Miltistage Fluidized Bed Collection of Aerosols

Dec 16, 1974 - In summary, we have demonstrated four continuous ad- sorptive ... gle-bed work of Yankel (1972) to include aerosols in sizes below 0.7 ...
0 downloads 0 Views 864KB Size
others or some small pores which hinder diffusion. The selective tailing observed in the case of p -diisopropylbenzene might plausibly be related to the relative size of adsorbent pores and this particular molecule. In summary, we have demonstrated four continuous adsorptive separations that resulted from pulse test information. This information is reasonably quantitative with respect to equilibria, semiquantitative and empirical with respect to rates, and qualitative with respect to tailing. Formal modeling of the test will permit more quantitative formulation of the information. However, even in its present form it is very effective in continuous process development.

Literature Cited Broughton, D. B.. Neuzil, R. W., Pharis, J. M., Brearly, C. S., Chem. Eng. Prog., 88 (9). 70 (1970). Broughton, D. B.. Chem. Eng. Prog., 84, (8),60 (1968). Chem. Marketing Reporter, 208,25 (Dec 16. 1974). Kiplinger, J. J., "Adsorption from Solutions of NonElectrolytes," Academic Press, New York, N.Y , 1965. deRosset, A. J., US. Patent 3 917 734 (Nov 4, 1975). deRosset, A. J. Neuzil. R. W., US. Patent 3 706 812 (Dec 19, 1972). Neuzil, R . W., U.S. Patent 3 558 732 (Jan 26, 1971). Neuzil. R . W., U.S. Patent 3 855 333 (Dec 17, 1974). Snyder, L. R. "Principles of Adsorption Chromatography", Marcel Dekker, Inc., New York, N.Y., 1968.

Receiued for review April 29, 1975 Accepted December 17,1975

Multistage Fluidized Bed Collection of Aerosols Daniel McCarthy, Anthony J. Yankel, Ronald G. Patterson, and Melbourne L. Jackson' Department of Chemical Engineering, University of Maho, Moscow, Maho 83843

The collection of liquid dioctyl phthalate aerosols in near monodispersed sizes of 1.4 to 0.06 pm was observed in a 6-in. diameter, multistage column using 1-in. deep beds of alumina granules (135pm size). Collection efficiencies were very high for all particle sizes in the single-stage, fixed-bed mode at gas velocities just below that for minimum fluidization. At higher gas flow rates, single-stage collections dropped until, at 2.5 times the minimum fluidization velocity, efficiencies were 1.4 pm, 70 YO,1 .O,55 YO,0.67, 42%, 0.37, 57 YO,0.28,58 YO, 0.13,67 % , and 0.06, 81% . For the larger particle sizes the predominant collection mechanism was interception, and for the smaller sizes Brownian diffusion became significant. Impaction was not significant for any P ) for n aerosol size or flow rate observed. For multistage collection the efficiency is given by E = lOO(1 stages where the penetration, P = (1 - E') per stage. The fractional efficiency per stage, E', was found to be constant and independent of particle concentrations of 0.1-1.0X lo6 particles/cm3. At twice the minimum fluidization velocity the lowest collection for three stages was 95% for the 0.28 and 0.37 p m size aerosols, 100 YO for 1 .O pm, and 99 % for 0.06 pm. Four stages showed higher collections with pressure losses being about 0.8in. of water per stage.

-

For some applications, a fluidized bed operating in one of several modes may offer performance advantages over traditional filtration devices for the collection of fine particles (below 1.5 bm). The present work demonstrates that a shallow, f i r e d bed of small granules (135 bm) can remove nearly all aerosols in sizes from 1.4to 0.06 pm diameter. In contrast, a single bed shows rapid decrease of collection efficiency upon fluidization. However, high efficiencies can be obtained by employing several shallow beds in series which overcome the bypassing of collection mechanisms in the fluidized state. The objective of the present study was to extend the single-bed work of Yankel (1972) to include aerosols in sizes below 0.7 grn and to demonstrate a predicted performance of multistage, shallow beds. It was further desired to better establish the collection mechanisms operative in the submicron size range extending into the region of significant Brownian diffusion. Jackson (1974b) reviewed the literature for the collection of submicron particles by fluidized beds and projected the potential merit of multistage operation. Such beds are limited to relatively low linear gas velocities, but high gas throughputs can be obtained by using larger bed plan areas. A fluidized bed can be operated at very low pressure drops of several inches of water which minimizes energy consumption. Continuous addition and removal of bed 266

Ind. Eng. Chem., Process Des. Dev., Vol. 15, No. 2, 1976

granules during collection operations is a possibility unlike the periodic shutdowns required for other types of filtration equipment. The use of porous granules permits extended operation for the collection of liquid aerosols which may be unsuited for collection on fabric filters.

Basis for Granule-Bed Collection Mechanisms of Filtration. Filtration of an aerosol from a gas stream is accomplished by a combination of forces that affect the motion of a particle and may cause it to move onto a collecting surface. Mechanisms of possible importance are: inertial impaction, interception, Brownian diffusion, electrostatic attraction, settling, and thermal precipitation. In the present study the last three were considered negligible. Inertial impaction is a mechanism wherein a particle moving with an air stream which flows around a collecting object resists the change in directioh and may impact on the collector. The inertial impaction number is the ratio of the distance a particle will penetrate into still air, when given an initial velocity of Uo, to the diameter of the collector, and is given by

Ranz and Wong (1952) reported inertial impaction efficiencies for spherical collectors as functions of the square root of the impaction number. Collection was shown experimentally to be negligible for (+)l/’ < 0.2. The interception mechanism is operative in combination with the impaction mechanism but can be important even when = 0. This may be visualized as a process involving a massless particle having finite size but no inertia. The center of the particle follows the fluid stream lines, but whenever the center approaches within a distance of D,/2 from the surface of the collector the particle will be caught. Ranz and Wong (1952) developed a relationship for target efficiency in terms of the interception number

+

7 = (1

+ NR)’- (1 + NR)-’

(3)

A third mechanism that plays an important role in collection, particularly when inertial forces and particles are small, is Brownian diffusion. If a particle passes close to a collector, the random, diffusional motion of the particle may bring it into contact with a collector. Friedlander (1957, 1967) and Stairmand (1950) agree that the efficiency of collection due to Brownian diffusion is a function of the Peclet number UODC Np, = (4) DBM The Peclet number is the ratio of the fluid resistance to the diffusive force caused by thermal motion. Levich (1962) developed the case for diffusion to a single isolated sphere and gave the efficiency as (5a) and Stairmand for a cylinder gives 7 = (~/NP,)~”

(5b)

Single-Stage, Fixed-Bed Efficiencies. Theoretical interception and diffusion target efficiencies calculated for a range of particle sizes are given in Table I. The impaction numbers ranged from 6.8 X down to 4.5 X indicating zero collection efficiencies for this mechanism. Interception is indicated to be the dominant collection mechanism for the 1.4, 1.0, and 0.67-pm diameter aerosols, while diffusion is the primary collection mechanism for the 0.28, 0.13, and 0.06-pm aerosols. The overall collection efficiency in a fixed bed can be estimated by an analysis similar to that suggested by Friedlander (1958) and developed by Yankel (1972). The theoretical target efficiency per interaction is low, but the number of granules encountered and the number of aerosolgranule interactions is large as the gas stream passes through the bed. The number of interactions of the gas stream with the bed granules is estimated from simple geometry, and the number concentration of particles, N, is reduced by the target efficiency for each interaction. For a single granule of diameter D,, the removal efficiency is given by 7 = (4a)/(7rDc2)

(6)

which serves to define a as the effective area swept of all particles and rDC2/4is the projected area of a granule. For a differential depth d H in the flow direction, there are (6A dHa)/(7rDc3)granules (cubic packing) and removal is -dN = (&/A) (number of granules) = (6aN d H a)/(7rDc3) (7) Rearranging and substituting for a from eq 6

Table I. Fractional Target Efficiencies per Granule Interception

Aerosol diameter, -~ NR I.lm Eq 2 1.1 X IO-’ 1.4 1.0 7.4 x 10-3 5.0 X 0.67 0.37 2.7 x 10-3 0.28 2.1 X 0.13 1.0 X 4.0 x 1 0 - 4 0.06

Target efficiency x IO2

Np,

Eq3

Eq4

3.1 2.2 1.6 0.8

0.6 0.3 0.1

Diffusion

1.0

X

Target efficiency x l o z

_____._~

lo5 104 lo“

7.0 x 4.4 X 2.1 x 104

1.5 X IO4

4.7 X l o 3 1.3 x 103

Eq5a Eq5b

._

0.2 0.2 0.3 0.5

0.7 1.4 3.4

-dN/N = (3/2)(7a/DC)d H

0.9 1.1 1.3 1.9 2.3 4.1 8.0

(8)

Integrating from H = 0 to H In NJNz = (3qaH)/(2DC)

(9)

The overall aerosol collection efficiencies in a 1-in. fixed bed attributable to interception and diffusion mechanisms are given in Table 11. The relative significance of the two mechanisms is apparent; with interception and diffusion occurring simultaneously, a net collection efficiency higher than either one separately would result. The efficiencies may not be directly additive because the mechanisms may either duplicate or augment each other. Also, entrapment in the wake region of a granule may add to collections. Multistage Efficiencies. Single-stage fluidized beds do not exhibit collection efficiemiss as high as those attained in fixed beds. Yankel (1972) determined that aerosol collection efficiencies decreased substantially once the gas velocity through the bed increased beyond the minimum fluidization velocity. The decrease in efficiency was attributed to some bypassing of collection mechanisms from the gas bubbles created during fluidization. The use of several stages in series permits the gas from each stage to become mixed before entering a subsequent stage. This stagewise concept is similar to the successive extraction technique in analytical chemistry. Assuming complete mixing between stages and a constant stage efficiency (correct for monodispersed aerosols), collection by n beds in series would result in an outlet aerosol concentration of N2 = N1(1 - E’)” = NIPn

(10)

and an overall collection efficiency of

E = lOO(N1 - Nz)/Nl = l O O ( 1 - P”)

(11)

Jackson (1974b), using a single-stage efficiency of 72%, predicted that three 1-in. beds a t a gas velocity of twice the minimum fluidization velocity would collect 0.7-pm aerosols a t 98% efficiency. Also, two 3-in. beds in series will collect 0.7-pm aerosols a t 98% collection efficiency. The pressure loss across the three 1-in. stages would be 2.4 in. of water while the two 3-in. stages would give a pressure loss of 4.8 in. of water. This demonstrates that energy losses for a given collection efficiency in a multistage fluidized bed can be minimized by optimizing the depths and the numbers of the fluidized beds. Previous Work Kunii and Levenspiel (1969) reviewed the extensive work on mass and heat transfer and catalytic reactions in fluidized beds and presented a model of the fluidized bed which describes a bubble and an emulsion phase. The first is esInd. Eng. Chem., Process Des. Dev., Vol. 15, No. 2, 1976

267

Table 11. Predicted Fixed-Bed Overall Collection Efficiencies0

D”,Pm 1.40 1.00

0.67 0.37 0.28 0.13

Interception,

Eq 5a

%

91 82 68 46 38 21

0.06 8 Bed depth = 1 in., U , solids fraction = 0.28. a

Diffusion, % based on use of Eq 5b

49 56 22 64 33 77 40 83 67 96 92 100 1.5 cmisec, 135-pm granules, 14 17

=

sentially solid-free and consists of fast-moving bubbles surrounded by thin clouds of granules and accompanying wakes and little or no chemical reaction or heat transfer occurs in this phase. In contrast, the emulsion phase is at the minimum fluidizing condition and is the region providing most interactions. Very shallow beds, as considered herein, do not result in the large bubble masses formed in deep beds and little attention appears to have been given to this situation in the literature. However, the model cited provides at least a qualitative explanation of the results of particle collection in a fluidized bed. Initial contact of the granules is probably with a gas jet at the bed support surface; the jets then develop into numerous small gas bubbles which cannot grow to large sizes before reaching the bed surface as for deep beds. Only partial contact of bed and gas phases thus occurs. Collection i n Fixed Beds. Fixed granular beds of various materials have been used to remove particulates from air to streams at high collection efficiencies which involve impaction, interception, and diffusion mechanisms. The emulsion phase of a fluidized bed is closely related to the fixed bed condition with respect to collection behavior. The same mechanisms are active for the collection of aerosols by fibrous and fabric materials. Juvinal et al. (1970) provide an extensive literature review of fixed bed collection. Blasewitz and Judson (1955) compared the performance of glass fiber filters and sand bed filters for the removal of methylene blue aerosols from gas streams flowing a t 1.5 to 2.0 cm/sec. Thomas and Yoder (1956) collected DOP aerosol in sand filters and on lead shot at low flow velocities; aerosol collection efficiencies increased when the bed operating mode was changed from an upflow pattern to a downflow pattern indicating gravity settling. Zahradnik et al. (1970) reported similar results for the collection of fly ash on alkalized alumina granules. Squires and Pfeiffer (1970) and Paretsky et al. (1971) collected particulates in sand beds with efficiencies attained exceeding 99%; they concluded that impaction was the primary collection mechanism active in the removal of aerosols with sand diameters exceeding 1.1 wm. Jackson (1973, 1974a) showed that very high velocities are necessary for the collection of submicron size particles by impaction and that diffusion becomes a factor for the collection of aerosols with diameters less than 0.3 wm in an impinger containing a liquid. Collection i n Fluidized Beds. The ability of fluidized beds to collect particles, although at less than desirable efficiencies, has been demonstrated by a few studies. Such beds provide a means of collecting aerosols with a minimal expenditure of energy, can operate in humid environments, and can be replaced continuously during operation. Singlepass fluidized beds exhibit lower aerosol collection efficiencies than fixed beds of granular material: Meissner and Mickley (1949), Scott and Guthrie (1959), Pilney and Er268

Ind. Eng. Chem., Process Des. Dev., Vol. 15. No. 2, 1976

ickson (1968). Some large scale operations have also been reported, Jugel et al. (1970) and Cook et al. (1971). Grace and Harrison (1970) inserted baffles in fluidized beds to effect bubble collapse which reduced bubble bypassing in the bed. Black and Boubel (1969) indicated that efficiency was primarily a function of interception with diffusion present to a small extent, and that impaction was not a factor for micron and submicron size aerosols. The analysis of Friedlander (1958) was employed to provide a correlation for a given bed depth and efficiency. In an effort to increase the collection efficiency, Boubel and Junge (1971) employed a fluidized bed in a centrifugal field. Glass spheres of 15-pm size collected sodium chloride particulate of “submicron” size at efficiencies not exceeding 60%. Anderson and Silverman (1958) collected small particles in shallow fluidized beds containing granules having an induced or applied electrostatic charge. The use of induced charges on the aerosol and bed granules produced collection efficiencies of 95-9996, considerably exceeding those for uncharged aerosols and bed granules. Pilney and Erickson (1968) suggested that fly ash collection might be improved by the use of a corona discharge upstream from a fluidized bed. Yankel (1972) investigated the collection of liquid dioctyl phthalate aerosols in fixed and fluidized beds of activated alumina granules. Collection efficiency was determined as a function of three variables: bed depth, gas velocity, and particle size with aerosol concentrations determined by measurement of laser light attenuation. Bed depths up to 4 in. in the fixed state gave collections of over 99% for aerosols of 0.67 or 1.4 wm diameter. Aerosol collection efficiencies decreased as the gas velocity increased above the minimum fluidization velocity, but collection efficiencies increased with bed depth. The predominant mechanism of collection for aerosols in the 0.67 to 1.4-micron range was considered to be interception. The porous alumina granules absorbed substantial amounts of dioctyl phthalate, up to 7% before bed freezing occurred. Experimental Procedures The experimental arrangement of particle generator, multistage collector, and particle counter is shown in Figure 1. A condensation aerosol generator for producing near monodispersed aerosols was constructed as described and evaluated by Liu et al. (1966) and Tomaides et al. (1971). The generator produces aerosols in sizes ranging from 0.04 to 1.4 pm by atomizing a liquid such as DOP (dioctyl phthalate) with filtered air, vaporizing the atomized droplets in the heated section of a glass tube, and condensing the vapor in a cool section of the tube. The homogeneity of the aerosol obtainable from the generator was increased by using a flow divider to collect the central 5% of the aerosol stream with the remainder discharged. The 5% central core was then mixed with diluting air to obtain the desired concentration of particles which ranged from 170 to 415 mg/ m3. The aerosol size could be varied by diluting the DOP in the atomizer with 100% ethanol to a specified concentration. A Battelle research-type cascade impactor was used to determine size distributions of the larger aerosols. The geometric standard deviation of the 1.4 fim aerosol was 1.16 with the flow divider in place and 1.17 without the flow divider; these compare to a value of 1.14 reported by Liu et al. (1966). The tower shell (Figure 1)consisted of seven vertical sections of 6-in. i.d. Plexiglas pipe. Aerosol-laden gas entered the bottom of the tower and was uniformly distributed over

h T V E N T

FILTERED

2.0

4

ATOMIZER

-

10-

\

CONTROLLED POWER SOURCE 95% EXCFSS

VENT

0

0908070 60 5-

\

b-;;;::; ,

,

GARDNERCNC

0 40 30 2-

ALUMINA GRANULES D, = 13% m

FILTERED DILUTION AIR

I

W

m

Figure 1. Generator, collection beds, and detector.

04

the cross section by a 7-cm depth of ceramic spheres (1-cm diameter) and a 20-cm vertical space between the ceramic spheres and the retaining screen of the first bed. The gas then traveled through five 1 in. deep alumina beds in series, each separated by a height of 6 in. of free volume. A 16-in. disengagement section followed the fifth alumina bed. The gas passed from the tower through a rotameter and was vented to a laboratory hood. The alumina retaining screens were 316 stainless steel, perforated plates, with a 6.2% free area and trapezoidal perforations measuring 535 pm by 280 pmbby 170 pm. The activated alumina, manufactured by Alcoa, had a measured mass mean diameter of 135 pm and a geometric standard deviation of 1.14. Gas flow rates through the bed ranged from 6 to 37 SCFH for the fixed state and from 37 to 100 SCFH in the fluidized state. The gas flow rate a t minimum fluidization corresponded to a gas velocity of 3.1 ft/min (1.56 cm/sec). The pressure drop across each bed reached a maximum of 0.8 in. of water as depicted in Figure 2 which shows typical fluidized bed behavior. The pressure drop of the ceramic spheres and the alumina retaining screens were 0.1-0.8 and 0.01-0.06 in. of water, respectively, for the range of gas flow rates employed. The number concentration of the DOP aerosol was measured with a condensation nuclei counter (CNC). Sample ports were located below the ceramic spheres, just below each of the granule retaining screens, and in the transport disengaging section of the tower. Samples were withdrawn from each test port and allowed to flow through the CNC for 1 min before a determination of number concentration was made. The background concentration from the alumina beds increased from 1200 to 10000 particles per cubic centimeter as the gas velocity increased from 3 to 9 ft/min. This was far below the aerosol concentrations which were from 0.3-1.0 X lo6 particles/cm3. Condensation nuclei counters employ the principle that small particles act as nucleation sites for the condensation of water vapor under certain circumstances. Fawcett and Gardner (1958) described the CNC used in this study which is currently produced by Gardner Associates, Schenectady, N.Y. Multistage Collection Performance The apparatus and procedures employed by Yankel (1972) were duplicated initially in this study to compare the aerosol collection efficiencies determined by the laser ight attenuation method against those obtained with the 2ardner CNC. Figure 3 gives the aerosol collection efficien:ies of two DOP sizes by a 2-in. deep alumina bed (175 pm yanules) as a function of gas velocity as determined by the wo methods. The efficiencies determined by the CNC nethod lie below those obtained by laser light attenuation. Phis difference is considered to be within the experimental

06 08 1

2

3

4 5

GAS VELOCITY (CM/SEC)

Figure 2. Typical pressure loss for the shallow fluidized bed.

l 95

--ap

90

o

o

r

-

>V

z

u

85-

LL IL W

z

o_ 80c W V 0 J 0

SOLID LINE

75-

- CNC

DOTTED LINE - L A S E R

( Y A N K E L , 1972 I

I

70 0

I O

I 5

VELOCITY R A T I O

20 ( U /Urn)

5

Figure 3. Comparison of detection methods, narrow bed.

AEROSOL DIAMETER ( f i n - )

Figure 4. First stage collections, 6-in. diameter beds, 1 in. deep.

errors inherent in each method, although the CNC method was considered more reliable as employed. For the 2-in. diameter bed and 175-wm bed granules, essentially 100% collections were observed for the 0.67 and 1.4-pm aerosols in the fixed-bed mode as shown in Figure 3. For a single 6-in. diameter bed (Figure 4), efficiencies were high at 96-100% for the 0.37 to 0.06-wm size range just a t Ind. Eng. Chem., Process Des. Dev., Vol. 15, No. 2, 1976

269

incipient fluidization (1.5 cmlsec). For the 0.67 to 1.4 fim sizes, collections approached 100% a t UIU, = 1.0. As indicated in Figure 5, collections dropped rapidly upon fluidization even a t values of UIU, = 1.1.Below incipient fluidization, and within the ability of the techniques employed to assess performance, the fixed-bed mode gave collections a t or near 100% for all sizes observed for the 135 pm size granules. Table 11, compared to Table I11 for the situation approaching the fixed-bed mode, indicates that observed collections are higher than might be expected using eq 5a because collections by the two mechanisms would not be directly additive. Preliminary attempts to assess a buildup of an electrostatic charge on the bed granules in the ungrounded plastic column indicated that probably any such effects were small and of minor significance. The charges were measured by a sensitive electrometer. Operation of the bed using aerosols passed through a krypton-85 radioactive source field to discharge the particles showed no differences from collection aerosols not so discharged. A high charge is not expected on aerosols produced by condensation processes. Thus the observed efficiencies are considered to agree with expected collections (Table 11) and wake collection mechanisms, which were not included in the analysis, would add to collections. The alumina granules are not perfect spheres and behavior would lie between that for eq 5a and 5b. The close packing of the granules would also reduce the flow channels from the situation assumed for single targets. A plot of collection efficiency across the lowest bed as a function of aerosol diameter and the velocity ratio in the tower is given in Figure 4. Aerosol collection efficiencies decreased as the aerosol diameter decreased from 1.4 to 0.67 pm and then increased as the aerosol diameter decreased further to 0.06 pm. This result is expected because significant Brownian diffusion effects occur for the smaller aerosol diameters (Table 11). The small diffusion distances in the granule bed appear to make this mechanism more significant than for a similar situation in lung transfer processes although the same trend with particle size is observed. The sharp decrease in aerosol removal efficiency upon fluidization is not the result of a change in the major collection mechanisms. The velocity of the gas in the emulsion phase fluidization is considered to be equal to the minimum fluidization velocity when bubbling of the bed occurs. The emulsion phase does represent a small expansion from the fixed-bed state, and the interception and diffusion effects are probably reduced a small amount. The major effect results from the jet-bubble passage of some of the gas wherein some particles by-pass the collection media. However, the gas within the bubble is usually considered to continually exchange with gas in the cloud and emulsion regions of the bed, and the aerosol within the cloud undergoes some interaction with the bed granules. Aerosol collection efficiencies are seen to decrease with increasing fluidization of the alumina bed for aerosols of any size because of the increasing fraction of the gas that traverses the bed in the bubble phase. Single beds that contain perforated plates, screens or baffles, which permit free passage of granules but which cause bubble collapse and reformation, might enhance aerosol collection. Alternately, the arrangement of multiple shallow beds in series was found to provide collection efficiencies exceeding considerably that of a single bed of equivalent total depth, as predicted by eq 11. For monodisperse aerosols the single-stage collection efficiencies are independent of particle concentrations because the collection mechanisms are not concentration de270

Ind. Eng. Chem., Process Des. Dev.. Vol. 15, No. 2, 1976

-s

100

I

30

20 0

VELOCITY

RATIO U / U v

Figure 5. Cumulative collections, minimum performance.

Table 111. Cumulative Collection Efficiencies First Aerosol ity stage, diameter, ratio, obsd, um UIU, %

1.4

1.o

1.1

1.2 1.9 2.5 1.1 1.2

1.9 2.5 0.67

1.1

1.2 1.9 2.5 0.37

0.28

0.13

0.06

1.o

1.5 2.0 2.7 1.0 1.5 2.0 2.7 1.0 1.5 2.0 2.7 1.0

1.5 2.0 2.7

81 78 74 70 80 74 61 55 71 60 52 42 96 74 64 57 97 76 68 58 100 83 73 67 100 93 88 81

Second stage

Third stage

Obsd,

Calcd,

Obsd,

Calcd,

%

%

%

%

95 95 97 94 93 91 85 74 92 87 79 73 99 91 88 80 99 92 86 82

96 95 93 91 96 93 85 80 92 84 77 66 100

93 87 82 100 94 90 82

100

100

96 92 88

97 93 89

100

100 100

98 98 96

99 96

100 100 100 100 100 100 98 95 100 98 96 92 100 97 95 92 100 98 95 91 100 98 98 96 100 99 99 99

99 99 98 97 99 98 94 91 98 94 89 80 100

98 95 92 100

99 97 93 100

99 98 96 100 100 100

99

pendent. Collections by stage with gas velocity are summarized in Table I11 where the observed efficiencies are compared with the stagewise efficiencies calculated for eq 11 using the aerosol collection efficiency across the first stage. The observed collection efficiencies approximate the calculated efficiencies, being sometimes larger and sometimes less, and are indicative of the experimental variation in the methods employed. Discussion a n d Design Considerations Figures 5 and 6 show typical behavior of the multistage, fluidized beds for two sizes of the aerosol representing a minimum collection condition (0.67 pm) and that for a very small particle size (0.13 pm). At twice the fluidization velocity, collections for three and four stages gave efficiencies of 95 and 9896, and 97 and 99%, for the two sizes; collections exceeded 99% a t the lower fluidization velocities, Collection efficiencies for other sizes are reported, McCarthy (1974), with a general behavior as indicated by Figures 5 and 6 and Table 111. The collections of the 0.06-fim DOP size aerosol were above 99% for three stages a t all flow velocities observed, and the 1.0 and 1.4-pm size particles were collected a t nearly 100%for velocity ratios below UIU, = 2. The collection performance of multistage fluidized beds can thus be predicted from single stage efficiencies which indicates for many applications that more than three stages would

60L Dp: 0 1 3 y m

0

d 0

plates. Behie and Kehoe (1973) and Yoon and Thodos (1973) developed models to explain the effect of distribution plate type, orifice size, and orifice jet velocity on the conversion in fluidized bed catalytic reactors. Results of the present study show that multistage, shallow fluidized beds have potential for high collection efficiencies of aerosol particles of sizes which are difficult to collect, Le., those below 2-km diameter. Single beds operating without fluidization, just below minimum fluidization, or with downflow of the carrier gas to attain high throughputs, can collect at high efficiencies for bed granules of small size. The decreasing collection efficiencies with increasing gas flows for fluidized beds can be overcome by the use of beds in a series with bed replacement during operation. An alternate procedure would be to employ the high collection effectiveness of fixed beds with bed mixing and replacement by periodic fluidization. Replacement or regeneration of bed granules is a separate consideration. Some literature reports the use of nonporous bed granules to collect liquid and solid particles but a t low bed capacities. Porous granules are indicated for use for liquid aerosols or for gas streams with a high moisture content, which might also be removed by the bed material, to extend bed capacity. Multistage fluidized beds have been used by industry for purposes other than particle collection: the activation of charcoal, Godel (1948); the calcination of limestone, White and Kinsalla, (1952); solids drying, Toei (1956); solids heating, Lurie (1959); and the reduction of iron ore, Labine (1960). A “fluid bed dry scrubber” has been employed on an industrial scale for air pollution abatement on aluminum potlines with a number of installations completed, Rush et al. (1973). The flouride recovered results in a net profit being obtained because of reduced raw material requirements. One instance is cited giving a profit of $5/ton of aluminum produced. The recovery includes submicron particulate and gaseous fluoride in the fluidized bed, Cook et al. (1971). Jugel et al. reported briefly on the large scale fluid collection of large size “dusts” having an induced electrostatic charged and affected by gas stream humidity. Bed performance is a function of gas velocity, bed collection material, bed depth, and type and size of particle to be removed. An optimization routine in terms of a specific situation, using the considerations provided above, is desirable for the evaluation of a particular application.

PO

30

I 0

VELOCITY

RATIO

U/Um

Figure 6. Cumulative collections, small aerosol size.

not appear to be necessary, and that five stages would almost seldom be indicated except for much higher gas flow rates. The effect of single-bed depth was demonstrated by Yankel (1972) for the 0.67-pm DOP size. At U/U, = 2.5, a 1-in. bed showed collections of 68%, a 2-in. bed 76%, and 3and 4-in. beds a t 81%. Similar results were obtained for the 1.4 pm size DOP with the corresponding bed depths showing 75, 82, and 89% collections. The results of the four bed depths for the 2-in. diameter column serve to indicate the undesirability of attempting to attain higher collections in fluidized beds by increasing bed depth. Concentration changes in transfer devices are best evaluated in terms of the transfer unit, Nt = In 1/(1- E ) , because it gives a direct measure of the effective equipment size. A collection of 63% corresponds to one transfer unit and 86% to two units meaning, for a given type of collection device and mechanisms, that the equipment must be doubled in size to increase the efficiency from 63% to 86%. The number of transfer units required increases rapidly for higher collection: three for 95%, four for 98, five for 99.3%, six for 99.7%, and seven for 99.9%. Thus, for the 0.67-pm aerosol and the 2-in. diameter column, a 3-in. deep single bed a t 81% was only half again as effective as a 1-in. bed a t 68%. For the 6-in. diameter beds, three 1-in. beds of granules in series (0.67-pm size, 1.0 U/U,) gave 95% collection and provided three transfer units compared to one unit for a single 1-in. bed a t 61%. This illustrates the gain in using multistage, shallow beds as compared to deeper single beds, wherein the mixing between stages makes maximum use of the collection processes. The 2-in. and 6-in. diameter bed performance are not directly comparable because of different characteristics of the bed granules and the different methods of assessing particle concentrations. A recent paper by Patterson and Jackson (1975) reports an extension of the work described herein to include glass bead collectors (515-pm size). Collection efficiencies were similar to those for the alumina granules at corresponding U/Umratios with U , a factor of 10 higher. The behavior of solid aerosols was also assessed. The higher flow velocities showed increasing collections with flow rate because of significant impaction effects. Liquid aerosols are best collected with porous granules to increase bed capacity; solid granules have shorter useful lives because with liquid aerosols, and perhaps also with gases of high humidity, freezing of the bed may occur after a period of collection. Thus, an ideal type of bed granule would be one of high density, porous, and with an optimum size such as to maximize collection among the three active mechanisms consistent with desired gas flow rates. The ability of the shallow beds of alumina granules to remove very substantial amounts of the DOP aerosol is consistent with reports which state that a significant portion of the heat and mass transfer, occurring in a fluidized bed, results from the processes operative near the gas distributor

Acknowledgment Support of this study was provided in part by Grant No. GK-39832 from the National Science Foundation, and by two awards from the Graduate School of the University of Idaho. The paper was presented in part a t the annual meeting of the Pacific Northwest International Section of the Air Pollution Control Association, Boise, Idaho, November 1974. Nomenclature A = bed plan area, cm2 a = area corresponding to a region of flow completely cleared of all particles by a granule, cm2 C = Cunningham slip factor, dimensionless DBM = Brownian diffusion coefficient, cm2/sec D , = diameter of collector granule, cm D , = diameter of particle, cm (or micrometers where des ignated) E = overall collection efficiency in bed, % E’ = fractional collection efficiency per stage H = bed depth, cm n = number of stages Ind. Eng. Chem., Process Des. Dev., Vol. 15. No. 2, 1976

27

403 (1970). JuvInaII, R. A,, Kessie, R. W., Steindler. M. J.. ANL-7683, National Technical Information Service, Springfield, Va., 1970. Kunii, D.. Levenspiel, O., "Fluidization Engineering", Chapter 6, Wiley, New York, N.Y., 1969. Levich, V. G., "Physiochemicai Hydrodynamics", Prentice-Hall, Englewood Cliffs, N.J.. 1962. Labine, R. A.. Chem. Eng.. 67, 96 (1960). Liu, Y. H.. Whitby, K. T.. Yu, H. H. S., J. Rech. Atmos., 3, 397 (1966). Lurie, Y. S., Portland Cement, (1959) McCarthy, D.. M.S.Ch.E. Thesis, University of Idaho, Moscow, 1974. Meissner. H. P., Mickley, H. S., ind, Eng. Chem., 41, 1238 (1949). Paretsky, L., Theodore, L., Pfeffer, R., Squires, A. M., J. Air Poll. Control Assoc., 21, 204 (1971). Patterson, R. G., Jackson, M. L.. Paper No. 22b, annual meeting, American Institute of Chemical Engineers. Los Angeles, Calif., Nov 18, 1975. Pilney, J. P., Erickson, E. E.. J. Air Poll. Control Assoc., 18, 684 (1968). (A detailed account is given in Department of Interior, Bureau of Mines, Contract No. 14-69-0070-375, multilih). Ranz, W. E., Wong, J. B.. ind, Eng. Chem., 44, 1371 (1952). Rush, D., Russell, J. C., Iverson, R. E., J. Air Poll. Control Assoc., 23, 98 (1973). Scott, D. S., Guthrie, D. A., Can. J. Chem. Eng., 37, 200 (1959). Squires, A. M., Pfeffer, R., J. Air PoU. Control Assoc., 20, 534 (1970). Stairmand, C. J., Trans. lnst. ofchem. Eng., 28, 130 (1950). Thomas, J. W., Yoder, R. E., Arch. lnd. Health, 13, 545, 550 (1956). Toei. E., Chem. Eng. (Japan), 20, 551, 695 (1956). Tomaides, M., Liu, 6. Y. H., Whitby, K. T., Aerosol Sci., 2,39 (1971). White, F. S., Kinsalla. E. L.. Mining Engineering, 4, 903 (1952). Yankel, A. J., M.S.Ch.E. Thesis, University of Idaho, Moscow, 1972. Yoon, P.. Thodos, G.. A.LCh.€. J., 19, 625 (1973). Zahradnik, R. L., Anyigbo, J., Steinberg, R . A,. Toor, H. L., Environ. Sci. rechnoi., 4, 663 (1970).

N = number concentration of aerosol, cm-3, N1 entering,

N Zleaving

P = fractional penetration per stage, = (1- E') U = superficial gas velocity in bed, cmlsec Uo = initial particle velocity, cmhec U , = superficial gas velocity at minimum fluidization,

-

cmlsec = solids fraction in bed 1 target efficiency per granule, % p = gas viscosity, P p m = micrometers CY

Literature Cited Anderson, D. M., Silverman, L., Air Cleaning Laboratory, Harvard University, to the U S . Atomic Energy Commission, NYO-4615, 1956. Behie, L. A., Kehoe, P., A.l.Ch.E. J., 19, 1070(1973). Black, C. H.. Boubel. R. W., lnd. Eng. Chem., Process Des. Dev., 8, 573 (1969). Blasewitz, A. G., Judson, B. F.. Chem. Eng. Prog., 51, 6-7J (Jan 1955). Boubel, R. W.. Junge, D. C., 64th annual meeting, A.I.Ch.E., San Francisco, Calif.. 1971. Cook, C. C.. Swany, G. R., Colpitts, J. W., J. AirPoll. ControiAssoc., 21, 479 (1971). Fawcett, H., Gardner, G., lnd. Eng. Chem.. 50, 87A (1958). Friedlander, S. K.. A.l.Ch.E. J., 3,43 (1957). Friedlander, S. K., hd. Eng. Chem., 50, 1161 (1958). Friedlander, S. K., J. Collokfhterfac. Sci., 23, 157 (1967). Godel, A., Chem. Eng., 55, 110 (1948). Grace, J. R.,Harrison, D., Chem. ProcessEng., 51, 127(1970). Jackson, M. L.. Patterson, R. G.. A.l.Ch.€., Symp. Ser., No. 747, 47 (1975). Jackson, M. L.. J. Air Poll. ControlAssoc., 24, 569 (1974a). Jackson, M. L., A.l.Ch.E. Symp. Ser. No. 741, 70, 82(1974b). Jugel. W., Reher, E. D.. Grobler, R., Tittman. A,, Chem. Tech. (Leipig), 22,

Received for reuiew May 5, 1975 Accepted October 28,1975

Thermodynamic Equilibria of Selected Heterocyclic Nitrogen Compounds with Their Hydrogenated Derivatives Joseph F. Cocchetto and Charles N. Satterfleld' Department of Chemical Engineering, Massachusetts lnstitute of Technology, Cambridge, Massachusetts 02 739

In the catalytic hydrodenitrogenation of liquid fuels the heterocyclic nitrogen compounds are those most resistant to removal. Thermodynamic analysis of the principal steps in the reaction of representative compounds (pyridine, pyrrole, quinoline, isoquinoline, indole, acridine, and carbazole) reveals that under some significant reaction conditions the overall rate may be at least partly governed by the equilibrium of the first step, the hydrogenation of the N-containing ring. There is no significant thermodynamic limitation on the principal subsequent steps or on the reaction as a whole. Unusual kinetic behavior that has been observed, such as the existence of a maximum in rate with increased temperature, can be well interpreted in terms of a thermodynamic limitation to the allowable concentration of the hydrogenated heterocyclic compound coupled with hydrogenolysis of the C-N bond being the rate-limiting step.

The removal of undesirable nitrogen compounds from petroleum and synthetic crudes derived from coal and oil shale is best achieved by catalytic hydrodenitrogenation (HDN).Most of this nitrogen is in the form of heterocyclic nitrogen compounds, which are the most resistant to HDN. Hydrodenitrogenation of heterocyclic nitrogen compounds proceeds in general via saturation of the heterocyclic ring, followed by ring fracture and subsequent removal of the nitrogen as ammonia. This HDN mechanism is exemplified below for pyridine (McIlvried, 1971; Sonnemans et al., 1972). 272

Ind. Eng. Chem., Process Des. Dev., Vol. 15, No. 2, 1976

3

=&

CsH~iNHz

cjHlz

+ "3

(l)

H Through experimental studies of both model nitrogen compounds and actual or simulated feedstocks, investigators have attempted to elucidate HDN mechanisms and to determine the kinetics of the various steps, as well as the kinetics of overall nitrogen removal. Little has been reported, however, about possible thermodynamic limitations