Skelland (1964)-Resistance to mass transfer inside droplets.pdf

7/22/2019 Skelland (1964)-Resistance to mass transfer inside droplets.pdf

1/6

Resistance to M a s s Transfer Inside DropletsA . H. P. SKELLANO and R. M. WELLEKI l l i n o i s I n s t i t u t e o f Technology , Chicago, I l l i no is

An experimental investigation of the effects of various physical propert ies on the dispersed-phase mass t ransfer coef f ic ient was carr ied out for both nonosci l lat ing and osci l lat ing liquiddroplets fal l i ng i n a s ingle s t ream through stat ionary cont inuous l iquid phases. The Colburn andWelsh two-component technique was used to isolate and measure the dispersed-phase resistanceto mass t ransfer. This technique l imi ted the experimental s tudy to systems wi th low interfac ialtensions, between 2.5 an d 5.8 dynes/cm. Solute was transfe rred into the droplets , and the dropletconcentrat ion was measured af ter droplet -fal l heights ranging f rom about 2 cm. to 103 cm. Pre-caut ions were take n t o minimize end effects.

The experime ntal mass t ransfer rates on nonosci l lat ing droplets i n general were greater thantha t pred ic ted by the Kronig and Br ink model for nonosci l lat ing c i rculat ing droplets. ExperimentalSherwood numbers for four l iquid systems were corre lated in terms o f a relat ionship involv ingthe dispersed-phase Schmidt number, the Weber number, and the t ime group, 4D~tJd2~.whichal lows for the t ime dependency of the t ransfer mechanism. The data were correlated wi th anaverage absolute deviat ion o f 34 . The Kronig and Br ink and Newman re la t ions f i t t ed the ex -perimental da ta for nono sci l lat ing droplets wi th an average absolute deviat ion of 46 and 54 ,respect ively . Th e e xperimental resul ts for th e osc i l lat ing droplets were Correlated by two relat ion-ships wi th an average absolute deviat ion of 10.5 . The Handlos and Baron model f i t te d theexperimental resul ts for osc i l lat ing drop lets wi th an average absolute deviat ion of 38 .

Liquid-liquid extraction is one of the important chemi-cal engineering operations. Many types of equipmenthave been designed for this operation: packed columns,mixer-settlers, spray columns, perforated-plate columnsand others. In spray and perforated-plate columns, masstransfer occurs between a liquid continuous phase and adispersed phase of liquid droplets. There are at least threemajor periods in the droplet life during which mass trans-fer takes place: (1) during formation of the droplets inthe continuous phase, 2 ) during free rise or fall of thedroplets through this continuous phase, and (3) duringcoalescence of the droplets a t the end of the free-riseperiod.Period (2) is of prime importance in spray columns.Periods (1) and ( 3 ) may be of more importance thanperiod ( 2) in a perforated-plate column, particularly ifthere are many stages 3 , 9, 23, 2 8 ) . In this work, onlythe mass transfer occurring during period 2 ) has beenstudied.Considerable work has been reported on the correlationof the continuous phase resistance to mass transfer. Muchof this material is summarized in references 13,25 and 29.Garner and Tayeban (8) separated the effects of oscillat-ing and nonoscillating droplets on the resistance offered bythe continuous phase.These findings, with the present work on dispersed-phase resistance, should contribute toward the ultimateobjective of the design of extraction columns from raterelationships, without the need for experimentally deter-mined efficiencies. The detailed procedure is to be pub-lished later.PREVIOUS W O R K AND T H E O R E T I C A L C O N S I D E R A T I O N S

For drops which are internally stagnant and with no re-sistance in the continuous phase, Newman's ( 2 6 ) theoreticalexpression leads toA. H. P. Skelland is at Notre Dame University, Notre Dame, Indiana.R. M. Wellek is at the Missouri School of Mines and Me tallurgy, Rolla,Missouri.

n = lWhen transfer resistance is zero in the continuous phase andlaminar circulation exists inside the drop, Kronig and Brink's2 1 ) expression leads to

d 3 * 16 D1.tk d = - - I n - Zt 8 [ Bn2exp(-kn-)] (d/2)2 2 )n = lBn and hn are given by Heertjes et al. 1 7 ) for 1and Elzinga and Banchero ( 5 ) or 1 6 3 .Handlos and Baron 1 6 ) proposed a mechanism in which aform of turbulence due to assumed random radial motions issuperimposed on circulation within the drop. Their expressionsare for zero resistance in the continuous phase and lead to

n = lPrevious workers (8, 1 7 , 1 9 ) have compared their mcasure-

ments of dispersed-phase resistance ( a total of eight differentsystems) with one or another of these three theoretical models.For stagnant drops there is good agreement with the Newmanrelationship. Some circulating drops have shown good agree-ment with the Kronig and Brink expression although notableexceptions occurred for the n-butanol-water system. Resultsfor the oscillating droplets deviated significantly from thcHandlos and Baron model.A P P A R A T U S AND E X P E R I M E N T A L P R OC E DU R E

The purpose of the apparatus is to disperse a single streamof liquid drops into a liquid continuous phase. The dispersed-phase fluid was stored in a 50 cc. constant-head buret ( A )(Figure l ) , itted with a Teflon needle valve (B) which eli-minates the problem of contamination with stopcock grease.The needle valve is designed for use with a range of nozzlesizes (C) . Th e larger nozzles were made of thin-walled glasstnbing with the tip ground so that th e plane of the tip was atright angles to the axis of the nozzle. These were attached

Vol. 10 No. 4 A.1.Ch.E. Journal Page 491


2/6

TABLE1.PHYSICALROPERTIES

SystemWater and ethyl acetateEthyl acetoacetate an d waterGlycol diacetate an d waterGlyceryl triacetate a nd water

Compo-sitionPuresat.Puresat.Puresat.Puresat.

Pd , g./CC.

0.99680.99571.01941.02201.09901.09761.15311.1472

pc , g./cc.0.89450.8990.99701.00540.99711.02260.99711.0002

* Phases mutually saturated.f Calculated from Olander s modification of Wilke-Chang correlation.

directly to thc Teflon nccdle valve with glass connectors. Thesni:illcr nozzles were stainless steel hypodermic needles withthe tips ground to a sharp, conical edge.Pyrex glass cxtraction columns ( E ) were used which had aninside diameter of about 7.5 cm. and effective height of 14,28, 54, 78 and 103 cm. At the bottom of each column was aconical section from which the droplets were removed foranalysis. The exit flow rate was controlled by means of a Teflonnccdle valve ( F ) at the bottom of the column.Materials

Thc Colburn a nd Welsh 2 ) two-component technique wasuscd in this study. The two most important requirements forthis procedure are that the saturation concentration in the con-tinuous phase must not be high or there will he a significantresiitance in this phase, and that the saturation concentrationin the dispersed phase must be large enough to obtain a meas-urable rate of mass transfer.The systcms studied, with the dispersed phase mentionedfirst, are water-ethyl acetate, ethyl acetoacetate-water, glycoldiacctate-water and glyceryl triacetate-water.Because surface-active agents can have a profound effect onthe mass transfer mechanism, considerable attention was givento ensuring the purity of all liquids used in the experiments.All organic liquids were purified hy vacuum distillation in anall-glass still with a reflux column. Only the middle cut (about75% of the total ) was used in the mass transfer an d eccentric-ity cxperirncnts. All connections in the distillation apparatuswere made by means of ground glass joints. The laboratorydistilled water was rcdistilled in an all-glass distillation flask.Physical Properties

Densities were measured with pycnometers, viscosities witha u-tube visconieter, and interfacial tensions with a ring ten-

lei T L F L O N NEEDLEDI CI I P I L L A HY

1 0 1

TIP

- I N C HCOLUMN

Fig. 1. Diag ram o f appa ra tus fo r mass t rans-fer experiments.

pd centi-poise0.8851.0601.4951.5802.622.4410.916.6

p c , centi- u,*poise dyne/cm.0.4450.460 5.780.8951.130 3.500.9171.350 2.330.9100.935 3.99

DL,tX d , X C , sq. cm./weight percent T , C. sec. x l o

25.6 0.8925.0 1.2624.2 0.75624.7 0.146

7.44 3.336.06 11.58.73 14.94.86 5.5

siometer. The interfacial tension was measured between mutu-ally satura ted phases, in view of the usual assumption of inter-facial equilibrium in two-phase transfer. Mutual sollubilities ofthe two-component systems were determined by tlhc conven-tional titration procedure.The saturation concentration of the continuous-phase liquidin the dispersed phase was determined with more precisionthan the concentration of the dispersed-phase liquid in thecontinuous phase. This is because the saturation coincentrationin the dispersed phase is used to calculate the dispersed-phasemass transfer coefficient. The continuous-phase saturation con-centration was determined only to ensure that it was not morethan about 15%, by mass, which has been considered themaximum permissible for the Colburn and Welsh technique tohe of value ( 6 ) .Diffusivities were calculated by the Wilke-Chang correlation(30) which was adjusted using the Olander modification (27when water was the solute.Physical properties appear in Table 1.Procedure: Mass Transfer

Droplets were formed at a frequency of 10 drops/lS sec. forall bu t the ethyl acetoacetate-water system, for which th e fre-quency was 5 drops/l5 sec. Between 5 and 15 cc. of dispersedphase were passed through the column in each run. A further7 to 1 0 cc. of dispersed phase were passed through the columnbefore each run during the start-up period and discarded be-fore sampling began. The outlet valve at the foot of the columnwas adjusted to give an interface a t the very apex of the cone.This eliminated inclusion of slugs of continuous phase whenwithdrawing the coalesced droplcts and minimized transferacross the coalesced interface.Droplet-fall velocities were obtained with an electric timeraccurate to 0.005 sec. for a free-fall distance of 45 or 65.5cm.Concentrations were determined by refractive indmex with apreviously determined calibration curve.The average drop volume in a given run was calculated fromthe measured duration of the run, th e total volumLe of dis-persed phase used, and the frequency of drop formation. Thetime from droplet detachment at the nozzle to arrival at thecoalesced layer was measured in order to calculate thLe time ofcontact during the free-fall period, t c .Calculation of dispersed-phase coef fic ient of mass tranrifer ka

A differential material balance on the dispcrscd-phase resultsin the following expression which defines the inside transfercoefficient:ddt- V C = k d Cp- C ) A (4)

where C is the average solute concentration in the droplet.When the two-component technique of Colburn and Welsh isused and equilibrium between the phases at the interface is as-sumed, the concentrations of each phase immediately adjacentto the interface are equal to the saturation concentration ofeach phase:

Page 492 A.1.Ch.E. Journal July, 1964


3/6

column, that is, t f o.With these boundary conditions onthe dro plet concentration and time of contact, Equ ation 4)was integrated to give the following expression for kd:

Fig. 2. Newman model, nonoxillating droplets.

Ci= C , saturationThu s, the interfacial concentrations are constant for a con-stant temperature and pressure) and are easily determinedfrom measurements of the saturation solubility. There is onedisadvantage in using the Colb um a nd Welsh techniq ue whichis not always mentioned in the literature. The partially misciblebinary systems in general have low interfacial tensions. Thus,if it is desired to test or obtain a mass transfer correlation forhigh interfacial tension systems, another technique must beused. It has also been stated that two-component systems donot exhibit th e complicating factors for example, interfacialturbulence) that have been observed for three component sys-tems 4 ) .Since the object of this work was to study the mass transfermechanism during the free-fall period, it was necessary toeliminate or minimize the effects of mass transfer during drop-let formation and coalescence. As an example of these effects,after a 2.0 cm. fall period, the total fractional extraction inethyl acetoacetate droplets was 0.12 a nd 0.36 for equivalentdroplet diameters of 0.95 cm. and 0.30 cm., respectively. Fora column heigh t of 78.0 cm., the total fractional extraction forthe above two droplet sizes increased to 0.43 a nd 0.64, respec-tively.At present no reliable method has been found for entirelyeliminating end effects. Thus the original graphical extrapolationprocedure to zero column height used by Sherwood, Evans,and Longcor (28) and by Licht and Conway 2 3 ) ,using aplot of th e log fraction solute unextracted vs. drop-fall timeor column height, has been shown to be invalid by Licht andPansing 2 4 ) .

The more sophisticated graphical extrapolation techniques ofJohnson and Hamielec, et al. (19, 20) can only be used inthose cases in which the data yield straight lines when plottedin their special manner. This condition was not fulfilled suffi-ciently to use the method in our case.The techn ique used in this work to correct for end effects isdescribed as follows. In order to reduce the transfer to thecoalesced drops, the interfacial area of the coalesced layerwas kept very small abo ut M sq. cm. ). The coalescence endeffect is reduced to a minimum; therefore the solute concen-tration in the exit stream closely approximates the average con-centration of the droplet just before it lands at the coalescedlayer. Hence, the droplet concentration after withdrawal fromthe shortest column approximately equals that after the samelength of fall in any longer column. The average droplet con-centration during the free fall in any longer column was thenconsidered to vary from the exit concentration of the shortestcolumn, C,, to the exit concentration from any longer column,Cf. he corresponding time of contact was the difference be-tween the total fall times in the shortest column and any longer

t c = tf oCf- oE , =z i- c,

and

then6)VA t ck = In ( 1 m )

The method used in this work is identical with that used byGarner, Foord and Tayeban 6),Garner and Tayeban 8 ) ,and Garner an d Skelland (12). This method, described above,definitely eliminates transfer occurring during formation anddetachment of the drop, leaving only the question of transferduring withdrawal of the drops at the end of the column asignificant coalescence interface is not present in this method).Transfer during withdrawal was assumed to be negligible insimilar work by Calderbank and Korchinski 1 ) on dropletheat transfer, and the results of this assumption agreed wellwith theoretical predictions. Our technique regarding end ef-fects is, therefore, at least in line with that used by severalother workers.The interfacial area A in Equation ( 6 ) was calculated asthe surface of a s pher e an d as the surface of an obla te spheroidwith the same volume as the droplet in both cases. The areaof the spheroid was readily calculated from that of the sphereusing the relationship presented by Garner and Tayeban (8 ) .Droplet Eccentr ic i ty

Eccentricity is defined in this work as dh/d,. Droplets werephotographed during free fall through a 7.6 cm. I.D. glasscolumn w ith a ce ntral viewing section mad e of optically planeglass to eliminate distortion effects. Eccentricities were meas-ured from enlarged projections of these photographs. Measure-ments were made in two separate sets, one in which the phaseswere mutually saturated and the other in which the dispersedphase was initially pure before introduction to the column.RESULTS: M A SS TRANSFER

The exper imenta l r e su l t s were divided i n to tw o c a t e -gories, nonoscil la t ing droplets and oscil la t ing droplets.

NSK observedFig. 3. Kronig and Brink model, nonoscillating droplets.

Vol. 10 No. 4 A.1.Ch.E. Journal Page 493


4/6

The original data are deposited with the A.D.1.OSherwood numbers were calculated from the experi-mental kd values by multiplying by de/DL. The values forthe nonoscillating droplets are compared with the theoreti-cal Sherwood numbers obtained from Equations (1) and2 ) for stagnant droplets and for circulating droplets asshown in Figures 2 and 3, respectively.A similar comparison for oscillating droplets appears inFigure 4, where the Sherwood number is obtained fromEquation (3) for droplets which are both circulating andoscillating.Nonoscillating Droplets

The discrepancies between experimental and theoreti-cal Nsh values in Figure 2 indicate that circulation cur-rents were present inside most of the droplets used in thisstudy. Figure 3 further shows that the rate of transfer veryoften exceeds that predicted for circulating droplets by theKronig and Brink relation, even though no visible oscilla-tion of these droplets could be detected. It may be noted,however, that the Kronig and Brink relationship was de-rived for conditions in which the circulation patterns with-in the drop are those given by Hadamard 1 4 ) . Thesepatterns in turn are restricted to the Stokesian regime offlow (that is, N R ~ 1) .Nevertheless, Hamielec and John-son 1 5 ) indicate theoretically that the Hadamard circula-tion patterns may be approximated under certain condi-tions up to Reynolds numbers of about 80. The dropletReynolds number range in the present work was 37 to 546.An empirical correlation of the results for nonoscillatingdroplets was, thercfore, attcmpted. Both the Newman andthe Kronig and Brink models [Equations (1 ) and (2 )1consider the internal mass transfer coefficient to be a func-tion of only 3 variables:kd f t c , D L , d e ) (7)

It appeared from graphical representations of the dataand past observations on droplet phenomena (7, 10, 11)that the function probably should include other variables.The following function exprcsses those which are possiblyrelevant:______Tabular material has been deposited as document 7923 with the.4nirrican Documentation Institute, Photoduplication Service, Library ofCongress, Washington 25 D. C. nd may be obtained for $3.75 forphntoprints or $2.00 for 35-mni. microfilm.

NSh, o b s e r v e dFig. 4. Handlos and Baron model, oscillating droplets.

NSh, o b s e r v e dAFig. 5. Equation 9 ) , nonoscillating droplets, Nsh = 31.4 N T ~ ~ - ~ .NSc-0.125We0.371.

k d = N t c , DL ,de U , fit, pd , pc , pa, U ( 8 )The data were correlated by twelve dimensionless vari-ants of Equation (8) , using the least squares techniquewhich has been programmed for use on the IBM-1620digital computer 2 2 ) . Three further correlation formswere tested, in which k d was based on the true area ofthe ellipsoidal droplets. In one of these forms, de was thediameter of the sphere of the same volume. In the othertwo forms, d, was the average of the major and minoraxes of the ellipsoid. The best correlation from the fifteen

examined was as follows:ANS, , = 31.4 N - 0 . 3 3 8 N-0.125 N0 371T , , S c We 9)

N T , is a dimensionless time group which appears inthe theoretical expressions of Newman and Kronig andBrink. It contains the time of contact, tc and therebymakes some allowance for the unsteady sta te nature ofthe mechanism.The average absolute deviation of the data from thisequation is 34%, the estimate of variance, S2 (log N s ~ ) ,is 0.035, and the multiple regression coefficient, A, is 0.83.The confidence limits of the exponents at the 95% prob-ability level are:

A

.338 S 0.075; 0.125 0.024; 0.371 S 0.12Experimental inaccuracies suggest rounding off of theseexponents to two decimal places.Equation (9 ) is for systems with low interfacial ten-sions (2.34-4.8 dynes/cm.) and low continuous-phaseviscosities , .L~< 1.35 centipoise). It covers the followingrange of variables:

14.2 6 N s ~ f 9340.03 x 10-3 6 N T ? ~ 32 x 10F856 6 Nsc 79,8000.48 Nwe 7.6

Values calculated from Equation (9) are compared withexperimental values in Figure 5. The estimate of the vari-ance and the average absolute deviation of this correlationPage 494 A.1.Ch.E. Journal July 1964


5/6

are compared with those of the theoretical relationshipsof Newman (stagnant drops) and Kronig and Brink (cir-culating drops) as follows:' AverageAbsolute Adeviation S 2 (log Nsh)

Equatiou (9) 34 0.035Newman 54 4.94Kronig andBrink 46 0.092

O s c i l l a t i ng DropletsThe transition from nonoscillating droplets occurred at

1 Reynolds number of about 600 for the water-ethyl ace-tate system and 360 for the ethyl acetoacetate-water sys-tem.The correlating procedure for oscillating droplets wassimilar to that described for nonoscillating droplets. Fromseventeen correlating forms examined, t w o alternativeequations were ultimately found to fit the data aboutequally well and are as follows:

N p is the P group which was used with success by Huand Kintner 1 8 ) in correlating droplet-fall velocity witha variety of variables.The average absolute deviation for Equations (10) and(11) is 10.5%; S2(logNsh) for Equations (10 ) and (11)is equal to 0.0042. The multiple regression coefficient Rfor both equations is 0.97.The confidence limits of the exponents at the 95%probability level are:

A

-0.141 t 0.060 -0.141 0.0590.684 0.220 0.760 .2460.100 ? 0.078 0.285 r 0.042The data on which Equations ( 10 ) and (11 ) are basedare for systems of low interfacial tensions (3.5-5.8dynes/

AFig 6. Equation ( l l ) , oscillating droplets, NSh = 0.142 N T ~ - ~ . ~ ~ ~We0.769 N 0.285.

cm.) and low continuous-phase viscosities p c < 1.35centipoise). The range of variables is:326 6 N s ~ 6 3,650411 N R ~ 6 3114

1.78 X l o s A N p 6 37.2 X lo86.38 6 Nwe 6 13.3Values calculated from Equation (11) are comparedwith the experimental values in Figure 6. Equation (10)

yields a closely similar plot.The estimate of the variance and the average absolutedeviation of these correlations are compared with thosefor the theoretical relationships of Handlos and Baron(circulating and oscillating drops) as follows:

0.0097 x 6 NT~% 2.19 x 10-3

AverageAbsolute Adeviation s2 log NSh)

Equation (10) 10.5 0.0042Equation (11 10.5 0.0042Ilandlos andBaron 38.3 0.040ACKNOWLEDGMENT

The financial support of the Sinclair Oil and Refining Com-pany in the form of a Fellowship and of the National ScienceFoundation in the form of a summer Fellowship, both to R. M.Wellek, is gratefully acknowledged.F. R. Anderson and R. F. Zabransky gave assistance inphotographing the droplet profiles.NOTATIONA = interfacial areas of droplet, sq. cm.Bn = coefficients in series expansionC = average solute concentration in the droplet, g.Ci = solute concentration on the dispersed-phase sideCf, o = average solute concentration in the droplet atdd,dh6D L sq. cm./sec.E m = fractional extractionf , n = functionsgkd, kd = dispersed-phase mass transfer coefficient, ob-served and calculated, respectively, g. mole/(sec.) sq . cm.) g . mole/cc.)

moIe/cc.of the interface, g. mole/cc.tf and to, respectively, g. mole/cc.

= diameter of spherical droplet, cm.= diameter of sphere having the same volume as= length of horizontal axis of droplet, em.= length of vertical axis of droplet, cm.= molecular diEusivity of solute in dispersed phase,

the droplet, cm.

= acceleration of gravity, cm./sq. sec.A

NRe = Reynolds number, deUpc/pcN s c = dispersed-phase Schmidt number, pd/pdDLN s h , Nsh = Sherwood number observed and calculated,A

Akdde kdderespectively, -, I , D L

Nwe = Weber number, deU2pc/aN ~ R viscosity ratio, p d p cN T ~ dimensionless time group, 4DLtc/deNpe, = modified Peclet group, NReNSh/(l + R)Vol. 10 NO.4 A.1.Ch.E. Journal Page 495


6/6

A AS2(logNsh) = estimate of variance of log Nsntt cto

= time of contact, sec.= time of contact during free fall= time of fall for the shortest distance (from de-

tachment at the nozzle to arrival at the coalescedlayer), sec.tT = temperature, C.UVX d persed phaseX c uous phaseAn = eigenvaluesPC, pd = viscosity of continuous and dispersed phase, re-pc, pd = density of continuous and dispersed phase, g./cc.= interfacial tension, dynes/cm.LITERATURE CITED

= time of fall for any other fall distance, sec.= gross terminal velocity of droplet, cm./sec.= volume of droplet, cc.= weight percentage of continuous phase in dis-= weight percentage of dispersed phase in contin-

spectively, poise

1. Calderbank, P. H., and I. J . 0. Korchinski, C h e m . E n g .2. Colburn, A. P., and D. B. Welsh, Trans . Am. Ins t . Chem.3. Coulson. J. M., and S. J. Skinner, Chem. Eng. Sci., 14. Davies, J. T., Tra m. Ins t . Che m. Engrs. , 38, 289 (1960).5. Elzinga, E. R., and J. T. Banchero, C h e m . E n g . Progr.6. Garner, F. H., A. Foord, and M. Tayeban, J Appl . Chem. ,

Sci. 6 , 65 (1956).Engrs., 38, 179 (1942).197 (1952).

Symposium Ser. N o . 29, 55, 149 (1959).9, 315 (1959).7 . Garner, F. H., and P. J. Haycock, Proc. Royal SOC . A252,457 1959).8. , and M. Tayeban, Anales de la Real SociedadEspanola, de Fisica y Quimica, Serie B-Quimica, TomoLVI B ) , 479 (1960).

9. - and A. H. P. Skelland, Ind. Eng. Chem., 461255 (i954). -10.11. ,C h e m . E n g . Sci., 4, 149 (1955).. Trons. Inst. Chem. Engrs ., Lo nd on ) , 29315 (1951).12. , Ind . Eng. Chem. , 48, 51 (1956).13. Griffith, R. M., C h e m . E n g . Sci., 12, 198 (1960).14. Hadamard, J., Compt. Rend. , 152, 1735 (1911).15. Hamielec. A . E.. and A. I. Tohnson, Can. J . Chem. Eng. ,4, 4 1 (1962). -16. Handlos, A . E., and Thomas Baron, A.1.Ch.E. Journal, 3,127 (1957).Sci., 3, 122 ( 1954).( 1955).

17. Heertjes, P. M., W. A. Holve, and H. Talsma, C h e m . Eng.18. Hu, Shengen, and R. C. Kintner, A.I.Ch.E. Journal, 1, 42,19. Johnson, A. I., and A . E. Hamielec, ibid. 6, 145 (1960).20. , D. Wood, and A . Golding, Can. J Chem.Eng., 221 (1958).

(19%).21. Kronig, R., and J. C. Brink, Appl . Sci. Res., A-2, 14222. Leeson, D., IBM-1620 Program Library, No. 6.0. 002,23. Licht, William, and J. B. Conway, Ind. Eng. Chem., 42,24. and W. F. Pansing, ibid., 45, 1855 (1953).25. Linton, M., and K. L. Sutherland, C h e m . E n g . Sci., 12,26. Newman, A. B., Trans. Am. Inst. Chem. Engrs., 27, 20327. Olander, D. R., A.IC1a.E. Journal, 7, 175 1961).28. Sherwood, T. K., J. E. Evans, and J. V. A . Longcor, Ind.29. Thorsen, G., and S. G. Terjesen, Chem. Eng. Sci., 17,30. Wilke, C. R., and Pen Chang, A.I.Ch.E. Journal, 1, 264

available prior to July 1962).1151 (1950 ).

214 (1960).(1931).

Eng. Chem., 31, 1144 (1939).137 (1962).( 1955).

Manuscript received February 11 , 1963. revirion received November18, 1963; paper accepted November ld, 1963. Paper presented atA.1.Ch.E. N e w Orleans meeting.

Selectivity in Hydrocarbon OxidationMAURICE SPIELMAN

Esso Research and Engineering Company Madison New JerseyIn many chemical reactions of commercial interest th e

desired product is an intermedia te in a sequence of reac-tions. For example in liquid phase hydrocarbon oxidationsit is often found that overoxidation results in a host ofuseless and undesirable tars and condensation products.These degrade feedstock to waste value and often causefouling and plugging in process vessels. On the other handlow conversion, underoxidation, reduces the amount ofundesirable materials made but increases the complexityand cost of operation by requiring large equipment tohandle high recycle rates.Operating conditions selected are often those near thehighest conversion possible without getting into processdifficulties. In many cases optimization with respect tobalancing the cost of recycle against the value of highselectivity to product is not adequately explored. Bymeans of kinetic studies optimum Conversion levels canbe shown to be well below those usually considered nor-mal.

This article presents an analytical study of two reac-tion sequences postulated for liquid phase hydrocarbonoxidation. Major differences between reactor types arefound. The relationship between conversion of feed andselectivity to intermediate product is strongly influencedby the type of reactor used. At any conversion level selec-tivity and yield of intermediate are higher in a batch orplug flow reactor than in a continuous flow stirred reactor.The undesirable result of conversion beyond t he pointof maximum yield is also shown. Choice of an optimumconversion level is thus simplified.Application of the calculated results to experimentaldata on ordinary and boric acid modified air oxidation ofcyclohexane are given.The results of this study are not limited to oxidation.They can provide a helpful tool for planning and inter-preting experimental studies, for visualizing fundamentalprocess limitations at an early stage of development, andfor making maximum use of existing data.

Page 496 A.1.Ch.E. Journal July, 1964

Documents

Skelland (1964)-Resistance to mass transfer inside droplets.pdf