Two Phase Fluid Solid Flow.pdf

TWO-PHASE FLUID-SOLID FLOW’ FREDERICK A. ZENZ’ New York University, University Heights, N . Y .

i)

Results of a series of experiments to determine the flow characteristics of particle-fluid mixtures are presented as a guide to more extensive work on the pressure losses in catalyst carrier lines and flowing fluid beds. Quantitative data are presented for the vertical and horizontal flow of three essentially uniform particles 0.0231,0.0366, and 0.066 inch in diameter, and a material of 0.0066-inch mean diameter exhibiting a fivefold variation in particle size. The experiments were all carried out in a 1.75-inch inside diameter Lucite tube with air as the fluid medium.

I

N R E C E N T years the advances in catalytic cracking and fluid bed processes have created new interest in the subject of the behavior of solid particles in a stream of carrier gas. The general subject would appear at first to be routine to those interested in the design of pneumatic conveying and separation systems. However, investigation of the published literature indicates that pneumatic conveying is an extremely empirical art. Rather high velocities are recommended for conveying specified grains (1, 17), and up until the past year very little basic information (7, 11, a4, 2%) was available concerning the effects of variation in particle size, velocity, loading, etc. The recent, more thorough, studies (9, 14, 83, 26) in this field have been carried out in pipes with rather small diameters at velocities considerably greater than the particle terminal and fluidization velocities thus entirely masking the relation of fluid-particle flow t o the “equilibria” a t fluidization. The experiments described in this paper were undertaken, therefore, with the hope of obtaining fundamental qualitative, and somewhat quantitative, data to establish certain regions of the so-called phase diagram for solid particle-gas systems.

cylinder. I n order to determine whether the flow rates would be affected by a flow of air past the valve tending t o pull material through a t possibly a faster rate, the calibrations were repeated with the hopper in operating position and the air control valve wide open. No effect of air rate was detectable. The main object of these experiments was to obtain a qualitative understanding of the relationships between the various phases of fluidization and the flow chatacteristics of solid-fluid suspensions and, therefore, the adequacy of the calming length prior to the test section was not taken into serious consideration. The length of pipe in which the solids undergo acceleration, thereby considerably increasin the pressure drop, is a t present being thoroughly investigate8 by Russ ( d 3 ) who reports it t o be a function of solids flow rate W , fluid velocity, and particle characteristics. The effect of the accelerating distance is important in design calculations inasmuch as Rusa reports acceleration t o persist in many instances as far as 15 feet from the solids feed point. I n the concurrent vertical flow arrangement, shown in Figure 1, A , the solids feed point was followed by a straight section 20 inches in length, then a long sweep elbow of 12 inches’ radius and another straight vertical section of 20 inches, prior t o the lower pressure tap. In the arrangement given in Figure 1, B, the horizontal distance between the solids feed point and the upstream pressure tap was 4 feet. It was planned to do a systematic study of the uniform spherical, angular, and mixed-size materials ranging in each category from 0.06 inch in diameter down to possibly 0.002 inch or smaller. However, the difficulty in obtaining small uniform spheres and the complexity of the changing flow characteristics of mixed-size material made i t advisable first to conduct exploratory tests with a few more easily obtainable particles for orientation purposes before altering the equipment to make more thorough quantitative investigations. The characteristics of the materials reported in this study are given in Table I. Note t h a t the rape seed, sand, and glass beads may each be considered essentially uniform size material, that they cover nearly a threefold change in density and particle size between the three different particles, and include one angular and two spherical shapes. Salt is the only quantitatively reported mixed-size material and according t o the screen analysis exhibits approximately a fivefold skewed probability distribution in particle size.

EXPERIMENTAL PROCEDURE

m

-

Figure 1 illustrates the arrangements of the apparatus for GRAPHICAL REPRESENTATION OF DATA studying vertical and horizontal flow. The experiments were carried out batchwise using a vacuum system in order t o simplify The experimental results for these four materials in vertical the construction of the apparatus. I n carrying out a run, the and horizontal flow are shown in Figure 2. The nearly vertical solids feed hopper was filled with the material to be investigated, dotted lines, in the four upper charts for vertical flow, represent the blower started, and the air control valve fully opened. The the disperse phase fluidization curves. This manner of plotting solids feed control valve was then opened a specified number of fluidization data enables clearer visualization of the relationships turns and pressure drops recorded a t various air rates as the air between particle-gas flow and the process of fluidization. The controlvalve was set to more closed positions. This resulted in a series of pressure drop and air velocity measurements for TABLE I. CHARACTERISTICS OF MATERIALS USED a given solids feed control valve position, corresponding t o a zu$$l Rape seed Sand Glass beads Salt Peter Henderson Port Washington, Minnesota Mining General Foods Corp., given value of W , established Seed Co. L. I. and Manufacturing Diamond Crystal by calibration of the valve. Co., Scotohlite The solids feed control valve Shape Spherical Sharply angular Spherical Granular was calibrated with a stop Density, p9, lb./ou. it. 68 165 155 131 Diameter, inch watch using various quantities 0,066 0.0366 0.0231 0,0066 ,of each material. It was found Screen analyses=,U. S. Standard Mesh No. that none of the materials in10 0.9 ... ... ... vestigated exhibited any fluid 12 98.6 ... property of increasing flow rate 14 0.1 10:3 ... ... 20 0.4 83.5 0.2 ... with increase in “head” of . . . 30 0.2 98.6 material in the feed hopper. 40 ... ... 1.0 0.2 60 ... ... 0.2 40.2 Each calibration run was re80 ... ... ... 5.8 peated several times with the 100 ... ... ... 27.0 .solids d i s c h a r g i n g t h r o u g h 120 ... ... ... 11.0 170 ... ... ... 13.6 the valve into a large open 200 ... ... ... 1.3 . . I

.

1 Present address, Hydrocarbon Research, Inc., 115 Broadway, New York. N . Y .

Through 200 ... Screen analyses before and after the flow experiments indicated no changes i n particle size. I

a

2801

.

.

.

I

.

.

I

0.9

INDUSTRIAL AND ENGINEERING CHEMISTRY

2802

w

f

Vol. 41, No. 12

unit length. This typical fluidizat,ion curve shown in Figure 3, G , also intersects the pressure drop line for the empty pipe a t the terminal velocity of t'he particles. As noted in Figure 3 , G, the fluidization curve becomes nearly vertical a t high voidages; it is this portion of the curve that is represented by the dotted curves in the upper charts of Figure 2. The lines labeied W = 0 (Figure 2 ) are the observed pressure drops for the flow of air only, with no suspended solids (corresponding to the E = 1.0 line in Figure 3, G).

A . D.

A.1 141 C O N C J R R E N T

VERTICAL

A.I.

FLOW

d

CONCURRENT VERTICAL FLOW

The upper charts of Figure 2 present the observed experimental data LEGEND on the concurrent flow of air-solids A I. Room A i r I n l e t AD Air Discharge suspensions in the vertical tube (Figure E C e n t r i f u g a l Blower c -T S 1, A ) recorded a t constant solids feed A I C Cyclone control valve position, or solids mass CV-I Air C o n t r o l v a l v e CV-2 Solldr Feed C o n t r o l Volve velocity, W . In these experiments i t H solids F e e d H o p p e r (8) H O R I Z O N T A L FLOW was noted t h a t a t velocities below that M A8r R o l a m e l e r R Sollds Receiver a t which the curves show minimum T.S. 44" T e s l S e c t i o n pressure drops, the particle velocities Figure 1. Arrangement of Equipment for Vertical and Horizontal Flow decrease, there is a considerable increascx Experiments in holdup of particles in the tube, and solids flow apparently continues through pressure drop-velocity curves for fluid beds ( 1 9 , ,936) are usually the increasingly dense mass of slowly moving particles float.ing in depicted as shown in Figure 3, F , in which the pressure drop across the tube, The weight of material holdup, slowly ascending in the the entire bed is plotted versus velocity. The E $ fixed-bed line tube, accounts for the increascd pressure drop. As the air velocrepresents the typical curve which may be computed from the i t y was decreased, the movement, of the particles became exwell known Chilton-Colburn ( 6 ) correlation. As the gas velocity is increased to the point a t which the bed is lifted, the particles bremely turbulent and the holdup of material increased unt,il may rearrange themselves in as loose as possible, but yet fixedeither the blower could no longer provide the pressure different,ial bed configuration. The voidage in this loosest possible conrequired to support the material and the bed collapsed causing figuration, designated as emf, has been correlated empirically with all flows to cease, or until slugging occurred to such an extent particle size and shape ( 1 9 , 10)and is illustrated diagrammatically in the comparison of Figures 3, A and B, where e 8 is a lower voidage that manometer readings, fluctuat,ing in some instances by as representative of a more densely packed or settled bed. At, much as' 10 inches of water, indicated extremely unst,eady and velocities (through the loosest possible fixed-bed configuration) undesirable flow conditions. In the vertical flow chart,s of great enough to give a pressure drop equal to the weight of maFigure 2, the onset of slugging is indicat,ed by the dashed lines terial in the bed per unit area of tube cross section, the bed expands so that all the particles are no longer touching and the bed representing the voidage, e,iu0, or rather solids concentration is in the so-called fluid state. Further increase in velocity is within the tube, at which flow becomes erratic and slugging is accompanied by further increase in bed expansion, or voidage, observed. Slugging is also prevalent in fluidization experiments as shown in Figures 3, C and D; the pressure drop, Lip, across (4,19, 13, 66). It has been observed that if the gas velocity the entire bed (recorded by the pressure taps on the left sides of the columns) in both cases remains equal to the weight of bed, through a dense fluid bed is increased, a velocity is finally reached as shown in Figure 3, F . The line in Figure 3, F , labeled E = 1.0 at which slugging sets in; if the velocity is still further increased is the pressure drop curve for the fluid flowing in the empty till the voidage becomes rather high (approximately within the tube as may be calculated from the Fanning equation. The range 0.9 to l.O), smooth disperse phase fluidization is established horizontal fluidization line in Figure 3, F , intersects the E = 1.0 curve a t the velocity, ut, required to balance or support a single again. The minimum values of E , corresponding to maximum particle in the tube. This balancing velocity usually is taken t o solids concentrations, for smooth disperse phase fluidization of the be equal to the terniinal or free-fall velocity as computed from particles investigated in this work are indicat,ed by the interdrag coefficient correlations (18). It is assumed that in calculasection of the dashed slugging loci with the dotted fluidization tions for any particular case, due corrections will be applied to the standard drag coefficient correlations. By due corrections curves in the upper charts of Figure 2 . The values of estuv are is meant taking into account the wall effect, as cited by Ladenindicated on the charts, The fact that a considerably denser burg (16); the relative tube and particle Reynolds number, to concentration (lower E ) of rape seed can be maintained in stable account for the controlling flow pattern (the difference between suspension possibly is associated wit,h the considerably lower actual balancing velocity and terminal or free-fall velocity can be density of these particles. Fluid bed studies have shown t h a t appreciable); and the e8ect of vessel height (the differences in velocities required t o support or balance single particles 2, 5 , the closer the density of the solid material to that of the fluidizing 10, 20, etc., feet up from the bottom of a vertical tube have been medium, the narrower the range of velocities within which slugreported t o vary considerably). ging is observed, The ratio of solids t o fluid density is, how-ever, I n a tube in which a continuous fluid-particle flow is mainonly one of the many factors affecting the slugging tendencies of tained, the pressure drop generally is discussed in terms of unit fluid-solid systems. length of tube. Because a fluid bed const,itutes a form of such continous flowing system, in which, however, there is only a net The vertical flow data of Figure 2 indicate t h a t the fluid flow of fluid, i t seems reasonable to make a plot of fluid bed presvelocity, uCh,at which a solids mass upward flow, W t ,chokes the lure drop on a unit-length-of-bed basis. Such a plot is illustube, may be related t o t,he superficial fluid velocity, U E , ~a~t ~ , trated in Figure 3, G, where the pressure drop, across the taps on which a disperse fluidized suspension begins to slug, by the equathe right sides of the columns depicted in Figures 3, A to E,, is tion: plotted versus velocity. As the bed expands, the voidage increases and there is less weight of charge material within a unit t = P P (1 - W V Q ) ( U c h - %slug) length of the bed, thus giving lower pressure drop across such

-

d

w

INDUSTRIAL AND ENGINEERING CHEMISTRY

December 1949

-

-Rape

Seed-

3----. Glass

Beads

-MIXED-SIZE

UNIFORM- S I Z E PARTICLES

d

2803

t

c--- Sand-

I MATERIAL I

Salt

L

Superficial A i r Velocity, u , ft./sec.

Figure 2.

Experimental Data on Vertical and Horizontal Flow of Dispersed Air-Solids Suspensions i n a 1.75-Inch lnside Diameter Lucite Tube

This indicates the desirability of establishing a fluidization correlation permitting prediction of the slugging range of a given particle-fluid system. This will become more apparent in the discussion of the so-called “phase diagram” for fluid-solids-flow. HORIZONTAL FLOW

A,

*

The charts in the lower half of Figure 2 present the observed experimental data for the horizontal flow (Figure 1,B)of air-solids suspensions. The breaks in the curves are due t o the settling out of particles in the tube. This is a very sharply defined phenomenon exhibiting a very definite relationship between the rate of flow of solids and the gas velocity, U S , at which particle settling begins. At a constant solids flow rate (fixed solids feed valve position), i t was found that when the air rate was decreased t o a certain velocity, us, the particles began t o settle out in the tube and continued t o settle out until the tube was nearly half full, before a steady state was again established and the discharge rate, W , equaled the feed rate. At velocities lower than us more material settled out, filling the tube with a deeper layer of particles and creating a n increase in pressure drop. The particles t h a t settled out in the tube remained stationary; there was no rolling of the material or pushing of the entire layer through the tube, though a few of the particles on the surface of the layer were bounced along in the direction of gas flow. The steady flow of solids, W , occurred in the well dispersed, solid-gas mixture flowing in the space above the settled layer of particles in the same manner as in flow a t velocities greater than uQ. The curves

were duplicated exactly when approached from low velocities, less than us. When the air rate attained the value at us, the tube became completely cleared of all settled material and steady flow of a well dispersed particle-gas mixture continued. Repeated experiments a t the settling velocities showed t h a t there was no consistent tendency for the material t o begin settling out either at several or at any one particular point in the tube. It is felt, however, t h a t this settling velocity may be somewhat dependent on the characteristics of the inner surface of the pipe, a rougher surface tending to induce settling a t possibly lower velocities due to increased turbulence at the pipe wall. The effect is probably small. The superficial horizontal fluid velocity at which material settles out of the flowing suspension has been termed the “saltation” velocity (28). Comparison of horizontal and vertical flow data for the uniform particles shown in Figure 2 resulted in the surprising observation t h a t choking in vertical flow occurred at the same velocity as saltation in horizontal flow. I n other words, at a given solids flow rate u& equals us for uniform particles. This was not observed in the salt experiments. Inasmuch as salt has a fivefold particle size range, this may be explained as a consequence of particle-to-particle contacts because of the differences in velocities between larger and smaller particles. I n horizontal flow it is conceivable t h a t the differences in the particle velocities may cause larger and smaller particles t o collide thus tending to produce a certain degree of aggregation which would tend t o cause saltation a t velocities higher than might otherwise be expected. A plot of W versus uRfor hori-

w,

INDUSTRIAL A N D E N G I N E E R I N G C H E M I S T R Y

2804

r-

r--

Vol. 41, No. 12

were carried out with water as the fluid medium. These water suspensions unfortunately were not investigated in both vertical and horizontal tubes thus not permitting any sort of comparison between us and u,,, values. I t is also interesting to note that the pressure drop curves for salt dropped t o the W = 0 line a t velocities greater than us; there was no detcctable difference in pressure drop for the flow of the mixtures and the flow of air.

i a

4

*

.@

4

SCHEMATIC PHASE DIAGRAMS FOR PARTICLE-GAS FLOW u-

Figure 4 presents a schematic siiininary of particle-gas flow characteristics. The notation used in the diagram for vertical tuhes is taken from the recently proposed fluidization nomenclature ( I O ) . The break in the fluid bed curve between dense and disperst phases ia typical of the experimental investigations of particlegas systems. Leva (21) and Rilhelm and Kwaulr ( 8 6 ) report that this break was less n evident in experiments with water as t’hefluida izing medium. Since it, was noted in the data (3 0 of Figure 2 that the choking velocity for a J given solids flow could be computed from a knowledge of t h c A voidage a t slugging and the corresponding fluidization velocity, n correlation for predicting the “width” and “position” (range of ealu;s) of the slugging break in the LOG u LOG u fluidization curve is essential to the proper design of vertical transport lines as well as Figure 3. Graphical Representation of Fluidization Data fluid-bed reactors. The curves labeled WI to W 3 shown , to the right of the dotted line representing disperse phase fluidization, are representative of the zontal flow of salt and W versus us for horizontal flow of any of experimental vertical flow curves of Figure 2 . The arrows the uniform particles indicated that W does not increase as rapidly attached to the curves indicate the direction of the solids flow with us for the mixed-size material as with the uniform particles. in the vertical tuhe. Countercurrent flow experiments with rape This difference in the relationship between the horizontal salseed in a n arrangement as illustrated in Figure 1, C, establish~d tation velocities for uniform- and mixed-size materials can also be the shape of t h e curves shown t o the left of the disperse phase noted by comparing the d a t a of Blatch ( 5 ) for a uniform size sand fluidization curve in Figure 3. The experimental data are not with those of Gregory ( I S ) and others ( 2 , 3, 12, 27) who worked presented here, becnuse the results obtained near the slugwith clay slurries, etc The experiments of these investigators f2

-Vertical

Tubes-

Horizontal

r

1

1

-Uniform-Size

Porticles-

Tubes

-

1

-Mixed-Sire

Material

P

a

c1 0 -I

LOO

Figure 4.

Sup.rlicbol

Gas

Velocity

LOG

Schematic Phase Diagrams for Particle-Gas Systems

fuperficlol

GOB

V*lnciti

December 1949

Figure 5,

INDUSTRIAL AND ENGINEERING CHEMISTRY

Pressure Drop in Horizontal Flow of Uniform Particle-Gas Mixtures

flow locus were rather inaccurate owing to a tendency for some of the smaller particles to reverse direction and flow concurrently upward with the air stream. It did appear, however, that the choking velocities in this region of countercurrent flow could again be calculated from the disperse phase fluidization curve by the relation:

The very limited countercurrent flow data of other investigators (8, 15) are in agreement with the observations reported in this study. The typical schematic diagrams of Figure 4 for the flow of suspensions in horizontal tubes are based on the data presented in the lower half of Figure 2. Unfortunately, none of the recent publications on particle-gas flow in either horizontal or vertical tubes have indicated close enough approach to the saltation or choking velocities to permit comparison with the observations presented in this study. CORRELATION OF PRESSURE DROP DATA

II

-

Figures 5 and 6 give empirical correlations of the pressure drop data for the uniform-size materials presented in Figure 2. Not much time has been given to a comparison of the data of other investigators with these correlations because of the lack of reported saltation and choking velocities. It is felt that no general correlations can be developed which do not take u s and uch into account. T h e work of Cramp and Priestly (7), for example, presents an excellent study of pressure drop in vertical transport; however, their final equation would indicate t h a t the pressure drop at all solids flows becomes infinite at the solids terminal velocity ut. They failed t o note an effect of W on ucb and assumed that Ueh equals ut for all values of W. The equation of Vogt and White ($6)also neglects t o account for saltation or choking velocities and in addition indicates an effect of pipe diameter by the factor (Dt/Dp)Zwhich has been found in practice to give extremely higher results for small size particles than experiments have indicated. The authors themselves warn against using the equation in cases involving fine particles in large pipe. Though Figures 5 and 6 can hardly be considered final correlations, it is felt that they represent a closer approach to the form which may eventually emerge when more data are available for correlation. It must be remembered that the pressure-drop data reported in Figure 2 are undoubtedly high owing to the rather short accelerating sestion provided in these tests. In addition, the effect of the pressure drop due to the static head of solids within the test section was not subtracted from the over-all pressure drops in the correlation of Figure 6 because no simple means of trapping the particles within the test section were provided for in

2805

Figure 6. Pressure Drop in Concurrent Upward Flow of Uniform Particle-Gas Mixtures

the construction of the apparatus, thereby not facilitating inventory measurements. Figures 5 and 6 are, therefore, very limited in general application and are presented merely as a summary of the pressure-drop measurements. Because of the effects of particle acceleration and size distribution, it is nevertheless felt that ultimately the frictional portion of the over-ail pressure drop in two-phase flow will require correlation in terms of specific friction, as originally suggested by Gasterstiidt (11), with proper account taken of the choking and saltation velocities in some manner similar to Figures 5 and 6. CONCLUSION

These experiments have given some indication of the flow characteristics of solid particle-gas mixtures and should serve as a guide to future investigations of the numerous facts of this general subject. The quantitative results should be applied with caution inasmuch as the correlations are probably not in final form. The main purpose of this investigation was t o establish the schematic phase diagrams shown in Figure 4 and thup illustrate the relationships between particle-gas flow and fluidiaation. An understanding of the typical so-called phase diagram is essential to the proper hydrodynamic evaluation of standpipe, carrier line, and fluid-bed reactor designs. ACKNOWLEDGMENT

This investigation was conducted while the author was in the employ of Hydrocarbon Research, Inc. The encouragement and guidance of Arthur Squires and Manson Benedict of this organization were invaluable in the development of this study. The author is also indebted to John Happel of New York University for his continual encouragement and suggestions. NOMENCLATURE

D, = particle diameter Dt

= tube diameter

e

=

€8 'ml

=

et

=

eszug

=

=

Ap =

AP= PP

=

u

=

u,

=

Ut UE

=a

=

fraction voids in bed of solids or in unit length of flowing system, cu. ft./cu. ft. fraction voids in densely settled fixed bed fraction voids in loosest possible fixed bed (above which the bed is fluid) fraction voids in fluid bed (e, >e,$) fraction voids in a dense or disperse phase suspension at the onset of slugging pressure drop across entire height of bed pressure drop per unit length of tube or bed densit of solid article, lb./cu. ft. superzial f l u i t velocity, ft./sec.-e.g., cubic feet of fluid flowing per second divided by the cross-sectional area of the empty pipe in square feet superficial fluidization velocities for dense or disperse phase fluid bed terminal, balancing, or free-fall particle velocity superficial fluid velocity at which saltation occurs (for a given W , in horizontal Row)

I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY

28836 &h

= superficial fluid velocity a t which choking occurs

71ta1u7

= superficial fluid velocity a t which

IV

= Ib. of

a given W, in vertical flow)

(for

a disperse phase

fluidized suspension begins to slug solids transported/(sec.)(sq. ft. of pipe cross section) LITERATURE CITED

(1) Alden, J. L., Heating and Ventilating, 35, 30-4 (August 1938). (2) Ambrose, H. A , and Loomis, A. G., Phasics, 4 , 265-73 (1933). (3) Babbitt, H. E., and Caldwell, D. H., Univ. Illinois Eng. E s p t . Sta. Bull. Ser., No. 319 (1939). (4) Benenati, R. F., and Cimler, E., private communication, Brook-

lyn Polytechnic Institute, 1949.

(5) Blatch, N. S.,Trans. Am. SOC.Civil Engrs., LVII, 400-8 (1906). (6) Chilton, T. H., and Colburn, A . P., Trans. Am. Inst. Chem. Engl-s., 26, 178 (1931). (7) Cramp, W., and Priestly, h.,Engineer, 137, 34-6, 64-5, 89-90,

112-13 (1924). (8) Docarmo, A. C. M., M.S. thesis, M.I.T., 1943. (9) Farbar, L., IND. ENG.C H E M . , 1184-91 ~~, (1949). (10) Friend, Leo, et al., Chem. Eng. iVewa, 27, 686, 726 (March 7, 1949). (11) Gasterstadt, H., Z . V e r . deut. Ing., 68, KO.24, B17-24 (1924). (12) Gradishar, F. J., Faith, W.L., and Hedrick, J. E., Trans. A m . Inst. Chem. Engrs., 39, 201-22 (1943). (13) Gregory, W. B., Mech. Eng., 49, 609-16 (1927).

Vol. 41, No. 12

(14) Hariu, O.H., and Molstad, M. C., ISD. ENG.CHEY.,41, 1148GO (1949). (15) Hettich, B. V., and Kean, A. H., M.S. thesis, M.I.T., 1943. (16) Ladenburg, R., Ann. P h y s i k , 23, 447-58 (1907). (17) Lambrette, il., Tech. maderne, 25, No. 22 (November 1933). (18) Lapple, C. E., and Shepherd, C. B., IKD.ENG.CHRM., 32, 63% 17 (1940). (19) Leva, M., Grummer, M., Weintraub, M.,and Pollchik, MI., Chem. Eng. Progress, 44, 619-26 (1948). (20) Leva, M., Grummer, M.,Weintraub, M.,and Storch, H. H., Ibid., 44, 707-16 (1945). (21) Leva, M., Weintraub, M., Grurnmer, M., and Pollchik. M . , IXD.ENG.CHEM.,41, 1206-12 (1949). (22) Lewis, W. K., Gilliland, E. R., and Bauer, W. C., Ibid., 1104-17 (1949). (23) Russ, G. H., private communication, Imperial College, P h c e Consort Road, London S.W. 7, England, 1948. (24) Segler, G., 2. Ver. deut. Ing., 79, 558-9 (1935). (25) Vogt, E. G., and White, R.R., IXD.EXG.CHEM.,40, 1731-8 (1948). (26) Wilhelm, R. H., and Kwauk, M., Chem. Eng. Progress, 44, 20118 (1948). (27) Wilheim, R. H., Wroughton, D. M.,and Loeffel, W. F., IND. ENG.CHCM.,31, 622-9 (1939). (28) Wood, S. A, and Bailey, A., Proc. Inst. Mech. Engrs. (London), 142, 149 (1939). RECEIVED February 2, 1945.

Vapor-Liquid Equilibrium Data FOR SYSTEM CARBON TETRACHLORIDE-n-PROPYL ALCOHOL JAMES F. CARLEY AND L. W. BERTELSEN, 111' Cornell University, Zthaca, N . Y . I7apor-liquid equilibrium data are given for the system carbon tetrachloriden-propyl alcohol. The system has an azeotrope boiling at 73.4" C., containing 81.8 mole YOcarbon tetrachloride. These data check fairly closely the values 73.2" C. and 81.3 mole Cj" obtained by Schicktanz, Etienne, and Steele ( 9 ) but disagree with those of Lecat (a), who gives the boiling point as 73.1' C. with a molar percentage of carbon tetrachloride of 75.0. Computed values of activity coefficients agree closely with those predicted by the van Laar solutions of the Gibbs-Duhem equation. Refractive index data for this system are also presented.

T

HE proper design of distillation and other contact equipment requires reliable vapor-liquid equilibrium data. Furthermore, the frequent necessity for estimating equilibrium relations from incomplete data by means of the Gibbs-Duhem equation requires that the data, though few in number, be accurate. One apparatus for obtaining such data which is generally thought to have very few sources of error is that designed by Jones, Schoenborn, and Colburn (6). This paper presents the results obtained on the system carbon tetrachloride-n-propyl alcohol, using that type of equipment. The apparatus used was twice as large as that used by the previous investigators (6). -41~0,it was found that the Nichrome helix inside the flash evaporator was attacked by the mixture of propanol and carbon tetrachloride, with a partial decomposition of the latter and the formation of hydrochloric acid. The effects of this reaction on the experimental results were disastrous. The Xichrome vias therefore replaced viit,h platinum, which proved satisfactory. The total pressure of the vapors was maintained 1

Present address, Harvard Law School, Cambridge, SIass.

a t 760.0 mm. of mercury by a barostat. The boiling points were measured to within 0.5' C. by a pair of copper-constantan thermocouples and a Leeds & Sorthrup portable precision potentiometer. The boiling point of the azeotrope was checked in a standard Cottrell apparatus, with an error of less than 0.2" C. Samples were analyzed by means of an Abbe refractometer and a large scale graph of refractive index against mole fraction carbon tetrachloride (Figure 1). The data for this graph were obtained from mixtures of the pure liquids; the weight of each component in the mixtures was measured with an analytical balance. The carbon tetrachloride was the technical grade supplied by the National Carbide and Carbon Company; from the Paragon Testing Laboratories a pure grade of n-propyl alcohol was obtained. Each was purified by distillation, using an adiabatic glass column packed to a height of 4 feet with iron spirals, a t a reflux ratio of about 30 to 1. The fraction distilling a t constant temperature was collected for use in each case. The refractive index of the purified propyl alcohol at 20" C. was 1.3860, in fair agreement with published values ranging from 1.38409 (a) to 1.38642 (10). The purified carbon tetrachloride had a refractive index of 1.3603; the published value is l.iG0 (4).

TABLE I. REFRACTIVE INDEX-COBIPOSITION DATA Mole Fraction Carbon Tetrachloride 0.0000

Refractive Index, n "no

0,0839

1.3958

0.3974 0.4932 0.6003 0.6929 0.7888 0.5034 i.0000

1.4411 1.4475 1.4542 1.4603

0.2108 0.2960

1.3860

1.4056 1.4126 1,4206 1 4278

1.4350