Fluid Phase Equilibria, 61 (1990) 89-98

Elsevier Science Publishers B.V., Amsterdam

89

LIQUID-LIQUID EQUILIBRIUM (C,,-C,,)-SULFOLANE SYSTEMS

ASHA MASOHAN, GUPTA SRIKANT and BACHAN M. NANOTI,

STUDIES

ON HYDROCARBON

KRISHAN *

G. SHARMA,

SOM

N. PURL

PUSHPA

S. RAWAT

Indian Institure of Petroleum,

(Received March 22, 1990;

Dehra Dun 248 005 (India)

accepted in final form July 4, 1990)

ABSTRACT Masohan, A., Nanoti, S.M., Sharma, studies K.G., Puri, S.N., Gupta, P. and Rawat, B.S., 1990.

Liquid-liquid

equilibrium

on hydrocarbon

(C,,-C,,)-sulfolane

systems.

Fluid

Phase Equilibria,
Equilibrium model tained

61: 89-98.
at 60 and 100 o C with sulfolane octylbenzene-cetane. by NRTL The data and five so obmodel

tie line data have been determined mixtures predicted consisting by UNIFAC octylbenzene-dodecane and

hydrocarbon have been

of butylbenzene-dodecane, and correlated

hexylbenzene-dodecane, and UNIQUAC

hexylbenzene-cetane, equations.

The alkylbenzenes

used in the study were prepared

in the laboratory.

INTRODUCTION

Until now the design and simulation of solvent extraction processes for medium petroleum fractions have been done mostly by fully empirical methods, which often require long and costly experimental work. The major problems in the mathematical simulation of the processes for such fractions are the complexity of the mixtures, the scarcity of basic data and the very limited availability of suitable thermodynamic models. Rational design of the processes requires characterization of the complex petroleum fraction and models for the prediction of their liquid-liquid and vapour-liquid equilibria (LLE and VLE). It has been shown (Mehrotra et al., 1986) that representation of a particular petroleum fraction, e.g. aviation turbine fuel (ATF), by model hydrocarbons, together with the equilibrium predictions by UNIFAC, forms a sufficiently accurate database for the simulation model. On similar lines, higher petroleum fractions like kerosene

* To whom correspondence 0378-3812/90, $03.50

should be addressed. 0 1990 - Elsevier Science Publishers B.V.

90 and gas oil can be represented by models consisting of key components of representative hydrocarbon types of average carbon numbers (Rahman et al., 1984). Design of extraction columns for systems containing these higher fractions makes it essential to generate equilibrium data on synthetic mixtures of known hydrocarbons (C,,-C,, range) with newer and industrial solvents. Various thermodynamic models such as NRTL, UNIQUAC, ASOG and UNIFAC are available for the prediction of liquid-liquid equilibria. The estimation of the binary NRTL or UNIQUAC parameters also requires the availability of experimental LLE, VLE or infinite dilution activity coefficient data. Such equilibrium data on higher hydrocarbon components (ClOC,,) and solvents are not available. The present paper describes the preparation of C,, C, and C8 alkylated benzenes and LLE studies on sulfolane-hydrocarbon systems involving the above alkylbenzenes and dodecane-cetane, which represent middle distillates. The experimental data were predicted by UNIFAC (Magnussen et al., 1981) and correlated by the UNIQUAC (Abrams and Prausnitz, 1975) and NRTL (Renon and Prausnitz, 1968) model equations.
EXPERIMENTAL

Synthesis of alkylbenzenes The butyl-, hexyl- and octyl-benzenes were prepared in the laboratory by the Friedel-Crafts reaction of benzene with the corresponding olefin (Olah, 1964); the latter in turn were prepared by dehydration of butanol, hexanol and octanol respectively (Eley et al., 1966). The final reaction product obtained in the preparation of butyl-, hexyl- or octyl-benzene was subjected to fractional distillation. The heart cuts representing the isomers of butyl-, hexyl- and octyl-benzene were the 170-175, 213-214 and 235-264C boiling fractions, respectively, for which the product yields were maximum. These cuts were taken for the present LLE studies. Purities Purities of the above three alkylbenzenes were around 99 wt.%, as determined by mass spectroscopy. As can be seen from Table 1, the density and refractive index of isomers of butyj-, hexyl- and octyl-benzenes determined experimentally matched fairly well with those reported (Rossini et al., 1953; Egloff, 1946; Hawley, 1971) for the corresponding pure isomers. Dodecane and cetane were obtained from Humphary Chemical Co., Connecticut, U.S.A. The purities of these two paraffins were better than 99

TABLE 1 of pure substances Density at 20 o C Refractive Literature 0.8621 0.8557 0.8582 0.7490 0.7734 1.2610 (Rossini et al., 1953) (Egloff, 1946) (Egloff, 1946) (Hawley, 1971) (Hawley, 1971) (Hawley, 1971) 1.4892 1.4868 1.4845 1.4218 1.4340 1.4807 value Experimental Experimental 0.8621 0.8598 0.8572 0.7486 0.7730 1.2605 index values are at 30C. value index at 20 C value Literature 1.4902 1.4859 1.4851 1.4221 1.4344 1.4810 value (Rossini (Egloff, (Egloff, (Hawley, (Hawley, (Hawley, et al., 1953) 1946) 1946) 1971) 1971) 1971)

Physical properties

Product

set-Butylbenzene 3-Phenylhexane I-Phenyloctane Dodecane Cetane Sulfolane

a Density and refractive

92

wt.%, as determined by gas chromatography. Philips Co., U.S.A. Determination of liquid-liquid equilibria

Sulfolane was obtained from

Liquid-liquid equilibria were determined at 60 and 100 o C in single stage batch equilibrium runs carried out in a double-walled mixer-settler of 250 ml capacity with a thermometer pocket and a syringe stirrer. The desired temperature was maintained to within + 0.05 o C using a thermostatic bath. Varying quantities of aromatic (alkylated benzene) and saturated (dodecane-cetane) hydrocarbons and sulfolane were taken in the mixer-settler and the contents were stirred for 15 min for establishment of equilibrium between the phases. The phases were allowed to settle for 15 min and then withdrawn separately for analysis. Solvent was removed from the raffinate phase by water washing and from the extract phase by azeotropic distillation with water. This analytical technique has been standardized and reported by Rawat and Gulati (1976). The aromatic contents in the solvent free extract and raffinate were determined by gas-liquid chromatography. For each particular system of alkylated benzene-paraffin-sulfolane, tie line data were obtained using different compositions of aromatic in saturate and keeping the solvent/feed ratio near 2.
RESULTS AND DISCUSSION

The experimental LLE data determined at 60 and 100 o C with sulfolane and model binary hydrocarbon mixtures consisting of butylbenzene-dodecane, hexylbenzene-dodecane, octylbenzene-dodecane, hexylbenzenecetane and octylbenzene-cetane are presented in Table 2. From the ternary LLE data of all the above systems, the binary NRTL and UNIQUAC parameters were estimated using the method of Sorensen (1980). The method starts with minimization of an activity objective function, Fa (eqn. (1)) which does not require qualified guesses of the parameters.

Once the convergence is achieved, the search shifts to minimize the concentration objective function, F, (eqn. (2)), which fits the calculated and experimental concentrations.

(2)

93 TABLE 2 LLE data of the systems alkylbenzene-dodecane/cetane-sulfolane Raffinate (mol%) Saturate Octylbenzene-cetane-sulfolane 60 79.69 80.13 79.59 63.84 60.52 52.25 46.66 30.85 26.35 11.00 100 80.25 75.92 63.87 66.90 Octylbenzene-dodecane-sulfolane 60 91.15 77.40 38.03 16.27 2.80 100 85.93 63.20 37.20 2.39 Hexylbenzene-cetane-sulfolane 60 79.63 58.17 45.75 31.85 10.88 100 79.63 58.63 44.17 31.55 Hexylbenzene-dodecane-sulfolane 60 87.46 65.89 58.00 27.98 9.32 2.02 10.65 30.94 38.92 65.67 80.73 83.81 1.89 3.17 3.08 6.35 9.95 14.17 0.24 0.38 0.37 0.22 0.11 0.00 0.73 2.32 3.49 6.04 9.44 10.61 99.03 97.30 96.14 93.74 90.45 89.39 (continued) 16.84 38.86 52.19 65.58 79.77 16.60 35.05 47.20 58.20 3.53 2.97 2.06 2.57 9.35 3.77 6.32 8.63 10.25 0.11 0.05 0.05 0.05 0.00 0.17 0.22 0.16 0.22 1.42 2.92 3.98 5.67 8.62 1.36 3.58 5.44 7.70 98.47 97.03 95.97 94.28 91.38 98.47 96.20 94.40 92.08 6.61 21.16 57.43 76.40 86.75 9.99 32.22 54.30 76.95 2.24 1.44 4.54 7.33 10.45 4.08 4.58 8.50 20.66 0.13 0.29 0.18 0.08 0.00 0.55 0.75 0.50 0.07 0.36 0.88 2.47 3.38 3.95 0.58 2.12 3.94 9.51 99.51 98.83 97.35 96.54 96.05 98.87 97.13 95.56 90.42 14.93 17.87 20.81 33.49 36.16 43.90 50.44 64.04 66.68 83.00 14.02 19.62 31.58 27.67 5.38 2.00 4.60 2.67 3.32 3.85 2.90 5.11 6.97 6.00 5.73 4.46 4.55 5.43 0.16 0.05 0.07 0.05 0.05 0.08 0.11 0.01 0.00 0.01 0.21 0.21 0.22 0.16 0.44 0.38 0.87 0.88 1.15 1.31 2.11 2.56 2.76 2.96 0.83 0.95 1.79 2.83 99.40 99.57 99.06 99.07 98.80 98.61 97.78 97.43 97.24 97.03 98.96 98.84 98.99 97.01 Aromatic Solvent phase composition extract phase composition (mol%) Saturate Aromatic Solvent

Experimental Temperature ( C)

94 TABLE 2 (continued) Temperature (C) Raffinate (mol%) Saturate Hexylbenzene-dodecane-sulfolane 100 83.47 61.88 43.23 22.30 13.33 6.18 Butylbenzene-dodecane-sulfolane 60 88.10 69.00 50.33 39.67 18.80 100 82.07 65.74 51.92 38.97 51.92 38.97 phase composition Aromatic 12.37 32.77 49.88 65.30 67.82 64.33 10.25 29.33 46.76 55.95 71.53 13.99 30.54 42.11 53.55 42.11 53.55 Solvent 4.16 5.35 6.89 12.40 18.85 29.49 extract phase composition (mol%) Saturate 0.64 0.79 0.65 0.66 0.67 0.53 Aromatic 1.65 4.00 6.23 10.32 14.62 21.15 Solvent 97.71 95.21 93.12 89.02 84.71 78.32

1.65 1.67 2.91 4.38 9.67 3.94 3.72 5.97 7.48 5.97 7.48

0.26 0.43 0.29 0.31 0.38 0.79 0.75 0.87 1.05 0.87 1.05

0.89 4.02 7.28 10.22 17.53 0.00 5.42 8.54 12.73 8.54 12.73

98.85 95.55 92.43 89.47 82.09 99.21 93.83 90.59 86.22 90.59 86.22

The model parameters, along with the root-mean-square deviation (RMSD) values, are given in Tables 3 and 4. Experimental LLE data at 60 and 100 o C were also predicted by the UNIFAC model, and RMSD values are given in Tables 3 and 4. For computation using UNIFAC, the hydrocarbons were represented by standard UNIFAC groups (Magnussen et al., 1981) while sulfolane was considered as one independent group. The hydrocarbon-hydrocarbon temperature-dependent interaction parameters have been taken from Rahman et al. (1984) and the sulfolane-hydrocarbon temperature independent parameters were taken from Magnussen et al. (1981). The RMSD between the experimental and calculated extract and raffinate phase compositions were calculated as [eqn. (3)] C C C ( x,J~ - X;jk)2/6M
k i 1

It is seen that, of all the models, the UNIQUAC model generally correlates fairly well with the experimental equilibrium data at both 60 and 100 C. The predicted values obtained by UNIFAC also match reasonably well with the experimental values at 60 o C. However, slightly higher RMSD values at

1
l/2

95 TABLE Model System 3 parameters obtained from ternary LLE data and RMSD NRTL Ag,,/R Dodecane-octylbenzene-sulfolane (1) (2) (3) (l)-(2) (l)-(3) (2)-(3) RMSD, mol% - 201.57 928.03 376.49 0.26 249.26 1295.20 852.37 - 99.52 353.81 303.75 0.21 150.16 145.46 - 9.32 0.30 ( OL = 0.2) Ag,,/R values at 60 o C UNIFAC Au,, /R

UNIQUAC AU,,/R

Dodecane-hexylbenzenesulfolane (1) (2) (3) (l)-(2) (l)-(3) (2)-(3) RMSD, mol% - 377.69 858.39 424.72 0.25 414.22 1321.80 582.09 - 11.27 358.58 294.47 0.22 92.86 110.97 - 23.32 0.51

Dodecane-butylhenzene-sulfolane (1) (2) (3) (l)-(2) (l)-(3) (2)-(3) RMSD, mol% 244.75 1082.70 476.91 0.22 151.79 1278.80 312.48 9.51 483.31 231.09 0.18 - 24.15 44.72 - 13.16 0.90

Cetane-hexylbenzene-sulfolane (1) (2) (3) (l)-(2) (l)-(3) (2)-(3) RMSD, mol% - 718.89 854.56 452.32 0.71 1118.10 1590.40 574.29 205.28 167.82 189.37 0.56 305.89 702.97 37.47 0.40

Cetane-octylbenzene-sulfolane (1) (2) (3) (l)-(2) (l)-(3) (2)-(3) RMSD, mol% - 219.91 626.37 497.21 0.61 290.92 1522.10 809.12 - 296.91 391.94 488.01 0.57 723.16 20.85 -45.76 0.20

100 o C are perhaps due to the limited number of available sulfolane-hydrocarbon parameters at higher temperatures for such systems. The experimental LLE data for a typical kerosene cut (140-240 o C) with sulfolane at 60 C were predicted using the detailed model suggested by Mehrotra et al. (1986). The RMSD value obtained was 0.1404 wt.%. This kerosene, when represented by propylbenzene-dodecane and butylbenzene-dodecane hydrocarbon model mixtures, showed RMSD of

96 TABLE 4 Model parameters System obtained from ternary LLE data and RMSD values at 100 o C UNIFAC AU,,/R

NRTL ((Y= 0.2) k, /R Ag,JR

UNIQUAC WJR

Dodecane-octylbenzene-sulfolane (I) (2) (3) (I)-(2) (I)-(3) (2)-(3) RMSD, mol% ~ 332.62 786.39 279.19 0.28 231.04 1298.80 819.86 - 63.43 426.16 309.04 0.24 49.09 31.22 - 32.73 0.87

Dodecane-hexylbenzene-sulfolane (I) (2) (3) (l)-(2) (I)-(3) (2)-(3) RMSD, mol% - 678.59 866.64 216.06 0.70 734.55 1442.80 657.88

_ 130.30
351.41 213.00 0.64

126.75 140.55 1.61 0.16

Dodecane-butylbenzene-sulfolane (1) (2) (3) (I)-(2) (I)-(3) (2)-(3) RMSD, mol% -51.15 762.67 759.88 0.57 -271.17 1358.30 95.46 143.00 431.23 375.43 0.58 - 172.27 29.06 - 105.29 0.95

Cetane-hexylbenzene-sulfolane (3) (I)-(2) (I)-(3) (2)-(3) RMSD, mol% Cetane-octylbenzene-sulfo!dne (I) (2) (3) (I)-(2) (I)-(3) (2)-(3) RMSD, mol% (1) (2) 139.85 716.33 404.00 0.21 - 231.58 1545.10 580.46 379.02 442.78 362.85 0.12 - 256.22 56.79 - 68.75 0.85

- 373.79 514.58 888.87 0.41

80.46 1610.40 343.95

_ 128.40
361.55 627.04 0.38

113.06 44.5 1 - 150.84

1.08

1.1350 and 2.6767 wt.%, respectively. These results suggest that the propylbenzene-dodecane model mixture represents this kerosene cut better than the butylbenzene-dodecane model.
CONCLUSIONS

The experimental dodecane-sulfolane

tie line data for the alkylated benzenes-cetane/ systems are correlated best by the UNIQUAC model.

97

These data have also been predicted by UNIFAC, and compare well with the experimental data. Kerosene (140-240 OC) is represented better by the propylbenzene-dodecane model mixture than by the butylbenzene-dodecane model mixture.
ACKNOWLEDGEMENTS

The authors wish to thank the gas chromatography group for analysing the extract and raffinate phases, the crude evaluation group for carrying out TBP distillation and the mass spectroscopy group for determining the purity of the alkylbenzenes used in the experiments.
LIST OF SYMBOLS

activity required ratio between the solute (U or L) concentrations phases I and II at infinite dilution A *,,, Ag,, NRTL parameters (J mall ) M number of experimental tie lines P parameter value constant Q R gas constant (J mall K-l) Aqj, A&$,UNIQUAC parameters (J mol- ) experimentally determined mole fraction X, ;
Greek letters

in

(Y NRTL non-randomness y activity coefficient

Superscript * estimated Subscripts i

parameter

values

j
k

I n U L

components phases tie lines data sets parameters upper component left component

98 REFERENCES 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. Egloff, G., 1946. Physical Constants of Hydrocarbons, American Chemical Society, Reinhold, New York, Vol. III. Eley, D.D., Pines, H. and Weisz, P.B., 1966. The Mechanism of Dehydration of Alcohols and Alumina Catalyst. Academic Press, New York, Vol. 16, p. 49. Hawley, G.G., 1971. The Condensed Chemical Dictionary, Van Nostrand Reinhold, New York, 8th edn. Magnussen, T., Rasmussen, P. and Fredenslund, Aa., 1981. UNIFAC parameter table for prediction of liquid-liquid equilibria, Ind. Eng. Chem. Process Des. Dev., 20(2): 331-339. Mehrotra, R., Garg, M.O., Chopra, S.J., Rawat, B. and Khanna, M.K., 1986. Liquid-liquid phase equilibria for dearomatisation of ATF fraction, Fluid Phase Equilibria, 32: 17-25. Olah, G.A., 1964. Advances in Catalysis: Friedel-Crafts and Related Reactions by Others, Interscience, New York, Vol. II, Part 1. Rahman, M., Mikitenko, P. and Asselineau, L., 1984. Solvent extraction of aromatics from middle distillates: equilibria prediction method by group contribution, Chem. Eng. Sci., 39(11); 1543-1558. Rawat, B.S. and Gulati, I.B., 1976. Liquid-liquid equilibrium studies for separation of aromatics. J. Appl. Chem. Biotechnol., 26: 425-435. Renon, H. and Prausnitz, J.M., 1968. Local compositions in thermodynamic excess functions for liquid mixtures. AIChE J., 14: 135-144. Rossini, F.D., Pitzer, KS., Arnett, R.L., Brawn, R.M. and Pimentel, G.C., 1953. Selected Values of Physical and Thermodynamic Properties of Hydrocarbons and Related Compounds. API, Carnegie Press, Pittsburgh, PA, U.S.A. Sorensen, J.M., 1980. Phase equilibria and separation processes. MAN 8106. ESTM: Estimation of UNIQUAC and NRTL parameters from ternary LLE data. Instituttet for Kemiteknik, Lyngby, Denmark.