1358 m
INDUSTRIAL AND ENGINEERING CHEMISTRY =
q
=
Q
=
d&l =
dQ2 =
dVl =
dV2 =
C
=
E
= =
+
electronic current density a t surface of dust layer, statamperes per sq. em., or amperes per sq. cm. charge carried by each cc. of dust as it is precipitated, statcoulombs per cc., or coulombs per cc. charge density upon exposed surface of dust layer, statcoulombs per sq. em. increase in charge density due to approaching charges, statcoulombs per sq. cm. decrease in charge density due to leakage (conductivity) statcoulombs per sq. cm. increase in potential of dust layer surface due to approaching charges, statvolts decrease in potential of dust layer surface due to leakage (conductivitx), statvolts capacitance of dust layer per unit area, statfarads per sq. cm. dielectric constant of precipitated dust total current density due to incidence Of electrons and charged dust, statamperes per Sq. em. or amperes per sq. cm.
Vol. 43, No. 6
LITERATURE C I T E D
(1) Hudson, C . M., Hoisington, L. E., a n d Royt, L. E., P h y s . Rea., 52, 664 (1937). (2) McWhirter, C. H., a n d Posey, R. F., Elec. Eng., 68, 783 (September 1949). (3) Parkinson, D. B., Herb, R. G . , Bernet, E. J., a n d McKlbben, J. L., Phvs. Rea., 53, 642 (1938). (4) Schmidt, W. A, ISD. ENG.CHEX.,41, 2428 (1949). (5) Schmidt, W. A., Sproull, TT. T., a n d X a k a d a , Y., paper presented a t A.I.M.M.E. Meeting, Kew York, Feb. 15, 1950. (6) Seite, F., “Modern Theoiy of Solids,” pp. 56, 67, etc., New Yolk, McGraw-Hill Book Co., 1940. ( i )White, 1%. J., and Anderson, E., unpublished laboratory repoi ts. (8) Zebrowski, S. P., Phys. Z., 33, 727 (1932). RECEIVED January 3, 1951. Material presented in part at 170th General Meeting, American Institute of Mining and Metallurgical Engineers, New York, February 15, 1950 (61, and in part at the Symposium on Dispersions in Gases, Division of Indujtrial and Engineering Chemistry, AXTERIC.4X CHEMICAL SOCIETY, Baltimore, RId., December 29, 1950.
Collection of Aerosols in a Venturi Scrubber F R A N K 0. EKMAN’
AND
H. F. J O H N S T O N E
UNIVERSITY O F ILLINOIS. U R B A N A , ILL.
T
HE Venturi scrubber a and T h e object of this work was t o determine t h e variables affecting t h e effisuccessful device for removing aerociency of t h e Venturi scrubber foscollecting aerosols. Oil aerosols of three difSoluble gases, and Odors from gas ferent drop sizes were used i n a laboratory unit constructed of Plexiglas, and streams. I t was first used on an induswater was introduced into t h e throat i n three different ways. trial scale in 1947to remove salt cake fume Thecollection efficiency of t h e scrubber depended on t h e operating conditions from kraft furnace gases ($9 s). Since and t h e nature of t h e aerosol. A method of correlating t h e efficiency w i t h then it has been applied successfully to the pressure loss is proposed. the collection of a wide variety of dusts This method i s useful for extrapolating t h e performance of pilot plant and and f ~ ~ m (e7s) . Johnstone and industrial installations t o determine t h e o p t i m u m design and operating con(6) have measured the rate of absorption ditions. of sulfur dioxide in the scrubber. This paper reports a study of the effect of several variables on the performance of a small unit for iemovDo = __ 16,000 + 1.4T,1.6 (1) ing oil aerosols. Methods of correlating the collection efficiency u2 and pressure loss are suggested and are applied t o the cxtrapoThis equation has been used successfully in the present work to lation of test data from large scale industrial installations. correlate collection efficiencies when the velocity of the gas through the throat varied from 200 to 500 feet per second. Although all collisions involving aerosol particles with water DEPOSITION BY I M P A C T I O N surfaces are assumed to be inelastic, factual proof of this is lackIf the particles in an aerosol are smal]-i.e., diameter of 0.1 ing. A nonsoluble particle, or one which is difficultly viet, such micron or less-they may be collected by the mechanism of as an oil droplet or a silica particle, may rebound after a glancing Brownian diffusion. If the diameter is much above 0.1 micron, collision more readily than a highly soluble particle. as in this work, the mechanism of impaction becomes predoniiIf it is assumed that each water droplet sweeps out an effective nant. The efficiency of impaction Won an object may be expath of length P , the effective volume of the tube from which all particles are removed by the passage of the drop is ?:?rD?P/4, pressed as the ratio of the cross-section area of the hypothetical where i t is the average impaction efficiency. The number of tube of gas from which the particles are all removed to the frontal droplets in L gallons of water is 6Lc‘/rD,“,where c’ is a converarea of the object, The efficiency is a function of the dimensionsion factor converting gallons to volume units consistent with Do. less group 0 = m u / k D ( 6 ) . For spherical Particles, m = “ d 3 P p / 6 The total volume swept out by all the drops is 3qtPLc‘/2D0. If L is the number of gallons of water introduced per 1000 cubic feet of and k = 3 r p d ; hence e = dapp,/18pD. The impaction efficiency gas, the specific area, 8,of the drops formed by atornization, in thus can be increased for any given particle size by increasing the square feet per cubic foot of gas, is 245L/D0. Thus the volume gas velocity and decreasing the size of the impacting object. swept by the droplets per cubic foot of gas is For atomization of water in a Venturi throat, thc average 3i$PSc’ V’ = diameter of the droplets may be estimated from an empirical (2) 2 X 245 X 1000 equation proposed by Nukiyama and Tanasawa (11) and conNow if the rate of collection of the aerosol particles is proporfirmed in the extensive studies made by Lewis and coworkers tional to the concentration of the particles and to the volume of (10): gas swept by the droplets per cubic foot of gas, it follows that 3YrPSc’ In-x ___ 1 Present address, Howard Smith Paper Mills, Ltd., Cornwall, Ontario, (3) no 490,000 Canada.
INDUSTRIAL AND ENGINEERING CHEMISTRY
June 1951
BY -PASS
SMOKE CENERATO R
I
TO ATMOSPHERE
WATER TO CVENTURI
n
Y
ing the main air stream. The size was not affected by the velocity of the air stream, within the range used. However, certain precautions were necessary t? assure uniform, reproducible particle sizes. The dibutylphthalate had t o be pure; oil that had previously been subjected to a high temperature proved unsatisfactory. If oil was left in the generator between runs it decomposed owing t o long exposure t o heat. Carbon particles would then appear in the aerosol, and the oil particles themselves would not be uniform. This condition was selfperpetuating, and could be remedied only by burning out the generator with hot air.
7
i
1359
1
ANALYTICAL
WATER OUT
The concentration of the aerosol in BLOWER the inlet gas stream was measured by SAMPLING T R A I N means of filters made from small copper Figure 1. Flow Sheet for Venturi Scrubber thimbles with a piece of copper screen soldered in one end. Glass wool obtained from the Corning Glass Works, with The removal efficiency, E, is expressed by fibers about 5 microns in diameter, was used as the filtering medium. The filters had collection efficiencies of about 95% for 3itPSC' -_ _ the smallest aerosol particles used. Three were used in series. no n 490,000 The two smaller aerosol sizes were sampled at convenient rates, (4) E=-- 1 - e no but isokinetic sampling wa8 necessary for the largest particles. BLOWER
WET T E S T METER
or In (1
- E ) = KijtPS
(5)
If the impaction efficiency and the effective path of the water drops through the gas are constant, a plot of In( 1 - E ) against S should give a straight line. However, it is expected that both the efficiency and the distance the droplets travel during acceleration are functions of the drop size and the velocity of the gas. Deviation from a straight line, therefore, may be expected in some cases.
Accurate sampling of the aerosol leaving the cyclone was possible only a t low water rates. At the high rates, coarse drops of water appeared in the stream from the cyclone, evidently due to blowoff from the end of the exit tube, and reproducible samples could not be obtained, especially a t high rates of gas flow. Accordingly, the collection efficiency was determined from the flow rates and the concentration of oil in the drain water found by extraction with ether. The ether extract was evaporated a t a low temperature and the dibutylphthalate remaining
EXPERIMENTAL
W
a
The Venturi shown in Figure 1 was made according to conventional design from '/*-inch Plexiglas sheet. The inside diameter of the cylindrical sections was 3 * o / inche:, ~ and the convergent and divergent angles were 25 and 7 , respectively. The throat was l a / 1 6 inches in diameter and 1.5 inches long. Water was introduced in three ways: from a single 0.125-inch jet directed downstream along the axis of the throat, radially outward from a brass pipe 0.125 inch in internal diameter with four 0.073inch holes, and radially inward from a single 0.125-inch jet at the side of the throat. In the first two arrangements, the water was injected 0.25 inch downstream from the convergent section. In the third arrangement, the jet was 0.75 inch downstream and flush with the inside surface. The Venturi was preceded by a standard orifice, 2.15 inches in diameter, as the primary metering element. The orifice was preceded by ten diameters of straight pipe, including straightening vanes, and was followed by a straight section seven diameters long. A cyclone was placed after the Venturi to remove the atomized water drops. This was designed to remove completely all droplets smaller than 5.5 microns (9) a t the lowest air flow used. The analytical method used required that the Venturi efficiency be defined as the fraction of the aerosol particles captured by water drops large enough to be removed by the cyclone. However, because the volume ercentage of water drops less than 10 microns is negligible, the egciencies measured should correspond approximately to the true collection efficiency by all of the droplets. The aerosol used was made from dibutylphthalate, an oil having negligible vapor pressure a t room tem erature. The generator consisted of an air heater, a small %enturi atomizer, a vaporizer, and a delay chamber to remove unvaporized oil drops. The mixture was introduced into the main stream through a Transite nozzle without appreciable heat loss. Here rapid condensation caused uniform aerosol droplets to form. A small gear pump, driven by a constant-speed motor controlled with a rheostat, delivered the oil to the throat of the vaporizing Venturi. Two homogeneous aerosols, with droplets 0.8 and 1 micron in diameter, respectively, and a heterogeneous smoke in which the average droplet size was 5.5 microns were produced. The droplet size was controlled by the temperature of the vapor enter-
Figure 2.
Collection Efficiency
Radial outward water injection
INDUSTRIAL AND ENGINEERING CHEMISTRY
1360
Vol. 43, No. 6
COLLECTION EFFlCl ENCY
0.41
0
I
I
1
I
4 8 12 18 S , S P E C I F I C A R E A OF D R O P S S ~ . F T . / C U FT. C A S .
Figure 3.
I
20
I
24
Collection Efficiency
Axial water injection
was dried and weighed. Data taken by this technique checked with those obtained by direct sampling in the range in which the latter were consistent. Aerosol samples were collected on a four-stage high-velocity glass impactor similar to that described by Sonkin ( l b ) . S o drops were ever found on the first two stages of the instrument. When the homogeneous aerosols were sampled, an occasional agglomerate was observed, averaging about three primary droplets, which were often attached to one another like a string of beads. Larger particles occasionally appeared on the third stage, but the mass of agglomerates and large particles was much less than 1% of the total mass of the aerosol collected on the fourth stage. When the larger nonuniform droplets of smoke were sampled, the average particle size on the third stage was greater than that on the fourth. The 5.5-micron average diameter reported for the larger aerosol vas obtained statistically from particle counts under a microscope. The diameter reported is the volume-surface mean diameter ( 4 ) .
I .o
0.9
I
1
The data in Table I show the effect of the size and concentration of the aerosol particles on the collection efficiency when the water injection into the Venturi was radially outward. The efficiency is greater the larger the particle size, and decreases somewhat as the concentration decreases. Table I1 shows the removal efficiencies for 1.0-micron oil droplets when the water was introduced axially and radially inward. Figure 2 shows the data for the higher range of aerosol concentrations in Table I plotted as log (1 - E ) against S, the specific drop surface. The data scatter considerably from the smooth curves for each particle size, since they represent a range of concentrations and throat velocities, both of nhich affect the rollection efficiency. The collection efficiencies for the 1.0-micron aerosol, when axial and radial inward water injection were used, are shown in Figures 3 and 4, respectively. The data are not sufficient t o show the curvature of the logarithmic efficiency plots, but it is apparent that the efficiency is a function of ua as well as S. By collecting the data into groups of approximately constant throat velocity, a separate line results for each range. I n Figure 5 , data values are replotted as log ( 1 - E ) against the product Sup.
Table I .
Effect of Particle Sire and Concentration on Collection Efficiency (Radial outward water injection)
Co
L,
I
nlg./Cu. Foot
FeZSec.
Ga1./1000 Cu. Feet
S, Sq. Feet/ Cu. Foot
Efficiency, Collection
%
O . ~ - ~ I I C R O -K4 E R O S O L
12.4 11.1 7.7 8.4 8.8 10.5 9.1 7.7 5.5 5,3 7.9 7.7 10.5 10.5 9.3 10.1 12.6 13.8 15.1 13.2 13.0 13.9 10.2 11.3 12.8 6.5
7.3
219 255 356 257 270 277
8.79 4.25 1.04 3.91 2.11 1.89
18.9 13.4 5.3 12.6 7.8 7.3
1.0->11CROS A E R O S O L , HIQHC O N C E S T R A T I O N 290 0.37 1.5 350 0.43 2.0 363 1.32 6.5 435 0.35 2.2 427 2.22 12.5 387 4.73 20.0 375 4.08 17.7 375 4.08 17.7 298 1.77 7.2 284 3.23 11.8 261 5.04 l5,4 245 6.45 17.2 226 8.32 18.7 348 1.48 7.2 341 2.75 12.3 325 3.92 15.4 327 3.84 15.3 321 4.96 17.9 308 5.51 18.5 486 0.90 6.2 46 1 1.98 12.2
29.1 16.1 7.2 22.6 18.0 11.6 10.4 14.4 19.0 16.9 35.3 75.3 63.6 57. I 22.1 43.6 47.6 63.4 70.7 23.1 40.7 52.7 49.2 61.9 68.1 29.5 44.2
1 ,0-MICRONAEROSOL,Low CONCESTRATION
2.8 3.2 2.6 2.8 3.0 2.9 2.6
269 209 344 345 328 284 492
6.0 6.1 7.1 9.2 5.6 4.5 4.9
284 271 255 223 391 379 363 346 297 488 286 245 221 398 402
8.1 6.0
S , S P E C I C I C A R E A OF D R O P S SQ.FT./CU. FT. G A S
Figure 4.
Collection Efficiency
Radial water inward Injection
5.2 7.3 9.8 11.6 6.6 4.9
1.95 9.24 1.53 1.57 3.83 7.04 1.10
7.3 18.8 7.4 7.6 15.2 20.0 7.6
10.0 33.0 12.1 11.1 35.0 46.7 13.4
5.5-?dICROX AEROSOL 1.90 7.5 3.31 11.6 4.88 14.8 9.16 19.5 1.35 7.4 2.41 11.7 3.57 15.7 4.80 18.5 6.90 20.5 1.15 7.9 1.73 7.0 . 6.46 17.2 8.91 19.1 2.32 12.2 2.29 12.1
53.6 53.4 68.9 75.8 40.4 79.3 69.0 69.7 83.1 41.5 40.0 61.2 67.4 49.1 45.9
INDUSTRIAL AND ENGINEERING CHEMISTRY
June 1951
Table II.
1361
Effect of Method of Water Injection o n Collection Efficiency (1.0-micron aerosol)
8, Collection L, Efficiency, Ga1./1000 Sq. Feet/ Cu. Feet Cu. Foot % AXIALINJECTION^ 11.2 262 1.37 5.1 8.4 19.3 14.3 249 4.76 7.3 20.1 18.3 224 8.02 8.0 16.4 19.5 213 10.21 8.5 22.9 1.58 7.4 337 7.4 31.8 14.9 322 8.6 3.79 34.0 6.20 19.2 302 7.8 32.7 8.24 21.8 290 9.6 22.1 0.98 6.8 491 7.6 36.3 22.7 7.41 10.5 330 35.3 1.52 10.0 482 7.0 44.8 452 3.31 17.8 6.3 RADIAL INWARD INJECTION 9.4 270 1.35 5.2 16.lb 31.9b 3.38 11.4 9.7 262 25.6b 1.30 6.5 7.8 365 11.8 57.9b 7.6 342 2 64 32.9b 1.07 6.7 6.3 442 69.9b 2.20 12.2 6.3 42 1 7.3 20 7 6.6 268 1.96 6.9 249 4.95 14.6 43.2 18.0 44.6 8.0 233 7.43 20.5 47.8 8.4 210 11.48 28.9 1.53 7.5 5.3 350 7.6 32.3 5.8 347 1.57 15.5 61.6 5.7 325 3.96 6.2 299 6.82 20.4 77.8 6.61 20.0 77.8 6.2 297 7.8 44.0 4.1 443 1.25 a Gas velocities shown are net velocities with respect to li uid leaving jet. b With a single l/a-inch water jet used; all others are wit% l/s-inch jet. C aI
Mg./Cu. Foot
.
u2,
Feet/Seo.
Here, a single straight line is obtained for each method of water injection, indicating that the product +,P in Equation 2 is approximately proportional to the relative velocity of the gas with respect t o the liquid a t the point of injection. PRESSURE LOSS
a
The pressure drop across the Venturi varies with the amount and method of water injection. The observed values are shown in Figure 6, expressed as the number of throat velocity heads lost plotted against the liquid rate L. Because the differential drop across the convergent section of a Venturi can be calculated, the values may be converted readily into over-all pressure loss ( 5 ) . The data for radial injection fall on straight parallel lines with somewhat higher loss for inward injection as compared with outward injection. Because of the momentum transfer from the axial water jet to the gas stream, the pressure loss with this method of injection is appreciably less for high water rates than for other methods of injection.
Figure 5.
sua Correlation of Collection Efficiency D a t a
drop as the result of a glancing collision, a more direct contact may be required for the capture of a hydrophobic particle. The effect of electrostatic charge on the particles upon collection efficiency is being evaluated in a further study of the Venturi scrubber, since it is known that large charges can be built up on spray droplets by atomization. A third factor that should be considered is the difference in performance of large and small Venturis. It has been observed that, for a given liquid-gas ratio, more uniform distribution of the spray may be obtained in large Venturis than in small. Consequently, large scrubbers frequently have higher collection efficiencies than the data from small pilot plants from which they were designed would indicate ( I ) . All the early industrial applications of the Venturi scrubber were for the collection of very small particles, mostly in the submicron range, which are difficult to remove by other means. Recently the equipment has been used on coarser and, consequently, more easily recovered aerosols. These aerosols have been collected with high efficiency a t low pressure losses, showing that a
CORRELATION O F DATA F R O M I N D U S T R I A L INSTALLATIONS
b
The collection efficiencies obtained in the small experimental scrubber used in this work are much lower than those usually obtained in industrial installations. Owing t o the logarithmic nature of the efficiency curves, data in the low ranges were necessary t o determine the effect of the variables studied. High efficiencies were not a prerequisite or a primary objective of the work. Xevertheless, because the efficiencies obtained were lower than those of industrial units operating under comparable conditions, a comment is warranted. I n this research the aerosol had a small particle size and entered the Venturi as primary particles. Industrial installations, however, have handled material of equal or smaller size, so that particle diameter alone cannot be the explanation for the difference. Physical characteristics of the aerosol not previously considered may play an important part. One of these may be the insolubility of the oil used and its hydrophobic nature. I n practice, oil aerosols are generally more difficult to collect with water scrubbers than aerosols of solid particles of the same size. Whereas a soluble and hydrophilic particle may attach itself t o a water
L , L I Q U I D RATE, GALilOOO CU. F E E T G A S
Figure 6.
Pressure Loss i n Venturi Scrubbers
I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
1362
Table 1 1 1 .
Performance of Venturi Scrubbers on Industrial Waste Gases Particle .$ize, icrons
Source Sulfuric acid plant
Aerosol His04 mist (synthetic)
Roaster gases
HzSOa mist
Chlorosulfonic acid plant Phosphoric acid plant Wood distillation plant Superphosphate den and mixer Silicon arc furnace
HzSOa mist HiPOa mist Tar and acetic acid Mist of fluorine compounds SiOz fume
Reverberatory lead furnace Blast furnace Steel open hearth (oxygen lanced) Blast furnace (iron)
Vol. 43, No. 6
L Gal./i000 Cu. Feet
Inches of Water
171 171 216 216 246 246 320
4.3 7.2 3.4 5.7 3.0 7.0 3.9
9.7 14.5 11.9 15.4 18.8 25.7 29.3
UZ 3
.....
Lead compounds Iron oxide Iron oxide Iron ore and coke dust
Throat Velocity
... ... ...
..... . . I . .
.....
o.oiL0:35
I
..*
...
... ...
...
267
3.9 4.7
...
411 729 401 161 259 766 192 1080 309
..
I
...
251
0.5-2 0.6-2.0 0.02-0.5 0.520
.
Gas Loading Mg./Cu. Foo't Inlet Outlet 190 0.6 309 1.7
-Z h ,
...
9.8 11.0
...
...
264
1560 65-389 195-1558
14.0
% 99.7 90.4 99.2 99.7 99.1 100.0 99 7 98.9 98 95 98 52.4 86.7 97 97 92-99 99
3.1 2.0 3.5 0.0 0.7 7.8 3.8 58 5.5
.. .. .. ...
...
...
...
Removal Efficiency,
51.8 65-389 0.5-3.2
+
c
The Venturi throat diameter here was 5.38 inches. Water was introduced by six 0.125-inch jets spaced equally around the periphery of the throat and set flush with the inner wall. Combustion gases from a B. and W.-Tomlinson recovery furnace were drawn into the Venturi scrubber after passing through cascade evaporators. The gaees entered the scrubber a t 275" to 300" F, and a t a wet-bulb temperature of about 163' F. The scrubbing liquid was recirculated with a bleed-off of a small amount, of saturated solution equivalent to the amount of salts recovered. Fresh water was fed to the system as make-up. Gas velocities through the Venturi throat varied from 100 to 325 feet a second, and water rates ranged from 3 to 9 gallons per 1000 cubic feet of gas. Inlet aerosol loadings averaged 130 rng. per cubic foot. Collection efficiencies varied with operating conditions with an average good performance of 94% recovery.
PRESSURE
Figure 7.
LOSS,
INCULS
OF
The results obtained with a commercial Venturi scrubber for collecting sulfuric acid mist (8) are shown in Figure 8. The graph covers the range of throat velocities used, but the cuives have been extrapolated with respect to the amount of scrubbing liquid, to show the expected effect of varying the operating conditions on the efficiency and pressure drop. Although the size of the droplets in the mist is not known, they were probably less than 1 micron in diameter, as the incoming gas was not presaturated with mater sprays to enlarge the mist particles, a pmctice that has been found to be beneficial in other plants. In addition to the mist, small amounts of acidic gases, including hydrogen fluoride, were present. The removal efficiencies for these were higher than for the mist, showing that, in general, gas absorption takes place in the scrubber to a higher efficiency than aerosol collection.
WATER
Venturi Scrubber Performance
Collection of salt cake f u m e f r o m kraft m i l l gases
large power consumption is not always required for this type of equipment. Commercial data taken on full scale or large pilot plant equipment operating under widely varying conditions are listed in Table 111. From a practical standpoint, the information of greatest importance is the relation between the pressure loss and the efficiency of collection. As is usual with industrial and pilot plant data, wide ranges of operating conditions and performance are seldom covered in the efficiency studies. The success of the methods proposed for correlating the data on the laboratory unit as shown in Figures 2 , 5 , and 6 suggesis that the same methods can be used t o predict the performance of industrial Venturi scrubber from small scale tests on the efficiency of collection of the same aerosol. Correlation of the data reported by Collins on the pilot plant tests for the recovering kraft mill fume (2) by a plot similar to Figure 2 has already been shown by Johnstone and Roberts (6). Pressure drop data on a large number of Venturi atomizers ranging in throat diameter from 0.25 to 25 inches agree substantially with the line for radial inward water injection in Figure 6. With this information, and using Equation 2 to estimate the specific surface of the drops, the data on several commercial scrubbers have been correlated and extrapolated. 01 0 2 Figure 7 is an extrapolation of the data from the pilot plant installation for recovering kraft mill fume, which was largely salt cake and soda ash.
d
I
I
1
1
I
I
I
I
6
a
I
4
10
I Z
I4
16
ta
PO
PRESSURE
LOSS,
Figure 8.
INCHES
OF
WATER
Venturi Scrubber Performance
Collection of sulfuric acid m i s t
June 1951
INDUSTRIAL AND ENGINEERING CHEMISTRY CONCLUSIONS
’
*
From the correlations of the data presented here, i t appears that a given collection efficiency can be obtained with the expenditure of less energy by using low throat velocities and high water rates than by using high throat velocities and low water rates. This generalization cannot be extrapolated beyond the range shown on the graphs, as a t very low throat velocities unsatisfactory atomization of the liquid and distribution of the spray in the gas stream are to be expected. Furthermore, the differences are marked only in the lower efficiency ranges, which are not important in a commercial installation. For the range of interest in plant installations-that is, 90% collection and higher-the lines of constant throat velocity tend to merge and the differences at very high efficiencies are small. Consequently, efficiencies in the higher ranges can be plotted directly against pressure loss for all practical purposes. These correlations also show that for highest efficiencies, high throat velocities must be used. The trend in several recent installations, where nearly complete collection of the aerosol is desired, has been to design for gas velocities between 300 and 400 feet per second through the throat. ACKNOWLEDGMENT
This work is a part of a n investigation on treatment of stack gases a t the Engineering Experiment Station of the University of Illinois. The authors wish to express their appreciation t o A. W. Anthony, Jr., for data on the performance of industrial installations of the Venturi scrubber and for helpful suggestions. NOMENCLATURE
C,
= =
d D
=
E
=
c’
=
Ah = AH =
upstream aerosol concentration, mg. per cubic foot conversion factor, cubic microns per gallon, 3.78 X l O I 5 diameter of aerosol particle diameter of object of impaction; subscript o refers t o volume-surface mean diameter of spray drops, microns efficiency of collection of aerosol, fraction pressure loss across Venturi, inches of water Venturi differential across convergent section, inches of water
1363
I C = Stokes’ law resistance coefficient
K L
= constant in Equation 5 = liquid-gas ratio, gallons per 1000 cubic feet
m
= mass of aerosol particle
n
= number of particles per cubic foot of gas; subscript o rep-
P
= length of effective path of water drop
u
=
V’
=
resents initial concentration
s = specific surface of drops formed in scrubber, -
qt
=
e
=
f i P
= =
PP
=
square feet per cubic foot of gas gas velocity; subscript 2 refers to gas velocity at Venturi throat, feet per second volume swept out by droplets per cubic foot of gas, - . dimensionless target efficiency of deposition of particles mu k ~dimensionless , quantity viscosity, poises gas density, pounds per cubic foot particle density, grams per cc. LITERATURE CITED
(1) Anthony, A. W., Jr., private communication, December 1950. (2) Collins, T. T., Jr., Paper Ind. and P a p e r W o r l d , 28, No. 5, 685; No. 6, 830; No. 7, 984 (1947). (3) Collins, T. T., Jr., Seaborne, C. R., and Anthony, A. W., Jr., P a p e r Trade J., No. 23, 45-9 (June 5 , 1947). (4) DallaValle, J. M., “Micromeritics,” 2nd ed., p. 47, New York, Pitman Publishing Corp., 1948.
(5) Dodge, B. F., “Chemical Engineering Thermodynamics,” p. 324, New York, McGraw-Hill Book Co., 1944. (6) Johnstone, H. F., and Roberts, M. H., IND.ENG.CHEM.,41, 2417 (1949). (7) Jones, W. P., Ibid., 41, 2424 (1949). (8) Jones, W. P., and Anthony, A. W., Jr., U. S. Technical Conference on Air Pollution, May 3 to 5, 1950, Washington, D. C. (9) Lapple, C. E., “Chemical Engineers’ Handbook,” Perry, J. H., ed., 3rd ed., p. 1026, New York, McGraw-Hill Book Co., 1950. (10) Lewis, H. C., Edwards, D. G., Goglia, M. J., Rice, R. I., and Smith, L. W., IND.ENQ.CKEM.,40, 67 (1948). (11) Nukiyama, S., and Tanasawa, Y . , Trans. Soe. Mech. Engrs. ( J a p a n ) , 5, No. 18, 68 (1939). (12) Sonkin, L. S., J. I n d . Hug. Toxicol., 28, 269 (1946). RECEIVED January 29, 1951. Previous papers in this series have appeared in IND. ENQ.CHEM.,27, 587, 659 (1935); 29, 286, 1396 (1937); 30, 101 (1938); 32, 1037 (1940); 39, 808 (1947); 41, 2403, 2417 (1849).
Performance of Wet Cell Washers for Aerosols
n
MELVIN W. FIRST, RALPH MOSCHELLA, LESLIE SILVERMAN, A N D EDWARD BERLY HARVARP SCHOOL OF PUBLIC HEALTH.
F
OR many years wetted beds of coarse fiber materials, notably glass, have been used as humidifiers in air-conditioning service. Air washers of this type have been considered of doubtful value for the removal of fine particles from air ( < 2 microns). Little evidence of a quantitative nature is available and there is a tendency to employ wet cell washer units primarily for air cleaning rather than humidification; in some instances these have been employed for air-cleaning tasks of a n exacting nature. The present study was undertaken to determine the range of usefulness of wet cell washers as air- and gas-cleaning devices and to evaluate quantitatively their performance with a number of aerosols under variations of air velocity, cell fiber diameter, and density of fiber packing. THEORY
Filtration.
Filtration of particles from aerosols is accomplished
EOSTON 1 5 , MASS.
by several mechanisms depending on their size and character. Particles greater than 2 microns are captured by interception, settling, and inertial forces-that is, impaction. Those between 1 and 2 microns are principally removed by streamline contact (interception) with the fibers. Particles smaller than 1 micron are effectively filtered by diffusional and electrostatic forces. These separating forces operate simultaneously during filtration of disperse dusts so that performance should be determined over a wide range of particle sizes. Fiber diameter is an important filter characteristic. The efficiency of impaction for spherical particles of uniform density is a function of the particle diameter squared and of the reciprocal of the impacting object size, Impaction efficiency has been shown by Sell and Albrecht, reported b y Johnstone and Roberts (9), to be a function of the dimensionless group
