You are on page 1of 19

This article was downloaded by: [Memorial University of Newfoundland]

On: 11 November 2013, At: 20:44


Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Chemical Engineering Communications


Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/gcec20

EXCESS ENTHALPY AND VAPOR-LIQUID


EQUILIBRIA WITH THE MHV2 AND SOAVE
MIXING RULES
a a
Åsa U. Burman & Krister H. U. Ström
a
Department of Chemistry and Biological Engineering , Chalmers
University of Technology , Göteborg, Sweden
Published online: 21 Jun 2011.

To cite this article: Åsa U. Burman & Krister H. U. Ström (2011) EXCESS ENTHALPY AND VAPOR-
LIQUID EQUILIBRIA WITH THE MHV2 AND SOAVE MIXING RULES, Chemical Engineering Communications,
198:11, 1435-1452, DOI: 10.1080/00986440903287874

To link to this article: http://dx.doi.org/10.1080/00986440903287874

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions
Chem. Eng. Comm., 198:1435–1452, 2011
Copyright # Taylor & Francis Group, LLC
ISSN: 0098-6445 print=1563-5201 online
DOI: 10.1080/00986440903287874

Excess Enthalpy and Vapor-Liquid Equilibria


with the MHV2 and Soave Mixing Rules

ÅSA U. BURMAN AND KRISTER H. U. STRÖM


Department of Chemistry and Biological Engineering, Chalmers
Downloaded by [Memorial University of Newfoundland] at 20:44 11 November 2013

University of Technology, Göteborg, Sweden

Calculations and predictions of excess enthalpy (HE) and vapor-liquid equilibrium


(VLE) were performed using the Gibbs energy mixing rules MHV2 and a modifi-
cation of it by Soave. The Soave-Redlich-Kwong equation of state was combined with
the UNIQUAC equation. Four sets of parameters estimated in the UNIQUAC model
were used for each of seven binary systems: the first estimated from VLE data, the
second and the third estimated from HE data for two versions of the UNIQUAC
equation, and the fourth estimated from both HE and VLE data simultaneously.
It was found that HE calculations can be performed with the mixing rules; the average
relative errors fell from around 200% for the conventional mixing rule to around 60%
for MHV2 combined with DECHEMA UNIQUAC parameters and was as little as
20% when the UNIQUAC parameters had been estimated from HE and VLE data
simultaneously. However, the approach suffers from the same shortcomings as far
as cross-prediction between HE and VLE data is concerned, as does the UNIQUAC
equation used alone. There is a discrepancy between values obtained with the mixing
rule and those obtained with the UNIQUAC equation directly. This discrepancy is
smaller for the Soave modification of the mixing rule.

Keywords Equation of state; Excess enthalpy; Mixing rule; Vapor-liquid


equilibria

Introduction
The last decade, new Gibbs energy mixing rules for cubic equations of state (GE-eos
models) have been developed (Huron and Vidal, 1979; Dahl and Michelsen, 1990;
Holderbaum and Gmehling, 1991; Wong and Sandler, 1992). With these, a mixture
parameter is obtained through a formal match between the expression for Gibbs
excess energy (GE) from the equation of state and that from a GE model (or activity
model). The mixing rules have proven to work excellently in calculations of
vapor-liquid equilibrium (VLE), even for mixtures containing polar compounds.
The aim of this study is to investigate if they can be used successfully in excess
enthalpy calculations. The issue is of both practical and theoretical interest. As HE
is related to the temperature dependency of phase equilibria, simultaneous compu-
tation of VLE and HE may be seen as a test of the thermodynamic consistency of
the model. The influence of the approximating equation, the GE model parameters,

Address correspondence to Krister H. U. Ström, Department of Chemistry and


Biological Engineering, Chalmers University of Technology, SE-412 96 Göteborg, Sweden.
E-mail: krister@chalmers.se

1435
1436 Å. U. Burman and K. H. U. Ström

and the parameters that describe the temperature dependence for the pure
component will be addressed. The investigation includes a comparison with the
conventional van der Waals one-fluid mixing rule, as a reference to a typical
engineering situation.
Two GE mixing rules with reference pressure zero are used in this study:
. MHV2 (Dahl and Michelsen, 1990), which is simple and much discussed.
. A modification by Soave (1992) of MHV2 that was designed to give a mixing rule
that was asymptotically correct as the temperature approaches infinity.
The equation of state chosen for this study is the Soave-Redlich-Kwong (SRK,
Downloaded by [Memorial University of Newfoundland] at 20:44 11 November 2013

Soave (1972)) with an expression for the temperature dependency of the energy para-
meter (a) from Mathias and Copeman (1983). This has been the most frequently used
equation in other MVH2 studies. The UNIQUAC equation (universal quasi chemi-
cal, Abrams and Prausnitz (1975)) is used as the GE model. Calculations were carried
out for seven binary systems with four parameter sets regressed in the UNIQUAC
model: one from the DECHEMA data series (Gmehling et al., 1977–1996) estimated
from VLE data, two estimated from HE data in two versions of the UNIQUAC
equation, and one estimated from HE and VLE data simultaneously. The binary
systems were chosen so that different kinds of HE behavior would be represented.
Systems for which there were both VLE and HE data available at two (or more)
temperatures in the DECHEMA data collection series (Christensen et al., 1984–
1991; Gmehling et al., 1977–1996) were preferred.

Theory
Vapor-liquid equilibrium can be modeled by two different approaches. With an
equation of state, the same model is used for both the liquid and the vapor phase.
The other approach uses a GE model for the liquid phase and an equation of state
(such as the ideal gas law) for the gas phase.
When an equation of state is used for calculations of properties for mixtures, the
model parameters are computed using a mixing rule. The most common one is the
van der Waals one-fluid mixing rule (also called the conventional or the classical):
XX pffiffiffiffiffiffiffiffi
a¼ xi xj aij aij ¼ ai aj ð1  kij Þ
i j
XX bi þ b j ð1Þ
b¼ xi xj bij bij ¼ ð1  lij Þ
i j
2

These mixing rules have a sound theoretical background and they have proven to
work well for mixtures that can be described by regular solution theory. When used
with a cubic equation of state, they cannot handle mixtures with polar compounds
nor mixtures for which there is a large difference in molecular size or critical tem-
perature (Sandler et al., 1994). With the Peng-Gasem-Robinson (PGR) segment-
segment interaction equation of state, asymmetric mixtures can be modeled applying
one-fluid mixing rules (Row et al., 2008). For HE calculations, the classical mixing
rules tend to be unreliable, and the results are extremely sensitive to the value of
the binary interaction parameter kij (Christensen and Rowley, 1988). In this study,
one interaction parameter kij ¼ kji will be estimated and used. The lij ¼ lji parameter
HE and VLE with MHV2 and Soave 1437

will be set to zero. A more elaborate version could have been applied. However, the
intention was not to investigate the possibilities of the van der Waals mixing rules,
but rather to use them as a reference to an ordinary engineering situation.
For polar systems at low pressure, GE models give a good representation of VLE.
Cross-prediction, that is prediction of VLE data from a GE model fitted to HE data or
vice versa, is typically not reliable or quantitative (Christensen and Rowley, 1988).
However, satisfactory results have been obtained for some systems (Pando et al.,
1983). With simultaneous regression of VLE and HE data, both properties can be
modeled (Burman and Ström, 2007; Kundu and Bandyopadhyay, 2007). This
approach has been applied even to solid liquid equilibrium and enthalpy in ice slurries
Downloaded by [Memorial University of Newfoundland] at 20:44 11 November 2013

(Cézac et al., 2009). As one model is used for the liquid phase and another for the gas
phase, problems arise when the GE models are to be used near the critical point. The
GE mixing rules were developed as an attempt to incorporate the information
contained in the GE model into a cubic equation of state. The basic idea behind rules
like MHV2 is to match the GE expression from the equation of state, at zero pressure,
with GE from a GE model. In order to obtain an explicit expression for the a
parameter, an approximating equation q(a) is applied. The dimensionless parameter
a is equal to a=bRT. For MHV2 it is a second-order polynomial expression
q(a) ¼ q0 þ q1a þ q2a2, where q1 ¼ 0.478 and q2 ¼ 0.0047. That is:

! !  
X X GE X b
2
q1 a  zi aii þ q2 a  zi a2ii ¼ þ zi ln ð2Þ
i i
RT i
b ii

Dahl and Michelsen (1990) found that MHV2 was successful in VLE calculations for
polar compounds using the UNIQUAC equation with parameters from DECHEMA
or the UNIFAC method. The model performed well in liquid-liquid (LLE) and
vapor-liquid-liquid equilibria (VLLE) calculations, provided that the GE model
gave a correct representation of the system (Dahl et al., 1991; Abdel-Ghani and
Heidemann, 1996). MHV2 has been applied to VLE and LLE predictions for solu-
tions containing a salt. Again, the results depend heavily on the GE model (Dahl and
Macedo, 1992). It could also successfully correlate VLE and VLLE of near-critical
and supercritical carbon dioxide-methanol-water (Yoon et al., 1993). The mixing
rule has been used for predictions of gas solubilities, Henry’s law coefficients, and
c1 (Dahl et al., 1991).
However, some serious criticism has been raised against the zero-pressure mixing
rules. Orbey and Sandler (1995) criticize them because they do not produce second
virial coefficients that are quadratic in compositional dependency; a known bound-
ary condition should be regarded and this violation will affect the fugacity coeffi-
cients computed for the vapor phase. One problem with MHV2 is that the fitting
of GEs for the liquid phase influences the vapor phase calculations too much, pro-
ducing too high GEs for that phase (Michelsen and Heidemann, 1996).
Several studies on extrapolation of VLE from low-pressure data to high pres-
sures have been presented. The MHV1 mixing rule (with a first-order linear approxi-
mation as approximating equation) was found to handle extrapolations better than
MHV2 (Fischer and Gmehling, 1996). It was puzzling that the first-order approxi-
mation, MHV1, produces better extrapolations than the second-order approxi-
mation, MHV2. Michelsen (1996) found that a very common inadequacy in the
temperature dependence of the GE models happens to be canceled out partially by
1438 Å. U. Burman and K. H. U. Ström

the lack of fit between the true q(a) and that computed with MHV1 at high tempera-
tures. This discrepancy in match is smaller for MHV2. According to Heidemann
(1996) reliable extrapolations are possible only when the GE model is accurate in
the temperature and pressure ranges of the extrapolation. Deviations between the
GE computed with the mixing rule and with the GE model are found to be caused
mainly by the approximating equation q(a) (Kalospiros et al., 1995). Plotting (@q=
@a) versus a, Kalospiros and coworkers visualized the problem. The MHV1 rule fits
only at a ¼ 13, the point it was fitted to, and the difference between the approximat-
ing equation and the real q(a) approaches infinity as the temperature approaches
infinity. Michelsen (1996) showed that if the difference in ai for the compounds in
Downloaded by [Memorial University of Newfoundland] at 20:44 11 November 2013

the mixture is large, i.e., for asymmetric systems, these errors quickly increase even
at ordinary temperatures.
According to the analysis of Kalospiros et al. (1995), an approximating equation
proposed by Soave (1992) resulted in an improved fit to the GE model. The Soave
q(a) expression is derived to be asymptotically correct when the temperature
approaches infinity:
 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
a lnð2Þ þ ða lnð2ÞÞ2 þ 69
qðaÞ ¼ 3:365  ð3Þ
2
Soave (1992) found that his expression performed as well in low-pressure VLE
calculations as did MHV2. For both MHV2 and Soave, a linear mixing rule is used
for the b parameter of the mixture.

Equations and Parameters


The equation of state most frequently used in MHV2 studies is the SRK equation of
state:
RT a
P¼  ð4Þ
v  b vðv þ bÞ

The pure component parameters are calculated as:


    2  3 2
ðRTci Þ2 1=2 1=2 1=2
aii ¼ 0:4286 1 þ c1 1  Tri þ c2 1  Tri þc3 1  Tri ð5Þ
Pci

RTci
bii ¼ 0:08664 ð6Þ
Pci
The expression chosen for the temperature dependency of the a parameter was
proposed by Mathias and Copeman (1983). It is applied in order to ensure a good
representation of pure component vapor pressure data. Dahl and Michelsen
(1990) obtained the so called Mathias-Copeman parameters in Equation (5), c1,
c2, and c3, by fitting the equation of state to the vapor pressure correlations of
Daubert and Danner (1986). In the work presented here, the Mathias-Copeman
parameters were estimated from experimental vapor pressure data published by
Boublı́k et al. (1984). The ratio of the difference in experimental and calculated
pressure to the experimental pressure was minimized. The resulting parameters are
presented in Table I. The correlations between the parameters are very high.
HE and VLE with MHV2 and Soave 1439

Table I. Mathias-Copeman parameters

Compound c1 c2 c3
Cyclohexane 0.788 0.021 0.251
1-Propanol 1.198 1.616 3.156
2-Methyl-1-propanol 1.083 2.033 2.926
Ethanol 1.374 0.591 2.188
Methanol 1.402 0.553 0.285
Water 1.057 0.465 0.345
Benzene 0.811 0.244 0.750
Downloaded by [Memorial University of Newfoundland] at 20:44 11 November 2013

Acetone 0.986 0.383 0.526


Tetrachloromethane 0.785 0.215 0.588
Trichloromethane 0.841 0.390 0.827

The incorporated GE model was the UNIQUAC equation. For the original
UNIQUAC equation, the molecular structure constants r and q are obtained from
the DECHEMA data series (Gmehling et al., 1977–1996). Anderson and Prausnitz
(1978) modified the equation by introducing a new external area parameter, q0 , in
the residual part of the equation for water and lower alcohols. In calculations with
the Anderson modification of the UNIQUAC equation, the same r and q values as
for the original equation were used, and the q0 values were taken from Anderson and
Prausnitz (1978).
The VLE calculations were performed as bubble point calculations following the
algorithm given in van Ness and Abbott (1982). Expressions for the compositional
derivative of the a parameter are given in the appendix. The HE calculations are also
described in the appendix.
In order to compare the MHV2 mixing rule and the one proposed by Soave,
predictions of HE were made with UNIQUAC parameters obtained from the
DECHEMA series (called DECHEMA in the tables) for both mixing rules. With
the MHV2 mixing rule, calculations were also performed with UNIQUAC para-
meters estimated from HE data, both for the original UNIQUAC equation (called
HE in the tables) and for the Anderson modification (called Anderson in the tables).
The mixing rule proposed by Soave was also combined with UNIQUAC parameters
estimated from HE and VLE data simultaneously (called HEGE in the tables). The
UNIQUAC parameters are found in Table II, and the procedure for their estimation
is described elsewhere (Burman and Ström, 2007).
For calculations with the van der Waals mixing rule, kij parameters for the
binary mixtures were estimated from the experimental VLE data listed in
Table III. The difference in the square of the fugacity between the liquid and the
vapor phase was minimized. The modified Marquardt algorithm (Bard, 1974) was
used for the minimization. The initial estimate of kij was 0.01, but the parameter esti-
mation generally converged to the same kij value for initial kij values ranging from 
0.1 to 0.1. The resulting values for the kij parameter are found in Table III. The lar-
gest root mean square deviation (RMSD) of the fugacity differences between the
phases was observed for acetone-water. This was expected for a system with a large
difference in pure component covolume (bii) parameters (see, for example, Knudsen
et al. (1993)). For the other systems, the fit is closer but there are definite trends in
the residuals. In the tables for the results, results from the van der Waals mixing rule
Downloaded by [Memorial University of Newfoundland] at 20:44 11 November 2013

Table II. UNIQUAC binary parameters

Acetone- Acetone- Benzene-2-


Model= Cyclohexane-1- Acetone- tetrachloro- trichloro- methyl-1- Methanol- Ethanol-
system Parameter propanol water methane methane propanol water water
DECHEMA n 12 [J=mol] 3894 1747 713.2 2974 2691 1375 547.8
n 21 [J=mol] 835.0 177.6 2010 4668 367.3 2119 381.7
Original W12 [J=mol] 5349 1907 2156 1350 14975 3315 5058
UNIQUAC W21 [J=mol] 5718 782.8 3027 1142 5035 3527 4822
HE l12 [J=mol K] 10.48 13.55 3.296 1.994 122.5 1.491 9.399
l21 [J=mol K] 17.16 21.43 3.371 5.869 3.584 7.301 18.33

1440
Anderson W12 [J=mol] 1.436  104 3150 1.499  104 3437 6625
modification W21 [J=mol] 3505 160.2 1937 3638 6715
l12 [J=mol K] 57.89 15.46 128.8 0.6222 15.12
l21 [J=mol K] 13.40 17.73 13.71 10.22 29.98
HEGE W12 [J=mol] 4085 2902 779.3 997.3 1723 1653 822.4
W21 [J=mol] 977.9 563.8 2146 22.18 98.44 2436 2190
l12 [J=mol K] 3.504 16.07 1.260 9.302 15.92 8.280 13.09
l21 [J=mol K] 2.914 12.20 1.776 4.486 7.868 28.91 35.06
For temperature-dependent parameters: n ij ¼ Wij þ lij(T  T Ref). The reference temperature (T Ref) used was 300 K for all systems.
HE and VLE with MHV2 and Soave 1441

Table III. Binary interaction parameters for the van der Waals one-fluid mixing rule
and observed root-mean-square deviation of the objective function

Mixture kij RMSD Reference


Cyclohexane-1-propanol 0.072 0.09 Smirnova and Kurtynina, 1969
Acetone-water 0.251 0.21 Lieberwirth and Schubert, 1979
Acetone-tetrachloromethane 0.0566 0.03 Brown and Smith, 1957
Acetone-trichloromethane 0.061 0.01 Kudryavtseva and Susarev, 1963
Benzene-2-methyl-1-propanol 0.081 0.04 Brown et al., 1969
Methanol-water 0.082 0.02 Ratcliff and Chao, 1969
Downloaded by [Memorial University of Newfoundland] at 20:44 11 November 2013

Ethanol-water 0.102 0.04 Mertl, 1972

with one binary interaction parameter will be referred to as kij-VLE. Calculations


were also performed in which kij was set to zero, referred to as kij ¼ 0 in the tables.
VLE and HE calculations were performed for seven binary mixtures and compared
to experimental data from the DECHEMA series (Christensen et al., 1984–1991;
Gmehling et al., 1977–1996). The systems chosen represent different HE behavior.
E
The endothermal systems are acetone-tetrachloromethane (HMax  300 J=mol), cyclo-
E E
hexane-1-propanol (HMax  700 J=mol), and benzene-2-methyl 1-propanol (HMax 
E
1600 J=mol). For these, the H versus composition curve is moderately skewed. One
E
of the exothermal systems, acetone-trichloromethane (HMin   2000 J=mol), is fairly
E
symmetric, whereas the other, methanol-water (HMin   900 J=mol) is strongly
skewed (minimum at xmethanol  0.2). Acetone-water is exothermal at high water
contents and endothermal at low. The HE behavior of ethanol-water changes from
exothermal at low temperatures to endothermal at high temperatures.

Results and Discussion


It was possible to use the equation of state-GE mixing rule approach to calculate HE
data and to predict it from VLE data. The results of the calculations are presented in
Table IV as root mean square deviation (RMSD), which is defined as the square root
of the arithmetical average of the squared deviation between the experimental and
the calculated values. The HE curves predicted by MHV2 combined with
DECHEMA parameters are qualitatively correct; they typically reproduce the shape
of the curve, but the values are generally too low. Astonishingly, the shape of the
curve computed with MHV2 was often correct even when the UNIQUAC model
itself with the same parameters did not catch the HE behavior (Figure 1). MHV2
with DECHEMA parameters often gave HE values closer to the experimental values
than did the UNIQUAC model upon which MHV2-DECHEMA was based. The
phenomenon is most likely a fortunate cancellation of errors. Fischer and Gmehling
(1996) report a similar behavior for VLE calculations with the MHV1 and predictive
Soave-Redlich-Kwong (PSRK) mixing rules. Michelsen (1996) concludes that a com-
mon error in the temperature dependence of the GE mixing rule appears to be can-
celed out by errors in the mixing rule. The RMSD between HE calculated with the
equation of state-GE model and the GE model alone are given in Table V. Dahl
and Michelsen (1990) fitted the approximating equation q(a) in MHV2 to a values
ranging from 8 to 18. The a value for water at 25 C is above 18. As may be seen from
Downloaded by [Memorial University of Newfoundland] at 20:44 11 November 2013

Table IV. Root-mean-square deviation between calculated and experimental HE [kJ=mol]

van der Waals MHV2 Soave Experimental data


Mixing rules
MixturenParameters kij ¼ 0 kij-VLE DECHEMA HE Anderson DECHEMA H EG E T [ C] Source
Ccyclohexane-1-propanol 0.28 1.4 0.38 0.48 0.08 0.24 0.05 25 III 1 570
0.21 1.2 0.44 0.34 0.06 0.40 0.19 50 III 1 572
Acetone-water 4.9 1.5 0.28 0.38 0.37 0.51 0.14 35 III 1 518
3.5 0.80 0.16 0.13 0.13 0.17 0.05 90 III 1 520
Acetone-tetrachloromethane 0.22 0.61 0.07 0.02 x 0.05 0.02 45 III 1 19
0.15 0.56 0.07 0.01 x 0.08 0.02 25 Fuchs et al., 1984
Acetone-trichloromethane 1.3 0.53 0.16 0.04 x 0.15 0.05 50 III 1 150
1.7 0.71 0.12 0.02 x 0.09 0.03 70 III 1 149

1442
Benzene-2-methyl-1-propanol 1.3 0.28 1.0 0.12 0.20 0.96 0.09 45 III 1 749
0.99 0.61 0.84 0.72 0.63 0.68 0.15 25 III 1 750
Methanol-water 2.0 0.89 0.40 0.05 0.06 0.51 0.04 25 III 1 304
2.0 0.75 0.28 0.08 0.10 0.38 0.06 53.5 III 1 305
Ethanol-water 2.2 0.65 0.19 0.09 0.10 0.26 0.06 70 III 1 458
2.9 1.2 0.59 0.19 0.25 0.74 0.09 25 III 1 457
1.8 0.33 0.09 0.19 0.18 0.10 0.15 110 III 1 459
Average 1.8 0.88 0.33 0.21 x 0.34 0.08
The Source column refers to the volume, part, and page number in the DECHEMA data series where the experimental data may be found. The
UNIQUAC parameters are found in Table II.
x ¼ the Anderson modification was not applicable for these systems.
HE and VLE with MHV2 and Soave 1443
Downloaded by [Memorial University of Newfoundland] at 20:44 11 November 2013

Figure 1. Excess enthalpy for acetone (1)-water (2) at 35 C. Experimental data (^) from
Belosouv and Sokolova (1966). With DECHEMA parameters the MHV2 mixing rule (solid
line) produces a better prediction than does the UNIQUAC equation (dashed line) used in
the mixing rule.

Table V, the discrepancies in HE between the combined equation of state-GE model


and the incorporated GE model are large for systems containing water.
When the MHV2 and the Soave mixing rules are compared, MHV2 appears to
be the better one. The arithmetical average of the RMSD from the experimental
values for all the data points in the study is 325 J=mol instead of 339 J=mol
(Table IV). However, the difference can be explained by the discrepancy described
above. The Soave mixing rule produced HE values closer to the GE model incorpor-
ated. For all the systems studied, the total arithmetical average of the RMSD from
the UNIQUAC equation decreased from 144 J=mol for the MHV2 mixing rule to
50 J=mol for the Soave mixing rule (Table V). For VLE calculations (Table VI),
the MHV2 and the Soave mixing rules with DECHEMA parameters were similar,
as reported by Soave (1992).
The MHV2 mixing rule combined with the original UNIQUAC equation with
parameters estimated from HE data (MHV2-HE) generally yielded HE values closer
to the experimental ones than the MHV2-DECHEMA model did (Table IV).
However, the improvement in the fit of the GE model to the experimental data
was not always reflected in the combined model because of the discrepancy between
the combined model and the GE model (see Figure 2).
Exchanging the original UNIQUAC equation for the Anderson modification,
still with parameters estimated from HE data, in the MHV2 mixing rule produced
similar results. The Anderson modification was much better for the cyclohexane-1-
propanol system.
The MHV2 mixing rule based on GE models that had parameters regressed from
H data alone (MHV2-HE and MHV2-Anderson) did not give reliable VLE predic-
E

tions (Table VI). The direct use of these GE models also failed to cross-predict VLE
accurately (Burman and Ström, 2007). Thus, it seems that the GE mixing rule
approach suffers from the same shortcoming on this point as the GE models do.
The UNIQUAC equation with parameters estimated simultaneously from HE
and VLE data (HEGE) combined with the Soave mixing rule produced the lowest
Downloaded by [Memorial University of Newfoundland] at 20:44 11 November 2013

Table V. Root-mean-square deviation between HE [kJ=mol] calculated with the GE mixing rules and with the UNIQUAC equation used
in the mixing rule

MHV2 Soave Experimental data


Mixing rule
MixturenParameters DECHEMA HE Anderson DECHEMA HEGE T [ C] Source
Cyclohexane-1-propanol 0.19 0.65 0.06 0.05 0.06 25 III 1 570
0.08 0.70 0.03 0.03 0.05 50 III 1 572
Acetone-water 0.35 0.38 0.37 0.09 0.09 35 III 1 518
0.12 0.11 0.11 0.08 0.08 90 III 1 520
Acetone-tetrachloromethane 0.012 0.014 x 0.004 0.005 45 III 1 19
0.02 0.005 x 0.013 0.013 25 Fuchs et al. (1984)
Acetone-trichloromethane 0.16 0.019 x 0.011 0.011 50 III 1 150

1444
0.07 0.018 x 0.032 0.033 70 III 1 149
Benzene-2-methyl-1-propanol 0.14 0.17 0.26 0.05 0.05 45 III 1 749
0.29 0.69 0.59 0.19 0.07 25 III 1 750
Methanol-water 0.13 0.037 0.054 0.019 0.016 25 III 1 304
0.14 0.085 0.099 0.037 0.038 53.5 III 1 305
Ethanol-water 0.20 0.19 0.20 0.04 0.04 25 III 1 457
0.13 0.07 0.07 0.05 0.06 70 III 1 458
0.05 0.19 0.20 0.05 0.05 110 III 1 459
Average 0.14 0.26 x 0.050 0.048
The Source column refers to the volume, part, and page number in the DECHEMA data series where the experimental data may be found. The
UNIQUAC parameters are found in Table II.
x ¼ the Anderson modification was not applicable for these systems.
Downloaded by [Memorial University of Newfoundland] at 20:44 11 November 2013

Table VI. Average absolute deviation between the calculated vapor phase composition and the experimental (%)

van der Waals MHV2 Soave Experimental data


Mixing rule
MixturenParameters kij ¼ 0 kij-VLE DECHEMA HE Anderson DECHEMA HEGE T [ C] Source
Cyclohexane-1-propanol 7.7 2.6 1.1 – 13 1.1 1.3 25 I 2c 544
13 4.2 1.7 – 20 1.7 1.6 55 I 2a 582
12 3.9 1.5 – 19 1.5 1.9 65 I 2a 583
Acetone-water – 4.1 0.9 2.8 2.3 0.9 0.9 35 I 1a 195
– 6.2 0.7 8.7 7.2 0.6 0.9 100 I 1a 194
Acetone-tetrachloromethane 5.6 1.5 0.6 8.3 x 0.6 0.8 45 I 3 84
Acetone-trichloromethane 6.6 1.1 0.6 2.8 x 0.6 0.3 15 I 3 108
6.3 0.9 0.6 2.8 x 0.6 0.3 20 I 3 109

1445
Benzene-2-methyl-1-propanol 3.8 1.0 1.2 3.2 5.4 1.2 1.7 25 I 2b 282
6.1 0.8 0.3 5.7 7.7 0.3 1.3 45 I 2b 283
Methanol-water 27 4.9 1.6 12 9.8 2.1 1.0 25 I 1 42
10 1.2 0.4 6.6 5.2 0.4 0.5 25 I 1a 49
21 2.2 0.7 14 11 0.8 1.4 60 I 1 41
Ethanol-water 22 3.5 0.8 19 19 0.8 1.8 25 I 1a 126
21 4.2 0.9 21 21 0.8 1.9 25 I 1a 125
19 2.6 0.9 21 21 0.8 1.4 55 I 1 173
Average – 3.2 0.9 – – 0.9 1.2
The Source column refers to the volume, part, and page number in the DECHEMA data series (Gmehling et al., 1977–1996) where the experimental
data may be found.
x ¼ the Anderson modification was not applicable for these systems.
– ¼ VLE calculations did not converge.
1446 Å. U. Burman and K. H. U. Ström
Downloaded by [Memorial University of Newfoundland] at 20:44 11 November 2013

Figure 2. Excess enthalpy for acetone (1)-water (2) at 35 C. Experimental data (^) from
Belosouv and Sokolova (1966). When UNIQUAC parameters fitted to HE data are used,
the discrepancy between HE computed with the GE model alone (dashed line) and with the
MHV2 mixing rule (solid line) is still large and deteriorates the results for the mixing rule.

average RMSD (81 J=mol) for HE calculations. It could also model the more com-
plicated HE behavior shown by some of the systems (Figure 3). The discrepancy
between the UNIQUAC equation and the mixing rule was similar for the
Soave-DECHEMA and Soave-HEGE models (Table V). The errors in calculated
VLE were somewhat larger than those computed with the DECHEMA parameters
(Table VI). The average absolute deviation in the vapor phase composition increased
from 0.9 to 1.2 molar%.
Theoretically, the pure component a(T) expression (the Mathias-Copeman para-
meters) would also be important to the calculation of HE. Following Michelsen and
Heidemann (1996), excess enthalpies for the system pentane-methanol were

Figure 3. Excess enthalpy for ethanol-water at 70 C. The solid line is computed with Soave’s
mixing rule and HEGE parameters and the dashed line with UNIQUAC and HEGE para-
meters. Experimental data (^) from Larkin (1975).
HE and VLE with MHV2 and Soave 1447
Downloaded by [Memorial University of Newfoundland] at 20:44 11 November 2013

Figure 4. Excess enthalpy for pentane (1) methanol (2) at 25 C computed with the MHV2
mixing rule, the Wilson equation, and binary interaction parameter from the DECHEMA
series. Three sets of Mathias-Copeman parameters were used: estimated in this study from
experimental data (dashed line), from Dahl and Michelsen (1990) (solid line), and from Huang
and Sandler (1993) (dotted line). Experimental data (^) from Collins et al. (1980).

computed using MHV2 in combination with the Wilson equation. Three different
sets of Mathias-Copeman parameters were used: first, parameters for pentane found
in this study from experimental data in Boublı́k et al. (1984) (giving c1 ¼ 0.8880,
c2 ¼ 0.2954, c3 ¼ 0.5850, the parameters for methanol are found in Table I); second,
parameters of Dahl and Michelsen (1990) estimated from the vapor pressure corre-
lation of Daubert and Danner (1986); and a third set from Huang and Sandler
(1993). It is not clear how the parameters used by Huang and Sandler were obtained.
As shown in Figure 4, the choice of parameters does influence the results signifi-
cantly. The difference was smaller for other binaries, most likely because the para-
meters were more similar for them.

Figure 5. Excess enthalpy for cyclohexane (1)-1-propanol (2) at 25 C. Experimental data (^)
from Vesely et al. (1979). The prediction of HE with the van der Waals mixing rule is not
improved when kij is estimated from VLE data (solid line) rather than set to zero (dashed line).
1448 Å. U. Burman and K. H. U. Ström

HE curves computed with the van der Waals mixing rules were symmetrical
and, thus, could not reproduce skewed HE versus composition curves. The
introduction of a binary interaction parameter estimated from VLE data did
improve the HE predictions for most of the systems. The average RMSD
from the experimental data decreased from 1767 to 877 J=mol. For some systems,
however, the results deteriorated (compare columns van der Waal kij ¼ 0 and kij-
VLE in Table IV and see Figure 5). According to Christensen and Rowley (1988),
HE calculations with the van der Waals mixing rule are very sensitive to the kij-
value.
As might be expected for mixtures with polar compounds, the bubble point
Downloaded by [Memorial University of Newfoundland] at 20:44 11 November 2013

calculations with the van der Waals mixing rule and no interaction parameter were
not successful: the calculations would sometimes even fail to converge (see Table VI
column van der Waals, kij ¼ 0). Introducing one interaction parameter improved the
results significantly (Table VI, column van der Waals, kij-VLE).

Conclusion
VLE and HE calculations were performed using the SRK equation of state with
MHV2 and Soave mixing rules combined with four different sets of UNIQUAC
parameters regressed in the UNIQUAC model. The MHV2 mixing rule with
UNIQUAC parameters from DECHEMA produces qualitatively correct HE
curves, sometimes closer to the experimental values than those predicted by the
UNIQUAC equation alone. The Soave q(a) function produces HE values closer
to those of the incorporated GE model. There is still a discrepancy between HE
values computed with the equation of state model and the GE model for both
mixing rules. The GE-eos approach seems to inherit the problems of the GE mod-
els as far as cross-prediction between HE and VLE is concerned. Simultaneous
estimation of parameters from VLE and HE data enables calculations of both
properties using one and the same model, but reduces the precision of the GE
model to some extent.
Improved GE models would be necessary if VLE and HE data are to be calcu-
lated successfully from the same model. More experimental measurements of HE
would be needed.

Nomenclature
a equation of state parameter, Nm4=mol2
b equation of state parameter, m3=mol
c Mathias-Copeman parameters
G molar Gibbs free energy, J=mol
H molar enthalpy, J=mol
k binary interaction parameter
l binary interaction parameter
n amount of substance, mol
P pressure, Pa
q external area parameter (UNIQUAC)
q auxiliary function (Equation (2))
R gas constant, J=mol K
HE and VLE with MHV2 and Soave 1449

T temperature, K
v molar volume, m3=mol
x molar fraction, liquid phase
Z compressibility factor
Greek Letters
a parameter (equation of state)
c activity coefficient
W binary interaction parameter (UNIQUAC), J=mol
H contact area fraction (UNIQUAC),
Downloaded by [Memorial University of Newfoundland] at 20:44 11 November 2013

l binary interaction parameter (UNIQUAC), J=mol K


n binary interaction parameter (UNIQUAC), J=mol
s parameter (UNIQUAC)
Superscripts
E excess property
O ideal state
Subscripts
c critical
i component
j component
k component
r reduced

Appendix
When VLE calculations are performed with an equation of state, the compositional
derivative of the a parameter can be obtained as:
      
@na @n @a @a @q
¼a þn ¼aþn ðA:1Þ
@ni @ni @ni @q @ni

where:
   
@q 1 b bi
¼ qðaÞ þ qðai Þ þ lnðcÞ þ ln þ 1 ðA:2Þ
@ni n bi b

With the MHV2 mixing rule

@a 1
¼ ðA:3Þ
@q q1 þ 2q2 a

whereas the Soave mixing rules gives

@a 1 17:25
¼  ðA:4Þ
@q lnð2Þ lnð2Þð3:365  qÞ2
1450 Å. U. Burman and K. H. U. Ström

When excess enthalpy is computed from the UNIQUAC model alone, the Gibbs-
Helmholtz relation is applied to the UNIQUAC equation. For a binary mixture, this
results in the following expression:
 0    
E 2 q1 x1 H02 @s21 q02 x2 H01 @s12
H ¼ RT þ 0 ðA:5Þ
H01 þ H02 s21 @T H2 þ H01 s12 @T

q0 xi n ij
H0i ¼ P i 0 sij ¼ eRT
qk xk
Downloaded by [Memorial University of Newfoundland] at 20:44 11 November 2013

For the SRK equation of state:


 @a  
H  Ho T @T a b
¼Z1þ ln 1 þ ðA:6Þ
RT RTb v
The compressibility, Z, and the volume, v, are the liquid roots from the equation of
state. The pressure used for the calculations was 1 atm.
With the van der Waals mixing rule, the derivative of a is:
          
@a 2 @a1 2 @a2 1  kij @a2 @a1
¼ x1 þ x2 þ x1 x2 pffiffiffiffiffiffiffiffiffi a1 þ a2 ðA:7Þ
@T @T @T a1 a2 @T @T

When the GE-mixing rules are used, the derivative is computed from the derivative
for a:
      
@a a @a a @a @q
¼ þ RTb ¼ þ RTb ðA:8Þ
@T T @T T @q @T
The derivative of a with respect to q are Equations (A.3) and (A.4) for the MHV2
and Soave expressions, respectively. The temperature derivative of q for the
MHV2 mixing rule is:
0  E 1
  G
@ RT X    2
@q @ A @ai @ai
¼ þ q1 x i þ q2 xi ðA:9Þ
@T @T i
@T @T

and for the Soave mixing rule:


  2 3
  @ GE
X xi ln 2 @ai 6
@q RT ai ln 2 7
¼ þ 41 þ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi5 ðA:10Þ
@T @T 2 @T 2
i ðai ln 2Þ þ69

In the two equations above, ai refers to the a value of pure component i.


The derivative of the pure component a parameter with respect to temperature is:

      2
@ai ðRTci Þ2 1=2 1=2
¼  0:4286 1 þ c1 1  Tri þ c2 1  Tri
@T Pci
 3     2  ðA:11Þ
1=2 1=2 1=2 1=2
þ c3 1  Tri c1 þ c2 1  Tri þ c3 1  Tri Tri
HE and VLE with MHV2 and Soave 1451

References
Abdel-Ghani, R. M., and Heidemann, R. A. (1996). Comparison of delta G excess mixing rules
for multi-phase equilibria in some ternary systems, Fluid Phase Equilib., 116, 495–502.
Abrams, D. S., and Prausnitz, J. M. (1975). Statistical thermodynamics of liquid mixtures: A
new expression for the excess Gibbs energy of partly or completely miscible systems,
AIChE J., 21, 116–128.
Anderson, T. F., and Prausnitz, J. M. (1978). Application of the UNIQUAC equation to
calculation of multicomponent phase equilibria. 1. Vapor-liquid equilibria, Ind. Eng.
Chem. Process Des. Dev., 17, 552–561.
Bard, Y. (1974). Nonlinear Parameter Estimation, Academic Press, New York.
Downloaded by [Memorial University of Newfoundland] at 20:44 11 November 2013

Belousov, V. P., and Sokolova, E. P. (1966). Heats of mixing for liquids. IV. Heat of mixing
for the binary systems acetone-water, methylethylketone-water, and cyclohexanone-
water, Vestn. Leningr. Univ. Fiz. Khim. ser. 3, 21, 90–93.
Boublı́k, T., Fried, V., and Hála, E. (1984). The Vapour Pressures of Pure Substances, Elsevier,
Amsterdam.
Brown, I., and Smith, F. (1957). Liquid-vapour equilibria: VIII. The system acetone þ
benzene and acetone þ carbon tetrachloride, Austr. J. Chem., 10, 423–428.
Brown, I., Fock, W., and Smith, F. (1969). The thermodynamic properties of solutions of nor-
mal and branched alcohols in benzene and n-hexane, J. Chem. Thermodyn., 1, 273–291.
Burman, Å. U., and Ström, K. H. U. (2007). Calculation of vapor liquid equilibrium and
excess enthalpy with the UNIQUAC equation, Chem. Eng. Commun., 194, 1029–1052.
Cézac, P., Castaing-Lavignottes, J., Reneaume, J. M., and Maunoury, A. (2009). Description
of ice slurry composition and enthalpy calculation from gE models: Application to
water-ethanol systems, Chem. Eng. Commun., 196, 715–728.
Christensen, C., Gmehling, J., Christensen, C., Rasmussen, P., Weidlich, U., and
Holderbaum, T. (1984–1991). Heats of Mixing Data Collection, DECHEMA Chemistry
Data Series, Frankfurt.
Christensen, J. J., and Rowley, R. L. (1988). Handbook of Heats of Mixing: Supplementary
Volume, Wiley-Interscience, New York.
Collins, S. G., Christensen, J. J., Izatt, R. M., and Hanks, R. W. (1980). The excess enthalpies of
10 (n-pentane þ an n-alkohol) mixtures at 298.15 K, J. Chem. Thermodyn., 12, 609–612.
Dahl, S., and Macedo, E. (1992). The MHV2 Model: A UNIFAC-based equation of state
model for vapor-liquid and liquid-liquid equilibria of mixtures with strong electrolytes,
Ind. Eng. Chem. Res., 31, 1195–1201.
Dahl, S., and Michelsen, M. L. (1990). High-pressure vapour-liquid equilibrium with a
UNIFAC-based equation of state, AIChE J., 36, 1829–1836.
Dahl, S., Fredenslund, A. A., and Rasmussen, P. (1991). The MHV2 model: A UNIFAC-
based equation of state model for prediction of gas solubility and vapor-liquid equilibria
at low and high pressures, Ind. Eng. Chem. Res., 30, 1936–1945.
Daubert, T. E., and Danner, R. P. (1986). DIPPR Data Compilation, AIChE, New York.
Fischer, K., and Gmehling, J. (1996). Further development, status and results of the PSRK
method for the prediction of vapor-liquid equilibria and gas solubilities, Fluid Phase
Equilib., 121, 185–206.
Fuchs, R., Krenzer, L., and Gaube, J. (1984). Excess properties of binary mixtures composed
of a polar component and an alkane, Ber. Bunsenges. Phys. Chem., 88, 642–649.
Gmehling, J., Onken, U., Arlt, W., Grenzheuser, P., Kolbe, B., Rarey, J., and Weidlich, U.
(1977–1996). Vapor-Liquid Equilibrium Data Collection, DECHEMA Chemistry Data
Series, Frankfurt.
Heidemann, R. A. (1996). Excess free energy mixing rules for cubic equations of state, Fluid
Phase Equilib., 116, 454–464.
Holderbaum, T., and Gmehling, J. (1991). PSRK: A group contribution equation of state
based on UNIFAC, Fluid Phase Equilib., 70, 251–265.
1452 Å. U. Burman and K. H. U. Ström

Huang, H., and Sandler, S. I. (1993). Prediction of vapor-liquid equilibria at high pressures
using activity coefficient parameters obtained from low-pressure data: A comparison of
two equations of state mixing rules, Ind. Eng. Chem. Res., 32, 1498–1503.
Huron, M. J., and Vidal, J. (1979). New mixing rules in simple equations of state for represent-
ing vapor-liquid equilibria of strongly nonideal mixtures, Fluid Phase Equilib., 3, 255–271.
Kalospiros, N. S., Tzouvaras, N., Coutsikos, P., and Tassios, D. P. (1995). Analysis of
zero-reference-pressure EoS=GE models, AIChE J., 41, 928–937.
Knudsen, K., Stenby, E. H., and Fredenslund, A A. (1993). A comprehensive comparison of
mixing rules for calculation of phase equilibria in complex systems, Fluid Phase Equilib.,
82, 361–368.
Kudryavtseva, L. S., and Susarev, M. P. (1963). Vapour-liquid equilibrium in the systems
Downloaded by [Memorial University of Newfoundland] at 20:44 11 November 2013

chloroform-hexane and acetone-chloroform, Zh. Prikl. Khim., 36, 1231–1237.


Kundu, M., and Bandyopadhyay, S. S. (2007). Thermodynamics of alkanolamine þ water
system, Chem. Eng. Commun., 194, 1138–1159.
Larkin, J. A. (1975). Thermodynamic properties of aqueous non-electrolyte mixtures. I. Excess
enthalpy for water þ ethanol at 298.15 to 383.15 K, J. Chem. Thermodyn., 7, 137–148.
Lieberwirth, I., and Schubert, H. (1979). Das isoterme Dampf-Flüssigkeits-Phasengleichgewichts-
verhalten des Systems Aceton=Wasser bei 35 C, Z. Phys. Chem. (Leipzig), 260, 669–672.
Mathias, P. M., and Copeman, T. W. (1983). Extension of the Peng-Robinson equation of
state to complex mixtures: Evaluation of the various forms of local composition concept,
Fluid Phase Equilib., 13, 91–108.
Mertl, I. (1972). Phase equilibria in the ternary system ethyl acetate – ethanol – water, Collect.
Czech. Chem. Commun., 37, 366–374.
Michelsen, M. L. (1996). Matching equation of state mixing rules to activity coefficient model
expressions, Fluid Phase Equilib., 121(1–2), 15–26.
Michelsen, M. L., and Heidemann, R. A. (1996). Some properties of equation of state mixing
rules derived from excess Gibbs energy expressions, Ind. Eng. Chem. Res., 35, 278–287.
Orbey, H., and Sandler, S. I. (1995). On the combination of equation of state and excess free
energy models, Fluid Phase Equilib., 111, 53–70.
Pando, C., Renuncio, J. A. R., Hanks, R. W., and Christensen, J. J. (1983). The prediction of
vapor-liquid equilibrium from heat of mixing data of binary hydrocarbon-ether and
hydrocarbon-aldehyde mixtures, Thermochim. Acta, 62, 113–124.
Ratcliff, G. A., and Chao, K. C. (1969). Prediction of thermodynamic properties of polar
mixtures by a group solution model, Can. J. Chem. Eng., 47, 148–153.
Row, K. H., Park, J. K., and Gasem, K. A. M. (2008). The modified PGR equation of state: Asym-
metric mixture VLE representation and predictions, Chem. Eng. Commun., 195, 492–510.
Sandler, S. I., Orbey, H., and Lee, B.-I. (1994). Equations of state, in Models for Thermodyn-
amic and Phase Equilibria Calculations, ed. S. I. Sandler, Marcel Dekker, New York.
Smirnova, N. A., and Kurtynina, L. M. (1969). Thermodynamic functions of mixing for a
number of binary alcohol-hydrocarbon solutions, Zh. Fiz. Khim., 43, 1883–1885.
Soave, G. (1972). Equilibrium constants from a modified Redlich-Kwong equation of state,
Chem. Eng. Sci., 27, 1197–1203.
Soave, G. (1992). A new expression of q(a) for the modified Huron-Vidal method, Fluid Phase
Equilib., 72, 325–327.
van Ness, H. C., and Abbott, M. M. (1982). Classical Thermodynamics of Non-electrolyte
Solutions, Pergamon, Oxford.
Vesely, F., Uchtil, P., Zabransky, M., and Pick, J. (1979). Heats of mixing of cyclohexane with
1-propanol and 2-propanol, Collect. Czech. Chem. Commun., 44, 2869–2881.
Wong, D. S. H., and Sandler, S. I. (1992). Theoretically correct mixing rules for cubic
equations of state, AIChE J., 38, 671–680.
Yoon, J.-H., Chun, M.-K., Hong, W.-H., and Lee, H. (1993). High-pressure phase equilibria
for carbon dioxide-methanol-water system: Experimental data and critical evaluation of
mixing rules, Ind. Eng. Chem. Res., 32, 2881–2887.

You might also like