Professional Documents
Culture Documents
1989,28, 324-328
Person, W. B. A Criteria for Reliability of Formation Constants of Van Geet, A. L. Calibration of the Methanol and Glycol Nuclear
Weak Complexes. J. Am. Chem. SOC. 1965, 87, 167. Magnetic Resonance Thermometers with a Static Thermistor
Renon, H.; Prausnitz, J. M. On the Thermodynamics of Alcohol- Probe. Anal. Chem. 1968,40,40.
Hydrocarbon Solutions. Chem. Eng. Sci. 1967, 22, 299. Van Geet, A. L. Calibration of Methanol Nuclear Magnetic Reso-
Renon, H.; Prausnitz, J. M. Local Compositions in Thermodynamic nance Thermometer at Low Temperature. Anal. Chem. 1970,42,
Excess Functions for Liquid Mixtures. AIChE. J. 1968,14, 135. 679.
Saunders, M.; Hyne, J. B. Trimer Association of &Butanol by NMR. Varian Associates XL-Series NMR Superconducting Spectrometer
J. Chem. Phys. 1958a, 29, 253. Systems Basic Operation Manual, 1984.
Saunders, M.; Hyne, J. B. Study of Hydrogen Bonding in Systems Vesely, Frantisek; Uchytil, P.; Zabransky, M.; Pick, J. Heats of
of Hydroxylic Compounds in Carbon Tetrachloride through the Mixing of Cyclohexane with 1-Propanol and 2-Propanol. Collect.
Use of NMR. J . Chem. Phys. 1958b, 29, 1319. Czech. Chem. Commun. 1979,44, 2869-2881.
Smirnova, N. A.; Kurtynina, L. M. Thermodynamic Functions of Vonka, P.; Svoboda, V.; Strubl, K.; Holub, R. Liquid-Vapor Equi-
Mixing for a Number of Binary Alcohol-Hydrocarbon Solutions. librium. System Cyclohexane-1-Butanol a t 50 "C and 70 "C.
Zh. Fiz. Khim. 1969, 43, 1883. Collect. Czech. Chem. Commun. 1971, 36, 18.
Thomas, E. R.; Eckert, C. A. Prediction of Limiting Activity Coef- Weltner, William; Pitzer, Kenneth Methyl Alcohol: The Entropy,
ficients by a Modified Separation of Cohesive Energy Density Heat Capacity and Polymerization Equilibria in the Vapor, and
Model and UNIFAC. Ind. Eng. Chem. Process Des. Dev. 1984, Potential Barrier to Internal Rotation. J. Am. Chem. SOC. 1951,
23, 194. 73, 2606.
Trampe, David M. Measurement and Applications of Limiting Ac- Wilson, G. M. Vapor-Liquid Equilibrium. XI. A New Expression
tivity Coefficients and Liquid-Liquid Equilibria. M.S. Thesis, for the Excess Free Energy of Mixing. J. Am. Chem. SOC.1964,
University of Illinois, Urbana, 1987a. 86, 127.
Trampe, David M. University of Illinois, personal communication,
1987b. Zong, Z.; Yang, X.; Zheng, X. Correlation of Vapor-Liquid Equilibria
Tucker, E. E.; Becker, E. D. Alcohol Association Studies. 11. Vapor of Associated Solutions. J . Chem. Eng. Jpn. 1984, 17, 71.
Pressure, 220-MHz Proton Magnetic Resonance, and Infrared
Investigations of tert-Butyl Alcohol Association in Hexadecane. Received for review May 24, 1988
J . Phys. Chem. 1973, 77, 1783. Accepted September 26, 1988
A simple method is proposed to determine the Wilson constants from binary vapor-liquid equilibrium
data. The method is based on a polynomial description of the variation of G E/RTwith composition
and finding the maximum value of the curve. The constants are determined by using a hand scientific
calculator or a personal computer. Prediction of vapor compositions is as good or better than that
obtained by complex optimization techniques.
Many equations have been proposed to describe the derived from the pertinent binary systems.
vapor-liquid equilibrium relationships. For homogeneous For a binary system, the activity coefficients are
systems, the two-parameter Wilson equation (Wilson,
1964) has been shown to be very suitable. In a multicom- In y1 =
ponent system, it reads
In y1 =
where
UjL
Ai, = exp[-(Xij - Xii)/RT] (2)
ui
Extensive tables are available (Hudson and Van Winkle,
To a first approximation the energy terms (Xi,- Xii) are 1970; Hirata et al., 1976; Gmehling and Onken, 1977),
assumed to be independent of temperature, although it is reporting the values of either the energy parameters (A,,
claimed (Nagata and Yamada, 1973) that a better fit is - A,,) or the constants A, and A,[.
obtained if a polynomial dependency is assumed: The most difficult problem in using Wilson's equations
is how to determine the two parameters A12and Azl from
Xij - Xii =u + bT + cT' + ... (3) a set of data ( y , , ~ , ) .Equations 4 and 5 are a pair of
Obviously application of eq 3 carries the penalty of a larger transcendental equations that can only be solved numer-
number of constants. ically. Czelej (1987) has developed a mathematical pro-
The Wilson equations are attractive because they have cedure to transform the Wilson equations into a polyno-
a built-in effect of temperature, lacking in previous models, mial form that allows an easier determination of the con-
and permit the calculation of multicomponent systems stants. Due to some serious mathematical errors present
from a combination of parameters A,, and AIi which are in Czelej's equations, their use is not recommended.
Several methods have been suggested for determining
the optimum constants AIz and Azl. According to Hirata
* To whom correspondence should be addressed. et al. (1976), the results of the methods depend upon the
088S-5885/89/2628-0324$01.50/0 0 1989 American Chemical Society
Ind. Eng. Chem. Res., Vol. 28, No. 3, 1989 325
1 - A21
22 = (10)
1 2 + A2lXl
Denoting g E = G E / R Tand defining the parameters z1and dP(xl)/dxl = C k w l k - ' = In (y1/y2) (18)
k=l
22
Thus, the values of gE(xl)and In (y1/y2)in eq 14 and
1 - A12 15 can be replaced by the polynomial P ( x l ) and its first
21 = (9)
x1 + '412% derivative, dP(xl)/dxl, respectively. Such representation
326 Ind. Eng. Chem. Res., Vol. 28, No. 3, 1989
of experimental data [eq 17 and 181 has obvious advantage a volume fraction term to the excess Gibbs energy of
in computer calculations. mixing:
As shown in Figure 1, gE(xl)has always a t least one G E / R T = GE/RT(Wilson) +
extreme value (maximum or minimum), its location x1 =
xl* and the corresponding extreme value gE(xl*)= P(xl*) x1 In ( x , + + +
rx,) x 2 In ( x l / r x,) (22)
can be calculated from eq 18 by taking into account that where r denotes the ratio of the molar volumes, uZL/ulL,
at this point the derivative @(xl)/dxl is nil. In this case, of the pure components and the G E/RT(Wilson) term is
eq 14 and 15 become defined in eq 8. Equation 22 is often called the T-K-
xl* In [ l + x2*zl] + x2* In [ l + x1*z2] - P(xl*) = 0 Wilson equation. The suggested algorithm can be used
(19) to determine the corresponding new pair of A12 and A,,
parameters, if gE(xl)= GE/RT is replaced by the function
and
Q(xl) = G E / R T - x1 In ( x , + rx,) - x 2 In ( x l / r x,)+
(23)
- In (1 + x1*z2)]+ x2*z2
where
The use of eq 19 and 20 for the evaluation of A,, and Q(xl) = x1 In (1 + xzzl) + x 2 In (1 + qz,) (24)
A,, will be illustrated later for a number of binary systems. Expressing Q(xl) in the polynomial form (17) permits
When the experimental data are good enough to permit calculation of the T-K-Wilson parameters from eq 19,20,
representation of the second and third derivatives of gE(x,) and 16.
in the polynomial form (17), evaluation of the Wilson Nagata et al. (1975) modified the Wilson equation by
parameters will reduce to solving the following algebraic adding the Scatchard-Hildebrand term
equations: (6, - 6,)2u1Lu,Lx1xz
~ 1 - 221
~ + 2 '22, - d2P(XJ/dX,2 = 0
1 ~ 2~ ~ - GE/RT = GE/RT(Wilson) + (X,VlL + X,V,L) (25)
+
3z12- 2xlzl3 - 3zZ2 2x,zZ3- d3P(xl)/dxl3= 0 (21) where a1 and 6, are the solubility parameters of pure
Once again z1 and z2 can be isolated because the first components.
equation in (21) is quadratic in z1 or z,. Evidently, this Once again, the calculation procedure for eq 25 is the
equation, if accurate, can replace the logarithmic rela- same as the one described for the T-K-Wilson equation,
tionship between z1 and z2 in eq 15 and 20. except that the auxiliary function is defined differently:
The Wilson equation in its original form has the sig-
nificant disadvantage that it cannot represent liquid-liquid
systems. A number of modifications have been proposed
to overcome this disadvantage. The simplest one was A more versatile treatment of partially miscible systems
introduced by Tsuboka and Katayama (1975), who added can be expected if the solubility parameters in eq 25 are
328 Ind. Eng. Chem. Res., Vol. 28, No. 3, 1989
replaced by an adjustable constant b or by the strictly 3. The mathematical routine described here is very fast
regular solution term, as suggested by Novak et al. (1974): and avoids the need of a medium- or large-size computer.
The required algorithms are normally available with
G E / R T = GE/RT(Wilson) + b x 1 x 2 (27) personal and hand scientific calculators.
If one-point data are considered, then the three param- 4. The procedure calls for data in the central part of
eters A12, Azl, and b present in eq 25 and 27 can be the composition range, where the experimental analytical
evaluated from a knowledge of G E and its first and second precision is the best.
derivatives. For the case where GE/RT is presented in the 5. For a given system, significantly different pairs of
polynomial form (17), the corresponding equations for the Wilson constants will have the same capability of pre-
case given in (27) become dicting the composition of the gas phase.
An additional important observation of the data re-
g E + bx,x2 = P(xJ ported in Table I is that the predictive ability for the vapor
composition of component 2 is normally much higher than
agE/dxl + b(xl - x,) = w(xl)/dxl that for component 1. We attribute this to the fact that
d2gE/dx12- 2b = d2P(xl)/dx12 (28) component 1 is usually the most volatile one and the
pertinent Antoine equation is being extrapolated beyond
where g E and its derivatives are given by eq 11-13. The its real range.
third parameter b can be easily removed from eq 28, but
there is no possibility to isolate z1 and z2. They should be Nomenclature
evaluated from the pair of transcendental equations (in- Ai. = Wilson constant
troduction of Q(xl) = gE/x1x2 + b where G E / R T = G k = excess Gibbs function
x1x2Q(x1)leads to more complex transcendental equations) R = gas universal constant
r = ratio ulL/uZL
T = absolute temperature, K
u f = liquid molar volume of component i
x i = mole fraction of component i in the liquid phase
y i = mole fraction of component i in the vapor phase
Greek S y m b o l s
(29) = solubility parameter of component i
yi = activity coefficient of component i
by ordinary numerical methods.
Quite clearly, the simplest form of these equations will Literature Cited
occur when the composition is equimolar (xl = x2 = 0.5) Czelej, M. The Polynomial Form of the Wilson Equation for Binary
and for the extreme value of P(xl*) where dP(xl)/dxl = and Ternary Systems. Int. Chem. Eng. 1987,27, 535-538.
0. Hirata, M.; Ohe, S.; Nagahama, K. Computer Aided Data Book of
The case represented by eq 25 can be treated in a similar Vapor-Liquid Equilibrium. Kodansha-Elsevier: Tokyo, 1976.
way. Hudson, J. W.; Van Winkle, M. Multicomponent Vapor-Liquid
Equilibriums in Miscible Systems from Binary Parameters. Znd.
Eng. Chem. Process Des. Dev. 1970, 9,466-472.
Results and Discussion Gmehling, J.; Onken, U. Vapor-Liquid Equilibrium Data Collection;
A number of systems were selected from the Hirata DECHEMA Chemistry Data Series; DECHEMA: New York,
(Hirata et al., 1976) collection, representing solutions with 1977; Vol 1.
Nagata, I.; Yamada, T. Parameter-Seeking Methods of Local Com-
strong positive or negative deviations from ideal behavior, positions Equations. J . Chem. Eng. Jpn. 1973, 6, 215-219.
with or without azeotropes, a t isobaric or isothermal con- Nagata, I.; Nagashima, M.; Ogura, M. A Comment on an Extended
ditions. The values of G E/RT were calculated and then Form of the Wilson Equation to Correlation of Partially Miscible
fitted with polynomials of different degrees (from 2 to 5). Systems. J . Chem. Eng. Jpn. 1975, 8, 406-408.
Equations 19 and 20 were solved and the Wilson constants Novak, J. P.; Vonka, P.; Suska, J.; Matous, J.; Pick, J. Applicability
were determined from relations 16. The computational of the Three-Constant Wilson Equation to Correlation of Strongly
Nonideal Systems. 11. Collect. Czech. Chem. Commun. 1974,39,
procedure was done fast and efficiently in a Mac 512 3593-3598.
computer using the Excel spreadsheet, the Statworks Silverman, N.; Tassios, D. The Number of Roots in the Wilson
statistical package for the polynomial algorithm, and the Equation in the Correlation of Vapor-Liquid Equilibrium. Znd.
Eureka program for the trial and error part. Overall Eng. Chem. Process Des. Dev. 1977,16, 13-20.
computational for one complete cycle took a few minutes. Tsuboka, T.; Katayama, T. Modified Wilson Equation for Vapor-
A scientific hand calculator (HP 15C) was also used for Liquid and Liquid-Liquid Equilibria. J . Chem. Eng. Jpn. 1975,
8, 181-187.
comparison purposes. Verhoeye, L. A. J. Remarks on the Determination of the Wilson
Results for the selected systems are presented in Table Constants in the Correlation of Vapor-Liquid Equilibrium Data.
I and compared with the results of Hirata et al. Inspection Chem. Eng. Sci. 1970,25, 1903-1908.
of Table I leads to the following conclusions. Wilson, G. H. Vapor-Liquid Equilibrium. IX: A New Expression
1. The accuracy of the suggested method (measured by For the Excess Free Energy of Mixing. J . Am. Chem. SOC.1964,
the ability to predict the vapor composition) is normally 86, 127-130.
as good or better than the complex search techniques. Received f o r review June 10, 1988
2. A third or fourth degree polynomial will usually Revised manuscript received November 1, 1988
suffice. Accepted November 7, 1988