equilibrio etilenglicol y agua

Isobaric VaporLiquid Equilibria for Ethanol + Water + EthyleneGlycol and Its Constituent Three Binary SystemsNaoki Kamihama, Hiroyuki Matsuda, Kiyofumi Kurihara, Katsumi Tochigi,*, and Shigeo Oba

Department of Materials and Applied Chemistry, Nihon University, 1-8 Kanda Surugadai, Chiyoda-ku, Tokyo 101-8308, JapanApplied Thermodynamics and Physical Properties Co., Ltd., 1-12-11-505 Sodegaura, Narashino-City, Chiba 275-0021, Japan

ABSTRACT: Isobaric vaporliquid equilibria were measured for the ternary systemethanol + water + ethylene glycol and its three constituent binary mixtures at 101.3 kPausing a modified RogalskiMalanoski equilibrium still. The thermodynamic consistencyof experimental binary data was checked using the point and area tests. The experimentalbinary data were then correlated by the Wilson, nonrandom two-liquid (NRTL), anduniversal quasichemical activity coefficient (UNIQUAC) equations. The ternary vaporliquid equilibria data were predicted using the binary parameters for the three equationswith good accuracy. The selectivity of ethylene glycol as entrainer for the separation of theazeotropic system ethanol + water by extractive distillation was discussed using the NRTLparameters.

INTRODUCTIONSince the 1970s, higher oil prices have driven the developmentof fermented ethanol as an alternative fuel, and concerns relatedto global warming have accelerated these efforts. The reductionof the energy requirements in the refining step of ethanolproduction is a major process development challenge inensuring that ethanol becomes an effective alternative energysource. The widely used distillation technology is not ideal forethanol dehydration owing to ethanol's azeotropic nature withwater. Azeotropic distillation or extractive distillation with anadditional solvent component (the entrainer) would thereforebe required. Azeotropic distillations are widely used in ethanoldehydration for nonfuel applications. Extractive distillation hasan inherent advantage in that it can reduce the energyrequirement based on the nature of the entrainers that do notneed to boil up to the top of the column. On the other hand,azeotropic distillation requires all entrainers to distill from thetop of the column.Ethylene glycol has been proposed by Washall1 as a potential

entrainer for alcohol dehydrations. Landisch and Dyck2

proposed several entrainers including glycol for ethanoldehydration, while ethylene glycol was proposed by Lee andPahl.3 These entrainers seem to possess favorable character-istics for ethanol dehydration that breaks the ethanolwaterazeotrope. However, the available vaporliquid equilibriumdata for the ethanol, water, and ethylene glycol system are notsufficient for the optimal design of the extractive distillationsystem with effective heat integration. This paper presents themeasurement of vaporliquid equilibria (VLE) for the ternarysystem ethanol + water + ethylene glycol and its constituentthree binary mixtures at 101.3 kPa. Some already reportedbinary and ternary VLE data35 were compared with theexperimental data. The thermodynamic consistency of the

experimental binary data was checked using the point and areatests,6,7 and then the data were correlated by the Wilson,8

nonrandom two-liquid (NRTL),9 and universal quasichemicalactivity coefficient (UNIQUAC)10 equations. The ternary VLEdata were predicted using the binary parameters for the threeequations.

EXPERIMENTAL SECTIONApparatus and Procedure. Figure 1 shows the apparatus

for measuring VLE data. A modified RogalskiMalanoskiequilibrium still with pressure control was used. The analysiswas similar to that described in the literature.11 In addition tothe still, the apparatus consisted of a data logger (34970A,Agilent Technologies Co., Santa Clara, CA), an personalcomputer, pressure controller, two buffer tanks, a refrigerator,and a vacuum pump.The equilibrium temperature was measured with a calibrated

platinum resistance thermometer with an accuracy of 0.01 K.The pressure in the still was determined with a pressurecontroller (DOI3-20, Druck Co., UK), and the accuracy wasestimated to be 0.03 kPa.Analysis. Vapor and liquid samples were analyzed with a gas

chromatograph (GC-4000, GL Sciences Co. Ltd., Tokyo,Japan) equipped with a thermal conductivity detector.Gaskuropack 54 60/80 was used as the column packing andhelium as the carrier gas. The compositions were determinedusing the relative area method with an accuracy of 0.1 mol %.Materials. Ethanol and ethylene glycol (special grade pure

from Wako Pure Chemical Industry, Ltd.) were used after

Received: August 3, 2011Accepted: December 19, 2011Published: January 13, 2012

Article

pubs.acs.org/jced

2012 American Chemical Society 339 dx.doi.org/10.1021/je2008704 | J. Chem. Eng.Data 2012, 57, 339344

removing traces of water with 3A molecular sieves. The waterwas passed through an ion exchanger and distilled. The materialpurity was checked by gas chromatography and found to begreater than 99.6 mass % for ethanol and 99.9 mass % forethylene glycol. The water contents of the purified chemicalswere checked by Karl Fischer titration, and the values were tobe 60 and 25 mass ppm for ethanol and ethylene glycol,respectively.Tables 1 and 2 show the experimental vapor pressures and

Antoine equation constants for ethanol, water, and ethyleneglycol, respectively. The average deviations between theexperimental and the calculated values using the literatureAntoine constants from Riddick and Bunger12 and Boublik etal.13 are 1.58 % and 0.98 % for the ethanol and ethylene glycol,respectively.

EXPERIMENTAL RESULTSBinary Systems. The experimental VLE data at 101.3 kPa

for ethanol + water, water + ethylene glycol, and ethanol +ethylene glycol are shown in Tables 3 to 5 and Figures 2 to 4.The activity coefficients, i, in Tables 3 to 5 were evaluated byeq 1:

= Py P x/i i i iS

(1)

where PiS is the vapor pressure of pure component i. Here the

nonideality of the vapor phase has been neglected, because thebinary coefficients in Tsonopoulos's procedure14 have not beenfound for ethylene glycol.The already published xy data for the binary systems

including ethylene glycol were given in Figures 5 and 6. Theagreement was good for ethanol (1) + ethylene glycol, but thedeviation was large for the water (2) + ethylene glycol (3)system.The experimental VLE data were examined for thermodynamic

consistency using the point test of Fredenslund et al.6 and thearea test of Herington as described by Gmehling and Onken.15

The results of these consistency tests are shown in Table 6. Theexperimental binary VLE data were found to be thermodynami-cally consistent according to the point and area tests.Ternary Systems. Table 7 and Figure 7 show the

experimental VLE data for ethanol + water + ethylene glycol

Figure 1. Apparatus for measuring VLE data (A). (B) Pressure controller; (C) data acquisition; (D) vacuum pump; (E) buffer tank; (F) personalcomputer; (G) modified RogalskiMalanoski equilibrium still; (H) thermometer; (I) stick heater; (J) slidac; (K) refrigerant unit.

Table 1. Experimental Vapor Pressures

P/kPa T/K

Ethanol40.00 329.6753.33 336.0666.66 341.2379.99 345.5993.32 349.3899.74 351.03

Water40.00 349.0953.33 356.1766.66 361.8779.99 366.6993.32 370.88100.0 372.76

Ethylene Glycol13.33 412.8120.00 422.7826.66 430.2633.33 436.3740.00 441.5046.66 445.9653.33 449.9360.00 453.5266.66 456.7873.33 459.7979.99 462.5886.66 465.1893.32 467.6298.66 469.48100.4 470.02

Table 2. Antoine Equation Constants (kPaK)a

substance A B C |Pb/P|av. (%)

ethanol 7.25298 1598.95 46.715 0.02water 7.23605 1754.35 37.731 0.04ethylene glycol 6.69332 1706.33 106.38 0.06

alog P/kPa = A B/(T/K) + C). bP = k|(Pexpt Pcalc)/Pexpt|k/N100, where N is the number of data points.

Journal of Chemical & Engineering Data Article

dx.doi.org/10.1021/je2008704 | J. Chem. Eng.Data 2012, 57, 339344340

at 101.3 kPa. The tails and heads of solid arrows in Figure 7represent the experimental liquid and vapor compositions onthe same tie line, respectively.

DISCUSSIONThe activity coefficients of the three binary systems werecorrelated by the Wilson, NRTL, and UNIQUAC equations.

The objective function in eq 2 was minimized duringoptimization of the parameters in the equations.

=

+

=

Fk

N

k k

obj1

1,calcd 1,exptl

1,exptl

22,calcd 2,exptl

2,exptl

2

(2)

where N is the number of data points. Table 8 lists theestimated parameters of the binary systems and deviations

Table 3. Experimental VaporLiquid Equilibria and ActivityCoefficients i for Ethanol (1) + Water (2) at 101.3 kPa

T/K x1 y1 1 2

373.15 0.000 0.000 1.000368.18 0.018 0.180 5.328 1.001360.50 0.079 0.418 3.731 1.013359.70 0.090 0.441 3.561 1.016357.27 0.147 0.510 2.764 1.045355.38 0.244 0.567 1.991 1.123354.55 0.311 0.589 1.676 1.209354.11 0.353 0.604 1.540 1.263353.45 0.425 0.627 1.362 1.375352.96 0.488 0.652 1.258 1.470352.39 0.570 0.688 1.162 1.606352.11 0.616 0.713 1.127 1.673351.82 0.670 0.736 1.082 1.812351.64 0.711 0.760 1.060 1.895351.37 0.802 0.819 1.024 2.109351.30 0.835 0.844 1.016 2.188351.27 0.869 0.873 1.011 2.246351.26 0.901 0.900 1.006 2.341351.26 0.914 0.912 1.004 2.372351.27 0.922 0.920 1.004 2.376351.29 0.940 0.937 1.002 2.431351.35 0.972 0.969 1.000 2.557351.44 1.000 1.000 1.000

Table 4. Experimental VaporLiquid Equilibria and ActivityCoefficients i for Water (2) + Ethylene Glycol (3) at 101.3kPa

T/K x2 y2 2 3

470.39 0.000 0.000 1.000462.57 0.020 0.227 0.900 1.000448.77 0.063 0.527 0.912 0.998438.69 0.104 0.681 0.914 0.997433.60 0.128 0.742 0.921 0.996426.70 0.167 0.810 0.924 0.994422.03 0.197 0.848 0.930 0.989415.27 0.250 0.892 0.931 0.988411.98 0.279 0.910 0.935 0.982408.00 0.317 0.928 0.942 0.983405.33 0.346 0.939 0.945 0.977400.18 0.409 0.956 0.952 0.982397.27 0.444 0.964 0.968 0.976393.53 0.501 0.973 0.974 0.972390.22 0.557 0.980 0.982 0.952384.51 0.674 0.989 0.989 0.945381.51 0.747 0.993 0.992 0.904379.47 0.797 0.995 0.999 0.895377.30 0.861 0.997 0.999 0.880376.05 0.900 0.998 1.000 0.873373.15 1.000 1.000 1.000

Table 5. Experimental VaporLiquid Equilibria and ActivityCoefficients i for Ethanol (1) + Ethylene Glycol (3) at 101.3kPa

T/K x1 y1 1 3

470.39 0.000 0.000 1.000389.79 0.177 0.961 1.406 1.020384.10 0.219 0.972 1.378 1.025377.16 0.293 0.982 1.308 1.046373.78 0.341 0.986 1.266 1.049371.65 0.376 0.988 1.239 1.068370.51 0.398 0.989 1.219 1.082369.25 0.427 0.990 1.189 1.110368.05 0.451 0.991 1.176 1.117366.65 0.482 0.992 1.158 1.140363.80 0.558 0.994 1.112 1.185361.99 0.611 0.995 1.086 1.250360.28 0.664 0.996 1.066 1.284358.41 0.734 0.997 1.036 1.364356.32 0.810 0.998 1.018 1.451354.00 0.901 0.999 1.002 1.613351.44 1.000 1.000 1.000

Figure 2. Temperaturecomposition diagram and activity coefficient-liquid composition for the binary system ethanol (1) + water (2) at101.3 kPa: (,) experimental values. Curves are calculated using theNRTL equation.

Figure 3. Temperaturecomposition diagram and activity coefficient-liquid composition for the binary system water (2) + ethylene glycol(3) at 101.3 kPa: (,) experimental values. Curves are calculatedusing the NRTL equation.



between the calculated and the experimental vapor-phasecompositions and bubble points using the activity coefficientequations. Table 9 shows the structural parameters ri and qi forUNIQUAC and liquid molar volume Vi

L for the Wilsonequation. Calculated results using the NRTL equation areshown by solid lines in Figures 2 to 4.

The VLE for the ternary ethanol + water + ethylene glycolwas predicted using the binary parameters listed in Table 8, andthe average deviations between the experimental and predictedvapor-phase composition and bubble point using the threeequations are given in Table 10. Predicted results using theNRTL equation are shown by dotted lines in Figure 7. The

Figure 4. Temperaturecomposition diagram and activity coefficient-liquid composition for the binary system ethanol (1) + ethylene glycol(3) at 101.3 kPa: (,) experimental values. Curves are calculatedusing the NRTL equation.

Figure 5. Comparison of xy of published data with the experimentaldata for the water (2) + ethylene glycol (3) system at 101.3 kPa.

Figure 6. Comparison of xy of published data with the experimentaldata for the ethanol (1) + ethylene glycol (3) system at 101.3 kPa.

Table 6. Resultsa of Thermodynamic Consistency Tests ofVLE for Three Binary Systems at 101.3 kPa

point testb area testc

system |y1|av 0.01 [+] D-J 10 [+]

ethanol (1) + water (2) 0.003[+] 28.82[+]water (2) + ethylene glycol (3)

VLE data of Lee and Pahl were calculated using the NRTLequation. The absolute deviations of ethanol vapor composi-

tions and bubble points were 0.013 mole fraction and 0.59 K,respectively.Figure 8 shows the effect of ethylene glycol for separating the

ethanol + water mixture evaluated using the NRTL equation.The azeotropic behavior will disappear at the 0.059 molefraction of ethylene glycol. Therefore, ethylene glycol can beused as an entrainer for the extractive distillation of the binaryazeotropic mixture ethanol + water.

CONCLUSIONThe vaporliquid equilibria for ethanol + water + ethyleneglycol at 101.3 kPa were measured, and the binary parametersof three activity coefficient equations, for example, Wilson,NRTL, and UNIQUAC, were determined. The ternary vaporliquid equilibria were predicted using the binary parameters.The determined parameters of these activity coefficientequations will be useful for the discussion of separation ofethanol aqueous mixtures using extractive distillation.

AUTHOR INFORMATIONCorresponding Author*Fax: +81-3-3293-7572. E-mail: [email protected].

REFERENCES(1) Washall, T. A. Separation of Water from a Single Alkanol byExtractive Distillation with Ethylene Glycol. U.S. Patent 3,464,896,1969.(2) Ladisch, M. R.; Dyck, K. Dehydration of Ethanol: New ApproachGives Positive Energy Balance. Science 1979, 205, 898900.(3) Lee, F.-M.; Pahl, R. H. Dehydration of Alcohol With ExtractiveDistillation. U.S. Patent 4,559,109, 1985.(4) Kireev, V. A.; Polov, A. A. Zh. Prikl. Khim. 1934, 7, 489.(5) Li, J.; Chen, C.; Wang, J. Vapor-liquid equilibrium data and theircorrelation for binary systems consisting of ethanol, 2-propanol, 1,2-ethanediol and methyl benzoate. Fluid Phase Equilib. 2000, 169, 7584.(6) Fredenslund, A.; Gmehling, J.; Rasmussen, P. Vapor-LiquidEquilibria Using UNIFAC; A Group-Contribution method; Elsevier:Amsterdam, 1977.(7) Herington, E. F. G. Tests for Consistency of ExperimentalIsobaric Vapor Liquid Equilibrium Data. J. Inst. Pet. 1951, 37, 457470.(8) Wilson, G. M. Vapor-liquid equilibrium. XI. A New Expressionfor the Excess Free Energy of Mixing. J. Am. Chem. Soc. 1964, 86,127130.(9) Renon, H.; Prausnitz, J. M. Local Compositions in Thermody-namic Excess Functions for Liquid Mixtures. AIChE J. 1968, 14, 135144.

Table 8. Activity Coefficient Model Parameters and Deviations between Experimental and Calculated Values

Aij Aji |T|ava |y1|avb |y2|avb

model system Jmol1 Jmol1 ij K mole fraction mole fraction

NRTL 1 + 2 227.56 5196.9 0.40 0.07 0.0021 + 3 1035.6 708.38 0.23 0.10

(10) Abrams, D. S.; Prausnitz, J. M. Statistical thermodynamics ofliquid mixtures: a new expression for the excess Gibbs energy of partlyor completely miscible systems. AIChE J. 1975, 21, 116128.(11) Kurihara, K.; Minoura, T.; Takeda, K.; Kojima, K. IsothermalVapor-Liquid Equilibria for Methanol + Ethanol + Water, Methanol +Water, and Ethanol + Water. J. Chem. Eng. Data 1995, 40, 679684.(12) Riddick, J. A.; Bunger, W.; Sakano, T. Organic Solvents PhysicalProperties and Methods of Purification, 4th ed.; John Wiley & Sons:New York, 1986.(13) Boublik, T.; Fried, V.; Hala, E. The Vapor Pressures of PureSubstances; Elsevier: Amsterdam, 1973.(14) Tsonopoulos, C. Empirical Correlation of Second VirialCoefficients. AIChE J. 1974, 20, 263272.(15) Gmehling, J.; Onken, U. Vapor-Liquid Equilibrium DataCollection; Chemistry Data Series; DECHEMA: Frankfurt, 19771982.



equilibrio etilenglicol y agua

Documents