EPA-650/2-74-102-a
August 1975 Environmental Protection Technology Series
PARTICULATE SIZING TECHNIQUES
R CONTROL DEVICE EVALUATION
a)
-------
EPA-650/2-74-102-a
PARTICULATE SIZING TECHNIQUES
FOR CONTROL DEVICE EVALUATION
by
W. B. Smith, K. M. Gushing, G. E. Lacey, and J. D. McCain
Southern Research Institute
2000 Ninth Avenue South
Birmingham, Alabama 35205
Contract No. 68-02-0273
ROAP No. 21ADM-011
Program Element No. 1AB012
EPA Project Officer: D. B . Harris
Industrial Environmental Research Laboratory
Office of Energy, Minerals, and Industry
Research Triangle Park, North Carolina 27711
Prepared for
U. S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, D. C. 20460
August 1975
-------
EPA REVIEW NOTICE
This report has been reviewed by the National Environmental Research
Center - Research Triangle Park. Office of Research and Development,
EPA, and approved for publication. Approval does not signify that the
contents necessarily reflect the views and policies of the Environmental
Protection Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for use.
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environ-
mental Protection Agency, have been grouped into series. These broad
categories were established to facilitate further development and applica-
tion of environmental technology. Elimination of traditional grouping was
consciously planned to foster technology transfer and maximum interface
in related fields. These series are:
1. ENVIRONMENTAL HEALTH EFFECTS RESEARCH
2. ENVIRONMENTAL PROTECTION TECHNOLOGY
3. ECOLOGICAL RESEARCH
4. ENVIRONMENTAL MONITORING
5. SOCIOECONOM1C ENVIRONMENTAL STUDIES
6. SCIENTIFIC AND TECHNICAL ASSESSMENT REPORTS
9. MISCELLANEOUS
This report has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series. This series describes research performed to
develop and demonstrate instrumentation, equipment and methodology
to repair or prevent environmental 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.
This document is available to the public for sale through the National
Technical Information Service, Springfield, Virginia 22161.
Publication No. EPA-650/2-74-102-a
11
-------
ABSTRACT
This report describes the results of laboratory and field work
related to inertial particulate size classifiers (impactors and
cyclones). The impactor work deals largely with non-ideal
behavior of impactors and problems encountered in field testing.
Preparation and handling procedures for using glass fiber
impaction substrates are discussed together with problems resulting
from SO2 reactions with certain types of glass fiber filter
media. The results of a brief series of tests of electrostatic
effects in impactor sampling are described which indicate that
these effects can be substantial under some circumstances. Design
and calibration data are given for two series cyclone size devices;
one designed to operate at a flowrate of 140 1pm (5 cfm) and
the second designed to operate at 28 1pm (1 cfm). Each provides
three size fractionation points in the 0.5 to 10 urn size interval.
The cyclone systems permit collection of larger quantities of size
fractionated particulates and are somewhat easier to use than
are impactors.
111
-------
TABLE OF CONTENTS
Contents Page
SECTIONS
I - Introduction and Summary 1
II - Studies Related To The Operation of Cascade
Impactors 3
A. Collection Substrate Materials 3
B. Comparison of EPA Method Five and Impactor
Measurements 16
C. Electrostatic Effects 21
D. Reentrainment Due To Improper Operation 29
III - Cyclone Sampling Systems 33
A. General Discussion 33
B. Cyclone Design 35
C. One ACFM Series Cyclone Sampling System 37
IV - Appendices 50
A. Andersen Filter Substrate Weight Loss Study.... 51
B. Glass Fiber Substrate Weight Gains 75
C. Cyclone Design and Calibration 95
D. Conversion Table for Units 121
References 122
FIGURES
1 - Comparison of weight of sulfate on blank Andersen
Impactor substrates and observed anomalous weight
increases 9
-------
TABLE OF CONTENTS
(Cont'd.)
FIGURES (Cont'd.) Page
2 - Anomalous weight increases of Andersen glass
fiber impaction substrates at different flue gas
temperatures 10
3 - Anomalous weight gains of various 47 mm diameter
glass fiber filters at different temperatures.
(60 minute samples at flowrates of 0.25 ACFM) 11
4 - Anomalous weight gain of 64 mm diameter Reeve Angel
900 AF glass fiber filters 13
5 - Anomalous weight gains of Andersen Impactor glass
fiber impaction substrates 14
6 - Grain loading by conventional (ASME, EPA) mass
train versus grain loading by cascade impactor for
thirteen sampling locations. All loadings
refer to control device outlets except for 9A 17
7 - Comparison of results of size distribution measure-
ments using isokinetic (cyclone) and non-isokinetic
(Brink) sampling rates 22
8 - System used to study the importance of charge
neutralization when sampling with impactors at
precipitator outlets 23
9 - Aerosol charge neutralization study-Brink Cascade
Impactor with filter substrates 25
10 - Aerosol charge neutralization study-Brink Cascade
Impactor - Bare Metal Plates 26
11 - Aerosol charge neutralization study-Brink Cascade
Impactor - No collection plates 27
12 - Electrostatic effects in probe and partial Brink
Impactor. All flowrates 0.1 cfm 28
13A- University of Washington (Pilat) Stage 6, 0.1 cfm,
DSQ = 1.0 micron. Four micron diameter monodisperse
particles. The uniform particle deposition is
apparent as compared to Figure 13B 30
-------
TABLE OF CONTENTS
(Cont'd.)
FIGURES (Cont'd.) Page
13B - University of Washington (Pilat) Stage 5
0.5 cfm, D50 = 1.0 micron. Four micron
diameter monodisperse particles. Poor
particle deposition is evident with scouring
around the point of impact 30
14 - Left-0.1 cfm U. of W. back-up filter. Right
0.5 cfm U. of W. back-up filter. Filter on
the right shows reentrained particles col-
lected after passing 1.0 micron cut off stage
shown in Figure 13B 31
15 - Series cyclone used in the U.S.S.R. for sizing
flue gas aerosol particles. ** 34
16 - Generalized cyclone design for the application
of Lapple' s equation 36
17 - Collection efficiency vs. particle diameter.... 39
18 - Percentage of total cyclone catch as found on
the cyclone's nozzle, body, cup, cap, and
back-up filter versus particle size 40
19 - One ACFM series cyclone showing the three
cyclones and Gelman back-up filter in the
preferred orientation. The nozzles were de-
signed for isokinetic sampling from 10 fps to
100 fps are also shown 42
20 - Cumulative mass loading vs. particle diameter.. 45
21 - Cumulative mass loading vs. particle diameter.. 46
22 - Cumulative mass loading vs. particle diameter.. 47
23 - Photomicrographs of cyclone catches acquired
at Bull Run Steam Plant 48
24 - Photomicrographs of cyclone and back-up filter
catches acquired at Bull Run Steam Plant 49
C-l - Collection efficiency versus particle diameter
for the SRI-1 cyclone 96
C-2 - Engineering dimensions (inches) for Cyclone
SRI-1 97
VI
-------
TABLE OF CONTENTS
(Cont'd.)
FIGURES(Cont'd.) Page
C-3 - Engineering dimensions (inches) for Cyclone
SRI-2 99
C-4 - Collection efficiency versus particle diameter
for the SRI-3 cyclone 100
C-5 - Engineering dimensions (inches) for
Cyclone SRI-3 101
C-5A- Engineering dimensions (inches) for
Cyclone SRI-3 102
C-6 - Collection efficiency versus particle diameter
for the SRI-4 cyclone 103
C-7 - Engineering dimensions (inches) for Cyclone
SRI-4 104
C-8 - Collection efficiency versus particle diameter
for the SRI-5 cyclone 105
C-9 - Engineering dimensions (inches) for Cyclone
SRI-5 106
C-10- Collection efficiency versus particle diameter
for Cyclone TIB 108
C-ll- Engineering dimensions (inches) for Cyclone TIB.. 109
C-12- Collection efficiency versus particle diameter
for Cyclone T2A 110
C-13- Collection efficiency versus particle diameter
for Cyclone T2A Ill
C-14- Collection efficiency versus particle diameter
for Cyclone T2A 112
C-15- Engineering dimensions (inches) for Cyclone T2A.. 113
C-16- Engineering dimensions (inches) for Cyclone T3A.. 114
C-17- Collection efficiency versus particle diameter
for Andersen Modified Pre-separator 115
C-18- Collection efficiency versus particle diameter
for Andersen Modified Pre-separator 116
VI1
-------
TABLE OF CONTENTS
(Cont'd.)
FIGURES (Cont'd.) Page
C-19- Engineering dimensions (inches) for the
Andersen Modified Pre-separator 117
C-20- DSO cut point versus cyclone flowrate for six
calibrated cyclones 119
C-21- Cyclone nozzle design specifications 120
TABLES
I - 5
II - Soluble Sulfate Analyses of San Juan Filters 8
A-l - Steel Ferrule and Washer Weight Changes Upon Desic-
cation 53
A-2 - Andersen Substrate Cumulative Weight Change Due
To Desiccation (Sets 13N, 15N, 16N) 55
A-3 - Cumulative Changes in Weight Due to Desiccation for
Andersen Substrates by individual set 57
A-4 - Moisture Absorption By Three Andersen Back-Up Filters 58
A-5 - Desiccation and Handling Weight Change of "Clean"
Andersen Substrates (7N, 8N, 9N) 60
A-6 - Desiccation and Handling Weight Change of
"Normal" Andersen Substrates (ION, UN, 12N) 61
A-7 - Desiccation and Handling Weight Change of "Clean"
Andersen Substrates By Sets 62
A-8 - Desiccation and Handling Weight Changes of "Normal"
Andersen Substrates By Set 63
A-9 - Desiccation and Sampling Weight Change of Andersen
Substrates (4N, 5N, 6N) After Sampling Filtered
Air At 0.5 ACFM for Six Hours at 24 C 64
A-10- Desiccation and Sampling Weight Change of Andersen
Substrates (IN, 2N, 3N) After Sampling Filtered
Air At 0.5 ACFM for Six Hours at 120 C 66
Vlll
-------
TABLE OF CONTENTS
(Cont'd.)
TABLES (Cont'd.) Page
A-ll - Desiccation and Sampling Weight Changes of
Andersen Substrates By Setting After Sampling
Filtered Air At 0.5 ACFM For Six Hours At
24°C 67
A-12 - Desiccation and Sampling Weight Changes of
Andersen Substrates By Set After Sampling
Filtered Air At 0.5 ACFM For Six Hours At
120°C 68
A-13 - Weight Changes of Andersen Substrate After
Sampling Filtered Effluent From A Wet
Precipitator At An Aluminum Reduction Plant.
125 F Gas Temperature 69
A-14 - Typical Average Amounts of Collection For An
Andersen Impactor 70
A-15 - Cumulative Weight Change of Teflon Filter
Substrate After Desiccation and Baking 72
B-2 - Citadel Cement 78
B-3 - Bull Run Steam Plant 79
B-4 - Bull Run Steam Plant 80
B-5 - Filter Blanks (Untreated) 83
B-6 - Citadel Cement 84
B-7 - Bull Run 86
B-8 - Sulfur Dioxide Pickup 94
-------
SECTION I
INTRODUCTION AND SUMMARY
The scope of work for this contract includes investigations
related to various techniques for the evaluation of pollution
control devices. This report deals exclusively with inertial
sizing techniques, cyclones and cascade impactors.
The impactor work done during this reporting period has to do
largely with nonideal behavior and sampling problems which are
encountered in field testing. Tests were run to determine the
optimum jet velocities for good deposition. Laboratory tests
with hard, dry, aerosols indicate that approximately 10 m/sec
is the practical upper limit for jet velocities if reentrainment
is to be avoided. Glass fiber substrate weight losses due to
handling, preparation, and sampling were studied and it was
determined that those losses are less than 0.2 mg per stage if
the procedures described in the test report are used. Anomalous
glass fiber substrate weight gains constitute a severe problem which
can completely ruin impactor tests. It was discovered that most
of those gains are due to sulfate uptake on the substrates.
Several materials were tested, and preconditioning was tried to
solve this problem. Teflon, Whatman GF/A and GF/D, and Reeve
Angel 934 AH are glass fiber materials which showed little weight
gains. Preconditioning the normal Gelman type A substrates
for several hours in flue gas also led to low weight gains. A
short study of electrostatic effects in impactor sampling was
done. The results were not conclusive, but indicate that there
are substantial interferences with bare collection plates but
very little with glass fiber substrates.
-------
Several small cyclones were evaluated during this period and the
results are included in this report. Three cyclones were arranged
in series and used to measure particle size distributions and to
collect size segregated samples for analysis. This system is
simple and reliable to operate and should be very useful in
accomplishing the design functions.
Section II and Appendices A and B describe the work related to
impactors while Section III and Appendix C give results for the
cyclone experiments.
-------
SECTION II
STUDIES RELATED TO THE OPERATION OF CASCADE IMPACTORS
The studies reported in this section are related to an assort-
ment of potential problems which can occur when stack sampling
with cascade impactors. Many of these problems were noted during
our own field test program, some were reported by other testing
groups, and some are operational problems which have been suspected
but never systematically investigated. The experiments and
results discussed in this section are not all complete, and in
fact some serve only to illustrate the degree and nature of the
problems, rather than to suggest solutions. A detailed descrip-
tion of our sampling systems and discussions of other research
related to impactors were given in an earlier report on this
contract.l
A. COLLECTION SUBSTRATE MATERIALS
Most impactors have collection stages which are too heavy to
obtain accurate measurements of the mass of the particles
collected in each size fraction. Weighing accuracy can be
improved by covering the stage with a lightweight collection
substrate made of aluminum foil, teflon, glass fiber filter
material, or other suitable lightweight materials, depending upon
the particular application. Some manufacturers now furnish light-
weight inserts to be placed over the collection stages. With
such arrangements, it is possible to make accurate determinations
of the masses collected on various stages without the risk of
reentrainment due to overloading.
The use of greased foils or lightweight glass fiber filter mats
as collection substrates tends to alleviate the problem of weighing
accuracy, but problems due to substrate weight changes are
-------
introduced. Although normal substrate preparation includes
baking and desiccation before the initial weighing, it is frequently
found that weight losses can occur when sampling clean air. Tests
were conducted to investigate this phenomenon in detail. It was
found that with careful handling, weight losses for glass fiber
substrates can be kept below 0.2 mg. This loss is attributed to loss
of fibers which stick to seals within the impactor and to "super-
drying" when sampling hot, dry, air. Weight losses of 0.2 mg
are small compared to normal stage catches when sampling particulate,
and thus are within a tolerable range for sampling errors. The
results of these experiments are given in detail in Appendix A.
When clean, hot, air is sampled with greased substrates, more
severe weight losses occur. Some of the weight lost on upper
impactor stages reappears as weight gained on the backup filters.
This is interpreted as an indication that grease flows or is blown
off the collection surface. Typical results are shown in
Table I from tests of two commonly used greases which were prebaked
and weighed before sampling filtered flue gas to investigate
weight losses. The changes shown in this table are about as large
as typical particulate catches and thus constitute serious inter-
ferences. We do not use greased foils often in our tests and
therefore have not studied this effect extensively. The limited
results shown in Table I indicate that a potentially serious
problem does exist and that further investigations are warranted.
Substrate weight gains have been a source of very large errors when
sampling industrial flue gases. Extensive studies were performed to
determine the magnitude of weight gains, their cause, and to
recommend solutions. These tests were initiated after it was dis-
covered that glass fiber substrates gained as much as 4-5 mg when
sampling filtered gas. These weight gains, due to gas phase
-------
TABLE I
SUBSTRATE DOW DOW SILICONS
MOLYKOTE HIGH VACUUM GREASE
111 COMPOUND
TEMP 300°F 280°F
FLOWRATE 0.46 ACFM 0.45 ACFM
SAMPLE DURATION 60 MIN. 60 MIN.
STAGE WEIGHT CHANGE(mg)
1 -1.1 -4.06
2 -0.74 -1.74
3 -0.34 -3.60
4 -0.36 -3.76
5 -0.46 -1.32
6 -0.32 -1.90
7 -0.46 -0.64
Filter +1.86 +2.68
AVERAGE LOSS
PER STAGE 0.54 mg 2.43 mg
NET TOTAL LOSS 1.92 14.34 mg
FILTERS - UNPRECONDITIONED GELMAN TYPE A (OLD TYPE)
EXPECTED FILTER WEIGHT CHANGE - - 0.2 mg
-------
reactions, seem to occur at all sites tested to some degree and
often are larger than normal particulate catches.
The nature and degree of these interferences are illustrated in
the following paragraphs where results obtained by running Andersen
Impactors with a prefilter to remove particulate (Blank Impactor
Runs) are discussed. These tests were run at the outlet of a
hotside ESP on a coal fired boiler (San Juan), the outlet of a
hotside ESP on a cement kiln (Citadel), and at the outlet of a
coldside ESP on a coal fired boiler (Bull Run).
Two series of tests were run at San Juan, one in June 1974 and
the second during January 1975. In the first series of tests
no anomalous weight gains were observed, while blank runs during
the second tests showed large weight gains. (It was subsequently
learned that Gelman Type A filter material had been changed after
the first test series.) Filter samples from the second San Juan
test and the Citadel tests were subjected to several types of
analyses, including carbon-hydrogen and soluble sulfate determinations,
weight loss at 110° and 600°C, and carbon disulfide extraction
followed by gas chromatographic analysis of the extract.
Because of the small size of the filters it was impossible to run
all the tests on each filter; therefore, composites of several
filters were used for some evaluations. Unfortunately, differences
in individual filters could not be detected in these samples.
From the limited data on the Citadel samples there is reason to
suspect the presence of an organic compound on one of the filters
and soluble sulfate on another. Considerably more data would be
required, however, before any firm conclusions could be drawn
regarding the Citadel tests.
-------
Preliminary data from the San Juan filters showed that soluble
•
sulfate levels were significantly higher than obtained from
unused filters from the same batch. Soluble sulfate determinations
were then made on all of the filters from the two sets showing
weight gains. The data in Table II clearly indicate that sulfate
is responsible for the majority of the observed weight gain of
each filter.
The formation of sulfate is presumed to be due to reaction of S02
on basic sites of the filters. Although no filters from the blank
run of the initial San Juan test were available, unused filters
from the same original batch were considerably less basic than
unused filters from the second San Juan test and, therefore, less
likely to cause sulfate formation.
Figure 1 shows in graphical form the results from San Juan Steam
Plant presented in the previous table as well as data from tests
at Bull Run Steam Plant. While the type of filter substrate
was the same the flue gas temperature was quite different (600 F
at San Juan and 300 F at Bull Run). The solid line indicates
a reasonable one to one correspondence. No correlation was found
between the flue gas S02 content and blank weight gains.
Figure 2 shows the weight gains of Blank Andersen Impactor
substrates versus the temperature of the flue gas. With the
exception of the Pre 6/74 point, a linear relationship seems to
hold. The Pre 6/74 represents data from the first San Juan
Steam Plant test shown in Table II. In this case the substrate
was quite neutral compared to the basic substrates in the second
test.
Figure 3 shows the results of tests of several types of 47 mm
filter substrate media tested at different flue gas conditions
and temperatures. It is again interesting to notice that the
-------
TABLE II
SOLUBLE SULFATE ANALYSES OF SAN JUAN FILTERS
Unused samples from batch of first San Juan test (6/74). No
samples from actual blank run (where no weight gains were observed)
are available.
Tgtal mg Reported
Sample No. pH SOi^/filter wt gain, mg
11 6.6 ^0.2
18 6.8 ^0.2
IF 6.9 ^0.2
Set 1 from second San Juan test (1/75)
Unused, perforated, unbaked 9.4 ^0.2
Unused, perforated, baked 9.4 ^0.2
Sll 7.6 4.4 4.98
S12 7.7 4.2 4.96
S13 7.4 5.1 5.58
S14 7.0 5.0 5.56
S15 7.1 5.6 6.06
S16 7.1 5.6 6.10
S17 7.0 6.0 5.94
S18 6.8 5.6 6.04
Set 2 from second San Juan test (1/75)
Unused, perforated, baked 9.3 -v/0.2
Unused, solid, baked 9.7 ^o!3
S20 6.8 7.3 8.34
S21 6.7 5.1 5.30
S22 7.1 4.6 5.22
S23 7.4 5.2 5.40
S24 7.2 4.5 4.80
S25 7.3 5.0 5.44
S26 7.2 4.6 5.24
S27 7.1 4.9 5.50
S28 7.0 4.6 4.86
S2F 8.2 5.5 6.24
a. pH determined after the filter sample was in contact with 10 ml
of distilled water (pH 5.6) for 1 hr.
b. The total soluble sulfate was determined by a Ba(C104)2
titration following a water extraction of the sample.
-------
cf
CO
0
_ BULL RUN STEAM PLANT
SAN JUAN STEAM PLANT
0
7
WEIGHT GAIN, MG,
FIGURE 1, COMPARISON OF WEIGHT OF SULFATE ON BLANK ANDERSEN IMPACTOR
SUBSTRATES AND OBSERVED ANOMALOUS WEIGHT INCREASES,
-------
S 3
D
VJ
- COAL FIRED POWER BOILERS
0 PORTLAND CEMENT KILN
f^
400 500
GAS TEMPERATURE, °F
FIGURE 2, ANOMALOUS WEIGHT INCREASES OF ANDERSEN GLASS FIBER IMPACTION SUBSTRATES AT
DIFFERENT FLUE GAS TEMPERATURES,
-------
1,5
1,0
,5
20J
I I
• GELMAN TYPE A (OLD)
O GELMAN SPECTRO GRADE TYPE A
* MSA 1106 BH
v REEVE ANGEL 900 AF
CEMENT
PLANT
50 PPM
S02
COAL FIRED POWER BOILER
550 PPM S0
0 o ^
I
I
30U
400
500
TEMPERATURE, °F
FIGURE 3, ANOMALOUS WEIGHT GAINS OF VARIOUS 47 MM DIA, GLASS FIBER
FILTERS AT DIFFERENT TEMPERATURES, (60 MINUTE SAMPLES AT
FLOWRATES OF 0,25 ACFM),
11
-------
weight gains seem to depend more on the temperature than on
the concentration of S02 present.
Figure 4 shows the change that preconditioning glass fiber
substrates can make. Also it can be seen that even without
preconditioning there appears to be a saturation limit at which
the weight increases stop.
Figure 5 shows the weight gains of Andersen Impactor substrates
versus exposure time for preconditioned and unconditioned substrates,
The dates at the top of the figure indicate the time at which
these substrates were acquired from Andersen 2000, Inc. The
6/74 Normal Substrates show an increase and leveling off with
exposure time while the 6/74 Preconditioned Substrates show
somewhat smaller weight gain. The "HOT" 6/74 Preconditioned
Substrates would seem abnormal compared to Figure 2 but apparently
the conditioning with hot flue gas reduced the weight gains for
this filter set. The 1/75 Normal Substrates show a possible
linear relationship, although certainly not conclusive. The
Preconditioned 1/75 Substrates indicate a satisfactorily low weight
gain versus exposure time.
The general procedure used to obtain samples and investigate
this problem in more detail was to pump filtered flue gas through
a number of stainless steel 47 mm Gelman filter holders arranged
in series. The first filter served to remove the particulate,
and the remaining five filters were then only exposed to the gas.
Ten different types of filter media were tested at 2 industrial
sites; the outlet of a hotside ESP on a cement kiln (Citadel), and
the outlet of a coldside ESP on a coal fired boiler (Bull Run).
The results of these tests are given in Appendix B and can be
summarized as follows:
• The pH of the filters varied widely from batch to batch
before testing.
12
-------
2,0 —
O - NORMAL
- PRECONDITIONED
0
A
A
5
EXPOSURE, HOURS
FIGURE 4, ANOMALOUS WEIGHT GAIN OF 64 MM DIAMETER REEVE ANGEL 900 AF
GLASS FIBER FILTERS,
13
-------
1,5
CJ3
• NORMAL SUBSTRATES
D PRECONDITIONED SUBSTRATES
A NORMAL SUBSTRATES
A PRECONDITIONED SUBSTRATES
'
6/74
6/74
1/75
1/75
1234
EXPOSURE, HOURS
FIGURE 5, ANOMALOUS WEIGHT GAINS OF ANDERSEN IMPACTOR GLASS
FIBER IMPACT I ON SUBSTRATES,
14
-------
• There was a definite correlation between high initial pH
and weight gains upon testing.
• The pH decreased during testing.
• A large fraction of the weight gain in every case was
found to be the result of sulfate formation on the filter
media.
• For a given temperature, the filters seem to "saturate" and
not gain additional weight after a period of time (2-6 hours).
It is presumed the sulfate was formed by the reaction of sulfur
dioxide with basic sites on the surface of the glass fibers. This
is a phenomenon which was known to occur in ambient sampling,2 but
which had been neglected or ignored in stack sampling.
Two approaches were attempted to avoid the problems of substrate
weight gains. Substrates were preconditioned by long exposure to
the flue gas so that most of the basic sites were neutralized before
using them, and a search was made for substrate materials which
do not react with the flue gas. Preliminary results indicate that
preconditioning overnight reduces the magnitude of the anomalous
weight gains by approximately a factor of ten. Also, in the
tests to date, Teflon, Whatman GF/D and GF/A, improved quartz, and
Reeve Angel 934 AH, are filter media which show little weight
change when exposed to flue gas. More tests should be made at
different sites and a check made on the reproducibility of results
from different batches before firm conclusions are drawn from
these data. None of these materials are currently being used in
cascade impactors and it is possible that their particulate
retention properties could differ slightly from standard substrates.
This could have the effect of altering the impactor calibrations
to some extent.
15
-------
B. COMPARISON OF EPA METHOD FIVE AND IMPACTOR MEASUREMENTS
If impactors performed ideally, their use should yield valid
particulate mass loadings in addition to particle size distribu-
tions. On the other hand, if impactors mass loadings do not
agree with the EPA Method Five it is questionable that the
particle size distribution is truly representative of the flue
gas aerosol.
Due to the sometimes wide disparity between grain loadings
obtained with mass trains and cascade impactors a study has
begun to try to isolate, explain, and to suggest possible solu-
tions to this situation. As can be seen in Figure 6, the
spread in the data is quite large, although the majority
lie within a rather broad band about the perfect agreement
line. The symbols on the figure can be associated with the
following list of sampling sites by using the corresponding number
indicated on the graph. A short explanation is given, if possible,
of possible reasons for the impactor - mass train disparity.
1. Marquette, Michigan Coal Fired Power Boiler
January 8-20, 1973
315°F 7.0% Moisture
No traversing by impactors at outlet. Stratification of dust
likely. S02/S03 gas phase reaction unlikely.
2. Harrisburg, Pennsylvania Refuse Incinerator
May 7-11, 1973
410°F 12% Moisture
Only a four point semi-traverse performed at outlet. Charred
paper present in exit gas stream with sizes greater than
impactor nozzle diameter.
16
-------
0.0001
0.0001
0.001 0.01
GRAIN LOADING - MASSTRAIN , gr/cu.ft.
O.I
1.0
Figure 6. Grain loading by conventional (ASME, EPA)
mass train versus grain loading by cascade
impactor for thirteen sampling locations.
All loadings refer to control device out-
lets except for 9A. Number references as
to location are presented in the text.
17
-------
3. Gorgas, Alabama Coal Fired Power Boiler
July 8-13, 1973
300°F 10% Moisture
Six point semi-traverse performed at outlet. Stratification
of dust likely. S02/S03 gas phase reaction unlikely.
4. Wood River, Illinois Coal Fired Power Boiler
September 12-20, 1973
315°F 15% Moisture
Definite S02/S03 gas phase reactions occurred. No traversing
performed.
5. Gillette, Wyoming Coal Fired Power Boiler
August 1 - September 1, 1973
300°F 15% Moisture
No traversing performed. SOa/SOa gas phase reaction unlikely.
6. Lone Star, Texas Open Hearth Steel Furnace
December 3-14, 1973
200°F 35% Moisture
Cyclonic air flow in duct. Dead air space in center.
Impactors did not traverse, but sampled two points, approxi-
mately 1/3 of duct diameter in from edge. Mass trains traversed
the duct.
7. San Juan, New Mexico Coal Fired Power Boiler
April-May 3, 1974
620°F 8.0% Moisture
Mass trains may have bumped sides of duct and vertical turning
vanes with nozzles pointing up. This could have caused fall
off of large chunks of fly ash. Impactors use 2 and 3 point
semitraverse. Stratification of coarse dust likely because
of turning vanes.
18
-------
8. Memphis, Tennessee Electric Arc Smelting Furnace
July 24-28, 1974
180°F 21.5% Moisture
Impactors used single point sampling. Wet stack might have
caused droplets to be lost in nozzles.
9. Outlet Aluminum Reduction Pot Lines
9A. Inlet August 19-23, 1974
120°F 5.3% Moisture
At this time there is not a good reason for the disparity.
Wet stack conditions allow for possiblity of water droplets
being lost in nozzles. Impactor sampled 2 points in duct ~ h
duct diameter in from duct wall. Condensate from badly
scaled duct walls was entrained in exit gas flow near duct
walls.
10. Buffalo, New York Asphalt Plant
October 12-19, 1974
170°F 8.5% Moisture
Water droplets and fog/mist in the duct. Did not include
material in impactor nozzles in loadings calculated from impactor
data.
11. Meremac, Missouri Coal Fired Power Boiler
November 1-5, 1974
315°F 7.0% Moisture
S02/S03 gas phase reactions did occur and cause impactor
substrate weight gains.
12. Pekin, Illinois Coal Fired Power Boiler
November 17-23, 1974
320°F 9.0% Moisture
SOa/SOs gas phase reactions did occur and cause impactor sub-
strate weight gains.
19
-------
13. Birmingham, Alabama Cement Kiln
December 16-20, 1974
550°F 25% Moisture
Some gas phase reactions may have occurred. Some indication
of condensable hydrocarbons.
Although it is not possible to clearly specify the causes of dis-
agreement between the mass loadings measured by impactors and mass
trains from the data presented above, tentative suggestions can
be made as follows:
1. The dust concentration in industrial flues and stacks
may vary radically with position and time. Vertical stratification
has been frequently observed and sampling positions downstream
from turns may introduce biased samples into either system.
Unfortunately it is difficult and tedious to take a large number
of isokinetic samples with impactors. Temporal variations in the
dust concentration are important because the sampling time for
impactors is dictated by the necessity of obtaining weighable samples
without overloading any single collection surface. From a
practical standpoint, this means that sampling times may vary
from 1 or 2 minutes up to 10 hours. If there are large variations
in concentration due to process changes, the short sampling times
may not give good averages of the true mass emissions. In some
cases this problem can be alleviated by using weighted averages of
a large number of such samples.
2. Gas phase-substrate reactions are a definite problem for
both impactors and mass trains. This is discussed in paragraph IIA
and in Appendix B.
3. It is difficult to always sample isokinetically when using
cascade impactors because the flowrate must remain fixed during
any single test, and because the selection of practical nozzle
sizes is limited. It is general practice to avoid the use of
20
-------
nozzles smaller than 2 mm in diameter even if smaller nozzles
are required for isokineticity. Figure 7, from our field test
data, shows how particle-size distribution measurements are
affected by anisokinetic sampling. In these three tests the
series cyclone arrangement was adjusted to isokinetic sampling
conditions, while the Brink impactor was operated with a 2.5 mm
nozzle, rather than the 1.0 mm nozzle required for isokinetic
sampling. It can be seen that particles larger than about 2 ym
in diameter are oversampled in the Brink and that the particle-size
distribution is severely distorted. In this case (which is not
untypical) the Brink data are very much in error with regard
to size distribution and total grain loading. This illustrates
that isokinetic sampling is essential for proper impactor operation
and that samples taken anisokinetically are not representative
of the flue gas aerosol.
The factors listed above all contribute to the scatter and uncer-
tainty in impactor results. There is clearly a need for a
systematic approach to determine the number and type of measure-
ments needed to obtain representative particle-size distribution
measurements, and to establish substrate preparation procedures
to minimize interferences due to gas phase reactions.
C. ELECTROSTATIC EFFECTS
Electrostatic forces due to charges which may exist on the flue
gas particulate are a potential source of error for any sampling
technique which extracts a sample through a nozzle or probe.
A short study was undertaken to determine the importance of
charge neutralization on aerosols prior to sampling by a cascade
impactor. Figure 8 shows a schematic of the setup of the impactor
and particle counter used in this preliminary experiment. A three-
stage Brink Cascade impactor was employed with a removable polonium
a-source attached to the nozzle of the impactor. A flowrate of
0.1 acfm was maintained through the impactor using the pump in the
21
-------
I03
1
Q
g
LU
>
=>
5
D
o
10
I01
I01
1 I I I I I I I
I 1
O.I
O BRINK
• CYCLONE
A BRINK
A CYCLONE
D BRINK
• CYCLONE
}TEST i —
J-TEST2
]• TEST 3 =
I I I J I I II
I I I I I I II
I I I I I I II
1.0 10.0
FV\RTICLE DIAMETER,
100.0
Figure 7. Comparison of results of size distribution measure-
ments using isokinetic (cyclone) and non-isokinetic
(Brink) sampling rates.
22
-------
Impaction
Plates
Removable Polonium 210
ex-Source
3.49 pm D50
2.03 ym D50
Polonium 210 a-Source
Valve
Absolute
Filter
To
Climet
Particle
Counter
Figure 8. System used to study the importance of charge
neutralization when sampling with impactors at
precipitator outlets.
23
-------
Climet Particle Analyzer, which operates at 0.25 acfm. Make-up
air to the particle counter was supplied by a controlled filtered
air supply. The aerosol exiting the impactor was charge neutralized
with a second polonium 210 a-source to minimize sample and instru-
mental losses due to electrostatic forces between the impactor and
particle counter.
The aerosol was produced by a "hobby" paint sprayer producing a
polydisperse OOP aerosol which was sampled by the impactor at the
outlet of the model wet-wall precipitator at Southern Research
Institute.
Four different configurations of the impactor were tested:
(1) with glass-fiber filter substrates, (2) with bare metal plates,
and (3) with no plates to study the effect of wall loss, and
(4) extractive sampling with a three foot, k" I.D. copper probe.
Each condition was tested with and without the charge neutralizer
prior to the impactor nozzle.
The data from this investigation are presented in Figures 9, 10, and
11. Figure 9 indicates that there is no appreciable effect of charge
neutralization when glass fiber filter substrates are used.
Figure 10 indicates that there is some effect on the number of
particles exiting the impactor with and without charge neutralization
when bare metal plates are used. Figure 11 shows that there is
essentially no wall loss within the impactor which can be attributed
strictly to the electrostatic effects.
The results are plotted another way in Figure 12. The data are
normalized to the concentration measured without the charge
neutralizer in place. Curves are shown for the probe alone, for
the impactor with glass fiber substrates (no probe), and for
the impactor operated with bare metal collection plates (no probe).
Losses in the probe and with bare substrates are seen to be rather
large. For the tests made using glass fiber substrates there was no
appreciable difference in the concentrations measured with and without
the charge neutralizer.
24
-------
o
0)
CO
(0
O
u
U
•H
4J
M
03
10'
T; TulLil^tLJjLLl j ^uJtiliiCJ^-^^li^ ^:''-- "• '-• ^•rT -r"-:"!"'
^pOTiWSFpwLimEf'^t^^ -
j^^-m^mm^t- ^™EMfflm^®itt-^-K1-"l-^i4..|
10'
_.
it3
-:-,^_ i~--^-r ._4.t^.L-,-.-
^41-^-: 7^-jT^T-r-4j
'
- I
LI
i . ;. i -l-"'. r-::-.l^l J-:j{S':r^STg£^_r^--^=E=rT?7--Ur--.=^i-- '^1-3-^-
r. - TI
] Figure 9]
._- .s-T---*:-?
Aerosol charge neutralization study
10J
Brink Cascade Impactor with filter substrates
o Without aerosol charge neutralization priorfi
to nozzle
ijr. D With aerosol charge neutralization prior
nozzle
One standard deviation indicated
.1 .2 .3 .4 .5 .6 .7 .8 .9 1.0 1.1 1.2 1.3 1.4
Particle Diameter, Microns
25
-------
10
U
0)
w
-p
o
U
QJ
H
U
3
M
&
iir/f-i-f' - -.";-:_' if-'-LijIi-r" i i: r
10J
';:-:' ""-• - - "' " "
[• r-
^LJrj-'-j'ib^tt:
tt-m-m
-^^t-Fr^lrjT;
F^^:
J-J-r- -•"-+-!">•
-r-t-4 -'-f-t-'-r—
r
•?—T—T "I—:—T ~j—r "T "•* ~~^~ 7 T ~-~ "I ' ' 4—
-:-:J
-M
TT-:;
rrr*~:"|"^' ~- . ;j-
IggllijJLi
-Figure 10. Aerosol charge neutralization study
Brink Cascade Impactor - Bare Metal Plates
o Without aerosol charge neutralization
prior to nozzle
D With aerosol charge neutralization prior
to nozzle
One standard deviation indicated
10
3-!
:t|ip" .1 .:r:.r±4±i:' Tt
.1 .2 .3 .4 .5 .6 .7 .8 .9 1.0 1.1 1.2 1.3
Particle Diameter, Microns
26
1.4
-------
O
U
0)
rH
U
-H
-P
M
rfl
C^
101
.-•: *. ~r. --
M
0
f
- —] —4—4-' - •*il4j:—
T.:
O
0)
w_ 10:
to
"1 '
10:
-
-T
^I-::.T;T
rfcjrtr:
-tm:
51
— : zj-- • i -
•r-T-Ttrrri
ciph:
~::'^V\
trr:--rl~:
, ' r ' j_V.. j.' ' 1 4— .. -I
J. ! I.:
. -JlTK~nL
'ff'I'T-
•
^d^
mm
,_..-* . r .^ ._ .
:— 1
- — -f-r-•—T--"
j-i-H--t-hH h 1~'~*TT" '.
4:,4^4^Ik-H^hp^:B
-:^ri4~T r-tt!l$43S3r~*£
f_, 1 ^-T , 1_" i 'T-
:f133EE"ife«:
. I ! .—J
U-'t; ..I—i-?"!:
-" -.;'•: .] --
; 7
$a-Tt-.
10
Figure 11. Aerosol charge neutralization study
Brink Cascade Impactor - No collection plates
Wall loss study
• Without aerosol charge neutralization prior
to nozzle
a With aerosol charge neutralization prior to
nozzle
One standard deviation indicated
-! Ir: i f r FF-F?TTI-1:i H ffliSS
0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0 1.1 1.2 1.3 1.4
Particle Diameter, Microns
27
-------
ro
oo
£.U
P
0
g
H^
H 1.5
LU
0
LU
I
wJ
^5^
LU
m It0
i— i
0
1—
LU
O
c^^ r™
>— ' 3
1 1 1 1 1 1 1 1 1
•—3 FOOT LONG, Is INCH I.D, PROBE
/\ 0- — EXIT OF STAGE 2 OF BRINK IMPACTOR
\ WITH METAL SUBSTRATES,
% .«*.
/ \ ° EXIT OF STAGE 2 OF BRINK IMPACTOR
/ \ WITH GLASS FIBER SUBSTRATES,
™^~ * \ ^™™
• \ ^^ ^™ ^^ ^^ ^ ^* v^^^"
O ^^ ^** »^^***
/ x"''\ X'' *""^^
/ si* \ / ^^
J*f' ^^
&/ ^.Q »
~ ^^8^^ o-x*^^*'"^^x^v ~~
^^^ -—^ll*1^" ^^x — ^
o •
I 1 1 1 I l 1 1 1
,2
Figure 12
,6 1,0
PARTICLE DIAMETER,
1,1
1,8
Electrostatic effects in probe and partial Brink Impactor. All
flowrates 0.1 cfm.
-------
These tests indicate that if cascade impactors are used without
charge neutralization at precipitator outlets, serious error may
occur in some cases, while accurate results could be obtained in
certain situations. The study was not extensive however, and the
need for a more quantitative investigation is apparent.
D. REENTRAINMENT DUE TO IMPROPER OPERATION
When impactors are operated at flowrates higher than some critical
value, particle bounce can lead to incorrect sizing of non-cohesive
particulates. This is especially true on the lower impactor stages
where high jet velocities often cause noticeable scouring. Visual
observation indicates that even with grease, erosion and scouring
can occur on the lower stages if jet velocities exceed about 65 m/sec.
The scouring and reentrainment problem can be minimized by reducing
the flowrate of the impactor. This change in flowrate increases
the cut point of each size fraction and decreases the amount of
information that can be obtained in the small particle size range.
However, the addition of a final stage to the impactor with the
proper combination of velocity and jet diameter would make it
possible to regain the information lost at the lower flowrate. This
has been done on the Brink impactor by adding a "6th" stage designed
and built at Southern Research Institute.
Laboratory tests were performed which indicate that flowrates of
about 10 m/sec or less are optimum for the best deposition on impaction
substrates. Figure 13 shows patterns of deposition obtained at
jet velocities of 5 and 12 m/sec, using a hard, dry,-laboratory
aerosol (ammonium fluorescein). The deposition of particles at
12 m/sec shows definite evidence of scouring while the pattern
formed at 5 m/sec is uniform and shows no tendency to smear out.
Figure 14 shows backup filters which contain particles reentrained into
the gas stream from these stages. The magnitude of the scouring
effect depends greatly upon the relationship between the particle size
and the stage D50, being larger for particles which greatly exceed
29
-------
Figure 13A
Figure 13B
Figure 13A. University of Washington (Pilat) Stage 6
0.1 cfm D50 = 1.0 micron. Four micron diameter monodisperse
particles. The uniform particle deposition is apparent as
compared to Figure 13B.
Figure 13B. University of Washington (Pilat) Stage 5
0.5 cfm Dso = 1.0 micron. Four micron diameter monodisperse
particles. Poor particle deposition is evident with scouring
around the point of impact. Figure 14 shows reentrained
particles which should have remained on this substrate.
30
-------
Figure 14.
Left - 0.1 cfm U. of W. back-up filter
Right - 0.5 cfm U. of W. back-up filter
Filter on the right shows reentrained
particles collected after passing 1.0
micron cut off stage shown in Figure 13B,
31
-------
the DSO. In the interpretation of this data, it should be clear
that the impactor was not operated in the conventional manner,
and the University of Washington Impactor was chosen only for
convenience in performing these tests. The object of this
experiment was to demonstrate the importance of maintaining
reasonable jet velocities when sampling with impactors.
32
-------
SECTION III
CYCLONE SAMPLING SYSTEMS
A. GENERAL DISCUSSION
Cyclones have been used less than impactors for making particle
size distribution measurements because they are bulky and give
less resolution. However, in applications where larger samples
are required, or where sampling times with impactors may be
undesirably short, cyclones are better suited for testing than
impactors. Cyclones are also frequently used as precollectors
in impactor systems to remove large particles which might overload
the upper stages, and to give a larger cut point than the first
impaction stage.
Chang3 developed an elaborate system of parallel cyclones which
separates particles into four size fractions. This system is very
large and complicated to operate, and requires extractive
sampling through a probe. Although the system is impractical for
stack sampling the discussion of cyclone design and calibration
included in Chang's report served as a starting point for the
work described in this section. Figure 15 shows a schematic of a
much simpler series cyclone system which was described by Rusanov1*
and is used in the U.S.S.R. for obtaining particle size information.
This device is operated instack, but because of the rather large
dimensions requires an eight inch port for entry.
Two series cyclone systems were designed during this reporting
period. One system was designed for approximately 5 acfm flowrate
and will be operated out of stack as part of a modified Aerotherm
high volume mass train. This system was not constructed at
Southern Research and consequently we have no data pertaining to
its calibration or performance. The second system was designed
33
-------
Inlet Nozzle
Cyclone 3
Cyclone 2
Cyclone 1
Figure 15. Series cyclone used in the U.S.S.R.
for sizing flue gas aerosol particles. **
34
-------
for 1 acfm flowrate and is compact enough to fit through a
6 inch port. Complete calibration and preliminary performance
testing of this system has been done. Both of these systems
have adequate resolution for many purposes and should be much
less susceptible to operator error and gas-substrate interferences
than impactors. The main advantage of each system, however, is
the ability to collect large, sized, samples for subsequent analysis.
B. CYCLONE DESIGN
Our experience has shown that an empirical approach to cyclone
design is useful if one has sufficient data. In this case several
small cyclones had previously been calibrated and the dependence
of the cut points upon the cyclone dimensions and sample aerosol
flowrate determined. This experience was used to estimate the
required dimensions for the new cyclones by relating their perfor-
mance to that of prototypes, and in some cases to use the unmodified
prototypes in the new systems.
In the case of specifications which were outside our range of
"hands on" experience, designs were derived from equations found
in the published literature. In addition to Chang's work there
are several other papers which are useful in developing cyclone
designs.5'* ' 7'8 Figure 16 shows the nomenclature that is used in
Lapple's equation to calculate the DSO, or cut point, for a
cyclone of arbitrary size. Lapple's equation is:9
Dso -
w \* v* o
where D50 = cyclone D5o (ym),
M = gas viscosity (poise),
p = density of particle (gm/cm3),
5
p = gas density (gm/cm3),
N = number of turns made by gas stream in the cyclone
body and cone,
35
-------
Gas
Gas out
T
I
I
T
VV4
D =D /2
s c
Hc=Dc/2
Lc=2Dc
Sc=Dc/8
Z =2D
c c
J =arbitrary,
usually D /4
Section A-A
1
i
Figure 16. Generalized cyclone
design for the application of
Lapple's equation.
-------
V = inlet air velocity (cm/sec), and
c
B = width of cyclone inlet (cm).
c
The square root relationship between D5o and V can be used to
G
predict the cyclone performance over a wide range of flowrates
if the cyclone has been calibrated at a known flowrate. In
this case, we can rewrite Lapple's equation as
Dsod) = /CVT
where D50(l) = the cyclone cut point at flowrate 1,
C = the cyclone calibration constant, and
Vi = aerosol flowrate 1.
Then the relationship between DSO{1) and D5o(2), the cut points
at the two different flowrates, is
DSO(2) = Ds
This relationship was verified experimentally and used to great
advantage throughout this work.
To arrive at final designs for cyclones having the desired cut
points three approaches were used: (1) Figure 16 and Lapple's
equation were used, (2) Previously calibrated prototypes were
used at various flowrates, and (3) Dimensions were selected by
extrapolation for cyclones which were between sizes or related
to the sizes of the prototypes which had been previously tested.
Dimensioned shop drawings and calibration curves, if available,
are shown in Appendix C for each cyclone studied.
C. ONE ACFM SERIES CYCLONE SAMPLING SYSTEM
A series cyclone system was developed during this study to satisfy
the specific objectives of achieving longer sampling times in high
grain loading situations and to collect gram quantities of size
fractionated particulate for chemical analysis.
37
-------
Several cyclone geometries were evaluated during these experiments.
Because of the empiricism involved in the design procedures it
was necessary to calibrate each cyclone and this was done using
the vibrating orifice aerosol generator as a source of standard
aerosols. Monodisperse ammonium fluorescein particles were
sampled and the internal surfaces of the cyclones were then washed with
0.1N NHuOH. The amount of particulate which had been caught on
each surface was determined by using absorption spectroscopy to
measure the concentration of ammonium fluorescein in the wash.
Figure 17 shows calibration curves which were obtained in
this manner for several cyclones which were considered as candidates
for use in the 3-cyclone arrangement.
In conjunction with this calibration, a detailed study was done
to determine the ultimate location of particles on the surfaces
of the SRI-1 cyclone. These surfaces are the nozzle, body, cup,
cap and outlet, and back-up filter. The results of this study
are shown in Figure 18. As smaller particles are sampled, the
amount collected by the back-up filter increases sharply for
diameters of less than 4 microns. The nozzle and body both show
single collection maxima at about 3 vim and then began to decrease
as smaller particles were sampled. The cup shows a minimum at
3 ym and then a large increase in collection efficiency between
2 ym and 1 vim. For particles less than 1 vim in diameter, the
cup collection efficiency decreases rapidly. The cap and outlet
fractional deposition peaked at about 3 vim, decreased to 1 vim,
then began to increase again as smaller particles were sampled.
For the system described here, cyclones SRI-5, SRI-1, and Chang's
T2A were selected, with cut points of approximately 0.5 vim,
0.95 ym, and 2.6 ym, respectively, at a sample flowrate of one
acfm and for particles having a density of 1.35 gm/cm3. Different
cut points could be achieved by the selection of other cyclones or
38
-------
u>
O
0)
•H
O
•r-l
IH
M-l
U
-P
O
0)
100
90
80
70
60
50
40
30
20
10
0
0.1 1.0
Particle Diameter, Micrometers
Figure 17. Collection efficiency vs. particle diameter
5 cyclones 1.0 acfm p=1.35 gm/cc 22°C/ 29.5 "Hg
10.0
• SRI-3
A "T2A"
SRI-1
SRI-5
"TIB"
-------
99.99
Cap and Outlet
Back-up filter
0.3 0.5 0.7 1.0
2.0
Figure 18.
Particle Diameter, Microns
Percentage of total cyclone catch as found on the
cyclone's nozzle, body, cup, cap, and back-up filter
versus particle size. Based on data taken in calibration
of the SRI-1 cyclone shown in Figure 12.
40
-------
by changing the flowrate through the system. As a matter of
fact, additional resolution could be obtained by sampling two
or more times with different sample flowrates each time.
Figure 19 shows the final configuration of the one acfm series
cyclone system. A "stacked" arrangement was used to join the
cyclones in order to minimize the diameter required for sampling
ports. ' This system can be used with six inch ports. A set of
ten nozzles was fabricated to allow isokinetic sampling in flue
gases with velocities from about 10 to 100 feet per second at the
one acfm design flowrate. The cyclone cups and bodies of this
prototype system are black anodized aluminum and the caps and
inlet- plates, which must withstand more stress, are made of
stainless steel. More metal could be machined from the outer
surfaces to lessen the total weight if a system made entirely
of 'stainless steel were desirable.
Although the cyclones have been thoroughly calibrated using
laboratory aerosols, field testing has been limited to date.
A.'field test was conducted at a coal fired steam plant to determine
the usefulness and accuracy of the one acfm series cyclone. The
aptual measurements were performed on the outlet side of an
electrostatic precipitator, approximately 20 feet upstream of the
sjtack. • Gas velocities were on the order of 65 feet per second
at: the sampling location. Six inch ports were available and the
series cyclone was held in a vertical orientation while sampling.*
One test was performed during each of three days. The pertinent
\
data are shown in Table III.
Concurrently during each series cyclone test a Brink Cascade
impactor was used to sample the flue gas to give a comparison
size distribution.
*Treaftis and Tomb have shown that small cyclone performance
is independent of orientation.10
41
-------
to
Figure 19.
One ACFM series cyclone showing the three cyclones and Gelman
back-up filter in the preferred orientation. The nozzles were
designed for isokinetic sampling from 10 fps to 100 fps are
also shown. The large cyclone is 4-3/4 inches long.
-------
TABLE III
SERIES CYCLONE TEST DATA
TEST
DATE
Amb. Pres. ("Hg)
Stack Pres. ("H20)
Gas Velocity (fps)
Imp. AP ("Hg)
Imp. Temp. (°F)
G. M. or
Orifice Temp. (°F)
Flowrate (acfm)
% H20
G. M. Vol. (ft3)
Start Time
Duration (min.)
Nozzle Dia. (mm)
WEIGHT GAIN (mg)
Cyclone 1
Cyclone 2
Cyclone 3
Filter
3/11/75
29.31
-1.6
65
3.2
290
74
.942
7.5
20.893
4:03
30
5.56
777.62
77.24
18.86
5.02
3/12/75
28.99
-1.2
65
3.1
280
64
.908
7.5
26.650
11:00
40
5.56
483.14
77.04
19.70
4.96
3/13/75
29.00
-1.0
65
3.7
300
80
.911
7.5
41.200
12:10
60
5.56
4276.21
584.90
123.76
19.56
43
-------
Actual sampling with the series cyclone was very simple. The
flowrate was maintained at its proper value using a dry gas
meter. The cyclone was allowed to warm up for 45 minutes before
sampling began and a nozzle was chosen to obtain isokinetic
sampling. For the Brink Cascade Impactor a 2.5 mm nozzle was
the smallest one available. At the 0.03 acfm flowrate a 1 mm
nozzle would have been necessary for isokinetic sampling. Because
of this, the larger particles were oversampled and a higher than
normal grain loading was indicated by the Brink, although the
small particle end of the distribution was apparently not affected
by the nonisokinetic sampling. Figures 20, 21, and 22 show the
results of the three tests. In each figure the series cyclone
and Brink agree well up to about 3 microns. Beyond this point
they diverge due to the anisokineticity of the Brink sample.
Figures 23 and 24 show micrographs of the material collected in
each cyclone and on the backup filter. Very effective size
segregation by the cyclones is clearly shown in these figures.
Great care is needed in cleaning the cyclones to assure that dust
is not lost. Cleaning the elbows connecting the cyclones is
important because a significant amount of material collects there.
The cyclone was brushed out to recover most of the collected dust,
although small amounts of residual dust remained on the brush and
cyclone walls. In these tests, the amount of dust lost was in-
significant in comparison to the large amount of dust collected.
This could represent a problem at sites where smaller samples
are collected.
These preliminary results indicate that the series cyclone concept
is very useful as a mechanism for obtaining particle size distri-
bution data and size fractionated samples, and that the system is
convenient to operate and somewhat more foolproof than cascade
impactors.
44
-------
1.0 10.0
Particle Diameter, Micrometers
10'
100.0
Figure 20. Cumulative mass loading vs. Particle diameter
March 11, 1975
-•• — Brink Cascade Impactor, 0.03 ACFM
* Series Cyclone, 1.0 ACFM
o
s
•H
W
0)
rt
s
0)
>
•H
-P
id
u
45
-------
s
I
•H
0)
•H
-P
CJ
0.1
Figure 21
1.0 10.0
Particle Diameter, Micrometers
100.0
Cumulative mass loading vs. Particle diameter
March 12, 1975
— -• Brink Cascade Impactor, 0.03 ACFM
A Series Cyclone, 1.0 ACFM
46
-------
o
*.
tr>
• <-t
Rj
o
en
I
0)
•H
4J
rH
3
u
1.0 10.0
Particle Diameter, Micrometers
100.0
Figure 22. Cumulative mass loading vs. Particle diameter
March 13, 1975
• Brink Cascade Impactor, 0.03 ACFM
A Series Cyclone, 1.0 ACFM
47
-------
CO
Cyclone 1
D5 o = 2.5 ym
Cyclone 2
D5 o = 1.0 ym
Figure 23. Photomicrographs of cyclone catches acquired at Bull Run Steam Plant,
-------
VD
I
•'•• *
I
Back-Up Filter
Cyclone 3
D5 o = 0.5 urn
Figure 24. Photomicrographs of cyclone and back-up filter catches acquired at
Bull Run Steam Plant.
-------
SECTION IV
APPENDICES
APPENDIX A - Andersen Impactor Weight Loss Study
APPENDIX B - Impactor Glass Fiber Substrate Weight Gains
APPENDIX C - Cyclone Design
APPENDIX D - Conversion Table for Units
50
-------
APPENDIX A
ANDERSEN FILTER SUBSTRATE WEIGHT LOSS STUDY
This study was undertaken to determine if weight losses from
the fiber filter substrates of the Andersen Impactor were
significant and to what extent they might affect field test measure-
ments. A procedure was set up to check for errors in the mechanics
of the weighing, moisture absorption by the substrates, errors
due to handling, and errors resulting from running an impactor.
These will be explained in detail later.
The filter substrates for the Andersen Impactors, with the
exception of the back-up filters, were baked for approximately
18 hrs in an oven at a temperature of 240 C. The back-up
filters were baked for about 6 hrs at the same temperature.
The substrates were from normal Andersen substrate stock.
The back-up filters were cut from glass fiber filter stock.
Mine Safety Appliance No. CT-75428.
After all substrates and filters had been baked, they were
allowed to cool to ambient temperature in room air. They then
were left sitting out for 4 days at room temperature and humidity
while their foil holders were cut. After being folded and
placed in the foil holders, the substrates, including
foil holder, were weighed. The process of placing the substrates
in the foils and then weighing them took 2 days, making a
total of 6 days elapsed time from the baking procedure. All
substrates were then placed in desiccators.
The balance used in the study was a Cahn Model G-2 Electro-
balance. The total weight of the substrates and foils was
greater than 200 mg but this scale was used with counter-
weights (tare weights) for improved accuracy. To keep
moisture at a minimum, the weighing chamber was loaded with
51
-------
a small pan containing Drierite, anhydrous CaSOi,. This is
our normal procedure in weighing Andersen substrates.
The average weights for the Andersen glass fiber substrates and
back-up filters were also measured so that the weight losses
could be evaluated relative to this total weight. Due to the
design of the Andersen Impactor, the filter substrates alter-
nate in pattern from stage 1 to stage 8; stage filters being
alike for stages 1, 3, 5, and 7, but different for stages 2, 4,
6, and 8. To obtain an average weight/ 50 substrate filters
of the odd type (1, 3, 5, 7), 50 of the even type (2, 4, 6, 8),
and 50 back-up filters were baked at 240°C for 6 hours, desiccated
for 18 hours, and then weighed on a Mettler Gram-atic balance.
The average weights are:
Odd Filter Stage Substrates (1, 3, 5, 7) 189.54 mg average,
Even Filter Stage Substrates (2, 4, 6, 8) 178.92 mg average, &
Back-up Filter 215.04 mg average.
Weighing-Balance Errors and Moisture Absorption Errors
Three sets of the substrates, Nos. 13N, 15N, and 16N, were
used to check the possibility of errors being made just in
the process of weighing the substrates. Also, this was a
check on the reproducibility of the balance in weighing an
object. In addition to sets 13N, 15N, and 16N, two
stainless steel ferrules and one flat washer were included
in the weighing as standards, since these should neither
lose nor gain significant mass due to absorption. They
were cleaned in benzene to remove any grease and kept in
clean containers in a desiccator throughout the test. Tweezers
were always used in handling the ferrules, the washer, and
the foil-wrapped substrates.
The results of the ferrule and washer weighings are given in
Table A-l. Approximately one weighing of each was made each
52
-------
TABLE A-l
STEEL FERRULE AND WASHER WEIGHT CHANGES UPON DESICCATION
A
B
C
Initial
Weight,
mg
494.12
422.50
413.80
Cumulative
6
+ .02
.00
.00
24
.00
.02
.02
30
+ .02
.00
.02
48
.00
.00
.00
Hours
54
+ .02
.00
.02
of Desiccation
72
.02
.02
+ .02
78
.00
.02
.02
96
.00
+ .02
+ .02
120
+ .02
.02
.02
144
+ .02
.00
+ .02
Weight change in mg as compared to initial weight after indicated
number of hours of desiccation.
A & B are stainless steel Swagelok ferrules; C is a zinc-plated
steel washer.
All weight changes are negative unless otherwise noted.
53
-------
day during the study. (All weights in all tables are negative
except those preceded by a + sign). The two ferrules were
made of 316 stainless steel and are listed as "A" and "B"
in the table. The washer was made of zinc-plated steel and is
listed as "C". The true weight of A, B, and C can be derived
since the tare weight used with the Cahn balance was weighed
also. Its weight was 366.7 mg, as weighed on a Mettler
Gram-atic balance.
True weight = [Cahn reading x 200 + 366.?]
mg
The "Cahn reading" is the reading taken from the mass dial
as a fraction of full scale, which is unity; 200 is the range
multiplier; and 366.7 is the tare weight correction. Using
a mass reading of 0.4321 as an example:
[0.4321 x 200 + 366.7] mg = 453.1 mg
True weights were not generally calculated because weight
losses could be monitored by differences of direct readings
on the Cahn.
The substrate sets 13Nf 15N, and 16N were the control group
of substrates. They were weighed periodically while the other
tests were being conducted. They served as a check on desic-
cation losses, balance calibration, and zero shift. These
sets remained wrapped in their foil covers throughout the
tests to avoid any loss of the filter material. The results
of weighing these sets are shown in Table A-2. The amount of
desiccation prior to each weighing is indicated in the table.
The table is structured as it is because Andersen substrates
are of two types, varying between even and odd impactor plates.
54
-------
TABLE A-2
ANDERSEN SUBSTRATE CUMULATIVE WEIGHT CHANGE DUE TO DESICCATION
(SETS 13N, 15N, 16N)
Desiccation Period
Initial
-
-
-
6 hrs
0.12 mg
0.03 mg
12
0.12 mg
0.04
12
0.05 mg
(0.01 mg)
3
24 hrs
0.15
0.03
12
0.15
0.03
12
0.09
(0.03)
3
48 hrs
0.14
0.02
12
0.14
0.02
12
0.09
(0.02)
3
72 hrs
0.14
0.03
12
0.14
0.02
12
0.09
(0.01)
3
96 hrs
0.15
0.02
12
0.15
0.03
12
0.07
(0.02)
3
120 hrs
0.15
0.02
12
0.16
0.02
12
0.10
(0.03)
3
Odd Numbered
Stage Substrates:
Average
o
No. of Filters
Even Numbered
Stage Substrates:
Average
a
No. of Filters
Back-Up Filters:
Average
a
No. of Filters
Cumulative weight change as compared to initial weight after indicated number of hours
of desiccation.
All weight changes are negative unless otherwise noted.
Parentheses indicate loss of accuracy due to insufficient data.
-------
The substrates are cut to fit the impactor plate without
obstructing the jets. Any difference in surface area of
the filter might cause a different evaporation amount. The
back-up filter is also separated from the even and odd substrates.
Table A-3 gives the raw data for each set of substrates by stage.
Table A-4 indicates the moisture pickup of substrates which, had
been baked and desiccated as previously described. Three back-up
filters were used, and instead of Drierite, a container of water
was placed in the balance weighing chamber to give maximum
humidity. Before the test, the balance was zeroed and calibrated.
Handling Losses
Handling losses in this case mean losses occurring in the
process of loading and unloading an impactor. The substrates
were taken from the foil covers and loaded into an impactor
just as if it were about to be run; but immediately after
assembling it, the impactor was unloaded and the substrates
placed in the foil covers.
Six sets of substrates were used. Three sets (ION, UN, 12N)
were treated entirely as typical substrates for a normal run.
The other three were cleaned. Each filter substrate was care-
fully checked for loose fiber pieces, which were removed,
and then each filter was blown off with dustless freon. These
were designated as the "clean" substrates (7N, 8N, 9N).
All six sets were desiccated for 48 hrs, weighed, loaded into
impactors, unloaded, and then placed in the desiccator for 48
additional hours.
During the unloading process, the jet plates were brushed down
to remove any bits of fiber which remained on them. Any
56
-------
TABLE A-3
CUMULATIVE CHANGES IN WEIGHT DUE TO DESICCATION FOR
ANDERSEN SUBSTRATES BY INDIVIDUAL SET
_________^_ Cumulative Desiccation Period
6 hrs 241
Set 13N
Set 15N
Set 16N
1
2
3
4
5
6
7
8
F
1
2
3
4
5
6
7
8
F
1
2
3
4
5
6
7
8
F
6 hrs
0.10 mg
0.04
0.10
0.08
0.10
0.10
0.08
0.10
0.06
0.16
0.14
0.10
0.14
0.16
0.14
0.16
0.14
0.06
0.14
0.16
0.12
0.16
0.14
0.10
0.10
0.08
0.04
24 hrs
0.15 mg
0.12
0.12
0.16
0.16
0.16
0.12
0.12
0.08
0.16
0.16
0.12
0.14
0.16
0.12
0.16
0.14
0.06
0.20
0.20
0.18
0.18
0.20
0.10
0.14
0.18
0.12
48 hrs
0.12 mg
0.10
0.12
0.14
0.18
0.16
0.14
0.12
0.10
0.14
0.14
0.12
0.14
0.18
0.12
0.14
0.16
0.06
0.14
0.18
0.16
0.16
0.18
0.12
0.12
0.12
0.10
72 hrs
0.10 mg
0.12
0.14
0.14
0.18
0.16
0.14
0.12
0.08
0.14
0.16
0.10
0.12
0.16
0.12
0.14
0.14
0.08
0.14
0.18
0.14
0.16
0.18
0.10
0.12
0.12
0.10
96 hrs
0.12 mg
0.10
0.12
0.16
0.18
0.18
0.12
0.10
0.10
0.16
0.18
0.12
0.14
0.16
0.12
0.12
0.16
0.06
0.16
0.18
0.18
0.18
0.16
0.12
0.16
0.16
0.06
120 hrs
0.14 mg
0.14
0.14
0.16
0.20
0.18
0.14
0.14
0.12
0.16
0.18
0.12
0.14
0.18
0.14
0.16
0.16
0.06
0.16
0.20
0.16
0.16
0.12
0.12
0.14
0.14
0.12
All weight changes are negative unless otherwise noted.
57
-------
TABLE A-4
MOISTURE ABSORPTION BY THREE ANDERSEN BACK-UP FILTERS
Elapsed Time (min) Cumulative Weight Changes, mg
#1 #2 #3
0 -
5 +.10 +.08 +.08
10 +.10 +.08 +.08
15 +.12 +.12 +.14
20 +.12 +.12 +.14
25 +.12 +.12 +.14
30 +.12 +.12 +.14
58
-------
loose fiber was placed with the substrate nearest it. Sets
9N and UN which were loaded into Andersen No.507 showed signs
of the substrates being partially cut through by the metal
o-rings. These two sets were the only ones with noticeable
pieces of fiber left on the jet stages, which were brushed
onto corresponding substrates.
Table A-5 gives average losses of the cleaned-up substrates.
Table A-6 gives the same for the normal substrates. Tables A-7
and A-8 give individual stage losses for all six sets.
Running Losses
Six sets were used for determination of weight change while
sampling filtered air. They were treated as normal substrates
in loading them into the foil covers. No special cleaning of
the filters was done. All sets were desiccated for 48 hours.
Three of the sets (4N, 5N, 6N) were loaded into impactors
after desiccation and the impactors were run at 0.5 acfm
for 6 hrs at room temperature, 24° C. A Gelman 47-mm filter
holder and filter were attached to the inlet of the impactors
to insure that clean air was sampled.
The other three sets (IN, 2N, 3N) were set up in the same
way and run at 0.5 acfm for 6 hrs in an oven at 120°C.
When these sets of substrates were unloaded, the same brushing
technique as described above was used. The substrates from the
impactors run at room temperature were similar to those that
had just been loaded and unloaded. The Andersen No. 507
impactor showed slight gasket cuts on several of the substrates.
Andersens No. 229 and 231 had no cut substrates, but were
crimped as required for a good seal.
59
-------
TABLE A-5
DESICCATION AND HANDLING WEIGHT CHANGE OF
"CLEAN" ANDERSEN SUBSTRATES (7N, 8N, 9N)
Desiccation Weight Change Average o Number in Average
Odd numbered substrates 0.02 mg 0.06 mg 12
Even numbered substrates 0.10 0.04 12
Back-up filters +0.01 (0.05) 3
Handling Weight Change
Odd numbered substrates 0.06 mg 0.04 mg 12
Even numbered substrates 0.09 0.04 12
Back-up filters 0.04 (0.06) 3
Handling Weight Change
by Stage
Stage
1 0.03 mg (0.04) mg 3
2 0.08 (0.02) 3
3 0.05 (0.02) 3
4 0.09 (0.03) 3
5 0.09 (0.05) 3
6 0.11 (0.01) 3
7 0.07 (0.03) 3
8 0.08 (0.05) 3
F 0.04 (0.06) 3
Numbers in parentheses indicate insufficient data for accurate a.
All weight changes are negative unless otherwise noted.
60
-------
TABLE A-6
DESICCATION AND HANDLING WEIGHT CHANGE OF
"NORMAL" ANDERSEN SUBSTRATES (ION, UN, 12N)
Desiccation Weight Change
Odd numbered substrates
Even numbered substrates
Back-up filters
Average
0.02
0.00
0.08
0.04
0.03
(0.07)
Number in Average
12
12
3
Handling Weight Change
Odd numbered substrates
Even numbered substrates
Back-up filters
0.06
0.04
+0.01
0.06
0.04
(0.08)
12
12
3
Handling Weight Change
by Stage
Stage
1
2
3
4
5
6
7
8
F
0.10
0.07
0.08
0.05
0.01
0.01
0.03
0.03
+0.01
(0.05)
(0.05)
(0.02)
(0.07)
(0.02)
(0.02)
(0.08)
(0.02)
(0.08)
3
3
3
3
3
3
3
3
3
Numbers in parentheses indicate insufficient data for accurate o.
All weight changes are negative unless otherwise noted.
61
-------
TABLE A-7
DESICCATION AND HANDLING WEIGHT CHANGE OF
"CLEAN" ANDERSEN SUBSTRATES BY SETS
Weight Change Weight Change After
During Desiccation Impactor Loading
Set 7N
Andersen 229
1 0.12 mg 0.00 mg
2 0.18 0.06
3 0.02 0.04
4 0.16 0.01
5 0.02 0.08
6 0.08 0.12
7 0.02 0.04
8 0.14 0.04
F 0.04 +0.02
Set 8N
Andersen 231
1 0.06 0.08
2 0.10 0.10
3 0.04 0.04
4 0.12 0.12
5 0.00 0.14
6 0.06 0.12
7 0.06 0.06
8 0.06 0.06
F +0.02 0.04
Set 9N
Andersen 507
1 +0.04 0.02
2 0.06 0.08
3 0.00 0.08
4 0.06 0.14
5 +0.02 0.04
6 0.10 0.10
7 +0.10 0.10
8 0.04 0.14
F +0.06 0.10
All weight changes are negative unless otherwise noted.
62
-------
TABLE A-8
DESICCATION AND HANDLING WEIGHT CHANGES OF
"NORMAL" ANDERSEN SUBSTRATES BY SET
Weight Change Weight Change After
During Desiccation Impactor Loading
Set ION
Andersen 229
1 0.00 mg 0.12 mg
2 +0.04 0.12
3 0.02 0.10
4 0.00 0.12
5 +0.02 0.04
6 +0.06 0.04
7 +0.06 0.12
8 +0.02 0.06
F 0.00 0.08
Set UN
Andersen 507
1 0.04 mg 0.14 mg
2 +0.02 0.04
3 0.00 0.08
4 0.00 0.04
5 0.04 0.00
6 0.02 0.00
7 0.00 0.02
8 0.04 0.02
F 0.10 +0.06
Set 12N
Andersen 231
1 0.10 mg 0.04 mg
2 0.02 0.04
3 0.04 0.06
4 0.02 +0.02
5 0.02 0.00
6 0.02 0.00
7 0.02 +0.04
8 0.00 0.02
F 0.14 +0.04
All weight changes are negative unless otherwise noted.
63
-------
The substrates run at 120°C showed more cutting. Again,
Andersen No. 507 had substrates which were cut by the seals
but slightly more severely. There were more pieces of
fiber on the jet stage than previously. Substrates from
Andersens No. 229 and 231 this time showed slight cuts in
some of the stages. This may have been due to heating the
impactor, or possibly to overtightening of the impactors,
although an attempt was made to tighten them equally each
time.
After sampling filtered air for 6 hrs, each set was unloaded,
brushed, foiled, and desiccated for 48 hours before weighing.
The results are shown in Tables A-9 through A-12.
Miscellaneous
In Table A-13, the results of two blank sets of substrates
loaded normally and run under stack sampling conditions at
an aluminum reduction plant with a Gelman filter in front
of the impactor, are given. The stack temperature was 125°F
and the running times are listed in the table. Also shown
is the average of a blank set run at a hot side precipitator
on a coal-fired boiler.
For comparison and to gauge the significance of impactor
weight losses, Table A-14 includes some averages of typical
Andersen Impactor stage net weight changes, including
possible filter weight loss, observed in sampling several
types of industrial particulate sources.
Also investigated was the possibility of using Teflon filter
membranes as substrates. The type of Teflon filter used was
64
-------
TABLE A-9
DESICCATION AND SAMPLING WEIGHT CHANGE OF ANDERSEN SUBSTRATES
(4N, 5N, 6N) AFTER SAMPLING FILTERED AIR AT 0.5 ACFM FOR SIX
HOURS AT 24 C
Weight Change During
Desiccation Average o
Odd numbered substrates 0.16 mg 0.05
Even numbered substrates 0.14 0.03
Back-up filters 0.17 (0.03)
Number in Average
12
12
3
Weight Changes Due
to Sampling
Substrates 1-4 average 0.06 mg 0.03
Substrates 5-8 average 0.07 0.03
Odd numbered substrates 0.07 0.03
Even numbered substrates 0.06 0.02
Back-up filters +0.03 (0.03)
12
12
12
12
3
Sampling Weight Change
by Stage
Stage
1
2
3
4
5
6
7
8
F
0.08 mg
0.06
0.05
0.05
0.06
0.06
0.07
0.07
+0.03
(0.02)
(0.02)
(0.05)
(0.03)
(0.02)
(0.02)
(0.03)
(0.03)
(0.03)
3
3
3
3
3
3
3
3
3
All weight changes are negative unless otherwise noted.
Parentheses indicate loss of accuracy due to insufficient data.
65
-------
TABLE A-10
DESICCATION AND SAMPLING WEIGHT CHANGE OF ANDERSEN SUBSTRATES
(IN, 2N, 3N) AFTER SAMPLING FILTERED AIR AT 0.5 ACFM FOR SIX
o;
HOURS AT 120 C
Weight Change During
Desiccation Average o
Odd numbered substrates 0.14 mg 0.04
Even numbered substrates 0.15 0.05
Back-up filters 0.15 (0.06)
Number in Average
12
12
3
Weight Change Due
to Sampling
Substrates 1-4 average 0.15
Substrates 5-8 0.17
Odd numbered substrates 0.17
Even numbered substrates 0.15
Back-up filters +0.03
0.04
0.04
0.04
0.04
(0.08)
12
12
12
12
3
Sampling Weight Change
by Stage
Stage
1 0.15
2 0.11
3 0.17
4 0.13
5 0.14
6 0.17
7 0.19
8 0.19
F +0.03
(0.01)
(0.02)
(0.04)
(0.02)
(0.00)
(0.02)
(0.06)
(0.01)
(0.08)
3
3
3
3
3
3
3
3
3
All weight changes are negative unless otherwise noted.
Parentheses indicate loss of accuracy due to insufficient data.
66
-------
TABLE A-11
DESICCATION AND SAMPLING WEIGHT CHANGES OF ANDERSEN SUBSTRATES
BY SETTING AFTER SAMPLING FILTERED AIR AT 0.5 ACFM FOR SIX
ur\i-mc aT ") L C
HOURS AT 24 C
Weight Change During Weight Change Due
Desiccation to Sampling
Set 4N
Andersen 229
1 0.20 mg 0.08 mg
2 0.12 0.04
3 0.16 0.08
4 0.16 0.06
5 0.08 0.08
6 0.08 0.08
7 0.04 0.12
8 0.12 0.08
F 0.14 0.00
Set 5N
Andersen 231
1 0.22 mg 0.06 mg
2 0.12 0.08
3 0.18 0.00
4 0.16 0.02
5 0.18 0.06
6 0.12 0.06
7 0.18 0.08
8 0.18 0.10
F 0.20 +0.04
Set 6N
Andersen 507
1 0.20 mg 0.10 mg
2 0.18 0.06
3 0.18 0.08
4 0.16 0.08
5 0.12 0.04
6 0.12 0.04
7 0.18 0.10
8 0.14 0.04
F 0.18 +0.06
All weight changes are negative unless otherwise noted.
67
-------
TABLE A-12
DESICCATION AND SAMPLING WEIGHT CHANGES OF ANDERSEN SUBSTRATES
BY SET AFTER SAMPLING FILTERED AIR AT 0.5 ACFM FOR SIX HOURS
AT 120°C
Weight Change During
Desiccation
Set IN
Andersen 229
1
2
3
4
5
6
7
8
F
0.14 mg
0.12
0.12
0.18
0.16
0.16
0.14
0.14
0.10
Weight Change Due
to Sampling
0.16 mg
0.10
0.16
0.10
0.14
0.14
0.24
0.18
+ 0.06
Set 2N
Andersen 231
1
2
3
4
5
6
7
8
F
0.10 mg
0.10
0.06
0.12
0.16
0.10
0.08
0.10
0.12
0.14 mg
0.14
0.14
0.14
0.14
0.18
0.12
0.18
+0.10
Set 3N
Andersen 507
1
2
3
4
5
6
7
8
F
0.16 mg
0.20
0.16
0.20
0.18
0.22
0.22
0.20
0.22
0.20 mg
0.10
0.22
0.14
0.14
0.18
0.22
2.20
0.06
All weight changes are negative unless otherwise noted.
68
-------
TABLE A-13
WEIGHT CHANGES OF ANDERSEN SUBSTRATE AFTER SAMPLING FILTERED
EFFLUENT FROM A WET PRECIPITATOR AT AN ALUMINUM REDUCTION PLANT.
125°F GAS TEMPERATURE
1
2
3
4
5
6
7
8
average
Sampling Time
240 nun. 103 min.
+0.06 mg
0.04
0.02
+0.08
,12
,08
0.
0,
0.12
+0.02 mg
0.04
+0.04
0.04
0.16
0.10
0.10
0.04 mg
All weight changes are negative unless otherwise noted.
Average weight change of Andersen substrates after sampling
filtered effluent from a hot side precipitator on a coal
fired boiler.
Temperature = 635 F
Sampling Time = 90 min.
Average = 0.02 mg
69
-------
TABLE A-14
TYPICAL AVERAGE AMOUNTS OF COLLECTION FOR AN
ANDERSEN IMPACTOR
Stage Site and Amount in mg
^•^Mta^B
1
2
3
4
5
6
7
8
F
1
0
0
0
0
1
1
2
22
A
.13
.78
.84
.80
.95
.25
.84
.72
.40
B
0.
0.
0.
0.
0.
0.
1.
1.
4.
37
27
30
28
32
29
10
68
51
C
1.70
1.29
1.08
1.17
2.12
4.96
5.06
2.64
5.53
15
6
11
7
8
5
2
0
3
D
.79
.94
.62
.23
.73
.56
.99
.66
.85
E
0.59
0.39
0.35
0.34
0.25
0.48
1.08
1.00
3.91
F
6.
2.
2.
1.
2.
5.
4.
2.
3.
74
45
08
63
52
30
99
52
01
G
9.70
6.97
8.23
8.93
12.62
14.48
8.61
4.39
8.56
Sampling
Time (min). 431 921 210 38 371 140 255
Stack Temp.
(°C) 121 41 335 166 66 152 167
A. Wet scrubber at submerged arc ferro-alloy furnace.
B. Wet precipitator on aluminum reduction pot lines.
C. Hot side precipitator on coal fired boiler.
D. Precipitator on coal fired boiler.
E. Wet scrubber on open hearth furnace.
F. Pilot scale precipitator on coal fired boiler.
G. Precipitator on coal fired boiler.
70
-------
Chemware filter membranes made of Zitex, which is a fibrous,
porous form of pure TFE-Teflon/Halon. They are produced by
Chemplast, Incorporated. A 75-mm, extra fine filter was
used. It was desiccated without being wrapped with anything
and weighed at various intervals. Then, it was baked for 6 hrs
at 205°C and weighed again, desiccated for 24 hrs, and weighed
one final time. The baking appeared to cause no damage to
the filter. Results are given in Table A-15.
CONCLUSIONS
Several conclusions can be drawn from this study of Andersen
filter substrates:
1. From Table A-l it can be seen that the Cahn Model G-2
Electrobalance used in this study as well as our normal
laboratory and field test work is quite stable in
day-to-day operation. Repetitive weighings of metal
objects agreed within 0.02 mg over a six day period.
This degree of uncertainty represents a very small
percentage of normal stage weights (e.g., Table A-14).
2. Tables A-2 and A-3 indicate that the substrates lose about
.10-.20 mg of absorbed moisture when desiccated for 24
hours at room temperature. Although further desiccation
does not remove additional moisture, losses do occur
when clean air is pulled through the impactors. These
losses are approximately .05-.10 mg for six hours
testing at 0.5 acfm at room temperature (24 C), and
.10-.20 mg at elevated temperatures (120 C). These
results are summarized in Tables A-9, A-10, A-ll and A-12.
Although a definitive reason for this has not been
71
-------
TABLE A-15
CUMULATIVE WEIGHT CHANGE OF TEFLON FILTER SUBSTRATE
AFTER DESICCATION AND BAKING
Original weight - 509.08 mg
Hours of Desiccation
24 -0.04 mg
48 -0.04
72 -0.04
96 -0.02
After baking 6 hrs
at 205°C -0.26
After 24 hrs desiccation -0.22
72
-------
established, one possibility is that the moving air
dries the substrates more than normal desiccation.
This explanation is consistent with the larger losses
which occur when clean, hot air is drawn through the
impactors.
3. Table A-4 confirms the reabsorption of water vapor by
previously desiccated substrates upon exposure to a
humid atmosphere. Within % hour, the weight lost
during a desiccation period of 24 hours is regained
(^.10-.20 mg). For this reason, the substrates and
stage catches from impactor tests should be desiccated
for at least 24 hours before weighing, and the weighing
should be done immediately upon removing each substrate
from the desiccator.
4. Tables A-5, A-6, A-7 and A-8 indicate that if careful
attention is maintained in weighing, loading, and
unloading substrates, then the handling losses are
not significantly better if the substrates are
"cleaned" before use. Poor handling and impactor
cleaning techniques would, of course, tend to magnify
these errors.
5. The field test data shown in Tables A-13 and A-14 indicate
that weight losses which may occur during impactor
measurements can amount to more than 50% of the particu-
late catch for any given stage, but more often are less
than 20%. The errors are most significant when
sampling sources with extremely low grain loadings
under circumstances that do not permit extended sampling
times.
73
-------
The data tabulated in Table A-15 shows that the weight
losses as indicated in Tables A-9 and A-10 can be either
significant or insignificant depending on the amount
of particulate collected per stage. Partial compen-
sation for this weight loss could be obtained by
adding approximately 0.1 mg per stage.
Although the weight losses were progressively higher
for impactor loading and unloading, sampling clean
cool air, and sampling clean warm air respectively,
it was not proven whether the losses were due to
drying, to loss of filter material, or both. One
can generalize all the results, including field tests,
as indicating that with prebaking, 24 hour desiccation,
careful weighing and loading, careful unloading and
impactor cleaning, post test desiccation for 24
hours, and careful final weighing, the weight
losses can be minimized, and will probably be
less than .20 mg per stage. This may or may not
seriously degrade the accuracy of the particle size
measurements, depending upon the amount of material
collected. It would be desirable to collect 2-3 mg
of material on every substrate, but this could result
in excessively long sampling times at some locations,
or overloading of other stages for some particle
size distributions.
74
-------
APPENDIX B
GLASS FIBER SUBSTRATE WEIGHT GAINS
Glass fiber filter media when exposed to stack-type environments
are subject to possible gas phase reactions with the flue gases.
In inertial impactors using glass fiber material as substrates,
the most important result of the reactions is the possible weight
changes which may occur in the substrates. Ideally, the only
weight change in a substrate should be the weight change caused
by the collection of particulate matter from the flue gas, but
field test results have shown that glass fiber is not the
"ideal" substrate, even though it may be a satisfactory one, if
the gas phase reaction weight change problem can be brought under
control.
With this problem in mind, tests were designed to gain an
understanding of the weight changes and to facilitate the selection
of a sufficiently inert filter material for impactor substrates.
A suitable substrate material would be one which has stable weight
characteristics and is mechanically strong to resist cutting,
tearing, and loss of material.
Several types of filter material were obtained from commercial
suppliers for testing: Gelman Types A, AE, and SpectroGrade
glass fiber filter material, Mine Safety Appliance 1106-BH glass
fiber filter material; Reeve-Angel 900 AF and 934 AH glass fiber
material; Whatman GF/A and GF/D glass fiber material; Chemplast
teflon filter membranes; and Pallflex Tissuquartz 2500 QAD (See
Table B-I). The method chosen to test the filter media was to
expose them directly to the flue gases for time intervals characteris-
tic of an impactor run. Several Gelman stainless steel 47 mm filter
75
-------
TABLE B-l
FILTER TYPES
Gelman Type A GA
Gelman Type AE GAE
Gelman SpectroGrade SA
Mine Safety Appliance 1106 BH MSA 1106 BH
Reeve-Angel 900 AF RA900AF
Reeve-Angel 934 AH RA 934 AH
Whatman GF/A GF/A
Whatman GF/D GF/D
Chemplast Teflon Filter Teflon
Pallflex Tissuquartz 2500 QAD Quartz
76
-------
holders were used to accomplish this. These holders were assembled
as a series filter arrangement and run just as an Andersen Stack
Sampler would be run. The first filter holder was a pre-filter
which removed the particulate. The remainder of the filters,
each in its holder, were exposed only to the flue gas.
Before the weighed filter samples were loaded in the Gelman holders,
they were cut to 47 mm size where necessary, baked in a laboratory
oven at 287°C (550°F) for two to three hours, and desiccated for
at least twenty-four hours. The samples remained in the desiccator
until just prior to use.
Two sampling sites were chosen for testing of the filter media.
Both sources were places where previous impactor runs had been
made, and the sources were different types. The first testing
was done at the outlet of a hotside ESP at Citadel Cement Company
between February 12th and February 21st, 1975. Two gas fired
kilns were in operation while the tests were being performed.
Outlet temperatures were in the neighborhood of 260°C (500°F).
Other data, including flowrates and weight gains, are listed in
Table B-2. Filter la includes the particulate catch in each case.
The second site was Bull Run Power Plant. The tests were performed
at the coal fired boiler precipitator outlet during two periods.
The first period was February 25 to February 28, 1975. Flue gas
temperatures ranged from 130°C (265°F) to 180°C (355°F). Other
data are listed in Table B-3. The second testing at Bull Run Power
Plant was April 1 to April 3, 1975, with flue temperatures varying
from 138°C (280°F) to 174°C (345°F). Table B-4 contains this data.
In the test at Citadel and the first test at Bull Run, there was
a problem with the filter material sticking to the o-ring and the
support screen of the holder. This created a nuisance and also added
77
-------
00
TABLE B-2
CITADEL CEMENT
cc-i
CC-2
CC-3
CC-4
CHg)
(°F)
(°F)
(ft1)
(«F>
(rain)
(ml)
(acfm)
CHg)
CHlO)
Date
Amb. Pres
Amb. Temp
Stack Temp
Gas Vol.
Avg . Gas
Meter Temp
Run Time
Ori. ID
Cond. HjO
% 8,0
Flow Rate:
- Ori.
- Gas
Meter
Avg . Probe
AP
Ori. AP
la
Ib
2
3
4
5
6
2-12-75
29.7
62
510
12.287
58
60
3348-
.059
74
(29.6) 221 used
for flowrate
0.4141 fl ,s
0.438f °-426
2.4
11.0
GA 3.22 mg
Silicone
o-rings
stuck
SA 0.28 to
filters
GA -0.48
-
-
2-18-75
29.5
68
480
23.808
74 '
120
3348-
.059
121
(26.6) 22.0
0.3831
0.390 1 "••»•'
3.0
10.3
GA 10.58 mg
GA 0.44
SA „.„ Slight
ring
RA 1.16
900 AF
MSA 1.02 flight
1106 BH Tin""
-
Teflon
o-rings
2-19-75
29.7
48
480
24.030
53
120
3348-
.059
131
(27.2) 22.0
o'Jiol °-400
3.0
10.4
GA 9.48 mg
GA 0.84
MSA 1.42
1106 BH
RA 1.70
900 AF
-, , ,, Stuck to
SA 1'66 metal
support
Teflon
o-rings
2-21-75
29.9
58
505
33.068
79
240
3348-
.059
97
(28.0) 22.0
15.0
2.6
GA 4.38
MSA 2.14
1106 BH
Teflon -0.02
GA 0.90
RA 2.40
900 AF
MSA 1.92
1106 BH
Teflon 0.00
Teflon
o-nngs
-------
TABLE B-3
BULL RUN STEAM PLANT
vo
CHg)
(8F)
(8F)
(ft'l
(°F)
dun)
(ml)
(acfm)
CHg)
CHiO)
Date
Amb. Pres
Amb. Temp.
Stack Temp.
Gas Vol.
Avg. Gas
Meter Temp.
Run Time
Ori. ID
Cond. U>0
I B.O
Flow Ratai
-Ori.
-Cas Meter
Avg Probe
AP
( + AP across
orifice)
Ori. AP
la
Ib
2
}
4
5
6
BRSP-1
2-25-75
29 08
54°
275
11 062
81
60
3148-
.059
6.0
(3 8) 7.5 used
In flow-
rate caL
culatiom
0.183 - ....
0.189 0'"6
a. a
6.5
GA 12.29mg
MSA 0 43
1106 BH,
GA 0.16
RA 0.37
900 AF
Teflon 0 01
MSA 0 27
1106 BH
GA 0 19
BRSP-2
2-25-75
29 08
61
265
18 767
73
60
1348-
.059
11.4
(4.9) 7.5
"•>« g 254
0.259 ' "*
12.65
14 9
GA 26.50
GA 0.44
RA 0.69
900 AF
MSA -0.14 "«*
"" BH S-rln,
t support
severely
Teflon 0.00
RA 0 56
900 AF
GA -0.74 Stuck
o-ring
& cut by
It.
All filters stuck
slightly to support
As euch as possible
recovered
BRSP-3
2-26-75
29.44
S3
275
80 928
81
240
3348-
.059
57.5
(5.6) 7.5
0.286 „ .„,
0 285 ° "'
12.6
18 5
GA
RA 0.77
900 AF
MSA 0.16
Teflon 0.01
SA 0.12
GA -0 0« Severely
cut
SA 0.31
This (and all that
follow) are Spectro
Grade A from batch
8192
BRSP-4
2-26-75
29 42
69
275
34.269
82
120
3348-
.059
23 8
(5.7) 7.5
0.211 , ,„
0.239 "•"'
12.7
12.6
GA 37 99
MSA 0 88
not BH
RA 0.83
900 AF
GA 0.24
SA 0.18
GA 0.41
Teflon 0.07
BRSP-5
2-27-75
29 36
64
100
140 245
79
4BO
3348-
059
90.8
15.7) 7.5
o!215 "-2J2
13.95
12.7
GA
GA 3.27 Moderate-
ly brown
on edge.
MSA 1 65 Moderate-
Hot BH ly brown
near
o-ring
SA 0.68 Severely
cut light
brown
GA 0 53 NO
discol-
oration
Teflon 0.00
MSA 1.00 Slight
1106 BH brown-
ing
BRSP-6
2-28-75
29 10
41
)55
8 431
74
10
1148-
059
4 8
(4 6) 7.5
°-"6 0 268
0.269 ° "8
12 S
12.9
GA 23.41
MSA 0.69
1106 BH
SA -0 12 Torn
GA 0.27
SA 0 66
Teflon -0 01
MSA 0 71
1106 BH
BRSP-7
2-28-75
29 10
50
155
-
_
0
_
-
-
-
-
GA 0 14
GA 0 38
MSA 0.77
1106 BH
SA -0 10 Torn
GA 0.18
MSA 0 70
1106 BH
Teflon 0 06
This set not ri.n plac
in stack and tdkcn out
-------
TABLE B-4
BULL RUN STEAM PLANT
BRSP-8
BRSP-9
BRSP-10
BRSP-11
BRSP-12
BRSP-13
("Hg)
(°F)
(°F)
(Qt3)
( F)
(rain. )
(ml.)
("H20)
("Hg)
(acfm)
*
Date
Amb. Pres.
Amb . Temp .
Stack Temp.
Gas Vol.
Avg . Gas
Meter Temp.
Run Time
Cond. H20
% H20
Ori. ID
Ori. AP
Avg . Probe
AP
Flowrate:
- Ori.
- Gas meter
la
Ib
2a
2b
3a
3b
4a
4b
ba
5b
6a
6b
4/1/75
29.28
84
315
52.172
103
240
48.9
4/2/75
29.15
78
315
13.004
92
60
11.0
(6.6) 7.5 used (5.2) 7.5
in flow-
rate
calculations
3348 - .059
9.7
7.1
0.242
0.238
GA
GA 3.68
GAE 8.54
IA '-02
SEE -°-50
GF/A
GF/A °'52
RA 934 AH - „
RA 934 AH °'46
3348 - .059
10.3
4.35
0.267
0.272
GA
GA 1.74
GAE
GAE 1-26
SA °-°°
S£ -1-96
GF/A
GF/A
RA 934 AH Q,
RA 934 AH °-°*
4/2/75
29.15
92
315
6.531
107
30
6.0
(5.9) 7.5
3348 - .059
10.3
4.75
0.261
0.262
GA
GA 1.00
GAE
GAE ! • 7 6
SA °-08
Quartz . .-
Quartz
GF/A
GF/A
RA 934 AH n -.
RA 934 AH °'"
4/2/75
29.15
89
315
6.516
99
30
5.8
(5.6) 7.5
3348- .059
10.3
4.65
0.262
0.266
GA
GA 0.84
GAE
GAE °-96
SA
SA ~U--L^
Quartz-1'40
GF/D
GF/D
£ in ™ °-°<
4/3/75
29.12
48
345
50.284
53.5
240
48.8
(6.1) 7.5
3348 - .059
10.3
6.6
0.272
0.266
GA
GA 2.74
GAE
GAE 6'66
st 4'°6
Juartz ~5'16
GF/D
GF/D
IA 934 AH Q _.
IA 934 AH
4/3/75
29.12
57
280
12.697
69
60
13.3
(6.2) 7.5
3348- .059
10.3
4.35
0.260
0.265
GA
GA 0.12
GAE
GAE 1 ' ° 8
IA o.io
Kz ••«
GF/D 0 18
GF/D °-18
RA 934 AH - ,-
RA 934 AH
00
o
NOTE:
GA - Gelman Type A
GAE - Gelman Type AE
SA - Gelman Spectro Glass Fiber, Type A
Quartz - Pallflex Tissuquartz 2500 QAD
GF/A - Whatman GF/A
GF/D - Whatman GF/D
RA 934 AH - Reeve Angel 934 AH
•Combined weight of the two filters per holder, except
for holder 1 where the weight of the prefliter (la)
has been eliminated.
-------
the possibility that some material might be overlooked. For
the tests-in April at Bull Run, teflon gaskets were cut which
fitted directly under the o-ring and on the support screen. The
filter sample was then between teflon gaskets and the gaskets
were preweighed with the filter sample.
Two identical 47 mm filters were run in each holder. Originally,
it was hoped that we might be able to obtain weight gain data from
each of these but because of a tendency of the filters to adhere
to one another the weight gains of both were lumped together.
All the results are shown in Table B-4.
The sampling times and flowrates used were supposed to approximate
the sampling time and flowrate of a typical Andersen Stack Sampler
test. Sampling times varied from thirty minutes to eight hours
which met the above requirement. The flowrates used, however, were
somewhat below that of the ideal Andersen flowrate of 0.5 cfm. On
all of the tests in which the teflon membranes were used, the
flowrates were only half as high as the desired flowrate of
0.5 cfm because the pressure drop across the membrane was large.
(Even so, the ratio between flowrate and filter surface area at
these lower flowrates approximated the same ratio of these for
an Andersen Stack Sampler.) No attempt was made at isokinetic
sampling since only the flue gas was of interest.
After each run, the filters were desiccated for at least 24 hours
and weighed. Any gas phase reaction was then detected by a
weight change. A report by Forrest Newman2 indicates weight
gains in glass fiber filters are possible by conversion of S02 to
various sulfates.
The results of all the tests are listed in the tables. From
these results, no definite trend emerged to indicate that the
weight gains depended upon flowrate. However, there seems to be
81
-------
a relation between weight gain and total exposure time of the
filter to the flue gas, regardless of the flowrate. Also, tem-
perature may be an important factor in the weight gains. At
Citadel Cement Company where the temperature was almost double
that of Bull Run Power Plant, the weight gains were much higher
than the Bull Run series. Since the two sources are different
types, this comparison may not be valid, but BRSP-6, a thirty-
minute run at 180°C (355°F), showed gains which exceeded those of
BRSP-3, a four hour run at 135°C (275°F).
Both Whatman types, GF/A and GF/D, and the Reeve Angel 934 AH
showed little tendency for weight gain, and seem to be candidates
for future use as impactor substrates. The Pallflex quartz
showed loss of weight, but it was very fragile and tended to
break and tear easily, which may have resulted in a loss of some
of the filter material.
Chemical analyses were performed on all the filter materials that
had been tested in the Gelman holders. This includes the tests
at Citadel Cement and those at Bull Run Steam Plant. Soluble
sulfate determination and pH tests were run on each sample,
except the teflon samples which had shown very little weight change.
These results are shown in Tables B-5, B-6, B-7.
The data in these tables indicate that sulfate is responsible
for the majority of the observed weight gain on each filter.
(Some filters which had small weight gains appeared to have
picked up little sulfate, but this may only reflect the limits of
our experimental accuracy.) In each case, the pH of the sample
was more acidic after it had been run, indicating possible
sulfate gain.
The work of Charles Gelman and J. C. Marshall of the Gelman Instrument
Company, makers of various filter media and equipment, seems to confirm
that S02 absorption is the cause of the weight gains.
82
-------
TABLE B-5
FILTER BLANKS (UNTREATED)
Type mg_SOjL~ pH
GA 0.04 8.6
MSA 0.18 9.3
RA 900 AF 0.06 9.8
SA Negligible 5.6
Quartz Negligible 6.6
GAE Negligible 9.2
RA 934 AH 0.06 7.2
GF/A Negligible 8.1
GF/D Negligible 7.0
83
-------
TABLE B-6
CITADEL CEMENT
cc-i
CC-2
CC-3
00
M9- wt- - Portion Mg. Wt. _ Portion M9- wt-
Gain Mg. SO» pH Analyzed* Gain Mg. SO* pH analyzed Gain M
la GA 3.22 2.1 7.1 fc
lb - - - -
2 SA 0.28 0.44 8.1 Whole
3 GA -0.48
4 _
5
6
Silicone
o-rings
stuck
to
filters
GA 10.58 5.02 8.1 \.
GA 0.44 0.10 8.0 Whole
SA 0.92 0.94 8.1 %
Slight
brown
ring
RA 1.16 0.83 7.7 «s
900 AF
MSA 1.02 0.76 7.7 «»
1106 BH Slight
brown
ring
- - -
_ _
Teflon
o-rings
GA 9.48
GA 0.84
MSA 1.42
1106 BH
RA 1.70
900 AF
SA 1.66
Stuck
to
metal
support
-
-
Teflon
o-rings
Filter
- c<-> = »,u Portion
g. SO, pH angled
4.83 7.7 k
0.27 7.2 1]
0.90 7.7 »j
1.45 8.1 >i
1.85 8.7 S
_
- -
'On those filters containing relatively large amounts of particulate, only a portion of the filter was analyzed.
Sulfate was calculated on a whole filter basis.
-------
TABLE B-6
(CONTINUED)
GA
MSA
1106 BH
Teflon
GA
RA
900 AF
MSA
1106 BH
Teflon
Mg. Wt. Gain
4.38
2.14
-0.02
0.90
2.40
1.92
0.00
Teflon
o-rings
CC-4
Mg. SOU"
2.0
1.67
-
0.30
2.27
1.52
^m
PH
7.4
6.9
-
7.1
7.0
7.5
"
Filter Portion
Analyzed
*
k
-
h
%
\
'
85
-------
TABLE B-7
BULL RUN
BRSP-l BRSP-2
Filter Portion
Mg. Wt. Gain Mg. SO* pH Analyzed Mg. Wt. Gain
la GA 12.29 0.56 5.6 *
Ib MSA 0.43 0.17 5.7 Whole
1106 BH
2 GA 0.16 0.06 5.5 Whole
00
0)3 RA 0.37 0.26 5.6 Whole
900 AF
4 Teflon 0.01
5 MSA 0.27 0.11 6.7 Whole
1106 BH
6 GA 0.19 0.04 6.4 Whole
•Same note as Citadel
GA 26.50
GA 0.44
RA 0.69
900 AF
MSA -0.14
1106 BH
Teflon 0.00
RA 0.56
900 AF
GA -0.74
All filters
stuck at least
slightly
Mg. S0i.= pH
0.91 5.4
0.04 5.9
0.20 6.9
Stuck
to
o-ring
and
support
0.26 7.7
Stuck
Filter Portion
Analyzed
*
Whole
Whole
Whole
-------
TABLE B-7
(CONTINUED)
BRSP-3
Mq. Wt. Gain Mq. SOn
PH
Filter Portion
Analyzed
BRSP-4
Mg. Wt. Gain Mg. SOi.
pH
Filter Portion
Analyzed
la
00 lb
^J
2
3
4
5
6
GA
RA
900 AF
MSA
Teflon
SA
GA
SA
_
0.77
0.16
0.01
0.12
-0.04
0.31
_ _
0.22 4.2
0.11 4.5
0.00 5.8
0.23 4.8
Severely -
cut
0.05 4.8
Whole
Whole
Whole
Whole
-
Whole
GA
MSA
RA
900 AF
GA
SA
GA
Teflon
37.99
0.88
0.83
0.24
0.18
0.41
0.07
_
0.26
0.25
0.06
0.29
0.03
4.4
5.5
5.7
5.1
5.3
5.1
k
Whole
Whole
Whole
Whole
3/4
-------
TABLE B-7
(CONTINUED)
BRSP-5
Mg. Ht. Gain Mg. SO*
Filter Portion
pH Analyzed
BRSP-6
Mg. Wt. Gain Mg. SOi."
Filter Portion
pH Analyzed
la
Ib
oo
00
2
3
4
5
6
GA
GA 3.27
MSA 3.65
1106 BH
SA 0.68
GA 0.53
Teflon 0.00
MSA 1.00
1106 BH
_ _ —
1.95 Moderately 3.0 %
brown on
edge
2.34 Moderately 3.0 %
brown near
o-ring
1.40 Severely 3.0 Whole
cut
0.103 5.6 Whole
_
0.52 Slight 5.8 *
brown
ring
GA
MSA
1106 BH
SA
GA
SA
Teflon
MSA
1106 BH
23.41
0.69
-0.12
0.27
0.66
-0.01
0.71
1.28 3.9 fc
0.24 6.4 Whole
Torn
0.02 6.5 Whole
0.10 5.7 Whole
_
0.28 6.4 Whole
-------
TABLE B-7
(CONTINUED)
BRSP-7
BRSP-8
Mg. Wt. Gain Mq. SO.."
Filter Portion
pH Analyzed
Mg. Wt. Gain
Mg. SO i.
Filter Portion
pH Analyzed
00
vo
la
Ib
2
3
4
5
6
GA
GA
MSA
1106 BH
SA
GA
MSA
1106 BH
Teflon
0.34
0.38
0.77
-0.10
0.38
0.70
0.06
This set not
heated
for 30
0.12
0.12
0.37
Torn
0.08
0.31
-
run; placed in
minutes , and
6.8
6. 9
7.8
-
7.2
7.8
-
stack.
taken out.
Whole
whole
Whole
-
Whole
Whole
-
Ib
2a
3a
4a
5a
5a
6a
& b
& b
& b
& b
& b
& b
GA
GAE
GSA
Quartz
GF/A
GF/D
RA 934
3.68
8.54
2.02
-0.50
0.52
_
0.24
1.62
5.16
1.54
Negligible
Negligible
_
Negligible
3.4
3.3
3.5
3.6
5.9
_
5.6
Ij
s each filter
•s each filter
2 wnole filters
_
2 whole filters
-------
TABLE B-7
(CONTINUED)
0 Ib
2a
3a
4a
5a
5a
6a
BRSP-9 BRSP-10
_ Filter Portion
Mg. Wt. Gain Mg. SO,," pH Analyzed Mg. Wt. Gain Mg. SCK
&
&
&
&
&
&
b
b
b
b
b
b
GA 1.74
GAE 1.26
GSA 0.00
Quartz -1.96
GF/A 0.02
GF/D
RA 934 0.02
AH
0.92 3.4 !j
0.46 5.8 k each filter
0.05 5.3 2 whole filters
_
Negligible 6.6 2 whole filters
_
Negligible 6.1 2 whole filters
GA 1.00 0.42
GAE 1.76 0.29
GSA 0.08 0.04
Quartz 0.12 Negligible
GF/A 0.08
GF/D
RA 934 AH 0.24 Negligible
Filter Portion
pH Analyzed
3 . 7 Whole
7.2 >s each filter
5.9 2 whole filters
3.5 2 whole filters
-
-
6.3 2 whole filters
-------
vo
TABLE B-7
(CONTINUED)
BRSP-ll
Mq. Wt. Gain Mg. SO..
pH
Filter Portion
Analyzed
BRSP-12
Mg. Wt. Gain Mg. SO..'
pH
Filter Portion
Analyzed
Ib
2a & b
3a & b
4a & b
5a & b
5a & b
6a & b
GA
GAE
GSA
Quartz
GF/A
GF/D
RA 934 AH
0.84
0.96
-0.12
-1.40
-
0.00
0.04
0.37
0.44
0.04
-
-
0.05
0.02
4.1
7.8
5.8
-
-
6.6
5.9
Whole
2 whole
2 whole
-
-
2 whole
2 whole
filters
filters
filters
filters
GA
GAE
GSA
Quartz
GF/A
GF/D
RA 934
AH
2.74
6.66
4.06
-5.16
-
-0.04
0.08
1.48
4.82
4.99
-
-
0.13
0.14
2.8
2.7
2.7
-
-
5.9
5.6
Whole
H each
*j each
-
-
2 whole
2 whole
filter
filter
filters
filters
-------
TABLE B-7
(CONTINUED)
BRSP-13
Mg. Wt.
Gain Mg. SGV
PH
Filter Portion
Analyzed
Ib
2a &
3a &
4a &
5a &
5a &
6a &
b
b
b
b
b
b
GA
GAE
GSA
Quartz
GF/A
GF/D
RA 934
AH
0.
1.
0.
0.
-
0.
0.
12
08
10
66
18
32
0
0
0
.03
.32
.05
Negligible
0
0
-
.06
.05
6.
5.
5.
3.
-
6.
5.
0
8
7
3
4
5
Whole
2
2
2
2
2
whole
whole
whole
-
whole
whole
filters
filters
filters
filters
filters
92
-------
They acknowledge that a high pH glass fiber can absorb sulfur
dioxide and thus cause erroneous, high, particulate weights.
Pate and Lodge's work using Na2CC>3 treated glass filters as
"dosimeters" for SOa exposure chambers, with weight gain of the
filters being a time function of exposure to S(>2, was mentioned.
According to this article, the S02 reaction on glass fiber
could cause "a 30% error in the measurement of total suspended
particulate matter" in an urban SO2 atmosphere. The new auto-
motive catalytic mufflers could increase this error. It is
possible that flue gases would give even higher errors, especially
if the gases have a high moisture content, because the reactivity
of SO2 appears to increase at higher humidity.
Both quartz and glass fiber filter material were tested by Gelman.
The quartz was found to be non-reactive to S02. The glass fiber
materials, Type II and SpectroGrade, prepared with HjSOi*, were low
in SOa pickup. The SpectroGrade glass, prepared with HC1, picked
up significant amounts of SOa (See Table B-8). Their explana-
tion is that the glass prepared with HaSOi, has reacted to form
CaSOi, to prevent further reaction with S02 to form sulfate.
(The test used for SOa reactivity was to expose the filters to a
water saturated atmosphere of S02 for 20 hours. Weight change
of the filter was measured.)
Another type of SpectroGrade coated with an organic silicone
resin showed low SOa pickup. This type of SpectroGrade with the
silicone treatment is now the standard type.
Use of the siliconized SpectroGrade at elevated temperatures may
result in the disappearance of the coating and S02 absorption by
the filter media since the filter is prepared with HC1.
93
-------
TABLE B-8
SULFUR DIOXIDE PICKUP
mg/Sheet - 20 Hour Exposure
SpectroGrade-HCl
Siliconized 3 7.1
SpectroGrade HC1 17 9.4
SpectroGrade
H2SCU 3 6.8
Type II Fiber
H2S04 3 6.8
Quartz 0 7.0
Quartz
Alkali Strengthened 23 (est.) 9.5
94
-------
APPENDIX C
CYCLONE DESIGN AND CALIBRATION
This appendix provides engineering data and calibration curves
that will be of assistance in the design of cyclone sampling
systems. Each of the cyclones listed below have been
calibrated and field tested except the SRI-2 and T2A cyclones.
These calibration data were obtained using aerosols generated
with a Vibrating Orifice Aerosol Generator. The particles
were ammonium fluorescein having a density of 1.35 gm/cc. For
particles of different mass densities, the particle size axis,
in the accompanying figures, should be shifted according to the
relationship
D(pa) = D(Pi
where PI = the density of the calibration aerosol (1.35 gm/cc),
pa = the density of the test aerosol (gm/cc),
D(PI) = any point on the particle size axis of the
calibration curve, and
D(PZ) = the corresponding point on the new graph.
Additional calibration data for cyclones TIB, T2A, and T3A
can be found in Chang's paper.3
A. CYCLONE SRI-1
Cyclone SRI-1 was designed for a nominal DSo of 1 ym at 1 acfm and
was derived by extrapolation using the previously calibrated SRI-5
and T2A cyclones which have cut points of approximately 0.5 pm and
2.6 vim at 1.0 acfm, respectively. The calibration curve for SRI-1
is illustrated in Figure C-l. Shop drawings for SRI-1 are provided
in Figure C-2.
95
-------
vo
100
90
80
70
u.
u.
UJ
o 40
Ul
30
10
50
SRI-I
O.I
I
0.96^ 1.0
PARTICLE DIAMETER, um
5.0
10.0
Figure C-l. Collection efficiency versus particle diameter for the SRI-1
cyclone. (22°C, 29.60"Hg, 1.35 gm/cc, 1.0 acfm)
-------
p
T
1
I
r
j
1
-s
"1
j
R
1/4 SECTION
COLLECTION CUP
NOZZLE ADAPTOR
A
B
C
D
E
F
G
H
J
K
L
M
0.5
1.75
0.25
0.40
1.223
0.425
1
1
750
223
0.80
18°30'
1.00
3.183
* 1
lii f.~S ii
M ;!
_, ILJ
F -4ol*-
c
1
i
VORTEX TUBE AND OUTLET
Z
on/o j *
*~"\7 r
c± !
CYCLONE
N
P
R
S
T
V
W
X
Z
AA
SS
- — G — «
n
%<= \
mm
i i \
4-,
J
i
1
L
1/2 SECTION
0.40
1.00
1.50
1.75
0.25
0.80
0.30
0.25
0.295
0.328
1.25
Figure C-2. Engineering dimensions (inches) for Cyclone SRI-1
97
-------
B. CYCLONE SRI-2
Cyclone SRI-2 was designed using Lapple's equation to achieve a
D50 of 10 urn at a flowrate of 5 acfm. This was designed especially
for use with the Aerotherm High Volume sampling system to obtain
size segregated samples for chemical analysis. No calibration data
are available at this time. Figure C-3 provides drawings for
cyclone SRI-2.
C. CYCLONE SRI-3
This cyclone design was also based on Lapple's equation and was
originally intended for use with a point to plane resistivity
probe. It has most recently been used as an externally mounted
pre-collector cyclone for a Brink Cascade impactor. Figure C-4
illustrates the calibration curve for this cyclone while
Figure C-5 provides the shop drawings.
D. CYCLONE SRI-4
This cyclone was designed for use as an inline pre-collector
cyclone for the Brink Cascade impactor. The design was modelled
after Chang's3 TIB cyclone. A calibration curve for cyclone
SRI-4 is illustrated in Figure C-6. The design drawing is shown
in Figure C-7.
E. CYCLONE SRI-5
Cyclone SRI-5 was originally designed as an externally mounted
pre-collector cyclone for the Brink Cascade impactor. In a more
recent application this design has been used for the third stage
of the SRI 3-stage cyclone. As in the case of cyclone SRI-4, this
cyclone design was modelled after Chang's3 TIB cyclone.
Figure C-8 illustrates the calibration curve for cyclone SRI-5,
while the shop drawing is shown in Figure C-9.
98
-------
12.913
12.913
1.614
Figure C-3. Engineering dimensions (inches) for Cyclone SRI-2
99
-------
o
o
100
90
80
*! 70
o
3 60
o
u_
u.
U 50
z
g
a 40
o
o
30
20
10
0
3 75 5 10
PARTICLE DIAMETER,
50
100
Figure C-4. Collection efficiency versus particle diameter for the SRI-3
cyclone. (22°C, 29.60"Hg, 1.35 gm/cc, 1.0 acfm)
-------
SOUTHERN RESEARCH INSTITUTE
Figure C-5. Engineering dimensions (inches) for Cyclone SRI-3.
-------
/-0-025T
o
to
»(>u^u
XSS®
•I
- ys£=~- — teame
,~~
.., ,.
t* r fasfJT~f
+(*-31. , iS-tt r.J
SOUTHERN RtSbARCH INSTITUTE
BtKMMGMAM, A1A1AMA 35305
Figure C-5A. Engineering dimensions (inches) for Cyclone SRI-3.
-------
o
OJ
&
100
90
80
70
60
ii.
b.
w 50
2
g
a 4°
30
20
10
0
SRI-4
I
Figure C-6,
10 12.5
PARTICLE DIAMETER,
100
fi H
29.60 "Hg,
Ver /US Particle diameter for the SRI-4 cyclone.
gm/cc , 0.03 acfm)
-------
.280
/t*s>/Wi/4E""
O. D.
*l.<*1!>
,28o 8or£~^cs.4ss(*
Tl&n T f IT 0t- TISCII "•>•
. 1
£T£:/t -
J/ii 5.-'.
F.'MiSH ftA~!'H&- .
3 i. flic KO HJCH K f*S
POLISH INTKKIOK S<-
8 riicfo/ucy R/4S
ALSO
- C-2.
Figure C-7. Engineering dimensions
(inches) for Cyclone SRI-4.
CTCTK ]-J
DCCMALS \\ 010
MNIW Kit
SOUTHERN RESEARCH INSTITUTE
8HMINGHAM. *LA»*MA 3J205
dfCL&>JE~ I v f- /rlPAi
SRI-4
10a?23-c-;
-------
o
m
—
O
100
90
80
70
60
50
§ 40
§ 30
20
10
O.I
SRI-5
0.5 0.6 1.0
PARTICLE DIAMETER, um
5.0
10.0
Figure C-8. Collection efficiency versus particle diameter for the SRI-5 cyclone,
(22°C, 29.60 "Hg, 1.35 gm/cc, 1.0 acfm)
-------
r
i
t
R
1/4 SECTION
COLLECTION CUP
:•- .-a ; I
VORTEX TUBE AND OUTLET
NOZZLE ADAPTOR
CYCLONE
1/2 SECTION
A
B
C
D
E
F
G
H
J
K
L
M
0.235
1.50
0.25
0.300
1.00
0.23
1,
1,
500
00
.556
24d
1.000
2.063
N
P
R
S
T
V
W
X
z
AA
SS
0.35
1,
1
1
000
50
,50
0.25
0.70
0.25
0.25
0.201
0.234
1.00
Figure C-9. Engineering dimensions (inches) for Cyclone SRI-5
106
-------
F. CYCLONE TIB
This design was originally developed by Chang while at McCrone
Associates, and was one of the four cyclones in his parallel
cyclone sampling system. Cyclones SRI-4 and SRI-5 were designed
after this particular cyclone. The calibration curve for cyclone
TIB is illustrated in Figure C-10 and the shop drawings are
depicted in Figure C-ll.
G. CYCLONE T2A
This design was also developed by Chang3 as part of his parallel
cyclone sampling system. The same internal dimensions were used
in the first stage cyclone of the SRI three-stage series cyclone
system. This cyclone design has been calibrated at three flow-
rates illustrated in Figures C-12, C-13, and C-14. Figure C-15 shows
shop drawings for the SRI version of this cyclone.
H. CYCLONE T3A
This is the largest cyclone in Chang's parallel cyclone sampling
system, with a nominal DSO of about 5 pm at 3 acfm. No calibration
data are available for this cyclone. This design was used as the
3 ym DSO unit for the Aerotherm 5 acfm sampling system. Shop
drawings are shown in Figure C-16.
I. ANDERSEN PRECOLLECTOR CYCLONE
Designed, built, and sold by Andersen 2000, Inc., for use as a
precollector for the Andersen Mark III Cascade impactor, this
cyclone has been calibrated by Andersen and by SRI. Calibration
data and dimensions are given for this cyclone in Figures C-17,
C-18, and C-19.
J. VARIATIONS IN D50 WITH AEROSOL SAMPLE FLOWRATE
As stated in Section IIIA, the D50 for a cyclone theoretically should
vary as the square root of the flowrate, and the D50 for a range
107
-------
o
CO
100
90
80
5s 70
o
o
u_
u.
60
50
40
o
0 30
20
10
O.I
Figure C-10,
0.5 1.0 1.25
PARTICLE DIAMETER, Mm
5.0
10.0
Collection efficiency versus particle diameter for Cyclone TIB,
(22 C, 29.60 "Hg, 1.35 gm/cc, 0.23 acfin)
-------
p
1
1
T
r
)
\
-s
s-
t
R
1/4 SECTION
COLLECTION CUP
c
_L
-IU
-E—J
VORTEX TUBE AND OUTLET
NOZZLE ADAPTOR
CYCLONE
1/2 SECTION
A
B
C
D
E
F
G
H
J
K
L
M
0.235
1.50
0.25
0.300
1.00
0.23
1,
1,
,500
,00
.556
24°
1.000
2.063
N
P
R
S
T
V
W
X
z
AA
SS
0.35
1.000
1.50
1.50
0.25
0.70
0.25
0.25
0.201
.156
1.00
Figure C-ll. Engineering dimensions (inches) for Cyclone TIB,
109
-------
v
z
UJ
u
u_
u.
LU
0
UJ
8
0.5 1.0
PARTICLE DIAMETER,
2.65
5.0
10.0
C-
12. Collection efficiency versus particle diameter for Cyclone T2A,
(22°C, 29.60 "Hg, 1.35 gm/cc, 1.60 acfm)
-------
0.5 1.0
PARTICLE DIAMETER,
3.2
5.0
10.0
Figure C-13,
Collection efficiency versus particle diameter for Cycl'one T2A.
(22°C, 29.60 "Hg, 1.35 gm/cc, .59 acfm)
-------
10
100
90
80
70
60
50
UJ
o
u.
li.
Ul
z
o
o 40
UJ
o
30
20
10
1
T2A
O.I
0.5 1.0
PARTICLE DIAMETER ,
2 65
5.0
10.0
Figure C-14
Collection efficiency versus particle diameter for Cyclone T2A.
(22°C, 29.60 "Hg, 1.35 gm/cc, 1.0 acfm)
-------
rssi
Y///A
1/4 SECTION
COLLECTION CUP
a
°H
-—J
VORTEX TUBE AND OUTLET
NOZZLE ADAPTOR
r
N
qjo
-©-
fjjn 1
w Q— . /
\ /
\ /
r-U._,
K
CYCLONE
1/2 SECTION
A
B
C
D
E
F
G
H
J
K
L
M
0.5
2.0
0.25
0.465
1.430
0.618
2.
1
1.
.000
.440
.080
14°5«
0.75
3.685
N
P
R
S
T
V
w
x
z
AA
SS
0.450
0.50
1.25
2.00
0.25
0.90
0.35
0.25
0.397
0.414
1.50
Figure C-15. Engineering dimensions (inches) for Cyclone T2A.
113
-------
0.82
3.219
2.141
4.00
SECTION A-A
Figure C-16. Engineering dimensions (inches) for
Cyclone T3A.
114
-------
100
90
80
70
u 60
a
Ld
Z
o
50
40
30
20
10
0
I ANDERSEN
I
Figure C-17
5 10
PARTICLE DIAMETER,
50
100
Collection efficiency versus particle diameter for "mdersen
.'io^ifie-x Pre-separator. 22°C, 29.5 "Hg, 1.35 gm/cj, 0.5 acfrr,
-------
100
90
80
70
uj 60
o
UJ
50
40
O
0 30
20
10
0
I
Figure C-18.
5 85 10
PARTICLE DIAMETER, Mm
50
100
Collection efficiency versus particle diameter for the
Andersen Modified Pre-separator.l * (0.75 acfm, 1.32 gm/cc,
particle diameter unknown).
-------
Figure C-19. Engineering dimensions (inches) for the
Andersen Modified Pre-separator.12
117
-------
of flowrates is related to the calibration D50 by the equation
D50(2) = D5o(l)V~vf •
where Vi = the calibration flowrate,
Va = the second flowrate of interest,
Dsod) = the calibration DSO at flowrate Vi, and
D5o(2) = the DSO at the second flowrate of interest.
Figure C-20 shows the theoretical relationships between D50 and
flowrate for each cyclone we have tested, and the experimental
data which is available for each.
K. NOZZLE SPECIFICATIONS
Figure C-21 provides specifications for the construction of
nozzles similar to those used with the SRI 3-stage series cyclone.
These ten nozzles allow a wide range of flue gas velocities to
be sampled while maintaining isokinetic sampling.
118
-------
100.0
10.0
o
Q.
O
O
10
0
0.01
Figure C-20.
O.I 1.0
CYCLONE FLOW RATE .ocfro
10.0
DSO cut point versus cyclone flowrate for six
calibrated cyclones.
119
-------
SOUTHERN RESEARCH INSTITUTE THREE-STAGE SERIES CYCLONE
NOZZLE SPECIFICATIONS
AMERICAN STANDARD PIPE THREAD
^Nominal \ Pipe Size
/ Five Threads
I
^.28^- 0.
1.
ALL DIMENSIONS IN INCHES
Nozzle
Number A
B
1
2
3
4
5
6
7
8
9
10
9°
9*
8°
8°
7°
6°
5°
4°
3°
2°
46'
19'
55'
21'
25'
30'
29'
36'
41'
45'
0.177
0.196
0.213
0.238
0.277
0.316
0.358
0.397
0.435
0.475
Figure C-21. Cyclone nozzle design specifications,
120
-------
APPENDIX D
CONVERSION TABLE FOR UNITS
To Convert From
°F
ACFM or acfm
"H20
inches
ft3
To
"C
m3/sec
mm Hg
mm
m3
Multiply By
(°F-32)5/9
0.000472
1.8682
25.4
0.02832
121
-------
REFERENCES
1. Smith, W. B., Gushing, K. M., and McCain, J. D. Particulate
Sizing Techniques For Control Device Evaluation, Report No. 21,
EPA Contract No. 68-02-0273, July 12, 1974.
2. Forrest, J. and Newman, L. Sampling and Analysis of Atmospheric
Sulfur Compounds For Isotope Ratio Studies, Atmospheric
Environment, 1_, pp. 561-573, 1973.
3. Chang, H. C. A Parallel Multicyclone Size-Selective Particulate
Sampling Train, Amer. Ind. Hyg. Assoc. J. 7j[(9) , pp. 538-45, 1974.
4. Rusanov, A. A. Determination of the Basic Properties of Dust
and Gases. In Ochistoko Dymovykh Gazov v Promshlennoi Energetike,
Rusanov, Urbakh, and Anastasiadi, "Energiya", Moscow, 1969.
5. Mushelknauz, E. Design of Cyclone Separators in the Engineering
Practice. Staub-Reinhalt Luft 30(1), 1970.
6. Blachman, M. and Lippman, M. Performance Characteristics of
Multicyclone Aerosol Sampler, Amer. Ind. Hyg. Assoc. J. 3!5_(311) ,
1974.
7. Leith, D. and Mehta, D. Cyclone Performance and Design.
Atmospheric Environment, 7, pp. 527-549, 1973.
8. First, M. W. Fundamental Factors in the Design of Cyclone Dust
Collectors. Doctoral Thesis, Harvard University, 1950.
9. Engineering Manual. Ed. by Perry, J. H. and Perry, R. H.
McGraw-Hill Book Company, Inc., New York. Pp. 5-61, 1959.
122
-------
REFERENCES
(CONTINUED)
10. Treaftis, H. N. and Tomb, T. F. Effect of Orientation on
Cyclone Penetration Characteristics. Amer. Ind. Hyg. Assoc. J.
39_(10), pp. 598-602, 1974.
11. Gelman, C. and Marshall, J. C. "High Purity Fibrous Air
Sampling Media", Industrial Hygiene Association Annual Meeting,
Thursday, May 16, 1974, Miami, Florida.
12. Andersen 2000, Inc. Advertising Literature.
123
-------
TECHNICAL REPORT DATA
(Please read Inunctions on the reverse before completing]
I REPORT NO ,
EPA-650/2-74-102-E
3 RECIPIENT'S ACCESSION-NO.
4 TITLE AND SUBTITLE
Participate Sizing Techniques for Control Device
Evaluation
5 REPORT DATE
August 1975
6 PERFORMING ORGANIZATION CODE
7 AUTHOR(S)
W. B. Smith, K. M. Gushing, G. E. Lacey, and
J. D. McCain
8 PERFORMING ORGANIZATION REPORT NO
SORI-EAS-75-369
9 PERFORMING ORGANIZATION NAME AND ADDRESS
Southern Research Institute
2000 Ninth Avenue, South
Birmingham, Alabama 35205
10 PROGRAM ELEMENT NO.
1AB012; RQAP 21ADM-011
11. CONTRACT/GRANT NO.
68-02-0273
12 SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13 TYPE OF REPORT AND PERIOD
Summary: 7/74 - 6/75
OD COVERED
14. SPONSORING AGENCY CODE
15 SUPPLEMENTARY NOTES
6 ABSTRACT
rep0rt des cribes results of laboratory and field work related to iner-
tial particulate size classifiers (impactors and cyclones). The impactor work deals
largely with non-ideal behavior of impactors and problems encountered in field test-
ing. Preparation and handling procedures for using glass fiber impaction substrates
are discussed, together with problems resulting from SO2 reactions with certain
types of glass fiber filter media. The results of a brief series of tests of electro-
static effects in impactor sampling are described: they indicate that these effects can
be substantial under some circumstances. Design and calibration data are given for
two series cyclone size devices: one designed to operate at a flowrate of 140 liters/
minute (5 cfm); and the other, at 28 liters/minute (1 cfm). Each provides three size
fractionation points in the 0. 5 to 10 jum size interval. The cyclone systems permit
collection of larger quantities of size fractionated particulates and are somewhat
easier to use than are impactors.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. COSATl Field/Croup
Air Pollution
Size Separation
Sulfur Dioxide
Evaluation
Impactors
Cyclone Separators
Tests
Field Tests
Glass Fibers
Substrates
Electrostatics
Air Pollution Control
Stationary Sources
Particulates
13B 14B
07A, 13H
07B HE, HE
11D
131 20C
8 D'STmmiTiON STATEMENT
Unlimited
19. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES
132
20 SECURITY CLASS (Thispage/
Unclassified
22 PRICE
EPA Form 2220-1 (9-73)
124
------- |