EPA-600/2-78-032
February 1978
Environmental Protection Technology Series
EVALUATION
OF THREE INDUSTRIAL
PARTICULATE SCRUBBERS
Industrial Environmental Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
-------
EPA-600/2-78-032
February 1978
EVALUATION OF THREE
INDUSTRIAL PARTICULATE SCRUBBERS
by
Seymour Calvert, Harry F. Barbarika,
and Gary M. Monahan
Air Pollution Technology, Inc.
4901 Morena Boulevard, Suite 402
San Diego, California 92117
Contract No. 68-02-1869
Program Element No. 1AB012
ROAP No. 21ADM-029
EPA Project Officer: Dale L Harmon
Industrial Environmental Research Laboratory
Office of Energy, Minerals, and Industry
Research Triangle Park, N.C. 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, D.C. 20460
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series pre:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Soct'oeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the ENVIRONMENTAL. PROTECTION TECH-
NOLOGY series. This series describes research performed to develop and dem-
onstrate instrumentation, equipment, and methodology to repair or prevent en-
vironmental degradation from point and non-point sources of pollution. This work
provides the new or improved technology required for the control and treatment
of pollution sources to meet environmental quality standards,
EPA BEVIEW NOTICE
This report has been reviewed by the U.S. Environmental Protection Agency, and
approved for publication. Approval does not signify that the contents necessarily
reflect the views and policy of the Agency, nor does mention of trade names or
commercial products constitute endorsement or recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
ABSTRACT
Field measurements on three full scale industrial scrubbers
were carried out to determine scrubber performance characteris-
tics, including the particle collection efficiency as a function
of particle diameter. The three scrubbers were different gas-
atomized spray types with pressure drops ranging from 54 to 178
cm W.C. Their performance on major sources of fine particle
emissions was compared to a mathematical performance model for
venturi scrubbers.
111
-------
CONTENTS
• * •
Abstract .......... , ............. • • 11:L
Figures ......... .................. v
Tables ............. i ............. vii
Abbreviations and Symbols. ... .............. i-x
Acknowledgment ....................... X1
1. Introduction
2. Conclusions
3. National Dust Collector Model 850 Variable Rod
Module Venturi .................... 4
4. American Air Filter Kinpactor Venturi ........ 16
5. Gas-Atomized Spray Scrubber ............. 31
6. Performance Comparison ................ 47
7. Performance Test Method ............... 64
8. Data Reduction and Computation Method ........ 74
References .................. ....... 85
IV
-------
FIGURES
Number Page
1 Schematic flow diagram of scrubbing system 5
2 Schematic of scrubber tower. 6
3 Penetration versus aerodynamic particle diameter for
ductile operation 12
4 Penetration versus aerodynamic particle diameter for
gray operation 12
5 Schematic diagram of scrubbing system 17
6 Particle penetration for run 2 24
7 Particle penetration for runs 4 and 5 24
8 Particle penetration for runs 6, 7 and 8 25
9 Particle penetration for runs 9, 10 and 11 25
10 Particle penetrations for runs 12 and 13 26
11 Schematic drawing of scrubber system ........ 32
12 Particle penetration for cascade impactor runs 1
and 2 41
13 Particle penetration for cascade impactor runs 3
and 4 41
14 Particle penetration for cascade impactor runs 5,
6 and 7 and diffusion battery runs 3-9 (inlet) and
10-12 (outlet) 42
15 Particle penetration for cascade impactor run 8. . . 42
16 Particle penetration for cascade impactor runs 9
and 10 43
17 Particle penetration for cascade impactor runs 11,
12 and 13, and diffusion battery runs 15 and 16
(inlet) and 13 and 14 (outlet) 43
-------
FIGURES (continued)
Number Pa&e
18 Particle penetration for cascade impactor runs 14
28
and 15,
44
19 Comparison of predicted and measured penetration for
the NDC venturi during gray iron operation 51
20 Comparison of predicted and measured penetration for
the NDC venturi during ductile iron operation 51
21 Comparison of predicted and measured penetration for
the average of runs 12 and 13 of the AAF venturi. ... 54
22 Comparison of predicted with measured penetration for
average of runs 5, 6 and 7, for the gas-atomized
scrubber 59
23 Comparison of predicted with measured penetration for
run 8 for the gas-atomized scrubber 59
24 Comparison of predicted with measured penetration for
average of runs 9 and 10 for the gas-atomized scrubber. 60
25 Comparison of predicted with measured penetration for
average of runs 11, 12 and 13 for the gas-atomized
scrubber 60
26 Comparison of predicted with measured penetration for
average of runs 14 and 15 for the gas-atomized
scrubber. 61
27 Modified EPA sampling train with in-stack cascade
impactor 67
Schematic diagram of diffusion battery system 69
VI
-------
TABLES
Number Page
1 Process conditions 8
2 Inlet and outlet size distribution summary using log-
probability analysis 10
3 Diffusion battery size distribution analysis 11
4 Overall penetration summary 13
5 Opacity summary 14
6 Cost data 15
7 Inlet process conditions 19
8 Outlet process conditions 19
9 Outlet average gas composition 20
10 Particle size distribution summary 22
11 Mass loading and overall penetration 27
12 Opacity 28
13 Scrubber conditions 34
14 Process conditions 34
15 Average gas composition 36
16 Cascade impactor data using log-probability analysis. 38
17 Diffusion battery particle size distributions . ... 38
18 Mass loading and overall penetration 40
19 Conditions for variable rod venturi performance
predictions 50
20 Conditions for AAF venturi performance prediction . . 53
VII
-------
TABLES (continued)
Number Page
21 Pressure drop comparison for the gas-atomized spray
scrubber b8
22 Conditions for gas-atomized spray scrubber perfor-
mance prediction 5°
23 Overall penetration comparison for the gas-atomized
spray scrubber "2
24 Measuring equipment and methods 65
Vlll
-------
LIST OF ABBREVIATIONS AND SYMBOLS
A,B = parameters in equation (21) or equation (29),
dimension!ess
b = Weibull slope, dimensionless
B = venturi parameter (equation 5), dimensionless
Cf = Cunningham slip correction factor (equation 9),
dimensionless
CDF = cumulative distribution function, dimensionless
C~ = drag coefficient at throat inlet, dimensionless
C = cumulative particle mass concentration, mg/DNm3
C t- = total inlet particle mass concentration, mg/DNm3
C = total outlet particle mass concentration, mg/DNm3
d" = differential operator
d = wire diameter of screen in screen diffusion battery, cm
w
d, = drop diameter, cm or ym
d. = jet diameter, cm
d = particle diameter, cm
P
d = particle aerodynamic diameter, ymA
pa
d = impactor stage cut diameter, ymA
pciC
d = diffusion battery stage cut diameter, cm
d = particle geometric mass mean aerodynamic diameter, ymA
Jr o
d = particle mass median aerodynamic diameter, ymA
d = particle geometric number (count) mean diameter, ym
d = Weibull minimum particle diameter, ymA
f(d ) = particle frequency distribution, dimensionless
K = inertial parameter equation (6) or equation (7),
P dimensionless
£. = venturi throat length, cm
m = mass of particles over a differential size element, g
N = cumulative number concentration of particles, #/cm3
P = cumulative mass fraction (equation 13), dimensionless
Pt = particle penetration, fraction
Ptf = overall penetration, fraction or %
Qr = gas volume flow rate, cm3/s
QT = liquid volume flow rate, cm3/s
Li
ix
-------
LIST OF SYMBOLS AND ABBREVIATIONS (continued)
S = solidity factor of screen diffusion battery stage,
dimensionless
u, * = ratio of velocity attained by the liquid drops at the
throat exit to the gas velocity, dimensionless
uGt = ^as velocity i-n throat, cm/s
u- = jet velocity, cm/s
u = superficial gas velocity, cm/s
j
x = length parameter (equation 3), dimensionless
x = parameter defined in equation (14), dimensionless
x,y = variables in equation (21), dimensionless
GREEK
AP = pressure drop, cm W.C.
0 = Weibull characteristic particle diameter, ymA
PQ = gas viscosity, poise (g/cm-s)
ymA = aerodynamic micrometers (equation 7)
p = particle density, g/cm3
PG = gas density, g/cm3
PT = liquid density, g/cm3
CT = geometric standard deviation, dimensionless
o
-------
ACKNOWLEDGEMENT
A.P.T., Inc. wishes to express its appreciation to Mr. Dale
L. Harmon, the EPA project officer, for excellent coordination
and assistance in support of our technical effort. The cooper-
ation provided by the plant personnel at the sites tested is
also greatly appreciated.
XI
-------
SECTION 1
INTRODUCTION
Air Pollution Technology, Inc. (A.P.T.) conducted performance
evaluations on three industrial particulate scrubbers under EPA
Contract No. 68-02-1869. The objective of the performance evalu-
ations was to determine fine particle penetration as a function
of particle size and scrubber parameters.
The project involved the following tasks:
1. Locate suitable scrubbers for evaluation and select
those which:
A. Have potential for high collection efficiency of
fine particles,
B. Control a process which is a major source of fine
particulate emissions,
C. Are widely used in industry.
2. Conduct appropriate field test programs to obtain neces-
sary performance data.
3. Use data obtained to evaluate the performance and pro-
bable economics of the scrubber.
4. Develop a useful performance model of the scrubber system,
The first task involved an extensive correspondence campaign.
Many leads were taken from listings in the National Emissions Data
System (NEDS) and contacts with state and local air pollution con-
trol districts. The result of this task was the identification of
the three scrubbers whose evaluations are presented in this report,
A separate detailed report has been issued for each of the
scrubbers tested (Calvert, et al. 1976, Calvert, et al. 1977a>
and 1977b). In this report each scrubber is discussed in a sepa-
rate section and comparison with a performance model is made in a
succeeding section. Also included in this report are discussions
of the testing and the data reduction methods that were used.
-------
SECTION 2
CONCLUSIONS
Field measurements on three full scale industrial scrubbers
were conducted to determine scrubber performance characteristics,
including the particle collection efficiency as a function of
particle diameter. A summary of the performance tests is given
in the tabulation below:
Pressure Drop
Control Device Source cm W.C.
Variable Rod Venturi Iron Cupola 178
(National Dust Collector)
Venturi Borax Fusing Furnace 110
(American Air Filter)
Gas-atomized Spray Iron Cupola 54-104
Scrubber
All three scrubbers were in the general classification of
gas atomized spray scrubbers and their performance was compared
to the mathematical model of Yung, et al. (1977a).
The program achieved the primary objective of evaluating
full scale industrial scrubbers, however some difficulty was en-
countered in finding suitable and willing facilities to evaluate.
The following conclusions regarding venturi performance are
based on the results of our evaluations.
1. Comparison of the measured particle collection efficiency
to the mathematical model led to the following conclusions:
(a) The comparison was good in the aerodynamic particle
diameter region around 1 ymA.
(b) In all cases the model predicted lower efficiency for
particle diameters less than 0.5 umA. To some extent
this is expected because the model considers only in-
ertial impaction as the collection mechanism.
-------
(c) The comparison was not always good for particle dia-
meters greater than 2 \irnK. The model prediction in
that size region was quite sensitive to the liquid-
to-gas ratio, which usually was not known with pre-
cision. Thus the model may have been adequate in
the range above 2 ymA, however knowledge of important
variables was not precise and the adequacy of the
model could not be fully proven.
(d) Predictions of overall efficiency were usually poor.
The model, which assumed a log-normal size distri-
bution, consistently underpredicted efficiency. Two
probable reasons are that the inlet distribution had
a geometric mass mean diameter greater than was mea-
sured with the heated cascade impactors and the size
distribution was not log-normal.
Generally the model and the measured performance
agreed. However, each source and scrubber had parti-
cular characteristics which requireed that some ana-
lysis be made to reconcile the model with the data.
Therefore, the model should only be used with a full
awareness of the assumptions on which it is based.
The model may need modification if used in situations
where conditions are significantly different from the
assumptions in the model.
2. High pressure drop scrubbers are vulnerable to leaks which
decrease the efficiency and waste power. Leaks generally develop
because of improper construction materials and/or inadequate main-
tenance.
3. The cyclone entrainment separators used on the venturi
and the gas-atomized spray scrubber were inefficient and caused
both emissions problems and blower maintenance problems.
-------
SECTION 3
NATIONAL DUST COLLECTOR MODEL 850
VARIABLE ROD MODULE VENTURI
SOURCE AND CONTROL SYSTEM
The emission source was an iron cupola which operated for
ten hours per day. The cupola melted both the relatively clean
ductile iron and the dirtier gray iron.
The primary component of the scrubbing system was a National
Dust Collector Model 850 variable rod module venturi scrubber.
The nominal capacity was 10.4 m3/s at 21°C (22,000 CFM at 70°F).
The variable rod module could be operated at pressure drops from
100 to 230 cm W.C. (40 to 90 inches W.C.). The venturi module
consisted primarily of several parallel rods positioned normal
to the gas flow. The gas was moved through the scrubber by two
induced draft Buffalo No. 91 blowers used in series. Each blower,
run by a 261 kW (350 HP) motor, was capable of drawing 114 cm
W.C. (45 inches W.C.).
The overall scrubbing system is shown in Figure 1 . Gas
flow from the cupola is controlled by adjusting the combustion
(tuyere) air flow. The charge door is open so air continually
leaks into the cupola. The temperature at the top of the cupola
is only about 175-205°C instead of the much higher temperature
encountered with most cupolas.
A series of quench sprays are introduced at the entrance of
the connecting duct between the cupola tower and the scrubber.
About 5.8 £/s of fresh water is used in the quench section. The
temperature of the gas is reduced to about 35°C by the quench
sprays. The gas is saturated when it reaches the scrubber.
As shown in Figure 2 , the scrubber consists of the follow-
ing sections:
1. A pre-spray section which sprays scrubbing liquor into
the gas stream at the rate of 7.8 £/s. This pre-spray has slight
4
-------
TO
ATMOSPHERE
t
S
•H
e
O I
to r^
rj LO
STACK
OUTLET
SAMPLING
PORT
t
VARIABLE
ROD
MODULE
VENTURI
SCRUBBER
V
c
•H
--. u
m o
S i/>
< to
o
o
QUENCH
WATER
350£/min
16°C
INLET
SAMPLIN
I PORT
CUPOLA
TUYERE AIR
I.D. BLOWERS
o
CN1
LO
z
AIR
Figure 1. Schematic flow diagram of scrubbing system.
-------
GAS OUTLET
18-21°C
COOLING WATER
SPRAYS
COOLING
SEPARATOR DECK
COLLECTOR DECK
VARIABLE ROD
MODULE
•.•.•.is!/-:- •«••. •.:•.•.-.•/.v ••/*%» v.VJ'Wid
DECKS ••.••:•.•.•••/••• :•;«%•..-..•.•.:.•. .•."••..•;•:
COOLING WATER
RETURN
SCRUBBER LIQUOR
RETURN
SCRUBBER LIQUOR
SPRAY
PRE-SCRUBBER
LIQUOR SPRAY
GAS INLET
1,010 Am3/min
35 C
QUENCH/SCRUBBER LIQUOR
RETURN
Figure 2. Schematic of scrubber tower.
6
-------
cooling effect, which may cause some condensation growth of the
particles, and it may collect some particles as it settles out
onto the bottom of the scrubber tower. The drops from these
sprays are not carried through to the variable rod module. The
liquor returned from this section is sent to the sludge tank.
2. The high energy section, which consists of a scrubber
liquor spray of 28.3 £/s and the variable rod module is next.
The pressure drop across the variable rod module is about 178
cm W.C. when the total pressure drop across the whole tower is
224 cm W.C. The temperature of the scrubber liquor is about 35°C
and the ratio of liquid to gas volume flow rates in this section
is about 1.68 £/m3.
3. The collector deck, consisting of 7.6 cm of loose 2.5 cm
diameter marbles, separates out the entrained liquid from the
high energy section. The returned scrubbing liquor is split
into streams, with 801 going to the recycle tank and 20% going to
the sludge tank.
4. Above the collector deck is a plate which collects the
water from the cooling section.
5. The gas is cooled prior to leaving the scrubber tower
in a cooling section consisting of sprays and two decks of single
layer 2.5 cm diameter marbles. The cooling water is supplied at
about 15°C at the rate of 112 £/s to cool the gas to about
18-21°C. There is no carryover of entrainment from the cooling
section.
-------
PROCESS CONDITIONS
The operating conditions of the variable rod venturi scrubber
for the period of sampling are shown below:
TABLE I. PROCESS CONDITIONS
CONDITION
INLET SAMPLE PORT
OUTLET SAMPLE PORT (1)
Temperature
Velocity
Am3/s
ACFM
DNm3/s @ 0°C
DSCFM § 21°C
Vol % H20 Vapor
Static Pressure
32-35°C
9.2 m/s (30.3 fps)
16.8
35,700
12.9
27,300
6.6
-1.1 cm W.C.
57-66°C
25.3 m/s (83.1 fps)
20.5
43,500
14.5
30,800
3.3
+0.38 cm W.C.
Pressure Drop Across Tower
Pressure Drop Across Venturi Module
Superficial velocity Through
Scrubber Tower (diameter 3.12 m)
based on inlet ACFM
224 cm W.C
178 cm W.C,
2,19 m/s
(1) The outlet sample port was located in the exit stack
after the straightening vane section which was used to reduce
the tangential spin of the gas. For this reason outlet flow
rates may not be as reliable as inlet flow rates.
-------
CASCADE IMPACTOR DATA
Sets of data were obtained from the variable rod module
venturi scrubber as will be described later in the test method
section. Fully developed flow enabled representative one-point
sampling. Identical one-point sample locations were used for all
the data points.
A summary of the inlet and outlet size distribution tests
is given in Table 2. The distributions approached log-normality
for diameters between 0.3 and 2 umA.
Average sample times for the inlet were three to four minutes
depending on the mass loading, while the outlet sample times
averaged approximately forty-five minutes to one hour.
No inlet data were obtained for run 1 due to the excessive
vacuum at the start of the run which overloaded the impactor.
Runs 9, 10, and 11 were not used as part of the data set be-
cause two of the five spray nozzles in the scrubber liquor spray
section were disconnected while these runs were being made.
The data from the inlet cascade impactor runs was analyzed
to determine the effect heating the impactors above the stack
temperature had on the size distributions. We could see no de-
finite trend in the heated impactor runs toward smaller size dis-
tributions than the unheated impactor runs.
DIFFUSION BATTERY DATA
Diffusion battery data were taken during the testing period.
The runs were made alternately on inlet and outlet sample loca-
tions as shown in Table 3, while impactor runs were being per-
formed .
Since operation of the scrubber was fairly constant over the
testing period the inlet and outlet samples were averaged result-
ing in one set of data each for the ductile and gray operations.
-------
TABLE 2 . INLET AND OUTLET SIZE DISTRIBUTION SUMMARY
USING LOG-PROBABILITY ANALYSIS
RUN OPERATION INLET OUTLET
N0- dpg,ymA ag dpg,ymA
1
2
3
4(D
5
6(2)
7(3)
8(4)
9(5)
10
11
12
IS^
14
15(7)
16
16P
17
Ductile
Ductile
Ductile
Ductile
Ductile
Gray
Gray
Gray
Gray
Gray
Gray
Gray
Gray
Gray
Ductile
Ductile
Ductile
Ductile
_ _
0.69
0.76
0.25
0.27
1.10
1.51
0.56
0.54
0.62
1.90
0.62
0.59
0.72
0.84
0.58
0.71
0.49
—
2.0
2.0
1.6
1.5
1.7
1.8
1.6
1.5
1.6
1.6
1.6
1.6
1.6
1.6
1.5
1.7
1.8
0.48
0.52
0.52
0.52
0.56
0.58
0.56
0.52
0.63
0.63
0.81
0.51
0.51
0.52
0.50
0.48
0.48
0.49
1.7
1.8
1.9
1.8
1.8
1.7
1.8
1.6
1.6
1.7
1.7
1.6
1.6
1.5
1.7
1.7
1.7
1.7
(1) Run 4 had 40% of the 5th stage holes plugged on the
inlet sample
(2) Run 6 had 30% of the 3rd stage holes plugged on the
inlet sample
(3) Run 7 had 30% of the 6th stage holes plugged on the
inlet sample
(4) Run 8 had 20% of the 6th stage holes plugged on the
inlet sample
(5) Run 9 had 30% of the 6th stage holes plugged on the
inlet sample
(61 Run 13 had 4 holes of the 5th and 6th stages plugged
on the inlet sample
(7) Run 15 had 50% of the 4th stage holes plugged on the
inlet sample
10
-------
TABLE 3. DIFFUSION BATTERY SIZE DISTRIBUTION ANALYSIS
Operation
Ductile
Gray
Inlet
d ~ ,,™ a
pn,ym g
0.023 3.0
0.040 3.1
d
_e
0.
0.
Outlet
n, ym
070
090
0
_!
2.2
2.0
PARTICLE PENETRATIONS
Particle penetration versus particle aerodynamic diameter
was computed and is shown in Figures 3 and 4 for ductile and gray
iron operations respectively. Penetrations for a few of the runs
are not shown because either their size distributions or their
total loadings were outside a standard deviation from the mean
of the set of data within which the run belonged.
The penetrations were calculated using a mathematical for-
mula based on the log-normality of the inlet and outlet size dis-
tributions .
Diffusion battery data yield penetration related to physical
size while cascade impactor data are in terms of aerodynamic size.
In order to put the results on the same basis, it is necessary
to know the particle density so that one can convert physical
size to aerodynamic size (or vice versa). In Figures 3 and 4
a value of 3 g/cm3 for density.has been used to convert the
physical diameter based on calculated diffusion battery penetra-
tions to penetrations based on aerodynamic particle diameter.
The penetration plots Indicate an increase in efficiency
for particles smaller than 0.2 ymA. These results are consistent
with published literature in that smaller, more highly diffused
particles experience increased collection efficiency.
An overall penetration summary for runs 1 through 17 is pre-
sented in Table 4. Total inlet and outlet mass loadings were
taken by cascade impactors.
11
-------
1.0
0.5
0.1
0.05
I
W
0.01
0.005
0.001
0.1 0.2 0.5 1.0 2.0 5.0
PARTICLE AERODYNAMIC DIAMETER, pmA
Figure 3. Penetration versus aerodynamic particle
diameter for ductile operation.
o.ooi
0.1 0.2 0.5 1.0 2.0 5.0
PARTICLE AERODYNAMIC DIAMETER, umA
Figure 4. Penetration versus aerodynamic particle
diameter for gray operation.
-------
TABLE 4. OVERALL PENETRATION SUMMARY
RUN NO.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
16P
17
OPERATION
(Gray)
(Ductile)
D
D
D
D
D
G
G
G
G
G
G
G
G
G
D
D
D
D
MASS LOADING *
mg/DNm3
INLET
2,050
2,100
1,430
1,810
2,060
1,910
2,230
2,310
1,880
2,100
2,290
2,260
2,320
1,960
1,790
2,250
1,390
OUTLET
23.6
20.9
24.5
28.0
27.0
32.0
30.7
35.8
58.5
39.3
52.9
27.9
24.0
23.0
21.2
24.8
24.8
21.2
OVERALL, PT
%
1.02
1.17
1.96
1.49
1.55
1.61
1.61
2.53
2.09
2.51
1.22
1.06
0.99
1.08
1.39
1.10
1.52
*N = 0°C, 1 atm
13
-------
Although Figures 3 and 4 show penetrations for ductile and
gray iron operations separately, there was no significant dif-
ference in the penetrations between the two operations. Likewise,
the average overall penetration for both the ductile operations
and the gray operation was 1.3%.
OPACITY
Opacity for the outlet stack of the National Dust Collector
Model 850 variable rod module venturi scrubber was taken by em-
ployees trained and certified by the California State Air Quality
Control Section. Readings were made five times a day from 8:00 a.m.
to 4:00 p.m. every two hours.
A summary of the daily average reading is presented below
in Table 5.
TABLE 5. OPACITY SUMMARY
Runs
2,3
4,5
6,7,8
9,10,11
12,13,14
15,16,17
Average
for Test
Average Opacity,
13
11
8
11
25
33
2
13
15
ECONOMICS
Cost data for the initial and annual cost was supplied by
the user. The unit was originally built in 1967 as a fixed
hole orifice venturi and in early 1975 was modified to a variable
rod module type to alleviate plugging problems. Shown on
Table 6 are the cost factors for the unit. F.O.B. shipping
and delivery charges were not available, but should be about
$15,000.
14
-------
TABLE 6 . COST DATA
Scrubber § Auxiliaries
A. Fans, motors and motor starters $136,200
B. Ducting 88,000
C. Liquid and solid handling and treatment 45,000
D. Instrumentation 7,800
E. Electrical material 55,000
Total $332,000
Scrubber Installation Cost
A. Erection $120,600
B. Plumbing 61,900
C. Electrical 36,000
D. Engineering 15,500
Total $234,000
Total Initial Cost (1967 prices) $566,000
Annual Costs include the following factors:
Operating Costs
A. Utilities $ 37,000
B. Labor 20,000
C. Supplies and materials 12,000
D. Treatment and disposal 4,000
Total $ 73,000
Maintenance Costs
A. Labor $ 10,000
B. Materials 6,000
Total $ 16,000
Plant overhead, space, heat,
light, insurance $ _7,000
Total Annual Costs $ 96,000
15
-------
SECTION 4
AMERICAN AIR FILTER KINPACTOR VENTURI
SOURCE AND CONTROL SYSTEM
The emission source was a large borax fusing furnace used in
continuous operation. The furnace was capable of producing 270
metric tons per day of anhydrous borax CNa281,07) from the penta-
hydrated form of the feed, but was not always operating at full
capacity during the testing. The particulates emitted were pri-
marily the hydrated and anhydrous forms of borax which escaped
during the drying and fusing processes.
The total scrubbing system is shown in Figure 5, with the
location of the sampling ports indicated. The gases from the fur-
nace at a temperature of 1,000°C to 1,100°C are quenched by scrub-
bing liquor to about 80°C. The gases then enter the American Air
Filter Kinpactor 10 x 56 venturi scrubber. The venturi is rec-
tangular in cross section with a throat height of 142 cm (56 in-
ches) . The throat width is automatically controlled to maintain
a pressure drop of about 110 cm W.C. (43 in. W.C.) across the
venturi. The throat width can vary from 0.64 cm (0.25 in.) to
25.4 cm (10 in.). Following the venturi the gases enter a cyclone
entrainment separator. The gas is moved through the scrubbing
system by a blower which is rated at 44 Am3/s at 84°C and 114
cm W.C. pressure differential (93,500 ACFM, 184°F, 45 in. W.C.).
The blower is powered by a 746 kW (1,000 HP) motor. The gases
exhaust to the atmosphere through a 21 m tall, 2.13 m diameter
stack.
The scrubbing liquor is recycled through a tank which is
fed about 28 l/s of fresh liquor or less err—amounts of fresh
water. About 22 H/s of concentrated liquor is pumped from the
bottom of the tank so that the borax concentration of the liquor
is maintained at from 10 to 15%.
16
-------
TO
ATMOSPHERE
FROM
FURNACE
>
1,000°C
LIQUOR
41 i/s
(MAX.)
80°C
20 m3/s
TO LIQUOR
RECYCLE 41 fc/s
TANK (MAX.)
TO
RECYCLE
TANK
BLOWER
STACK
70 C
25 m3/s
Figure 5, Schematic diagram of scrubbing system.
-------
PROCESS CONDITIONS
The gas conditions at the inlet and outlet sampling locations
for each run are presented in Tables 7 and 8. The conditions
during runs 9 through 13 were thought to be most representative
of normal conditions because of the amount and consistency of
the product from the fusing furnace during these runs. The
average barometric pressure during the testing period was 93.63
kN/m2 (27.65 in. Hg) .
Close examination of the inlet and outlet flow rates indi-
cates a discrepancy in the measured flow rates. The outlet flow
rate is greater than the inlet flow rate while the inlet temper-
ature is greater than the outlet temperature. The static pressures
are practically equal. There are two explanations for this flow
rate discrepancy: (1) As noted previously, the inlet flow is not
well developed and thus the inlet flow rate may not be accurate
and, (2) air could be leaking into the system because of the low
pressure (-110 cm W.C. gage) within the system between the verituri
and the blower.
The results on the Orsat analysis of the outlet gas are pre-
sented in Table 9.
CASCADE IMPACTOR DATA
Particle size distribution data were obtained for the Ameri-
can Air Filter Kinpactor venturi scrubber as will be described in
the test method section. Identical single point sampling at the
average velocity location was performed at both the inlet and out-
let. The sampling time of each run depended on the mass loading.
The average sampling times for the inlet and outlet were nine and
fifty minutes respectively.
Because of the number of very large particles in the inlet
gas and the entrained water droplets at the inlet and outlet, pre-
cutters were used. The aerodynamic cut diameters of the inlet and
outlet precutters were approximately 12 ymA and 4.5 ymA respective-
ly. The inlet sampling was approximately isokinetic. However,
because of the large amount of water droplets in the outlet gas,
a special "rain can" had to be used on the outlet sampling nozzle.
18
-------
TABLE 7 . INLET PROCESS CONDITIONS
Run
1
2
3,4
6,7
9,10
12,
,5
,8
,11
13
Temp. Water Volume
°C Percent*
75
108
73
78
79
81
36
57
14
23
49
41
Static Press. Flow Rate
cm W.C. ArnVs (ACFM)
-8
-6
-6
-7
-7
-8
.3
.5
.1
.2
.5
.6
24
21
16
18
20
20
(51
(45
(35
(38
(42
(43
,000)
,000)
,000)
,000)
,000)
,000)
TABLE 8 . OUTLET PROCESS CONDITIONS
Run
1
2
3,4
6,7
9,10
12,
,5
,8
,11
13
Temp. Water Volume
°C Percent*
80
73
72
68
71
69
31
33
38
28
23
21
Static Press. Flow Rate
cm W.C. Am3/s (ACFM)
-0
-0
-0
-0
-0
-0
.5
.4
.4
.4
.4
.4
29
25
25
25
25
25
(62
(54
(54
(54
(53
(53
,500)
,000)
,000)
,000)
,000)
,000)
*Based on wet and dry bulb temperatures
19
-------
TABLE 9. OUTLET AVERAGE GAS COMPOSITION
Gas Component
N2
02
C02
CO
H20
Molecular Wt.
Volume Percent
(Wet)
60
6
5
0
29
26
Volume Percent
(Dry)
85
8
7
0
29
20
-------
Also, the outlet sampling nozzle was oriented perpendicular to
the gas flow. The velocity through the outlet nozzle was main-
tained at the velocity of the outlet flow at that location.
To minimize the possibility of condensation in the impactors
and to collect only the dry particles, the impactors were main-
tained at about 15°C above the gas stream temperature by using
heating blankets. The precutters were not heated.
The fact that the impactors were heated should be noted when
interpreting the size distribution data. Some unpublished data
taken with heated and unheated cascade impactors in a stream of
wet borax particles have shown differences in the size distribu-
tions. The primary difference was that the mass median aerody-
namic diameter of the borax particles collected in heated impac-
tors was as small as only 70% of the mass median aerodynamic
diameter of the particles collected in unheated impactors. Thus,
the venturi scrubber may be encountering larger aerodynamic size
particles than the heated cascade impactor data indicate.
Data from run 3 are not included because the outlet cascade
impactor stages had become wet. This happened because an attempt
was made to sample with an outlet nozzle orientation parallel to
the gas stream for this run.
Size Distributions
The inlet and outlet size distributions both seem to be bi-
modal. Thus, no attempt was made to fit all the data points to
a log-normal curve. The size distributions were approximately
log-normal in the region below 2 ymA. The mass mean geometric
diameters and geometric standard deviations for the log-normal
parameters as well as the mass median diameter (from the data
points) runs are presented in Table 10.
Diffusion Battery Data
Diffusion battery data were taken during the fifth and sixth
days of testing. The inlet data were taken during cascade impac-
tor runs 9, 10 and 11, while the outlet data were taken during
runs 12 and 13.
21
-------
TABLE 10. PARTICLE SIZE DISTRIBUTION SUMMARY
Run
1
2
4
5
6
7
8
9
10
11
12
13
pm
ymA
0.80
0.94
0.86
0.92
0.82
0.66
0.86
0.76
0.94
0.75
Inlet
dpg
ymA
0.84
0.99
2.3
0.77
1.9
0.85
0.79
0.55
0.83
0.69
1.2
0.72
ag
2.1
3.3
3.7
2.7
3.3
4.7
3.3
2.7
3.3
2.7
3.3
3.0
pm
ymA
0.40
1.1
0.32
0.33
0.23
--
0.25
Outlet
Sg
ymA
0.55
0.78
0.32
0.22
0.41
0.20
0.22
0.13
0.16
0.25
0.16
0.19
ag
2.0
3.2
2.6
2.9
2.1
2.8
3.4
3.6
3.5
5.2
3.6
3.5
Note: d = mass median aerodynamic particle diameter
p from data
d = log-normal geometric mass mean aerodynamic
P8 particle diameter
0 = log-normal geometric standard deviation
22
-------
Because of the large amount of water vapor (40% by volume)
in the gas, the lenses in the condensation nuclei counter of the
diffusion battery would fog when the diffusion battery was opera-
ted in the normal manner. Increased dilution did not solve the
problem because the particle count would then drop below the
threshold of the counter. Also, heating the diffusion battery
to its maximum allowable temperature in an effort to reduce con-
densation on the lenses did not help.
The system was modified to allow data to be taken by routing
the incoming source gas through a glass flask before entering the
diffusion battery. Enough of the water vapor condensed in the
flask so that the condensation nuclei counter did not become in-
operable. Any condensation of water vapor would cause collection
of submicron particles by diffusiophoresis. Thus, a fraction of
the particles did not reach the diffusion battery. For this rea-
son the diffusion battery data may not be accurate.
However, since the configurations for taking the outlet data
and the inlet data were the same and the process was constant
during the testing period, the data still give an indication of
the relationship between the distribution of submicron particles
at the inlet and outlet. The data show that the outlet submicron
particles are larger and more monodisperse. This is an indication
that particle growth by condensation may be occurring in the scrub-
ber .
The data are only qualitative because condensation also oc-
curred within the measuring system. However, mechanisms for par-
ticle growth are present in the scrubber system. These mechanisms
are discussed later.
PARTICLE PENETRATIONS
Particle penetration versus particle aerodynamic diameter
was calculated from the cascade impactor data. The results are
shown in Figures 6 through 10 for each day of testing. Pene-
trations for run 1 are not shown because they are much larger
than the penetrations for all the other runs, indicating anomalous
23
-------
ts)
O
I—I
E-i
U
O
I—I
H
W
W
0.05
0.01
0.005
0.001
0.3 0.5
PARTICLE AERODYNAMIC DIAMETER, d ,ymA
pa
Figure 6. Particle penetration for run 2,
O
i—i
H
O
i—i
H
w
0.05
SS o.oi
0.005
0.001
0.3 0.5 1 23
PARTICLE AERODYNAMIC DIAMETER,d ,
' pa'
Figure 7. Particle penetration for runs
4 and 5.
-------
ts)
tn
o
I—I
H
i
PL,
A
I—1
H
H
W
0.05
0.01
0.005
0.001
0.3 0.5 1 23
PARTICLE AERODYNAMIC DIAMETER,d ,ymA
pa
Figure 8. Particle penetration for runs
6, 7, and 8.
0.05
O
H
§
O
I—i
H
W
0.01
o.oos £EE:
o.ooi
0.3 0.5 1 23
PARTICLE AERODYNAMIC DIAMETER,d umA
pa*
Figure 9. Particle penetration for runs
9, 10 and 11.
-------
2
O
CJ
2
p.,
«*
O
I—I
H
2
H
w
0.1
0.05
0.01
0.005
0.001
0.3 0.5 1 23
PARTICLE AERODYNAMIC DIAMETER, d
pa
Figure 10. Particle penetrations for runs 12 and 13.
26
-------
behavior. Penetrations for runs from the last two days, when
operations were most smooth, are quite consistent.
Because the size distributions were not log-normal all of
the penetrations were calculated manually. Particle penetration
based on diffusion battery data is not presented because of the
inaccuracies incurred during the data acquisition. Diffusion
battery data are discussed in the previous section.
The total mass loadings and overall penetrations for the
runs are presented in Table 11. The total mass loading was deter-
mined from analysis of the cascade impactor data. Run 1 had ano-
malously high overall penetration and high flow rates which indi-
cated that something may have been wrong with the data or that
the venturi was not operating properly during that run. The
average penetration for all of the runs, exclusive of run 1, was
2.5%.
TABLE 11. MASS LOADING AND OVERALL PENETRATION
Run
1
2
4
5
6
7
8
9
10
11
12
13
Inlet Mass Loading
mg/DNm3
332
653
1,150
841
690
2,310
882
829
1,040
863
1,210
1,000
Outlet Mass Loading
mg/DNm3
40.4
18.3
42.1
11.0
21.9
18.4
20.5
22.7
19.3
31.0
36.9
27.3
Penetration
1
12.2
2.8
3.7
1.3
3.2
0.8
2.3
2.7
1.9
3.6
3.0
2.7
27
-------
OPACITY
Opacity for the outlet stack of the American7Air Filter
Kinpactor 10 x 56 venturi scrubber was determined by personnel
trained and certified by the California Air Resources Board.
Readings were made hourly during the testing period. The opa-
city determinations were made somewhat difficult by the presence
of steam condensation in the plume and the proximity of other
stacks emitting similar particulates.
Table 12 presents the daily average opacity readings.
TABLE 12. OPACITY
Date
Runs
Average Opacity,
Average Outlet
Loading, mg/m
8/19/76
8/20/76
8/21/76
8/22/76
8/23/76
2
3-5
6-8
9-11
12,13
10-15
15
20
15-20
20
9
12
11
14
19
28
-------
ECONOMICS
Data for the initial costs of the venturi-cyclone scrubbing
system, purchased in 1970, were supplied by the user:
Approximate scrubber purchase cost $200,000
Scrubber auxiliaries:
1. Fans, motors, etc. 30,000
2. Ducting 47,000
3. Liquid and solid handling and
treatment 50,000
4. Instrumentation 45,000
5. Electrical material 36,000
Scrubber installation:
1. Site preparation 108,000
2. Installation 100,000
3. Startup and modification 51,000
4. Engineering 63,000
Total Initial Cost - $730,000
The operating costs were not available. However, the power
costs can be estimated. The major power user is the large blower,
The blower is powered by a 746 kW (1,000 HP) motor. At $0.03
per kW-hour the fan power would cost $22 per hour, or $537 per
day.
29
-------
OPERATING PROBLEMS
The primary operating problems with the system are the plug-
ging of nozzles -et&d piping and scale build-up in the system.
These problems are all caused by calcium carbonate and sulfate
deposits. The local water is very hard and the feed to the fusing
furnaces may also contain these mineral impurities. To combat
this problem special reamer nozzles have to be used and the system
has to be shut down periodically to chip away the built-up scale.
Although a large amount of entrained water was carried-over
from the entrainment separator through the fan, and out the stack,
it was not considered a problem by operating personnel.
30
-------
SECTION 5
GAS-ATOMIZED SPRAY SCRUBBER
SOURCE AND CONTROL SYSTEM
The emission source for this test was a no. 7 gray iron
cupola used for an 8 to 9 hour/day operation. The cupola nor-
mally is run for 3 days/week, but during the test period it
was run for as many days as possible (3 days in the first week,
5 days in the second week). The operation produces 10 to 12
tons per hour of metal from a feed of scrap metal, coke, and
limestone.
During the testing period the cupola ran at normal condi-
tions. The particulate emissions result from the melting of the
scrap and burning of the coke in the cupola. Figure 11 is a
schematic diagram of the entire operation including the scrubber
system.
The hot gases from the melting operation enter the after-
burner where it reaches approximately 700 - 1,000°C. The gas
is then quenched by city water, which during the test had a
flow rate varying between 0 and 2.6 £/s (4.1 GPM) giving a
temperature range of 700°C to 60°C. Scrubber liquor is added
just ahead of the throat section (the rate was varied during the
testing period from 2.5 £/s to 15.8 £/s). Next the gas enters
a variable throat scrubber section which is a rectangularly cross-
sectioned orifice with a throat height of 91.4 cm (36 inches).
The throat width can be varied from 4.2 cm (1.7 in.) to 11.4 cm
(4.5 in.). Following the venturi the gas enters a cyclone type
entrainment separator. An induced draft blower moves the gas
through the scrubbing system, powered by a 298 kW (400 HP) motor.
From this blower the gas is exhausted through an 11.9 m (39 ft)
high, 0.91 m (3 ft) diameter stack.
At the venturi section, scrubber liquor is pumped in through
pipes placed directly upstream of the throat. This liquor
31
-------
WET CAP
CHARGING
DOOR
CM
to
QUENCH
SPRAYS
SCRUBBING
LIQUID
C
ENTRAINMENT
SEPARATOR
CUPOLA
AFTERBURNER QUENCHER
SCRUBBER SUMP
t
STACK
Figure 11. Schematic drawing of scrubber system.
-------
begins as clean water every day from a sump below the cyclone.
No fresh water feed is added to this liquor so that as the day
progresses the solids concentration increases. When the system
is shut down the sump is neutralized, cleaned out, and fresh
water pumped in for the next day of operation.
The water used for the quench sprays, however, comes directly
from the city water lines and no recirculation is used.
PROCESS CONDITIONS
The scrubber process conditions for the inlet and the outlet
are shown in Tables 13 and 14 . Runs 1 through 4 represent the
scrubber running with a number of leaks, as can be seen by com-
paring the inlet and outlet flow rates. Between runs 4 and 5
many of the large leaks were patched, resulting in much smoother
operation as well as a reduction of the fan amperage, an increase
in the inlet flow rate, and reduction in the outlet flow rate.
After patching most of the leaks, more control over the
scrubber operation could be maintained. It was then possible
to adjust water flow rates as well as the pressure drop across
the venturi. This enabled a number of test conditions to be
examined. For the most part, the charging operation was fairly
constant and did not vary significantly between the runs, except
for run 1 when the crane broke and only one charge was added
during the entire run. The average barometric pressure during
the testing period was 99.36 kPa (29.34 in. Hg).
The top of the cupola was open, except for a "wet cap."
The wet cap consists of baffles and sprays at the top of the
cupola and it was operated during the test period. Depending
upon the scrubber conditions, varying quantities of visible
emission were observed flowing from the wet cap.
The tests were conducted during the winter so that the am-
bient temperature averaged about 0°C during the testing. This
low ambient temperature probably caused the temperature of the
gases entering the scrubber to be lower than normal because of
heat transfer through duct walls, leakage of cold air, and use
of cooler than normal quencher and scrubber water.
33
-------
TABLE 17,. SCRUBBER CONDITIONS
TABLE 14. PROCESS CONDITIONS
Run
No.
1
2
3
4
5
6
1-B
7
8
2-B
9
10
11
12
13
14
15
Pi
*
-1.
-1.
-3.
-2.
-6.
-6.
-6.
-6.
-3.
-3.
-5.
-4.
-2.
-2.
-2.
-5.
-5.
8
3
0
0
4
4
6
1
8
6
1
3
3
5
8
8
8
P2
*
- 77
- 81
- 57
- 64
- 83
- 85
- 87
- 87
- 86
- 85
- 87
- 85
-105
-105
-107
- 94
- 92
AP
*
.5 75
.3 80
.2 54
.8 62
.8 76
.1 78
.6 81
.6 81
.4 82
.1 81
.6 82
.1 80
.0 103
.0 104
.0 104
.0 88
.7 86
.7
.0
.2
.8
.4
.7
.0
.5
.6
.5
.5
.8
.0
.0
.0
.2
.9
Throat
Width
(cm)
11
4
5
4
11
11
11
11
S
5
5
5
4
4
4
11
11
.4
.4
.6
.4
.4
.4
.4
.4
.1
.1
.6
.6
.2
.3
.3
.4
.4
Scrubber
Liq.Flow
(Vs)
3
11
< 2
< 2
12
12
12
12
9
9
9
9
12
12
12
15
15
.2
.4
.5
.5
.6
.6
.6
.6
.5
.5
.5
.5
.6
.6
.6
.8
.8
Quench
Flow
O/s)
1
0
1
1
1
1
1
1
1
1
0
0
2
2
2
1
1
.1
.0
.6
.6
.6
.6
.6
.6
.6
.6
.0
.0
.6
.5
.5
.7
.7
Pj = static pressure before venturi
P2 = static pressure after venturi
AP = Pi- P2
* = numbers expressed in cm W.C.
Run
No.
1
2
3
4
5
6
1-B
7
8
2-B
9
10
11
12
13
14
15
Inlet
Temp.
°C
246
496
132
132
177
149
149
177
177
177
699
677
68
60
71
66
177
Outlet
Temp.
°C
56
56
57
57
65
65
65
54
54
54
58
58
51
51
51
51
51
Inlet Outlet Inlet
Water Water Flow Rate
Vapor Vapor Am /s
Vol.% Vol.% (MACFM)
**
21
11(6*)
27
27
25
26
26
25
25
25
2*
2*
27
20
32
25
25
7
7
17
17
20
20
20
5
S
5
S
5
8
8
8
11
11
5.2(11.0)
6.2(13.2)
6.5(13.7)
6.5(13.7)
11.4(24.1)
11.0(23.4)
11.0(23.4)
10.8(22.9)
10.8(22.9)
10.8(22.9)
12.6(26.7)
12.5(26.4)
4.1( 8.7)
4.0( 8.4)
4.2( 8.8)
9.1(19.2)
10.4(22.2)
Outlet
Flow Rate
Am /s
(MACFM)
** A
11.7(24.8)
11.7(24.8)
12.5(26.5)
12.5(26.5)
10.5(22.2)
10.5(22.2)
10.5(22.2)
10.0(21.1)
10.0(21.1)
10.0(21.1)
8.1(17.1)
8.1(17.1)
7.0(14.8)
7.0(14.8)
7.0(14.8)
9.8(20.7)
9.8(20.7)
* Based on sampling train water catch
** Based on wet and dry bulb temperatures
*** MACFM = thousand actual cu. ft/min
-------
The cyclone separator was not operating efficiently during
the testing period. It was patched along with the rest of the
system after run 4 but it could not be sealed completely as it
had been repeatedly patched previously. Any leakage into the
cyclone would disrupt the flow of liquid down the walls and
would cause reentrainment. Entrainment emission was clearly
noticeable as rainout from the stack. The entrainment emission
was due to several factors such as leakage, poor flow distribu-
tion at the inlet, and internal roughness due to scale buildup.
The Orsat analyses of the outlet gases are presented in
Table 15.
CASCADE IMPACTOR DATA
Particle size distribution data were obtained for the gas-
atomized spray scrubber as described in the test method section.
Identical single point sampling at the average velocity location
was performed at both the inlet and outlet. The sampling time
for each run depended upon the mass loading. The average sampling
times for the inlet and outlet, respectively, were 6.5 and 31
minutes.
The inlet sample ports, because of a limited length of
ducting after quencher and before the venturi throat, were loca-
ted 0.73 equivalent duct diameters downstream of a 78° bend and
0.73 equivalent duct diameters from the beginning of the venturi
section. The inlet duct was rectangular (0.91 m x 0.61 m) and
its equivalent duct diameter was 0.73 m as calculated by EPA
Method 1.
The outlet sample ports were located in a 0.91 m (3 ft)
diameter round stack, about 8 stack diameters downstream of the
inlet from the fan and 3 diameters upstream of the stack air.
The velocity traverses indicated a well-developed flow pattern.
Entrained water drops were a problem at both the inlet and
outlet, so precutters for the inlet and outlet were approximately
11.6 ymA and 5.6 ymA, respectively. Both the inlet and outlet
sampling were approximately isokinetic.
To minimize the possibility of condensation in the outlet
35
-------
TABLE 15. AVERAGE GAS COMPOSITION
Gas
Component
N2
02
C02
CO
Molecular Wt.
Inlet
Volume,!
Dry
80.1
6.4
13.2
0.3
29.9
Outlet
Volume, 1
Dry
(Runs 1-4)
80.1
16.9
3.0
0
28.8
Outlet
Volume,!
Dry
(Runs 5-15)
79.6
14.6
5.8
0
29.2
36
-------
impactors and to collect dry particles, the impactors (except
run 9 which was below the stack temperature due to a malfunction
of the heating blanket) -were maintained at about 13.5°C above the
gas stream temperature using heating blankets. Because of the
temperature fluctuation at the inlet the impactor was kept out of
the stack and heated by a heating-blanket to an average tempera-
ture of 86°C. In both cases (inlet and outlet) the precutters
were not heated.
The fact that the impactors were heated should be remembered
when interpreting the size distribution data. Because of the
presence of a quencher upstream of the scrubber section, the
particles may be wet due to condensed or absorbed water when
they enter the scrubber. The wet particles would have a different
size distribution than those collected in a heated impactor. The
wet particle size distribution would probably have a larger
mass median diameter. The runs with the most favorable con-
ditions for wet particles were runs 11-14.
Table 16 summarizes the "d " and "a " results for the
to o
cascade impactor runs using log-probability parameters. The
distribution data include only the results of the cascade impac-
tor stage analysis. The precutter and probe weight gains were
not used. The log-normality approximation was adequate only for
diameters between 0.2 and 2 ymA.
DIFFUSION BATTERY DATA
Diffusion battery data were taken during cascade impactor
runs as shown in Table 17.
A dilution system was used for this testing along with the
A.P.T. diffusion battery. A 208 liter (55 gallon) drum was used
upstream of the diffusion battery as the dilution system.
Samples were not taken continuously from the stack because
the intermittent charging operation can result in rapidly chan-
ging readings. A syringe was inserted into the stack several
times to withdraw gas samples and these were expelled into the
dilution tank. Later this system was switched to a calibrated
hand pump. These gas samples were diluted with dry filtered air
37
-------
TABLE 16. CASCADE IMPACTOR DATA USING
LOG-PROBABILITY ANALYSIS
Run No .
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Inlet
dpg
(ymA)
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.47
.45
.40
.26
.19
.33
.39
.45
.58
.37
.40
.41
.31
.20
.18
Inlet
°g
2
2
2
2
3
2
2
2
2
2
2
2
2
4
4
.6
.9
.1
.7
.4
.0
.0
.0
.0
.7
.4
.4
.7
.4
.4
Outlet
dpg
(umA)
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.86
.76
.38
.58
.36
.43
.41
.39
.49
.35
.53
.43
.41
.43
.49
Outlet
°g
1
4
2
2
2
2
2
2
2
2
2
2
2
2
2
.9
.0
.3
.3
.3
.2
.2
.5
.3
.7
.0
.7
.4
.4
.2
oo
TABLE 17. DIFFUSION BATTERY PARTICLE
SIZE DISTRIBUTIONS
i
D.B.
Run
No.
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Impactor
Run No.
5
5
6
1-B
8
8
2-B
2-B*
2-B*
2-B*
11
11
12
13
Type
of
Run
In
In
In
In
In
In
In
Out
Out
Out
Out
Out
In
In
V
Cum)
0.070
0.092
0.095
0.128
0.072
0.095
0.094
0.057
0.061
0.085
0.090
0.090
0.086
0.059
a
g
3.4
12.0
12.0
14. '0
6.0
2.9
11.5
1.0
2.5
4.5
6.0
8.9
9.3
6.0
Total
Particle
Count**
No. /cm3
9.0 x 106
1.1 x 107
1.1 x 107
3.9 x 106
2.3 x 106
1.3 x 10s
1.1 x 107
5.1 x 106
8.0 x 106
8.3 x 105
7.8 x 105
1.9 x 106
2.2 x 107
1.5 x 107
*Same condition but not simultaneous
**Gas conditions: 20°C, 1 atm
-------
in the tank and run through the diffusion battery with particle
counts being measured using the condensation nuclei counter.
The particle size distributions determined by diffusion
battery analysis are given in Table 17.
PARTICLE PENETRATIONS
Particle penetrations for various particle aerodynamic dia-
meters were calculated from cascade impactor and diffusion bat-
tery cumulative loading data. The penetrations based on diffusion
battery data were calculated from log-normal inlet and outlet
cumulative distributions. The physical (actual) size distribu-
tions from the diffusion battery analysis were converted to
aerodynamic size by assuming the cupola dust had a particle
density of 2.5 g/cm3.
Since the size distributions were not log-normal, the
cascade impactor penetrations presented in Figures 12 through
18 were calculated graphically.
The total mass loadings and overall penetration for the
runs are presented in Table 18. Mass loading and penetration
are also shown without the outlet precutter catch. The reason
is that the precutter loading was mostly entrainment carryover
from the cyclone and was not due to scrubber inefficiency.
39
-------
TABLE 18. MASS LOADING AND
OVERALL PENETRATION
Run
No.
1
2
3
4
5
6
1-B
7
8
2-B
9
10
11
12
13
14
15
Mass
Inlet
1,570
2,120
1,100
2,060
3,190
2 , 340
7,030
2,530
1,913
1,600
5,070
2,620
1,750
2,370
2,500
1,450
1,240
Loading, mg/DNm *
Outlet i
212
74.3
280
176
609
341
968
234
320
177
198
589
159
214
797
278
277
Outletz
46.3
36.5
267
175
514
307
-
234
98.2
-
156
159
153
159
174
186
223
Penetration, \
Pt1
13.5
3.5
25.5
8.5
19.1
14.6
13.8
9.3
16.7
11.1
3.9
22.5
9.1
9.0
31.8
19.2
22.2
Pt2
2.9
1.7
24.4
8.5
16.1
13.1
-
9.2
5.1
-
3.1
6.1
8.7
6.7
6.9
12.8
17.9
Including outlet precutter dry weight gain
Without outlet precutter dry weight gain
* N = 0°C, 1 atm
40
-------
1.0
0.01
10
PARTICLE AERODYNAMIC DIAMETER, d
.
pa
§
1-1
H
%
1.0
0.5
0.2
0.1
H
w
§ °-05
Cu
0.02
0.01
Figure 12. Particle penetration for cascade impactor
runs 1 and 2.
0.2 0.5 1 2 5 10
PARTICLE AERODYNAMIC DIAMETER, d . ymA
pa
Figure 13. Particle penetration for cascade impactor
runs 3 and 4.
-------
1.0
1.0
0.03
0.05 0.1
0.2
PARTICLE AERODYNAMIC DIAMETER, d
lpa'
Figure 14. Particle penetration for cascade impactor runs
5, 6 and 7 and diffusion battery runs 3-9 (inlet)
and 10-12 (outlet).
0.01
O.OS 041 0.2 0.5 1 2 5
PARTICLE AERODYNAMIC DIAMETER, d , pmA
pa
Figure 15. Particle penetration for cascade impactor run 8.
10
-------
1.0
CM
1.0
0.01
0.2
0.5 1 2 5
PARTICLE AERODYNAMIC DIAMETER, d .
pa
Figure 16. Particle penetration for cascade impac-tor
runs 9 and 10.
0.01
.005
.05 0.1 0.2 0.5 1 2 5 10
PARTICLE AERODYNAMIC DIAMETER, d , ymA
pa
Figure 17. Particle penetration for cascade impactor runs 11,
12 and 13, and diffusion battery runs 15 and 16
(inlet) and 13 and 14 (outlet).
-------
0.01
10
PARTICLE AERODYNAMIC DIAMETER, d , umA
' pa' ^
Figure 18. Particle penetration for cascade impactor
runs 14 and 15.
44
-------
OPERATING PROBLEMS
The primary operating problem with the system at first was
the limiting capacity of the fan. Great amounts of excess air
were getting into the system through all of the leaks, resulting
in the fan running at its maximum capacity. For this reason the
system could not pull enough gas through the scrubber and a large
portion left through the top of the cupola. The amount of water
that could be added to the system was also limited, which in turn
affected the scrubber efficiency.
Once most of these leaks were patched, however, the fan
amperage dropped well below the limit, and the system could be
adjusted easily.
The scrubber liquor was another operational problem. Since
the cyclone entrainment separator was not working effectively, much
of the water which captured the particles was carried over, through
the fan and out the stack. The scrubber liquor was recycled and
changed only at the end of each day which meant that the solids
concentration increased as the day progressed. Reentrainment of
this water, therefore, would mean higher outlet loadings since
this extra weight was considered in the emission rates.
Corrosion had a severe effect on the system operation because
it was the cause of the excessive leakage of air into the scrub-
ber and entrainment separator.
Another problem with the system was the absence of a cap
on the cupola top. This allowed the gases to escape through the
top. Some of the gases do not pass through the scrubber and
therefore are not treated effectively, which in turn caused a
visible emission problem. This was alleviated by capping the
cupola top subsequent to our tests.
ECONOMICS
The scrubber system was built about ten years ago with sub-
stantial participation by the foundry. Consequently, the cost
45
-------
was distributed among several purchased components, contracted
work, and foundry-performed work. Records of the costs were
not kept in one place and it was not possible to obtain precise
data on the sub-system costs.
The operating cost for the blower (300 kW motor] was about
$72.00 per day, based on eight hours use per day.
46
-------
SECTION 6
PERFORMANCE COMPARISON
INTRODUCTION
All three scrubbers evaluated fall within the general classi-
fication of gas atomized spray scrubbers. The AAF venturi is
most like a conventional venturi scrubber. The performance of
gas atomized scrubbers depends on many factors which involve
both the properties of the dust to be collected and the operating
conditions of the scrubber. Yung, et al. (1977a) have modeled
the performance of Venturis assuming collection results from par-
ticle impact ion on the liquid drops atomized in the venturi throat.
The important parameters are particle aerodynamic impaction dia-
meter, liquid drop diameter, amount of liquid, and relative velo-
city between the particles and the liquid. The energy used by a
venturi goes primarily into achievement of high relative velocities
and is manifested in the pressure loss.
PRESSURE LOSS
Yung, et al. (1977b) have derived a relatively simple model
for the pressure loss in a venturi:
AP = 1 x 1(T3 u*e u^ [-iij (1)
where: AP = pressure drop, cm W.C.
ratio of the velocity ;
at the throat exit to the gas velocity, dimensionless
Ul = ratio of the velocity attained by the liquid drops
de
Gt
= gas velocity in throat, cm/s
QT = liquid volume flow rate, cm3/s
Qr = gas volume flow rate, cm3/s
47
-------
This model is applicable to large Venturis in which the predomi-
nant energy loss mechanism is the acceleration of the liquid
drops. The term "u* " is not ususally measured and can be con-
sidered an empirical constant. It has a direct mathematical
relation to the venturi throat length under certain conditions,
however. If the drop acceleration takes place in a constant area
section and the initial drop velocity is insignificant then,
u*e = 2[1 - x2 + (x* - X2)0'5] (2>
3£. C- pr
where: x = 1 + t Do G (3)
16 dd PL
i. = venturi throat length, cm
C,, = drop drag coefficient at throat inlet, dimensionless
PG = gas density, g/cm3
d, = drop diameter, cm
p, = liquid density, g/cm3
For simplicity the drop diameter is taken as the Sauter mean
diameter resulting from gas atomization as predicted by the
Nukiyama-Tanasawa relation.
PARTICLE COLLECTION
Yung, et al. (1977a) found that the assumption that particle
collection occurs primarily in the constant area throat section
is often valid and greatly simplifies the performance prediction.
In the throat section the particle penetration reduces to:
48
-------
In Pt(d )
= . 4 K n * ^ I-5 *
B K fl-ii* 1 + n i PO u~udej + 4.2 (1-u, )°'5
0.7
de-
(4)
where: B =
Kp0 ~ = inertial parameter based on
G d throat velocity, dimensionless
(6)
and, Cr = Cunningham slip factor, dimensionless
d = particle diameter, cm
p = particle density, g/cm3
Ug = gas viscosity, poise
Equation (4) sometimes slightly underestimates the particle
collection occurring in a venturi scrubber. For most industrial
venturi scrubbers, particle collection can be predicted closely by
neglecting the first term in the right hand side of equation (4).
The overall penetration prediction requires the integration
of equation (4) over the inlet dust particle size distribution.
VARIABLE ROD MODULE VENTURI SCRUBBER COMPARISON
The variable rod module venturi is essentially several rec-
tangular jets in parallel.
49
-------
Pressure Drop
The exact geometry of the venturi rod module was not made
available. Thus, throat length and throat gas velocity were not
known, and the measured pressure drop could not be compared to
that predicted by equation (1).
Particle Collection
Since "u*i " was not known from the geometry, penetration pre-
dictions for a range of values were made and are shown in Figures
19 and 20. The conditions used are shown in Table 19. The closest
fit to the data occurred for Uj =0.75 which corresponds to a
throat length of 27 cm and a throat gas velocity of 11,700 cm/s,
using equations (1) to (3).
TABLE 19. CONDITIONS FOR VARIABLE ROD VENTURI
PERFORMANCE PREDICTIONS
AP = 178 cm W.C.
QL/QG = 0.00168
PL = 1 g/cm3
pr = 0.9 kg/m3
yG = 1.89 x 10"
ude =
uGt '
dd '
CDo '
B =
*t =
1.0
10,140
107
0.56
3.32
oo
g/cm-s
0.75
11,710
101
0.55
3.42
27
0.5
14,340 cm/s
93 ym
0.53
3.52
6.3 cm
Based on an approximate angle of free jet divergence of 20°
and a spacing between rods equal to the rod radius, it was es-
timated that the rod diameter was 7.2 cm (2.8 in.), which was
within reason.
The overall penetration was calculated on averaged cascade
impactor data. For gray iron operation (d = 0.91 ymA, a = 1.6)
XT O O
50
-------
O
U
o
l-
w
w
IX
0.1
0.01
0.001
0.1
PARTICLE AERODYNAMIC DIAMETER, pmA
Figure 19. Comparison of predicted and measured penetrations
for the NPC venturi during gray iron operation.
0.1
o
W
W
0.01
0.001
0.1
PARTICLE AERODYNAMIC DIAMETER, ymA
Figure 20. Comparison of predicted and measured penetration
for the NDC venturi during ductile iron operation
-------
a penetration of 4.1% was predicted and for ductile iron operation
(d = 0.57 ymA, a = 1.7) a penetration of 16.2% was predicted.
fT ^? v5
The data indicated the overall penetration was between 1% and 2.51,
which meant that the size distribution used in the predictions
was not correct for sizes larger than about 3 ymA.
Discussion
The model assumes that collection is by inertial impaction
on drops, occurring only in the throat region of the venturi. The
data agree with the model in the range of 0.8 to 2.0 umA diameters
but show greater efficiency below 0.8 ymA. Thus, the model is
fairly close for the larger particles of interest. However, as
particle diameter decreases below 1 ymA diameter, forces other
than inertia, such as flux forces and Brownian motion, become
important.
Diffusiophoresis, and thermophoresis, two of the flux force/
condensation (F/C) mechanisms are effective in regions where the
liquid spray drops are cooler than the gas and the gas is satura-
ted. Some condensation may have caused particle growth between
the inlet sampling port and the variable rod module. Here the pre-
scrubber sprays may have been a few degrees cooler than the gas,
so that enough particle growth could have occurred prior to the
venturi rod module to increase the collection efficiency of the
smaller particles.
F/C effects could have caused the cooling spray drops to
collect fine particles in the cooling section of the tower. The
cooling spray water temperature was about 20°C lower than the gas
temperature. Collection by Brownian diffusion would not be im-
portant in the size regime that was measured in these tests.
Another section of the system where small particle collection
could have occurred was the section containing the two blowers
in series. The high velocity and extreme turbulence in the blow-
ers may have been effective in causing particle collection.
AAF KINPACTOR VENTURI SCRUBBER COMPARISON
The American Air Filter venturi is a classical venturi with
a rectangular throat. The throat width is automatically controlled
52
-------
to maintain a pressure drop of about 110 cm W.C.
Pressure Drop
Since the throat width was automatically controlled the
throat area and gas velocity were not known with any precision.
The maximum throat area possible was 0.361 m2. The throat length
was not known either. Thus, direct comparison of pressure drop
measured with equation (1) was not possible.
Particle Collection
In many Venturis the ratio of drop velocity to gas velocity
at the throat exit (u*[e) is about 0.8, and this value will be
used in the prediction. The liquid to gas ratio used is based
on the gas flow rate at the outlet for runs 12 and 13 (20 Am3/s)
and 80% of the stated liquid flow rate (33 £/s) . Table 20 presents
the conditions used for the performance prediction.
TABLE .20. CONDITIONS FOR AAF VENTURI
PERFORMANCE PREDICTION
AP = 110 cm W.C.
QL/QG = 0.0017
PL = 1.0 g/cm3
PG = 0.79 kg/m3
yfi = 1.6 x 10"" g/cm-s
ude = °'8
ur+ = 8,860 cm/s
bt
dd = 104 urn
CDO" Q'58
B = 3.37
j, = 41 cm
T-
The comparison of prediction with the data is shown in
Figure 2L The measured penetration is also shown corrected for
53
-------
2
O
t—i
H
U
2
O
i—i
H
H
W
0.05
PREDICTION
CORRECTED FOR
WET PARTICLES
WITH DILUTION
MEASURED FOR
DRY PARTICLES
0.01
0.005
0.001
0.3 0.5 1 23
PARTICLE AERODYNAMIC DIAMETER, d__,
pa
Figure 21. Comparison of predicted and measured
penetration for the average of runs 12
and 13 of the AAF venturi
54
-------
dilution and for a diameter increase to account for the fact
that the measurements were made with heated impactors. Based
on an average dp? = 1 umA and ag = 3 for the inlet dust the
overall penetration was predicted to be about 15% while the
measured penetration was about 3%.
Discussion
In order to compare the model with experiment a few important
factors need to be noted:
1. Heated impactors - As explained previously the inlet and
outlet size distributions were determined using heated cascade
impactors. Based on previous experience it was expected that the
mass median diameters of the actual wet borax particle size dis-
tributions were up to 1.5 times the sizes measured in the heated
impactors. The high solubility of borax was the primary reason
for the difference. Thus the venturi was collecting larger par-
ticles than those measured in the heated impactor. In order to
compare the experimental results with the prediction it is neces-
sary to plot experimental penetration against the actual (wetj
particle size rather than the dried particle size. This correc-
tion causes the measured penetration curve in Figure 21 to shift
toward larger particle diameters because Figure 10 is on the dry
particle size basis.
2. System leaks - The actual scrubber penetrations could
also be different from those based on the impactor data because
of dilution of the outlet by air leaks into the system, which
probably did occur. Dilution of the outlet stream would cause
the actual penetrations to be greater than those measured by a
factor equal to the dilution factor. The outlet concentration
may have been diluted as much as 251. The dashed line on Figure
21 includes the effects of 25% dilution and 1.5 times particle
growth on measured data.
3. Collection mechanisms - The model for venturi performance
assumes particle collection by inertial impaction on drops in
the throat region of the venturi. This model does not account
55
-------
for forces other than inertia which can effect submicron par-
ticle collection. Thus, in the region below about 0.5 ymA the
predicted penetration and the experimental penetration are expec
ted to differ.
Another important phenomenon is the collection of particles
of all sizes due to solution-induced condensation. Borax (Na2lU
is highly soluble in water. If the scrubber liquor became more
concentrated than would be in equilibrium with the vapor, it is
conceivable that particles could be swept toward the liquid as
water vapor moves to condense on it. The liquid drops which
contain dissolved borax have a reduced vapor pressure which in-
duces the condensation which causes the growth in particle size
manifested in the differences between wet and dry particle sizes.
Some of this dissolution and condensation will have occurred
in the quench section upstream of the venturi. However, because
of the large amount of liquid injected at the venturi these
mechanisms may still be causing particle collection and growth
downstream of the venturi. Growth or collection of the submicron
particles after the inlet sampling point, either before or after
the venturi would explain some of the differences between the
predicted and the experimental penetrations.
Condensation can also occur when the vapor pressure of the
water is reduced because it is at a lower temperature than the
adiabatic saturation temperature of the gas. This cause of con-
densation was not significant here because there was very little
cooling of the scrubber liquor.
4. Sensitivity of prediction to L/G ratio - The predicted
curve is very sensitive to the liquid to gas volume flow rate
ratio (L/G) for particle diameters above 1.0 ymA. A 25% decrease
in the L/G ratio would cause the prediction to agree closely with
the data. Since neither the inlet gas flow rate nor the liquid
flow rate are known precisely, it is possible that the actual
L/G ratio was lower than the value used in the prediction.
5. Entrainment - The carry-over from the cyclone was so heavy
that precutters had to be used on the outlet sampling probe. It
56
-------
was possible that some smaller entrapment drops, containing dis-
solved borax, penetrated the precutters or were shattered in the
precutters and were collected in the heated impactors. Since
the venturi model assumed that no entrainment carry-over occurred,
the actual penetrations would be greater than the model predictions
as seen in Figure 21.
These five factors help explain the differences between the
model prediction and the data. The many factors and uncertainties
involved are enough to preclude any judgment of the accuracy of
the model.
GAS-ATOMIZED SPRAY SCRUBBER COMPARISONS
The gas-atomized spray scrubber had a throat section that
was a rectangular orifice, whose width could be varied to adjust
the pressure drop.
Pressure Drop
The geometry of this scrubber was well known except for the
"throat length." Since the throat was an orifice the physical
length was only a fraction of a centimeter. However, a free
jet will not expand immediately but will continue straight for a
few centimeters. A straight section length of 10 cm was chosen
and as seen from Table 21 a reasonable pressure drop was predicted
for some of the run sets. Predicted pressure drop for two of
the run sets was very poor indicating either an error in measure-
ment or the assumed straight section throat length was too large.
Particle Collection
The scrubber parameters are known or estimated from the
average of the conditions for the sets of runs and shown in
Tables 21 and 22. The predicted particle penetrations are shown
compared to the average of the run sets in Figures 22-26. The
solid (measurement) curves represent the measured penetration with-
out accounting for leakage. The dashed curves assume that the
outlet was diluted by the factor listed in Table 22 which is
based on the data in Table 14.
The overall penetration predictions for the average log-nor-
mal inlet dust distribution based on the cascade impactor data
are shown in Table 23.
57
-------
TABLE 21. PRESSURE DROP COMPARISON FOR THE
GAS ATOMIZED SPRAY SCRUBBER
Run Set uct
cm/s
5,6
8
9,
11,1
14,
Note:
,7 10,
23,
10 24,
2,13 10,
15 10,
Assumed
600
200
400
400
400
tt
QL
0.
0.
0.
0.
0.
= 10
/QG
0011
0009
0008
0031
0014
cm
ul Measured AP
de cm W.C.
0.
0.
0.
0.
0.
61
70
70
43
58
79
83
82
103
88
Predicted AP
cm W.C.
78
349
342
148
91
TABLE 22. CONDITIONS FOR GAS-ATOMIZED SPRAY
SCRUBBER PERFORMANCE PREDICTION
Run Set
5,6,7
8
9,10
11,12,13
14,15
PL
g/cm3
1
1
1
1
1
.0
.0
.0
.0
.0
PG
kg/m3
0
0
0
0
0
.72
.70
.36
.93
.80
2
2
4
1
2
yG
g/cm-s
.1x10"'
.2x10'"
.1x10'"
.7x10"*
.0x10"*
dd
ym
79
45
40
200
94
0
0
1
0
0
CDo
.68
.64
.10
.46
.61
B
2.26
2.00
2.02
7.30
2.86
Oil1
1.23
1.27
1.88
1.79
1.22
!Dil is the ratio of measured outlet to inlet volume flow rates at
standard conditions.
58
-------
1,0
Cn
WITH DILUTION FACTORH
1.
0.01
0.1 0.2 0.5 1 2
PARTICLE AERODYNAMIC DIAMETER, d
pa'
5
umA
10
Figure 22. Comparison of predicted with measured penetration
for average of runs 5, 6 and 7, for the gas-
atomized scrubber.
WITH DILUTION FACTOR
0.01
0.1 0.2 0.5 1 2 5 10
PARTICLE AERODYNAMIC DIAMETER, d . pmA
pa
Figure 23. Comparison of predicted with measured penetra-
tion for run 8 for the gas-atomized scrubber.
-------
1.0
g
1.0
WITH DILUTION FACTOR
0.01
0.1
0.2
0.5
PARTICLE AERODYNAMIC DIAMETER, d , ymA
pa
Figure 24. Comparison of predicted with measured penetra-
tion for average of runs 9 and 10 for the gas-
atomized scrubber.
WITH DILUTION FACTOR
.001
PARTICLE AERODYNAMIC DIAMETER, d ft, umA
Figure 25. Comparison of predicted with measured penetra-
tion for average of runs 11, 12 and 13 for the
gas-atomized scrubber.
-------
1.0
0.5
H
U
o
I—I
H
0.2
0.1 :
0.05
=; WITH DILUTION FACTOR
0.02
0.01
0.1 0.2 0.5 1 2 5
PARTICLE AERODYNAMIC DIAMETER, d , ymA
Figure 26. Comparison of predicted with measured penetra-
tion for average of runs 14 and 15 for the gas
atomized scrubber.
61
-------
TABLE 23. OVERALL PENETRATION COMPARISON FOR
THE GAS-ATOMIZED SPRAY SCRUBBER
Run Set
5,6,7
8
9,10
11,12,13
14,15
Measured *
Ft, %
13
5
5
7
15
Predicted
Ft, %
47
26
33
44
45
Note: Assumed d =0.4 ymA, a =2.5
ir o o
* Neglecting outlet precutter dry weight gain.
Discussion
The comparison of predicted and measured penetration leads
to the following observations:
1. The predictions are generally close to the experimental
results for particles around 1 umA diameter. Runs 5, 6, and 7
have the poorest fit. Although the operating conditions for
these runs were similar to those of runs 14 and 15, the exper-
imental penetration was greater. This may be due to the fact
that the gas flow rate was higher in runs 5,- 6, and 7 which could
mean more leakage and could cause more entrainment carryover at
the outlet. The effect of L/G ratio on the model predictions
is shown by comparing the predicted penetrations of these two
run sets. The main difference in the operating conditions was
that runs 14 and 15 had a 45% higher L/G ratio than runs 5, 6
and 7.
2. Predicted penetration for sub-micron particles is gener-
ally higher than measured. This may be due to condensation caused
by the cool duct walls and the cool scrubber liquid which had low
temperatures due to the cold ambient air (0°C). Leaks of the
cold ambient air into the system would also have caused conden-
sation.
62
-------
3. Predicted penetration is generally lower than measured
for particles larger than about 1 ymA diameter. There are several
plausible causes of this disparity:
a. Entrainment not collected in the entrainment separator
would yield particles in the larger size range. The cyclone
separator was known to be inefficient although its performance
was improved by internal modifications and leak patching. This
inefficiency was clearly demonstrated by noticeable liquid drops
at the outlet stack where the sampling crew was working, and by the
precutter mass collections.
b. Imprecise knowledge of the liquid flow rate can cause
the prediction to be substantially incorrect in the larger par-
ticle size range. This effect becomes more serious as the liquid/
gas ratio decreases, as seen by comparing the prediction of runs
4, 5, and 7 with that of runs 14 and 15.
4. Predicted overall penetration was generally higher than
measured. This difference is probably due primarily to the
assumption that the size distribution is log-normal with the same
standard deviation for all particles less than 0.4 ymA (dp ) .
Condensation probably affected the log-normality of the small
size distribution. Also, the penetration model over-predicted
the penetration of the smaller particles which would definitely
cause the overall penetration prediction to be too high.
In conclusion, with all the real effects considered, the
model provides an adequate prediction. Predictions for several
of the runs, notably 8, 11, 12, 13, 14, and 15, were quite good.
63
-------
SECTION 7
PERFORMANCE TEST METHOD
The approach used in these scrubber evaluations involved
obtaining experimental collection efficiency data, acquiring
information on system characteristics and behavior, and per-
forming computations which made use of the performance data
and mathematical models. Over the course of the program the
methods and apparatus used generally were improved and modified
to suit each specific test situation; however the main features
remained similar and are described here.
The most important experimental measurements were those
regarding particle size and concentration. Cascade impactors
were used to determine particle diameters larger than 0.3 ymA.
The Air Pollution Technology, Inc. portable screen diffusion
battery (A.P.T. - S.D.B.) was used for particle size determin-
ation in the diameter range from 0.01 ym to 0.1 ym (actual).
The apparatus and methods used are outlined below:
1. Gas velocity distribution and parameters were measured
at the inlet and outlet of the scrubber in order to define the
following:
a. Conditio-ns required for isokinetic sampling.
b. Particle mass concentration per unit volume of
dry gas, which is needed to provide a consistent basis for com-
paring inlet with outlet size distributions in the computation
of efficiency.
c. Gas flow rate.
d. Amount of liquid entrainment in the scrubber outlet
Methods to measure these parameters are tabulated in the
table on the following page.
64
-------
TABLE 24. MEASURING EQUIPMENT AND METHODS
Parameter Equipment
Gas velocity
and flow rate
Gas temperature
Humidity
•••- -—•—•—- - . - — _
Standard pitot tube or cali-
brated type "S" pitot tube;
differential pressure gauge.
Calibrated thermocouple or
mercury filled glass-bulb
thermometer.
Wet bulb - dry bulb
thermometers.
Method
- • ' »-»^»^^«^»
EPA Method 1;
EPA Method 2.
Pressure
Inclined water manometer
or a pressure gauge.
Wet and dry bulb
temperature mea-
surement on a
flowing sample
withdrawn from
the duct.
Measured by means
of a static pres-
sure tube inser-
ted in the duct.
2. The most essential part of a scrubber performance
test is the determination of particle size distribution and
concentration (loading) in the inlet and outlet of the scrubber.
For accurate determination of particle size distribution, a
collection mechanism that collects particles and causes neither
formation nor breakup of aggregates is necessary. Cascade
impactors come close to meeting these requirements.
In a cascade impactor particles are classified by inertial
impaction according to their mass. The larger ones are collected
on the plate opposite the first stage and the smallest on the
plate opposite the last stage. A.P.T. used the University of
Washington Mark III source test cascade impactor for particle
size fractionation. This impactor classifies particles into
seven size ranges and is capable of sizing particles down to
about 0.1 ym diameter (actual). All impactors were calibrated
in the laboratory according to EPA guidelines, see Calvert, et
al. (1976) and Harris (1977) on calibration method.
In order to minimize probe losses, tests were made with
the impactors in the duct and with the inlet nozzles appro-
65
-------
priately sized to give isokinetic sampling. A modified EPA Method
5 sampling train was used to monitor the sample gas flow rate.
Figure 27 shows the sampling train arrangement.
In some tests, a precutter was used to remove either the
large particles from inlet samples or the entrained liquid from
outlet samples. A round jet impactor was found to work satis-
factorily as the precutter and was adopted for use for both inlet
and outlet sampling. The impactors were either given time to
reach the duct gas temperature or heated to prevent condensation.
To increase the weighing accuracy, lightweight substrates
were used on the collection plates. Generally, either greased
aluminum foil or a glass fiber filter paper substrate was used.
Impactor substrates and backup filters were weighed to the nearest
tenth milligram (10~"g) by using an analytical balance.
Particle size distribution and loading measurements were con-
ducted simultaneously at the scrubber inlet and outlet. This
minimizes the effects of particle size distribution changes which
may result from fluctuations in the operation parameters. The
sampler was held at one location in the duct for the duration of
each sampling run. This is an adequate technique for obtaining
good samples of particles smaller than a few microns in diameter
because they are generally well distributed across the duct.
Blank impactor runs were performed periodically to assure
that the greased aluminum substrates did not react with the stack
gases or lose weight. A blank impactor run consists of an im-
pactor preceded by two glass fiber filters and run at identical
sample conditions as the actual smapling runs.
3. In-stack filter samples were taken to obtain total par-
ticulate loadings and overall collection efficiencies of the
system. The sampling train arrangement was the same as the cas-
cade impactor train (Figure 27) with the cascade impactor replaced
by a filter.
4. Inertial impaction devices such as cascade impactors are
normally insufficient for sizing particles smaller than 0.1 to
66
-------
PRECUTTER
AND
NOZZLE
u
THERMOMETER IMpINGER TRAIN
HEATED
CASCADE
IMPACTOR
ORIFICE
METER
MANOMETER
DRY GAS
METER
VACUUM
PUMP
SILICA
GEL
DRYER
Figure 27- Modified EPA sampling train with in-stack cascade impactor
-------
0.3 vim (actual diameter). Size fractionation of these particles
may be accomplished by diffusional collection devices (diffusion
batteries) which may consist of closely spaced parallel plates,
long, thin tubes, or an array of wire screens. Parallel plate
and tube diffusion batteries require many pumps, dilution appara-
tus, and other bulky equipment which prove to be cumbersome in
field use. For this and other reasons, Air Pollution Technology,
Inc. developed and uses a portable screen diffusion battery which
is lighter and more portable than previous devices.
The screen diffusion battery utilizes a series of layered
screens intermittently separated for sampling purposes (Figure
28). Size fractionation by the diffusion battery is detected by
measuring the overall particle number concentrations of the gas
stream entering and leaving a known number of screens. A con-
densation nuclei counter (CNC) is used to determine number con-
centrations. Concentrated aerosol samples must be diluted until
compatible with the CNC (~106 particles/cm3). Screen penetration
data are then analyzed to determine size distribution and cumula-
tive mass loading of the particulates in the stream. Cascade im-
pactor and diffusion battery analyses often can be combined to
obtain an overall characterization of the particulate size dis-
tribution (and scrubber penetration).
Fine particle size measurements with the diffusion battery
were not taken simultaneously at the inlet and outlet of the
scrubber system. During an impactor run, several inlet and out-
let measurements were taken alternately with the S.D.B. Since
the system remained fairly constant during each run, alternate
inlet and outlet S.D.B. measurements were considered to approxi-
mate simultaneous sampling.
Each S.D.B. run consisted of a continuous series of CNC
readings. Normally, CNC counts were taken at each diffusion
battery stage in order of increasing number of screens and then
the process was repeated until three to four sets of readings
68
-------
r
STACK
GAS
1st DILUTION
FLASK WITH CHARGE
NEUTRALIZERS
CASCADE
IMPACTQR
ROTAMETERS
FILTER
9
ROTAMETER
VACUUM
PUMP
FILTER
FLOW
METER
1
DIFFUSION
BATTERY
2nd
DILUTION
FLASK
TO
CNC
PUMP
DESSICANT
VACUUM
»« > PUMP
Figure 28, Schematic diagram of diffusion battery system,
-------
were obtained. Continuous monitoring of flow, temperatures, and
pressures enabled steady operation of the diffusion battery. Con-
ditions in the duct (pressure, gas velocity, temperature, and
water vapor content) were obtained during the impactor tests.
The size of particles entering the diffusion battery was
limited by using a cascade impactor precutter on the instack end
of the probe. loskinetic sampling was not maintained because
the particles to be measured were too small to be segregated by
inertial effects from bends in the gas stream. The sample stream
entering the diffusion battery was immediately diluted with heated,
dried, filtered air to control condensation. Two Polonium 210
charge neutralizers were inserted into the flask to eliminate
electrostatic effects.
A portion of the resultant aerosol was directed through the
diffusion battery and the outlet was diluted to a concentration mea-
surable by the Gardner CNC. The aerosol from the second dilution
flask was sampled with the Gardner CNC. The excess aerosol was
exhausted through the vacuum pump. The excess aeros.ol from the
first dilution was passed through an absolute filter and pumped
to the atmosphere. The Gardner CNC was calibrated daily against
a standard B.G.I. Pollack, Model P, CNC and found to read con-
sistently 33% lower than the Pollack CNC for the concentration
range used in these tests.
ERROR ANALYSIS
Sample Bias
It is important to note that the program objective is to
investigate scrubber performance on fine particles and, conse-
quently, it is not essential that the methods used be accurate
for large particles. This makes the sampling simpler in the
following ways:
1. Isokinetic conditions are not important for fine particles.
For example, the error caused by sampling 4 ymA particles at a
velocity 50% higher or lower than the gas stream velocity would
only be about 2 or 3% of the concentration.
70
-------
2. The fine particles will be well distributed in the gas
stream, except in cases where streams with different particle
concentrations have not had time to mix, so that generally single
point sampling is sufficient. To illustrate, we may note that the
Stokes stopping distance of a 3 ymA particle with an initial
velocity of 15 m/s (50 ft/s) is about 0.04 cm (0.016") and for
a 1 ymA diameter particle it is one ninth of that. Since the
stopping distance is the maximum distance a particle can be dis-
placed from a gas stream line by going around a right angle turn,
it is obvious that fine particle distribution in the gas stream
will be negligibly affected by flow direction changes.
3. The effect of a precutter on the size resolution of a
cascade impactor is not significant in the size range of interest,
so long as the precutter has a cut diameter larger than several
microns.
Diffusion Battery
The screen diffusion battery was calibrated in the A.P.T.
small particle laboratory. An aerosol of known size distribution
was generated and passed through the diffusion battery. The
total number concentration was measured with a condensation
nuclei counter at the inlet and outlet of each S.D.B. stage.
The penetration (percent) of particles was then calculated and
plotted against solidity factors on semi-logarithmic paper. The
experiment was repeated with the same aerosol until a smoothed
average curve relating number penetration to solidity factor
was obtained. From the smoothed curve, a correlation factor was
found for computing the theoretical diffusion battery performance.
The scatter of data points about the smoothed (fitted)
calibration curve represents the experimental error in the pene-
tration measurement. This measurement error included meter reading
error and the accuracy of the CNC. The measurement error was de-
fined in terms of relative error, or the deviation from the aver-
aged penetration value divided by the averaged value.
This procedure was repeated on other aerosols of known
size distribution. The maximum relative error was then determined
71
-------
from these experiments for each solidity factor. The maximum
relative error for the screen diffusion battery determined by
this method is 10.4% for solidity factors of 13, 26, and 40.
Cascade Impactors
Cascade impactors were used as the principal means of obtain-
ing information about the inlet-outlet size distributions. It
was important to understand the sources of error and how the
error can be minimized.
The procedural errors include the accuracy of the weighing
of the deposits, and reading of the test data such as temperature,
gas volume, time, and pressures. The errors from the impactor
design and construction include wall losses, accuracy and pre-
cision in construction of critical components, and particle re-
entrainment from the collection surface.
Some of the design and construction limitations can be re-
duced by procedures such as recovering the wall losses and by
sampling at certain flow rates and times to reduce reentrainment
errors and by calibration of the impactor. The experimental
data obtained with commercial impactors were reported by Smith,
et al. (1974). Smith, et al. (1974) reported that all impactors
tested had appreciable wall losses for particle diameters above
10 microns. This error can be reduced by brushing the material
from the wall onto appropriate collection disks. The flow velo-
city through the impactor jets should not be above 65 m/s to be
absolutely certain of avoiding reentrainment of particles from
the collection substrates. The extent of reentrainment will de-
pend on the properties of the material and the amount of deposit.
However, Rao (1975) reported that collection efficiency increased
with increased particle load. When the particle weight is over
10 mg, part of the deposit may break away from the surface and mi-
grate within the impactor. The lightweight deposit places im-
portance on accurate weighing. The analysis of impactor errors
was limited to the weighing error and in the calculation of
collection efficiency error. The effects of weighing errors on
the results of impactor tests have been analyzed by Sparks (1971),
72
-------
An analysis of the weighing error using three different estimations
was reported by Fegley, et al. (1975). The results indicate that,
when the weight of sample per stage is less than 1 mg when weighed
with a balance with a precision of 0.05 mg, the error in the frac-
tional mass will be greater than 10%.
73
-------
SECTION 8
DATA REDUCTION AND COMPUTATION METHOD
CASCADE IMPACTOR DATA ANALYSIS
In a cascade impactor particles are classified by inertial
impaction according to their mass. The larger ones are collected
on the plate opposite the first stage and the smallest on the
plate opposite the last stage.
Once the stage "catches" have been measured, usually by
weighing particle collection foils or papers, the data analysis
is relatively simple. Generally the objective is to make a
plot of particle diameter versus mass percent oversize or under-
size and to represent the size distribution in terms of log-
normal distribution parameters if possible. Thus, it is neces-
sary to do the following:
1. Add all of the stage and filter collection weights to
get the total particle mass collected.
2. Compute either:
a. Cumulative percent collected as the gas flows through
the succeeding stages. This is "percent oversize."
b. Cumulative percent penetrating as the gas flows
through succeeding stages. This is "percent under-
size."
3. Compute the cut diameters for the impactor stages,
taking into account gas viscosity (or temperature) and gas
sampling flow rate. The equation is:
0,5
74
-------
where: dpac = impactor stage cut diameter, ymA
yG = gas viscosity, poise
d. = jet diameter, cm
u. = jet velocity, cm/s
The particle diameter used is called "aerodynamic diameter"
and it has the unit of "aerodynamic microns," ymA. This is the
effective diameter for particle separation by inertial impaction
and it takes into account the effects of particle density and
particle "slip" between gas molecules. It is related to the
actual physical size of the particle by the following equation:
where: d = aerodynamic particle diameter, ymA
pa
d = actual particle diameter, ym
p = particle density, g/cm3
C' = Cunningham slip correction factor, dimensionless
At room temperature for air the Cunningham slip correction
factor, C', is given by:
C» - 1 + ---- (9)
If the particle distribution follows the log-normal law,
a straight line will result on log-probability paper. The 501
value of particle diameter is the geometric mass mean diameter,
d , and the geometric standard deviation, a , is given by:
Pg 8
- !i:**value of-^ (io)
g 50% value of d
pa
75
-------
Overall Particle Penetration
Overall particle penetration is defined as:
y * dP cn>
mass concentration out
massconcentration in
where: Pt = overall particle penetration, fraction
Pt(d ) = penetration for particles with diameter
" d , fraction
f(d ) = particle frequency distribution
C . = total inlet particle concentration, mg/DNm3
The overall particle penetration can be computed using the
data from a simulatneous inlet and outlet cascade impactor or
filter run.
Particle Penetration as a Function of Particle Diameter
The particle penetration for particle diameter, d ., or
grade penetration curve is given by:
"dC
PtCdp)
dC
outlet
(12)
P_
\
inlet
dC
where: — *— = the slope of the cumulative mass concentration
d(d )
P versus particle diameter curve at d ,
mg/DNm3- ymA
Particle penetration as a function of particle size is com-
puted from inlet and outlet particle size distributions and
concentration data. The major steps involved in the computation
are as follows:
76
-------
1. Reduce cascade impactor data to the form of cumulative
particle mass concentration for each impactor cut diameter.
2. Determine the slopes of the cumulative mass distribution
curves at several values of particle diameter for both the inlet
and outlet and then compute penetration at each particle diameter.
There are several techniques to determine the slope of the
cumulative mass distribution curve. Some of the techniques are
discussed below.
1. Graphical technique - Cumulative mass concentration versus
aerodynamic particle diameter data may be fitted with a curve
by eyeball method. The slopes of the curve at various values of
particle diameter are then measured graphically.
2. Curve fitting - Curve fitting to the data points and
the measurement of curve slopes by eye involves subjective judg-
ment. To eliminate the judgment errors, it is possible to fit
the data with a mathematical function and then evaluate the slope
analytically. We have tried fitting the cumulative mass curves
to log-normal distribution functions, and the Weibull distribution.
Log-normal Distribution -
If the inlet and outlet size distributions are nearly
log-normal, then a purely mathematical particle penetration is
used. This mathematical log-normal penetration is based on the
following:
The cumulative distribution of particle sizes is:
(13)
where P(x) is the cumulative mass fraction of sizes smaller
than "d ", and
pa '
In d - In d
-\r "^ J» | I,, ~ V. s
lnag
77
-------
The derivative of the distribution function is:
d P _ d P d x _ exp (-x2/2) , ->
* dpa " d x d dpa " C2iO* dpa in crg
thus,
d C C . exp(-x2/2)
= E = -^ (16)
d d f 2ir) d In o
pa pa g
Using equation (12) , the penetration is then:
)
:pti Cln Vout
exp
Weibull Distribution -
The Weibull distribution, Lipson and Sheth (1973), offers
two advantages over the log-probability distribution. The first
is that it has three parameters rather than two. The second is
that the cumulative distribution function (CDF) is explicit and
does not have to be approximated by multi-termed polynomials.
Cumulative distribution function -
CDF = 1 - exp
JE_lV\ | - (18)
6 " dpo
where d = particle diameter , cm
d = minimum particle diameter, cm
0 = characteristic diameter , cm
b = Weibull slope , dimensionless
78
-------
b = B
The CDF has the property that:
CDF (0 = d ) = 0.632 (19)
The median particle diameter occurs when the CDF = 0.5, so:
V • dpo + <9 - dpo) (in 2)1/b (20)
Linear transformation - Transformation to a linear form:
y = A + B x
requires that:
y = In In ( ^CDF) C22)
1 (23)
therefore,
A - -b In (0 -dpQ)
B = b
and,
79
-------
Least squares curve fit - The minimum particle diameter,
d , is that which results in the highest linear correlation co-
efficient based on the above linear transformation, when a least
squares linear regression is performed on the size distribution
data. Note that,
0 < d < smallest diameter found in the distribution
- po
Density function - The Weibull density function is the de-
rivative of the CDF:
'd -d
ib-1
f(dp) = ---i exp
p e-d \ e-d
po \ poy
dP°
(26)
Penetration - The penetration is the ratio of the cumulative
mass loading distribution derivatives,
Pt (d ) to fout <
Cpti £in
-------
"average"size of the particles in the distribution. The median
particle diameter is directly related to "6" by equation (20).
Weibull slope, physical interpretation - "b" is analogous
to the geometric standard deviation of the log-probability dis-
tribution. It indicates the "spread" of the size distribution.
The larger the Weibull slope, bs the more uniform (monodisperse)
the particle sizes.
SCREEN DIFFUSION BATTERY DATA ANALYSIS
Screen diffusion battery data consist of particle number
concentrations which are obtained after each stage in the diffu-
sion battery. Particle penetrations are calculated from the
ratio of the number concentration taken at a given SDB stage to
the inlet number concentration. The particle size distribution
may then be determined from the penetration data.
Most particle formation processes result in a particle size
distribution which is log-normal. Log-normal size distributions
are conveniently represented by two parameters: the number geo-
metric mean diameter (d ) and the standard deviation (a ).
Typically, process created aerosols fail log-normality only at
the extremes of large and small particles which represent only
a small percentage of the total particulates.
Calvert, et al. (1972) describe a method for converting log-
normal size distributions to overall penetrations using the
relation between particles of a discrete diameter, and penetration
of those particles through the device.
The equation is a discrete form of the following equation:
f
Pt = J Pt f (dp) a~(ln dp) (28)
~ OO
where, for a screen diffusion battery:
B
Pt = exp (-A S dp ) (29)
81
-------
and, Pt = penetration fraction of a particle of a given diameter
through S
S = solidity factor, a dimensionless parameter
d = particle diameter, ym
A,B = constants established by theory and laboratory
experimentation, dimensionless
and for a log-normal distribution:
f(dp) -
y/2/n In
exp
g
2 -i
dP - ln V
In a
g
(30)
Calvert and Patterson (1977) have shown that for particles
e size range of 10" um
through the SDB is given by:
in the size range of 10~3um < d < 10" um the penetration
Pt = exp | -1 x 10"" S (us dr)"°"67d ~1>29
s c-
(31)
where: u = superficial gas velocity, cm/s
d = wire diameter of screen, cm
Thus, the variables in equation (29) have the following values:
A = 1 x 10"* (us dc)~0>67
B = -1.29
The particle diameter that penetrates a SDB stage with 501
efficiency is found by substituting 0.5 for Pt in equation (29):
1/B
pc
-In (0.5)
AS
(32)
82
-------
Each stage will have a certain "d " for a given flow rate.
In order to determine the number geometric mean diameter,
dpn, and standard deviation, og, of the particles entering the
SDB a curve matching technique must be used. The data consist
of overall penetration, Pt (ratio of outlet to inlet particle
concentration), and stage cut diameter, d , for each screen.
A plot of these data points must match a curve that is a numeri-
cal solution to equation (28) such as one of those presented
by Calvert, et al. (1972). The curve most closely matched will
determine the "d " and "a " of the particles entering the SDB.
Non-log normal data must be handled by a graph stripping
technique outlined by Sinclair (1972) which entails tedious
graphical integration and mathematical conversions. In our ex-
perience the data have fallen sufficiently close to log-normality
that the curve matching technique is acceptable.
Conversion of number distribution to mass distribution is
necessary in order to put the diffusion battery and cascade
impactor data on the same basis. The method used to make this
conversion is a graphical integration of the following equation:
N . /-Npi dm /N_ x 100
c . . _Bi f _P 3 [ _P | (33)
/-
f
J
P1 100 J dN \ N
O r \
where: N = cumulative number concentration of particles
smaller than "dp", #/cm3
N = total number concentration of particles, #/cm3
pt
d = particle diameter, ym
d P = mass of particles in the infinitesimal size range
P1 (dp + 3dp), g
C . = cumulative mass concentration of particles
smaller than "dpi", g/cm3
83
-------
The quantity (dm /dN ) is simply the mass per particle of diameter
"d ". The quantity (N x 1QO/N ) is the number percent of
particles smaller than ud ". Thus, equation (33) can be evaluated
from a plot of mass per particle versus cumulative number percent
of particles, both quantities being evaluated at the same par-
ticle diameter to provide a point on the plot. The total and
cumulative number concentration data are obtained as described
previously.
For a log-normal distribution the mass and number distribu-
tion are related by:
a (number) = a (mass) (34)
o o
ln dPg = ln V + 3 ln
Finally, since impactor data are usually reported as mass aero-
dynamic impaction diameters,equation (8) must also be used to
complete the comparison of SDB and cascade impactor data.
84
-------
REFERENCES
Calvert, S. "Engineering Design of Fine Particle Scrubbers",
J. of APCA. 24: 929, 1974.
Calvert, S., C. Lake, and R. Parker "Cascade Impactor Calibra-
tion Guidelines" EPA-600/2-76-118, 1976a.
Calvert, S. , H.F. Barbarika, and C.F. Lake "National Dust
Collector Model 850 Variable Rod Module Venturi Scrubber
Evaluation" EPA-600/2-76-282, 1976b.
Calvert, S., H.F. Barbarika, and G.M. Monahan "Gas Atomized
Spray Scrubber Evaluation" EPA-600/2-77-209a, 1977a.
Calvert, S., H.F. Barbarika, and G.M. Monahan "American Air
Filter Venturi Evaluation" EPA 600/2-77-209b, 1977b.
Calvert, S. and R.G. Patterson "Submicron Particle Size Measure-
ment with a Screen Diffusion Battery", Final Report, EPRI Con-
tract RP 723-1-760205, 1977.
Fegley, M.J., D.S. Ensor, and L.E. Sparks "The Propagation of
Errors in Particle Size Distribution Measurements Performed
Using Cascade Impactors" Paper 75-32.5 presented at the 68th
Annual Meeting of APCA, Boston, MA, June 15-20, 1975.
Harris, D.B. "Procedures for Cascade Impactor Calibration and
Operation in Process Streams" EPA 600/2-77-004, January 1977.
Lipson, C. and N.J. Sheth "Statistical Design and Analysis of
Engineering Experiments", McGraw-Hill, 1973.
Lundgren, D.A. "An Aerosol Sampler for Determination of Particle
Concentration as a Function of Size and Time", J. of APCA, 17:
225, 1967.
Rao, A.K. "Sampling and Analysis of Atmospheric Aerosols",
Particle Technology Laboratory, Mechanical Technology Labor-
atory, University of Minnesota. Publication 269, 1975.
85
-------
REFERENCES (continued)
Sinclair, D. "A Portable Diffusion Battery" American Ind.
Hygiene Assoc. Journal 55: 729-755, 1972.
Smith, W.B., K.W. Gushing, and J.D. McCain "Particle Sizing
Techniques for Control Device Evaluations" EPA 650/2-74-102,
NTIS PB 240670/AS, October 1974.
Sparks, L.E. Personal Communication, 1971.
Yung, S., S. Calvert and H.F. Barbarika "Venturi Scrubber
Performance Model" EPA-600/2-77-172, 1977a.
Yung, S., H.F. Barbarika and S. Calvert "Pressure Loss in
Venturi Scrubbers" J. Air Poll. Control Assoc., 27: 548-551,
1977b.
86
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
! REPORT NO. 2
EPA- 600/2 -78-032
4. TITLE AND SUBTITLE
Evaluation of Three Industrial Particulate Scrubbers
7. AUTHOR(S)
Seymour Calvert, Harry F. Barbarika, and
Gary M. Monahan
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Air Pollution Technology, Inc.
4901 Morena Boulevard, Suite 402
San Diego , California 92117
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
3. RECIPIENT'S ACCESSION- NO.
5. REPORT DATE
February 1978
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO.
1
10. PROGRAM ELEMENT NO.
1AB012; ROAP 21ADM-029
11. CONTRACT/GRANT NO.
68-02-1869
13. TYPE OF REPORT AJMD PERIOD COVERED 1
Final; 3/75-12/77
14. SPONSORING AGENCY CODE
EPA/600/13
is. SUPPLEMENTARY NOTESIERL_RTP project officer is Dale L. Harmon, Mail Drop 61, 919/ 1
514-2925. Earlier reports in this series are EPA- 600/2 -76 -2 82 and EPA-600/2-77-
209a and -209b. |
16. ABSTRACT
The report gives results of field measurements, carried out on three full scale indus-
trial scrubbers to determine scrubber performance characteristics, including par-
ticle collection efficiency as a function of particle diameter. The three scrubbers
were different gas-atomized spray types with pressure drops ranging from 54 to 178
cm W. C. Their performance on major sources of fine particle emissions was com-
pared to a mathematical performance model for venturi scrubbers.
17. KEY WORDS AND DOCUMENT ANALYSIS |
'• DESCRIPTORS
Pollution Atomizing
Dust Spraying
Scrubbers
industrial Processes
Measurement
Gases
18. DISTRIBUTION STATEMENT
Unlimited
b.lDENTIFIERS/OPEN ENDED TERMS
Pollution Control
Stationary Sources
Particulate
Collection Efficiency
19. SECURITY CLASS (This Report)
Unclassified
20. SECURITY CLASS (This page)
Unclassified
c. COSATI Field/Group I
13B
11G
07A
13H
14B
07D
21. NO. OF PAGES I
99
22. PRICE I
EPA Form 2220-1 (9-73)
87
------- |