See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/254292464

Venturi Scrubber Performance

Article  in  Journal of the Air Pollution Control Association · July 1972

DOI: 10.1080/00022470.1972.10469674

CITATIONS READS

42 316

3 authors, including:

Dilip Mehta
Synergia Life Science, Mumbai, India
27 PUBLICATIONS   291 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Reverse pharmacology for natural drug discovery View project

Water pollution View project

All content following this page was uploaded by Dilip Mehta on 08 December 2014.

The user has requested enhancement of the downloaded file.

 Venturi Scrubber Performance
Seymour Calvert, Dale Lundgren, and Dilip S. Mehta
University of California at Riverside

Nomenclature The object of this program was to study particle collection efficiency in venturi
0" Cunningham correction factor scrubbers in order to account for the effects of particle size and wettability, ven-
c Concentration of particles in gas
(lb/ft3 or g/cm 3 ); ct- = inlet; c/ = turi size, and fluid flow rates. The body of information is directed to those in-
outlet terested in the prediction of particle collection efficiency in venturi scrubbers.
Vr Experimental determinations of particle collection efficiency were made for 1
/ Velocity ratio =
in., 2 in. diameter (throat) Venturis and for a 4 in. X 12 in. rectangular cross-
fa An empirical factor defined as the fac- sectioned venturi of 5000 CFM capacity. The test aerosols were spray-dried
tor " / " defined at "atomization ve-
locity" uranine alone, uranine mixed with boric acid to reduce hygroscopicity, and
methylene blue. Particle diameters ranged from 0.5 to 4.6 microns. Calvert's
K Inertial impaction parameter =
equation for particle collection efficiency in atomizing scrubbers was refined to
2C'VrPpRP2
9 M Re include a continuous integration with point values of collection efficiency. Data on
experimentally determined collection efficiency and a computed correlative
L Ratio of liquid to gas flow rates
(gal/1000 ft3) = (gal/MCF) parameter "f" are presented in tables and plots. The results of this research in-
Q Volumetric flow rate (ft3/sec or m 3 /- clude the following:
sec), Qw = liquid; Qa = gas
R Radius (cm or /an); Rc = drop; Rp 1. The design equation is more rigorous after modification and can be solved
= particle with a slide rule, without need for additional charts.
Nile Reynolds number (dimensionless)
Vg Gas velocity relative to duct (ft/sec) 2. Particle wettability is very important and collection efficiency is higher for hy-
Vr Drop velocity relative to gas (ft/sec drophilic than hydrophobic particles.
or cm/sec)
p Density (lb/ft3) or (g/cm3); PL = 3. At low water to air ratios (i.e., below 2 gal/MCF) there is an increase in the
liquid; pp = particle effectiveness of water.
n Viscosity (poises or lb/sec-ft)
x Drag coefficient (fraction) 4. Data from the literature are utilized along with the present results to yield
7? Collection efficiency of a single drop recommended values of "f" for design purposes.
r)a Efficiency, "rj," evaluated at "atom-
ization velocity" 5. It was not possible to obtain reliable data from the 1 in. dia. venturi after
many dozens of meticulous experiments.

Venturi scrubbers are used extensively for theoretical analysis are almost al- with well defined particle properties and
for the collection of small particles and ways lacking. larger capacity than prior studies. Two
less frequently for mass transfer from round and one rectangular cross-section
gas streams. This paper reports an ex- A mechanistic theory describing col- Venturis were studied, but only the data
perimental study of particle collection lection of particles in a venturi type obtained on the 2 in. (throat) diameter
and an analysis of the data in terms of scrubber has been developed by Cal- round and the 4 in. X 12 in. (throat)
theory. Useful data on venturi per- vert. 1 It involves an empirical constant rectangular Venturis are presented.
formance which can be used for develop- which requires experimental evaluation. The data from a 1 in. diameter venturi
ing a general relationship are very scarce. As first presented, it utilized the data of were not used because of discrepancies
Data given by a number of investigators Ekman,2 Brink,3 and Yocum4 to eval- in the material balances for aerosol and
are of limited value because they are on uate the constant. The present study other inconsistencies. Our opinion is
an overall performance basis and im- extended over a two year period and was that it is very difficult to get good data
portant items of information necessary undertaken to obtain data on systems on such a small venturi.

July 1972 Volume 22, No. 7 529

Particle Collection Mechanism Downstream-j r-Pressure Downstream
In a venturi scrubber, liquid is atom- sampling J / point Throat sampling
point point
ized by high velocity air at the entrance
to the throat section. Particles from
air are collected by water drops. Water
is normally removed from air by a cy- Water Exhaust
clone downstream from the venturi Pressure in ection
i
throat. point Pressure
point
Calvert described particle collection
in a venturi as follows:
dc Qw 4RC Venturi-water
injection points
= /Vvdf (1)
•—Aerosol
generator
Equation (1) is modified for particle Figure 1. Sketch of small venturi apparatus.
collection by water drops, based on the
following assumptions:
We obtain: scale unit as shown in Figure 1, with a
1. Particles are collected only by 2 in. dia X 1 in. long throat with a 12 in.
atomized liquid, and the average liquid long diffuser to a 4 in. dia duct. The
drop diameter is as given by the em- (9)
other was a 4 in. X 12 in. (throat)
pirical correlation of Nukiyama and rectangular venturi with a capacity of
Tanasawa.5 and:
5000 CFM. The large venturi had a 4
Rc _ ri6,ooo | K,f in. long throat and a 32 in. long diffuser
to a 12 in. X 12 in. duct. (See Figure
2.) These two systems were used to
Substituting equation (10) in (1) gives: study the effect of various parameters.
c/ Particles were generated by either a
dc 4 QJtepry^ spinning disc atomizer or a 12-hole
2. The collection of particles by single X Collision-type air jet atomizer. While
drops is inertial and based on the rela-
tive velocity between the drop and the
air. The collection efficiency of the
pr / T df (11) the spinning disc (S.D.) provided close
to monodisperse particles, the Collision
J, \_K,f + 0.7 J atomizer (C.A.) provided the high load-
drop is approximated by Equation (3). Integration of Equation (11) yields: ing required where the analytical method
was not sensitive enough, although the
(3) size distribution produced was wider.
'" [KT^J Ci \ ObQ ) The collection efficiency was measured
by withdrawing a gas sample by vacuum
Where K is the inertial impaction pa-
rameter: Where: pump through a filter and a dry gas
meter. The concentration of the aero-
K= (4)
f(K,,f) = i sol was determined by relating the mass
of particles on the filter with the air flow
through the filter. Methylene blue was
3. The acceleration of liquid drops is 1AinM_±°l+ 0.49 extracted from thefilterand analyzed by
0.7 0.7 + K2f_
calculated from a linear approximation a Turner Fluorometer. Size distribu-
of Ingebo's data6 on drag coefficient for Equation (12) was used for the inter- tion was determined by Anderson
accelerating drops. pretation of the experimental data. sampler. The results were confirmed
by microscope count and are accurate
55 (5)
Experimental within ±10%. Densities were deter-
x= In the present investigation two ven- mined by liquid displacement and
turi systems were used. One was a lab- weighing of the bulk materials.
Assuming efficiency varies linearly with
f.
Va _fa -25'6'
(6)
V ~f Inlet
filters
The use of Equation (6) in (1) re- Upstream
sulted in the following Equation: sampling
point
13,500L + 1.2Ls-% X
J
Aerosol
(*f) X 10-" (7) inlet

Rather than using the assumption of

Equation (6), Equation (7) has been
modified to account for the point-by-
point variation of collection efficiency
with relative velocity.
Utilizing the definition: -H h-13"

Vr = fVg (8) Figure 2. Plan sketch of large venturi apparatus.

530 Journal of the Air Pollution Control Association

Table I. Experimental conditions. duced. These small particles were still Table III. Data for large venturi.
A. 2 in. dia throat. present at the downstream sampling L
Throat Water point, when the venturi was operated part. ft/sec gal/MCF E% f
air vel. Rate dia Particle without the introduction of test aerosol.
(ft/sec) gal/MCF (Mm) Mat. Gen. 200 2.0 40 0.46
Similar tests showed that the B&U 200 2.7 51 0.47
50-127 1.0-34.0 0.55 SD particles did not grow in size in passing E 200 5.0 65 0.40
TO <D
1.0 SD through the scrubber. However, the ef- ••u.E
150 1.0 9 0.32
2.0 SD ficiency study with this test aerosol in- E TO 150 2.0 31 0.46
4.6 SD 150 3.0 39 0.44
dicated that the particles might be grow- o 100 2.0 8 0.31
B. 4 in. X 12 in.throat ing in size. This problem was resolved I—1
100 3.0 15 0.36
100-250 1.0-8.0 1. 0 U SD by looking at the particles with a 100 4.0 19 0.35
0. 75 U CA microscope while exposing them to 250 1.2 44 0.72
0. 75 B&U CA various relative humidities. It was
0. 8 MB CA 250 1.7 59 0.78
found that B&U particles did grow in 250 3.0 72 0.69
3 250 4.3 79 0.64
Note:Densities are U = 1.3 g/cm3, B&U = 1.4,
size with increasing humidity, but that «8 200 1.5 30 0.62
MB = 1.25. they returned to the original size as 200 2.3 47 0.68
conditions were reversed with dry air. m
E 200 3.7 57 0.59
Methylene blue (M.B.) particles did not TO 200 5.2 68 0.56
Table I I . Data for large venturi. change in size with humidity. TJ
150 2.0 22 0.58
E
a. 150 3.0 36 0.62
Residence time of the particles in the in 150 5.0 45 0.54
Part. ft/sec gal/MCF E% venturi, between the two sampling 150 7.0 45 0.50
250 1.2 41 0.68
points, is a fraction of a second. If the 100 3.0 4 0.4
<D
250 1.7 53 test aerosol particles grow rapidly as the 100 4.5 8 0.43
c
0.66 100 7.5 17 0.47
'E 250 3.0 68 0.62 air contacts and atomizes the liquid in
2 250 4.2 75 0.56 the venturi throat, the collection effi-
200 1.4 26 0.52
E 200 2.2 36 0.51 ciency will be affected and will be higher
Table IV. Data for small venturi.
.2 200 3.7 50 0.50 than if the particles had retained their (All Uranine).
"D
200 5.2 58 0.32 shape and size.
3. 150 2.0 9 0.35 Experimentally determined data on dp L
in 150 3.0 25 0.47 ft/sec gal/MCF
o 150 5.0 45 0.52 collection efficiency are given in Tables
150 7.0 46 0.47 II, III, and IV. Efficiency of collection 0.55 50 9 0.69
for 1.0 yum U is higher than that for 0.75 0.55 50 15 0.73
<U 250 1.2 40 0.72 0.55 50 23 0.97
3
250 1.8 43 0.52 Aim U. Particle collection efficiency for 0.55 50 28 0.99
m 250 3.0 55 0.47 0.75 Aim B&U is slightly higher than that 0.55 50 34 1.2
6 250 4.2 62 0.43 for pure U particles of the same size. 1.0 50 3 0.33
150 2.0 16 0.41 Collection efficiency for 0.75 /*m MB 1.0 50 9 0.43
E 150 3.0 25 0.43 1.0 50 15 0.57
TO
150 5.0 31 0.40 particles is illustrated in Figure 3 and is 1.0 50 23 0.67
"O
150 7.0 44 0.42 lower than either pure U or the B&U 1.0 50 28 0.72
c 100 3.0 11 0.47 particles of the same size. This is con- 1.0 50 34 0.77
a.
oq 100 4.5 15 0.46 sistent with the observation that the 2.0 50 3 0.29
o 100 8.0 20 0.43 2.0 50 9
MB particles do not change in size with 0.31
2.0 50 15 0.37
humidity. 2.0 50 23 0.44
Equation (12) was used to interpret 2.0 50 28 0.43
the efficiency results in terms of /. If 2.0 50 34 0.44
Experimental Results 2.0 100 1.5 0.48
all the assumptions listed in the deriva- 2.0 100 4.6 0.32
The experimental data were taken in tion of equation (12) were correct, the 2.0 100 7.6 0.43
the two systems under conditions as value of / would be the same for a par- 2.0 100 11.0 0.44
given in Table I. All aerosols were ticular air and liquid rate and would 2.0 100 17.0 0.38
made with either a spinning disc gen- 2.0 127 1.2 0.74
represent the effects of several factors 2.0 127 2.8 0.63
•erator or a two-fluid atomizer, by atom- such as when atomization occurs, how 2.0 127 6.0 0.65
izing a solution and then drying the mist well the drops are distributed, liquid 2.0 127 8.4 0.56
to form solid particles. The first runs loss to walls, and others. This equation 2.0 127 11.0 0.50
were made with uranine particles (U) in 2.0 127 13.4 0.44
does not account for the growth of par- 4.6 100 2.2 0.30
the expectation that they would remain ticles. Computed values of / are given 4.6 100 7.6 0.29
dry during the period of passage through in Tables II, III, and IV. An illustra- 4.6 100 10.8 0.28
the venturi. tive plot of / vs L for 0.75 ^m MB is 4.6 100 14.0 0.24
When it was suspected that the 4.6 100 17.0 0.22
given in Figure 4: Representative
uranine particles were absorbing water values of / computed for Ekman and
and increasing in size a mixture of 90% Johnstone's data2 are plotted against
boric acid and 10% uranine (B&U) was liquid to gas ratio in Figure 5. The it is the only condition for which / in-
substituted. Tests of particle growth Ekman and Johnstone data are for 0.8, creases with L. At such low air velocity
in passing through the 4 X 12 in. ven- 1.0, and 5.5 jum dia particles of di- the prediction of water drop size is very
turi and the entrainment separating octyl-phthalate (D.O.P.). uncertain. All systems show a de-
cyclone were made by sampling the out- crease in / as particle size increases.
let aerosol from the cyclone with a cas- Discussion The influence of air velocity is less con-
cade impactor. Because the inlet aero- Inspection of the data shows that sistent and is most pronounced for the
sol is nearly monodisperse, any change there is a considerable spread of the U and the B&U at low liquid rates,
in size can be readily detected. computed values of / and that there are where / increases with increasing air
Uranine particles grew considerably trends of variation with particle size, velocity. When the liquid to gas ratio
in passing through the scrubber. An gas velocity, and liquid rate. The most is less than about 2 gal/MCF, there is
unexpected finding was a larger number variation is shown for the small venturi an increase in / as L decreases. This
of smaller particles than those intro- at very low (50 ft/sec) air velocity and latter effect may be due to better uti-

July 1972 Volume 22, No. 7 531

 1.0 Conclusions
0.8 ^m dia Methylene Blue 4 " x l 2 " venturi
Carefully taken experimental data on
several particle materials show a dis-
!250fps
tinct influence of wettability and hygro-
scopicity. Interpretation of these and
other data in terms of a modification of
f 0.5 Calvert's equation correlates most of the
r=rO variation in an empirical constant, /.
Z~inr_r--
150fps However, there remain some influences
of particle size andfluidflowrates which
create uncertainty in the accuracy of
predictions based upon this method.
Based upon our data and those cited
0.0
0 1 2 3 4 5 6
in the discussion above,-we recommend
for design purposes that the / values
L (gal/MCF)
used with equation (12) should be 0.45
Figure 3. Collection efficiency for methylene blue. for hydrophilic particles and 0.2 for hy-
drophobic particles when L is greater
100 r
than 5 gal/MCF. When L is lower
0.8 pm dia Methylene Blue 4" x 12" venturi than about 2 gal/MCF a higher value
of/ should be used, as shown in Figure 5.
Where hygroscopic particles are con-
cerned, there is greater uncertainty as to
their effective particle size and the con-
servative approach is to use the / value
£50 150 fps for methylene blue, which will give a low
prediction for collection efficiency.
Despite the use of meticulous experi-
mental techniques and discarding the
results of many runs because of incon-
sistencies in the material balance, there
0 are still undefined variations in the
0 1 2 3 4 5 6 7 data. We conclude that one must be
L (gal/MCF) very careful in using experimental re-
sults, especially those taken on small
Figure 4. Computed "f" from methylene blue experiments. apparatus and with water soluble test
aerosols. Further research is needed to
1.0 identify and define quantitatively any
0.9 presently unknown relationships be-
tween / and system variables, including
0.8 venturi size.
0.7
Acknowledgment
0.6
This work was supported by a re-
f0.5 search grant from the National Air
0.4 Pollution Control Administration (EPA).
0.3 References
0.2 1. S. Calvert, J. Amer. Inst. Chem. Eng.,
0.1 16: 392(1970).
2. F. O. Ekman and H. F. Johnstone, Ind.
0.0 Eng. Chem., 43: 1358 (1951).
0 3 4 5 3. J. Brink and C. E. Constant, Ind. Eng.
L (gal/MCF) Chem. 50: 1157(1958).
4. J. E. Yocum and Chapman, Air Repair
Figure 5. Comparison of "f" for several systems. 4: 154(1955).
5. S. Nukiyama and Y. Tanasawa, Trans.
Soc. Mech. Eng. (Tokyo) 4: 86 (1938).
6. R. Ingebo, NASA Tech. Note 3762
(1956).

lization of the water, perhaps because of particle size upon which this value is
lower wall losses and of less agglomera- based ranged from 0.5 /*m to 1.5 /xm dia
tion of drops. (density about 1.65 g/cm3).
The Ekman and Johnstone data2 yield The general trends and approximate Dr. Calvert's present address is
Applied Science, University of Cali-
distinctly lower values of / than either magnitudes of computed / values for our fornia at Riverside, Calif. 92502.
our data or those of Brink and Con- data and Ekman and Johnstone's as Mr. Lundgren is now at the Uni-
stant.3 Computation of / for Brink and shown in Figure 5 show a great de- versity of Minnesota and Dr. Mehta
Constant's data on the collection of 80% pendence upon particle material. The is in India. The work for this paper
was done at U. C. R. The paper
orthophosphoric acid mist in a 6 X 34 in. variation due to all factors is greatest at was presented as Paper No. 70-8 at
venturi at a gas velocity of 218 ft/sec the low liquid rates and levels out when the 63rd Annual Meeting of APCA
and L about 11 gal/MCF yields an L approaches the usual operating range, in St. Louis in 1970.
average value of about 0.45 for /. The around 10 gal/MCF.

532 Journal of the Air Pollution Control Association

View publication stats