbation theory. Tavares et al. Ž1995. also
h s r UsBH3
proposed equation of states for square-well chains using the
6 SAFT approach. In addition to properties of monomer seg-
1
ments, they included dimer properties in the spirit of the
sBH s Ý DiT ir2 q D ln ln T ß
i sy2
TPT-D1 equation ŽEq. 38.. The concepts have been extended
to hetero-segmented molecules ŽAdidharma and Radosz,
1998.. Recently, six different SW-SAFT models have been
where Ci and Di , Ci j and D ln are numerical constants and reviewed ŽAdidharma and Radosz, 1999.. According to the
adjustable parameters, hBH stands for hybrid Barker- comparison reported by Adidharma and Radosz Ž1999., the
Henderson, and details were given by Kolafa and Nezbeda addition of dimer structure does not improve the usefulness
Ž1994..
of SW-SAFT equations for representing real fluids signifi-
For the dipole-dipole term, the resulting Helmholtz func- cantly. Originally, the validity of SW-SAFT equations were
tion is limited to a square well with a width of 1.5s . This limitation
has been addressed by subsequent work ŽChang and Sandler,
Adipole A2 1 1994; Gil-Villegas et al., 1997..
NkT
s
T ž 1y Ž A 3rA 2 . / Ž 69.
By providing an additional parameter which characterized
the range of the attractive part of the monomer-monomer
where potential, Gil-Villegas et al. Ž1997. proposed a general ver-
sion of SAFT for chain molecules formed from hard-core
A 2 sy
2p r UmU 4
J Ž6.
¶ monomers with an arbitrary potential of variable range ŽVR..
As discussed above, the general form of the SAFT equation
U
3 T of state can be written as
32p 3 14p r U 2mU 6
A3 s
135 5
( TU2
333
K 222 A
s
Aideal
q
Amono
q
Achain
q
Aassoc
Ž 72.
T Tk B
• Ž 70. NkT NkT NkT NkT NkT
TU s s
T˜ e
In VR-SAFT ŽGil-Villegas et al., 1997., the ideal Helmholtz
mU s mr'e m s 3 function Aideal and the contribution to the Helmholtz func-
rU s r b tion due to interaction association Aassoc are the same as in
bs NL s 3 ß Eq. 50. The contribution due to the monomer segments Amono
is given by
333
The coefficients J Ž6. and K 222 are integrals over two-body Amono Am Ah s
q b A1 q b 2 A 2
and three-body correlation function for the Lennard-Jones
fluid and have been calculated by Twu and Gubbins Ž1978a,b.. NkT
sm
NkT
sm
ž NkT / Ž 73.
c1 2.25855 y1.50349 0.249434 1 where the physical term is that of the Soave-Redlich-Kwong
equation of state and the associating term is taken from SAFT
0
c 2 s y0.669270
c3 10.1576
1.40049
y15.0427
y0.827739
5.30827 0 0
l
l2
Ž 78.
equation ŽHuang and Radosz, 1990.. Kontogeorgis et al.
Ž1996. applied this new equation of state to pure components
and obtained good correlations of both vapor pressures and
saturated liquid volumes for primary-alcohols, phenol, tert-
butyl alcohol, triethylene glycol, and water.
In the second-order term ŽEq. 76., K h s is the isothermal Voutsas et al. Ž1997. applied the CPA equation of state to
compressibility of the hard-sphere fluid and was given by liquid-liquid equilibrium calculations in alcohol q hydro-
carbon mixtures. They used the conventional van der Waals
Ž 1yh . 4 one-fluid mixing rules for the attractive parameter a and the
K hss Ž 79. co-volume parameter b. Satisfactory results were obtained in
1q4h q4h 2
all cases using only a single temperature-independent binary
interaction parameter. They also compared the performance
The analytical expressions of the A1 and A 2 for Sutherland of CPA equation of state with that of the SRK and SAFT
fluids and Yukawa fluids over variable range were also given equations of state and concluded that the CPA equation pro-
by Gil-Villegas et al. Ž1997.. vided an improvement over the SRK equation and performed
The VR-SAFT equation broadens the scope of the original similar to the SAFT equation of state.
SAFT equation ŽHuang and Radosz, 1990. and improves the
chain contribution and the mean-field van der Waals descrip-
tion for the dispersion forces of the HS-SAFT treatment. Heteronuclear chains
Gil-Villegas et al. Ž1997. demonstrated the adequacy of the The preceeding equations of state can only be applied to
VR-SAFT equation in describing the phase equilibria of chain homogeneous chains, that is, chains composed of identical
molecules such as the n-alkanes and n-perfluoroalkanes. segments. Instead, many real chain-like fluids are heteroge-
Davies et al. Ž1998. showed the VR-SAFT provided a simple neous being composed of different alternating segments.
and compact equation of state for Lennard-Jones chains, and Furthermore, we are often interested in studying mixtures of
it is valid for ranges of density and temperature of practical different types of chains. Several equations of state have been
interest. McCabe et al. Ž1998. used the VR-SAFT equation proposed relatively recently that tackle the problem of het-
to predict the high-pressure fluid-phase equilibrium of binary eronuclear chains.
mixtures of n-alkanes and obtained good agreement with ex- Amos and Jackson Ž1992. used a bonded hard-sphere ap-
periment. Galindo et al. Ž1998. provided a detailed analysis proach to devise an equation of state for the compressibility
o f th e V R -S A F T a p p r o a c h fo r m ix t u r e s o f factor of hard spheres of different size. The compressibility
fluids with nonconformal intermolecular potentials, that is, factor for a fluid mixture of heteronuclear hard chain
which have attractive interactions of variable range. They ex- molecules is
amined the adequacy of the monomer contribution to the
Helmholtz function by comparing with computer simulation r
identify some of the reasons why an absolute quantitative the evaluation of alternatives, particularly if the alternatives
judgment is difficult: are more complicated.
Ž1. The very large number of equations of state ŽSandler, Partly because of the above factors, there are relatively few
1994. is an obvious factor that makes the assessment of dif- reports in the literature ŽSpear et al., 1971; Soave, 1972; Car-
ferent equations of state difficult. In this context, it should be nahan and Starling, 1972; Beret and Prausnitz, 1975; Peng
noted that the problem is compounded by the practice of and Robinson, 1976; Abbott, 1979; Martin, 1979; Elliott and
equation of state developers testing their equation of state Daubert, 1985; Kim et al., 1986; Han et al., 1988; Mainwaring
against experimental data, but not offering an identical com- et al., 1988; Sadus, 1992; Plackov et al., 1995; Plohl et al.,
parison with other equations of state. Nonetheless, the close 1998. comparing the predictive properties of two or more
inter-relationship between the different equations of state al- equations of state over the same range of physical conditions
lows some scope for a category by category evaluation. and experimental data.
Ž2. The accuracy of equations of state is often dependent Equations of state are used frequently to correlate experi-
on highly optimized equation of state parameters. These mental data rather than to provide genuine predictions. It is
equations of state parameters are typically tuned to a particu- well known that the van der Waals equation cannot be used
lar region of interest and breakdown outside this region. For to correlate accurately the vapor-liquid coexistence of pure
example, an equation of state tuned for atmospheric pres- fluids. In contrast, the addition of further adjustable parame-
sures is unlikely to predict high-pressure phenomena with ters and temperature dependence to the attractive term used
equal accuracy ŽSadus, 1992a.. in the Redlich-Kwong, Soave-Redlich-Kwong and Peng-
Ž3. Comparison of the effectiveness of the equation of state Robinson equations of state result in accurate correlations
for mixtures is hampered by the additional uncertainties in- ŽAbbott, 1979; Han et al., 1988. of the vapor pressure. How-
troduced by different mixture prescriptions, combining rules ever, a large improvement in the prediction of vapor pres-
and unlike interaction parameters. These topics are dis- sures can be obtained ŽPlackov et al., 1995. by using either
cussed in detail in the section discussing mixing rules. the Guggenheim or Carnahan-Starling van der Waals
Ž4. Users of equations of states often adopt a favorite ŽCSvdW. equations of state. The predictions of the Guggen-
equation of state with which they become expert in using. heim and CSvdW equations cannot compete with the accu-
This can result in considerable inertia to change that hinders racy obtained from empirical correlations with cubic equa-
The ©an der Waals mixing rules Impro©ed ©an der Waals mixing rules
The most widely used mixing rules are the van der Waals Many workers ŽAdachi and Sugie, 1986; Panagiotopoulos
one-fluid prescriptions and Reid, 1986; Stryjec and Vera, 1986a,b; Schwartzentruber
et al., 1987; Sandoval et al., 1989. have proposed modifica-
tions for the van der Waals prescriptions. A common ap-
as Ý Ý x i x j ai j Ž 86.
i j
proach is to include composition-dependent binary interac-
tion parameters to the a parameter in the van der Waals
bs Ý Ý x i x j bi j Ž 87. mixing rule and leave the b parameter rule unchanged. Some
i j of examples are summarized in Table 4.
as Ý Ý x i x j ai j Ž 88.
where g`E is the value of the excess Gibbs energy at infinite
1r2 pressure and can be calculated from ŽRenon and Prausnitz,
a i j s Ž a ii a j j . 1y l i j y m i j Ž x i y x j . Ž 89 .
1968.
Adachi and Sugie Ž1986. showed that their mixing rule can be n
Ý
n
k s1
x k G ki
Ž 92.
fi s
bi ¶ and b are
®i
ai A`E
b
fs • Ž 96.
as b Ý xi b
i
q
C
Ž 100.
®
®i ® a
fc s
ž / ž /ß
bi
y1
b
y1
bs
Ý Ý xi x j ž by
RT / ij
Ž 101.
A`E ai
The mixing rule ŽEq. 95. depends on the liquid-phase volume
1q
RT
y Ý xi
ž /
bi RT
of the mixture and individual components, and is a less re-
strictive assumption than the Huran-Vidal mixing rule. where C is a constant dependent on the equation of state
This basic concept ŽMollerup, 1986. was implemented by selected Žfor example, C is equal to 1r'2 ln Ž'2 y1. for the
Michelsen Ž1990.. Using a reference pressure of zero and the Peng-Robinson equation of state. and A`E is the excess
Soave-Redlich-Kwong equation of state ŽEq. 5., Michelsen Helmholtz function at infinite pressure, and
ŽMichelsen, 1990; Dahl and Michelsen, 1990. repeated the
matching procedure of Huron and Vidal resulting in the fol- a Ž 1y k i j . ai aj
lowing mixing rule called the modified Huron-Vidal first or-
der ŽMHV1.
ž by
RT / ij
s
2 ž bi y
RT / ž
q bj y
RT / Ž 102.
2
ß jj
2
ß jj
bi j s 1r4 Ž 2 1r3 . 4 Ž bii1r3 q bj1r3
j .
2
Ž bii q bj j .
1r3
Ž 108.
Sadus Ž1993. reported that the new combining rule ŽEq. 108.
In Eq. 104, the j i j Žalso commonly defined as 1y k i j . and is generally more accurate that either the Lorentz, arith-
z i j Žalso commonly defined as 1y l i j . terms are adjustable metic, or geometric combining rules.
parameters which are used to optimize agreement between
theory and experiment. The z i j term does not significantly
improve the analysis of high-pressure equilibria, and it can be Conclusions
usually omitted Ž z i j s1.. The j i j term is required, because it Considerable progress has been achieved in the develop-
can be interpreted as reflecting the strength of unlike inter- ment of equations of state. Many highly successful empirical
action except the simple mixtures of molecules of similar size. equations of state have been proposed that can be used to
This interpretation is supported by the fact that values of j i j calculate the phase behavior of simple fluids. However, a
obtained from the analysis of the critical properties of many more sophisticated approach is required for complicated
binary mixtures consistently decline with increasing size dif- molecules. To meet the challenge posed by large and compli-
ference between the component molecules as detailed else- cated molecules, equations of state are being developed in-
where ŽSadus, 1992a, 1994.. creasingly with an improved theoretical basis. These new