Ind. Eng. Chem. Process Des. Dev. 1988, 25, 728-733
728
)b=O A1 A2
The optimization criterion requires the proposition of different values for until up in (B-4) is minimized. Registry No. Methyl oleate, 112-62-9; methyl elaidate,
= -1
(A-18)
= -l/p
B. The experimental data may be dealt with in a transformed version of (A-17) where
i = 1,...,n
y i = In (wT - wC);
xi = In [ ( S ~ T- OC)/(@.WO q = In [wTo
P =
- wc0)li
+ wcO]
(B-1)
Al/A2
Equation A-17 becomes then ~i
= Q + Pxi
03-2)
Both xi and y i are subject to the same type of random errors. Their deviations from the "true" values x / and yi' are given by {i
= xi - x
i
6i = yi - yi'
(B-3)
Several estimation methods for p , in the linear function (B-21, are given by Guest (1961). Assuming that the standard deviations are the same for all the experimental points and that at: = ut12,the slope p and its standard deviation are given by p =m
+ (1+ m2)1/2
(B-4)
where
m = ( C y i * 2- ,&*2)/2Cxi*yi* xi* = xi - R
z = &/n
1937-62-8.
Literature Cited Albright, L. F.; Wlsniak, J. J. Am. Oil Chem. SOC. 1962, 3 9 , 14. Albright, L. F. Chem. Eng. Sci. 1967, 11, 197. Allen, R. R.; Kiess, A. A. J. Am. 011 Chem. SOC. 1955, 3 2 , 400. Allen. R. R.; Kless, A. A. J. Am. Oil Chem. SOC. 1958, 3 3 , 355. Beckmen, H. J. J. Am. OllChem. SOC. 1983, 6 0 , 2, 282. Benson, S. W.; Cruickshank, F. R.; Golden, D. M.; Hougen, G. R.; O'Neal, H. E.; Rodgers, A. S.;Shaw, R.; Walsh, R. Chem. Rev. 1969, 6 9 , 279. Cordova, W. A.; Harriott, P. Chem. Eng. Sci. 1975, 3 0 , 1201. Draguez de Hauit, E.; Demoulin. A. J. Am. Oil Chem. SOC. 1984, 6 1 , 2, 195. Duncan, D. P. J. Am. OllChem. SOC. 1984, 6 1 , 2, 233. Eldlb, I. A.; Aibrlght, L. F. Ind. Eng. Chem. 1957, 49, 825. Grau, R. J.; Cassano, A. E.; Baltanas, M. A. Ind. Eng. Chem. Fundam., in press. Guest, P. G. "Numerical Methods of Curve Fitting"; Cambridge University Press: New York. 1961. Gut, G.; Kosinka, J.; Prabucki, A.; Schuerck, A. Chem. Eng. Sci. 1979, 3 4 , 1051. Hashlmoto, K.; Katsuhiko, K.; Nagata, S. J. Am. Oil Chem. SOC. 1971, 48, 291. Horluti, J.; Polanyl, M. Trans. Faraday SOC. 1934, 3 0 , 1164. Larsson, R. J. Am. OilChem. SOC. 1983, 6 0 , 2, 275. Lichtfleld, C.; Lord, J. E.: Isbell, A. F.; Relser, R. J. Am. Oil Chem. SOC. 1963, 40, 553. Lichtfield, C.; Harlow, R. D.; Isbell, A. F.; Reiser, R . J. Am. Oil Chem. SOC. 1965, 42, 73. Lidefelt, J. 0.; Magnusson, J.; Schoon. N. H. J. Am. Oil Chem. SOC. 1963, 60, 3, 588. Lombardo, E. A.; Hall, W. K. AIChE J. 1971, 17, 1229. Magnusson, J. Ph.D. Thesis, Chalmers University of Technology, Goteborg, Sweden, 1983. Pihl, M.; Schoon, N. H. Acta Polytech. Scand.. Appl. Phys. Ser. 1971, 100, 3. Puri, P. S. J. Am. OilChem. SOC. 1080, 850A. Rogers, D. W.; Hoyte, 0. P. A.; Ho, R. K. C. J. Chem. Soc., Faraday Trans. 11977, 74, 46. Smlth, R. L.; Prater, C. D. Chem. Eng. Prog. Symp. Ser. 1968, 63, 73, 105. Stingley, D. V.; Wrobel, R . J. J. Am. OilChem. SOC. 1961, 3 8 , 201. Swern, D. "Industrial Oil and Fat Products"; Interscience: New York, 1964. Van der Planck, P.; Linsen, B. 0.; Van der Berg, Proceedings of the 5th EuropeanlPnd International Symposium on Chemical Reaction Engineering, Amsterdam, May 2-4, 1972. Wei, J.; Prater, Ch. D. Adv. Catal. 1962, 13, 203. Wisniak, J.; Albright, L. F. Ind. Eng. Chem. 1961, 5 3 , 5 , 375.
Received for review May 29, 1985 Accepted December 9, 1985
Capacity and Hydrodynamics of an Agitated Extraction Column An11 Kumar, Ladldav Stelner, and Stanley Hartland' Department of Chemical Engineering and Industrial Chemistty, Swiss Federal Institute of Technology, 8092 Zurich, Switzerland
The influence of operating variables on the hydrodynamics of a Kuhni-type agitated extraction column is considered. The effect of the phase ratb and agkation intensity on the column capacity with and without mass transfer is detailed. The influence of the small drop population on capacity of the column is descrlbed, and equations for the prediction of maximum throughput are suggested. The capacity may be increased by coalescing the small drops entrained in the continuous phase on packing inserted into the column base.
Although extraction columna have been in industrial use for many years, there is still a need for reliable scale-up data. Behavior of agitated extraction columns is particularly difficult to simulate. There are many independent variables which provide flexibility in the operation of the column but which complicate the design and optimization of the process variables. It is still commonly presumed that the dispersed-phase holdup and interfacial area may be characterized by mean 0196-4305/06/1125-0720$01.50/0
values though departures from this assumption have been reported by Strand et al. (1962),Misek (1964),Defives and Schneider (1961), and Bell and Babb (1969) in different types of extraction columns. The relative velocities of the phases and flood points were investigated in a Kuhni-type agitated extraction column and predicted by using mean values of holdup and drop size. The entrainment of small drops may lead to flooding of the column. These drops can be coalesced on 0 1986 American Chemical Society
Ind. Eng. Chem. Process Des. Dev., Vol. 25, No. 3, 1986 729
~~~
~~
Figure 1. Column internal construction: shaft diameter, 20 mm; hole diameter, 15 mm; three spacer rods each of diameter 8 mm. (1) Teflon bearing; (2) turbine; (3) shaft; (4) perforation; (5) shaft support.
6
I:
Table I. Dimensions of Kuhni Extraction Column Type E 150/18
column diameter, Dk,m active column height, Hk, m no. of compartments compartment height, H,,m fractional free area of stator plate, 4 shrouded six-bladed turbines diameter, D,, m height, H,,m end section length, HE, m
0.150 1.260 18
diameter, DE, m packing (in bottom end section)
0.200
0.07
0.235, 0.40
0.085
0.010
0.600 (upper) 0.550 (lower) Melapack (Sulzer AG, Switzerland)
orientated packing inserted into the column base, which may be used to increase the capacity of all types of columns. The packing used also distributed the incoming dispersed phase over the column cross section. Additionally, the axial variation of holdup and drop size was investigated.
Experimental Work Figure 1 shows the column internal construction, with detailed specifications given in Table I. Perforated plates with 23.5% and 40% free area were used. The column is equipped with 18 shrouded 6-bladed turbines which provide efficient radial dispersion. In the column, there is countercurrent flow of the light and heavy phases with circulation from the turbines superimposed on the main flow. Axial variation in the holdup may be expected: The test system used in the present study was deionized water and distilled o-xylene. The solute was acetone which was transferred in both directions, o-xylene being always dispersed. The sampling arrangements for the measurement of holdup, drop size, and other hydrodynamic pa-
Figure 2. Experimental column showing measuring devices: (1) holdup (direct sampling);(2) drop size (capillary with window); (3) dispersed-phase axial mixing (LED probes); (4) continuous-phase axial mixing (stainless steel electrodes); (5) concentration (2 mm i.d. tubes; dispersed-phase tube with Teflon coalescer);(6) tracer injection.
rameters are shown in Figure 2. To measure the holdup, pressure drop, conductivity, and direct sampling methods were tried. The pressure drop was found to be a function of the operating variables. It was possible to allow for the effect of agitation intensity and continuous-phase flow on the pressure drop but not for that of the dispersed-phase flow. Conductivity was found to be sensitive to the location of the electrodes due to the internal metal construction. Further development of this method could lead to its successful application. Direct sampling was thus used to determine the holdup profiles in the present study because of its simplicity and reproducibility. After flushing the valves (Figure 2), instantaneous samples of about 100 mL each were taken from stages 2, 5, 8,11,14, and 17 counted from the dispersed-phase inlet. To facilitate measurement of the drop size, the drops were photographed through a flat optical glass window separated from the column wall by the continuous phase to minimize optical distortion. This method was suitable for low agitation speeds, low holdups, and mean drop sizes (&) greater than 4 mm. For higher holdups and smaller drop sizes, the drops in the column were collected in a specially shaped funnel and sucked into a capillary with an inner diameter of approximately 2 mm, in which they were photographed outside the column. This method could be used for all drop sizes between 0.05 and 5 mm. In order to avoid optical distortion, the photographs were taken through a flat optical glass window fixed in a ca-
730
Ind. Eng. Chem. Process Des. Dev., Vol. 25, No. 3, 1986 0.5,
pillary tube housing filled with the continuous phase.
Correlating Holdup with Phase Flow Rates For the plug flow, the relative velocity V, of drops with respect to the counterflowing continuous phase is defined by
The constriction factor 4 is the minimum value of the fractional-free cross-sectional area in the column available for flow. The relative velocity V, is normally expressed in terms of a characteristic velocity v k and a holdup function. v k is the relative velocity of droplets extrapolated to zero holdup; it depends on the drop size and physical properties and is related to the drop terminal velocity. Thomton (1956) and Strand et al. (1962) correlated their results with (2). v, = vk(1- x d ) (2) This equation applies when the mean drop size is independent of the flow rates and dispersed-phase holdup. This was also observed experimentally for the present column and system used. For the axisymmetric rotating-disk extractor (ARD), Misek (1963) and Misek and Marek (1970) proposed v, = vk(1 - x d ) exp[Xd(Z - 4.1)] (3) where vk
= 0.249d4,(g2Ap2/p,c1c)1/3
(4)
and the coalescence factor 2
= 0.0159[ (P$k/pc) ( (5) Koide et al. (1967) obtained an approximate expression for the Zenz (1957) correlation in the form v, = Vk[0.27 + 0.73(1 - Xd)’”] (6) where v k is obtained from equations proposed by Klee and Treybai (1956) using the Sauter mean diameter. A simple equation was presented by Kumar et al. (1980) for fractional holdups up to 0.75 v, = vk[(1 - Xd)/(l + xd113)]n/2 7 5 Re 5 2450 (7) where vk
= [KApgd/pcl”’
(8)
with
K = 2.725 n = 1.834 This was derived by using data for spray columns but should in principle apply to all cases of countercurrent flow. Bibaud and Treybal(1966) gave a correlation to predict the characteristic velocity for a mechanically agitated extractor as vk = 1.77 X 10-4(ai/~c)(g/Dfl)(A~/~c)0.9 (9) From the previous works on the Kiihni column, no relevant data are available for mass-transfer experiments. In the absence of mass transfer, Fischer (1973) suggests the use of a complicated method to determine relative velocity. The method uses a modified plot of the Archimedes number ( A F )against the Froud number (FF)for solid spheres. An empirical equation is then suggested which converts the values for solid spheres to that applicable to drop swarms. His own analysis shows an accuracy of *30%. Simple correlations are suggested below which predict the relative velocity and throughputs with better accuracy.
$
,
,
,
,
t _Symbol + 100 1 17 1 4 0 , x 180
,
,
,
,
,
2LE
0. 4
0 . 00
0.05
i
/
0 . 10
0. 15
0. 2 0
0 . 25
XdCl-Xd)
Figure 3. Evaluation of characteristic velocity from slope of straight line plots of eq 1 and 2. $I = 23.5%. Table 11. Average Errors between Experimental V, Values from Equation 1 and Predicted Values in Absence of Mass Transfer. Plate Free Area 23.5% V, predicted av error V, predicted by using eq by using eq (fraction) 10 2 0.076 4 2 0.116 4 3 0.255 8 7 0.413 Klee and Treybal (1956) 6 0.239
Correlating Present Experimental Results The ratio of the Weber number (We)to the Reynolds number (Re) could be expressed for the present extractor and system by v k = 0.82 x io-4(ai/pc)(g/Dfl) (10) with 4.6
< ( g / D P ) < 41.6
The dependence of characteristic velocity on agitation intensity given by eq 10 was obtained from a plot, Figure 3, of the experimental v d + V,&/(l- x d ) values against Xd(1 - Xd). The relative velocities obtained from eq 1 using experimental mean holdup values were compared to values obtained from eq 2 with v k values from eq 10. The use of eq 10 results in an average error of only 7.6%, and knowing the operating variables (V,, vd, and N), it can be used with eq 1 and 2 to obtain mean holdup values. The use of other equations gave much higher average errors and is not recommended for the present column and system. Table I1 summarizes these results. Equations 4, 8, and 9 (and also that of Logsdail et al. (1957))show that V, depends on the density difference and continuous-phase density, the exponent on Ap taking a value between 0.5 and 1.0 (the upper limit corresponding to Stokes law) and on pc between -0.33 and -0.9. The use of only one system in the present study did not justify the inclusion of Ap and pc in eq 10. The effect of densities is included in the value of the constant.
Generalized Correlation The relative velocity may be expressed as a function of the physical properties and operating variables through the dimensionless equation
The constant K1 in eq 11 for the present column and
Ind. Eng. Chem. Process Des. Dev., Vol. 25, No. 3, 1986 731 Table 111. Values of Constant K i n Equation 11 with and without the Holdup Term and Average Errors between Experimental V, Values from Equation 1 and Predicted V, Values from Equation lla case i case ii case iii K1 0.77 X lo-' 0.96 X lo4 1.67 X lo-' 0.153 0.192 fractional av error 0.059 Kz 0.70 X lo4 0.88 X lo-' 1.58 X lo4 0.137 0.185 fractional av error 0.069
m
II
Flooding Holdup can be used as a measure of the approach of the extractor operating conditions to the flooding point since holdup increases with increasing flow rates V, and v d . Flooding was normally characterized by rejection of the light phase at the aqueous outlet and formation of a second interface or dense packed layer of small droplets a t the column base. In certain cases, flooding occurred by phase inversion in a particular stage. In the column, very small drops are carried down with the continuous phase and accumulate below the lowest stage. A suitable packing placed in the base of the column reduced the local turbulence, thus allowing the small drops to coalesce and rise back into the column, hence delaying the onset of flooding and thereby increasing the capacity of the column. Flooding with the packing present was indicated with the rejection of coalesced organic phase and
I
/
/
,
,
:
I
x
\
'K1= constant in eq 11 with 1 - Xdterm. K 2= constant in eq 11 without 1 - Xd term. Case i = no solute transfer. Case ii = solute transfer continuous to dispersed phase. Case I11 = solute transfer dispersed to continuous phase. system was obtained by linear regression of the data from experiments using 23.5% and 40% free area perforated plates. Since the system behaves differently in the presence and absence of solute transfer, the data analysis was divided into three cases: ti) no solute transfer; (ii) solute transfer from the continuous (c) to the dispersed (d) phase, Le., noncoalescing system; and (iii) solute transfer from the (d) to the (c) phase, i.e., coalescing system. For experimenta in the absence of mass transfer, only 23.5% free area plates were used. The values of the constants and the fractional average errors between the measured and predicted values of the relative velocity are summarized in Table 111. The use of eq 1with eq 2 or 11with constant K1results in a cubic equation for the holdup. Since the mean dispersed-phase holdup was less than 30% under all operating conditions except at flooding, the constant in eq 11 was evaluated without the 1- Xdterm. Regression results of the experimental data for all three cases with the new constant K2are shown in Table 111. It is seen that the average errors still have the same order of magnitude. It is easier to solve eq 1with eq 11without the holdup term as now only a quadratic holdup term appears. This avoids handling of the complex roots occurring sometimes during the solution of the cubic equation. The results of dimensional analysis may furthermore be compared with those obtained by the method used for the evaluation of the constant in eq 10. For the no masstransfer case, the constant (0.82 X 10-9 in eq 10 and the constant K in Table I11 carry similar values. The constant for the coalescing system is bigger and different than that for the noncoalescing system and for the case without mass transfer. This is to be expected since under the same operating conditions, much bigger drops with higher rise velocities are present in the column for the coalescing system.
40
N
I
/
d r s p . >oont.
I
\
/,,,,,,,,,I
0 100
140
180
Agitator
220 speed,
260
N
300
Crpml
Figure 4. Decrease in capacity with an increase in agitation intensity during solute transfer. Plate free area 40%.
cont. > d i e p . Y
m
100
140
180
220
260
300
Agitotor speed, N Crpml
Figure 5. Decrease in capacity with an increase in agitation intensity during solute transfer. Plate free area 23.5%.
formation of a clear interface in the lower end section. Earlier experiments performed without the packing showed lower maximum throughputs.
Column Capacity Figure 4 shows that the capacity for plates with 40% free area is higher when mass transfer occurs from the dispersed to the continuous pahse than in the reverse direction. This is true also for plates with 23.5% free area (Figure 5) at higher agitator speeds but not at low agitation intensities where mass transfer from the dispersed to the continuous phase enhances coalescence. At low agitator speeds, turbine action is not sufficient to disperse the light phase as a result of which the volume of the coalesced organic phase increases in a particular stage (see also the section on axial variation) and, due to low free area, floods the column a t low throughputs. As shown in Figure 6, throughput of the column could be increased considerably when a suitable packing was placed in the column base. Prediction of Column Capacity at Flooding Flow rates of both phases a t which flooding occurs can be found by equating eq 1to eq 2 and differentiating with respect to the holdup. The differentials dV,/dXd and dVd/dXd are set equal to zero to obtain expressions for limiting flow rates: = 24Vkxdf"(l - xdf) when dV,/dXd = 0 (12) v,., = 4vk(l - 2x,3)(1- x d f ) 2 when dVd/dXd = 0 (13) For a given phase ratio P (=v d / VJ,the value of the holdup vdf
Ind. Eng. Chem. Process Des. Dev., Vol. 25, No. 3, 1986
732
- 4 0 1 i "E \
"E
u
32
1
I
I
,
,
,
,
,
,
I
,
,
,
,
,
t
i
I
T ~ L '
8.38
140
(2/ m z h > -
S y m b o l Vd
I
i-
,-.
2 "
j 24 3 CL
+
2.55 3 . 28
0
3. 79 4. 64
0
0.6-
*
a
II
2
,
,
V o (m3 /m2 h )
m
D
,
1 . 0 , ,
,
-
1
16
i i 4
-
8
4 0
+ 01 :00
1
'
'
140
1
180
'
1
'
1
220
A g i t o t a r speed, N
'
1 300
250
Figure 6. Increase in capacity of the column by introduction of packing in the absence of solute transfer. Plate free area 23.5%.
0. 8
I. 0
Figure 8. Axial variation in holdup with an increase in dispersedphase flow at constant agitation intensity. Plate free area 23.5%.
,
1.2--
~
5~
0. 4 0. 6 H e i g h t (h/H)
0. 2
0. 0
I r ~ m l
,
,
1
I
,
,
'
'
,
,
1
I
I
Symbol r p m
I 220 260
" 0.61 2 L
f
-j0 . 4 L *\
0.
a
0. 2
0. 6 H e 1 g h t (h/H) 0.4
0. 8
I. 0
0.01 0.0
I
I
0 2
1
0.4
O.E
P.8
1
:.O
H e i g h t (T/H)
Figure 7. Axial variation in drop size at different agitation intensities in the absence of mass transfer. Plate free area 23.5%.
Figure 9. Axial variation in holdup with an increase in dispersedphase flow at constant agitation intensity. Plate free area 23.5%.
a t flooding, obtained by eliminating qJvkfrom eq 12 and 13 (Thornton, 1956), is
space around the shaft for the flow of the accumulated phase into the next stage. This, coupled with coalescence of the organic phase of the Teflon bearings, led to higher holdups (leading to humps in the holdup profiles) near the shaft support, Figure 9. Plates without bearings have space around the shaft which allows the dispersed phase to flow through.
Xdf
=
(P2+ 8P)0.5 - 3P 4(1 - P)
(14)
This can be used with vk from eq 10 to predict the dispersed-phase flow rates a t the flooding point from eq 12. This equation predicts the flow rates a t flooding with an accuracy of f 1 5 % .
Axial Variation of Drop Size and Holdup The light phase is broken down by the impellers as it moves along the column length until an equilibrium drop size is attained. Under most of the operating conditions, about a third of the column length was needed to attain the equilibrium drop size. Figure 7 shows the axial variation in the drop size. I t is less a t higher agitation intensities than a t low agitator speeds. Measurements show that holdup also varies along the column length. Centrifugal forces cause the light phase to accumulate along the shaft, and a t higher agitation intensities, holdup profiles with a maximum near the middle of the column were observed and are shown in Figure 8. After a critical value of holdup was reached in any one of the stages, dispersed-phase built up in that stage or in the lower stages until the column finally flooded. At low speeds, however, the agitation was insufficient to thoroughly disperse the light phase accumulated around the shaft. It was more pronounced in stages near the shaft support bearings which occupied about 12% of the column cross-sectional area, Figure 1, and provided no immediate
Conclusions Equation 10 with eq 2 or eq 11 alone allows successful prediction of the relative velocity in a Kuhni column. Knowing the operating variables, one can thus also predict mean values of holdup. Equations 10 and 12 predict the flooding velocities reasonably well for the present column geometry and liquid system. A better design of the shaft supports would eliminate the irregularities obtained in the holdup profiles. Packing in the column base helps to reduce entrainment and increases the capacity of the column. Acknowledgment Financial support was provided by the "Kommission zur Forderung der Wissenschaftlichen Forschung" and equipment was supplied by Kuhni AG (Switzerland). Nomenclature Ar = Apgd3/p,v:, Archimedes number B = total throughput per unit area, V , + Vd, m/s Dk = column diameter, m D, = agitator diameter, m d = drop size, m d3* = Sauter mean drop diameter, m
733
Ind. Eng. Chem. Process Des. Dev. 1886, 25, 733-736
= dispersed-phase viscosity, kg/ (ms) v, = kinematic viscosity, m2/s ai = interfacial tension, kg/s2
Fr = dV/g = D P / g = Froude number g = acceleration due to gravity, m/s2 h = distance from bottom plate, m H = distance from first to last plate, m N = agitator speed, rev/s Re = dVp/p, Reynolds number V, = continuous-phasesuperficialvelocity based on the empty column cross section, m/s v d = dispersed-phasesuperficial velocity based on the empty column cross section, m/s V, = characteristic velocity, m/s V, = relative velocity, m/s We = dVp/a, Weber number Xd = dispersed-phase holdup, fraction Z = coalescence coefficient in eq 3 and 5 Greek Symbols 4 = fractional plate free area Ap = density difference, kg/m3 pc = continuous-phase density, kg/m3 Pd = dispersed-phase density, kg/m3 p, = continuous-phase viscosity, kg/ (ms)
kd
Literature Cited Bibaud, R. E.; Treybai, R. E. AIChf J . 1866, 12, 472. Bell, R. L.;Babb. A. L. Ind. Eng. Chem. Process Des. D e v . 1988, 8 , 393. Deflves, D.; Schnekler, G. Genie Chlm. 1861, 8 5 , 246. Fischer, E. A. Ph.D. Dissertation, Swiss Federal Institute of Technology, Zijrich, 1973, Dlssertatlon No. 5016. Koide, K.; Hirahara, T.; Kubato, H. Kagaku Kogaku 1867, 5 , 38. Klee, A.; Treybal, R. E. Chem. Eng. J . 1856, 2 , 444. Kumar, A.; Vohra, D. K.; Hartland, S. C a n . J . Chem. f n g . 1880, 5 8 , 154. Logsdaii, D. H.;Thornton, J. D.; Pratt, H. R. C. Trans. Inst. Chem. f n g . 1957, 3 5 , 301. Misek, T. Collect. Czech. Chem. Commun. 1864, 2 9 , 1755. Misek, T. Collect. Czech. Chem. Commun. 1963, 2 8 , 1613. Misek, T.; Marek, J. Br. Chem. Eng. 1870, 15, 202. Strand, C . P.; Olney, R. 8.; Ackerman, G. H. AIChE J. 1962, 8 , 252. Thornton, J. D. Chem. Eng. Sci. 1856, 5 , 201. Zenz. F. A. Pet. Refln. 1957, 3 6 , 8.
Receiued for review February 19,1985 Revised manuscript received November 12,1985 Accepted January 9,1986
L-L-E Data for Aromatics Extraction Calculations Using a Modified UNIFAC Model Mamata Mukhopadhyay' and Avlnash S. Pathak Chemical Engineering Department, I.I.T. Bombay, Bombay-400
076, Indla
A modified UNIFAC model has been used for prediction of L-L-E data of multicomponent aromatics extraction systems. The isobaric activity coefficients at infinite dilution have been employed for evaluation of the group-interaction parameters between the CH,-sulfolane pair of groups and their temperature coefficients. These parameters along with the other group-interaction parameters evaluated earlier have been employed in the modified UNIFAC model. The valldity of these parameters has been tested by comparing the predicted results with the corresponding experimental data from the literature. The good agreement justifies the capability of the modified UNIFAC model and improvement of interaction parameters evaluated from the infinite dilution activity coefficients over those evaluated earlier by utilizing the mutual solubility data.
The applicability of the UNIFAC model (Fredenslund et al., 1977) in the prediction of multicomponent L-L-E data needed in the process engineering calculations for extraction of light aromatics from reformed naphtha with sulfolane as the polar solvent were discussed in an earlier publication (Mukhopadhyay and Dongaonkar, 1983). The binary-group-interaction parameters for all possible pairs of groups involved in the calculations except the CH2sulfolane were evaluated from the V-L-E data of the miscible binaries, whereas those for the latter pair were evaluated from the mutual solubility data (Karvo, 1980) of the partially miscible binaries. In the present work, isobaric activity coefficients a t infinite dilution have been employed for evaluation of the CH2-sulfolane pair of group-interaction parameters. The modified UNIFAC model (Kikic et al., 1980, 1982) with the temperature dependency on the interaction parameters as suggested by Larsen et al. (1983) has been considered for the prediction of the multicomponent L-L-E data.
* To whom the correspondence should be addressed. 01 96-430518611 125-0733$01.50/0
Table I. Experimental vs. Temperature Data for the Systems Sulfolane-Cyclohexane, Sulfolane-n -Heptane, and Sulfolane-n -Hexane temp, K 347.6 352.6 354.6
7; temp, K Sulfolane(l)-Cyclohexane(2) 129.90 546.2 98.87 552.7 90.00 555.6
4.768 4.329 3.842
363.3 368.1 370.5
Sulfolane(1)-Heptane( 2) 120.68 546.2 108.30 552.5 102.84 555.5
5.615 5.114 4.592
334.6 339.2 341.4
Sulfolane(1)-Hexane (2) 139.85 546.3 121.43 552.7 110.58 555.6
4.693 4.609 4.390
7;
Group-Interaction Parameters The isobaric infinite dilution activity coefficients have been determined by using the differential ebulliometric technique for three partially miscible binaries, n-hexanesulfolane, n-heptane-sulfolane, and cyclohexane-sulfolane @ 1986 American Chemical Society
