Volatiles loss in the nozzle zone during spray drying of emulsions

Volatiles loss in the nozzle zone during spray drying of emulsions. J. Aleksonis Zakarian, and C. Judson King. Ind. Eng. Chem. Process Des. Dev. , 198...
4 downloads 0 Views 887KB Size
Ind. Eng. Chem. Process

Des. Dev. 1982, 27,

Satterfield, C. N.; Reid, R. C. J . Chem. Eng. Data 1981, 6 , 302. Semenov, N. N. "Some Problems in Chemical Kinetics and Reactivity"; Princeton University R e s : Princeton. N.J., 1958. 1959: Vol. I , .DD . 1-114. voi. 2, pp 43, 217. Snow, R. H. J . phvs. Chem. 1988, 70, 2780. Tayh, J. E.; Kulich, D. M. Int. J . Chem. Klnet. 1973, 5 , 455. Trotman-Dlckenson, A. F.; Milne, 0. S. "Tables of Bimolecular Gas

107-113

107

Reactions"; NSDS' NBS 9, U S . Dept. of Commerce, 1967 Vardanyan, J. A. Combust. Flame 1974, 22, 153.

Received for review October 10, 1980 Revised manuscript received June 5, 1981 Accepted July 31, 1981

Volatiles Loss in the Nozzle Zone during Spray Drying of Emulsions J. Aleksonls Zakarlan' and C. Judson Klng' Depertment of Chemical Englnmrlng, University of California. Berkeley. California 94720

The effect of a dispersed oil phase on the retention of volatile acetates was investigated for spray drying of aqueous

sucrose solutions. Particular attention was given to the region close to the atomizer. Suspended oil droplets were found to affect volatiles retention In two ways-through extraction and sequestering the volatiles against loss, and indirectly through an increase In both the average drop size and the spread of drop sizes. The effects of drop-size distribution c a n be largely accounted for by correlating volatiles retention with the percent water evaporation or by using a computerized model based upon the transport Characteristics of drops of different sizes. The extractive effect c a n be accounted for by a simple model based upon a rate limb due to diffusion in the aqueous phase. This model c a n also incorporate the effect of the oil phase in suppressing aroma response in a reconstituted food product.

Introduction Spray drying is used on a large scale for dehydration of a number of liquid- and slurry-form foods, such as coffee, milk, and purees. The principal barrier to more widespread use of spray drying for foods is loss of product quality. One principal quality attribute is retention of volatile flavor and aroma substances. These volatile substances nearly all have quite large activity coefficients in aqueous solution (Bomben et al., 1973), with the result that they partition to a very high extent into an oil or fat phase, if one is present. Equilibrium distribution coefficients of lo2to lo4are common (see, e.g., Leo et al., 1971). Foods often contain an emulsified organic phase, as in milk and citrus juices. In other cases, such as coffee extract, it is possible to control the extent to which a residual or added oil phase is present. The goal of the present work was to determine experimentally the influence of an emulsified oil phase upon spray characteristics and retention of volatile acetates during spray drying of aqueous sucrose solutions. Previous measurements and interpretations of volatiles retention during spray drying have been reviewed by Bomben et al. (1973), Bruin and Luyben (1980), and King et al. (1981), among others. For the most part, measurements have been restricted to overall volatiles loss as evidenced by reconstitution of the product collected from a spray dryer. Thijssen and co-worker8 (Thijssen and Rulkens, 1968; Rulkens and Thijssen, 1972; Kerkhof and Thijssen, 1977) have demonstrated the utility of the concept of "selective diffusion", whereby development of a surface layer of high dissolved-solids content sharply retards volatiles loss. In a prior study to the present one, Kieckbusch and King (1977,1980) developed a method of sampling liquid locally near the atomizer in a spray chamber and demonstrated quite large losses of volatiles Chevron Research Corp., Richmond, CA 94802. 0196-4305/82/1121-0107$01.25/0

within a few centimeters of a spray nozzle for sucrose solutions up to 40% w/w and maltodextrin solutions up to 20% w/w. Various patents (e.g., Clinton et al., 1966; Huste et al., 1968) suggest use of an emulsified oil phase to improve retention of volatile flavor and aroma substances during spray drying. King and Massaldi (1974) used a simple transport analysis to estimate the influence of low levels of an extractive, emulsified phase on volatiles retention during drying. If the extractive phase remains within the spray drops, rather than at the surface or in separate drops, a considerable increase in the overall retention of the extracted components should result. However, a compensating effect comes from the effect of extraction into the emulsified phase suppressing the partial pressure-and hence the aroma response-of the volatiles over the reconstituted product. The analysis showed that the degree of dilution upon reconstitution must be substantially greater than that of the feed to a spray dryer in order for the presence of an emulsified extractive phase to give an appreciable increase in product aroma response. This is the case for coffee, which is much more dilute in the cup than in the extract fed to a spray dryer. Ban (1978, 1979) measured losses of methyl benzoate, L-carvone, benzyl alcohol, and benzaldehyde at high concentrations (3 to 12% w/w) during drying of suspended macrodrops of gum arabic solutions. Losses of the more soluble compounds were greater than those for less soluble volatiles. This result is in accord with the concept of the rate-limiting step being transport of dissolved volatile substance through the aqueous continuum, as put forward by King and Massaldi (1974). Kerkhof (1977) measured retentions of l-hexanol dispersed in maltodextrin solution during spray drying, and found that a surfactant, Tween, was effective in promoting volatiles retention, presumably because it hindered aggregation and /or coalescence of drops. With Tween, the percentage hexanol retention increased as the ratio of suspended to dissolved hexanol 0 1981 American Chemical Society

108

Ind. Eng. Chem. Process Des. Dev., Vol. 21, No. 1, 1982

increased, again in accord with the concept of transport in the aqueous continuum being the rate-limiting step. Blakebrough and Morgan (1973) measured retentions of ethyl caproate and ethyl propionate during spray drying of milk, finding that the retention was better in skim milk than in whole milk. They postulated that the presence of fat in whole milk altered the permeability of the surface layer of a drop, allowing greater volatiles loss. A similar interpretation was made by Reineccius et al. (1978) for measurements of the loss of methyl ketones during spray drying of skim milk, with varying levels of butter added. Increasing butter level up to 21% w/w increased retentions of the volatile ketones, but beyond that point a greater butter content decreased retentions. This was attributed to continuous fat-phase paths for volatiles transport within a drop. In the present work, retentions were measured locally in a spray chamber for volatile n-acetates (ethyl through pentyl) present at very dilute levels in aqueous sucrose solutions, with and without a small amount of dispersed peanut oil phase. This system was chosen since vaporliquid-liquid phase equlibrium data were available (Kieckbusch and King, 1979a) and since it was the system (without oil) used in the previous work of Kieckbusch and King (1980). Liquid samples were collected at various axial and radial distances from the atomizer, with particular attention to the region immediately adjacent to the atomizer. Simultanwus measurementa of drop-size distribution were made, so that the effects of oil on the spray characteristics could be separated from the extractive effects. The results were interpreted in terms of thermodynamic, mass-transfer, and drop-dynamics models in an effort to identify the underlying mechanisms controlling retention. Experimental Apparatus and Procedure Spray Apparatus. The spray chamber and air and liquid circulation systems were the same as those used and described by Kieckbusch and King (1980). Standard operating conditions for experimental runs were as follows: liquid pressure at nozzle = 0.79 MPa (100 psig); liquid temperature at nozzle = 42 to 44 "C; air flow rate = 100 kg/h; air temperature outside spray = 150 to 160 OC; concentration of acetates in feed = 100 ppm w/w (each); and concentration of sucrose in feed = 40% w/w. The liquid-feed temperature corresponds to the wet-bulb temperature for the system. The atomizer used was a Spraying Systems Co. UNIJET T400017 pressure nozzle, which produces a flat, fan-shaped spray. The water capacity at 0.79 MPa is 6.1 L/h. The measured orifice area was 6.0 X lo4 cm2. This nozzle was used instead of the cone-spray nozzles primarily used by Kieckbusch and King (1980) so that the drop-size distribution would be relatively independent of sampling position. In cone-spray pressure nozzles, entrainment of air due to droplet deceleration causes a fractionation of drops by size (Rothe and Block, 1977), with the result that the drop-sue distribution changes with position. It was found that it is important to align the fan-spray nozzle precisely before a run, so that the sample flow collection rate at a given sampler location would be reproducible. Sampler for Volatiles Retention. The collection device for liquid samples used for measurements of percent evaporation and volatiles retention was the methanolchilled, continuous-flow sampler described and used by Kieckbusch and King (1977, 1980). Determination of Drop-Size Distribution. Spray drops were collected in silicone oil for microscopic examination using the device shown in Figure 1. The collector was designed to have nearly the same overall dimensions

-

drop collecting cell

aperture

inserted here

glass

i t rl.Ilq6cm Y IY

4

1.5 cm

c

T'

Figure 1. Collector for measurement of drop-size distribution.

and geometrical shape as the sample collector used for measurements of volatiles retention, so that the two devices would have essentially the same collection efficiency as a function of drop size. This would allow the measured drop sizes to be used in a meaningful way for interpretation of the volatiles-retention data. A stainless steel tube, 1.3 cm 0.d. and 0.8 cm deep, with a piece of glass centered at the bottom, served as the drop-collection cell. The cell was filled with Dow Corning 510 silicone fluid. A flexible tape, 1.6 cm wide and made from a Teflon derivative, served as a shutter which allowed exposure of the silicone oil to the spray for a brief period as the shutter was moved. Drops retained their spherical shape while sinking slowly through the fluid. The cell was removed and photographed several times under different microscope conditions. Drop sizes were then measured and counted from the clearest photograph. Preparation of Emulsions. Sucrose and peanut oil were purchased from a local supermarket. Emulsions of peanut oil in 40% w/w sucrose solution were made by first mixing the desired amount of oil with a small amount of sucrose solution in a laboratory blender. This mixture was then added to a holding tank containing the rest of the sucrose solution. The suspension was circulated for 15 min through a Model ND-1 Charlotte colloid mill (Chemicolloid Laboratories, Inc., New York, NY), with the clearance between rotor and stator of the mill set at 25 pm. The resulting emulsion was placed in a 19-L Pyrex feed tank, the acetates were added by pouring standard solutions from a graduated cylinder, and the entire mixture was agitated for another 15 min. Some settling of oil to the surface occurred during handling and during a run. In some cases an emulsifying agent, Tween 81, was added at either 1or 10% v/v, based on the oil, to see if this would stabilize the emulsion. Microscopic observation revealed that, with 1?% Tween 81 in emulsions of up to 1% w/w oil, the oil droplets were predominantly 1to 10 pm in diameter, with a few ranging up to 100 pm. With 10% Tween 81, all droplets were less than 20 pm, with most about 5 pm. However, presence or absence of Tween 81 did not seem to affect spray characteristics and measured volatiles retentions significantly. Analytical Techniques. Oil concentration levels within the feed were measured using a spectrophotometric technique with Sudan Blue dye, developed and described by Zakarian (1979). When there was a tendency for the oil concentration to change during a run, experimental results for volatiles retention were related to the measured

Ind. Eng. Chem. Process Des. Dev., Vol. 21, No. 1, 1982

oil level at the time of volatiles sampling. Two liquid samples were taken with the methanolchilled collector at each sampling position-one for measurement of water evaporation and one for measurement of volatiles retention. The amount of water evaporation was measured using a standard Abbe refractometer, with known calibrations for sucrose solutions. Results with the refractometer agreed within f0.05% sugar with the results of gravimetric measurements of water content, made using overnight drying in a vacuum oven. From a knowledge of the percent water evaporation, the appropriate amount of water was added to return the second sample to the feed dilution. Concentrations of the volatile acetates were then measured using flame-ionization, vapor-headspace gas chromatography with the water-jacketed injector and other techniques described by Kieckbusch and King (1979b), except that their needleless injector was not used. The sample bottles used for volatiles analysis were thermostatted at 25 "C in a bath and were stirred magnetically. A Perkin-Elmer Model 3920 gas chromatograph, equipped with a Perkin-Elmer Model M-2 digital integrator, was used with a 0.5-m Porapak Q column, maintained at 180 "C. Each sample was analyzed at least twice, with replicate analyses often agreeing within 1% and never differing by more than 6%. The ratio of peak areas for headspace analysis of collected spray and feed samples, adjusted to the same dilution, was taken as the fraction retention.

Results and Discussion Fan-Spray Characteristics. Photographs were taken of the nozzle and film-disintegration zone for sprays of 40% w/w sucrose solution, with and without added peanut oil. Data are reported elsewhere (Zakarian, 1979). Measurements made to the point of first visible disintegration on these photographs showed that the film length dropped sharply from about 1.4 cm to 0.8 cm, in a roughly linear fashion, as the oil level was increased from zero to 0.1% w/w. Beyond this level, up to 0.3% oil, the film length decreased much more slowly, down to 0.7 cm. The spray angle, determined by extrapolation of tangents to the fan-spray boundaries back into the nozzle, increased from 44" with no oil to 57" with 0.3% oil. The average distance from the point of first visible film disintegration to the point of apparent complete disintegration was 0.5 cm. The uncertainties of measurement were about f0.15 cm for the film lengths and f1" for the spray angles. Since the film thins as it proceeds outward from the nozzle, and since the drop size is directly related to the thickness of the film at disintegration, one would expect the shorter film lengths in the presence of oil to correspond to larger average drop sizes. Figure 2 shows drop-size distributions measured using the collector shown in Figure 1, for cases of no oil and of 0.25% w f w oil. In the former case, 4721 drops were counted and measured; in the latter case, 7085 drops were counted and measured. Both cases correspond to the standard operating conditions, with the air temperature at 160 "C. The data are plotted in Figure 2 according to the square root-normal distribution function (Tate and Marshall, 1953). From these results, one can s e that the presence of oil increases both the average drop size and the spread of drop sizes. The largest drops observed were 140 pm and 320 pm in diameter for the no-oil and oil cases, respectively. The Sauter-mean (volume/ surface average) diameters were 39 pm and 58 pm for the no-oil and 0.25% -oil cases, respectively. Measurements of drop-size distributions showed that the Sauter mean diameter rose sharply as the amount of suspended oil was increased from zero to 0.1 90,and then

109

PLOT OF DROP-SIZE DISTRIBUTION

5

,

,

,

,

,

I

I

1

31

15 2535 55 75 95 125 155 185 DROP DIAMETER, x

CM

Figure 2. Drop-size distributions, with and without emulsified oil.

increased only slightly with further increases in the level of oil. This behavior is in accord with the observed effects of oil on the film length before disintegration. As is shown by Dombrowski and Munday (1968), among others, the mechanism of film disintegration in the absence of suspended droplets or bubbles is one of break-up into undulating sheets, then wavy filaments, and finally drops. In the present work, calculations were made using the method of Dombrowski et al. (1960) to find the thickness of the liquid sheet at the point of initial disintegration (Zakarian, (1979). These indicated a thickness of about 6 pm in the absence of oil, increasing to about 9 pm in the presence of 0.08% or more oil. Photomicrographs of the feed emulsions used in this work showed that the average size of oil droplets was in the range 5 to 10 pm. The results therefore support a mechanism whereby film disintegration occurs when the film thins to a thickness of the order of the oil-droplet diameter. At this point, oil droplets reach the film surface and may spread rapidly, giving disintegration by a perforation mechanism. Dombrowski and Fraser (1954) indicate that no perforation occurs for spraying of emulsions when the emulsified-droplet size is smaller than the film thickness at the point of aerodynamic disintegration. Andrew et al. (1972) showed that it is necessary for the dispersed phase to have the lower surface tension, as is the case in this work, in order for a perforation disintegration mechanism to be operative during spraying of emulsions. Volatiles Retention. Measurements made of the retentions of ethyl, butyl, and pentyl acetates for 40% sucrose solution with no oil (Zakarian, 1979) agreed substantially with the results of Kieckbusch and King (1980) for the fan-spray nozzle. The retention increased with increasing carbon number of the acetates, indicating liquid-phase mass-transfer control, as found by Kieckbusch and King (1980). The effect of added oil on the retention of ethyl acetate is shown in Figure 3. When plotted against axial distance from the nozzle, the retention without oil follows one curve, while retentions in the presence of various amounts of added oil cluster about another curve. The equilibrium distribution ratios for ethyl acetate between peanut oil and 40% sucrose solution measured by Kieckbusch and King (1979a) indicate that only 2.6% of the ethyl acetate should be extracted into the oil phase at 0.37% oil; hence the

110

Ind. Eng. Chem. Process Des. Dev., Vol. 21, No. 1, 1982 ETHYL 1

I

I

ACETATE I

I

I

i

PENTYL ACETATE

RETENTIOh

I

I

l

W T % PEANUT OIL 0 = 000 A = 005 v = 008 v 1025 m

9

60

-

55

-

50

-

c z

I

RETENTION

I

I

i

037

I 85

W

a

45-

bo

40

4 0 % SUGAR FAN SPRAY

0

12

8

4

AXIAL

16

DISTANCE

(cm)

Figure 3. Retention of ethyl acetate, aa a function of axial distance. E T H Y L ACETATE Y

I

RETENTiOh

!

l

75

m:O25 A = 008

70

T i 037 n = 081

0

:

0035

= 005

OPEN SYMBOLS WITH I IN OIL

TWEEN

A

z

0

60 w z

c W

a

55

bo

50 NO OIL

45

FAN SPRAY 40% SUGAR I

I

1

1

1

2

3

4

% EVAPORATION

Figure 4. Retention of ethyl acetate, as a function of percent evaporation.

change in retention behavior in the presence of oil should not be attributable to extraction. Kieckbusch and King (1980) found that correlating the retentions of volatiles against percent evaporation from the spray (instead of axial distance) served to eliminate most of the effects of changes in drop-size distribution upon retention. Figure 4 shows that this is also the case for the retentions of ethyl acetate observed in the present work. Although there is substantial scatter, the retentions fall around one curve for retention w. percent evaporation, without a discernible trend with changing oil level. Here percent evaporation is defined as the percentage of the total feed material (water sucrose), mass basis, that has been evaporated. A similarity analysis (Zakarian, 1979) shows that both percent volatiles retention and percent evaporation are unique functions of t/R:, where t is exposure time and Ro is drop radius, provided that the gas phase Sherwood number is constant (Le., product of radius and external mass-transfer coefficient = constant, with constant gas

+

0

4

E

12

16

20

A X I A L DISTANCE ( c m j

Figure 5. Retention of pentyl acetate, as a function of axial distance.

/

WT % OIL 0 = 0005

65

NO OIL

20

diffusivity). Thus for uniform drop size, uniform residence time, and various other less restrictive assumptions the percent retention of a volatile component should be a unique function of percent evaporation for any drop size, since both are unique functions of t/Ro2. For pentyl acetate, the phase-equilibrium data of Kieckbusch and King (1979a) indicate that about 53% of the total pentyl acetate should reside in the oil phase at 0.37% oil. Hence the extractive effect of the oil should be substantial. This prediction is borne out by the observed retentions, shown as a function of axial distance in Figure 5. Retentions of pentyl acetate in the presence of oil are substantially greater than those in the absence of oil. King and Massaldi (1974) present an analysis of the effect of an emulsified, extractive oil phase on the retention of volatile components when the loss is rate-limited by diffusion in the continuous liquid phase. If the emulsified phase is present as very fine droplets, it can be pictured as a continuous extractive sink for a volatile component, reducing thereby both the concentration and the cocentration gradient of that component in the continuous liquid phase. The diffusional loss process is thereby describable by the diffusion equation for the equivalent system with no emulsified droplets, but with an effective diffusivity D', given by D D'= l + P

where D is the diffusivity in the continuous liquid phase and p is the extraction factor, cwS/W , where is the equilibrium distribution coefficient (weight fraction of volatile component in dispersed oil phase + weight fraction of volatile component in continuous aqueous phase, at equilibrium) and S/ W is the phase ratio (massoil + mass aqueous phase). As was pointed out earlier, another factor which should be taken into consideration is the effect of the oil phase in suppressing the partial pressure, or aroma, of volatile components over a reconstituted product. Assuming Henry's law and a concentration-independent equilibrium distribution coefficient we have 1 P"f -=(2) P:, 1 + P'

cw

cw,

Ind. Eng. Chem. Process Des. Dev., Vol. 21, No. 1, 1982 111 I

I

..

Oilution of product = Dilution of fe8d

,

000

D t/ R2-0

a

001

OCQ

005010

030 =

Ks3/W

Figure 6. Fraction retention of aroma baaed upon sphere diffusion model; @’ = 8.

.

IO

b>0 9 30 times dilution

r

.x>

5+

oe *A EVAPORATION 07

Figure 8. Retention of pentyl acetate, as a function of percent evaporation.

w 06

05

Bg

Table I. Comparison of Predicted and Experimental Effects of Emulsified Oil on Retention of Pentyl Acetate at 2.0%Evaporation fraction retention in emulsion

0 4

4

g L 2

03

02

01

02 0 3

% oil wlw

0 5 0 7 10

100

f

=

g / w

Figure 7. Fraction retention of aroma baaed upon sphere diffusion model; 8’ = 8\30.

where Pd is the equilibrium partial pressure of a volatile component over the reconstituted product, P i , is the equilibrium partial pressure that would occur if there were no extraction into the emulsified oil phase, and B’ is the extraction fador for the reconstituted product, which may differ from 8 because of a different phase ratio in the reconstituted product from that in the feed to the dryer. King and Massaldi (1974) present solutions for percent retention of aroma, based upon a slab geometry. Figures 6 and 7 show solutions for the sphere geometry, expressed as fraction retention of aroma (partial pressure) as a function of 8 and Dt/R2,where t is exposure time and R is sphere radius. Two cases are shown; in Figure 6 p’ = 8, and in Figure 7 p’ = 8/30, such as might occur for spray-dried coffee, where the degree of dilution of the reconstituted product is much greater than that of the aqueous solution (extract) fed to a dryer. The solutions in Figures 6 and 7 are obtained by taking the constantdiffusivity solution for diffusion in a stagnant, single-phase sphere (see, e.g., Sherwood et al., 1975), correcting the diffusivity by eq 1, and correcting the indicated fraction retention by eq 2. Cases of concentration-dependent diffusion coefficient and changing sphere radius may, to a good approximation, be interpreted as corresponding to a certain effective value of Dt/R2 in Figures 6 and 7 after exposure time t. From Figure 6, it can be seen that the fraction aroma retention largely decreases with increasing 8 for 0‘ = 0, indicating that the depressant effect of the oil on the aroma of the reconstituted product dominates over the extractive effect during drying. However, for much greater product dilution (p’ = 8/30]the aroma retention for a given degree of exposure (Dt/R2)can reach a substantial maximum at intermediate values of 8. Under this condition, oil present

0 0.05 0.08 0.25 0.37

P 0 0.15 0.24 0.76 1.13

exptl 0.635 0.70 0.74 0.73 0.81

pred. 0.635 0.67 0.68 0.71 0.74

during drying can give a useful protective effect on the aroma of the reconstituted product. The effect of emulsified oil on retention of pentyl acetate can now be interpreted in terms of the solutions shown in Figures 6 and 7. For this purpose it is desirable to use results expressed as percent retention vs. measured percent evaporation, so as to remove much of the effect of changes in drop-size distribution. This form of plotting is shown in Figure 8. The curves cross one another, which may be the result of residual effects of drop-size distribution or directional experimental error. Note that the points for retention of ethyl acetate at 0.05 and 0.08 w t 7” oil are also relatively high at low % evaporation (Figure 4). The model expressed in eq 1and 2 and shown in Figures 6 and 7 may be used to predict fraction retentions for comparison with the results shown in Figure 8. Fraction overall retentions of pentyl acetate in the emulsion may be predicted from the stagnant-sphere model using eq 1 as a correction. This is equivalent to taking results from Figures 6 and 7 and multiplying them by (1 + @’), thereby removing the factor given by eq 2. Results of this comparison are shown in Table I, for the case of 2.0% evaporation. In this case, the retention is 63.5% with no oil, corresponding to Dt fR2 equal to about 0.015 for 8 approaching zero in Figure 6. The predicted results are obtained by holding this value of D t / R 2 constant. It can be seen from Table I that the order of magnitude of the predicted effect of extraction into an emulsified oil phase agrees roughly with the experimental data. Computer Modeling. A computerized theoretical model was developed, incorporating various mechanisms of mass transfer which should be operative in the vicinity of the atomizer and coupling these with the measured

112

Ind. Eng. Chem. Process Des. Dev., Vol. 21, No. 1, 1982 P E N T Y L ACLTATE

RETENTION

P E N T Y L ACETATE RETENTION

YO 31-

F A N SPRAY

70,

I '

i

n

I

I

I

I

I

Run 23 Dec 78

85

0.25%

011

Computer model using drop- size dlstrlbutlon

70

using D3* = 0 0058 cm

-

6

8

0

12

'

A.

14

.--. 16

AXIAL DISTANCE, C M

*"

6

8

10

12

AXIAL DISTANCE

I4

,

16

CM

Figure 9. Pentyl acetate retention: experimental data and computer predictions for no oil.

drop-size distributions and a model of aerodynamics and drop dynamics in the same region. The computer model is described in detail by Zakarian (1979). The aerodynamic and drop-dynamic equations are the same as those presented by Rothe and Block (1977), except that the initial momentum of the air and the effect of the gas velocity outside the spray are not neglected. Mass transfer from the expanding liquid film was calculated using the model of Hasson et al. (1964) as adapted to mass transfer by Simpson and Lynn (1977). Mass transfer during drop formation was estimated by adapting the extraction model of Skelland and Minhas (1971) to the gas-desorption situation and by taking a formation time (2 X 10" s) crudely estimated from the nozzle velocity and the photographically measured distance between initial and complete disintegration of the film. The effect of circulation superimposed upon diffusional mass transfer within the drops was estimated using the model of Kerkhof and Schoeber (1974) with the surface-tension gradient assumed to be 0.1 N/m2. Gas phase resistance to mass transfer was neglected, in view of the very high Henry's law constants of the acetates. Calculations were made both (1) for the actual measured distribution of drop sizes, with the fallvelocity and mass-transfer characteristics of different drop-size increments considered separately and then weighting the volatiles contents of the different fractions together, and (2) for uniform drops with a diameter equal to the Sauter mean diameter of the measured distribution. Equation 1 was used to allow for the extractive effect of the suspended oil phase. Diffusion coefficients were estimated by applying temperature corrections and the Wilke-Chang (Sherwood et al., 1975) correction for molar volume of the transferring solute to the diffusivities of ethyl acetate reported by Chandrasekaran and King (1972). Values used at 43 "C were 0.45,0.40,0.37,and 0.34 X lo+ mz/s for ethyl, propyl, butyl, and pentyl acetates, respectively, in 40% sucrose solution. Comparisons of the predictions of the model with experimental results are shown in Figures 9 and 10 for the cases of pentyl acetate with and without oil added. Similar trends and degrees of agreement were found for ethyl acetate with and without oil (Zakarian, 1979). In general, the agreement is close-within 5% of the experimental values. A common feature is that the retentions calculated using the Sauter-mean diameter are lower than thrxe made allowing separately for drops of different size ranges. This

Figure 10. Pentyl acetate retention: experimental data and computer predictions for 0.25% oil. Table 11. Predicted Contributions o f Different Mechanisms to Ethyl Acetate Loss in First 16 cm from Atomizer amount of loss (% of ethyl acetate in feed) source expanding liquid film drop formation internal circulation of drops (increase over stagnant-drop diffusion) diffusion in drops, if stagnant

no 0.25% oil oil

10 7 6

5 7 4

31 20 54

36

occurs because larger drops fall faster and reach a given axial position with a disproportionately high volatiles retention. Also, when drops smaller than the Sauter mean diameter become nearly fully depleted of volatile solute, they no longer contribute additional loss. The contributions of the various mass-transfer mechanisms considered are all important. Table I1 shows a breakdown of loss mechanisms for ethyl acetate, with and without oil, in the first 16 cm from the fan-spray nozzle at standard conditions. Losses are approximately double what would be predicted for mass transfer from instantly formed, noncirculating drops. Uncertainty exists in some of the predictions incorporated in the model. Interactions between drops of different sizes are not considered in the Rothe-Block aerodynamic model. This and other assumptions in that model are probably the main source of uncertainty. Drop oscillation is neglected, but was estimated to have only a small effect. The surface-tension gradient in the Kerkhof-Schoeber circulation model is unknown, but changing it over a range from zero (full circulation) to 0.5 N/m2 gave at most a 3% change in calculated retention. The formation time is poorly known for the Skelland-Minhas model; however, an extreme upper limit interpreted from photographs is 2 X s, which leads to no more than 7% change in retention over the extreme range of conditions. The velocity of the liquid emerging from the nozzle could not be measured, but was estimated from nozzle dimensions to be 3260 cm/s. Decreasing this by 1/3 gave no more than a 5 % decrease in predicted volatiles retention in the range of axial distances of interest. The drop-size distributions were established by counting a finite number of drops; for the numbers counted, Bowen and Davies (1951) indicated

Ind. Eng. Chem. Process Des. Dev. 1082, 21, 1 1 3-1 17

that the 95% confidence limits for the Sauter-mean diameter are about &5%. Probably a greater source of error is the very strong effect of the small number of very large drops on the derived Sauter-mean diameter. Also, the smallest drops may have evaporated before being photographed. Conclusions The effect of an added oil phase in spray drying can be separated into two parts-physical and chemical. The physical effect of oil consists of a decrease in the length of the coherent spray sheet, an increase in spray angle, and both an increase and a greater spread in drop sizes. The chemical effect of oil is to extract volatile compounds, according to their oil-water partition coefficient, into the dispersed oil phase. This increases the total amount of volatile component retained during drying, but decreases the retention of aroma (partial pressure), if the retention is referenced to an oil-free feed and the product dilution is the same as the feed dilution. A much greater dilution of the product and/or reference to a feed with oil already present improves the retention of aroma. Correlation of the volatiles retention with respect to percent water evaporation serves largely to remove effects of changes in drop-size distribution. The retentions of pentyl acetate, which is extracted to a substantial extent into the oil phase, are in at least qualitative agreement with the predictions of the dispersed-phase model of King and Massaldi (1974). A computer model was developed to describe volatiles retention in the nozzle zone. The model, which accounts for spray aerodynamics and drop-size distributions, gives reasonably good agreement with experiment. Literature Cited Andrew, S. P. S.;Darnani. M. A. K.; Dombrowskl, N. J. COAbM Interface Sei’. 1972, 4 1 , 445. Ban, T. Kagaku Kogaku Ronbunshu, 1978, 4 , 515. Ban, T. Kagaku Kogaku Ronbunshu, 1979, 5 , 213. Biakebrough, N.; Morgan, P. A. L. Birmingham Univ. Chem. €ng. 1973, 24, 57. Bomben, J. L.; Bruin, S.; Thijssen, H. A. C.; Merson, R. L. Adv. Food Res. 1973, 20, 1. Bowen. I.G.; Davies, G. P. Shell Technical Note ICT 28. 1951.

113

Bruin, S.; Luyben, K. Ch. A. M. ”Drying of Food Materials; A Review of Recent Developments”. in “Advances In Drying”; Mujumdar, A. s., Ed.; Hemisphere Publ. Co.: New York, 1980; Vol. 1. Chandrasekaran, S. K.; King, C. J. AIChEJ. 1972, 18, 513. Clinton, W. P.; Munsey, N. Y.; Kurant, T.; Pitchon, E. U.S. Patent 3244533, Apr 5, 1966. Dombrowski, N.; Fraser, R. P. milos. Trans. R. Soc. London 1954, 247A, 101. Dombrowski, N.; Hasson, D.; Ward, D. E. Chem. Eng. Scl. 1960, 12, 35. Dombrowski, N.; Munday, G. “Spray Drying”, in “Biochemical and Biological Engineering Science”; Blakebrough, N., Ed., Academic Press: New York. 1968 vol. 2. Hasson, D.;Luss, D.; Peck, R. Int. J . Heat Mass Transfer 1984, 7 , 969. Huste, A.; Breza. R. J.; Kohler, R. B. Canadian Patent 786280, May 28, 1968. Kerkhof. P. J. A. M. “Preservation of Coffee Aroma During the Drying of Coffee Extracts”. paper presented at 8th Int. Sci. Colloquium on Coffee, Assn. Sci. Inti. Caf6, Abidjan, 1977. Kerkhof, P. J. A. M.; Schoeber, W. J. A. H. “Theoretical Modelling of the Drylng Behavbr of Droplets In Spray Dryers”, in “Advances In Preconcentratlon and Dehydration of Foods”; SDlcer. A.. Ed.; A.~. ~ l i eScience d PubIishers, Ltd., Lodon, 1974. Kerkhof, P. J. A. M.; Thijssen, H. A. C. AIChE Symp. Ser. 1977, 73(163). 33 Kieikbusch, T. G.; King, C. J. Proc. 2nd Paclffc Chem. Eng. Congr. (PA. CMC-77) 1977, 1 , 216. Kleckbusch. T, G.; Klng, C. J. J . Agr. FoodChem. 1979a, 27, 504. Kieckbusch. T. G.; King, C. J. J . Chromtogr. Sci. 1979b. 17, 273. Kieckbusch, T. 0.;King, C. J. AICM J. 1980, 26. 718; Erratum, IbM. 1981, 2 7 , 528. King, C. J.; Kieckbusch, T. G.; Greenwald, C. G. I n “Advances in Drying”; Vol. 3, Mujumdar, A. S.,Ed.; Hemisphere Publ. Co.: New York, 1981, in press. King, C. J.; Massaldi, H. A. Proc. IV Int. Congr. Food Sei. Technoi. 1974, I V , 183. Leo, A.; Hansch, C.; Eikins, D. Chem. Rev. 1971, 71, 525. Reineccius, G. A.; Anderson, H. C.; Felska. B. J. J . Food Sci. 1978, 4 3 , 1494. Rothe, P. J.; Block, J. A. Int. J. Mult@haseFlow 1977, 3, 263. Rulkens. W. H.; Thijssen, J. A. C. J . Food Techno/. 1972, 7 , 95. Sherwood, T. K.; Pigford, R. L.; Wiike, C. R. “Mass Transfer“, McGraw-Hili: New York, 1975. Simpson, S.G.; Lynn, S. AICh€J. 1977, 23, 686. Skeiiand, A. H. P.; Minhas. S . S. AIChE J . 1971, 17, 1316. Tate, R. W.; Marshall. W. R., Jr. Chem. Eng. frog. 1953, 49, 169. 226. Thijssen, H. A. C.; Rulkens. W. H. Ingenieur (The Hague) 1968, 80, 45. Zakarlan, J. A., Ph.D. Dissertation in Chemical Engineering, University of California, Berkeley, 1979.

Received for review October 27, 1980 Accepted July 10, 1981

This work was supported by Grant No. ENG76-17270 from the National Science Foundation as well as by a supplementary grant from the General Foods Corporation.

Pyrite Catalysis in Coal Liquefaction Mwakar Garg’ and Edwln N. Givens Corporate Research and Development Department, Air Products and Chemicals. Inc.. Allentown. Pennsylvania 18 105

Significant liquefaction catalysis by pyrite was observed for both Elkhorn No. 3 and Kentucky No. 9 coals. For both coals conversion and oil yield increased on addition of 10% pyrite to the feed slurry. Oil yields increased from 27.3 to 41.0% for Elkhorn No. 3 and from 15.3 to 34.9% for Kentucky No. 9. Gas yields increased slightly for Elkhorn No. 3 and were essentially unchanged for Kentucky No. 9. Hydrogen consumption, after correction for the increased hydrogen sulfide make, was likewise favorable. Sulfur contents in the residual SRC material increased in both cases. Solvent hydrogen content remained constant in the presence of pyrite, whereas it decreased significantly in the absenceof pyrite, Solvent In the presence of pyrite without any cod present showed llttle change at process conditions.

Introduction Coal is a complex mixture of organic and inorganic constituents each of which has a unique response during liquefaction. Indigeneous coal minerals are not inert ingredients, but they undergo chemical and physical changes as well as catalyze the transformation of the organic phase. 0196-4305/82/1121-0113$01.25/0

The conversion of the coal to liquid products has been found to increase as the mineral matter content and the concentration of the iron and titanium in the coal increase (Mukhurjee and Choudhury, 1976). The catalysis by iron compounds in the coal liquefaction reaction has been known for a long time. The Germans 0 1981 American

Chemical Society