You are on page 1of 6

Ind. Eng. Chem. Process Des. Dev.

1988,25, 531-536 531

performed at the highest temperature (242 “C) with low carbon monoxide oxidation. The content of carbon mon-
excess of oxygen in run 2 and high excess in run 6. The oxide thus increases rapidly and reduces the adsorbed
oscillations are very complicated and almost aperiodic in sulfur trioxide to sulfur dioxide which, on rapid desorbing,
run 2. In run 6 at the high oxygen excess, the frequency gives a fresh active surface. Subsequently, the carbon
is very high and the oscillations are almost periodic. High monoxide oxidation ignites and sulfur dioxide is oxidized
excess of oxygen, together with low temperature (runs 3 to sulfur trioxide, until the active sites are deactivated by
and 4), will obviously favor the suppression of the emission adsorbed sulfur trioxide again.
of sulfur trioxide. The proposed explanation given here is partly based on
It has been known for the past 10 years (Scheintuch and a study by Yao et al. (1981). They found that the de-
Schmitz, 1977) that the oxidation of carbon monoxide can creased oxidation rate of carbon monoxide on platinum
oscillate under isothermal conditions and under conditions in the presence of sulfur dioxide was attributed to the
where the surface reactions do not interact with physical effect of surface sulfates formed by sulfur dioxide ad-
transport steps. Various explanations of this behavior have sorption and oxidation, on the chemisorption capacity for
been proposed and tested by mathematical simulations carbon monoxide. As shown from infrared adsorption
(Scheintuch and Schmitz, 1977; Olsson, 1984). studies, the surface sulfates inhibited the associative
Of special interest for the present investigation are the chemisorption of carbon monoxide, which resulted in a
observations made by Mukesh et al. (1983),who found that lower activity for carbon monoxide oxidation.
carbon monoxide began to oscillate under isothermal The effect of the supporting material on the oscillations
conditions when 1-butene was added to the feed. Unlike was discussed by Mukesh et al. (1983). We found that the
the present study, their reactor was a CSTR and the supporting material had little influence on the oscillatory
content of 1-butene was as high as 1%, i.e., half the content behavior of the reaction system. The same type of oscil-
of carbon monoxide. In our study, the content of carbon lations as in the present study was found when the oxi-
monoxide was about 1% and the content of sulfur dioxide dation was performed in an unsupported platinum catalyst
only 0.002%. Moreover, the temperature was rather low tube used in a previous study (Olsson and Schoon, 1985).
(150 OC) in their experiments, resulting in relaxation os- Acknowledgment
cillations with low frequency. Cutlip and Kenney (1978)
considered that 1-butene competed with carbon monoxide The financial support of the Swedish Board for Tech-
for the same active sites and thus delayed the ignition of nical Development is gratefully acknowledged.
the rapid oxidation of carbon monoxide. During the ig- Registry No. SO3, 7446-11-9; SO2, 7446-09-5; CO, 630-08-0;
nition, carbon monoxide is removed both from the gas 02, 7782-44-7.
phase and from the catalyst surface. Consequently, 1- L i t e r a t u r e Cited
butene will be further adsorbed followed by some oxida- Cutlip, M. 0.; Kenney, C. N. ACS. Symp. Ser. 1978, No. 65, 475.
tion, and this increased adsorption eventually extinguishes HiavZek, V.; van Rompay, P. Chem. Eng. Sci. 1981, 36, 1587.
Mukesh, D.; Kenney, C. N.; Morton, W. Chem. Eng. Sci. 1983, 38, 69.
the oxidation of carbon monoxide. Carbon monoxide be- Nayfeh, A. H.; Mook, D. T. “Nonlinear Oscillations”; Wiley: New York, 1979.
gins again to adsorb and accumulates on the surface until Oisson, P.; SchMn, N.-H. Chem. Eng. Scl. 1985, 4 0 , 1123.
Oisson, P. Doctoral Thesis, Chaimers University of Technology, Gothenburg,
the next ignition starts. 1984.
A similar explanation may be valid for the initiating Pierson, W. R. Chemtech 1978, 6, 333.
effect of sulfur dioxide on the oscillatory carbon monoxide Scheintuch, M.; Schmitz, R. A. Catal. Rev. Sci. Eng. 1977, 15, 107.
Yao, H. C.; Stepien, H. K.; Gandhi, H. S . J. Cafal. 1981, 6 7 , 231.
oxidation. Since the oscillations are very rapid, it is
possible that only the most active sites take part in the Received for review July 27, 1984
oxidation of sulfur dioxide. These sites are rapidly deac- Revised manuscript received August 19, 1985
tivated by adsorbed sulfur trioxide which quenches the Accepted August 27, 1985

A Cubic Equation of State for Polar and Apolar Fluids


Hosseln Toghlanl and Dablr S. Vlswanath’
Department of Chemical Engineering, University of Missouri-Columbia, Columbia, Missouri 652 1 1

A generalized cubic equation of state is presented which utilizes the associating parameter, x,of Halm and Stiel.
The substancedependent critical compressibility factor, re,
as well as the adimensional factor, a(T J , was obtained
by minimizing the deviatlon in calculated liquid volume, while maintaining the fugacity equivalence of the two phases
in equilibrium. These parameters were generalized as functions of Pitzer’s acentric factor, w , and the associating
parameter, x. Thus, for pure substances, calculations require knowledge of only the critical temperature and
pressure, w , and x. This incorporation extends the applicability of the equation to highly polar substances, such
as ammonia, water, acetic acid, and methanol. This equation may also be used for aromatics including naphthalene
and isoquinoline and for high molecular weight paraffins such as hexadecane and eicosane. Ninety-six substances
of different chemical nature were tested and gave average absolute percentage deviations in vapor pressure of
1.38%, 2.14%, and 2.08%, respectively, for the proposed, Soave, and Peng-Robinson equations. The deviations
in liquid volume are 3.9%, 17.61 % , and 7.71 % , respectively, for the three equations.

Equations of state have proven to be useful tools in tively small amount of input data. Although the com-
industrial applications, allowing prediction of equilibrium plexity of any equation of state no longer presents a com-
phenomena and sizing of process equipment from a rela- putational problem, the majority of authors proposing
0196-4305/86/1125-0531$01.50/0 0 1986 American Chemical Society
532 Ind. Eng. Chem. Process Des. Dev., Vol. 25, No. 2, 1986

equations of state have attempted to retain the simplicity constraints. The third constraint arises from incorporation
found in the van der Waals’ cubic equation while trying of Abbott’s hypothesis (1973) and defies the apparent or
to improve its accuracy through modifications. In spite calculated critical compressibility. The constraints im-
of its simplicity, the van der Waals’ equation is tremen- posed are
dously successful in representing thermodynamic prop-
erties qualitatively.
Most of the semiempirical equations of state set forth
in recent years have retained the van der Waals’ thermal
pressure term. The differences lie in the expression of the
attractive pressure term. One of the earliest attempts to (3)
modify this attractive term was by Redlich and Kwong
(1949). Several authors have modified this term, assuming p ,,,calcd
a temperature dependence for a and replacing the de- (4)
nominator by a quadratic polynomial in volume. Two
important modifications which have enjoyed widespread The third constraint results in a P-V-T surface which has
acceptance are the Soave (1972) and the Peng-Robinson a calculated critical compressibility factor different from
(1976) equations of state. the experimental critical compressibilityfactor. This leads
Both the Soave and Peng-Robinson equations have been to better quantitative prediction of properties in regions
widely tested to determine their predictive capabilities in removed from the critical point while sacrificing predictive
the liquid region and for vapor-liquid equilibria (VLE) for capabilities in the region very near the critical point. This
a large number of systems. Both perform reasonably well does not lead to the impasse that it would appear to, be-
in VLE calculations, but the Soave equation typically cause methods of treating the critical region and its no-
overestimates liquid volumes, whereas the Peng-Robinson nanalytic behavior are under investigation such as that
equation does better in the prediction of liquid densities described by Fox (1983), which may well provide a pow-
for normal fluids. However, the use of these equations for erful tool in the future.
systems containing heavy hydrocarbons and polar fluids The parameters obtained from applying these three
leads to discrepanciesbetween calculated and experimental constraints to the equation of state after algebraic ma-
liquid densities. The major drawback in both the Soave nipulation are
and Peng-Robinson equations is that the critical com-
pressibility takes on values of 0.333 and 0.307, regardless
of the substance. (5)
In an effort to overcome this fixed critical compressi-
bility factor, Abbott (1973) anticipated that for many ap-
plications, the fixed value should be replaced by a sub-
stance-dependent adjustable parameter, CC. Typically, this
calculated critical compressibility factor is larger than the 1 - 3lc
experimental critical compressibility factor. Many authors cc = - (7)
such as Harmens and Knapp (1980), Schmidt and Wenzel Lvc
(19801, and Patel and Teja (1982) have incorporated this where
adjustable factor in their models, which resulted in im-
proved predictive capabilities for both vapor pressure and nb(w,x) = vctc (8)
liquid volume. All the above equations can be generated
from the most general cubic equation given by Martin fia(u,x) = (1- t c ( 1 - vc)I3 (9)
(1979) and also can be obtained from the equation of The value of qc may be determined analytically or nu-
Schmidt and Wenzel(l980) with an appropriate choice of merically by using the equation
the parameters.
In this work, we have used the general cubic equation 17; + v2(2/lc - 3) + 3vc - 1 = 0 (10)
of Schmidt and Wenzel (1980) along with the constraint
U + W = 1. However, instead of setting W = -3w, we set once the value of lcis set. If the value of c is set to 3w in
it equal to -c, where c is one of the equation parameters. eq 7, the above results are identical with those of Schmidt
We will also incorporate the polar factor of Halm and Stiel and Wenzel (1980). The critical properties used in ap-
(1967) in our modification so that polar and associating plying the constraints at the critical point were taken from
substances can be treated. These choices result in the Reid et al. (1977).
predicted critical compressibility factor being substance- Now the problem arises of selecting the optimal value
dependent. of the calculated critical compressibility factor, lc. The
optimal value of Cc was determined by following the path
Proposed Cubic Equation taken previously by Patel and Teja (1982) and Adachi et
The functional form of the proposed cubic equation of al. (1983). The following steps are taken to evaluate the
state is optimum value.
RT aa(Tr) 1. A value of l, in the neighborhood of the experimental
G(P,u,T,a)= -P + -- = 0 (1) critical compressibility factor is assumed.
-b u2 + bv + cbu - cb2 2. The critical parameters are evaluated by using eq
The three pure component parameters a, b, and c are 5-10.
determined based on the observation of the pressure- 3. For each data point along the saturation dome, a
volume behavior at the critical point, where a(T,) is equal value of a( T,) is determined by using the fugacity equiv-
to unity. Three constraints are imposed on eq 1 at the alence of the two phases in equilibrium.
gas-liquid critical point. The standard requirement of the f=t“ (11)
disappearance of the first two pressure-volume derivatives
a t the critical point comprises the first two of these three Utilizing the proposed equation of state in the relation
Ind. Eng. Chem. Process Des. Dev., Vol. 25, No. 2, 1986 533

A
1 n [ i ] = ~ " [ ~ - & ] d u + Z - l - l n Z (12)

the fugacity coefficient takes the form PROPANOL A

1.2 -
PENTANOL \ A A

where B = bP/RT, H = ((b + cb)' + 4 ~ b ' ) ~FI ~=, P(b +


cb - H)/(2RT), and G = P(b + cb + H)/(2RT).
A
4. The average absolute deviation in the liquid volume

14
is calculated. A
5. The assumed value of Cc is incremented by 0.001, and
steps 2-4 are repeated until a minimum in the average A
absolute deviation in the liquid volume is found. The value .
E 1.0
of lCwhich provided this minimum is used in subsequent t-
calculations. L

The temperature dependence of a ( T ) is established by


following an approach identical with that of Soave by
assuming that a ( T ) is composed of two parts:
a(T) = aca(Tr) (14) C I
A
Thus, like Soave's equation, the term a(Tr)must equal A
I- CACETlC
unity along the critical isotherm. This provides some V
a A
insight into the functional form of a(T,), but after many
attempts to find a suitable form for a(Tr),it was decided
to use the following two expressions given by Soave (1972) LACETONE
and Mathias and Copeman (1983), respectively.
a(Tr)= (1+ m(1 - T,1/2))2 (15)
0.6 1 A
A

a'(Tr) = A
(1+ C1(l - Trl/') + C2(l - Tr1/2)2+ C3(l - Tr1/2)3)2
(16)
Mathias and Copeman (1983) found it necessary to include 0 DIPOLE MOMENT > 0 . 8
higher order terms in the expansion in order to correlate A DIPOLE MOMENT 0.8 <
the vapor pressures of highly polar substances. We de-
termined the constants in both of these correlations for 0 0.2 0.4 0.6 0.8 1.0
each substance by using the Britt-Luecke (1973) gener- ACENTRIC FACTOR, W
alized least-squares regression algorithm. Figure 1. Characteristic constant vs. acentric factor.
The coefficient m in eq 15 is plotted vs. the acentric
factor in Figure 1. It is seen that the associating and given by Thompson (1966), the parameter Z,, presented
highly polar substances do not follow the same trend as by Spencer and Adler (1978), the molar polarization of
the normal fluids. This has been corrected as shown in Viswanath and Prasad (1970), the polar parameter of
the next section. Nishiumi (1980), and the polar factor of Halm and Stiel
(1967). Of these parameters, only the polar factor of Halm
Generalization of a ( T , ) and Cc
and Stiel can take on both positive and negative values.
In order to better predict liquid properties, investigators The others are limited to values equal to or greater than
such as Zudkevitch and Joffe (1970) who modified the zero. Thus, direct incorporation of any of these other
original Redlich-Kwong equation, Fuller (1976) who parameters would lead to a shift in one direction only
modified the Redlich-Kwong-Soave equation, and Heyen instead of making both positive and negative corrections.
(1980) who modified the Peng-Robinson equation have The polar factor of Halm and Stiel was developed as an
made the covolume parameter as well as the attraction extension of the existing three parameter corresponding
parameter temperature-dependent. However, as stated by states principle of Pitzer et al. (1955) to account for de-
Bazua (1983), this approach may lead to difficulties in viations from normal fluid behavior seen in polar sub-
calculating thermodynamic properties that require deriv- stances. This parameter is defined as
atives of the equation of state. Although this approach
has proved moderately successful in the prediction of liquid x = log Prl~,=0,8+ 1.7(- log PrlT,=0,7 - 1) + 1.552 (17)
densities for polar substances, we have taken a somewhat or utilizing the definition of the acentric factor
different approach, similar to that of Pate1 and Teja (1982).
Their generalized equation allowed prediction of both x = log P r l ~ , = 0 . 6+ 1 . 7 +
~ 1.552 (18)
vapor and liquid properties of nonpolar substances. In Expressions for various thermodynamic properties as a
their work, they also studied polar substances such as function of reduced temperature, the acentric factor, and
water, ammonia, and alcohols. However, these substances the polar factor have been formulated by Halm and Stiel
could not be included in their generalization as was ex- (1967,1970,1971). One of the properties generalized was
pected. Thus, in our approach, we include a fourth tem- the experimental critical compressibility factor as a
perature-independent parameter, in an attempt to include function of w and x. Their expression is
the associating and highly polar substances in our gener-
alization of CC and m. z,=
As a possible fourth parameter, we examined the para- 0.291 - 0 . 1 1 4 ~- 1.42%+ 0.069~'- 7 . 0 5 ~ '+ 1 . 5 1 ~ ~
chor developed by Quayle (1953), the radius of gyration (19)
534 Ind. Eng. Chem. Process Des. Dev., Vol. 25, No. 2, 1986

Table I. Constants for Equation 22 Table 11. Constants for Equation 23


A , = 0.441926073 A2 = 1.342 755 128 El = 0.324020789 E, = -0.056 675 895
A3 = -0.328431 972 A, = -15.020 572 758 E , = -0.001 268 996 E, = -3.259 131 762
A , = -2.226936 391 A, = 21.112706213 E5 = 1.399475 880 = 4.769 592 801
A: = 0.015 079 204 A, = 234.965 900 518 B, = 0.003 431 898 B, = 59.450 596 850

Since tabulated values of the polar factor for some sub-


stances in our study were unavailable, we undertook an
effort to obtain internally consistent values for both w and
x . Three different methods were used to obtain the
acentric factor: (1)interpolation of available experimental
data on vapor pressure, using Aitken's technique; (2) the
Wagner vapor pressure equation utilized by McGarry
(1983); and (3) the Antoine equation. The Wagner equa-
tion is
In P, =
1
-{A(1 - T,)
Tr
+ B ( l - Tr)lJ2+ C(1 - + D ( l - T,)') METHANOL

(20)
Values for the coefficients in this equation are available
for 251 substances. The coefficients used for the Antoine ACID
equation are given in Reid et al. (1977). It was seen that
the values for the acentric factor obtained from the Wagner
equation were more consistent with those given by Reid
et al. (1977)than those obtained from the Antoine equation
or from interpolation. Since the temperature range where
the Wagner equation was applicable extends below a re-
duced temperature of 0.6, we also evaluated the polar
factor. For higher molecular weight hydrocarbons where
the Wagner coefficients are not available, the values of the
acentric factor given by Reid et al. (1977) were used.
A variety of definitions for normal fluids appear in the
literature. Curl and Pitzer (1958) define a normal fluid
as one whose surface tension ratio, aovo2J3/RT,, can be
calculated to within 5% by
aovo213
- - - 2.24
RT,
X lo-' + 1.42 X lo-'~

Stipp et al. (1973) define a normal fluid to be that whose


polar factor has an absolute value not exceeding 0.005, and
0.4
P
0 0.2 0.4
DIPOLE
DIPOLE

0.8
MOMENT
MOMENT

0.8
> 0.8
0.8

1.0
they state that slightly polar substances satisfying this ACENTRIC FACTOR, W
definition ucan be considered to follow normal fluid be- Figure 2. Correlation between characteristic constant and acentric
havior exhibited by nonpolar substances". factor.
Examining the plot of m vs. w in Figure 1, we see that
the points for polar substances with dipole moments of less fashion, we generalized our calculated critical compressi-
than approximately0.8 and nonpolar substances follow the bility factor by using the same expression.
same trend. Thus, we elect to use the polar factor for only L=
those substances whose dipole moment is greater than 0.8, Bl + B2w + B3w2+ B4x + B6x2+ Bswx + B7w3 + Bsx3
and for all other substances, we set the polar factor to 0.
It is interesting to note from Figure 1 in the case of (23)
methanol, ethanol, and propanol that all have a dipole The constants for eq 23 are listed in Table 11. To
moment near 1.7 D, but methanol lies below, ethanol lies illustrate the effect of including polar-factor-dependent
on, and propanol lies above the nonpolar trend. The Stiel terms in eq 22, a plot of m - A4x - A5x2- Aswx - A8x3vs.
polar factor for these follow a similar trend that the value w is presented in Figure 2. As is apparent, use of the polar
of x is 0.0373 for methanol, 0.0003 for ethanol, and -0.0499 factor forces points for the substances with dipole moments
for propanol. This shows that a combination of x and w greater than 0.8 to approach the same curve as those with
may work better compared to either x or w alone. dipole moments less than 0.8. This justifies our expressing
On the basis of these observations, the value of m was m in terms of w and x . We also attempted to generalize
fitted to the following expression which is identical with the coefficients in the Mathias and Copeman (1983) ex-
that used for Z,, eq 19, with an additional higher order pression, but the results were not encouraging.
term Data Base
m= Data used in establishing the performance of the pro-
A , + Azw + A3w2 + A4x + A5x2 + A ~ W+X A703 + A,x3 posed equation of state were taken from a variety of
(22) sources. References for each substance appear in Table
I11 in the supplementary material. For those substances
and the coefficients established by using the Britt-Luecke where the liquid volume data were unavailable in the
algorithm (1973) are shown in Table I. In a similar temperature range corresponding to the available vapor
Ind. Eng. Chem. Process Des. Dev., Vol. 25, No. 2, 1986 535

Table IV. Comparison of Vapor Pressures and Liquid Volumes


A.A.D. (%) in vapor A.A.D. (70)in liquid
no. of no. of pressure volume
data substances PE RKS PR PE RKS PR
monoatomic 92 3 1.26 1.77 1.84 4.97 5.34 9.33
diatomic 83 5 0.65 0.80 1.07 4.55 4.79 8.25
paraffins 613 29 1.34 1.96 2.43 3.44 17.70 6.50
olefins 217 10 0.93 2.08 1.40 2.99 13.64 4.31
diolefins 36 1 0.91 1.75 0.89 3.34 12.73 3.82
acetylenes 37 2 3.33 3.17 2.79 2.71 12.13 3.48
naphthenes 56 2 1.39 1.26 1.98 3.52 11.74 4.25
halogenated benzenes 53 2 1.38 1.41 1.55 4.74 15.06 2.06
aromatics 285 10 1.19 2.04 1.99 3.31 14.17 3.73
aldehydes and ketones 42 2 1.37 2.36 0.91 2.18 37.62 21.66
amines 38 2 0.82 1.87 1.57 5.44 7.36 7.54
hydrogen bonding 175 6 2.66 4.21 3.81 3.74 27.76 13.35
ethers 37 3 1.55 2.85 2.33 4.21 17.37 4.91
sulfur compds 87 4 1.22 1.73 1.41 5.23 12.79 4.84
nitrogen compds 75 3 1.50 2.81 1.82 8.59 43.44 26.78
halogenated hydrocarbons 73 3 1.74 1.51 0.97 2.34 21.39 8.29
acetates 71 3 0.99 4.00 3.01 5.71 33.01 17.68
fluorocarbons 56 3 1.23 1.01 2.05 3.28 9.93 4.44
oxides 64 3 1.28 1.19 1.48 5.89 20.01 7.74
grand total 2190 96 1.38 2.14 2.08 3.90 17.61 7.71

0 * - -
0
EQUATION 1
0 0
0
0
-10 0
w -10
0
3
A
0
o n o n 0 n o n o o 0
0 Y
W
2
- 20 P E N G - ROBINSON EQUATION 0 - 0
0 U
0 >
i -20 0
0 9
0
5 0
0

z
0 ’ -30
0
0
0 AA A A
0
5
> -30
W
SOAVE EOUATION ’ A b
2
CI

8 & A 2 -40
0 a
A
A 9 P 0
AMMONIA
W
-40 n P

A # -50 a
0.6 0.7 0.8 0,s 1.0
n
METHANOL
REDUCED TEMPERATURE P

Figure 3. Comparison of percent deviation in saturated liquid molar -60 n


volume vs. reduced temperature for ammonia.
pressure data, the modified Rackett equation of Spencer 0,6 0.7 0.8 0.9 1.0
and Danner (1972) was used. For this equation, both the REDUCED T E MPERATURE
parameter Z,, and the critical properties given by Spencer
Figure 4. Comparison of percent deviation in saturated liquid molar
and Adler (1978) were used in generating liquid volume volume vs. reduced temperature for methanol.
data. However, for some substances, the parameter 2, was
not available, and in these cases, the value of Z,, given by Robinson equation (PR) in calculated vapor pressure and
Hankinson and Thomson (1979) along with the critical liquid volume and the overall weighted average for each
properties given in Reid et al. (1977) were used. Also, we class of substances is presented in Table 111, which has
have generated vapor pressure data for some of the sub- been submitted as supplementary material (see paragraph
stances by using either the equations of Ambrose et al. at end of paper regarding this material). Table IV sum-
(1974,1975) or the equation of Wagner given by McGarry marizes the results by class of compound. This compilation
(1983). In this phase of our work, we have accepted the indicates that incorporation of the parameter x into our
reliability of the data generated in this manner and have proposed equation generally results in better predictive
assumed it to be accurate to within experimental error. capabilities than either the Redlich-Kwong-Soave or
During our analysis, it became apparent that some ex- Peng-Robinson equations. For high molecular weight
perimental data points were obviously in error, and these hydrocarbons, much better results were obtained for the
points were eliminated from our study. calculated liquid volume by using the proposed equation
Results rather than either the Peng-Robinson or Redlich-
A summary of the average absolute percentage devia- Kwong-Soave equation.
tions, A.A.D.( %), for the proposed equation (PE), the The deviation in liquid molar volume as a function of
Redlich-Kwong-Soave equation (RKS), and the Peng- reduced temperature is depicted graphically in Figure 3
536 Ind. Eng. Chem. Process Des. Dev., Vol. 25, No. 2, 1986

for the highly polar substance ammonia and in Figure 4 u = surface tension
for methanol which exhibits hydrogen bonding. As is 7 = smallest positive root of eq 10
apparent from these figures, the proposed equation per- p = fugacity coefficient
forms better than the Peng-Robinson equation and the x = association factor
Redlich-Kwong-Soave equation over the whole range of R = dimensionless premultiplier defined by eq 8 and 9
w = acentric factor
reduced temperature. The proposed equation gives very
good results up to reduced temperatures of approximately Subscripts
0.80. a = refers to parameter a
Conclusion b = refers to parameter b
c = critical property
Preliminary findings for the proposed equation of state 0 = hypothetical property at 0 K
which utilizes the associating factor of Halm and Stiel r = reduced property
indicate a significant improvement in the representation
of liquid volumes for the majority of substances studied Superscripts
compared to the highly regarded Redlich-Kwong-Soave 1 = liquid phase
and Peng-Robinson equations of state. Generalized ex- v = vapor phase
pressions for the apparent critical compressibility factor Literature Cited
and the adimensional factor in the attraction term have Abbott, M. M. AIChE J . 1973, 19, 596.
been presented which are suitable for all the substances Adachi. Y.; Lu, B. C.-Y.; Sugie, H. fluid Phase Equillb. 1983, 13, 133.
Ambrose. D.; Sparke, C. H. S.; Townsend, R. J. Chem. Thermodyn. 1974.
studied, both polar and nonpolar. Even though it is 6, 693.
possible to develop correlations for each class of com- Ambrose, D.; Sparke, C. H. S.;Townsend, R. J . Chem. Thermodyn 1975,
pounds to better predict saturated liquid volume and vapor 7, 185.
API Monogr. Ser. 1978, No. 707, 1.
pressure, we think it more desirable to have a single gen- API Monogr. Ser. 1979, No. 712, 1.
eralized expression to facilitate use of the equation. The Bazua, E. R. "Cubic Equation of State for Mixtures Containing Polar
performance of the equation as with any cubic equation Compounds"; Newman, S. A., Ed.; Ann Arbor Science: Ann Arbor, MI,
1983; p 195.
of state is limited to regions removed from the critical Britt, H. I.; Luecke, R. H. Technometrics 1973, 15, 233.
point. However, the proposed equation could be used a t Canjar, L. N.; Manning, F. S. "Thermodynamic Properties and Reduced Cor-
relations for Gases"; Gulf Publishing Co.: Houston, 1967.
temperatures much closer to the critical point compared Curl, R. F., Jr.; Pitzer, K. S. Ind. Eng. Chem. 1958, 50, 265.
to the Redlich-Kwong-Soave and Peng-Robinson equa- DeSelms, B. C. M.S. Thesis, University of Missouri-Columbia, 1984.
Fox, J. R. Fluid Phase Equilib. 1983, 14, 45.
tions as is evident from Figures 3 and 4. Fuller, G. G. Ind. Eng. Chem. Fund" 1976, 15. 254.
Preliminary results of VLE calculations for systems Griguii, U. "Properties of Water and Steam in S I Units", 2nd ed.; Springer-
containing water and alcohols are very encouraging. Work Veriag: Berlin, 1979.
Hankinson, R. W.; Thomson, G. H. AIChE J . 1979, 2 5 , 653.
is under way to determine suitable mixing rules for the Haim, R. L.; Stiei, L. I. AIChE J . 1967, 13, 351.
equation of state parameters, which involve only the pure Haim, R. L.; Stiel, L. I.AIChE J . 1970, 16, 1.
component parameters without introduction of an addi- Haim, R. L.; Stiel, L. I. AIChE J . 1971, 17, 259.
Harmens, A.; Knapp, H. Ind. Eng. Chem. fundam. 1980, 19, 291.
tional parameter such as the empirical binary interaction Heyen, G.; Liege, 6 . Paper presented at the 2nd International Conference on
parameter, KIJ. Preliminary work in predicting P-V-T Phase Equilibria and Fluid Properties in the Chemical Process Industry,
Dechema, Frankfort, 1960, p 9.
behavior of binary systems looks optimistic in this direc- Maitland, G. C.; Rigby, M.; Smith, E. 8.; Wakeham, W. A. "Intermolecular
tion without the use of KLJ.This is further supported by Forces"; Ciarendon Press: Oxford, 1981; p 460.
Martin, J. J.; Campbell, J. A.; Seidei, E. M. J. Chem. Eng. Data 1963, 8 ,
a study of potential parameters for the Lennard-Jones 560.
potential function by DeSelms (1984). This study which Martin, J. J. Ind. Eng. Chem. fundam. 1979, 18, 81.
compared 83 different mixing rules led to the conclusion Mathias, P. M.; Copeman, T. W. Nuid Phase Equilib. 1983, 13, 91.
McGarry, J. Ind. Eng. Chem. Process Des. Dev. 1983, 22, 313.
that use of the harmonic mean for both of the potential Nishiumi, H. J. Chem. Eng. Jpn. 1980, 13. 178.
parameters gives better results. Further, the ratio of the Patel, N. C.; Teja, A. S. Chem. Eng. Sci. 1982, 37, 463.
Peng, D. Y.; Robinson, D. B. Ind. Eng. Chem. fundam. 1976, 15, 59.
geometric mean to the arithmetic mean for the energy Pitzer, K. S.; Lippmann, D. 2.; Curl, R. F., Jr.; Huggins, C. M.; Peterson, D. E.
parameter yields values in good agreement with the J. Am. Chem. SOC. 1955, 77, 3433.
quantity (1- K,) as given by Maitland et al. (1981). Work Quayie, 0.R. Chem. Rev. 1953, 53, 439.
Rediich, 0.; Kwong, J. N. S. Chem. Rev. 1949, 4 4 , 233.
is continuing in this area, and findings will be reported in Reid, R. C.; Prausnitz, J. M.; Sherwood, T. K. "The Properties of Gases and
future publications. Liquids", 3rd ed.; McGraw-Hili: New York, 1977; p 630.
Soave, G. S. Chem. Eng. Sci. 1972, 27, 1197.
Acknowledgment Schmidt, G.; Wenzei, H. Chem. Eng. Sci. 1980, 35, 1503.
Spencer, C. F.; Adier, S. B. J . Chem. Eng. Data 1978, 23, 82.
This work was funded by the National Science Foun- Spencer, C. F.; Danner, R. P. J . Chem. Eng. Data 1972, 17, 236.
dation under Contract PE8117031. Stipp, G. K.; Bai, S. D.; Stiei, L. I. AIChE J . 1973, 19, 1227.
Thompson, W. H. Ph.D. Thesis, The Pennsylvania State University, University
Nomenclature Park, 1966.
Timmermans, J. "Physic0 Chemical Constants of Pure Organic Compounds";
a, b, c = equation of state parameters, eq 1 Eisevier: New York, 1950.
Vargaftik, N. B. "Tables on the Thermophysicai Properties of Liquids and
f = fugacity Gases", 2nd ed.; Hemisphere Publishing Co.: Washington, DC, 1975.
K,, = binary interaction parameter Viswanath, D. S.; Prasad, D. H. L. "A New Third Parameter for the Correia-
m = characteristic constant tion of Thermodynamic Properties"; ASME, United Engineering Center:
P = pressure New York, 1970.
Zudkevitch, D.; Joffe, J. AIChE J . 1970, 16, 112.
R = universal gas constant
T = absolute temperature
v = molar volume Received for review December 26, 1984
Z = compressibility factor Revised manuscript received August 29, 1985
Z,, = Spencer and Danner (1972) 2 factor Accepted September 19, 1985

Greek Letters Supplementary Material Available: Table I11 (com-


a: = adimensional factor, eq 15 parison of vapor pressures and liquid volumes by compound)
a' = adimensional factor, eq 16 (7 pages). Ordering information is given on any current
{c = apparent critical compressibility factor masthead page.

You might also like