EPA-600/2-76-065
March 1976
Environmental Protection Technology Series
ASSESS!
PARTICLE CONTROL TECHNOLOGY
:NCLOSED ASBESTOS SOURCES
Phase II
fatostrial Envtrounwirt^ Rtswrcfc Latoratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into five series. These five broad
categories were established to facilitate further development and application of
environmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The five series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
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.
EPA REVIEW NOTICE
This report has been reviewed by the U.S. Environmental
Protection Agency, and approved for publication. Approval
does not signify that the contents necessarily reflect the
views and policy of the Agency, nor does mention of trade
names or commercial products constitute endorsement or
recommendation for use.
This document is available-to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/2-76-065
March 1976
ASSESSMENT OF PARTICLE CONTROL
TECHNOLOGY FOR ENCLOSED ASBESTOS
SOURCES--PHASE II
by
Paul C. Siebert, Thomas C. Ripley, and Colin F. Harwood
IIT Research Institute
10 West 35th Street
Chicago, Illinois 60616
Contract No. 68-02-1353
RQAP No. 21AFA-006
Program Element No. 1AB015
EPA Project Officer: D.K. Oestreich
Industrial Environmental Research Laboratory
Office of Energy, Minerals, and Industry
Research Triangle Park, NC 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, DC 20460
-------
ABSTRACT
The report gives results of an experimental study to
optimize control of emissions of asbestos fibers using a bag-
house. Baghouse operating parameters that were studied in a
statistically designed experimental plan were: (1) filter
fabric, (2) air-to-cloth ratio, (3) dust loading, (4) relative
humidity, (5) shaking amplitude, (6) frequency, (7) duration
and time between shaking cycles, and (8) bag series configuration.
Operating parameters which were found to be statistically
significant in causing reductions in asbestos emissions were:
(1) bag fabric, (2) waste type, (3) air-to-cloth ratio,
(4) relative humidity, (5) period between shakes and shaking
duration, and (6) shaking amplitude. The values of these
operating parameters that are recommended for industry usage
to significantly reduce outlet concentrations of asbestos are:
(1) cotton sateen bags, (2) an air-to-cloth ratio of
32 2
1.22 m /min/m (4.0 cfm/ft ), (3) a combination of period
between shakes of 120 min with a shaking duration of 20 sec,
and (4) a shaking amplitude of 3.500 cm. These operating
conditions resulted in pressure drops across the fabric filter
that were quite reasonable (_<2.0 in. t^O) . Thus, the most
economical alternatives of cotton sateen bags, high air-to-
cloth ratio, and low pressure drop operating conditions were
found to be among the most significant in reducing asbestos
emissions.
This report was submitted in fulfillment of IITRI Project
No. C6291, Contract No. 68-02-1353, by the IIT Research
Institute, under the sponsorship of the Environmental Protection
Agency. Work was completed as of June 1975.
iii
-------
CONTENTS
Page
Abstract iii
List of Figures v
List of Tables vii
Acknowledgements x
Sections
1 Conclusions 1
2 Recommendations 4
3 Introduction 6
4 Development of Experimental Plan 8
5 Experimental Apparatus 28
6 Experimental Procedure 41
7 Discussion of Results of Testing Based on
Statistical Analysis of Data 56
8 References 118
Appendix A 121
iv
-------
FIGURES
No. Page
1 Schematic of Automatic Timer System 31
2 Calibration Curve of Vibra-Serew SCR-20 Dust
Feeder for Asbestos Cement Waste using 1/2 in.
Screw 34
3 .Porous Tube Diluter 37
4 Baghouse Testing Apparatus 38
5 Non-Shaking Baghouse 39
6 Phase III Royco Particle Counter Traces 53
7 Estimates of the Geometric Mean and Their 90%
Confidence Intervals for Outlet Concentration
of Asbestos Fibers Greater than 0.06 ym by
Type of Bag - Phase I - Subsample 1 66
8 Estimates of the Geometric Mean and Their 90%
Confidence Intervals for Outlet Concentration
of Asbestos Fibers Greater than 1.5 ym by
Type of Bag - Phase I - Subsample 1 67
9 Estimates of the Geometric Mean and Their 90%
Confidence Intervals for Outlet Concentration
of Asbestos Fibers Greater than 6.0 ym by
Type of Bag - Phase I - Subsample 1 68
10 Estimates of the Geometric Mean and Their 90%
Confidence Intervals for Outlet Concentration
of Asbestos Fibers Greater than 0.06 ym by
Type of Bag - Phase I - Subsample 2 78
11 Estimates of the Geometric Mean and Their
Confidence Intervals for Outlet Concentration
of Asbestos Fibers by Type of Bag - Phase I -
Subsample 2 79
12 Estimates of the Geometric Mean and Their 90%
Confidence Intervals for Outlet Concentration
of Asbestos Fibers Greater than 0.06 ym by
Type of Waste - Phase I - Subsample 2 80
v
-------
FIGURES (cont.)
No. Page
13 Estimates of the Geometric Mean and Their 90%
Confidence Intervals for Outlet Concentration
of Asbestos Fibers by Type of Waste - Phase I -
Subsample 2 81
14 Geometric Means, 90% Confidence Intervals, and
the Regression Line for Outlet Concentration
of Fibers Greater than 1.5 ym by Air-to-Cloth
Ratio - Phase II 93
15 Geometric Means, 90% Confidence Intervals, and
the Regression Line for Outlet Concentration
of Fibers Greater than 6.0 ym by Air-to-Cloth
Ratio - Phase II 94
16 Geometric Means, 90% Confidence Intervals, and
the Regression Lines for Outlet Concentration
of Asbestos Fibers by Shake Period for an
Amplitude = 3.50 cm - Phase III 104
17 Geometric Means, 90% Confidence Intervals, and
the Regression Lines for Outlet Concentration
of Asbestos Fibers by Shake Period for an
Amplitude = 0.875 cm - Phase III 105
18 Estimates of Geometric Mean and Their 90%
Confidence Intervals for Outlet Concentration
of Asbestos Fibers Greater than 1.5 ytn by
Stabilization Period - Phase IV 112
19 Estimates of Geometric Mean and Their 90%
Confidence Intervals for Outlet Concentration
of Asbestos Fibers Greater than 5.0 ym by
Stabilization Period - Phase IV 113
20 Estimates of Geometric Mean and Their 90%
Confidence Limits for Outlet Concentration
of Asbestos Fibers by Stabilization Period -
Phase IV 116
VI
-------
TABLES
No. Page
1 Complete List of Control Options for Baghouse 9
2 Test Fabric Characteristics 12
3 Effect of Relative Humidity on Outlet Dust
Concentration and Efficiency 13
4 Efficiencies for Cloths of Different Weaves 17
5 Reduced List of Options for Baghouse 23
6 Final List of Options for Baghouse 24
7 Shaker Assembly Motor Calibration 29
8 Filter Bag Characteristics 32
9 Phase I Results for Asbestos Cement Waste 44
10 Phase I Results for Fibrous Asbestos Waste 46
11 Data and Results for Cotton Sateen, Phase II 48
12 Phase III Fiber Counts 51
13 Phase IV Fiber Counts 54
14 Numeric Coding of Waste Type 58
15 Numeric Coding of Bag Type 58
16 The Independent Variables and Their Desired Levels
for Subsample 1 of Phase I 61
17 Data Base for Phase I 62
18 Correlations Between Phase I Variables for
Subsample 1 (N = 20) 63
19 Geometric Means and 90% Confidence Limits for
Phase I - Subsample 1 65
20 Results of Regression Analysis of Subsample 1 -
Phase I for Fibers Greater than 1.5 ym 70
Vll
-------
TABLES (cont.)
No.
21 Results of Regression Analysis of Subsample 1 -
Phase I for Fibers Greater than 6.0 ym 71
22 Computed Geometric Means of Outlet Concentration 72
23 The Independent Variables and Their Desired
Levels for Subsample 2 of Phase I 73
24 Correlations Between Phase I Variables for
Subsample 2 (N = 13) 75
25 Geometric Means and 90% Confidence Limits for
Phase I - Subsample 2 76
26 Results of Regression Analysis of Subsample 2 -
Phase I for Fibers Greater than 1.5 ym 82
27 Results of Regression Analysis of Subsample 2 -
Phase I for Fibers Greater than 6.0 ym 83
28 Phase II Independent Variables and Their Desired
Levels 86
29 Data Base of Phase II 87
30 Correlations Between Phase II Variables 89
31 Results of Regression Analysis of Phase II for
Fibers Greater than 1.5 ym 91
32 Results of Regression Analysis of Phase II for
Fibers Greater than 6.0 ym 92
33 Data Base for Phase III 98
34 Correlations Between Phase III Variables 100
35 Results of Regression Analysis of Phase III for
Fibers Greater than 1.5 ym 102
36 Results of Regression Analysis of Phase III for
Fibers Greater than 5.0 ym 103
37 Phase IV Independent Variables and Their Desired
Levels 108
Vlll
-------
TABLES (cont.)
No.
38 Data Base for Phase IV 110
39 Geometric Mean and 90% Confidence Limits of Outlet
Concentration for Different Stabilization Periods 111
40 Geometric Mean and 90% Confidence Limits of Outlet
Concentration for One and Two Bag Baghouses 115
IX
-------
ACKNOWLEDGEMENTS
guidance and encouragement of the Environmental
•Protection Agency Project Officer, Mr. David Oestreich, is
gratefully acknowledged. His enthusiasm and concern for the
project contributed much to its success. Dr. James Turner,
also of the EPA, gave valuable consultancy on the fabric
filter operating fundamentals.
IITRI personnel who contributed to the program were:
Paul Siebert, who was the principle investigator, and
Thomas Ripley, who undertook the statistical design and
analysis. Dr. Colin F. Harwood was the Project Leader,
while John D. Stockham, Manager of the Fine Particles Research
Section, had administrative responsibility. Other IITRI
personnel who contributed to the program were Erdman Luebcke,
M. Ranade, and Dr. Earl Knutsoh.
x
-------
SECTION 1
CONCLUSIONS
For all the fabrics and values of the baghouse oper-
ating parameters tested, the mass efficiencies of asbestos
collection exceeded 99.99%. However, as noted in the
Phase I report, extremely high numbers of small fibers may
still be emitted while attaining such high mass efficien-
cies. Typical outlet concentrations of asbestos fibers on
57 3
the order of 10-10 fibers/m (for fibers > 1.5 ym) and
893 ~~
10 -10 fibers/m (for fibers >_ 0.06 ym) were found to be
emitted.
Operating parameters which were found to be statis-
tically significant in causing reductions in asbestos
emissions were: (1) bag fabric, (2) waste type, (3) air-
to-cloth ratio, (4) relative humidity, (5) period between
shakes and shaking duration, and (6) shaking amplitude,
The following conclusions were drawn with regard to the
effect of these variables on fiber outlet concentration:
1. Cotton sateen was as efficient or more efficient
than all other fabrics tested in reducing
emissions in all size ranges of fibers measured.
2. Raw asbestos fibrous waste emits fewer fibers of
length >_ 6.0 ym than does asbestos cement waste
for equal dust loadings by mass in the air stream.
3. For the air-to-cloth ratios studied (0.46-
1.22 m^/min/m2 or 1.5-4.0 cfm/ft2), the optimum
ratio was 4.0 cfm/ft2.
-------
4. There is some evidence to indicate that a decrease
in relative humidity may reduce the outlet con-
centration of fibers :> 6.0 jjm.
5. The combination of long period between shakes and
short shaking duration produces significantly
lower outlet concentrations than does that of
short period between shakes and long shaking
duration.
6. Higher shaking amplitudes produce lower outlet
concentrations.
7. Outlet concentration is not a significant function
of stabilization period for periods greater than
24 hours.
8. A bag series system of two baghouses in series is
not significantly more efficient than is a single
baghouse in a stabilized condition.
9. Recycling the exhaust from a section of stabil-
izing new bags to a previously stabilized section
may drastically reduce the high initial outlet
concentrations from a new bag.
The pressure drop across the fabric filter was found to
be prohibitively high (>_ 5.0 in. ^0) in the stable con-
dition in most tests at the low values of shaking amplitude,
frequency, and duration of 0.875 cm, 1.0 cps, and 20 sec,
respectively. However, it was found that when the high
values of either shaking amplitude (3.500 cm) or frequency
(5.0 cps) were employed, the resulting pressure drops were
quite reasonable (<_ 2.0 in. H20) . Thus, the most economical
alternatives of cotton sateen bags, high air-to-cloth ratio,
and low pressure drop operating conditions have been shown
to be among the most significant in reducing asbestos
emissions.
It should also be noted that the results from the
present sampling and analysis methodology for counting as-
bestos fibers in a gas stream is highly unreliable. Many
inconsistencies were found in the data and results; it was
-------
only through extensive use of statistical techniques that
the relationship between the operating parameters and con-
trol efficiency could be established.
-------
SECTION 2
RECOMMENDATIONS
The study has shown that several baghouse operating
parameters significantly affect the outlet concentrations
from baghouses controlling asbestos emissions. Values of
these operating parameters recommended for industry usage
are: (1) cotton sateen bags, (2) an air-to-cloth ratio of
1.22 m3/min/m2 (4.0 cfm/ft2), (3) a combination of period
between shakes in excess of 120 min with a shaking duration
of 20 sec, and (4) a shaking amplitude of 3.500 cm.
Further study of all of these parameters in extended
ranges and with more intermediate values would be valuable.
An understanding of the interactions of the mechanical
shaking variables with either the bag fabrics or the air-
to-cloth ratio could prove to be very worthwhile. Initial
studies of bag fabric, air-to-cloth ratio, relative humidity,
and dust loading were made under the assumption that the
maximum dust caking conditions of the shaking variables
would produce the lowest outlet concentrations. This was
shown to be incorrect in the study of the mechanical shaking
variables. After establishing the most desirable operating
parameters, it would be of great value to perform a field
demonstration at an existing industrial installation.
Methods of sample preparation and counting of asbestos
fibers should also be statistically studied to improve
reliability and repeatability. This would be especially
valuable if the smaller fibers are found to be a major
-------
health problem. The present method of counting by optical
microscopy has been shown to be subject to high variability
in the presently regulated size range ^5.0 ym and is even
less reliable for smaller fibers. Studies to improve
analytical methodology are presently being undertaken in the
electron microscope range, but not in the range of optical
microscopy.
-------
SECTION 3
INTRODUCTION
Asbestos has been shown to be a health hazard and a
carcinogen. Control of atmospheric emissions of asbestos
has been made mandatory under Section 112 of the Clean Air
Act. It is necessary for the Environmental Protection
Agency to assess control methodology for these emissions
and to establish the best available technology based on
optimum operating conditions. It is then possible to intro-
duce legislation and promulgate regulations that will
require the application of operating practices capable of
protecting the public health. The applicability and effec-
tiveness of these practices must be supported by sound
scientific procedures and experimental evidence.
During Phase I of this study, it was shown that cur-
rent control devices emit very large numbers of small
fibers. Diffusion modelling inferred that these fibers
travel large distances from the source, and it is suspected
that the very small submicron fibers may remain suspended
indefinitely. As baghouses were found to be the accepted
best method of reducing asbestos emissions, it was decided
to conduct an experimental study to optimize baghouse per-
formance for controlling emissions of asbestos fibers.
Baghouse operating parameters were varied to establish
optimum operating conditions for minimizing the number of
fibers in the outlet. Parameters that were studied in a
statistically designed experimental plan were: (1) filter
-------
fabric, (2) air-to-cloth ratio, (3) dust loading, (4) rela-
tive humidity, (5) shaking amplitude, (6) frequency, and
(7) duration and time between shaking cycles. Also included
in the original test plan were the alternative option of a
cyclone pre-cleaner and double filtration by a series bag
arrangement.
The first sets of samples were analyzed by optical and
transmission electron microscopy. After these methods were
found to be unreliable from the standpoint of reproduci-
bility (especially for fibers in the 1.5-5.0 vim range),
additional real time data were taken using the Royco light
scattering instrument. The latter method gives outputs
that are only suitable for comparison on a total particulate
basis. The Royco instrument is calibrated for spherical
rather than fibrous particles.
-------
SECTION 4
DEVELOPMENT OF EXPERIMENTAL PLAN
Fabric filtration operating parameters were first
studied in the literature and by contacts with the asbestos
industry to determine the parameters most likely to have a
major influence on efficiency and the operating conditions
most commonly in use. The complete literature search on
control methods and compilation of control equipment user's
data in the asbestos industry was reported in the Phase I
report (EPA-650/2-24-74-088). In this report, the operating
parameters of fabric filtration were listed and evaluated on
the basis of the literature to determine those variables
which would be most worthy of experimental study from both a
technical and economic standpoint.
Table 1 contains the list of control options. The
major discussion of each option is given in the following
sections.
OPTIONS BEFORE THE BAGHOUSE
It has been shown by Timbrell2 that asbestos fibers can
be aligned by the use of an electrostatic field. This pre-
sents the possibility of aligning asbestos fibers so that
they would all strike the fabric filter broadside, and thus
their high aspect ratio could always be utilized to increase
efficiency. However, due to the experimental nature of this
development and the high efficiencies (> 99%) reported by
industry for the baghouses presently in use, this option was
not studied.
8
-------
Table 1. COMPLETE LIST OF CONTROL OPTIONS FOR BAGHOUSE
A. Options Before the Baghouse
1. Alignment of fibers by the use of electromagnetic
fields .
2. Alteration of the state of agglomeration by:
a. the use of ultrasonics
b. aerodynamic changes
3. Changing the fibers' surface properties by altering
the relative humidity of the system.
4. Optimize the air stream in terms of:
a. temperature
b. flow rate
c. flow rate fluctuations
5. Optimize the dust loading in terms of:
a. fluctuations
b. total dust loading
6. Use of secondary pre-cleaner to remove major fraction
a. cyclone
b. scrubber
c. impinger device (e.g., Pentapure)
B. Options Within the Baghouse
1. Bag construction factors:
a. fabric weave
b. fabric denier
c. type of thread (staple or filament)
d. type of fiber
e. thread count
f. fabric texture
g. physical properties (e.g., tensile strength,
wear rate, electrostatic charging)
h. quality control in manufacture
i. pre-treatment of the fabric
j. bag seams (stitched or bonded)
2. Improve baghouse design:
a. optimize physical arrangement of bags
b. optimize bag dimensions
c. install
-------
Electrostatic charging of both the filter fabric and
the particles being collected was discussed by Strauss3. It
has the effect of increasing particle agglomeration; however,
dust cake release may be hindered depending on the type of
filter fabric, humidity, and conductivity of the particles.
Another method of altering the state of agglomeration
of particles before collection is by the use of ultrasonics.
Strauss3 discusses the use of sonic agglomeration as a
primary collection device. It achieved reasonable efficien-
cies for sulphuric acid mist (96-99.5%), carbon black (82%),
zinc oxide fumes (78%), and lead oxide fumes (95-98%). No
attempts of using sonic agglomeration for collection of a
fibrous material such as asbestos were reported. It is,
therefore, conceivable to use sonic agglomeration as a pre-
cleaner; however, high cost compared to a cyclone would make
it uneconomical. Installation costs were reported as 15%
less than an equivalent electrostatic precipitator. Oper-
ation and maintenance costs are high, and high efficiency
muffling devices are required, which puts this option out of
the range of practicality.
The state of agglomeration can also be altered by aero-
dynamic changes before the primary collector. The most com-
mon means of increasing agglomeration is by inducing a state
of turbulence. However, high turbulence already exists both
in the ductwork before the fabric filter, and within the
filter itself. This option, therefore, is inherently includ-
ed and very elaborate, and long ducting systems would be
necessary to bring about any significant improvement in the
state of agglomeration.
The effect of relative humidity on fabric filter per-
formance was studied by Durham and Harrington1*. The rela-
tive humidity was controlled between 20 and 60 percent.
Using 4.0 ym median diameter fly ash as a test dust, 11
10
-------
different fabrics were evaluated. Details of the fabrics
are given in Table 2. The results obtained for the variation
in outlet concentration and efficiency are shown in Table 3.
It can be seen that cotton is by far the most efficient fab-
ric tested having a mass collection efficiency of greater
than 99.99% for all humidities tested. Humidity can be seen
to, have a marked effect on the continuous filament fabrics,
that is, all except cotton. It is conjectured that the
fibrous projections on spun cotton yarn are responsible for
the high mass collection efficiency at all humidities. The
efficiency for cotton bags is so high that the effect of
humidity was beyond the sensitivity of the experiment to
detect. However, it is reasonable to suppose that with
increased experimental sensitivity, an increase in collection
efficiency with increase in humidity would be observed with
cotton bags.
An interesting finding of Durham and Harrington was
that, while humidity had a marked effect on the collection
efficiency using fly ash as the test dust, there was no
apparent effect when using cement dust, pulverized lime-
stone, or amorphous silica.
It can be seen from the above that humidity has the
capability of drastically modifying collection efficiencies.
Therefore, it was decided to include humidity as an import-
ant variable. Experiments were done using asbestos dust to
establish the effect of humidity on collection efficiency as
a function of size.
The air stream may be optimized with respect to temper-
ature. Resistance to high temperatures is one of the pri-
mary considerations in the choice of filter fabric. However,
in most asbestos processing applications, high temperature is
not a problem. In conjunction with the relative humidity,
the temperature affects the adhesion of particles as reported
11
-------
Table 2. TEST FABRIC CHARACTERISTICS
Fiber
Composition*
Nylon
Creslan
Dacron
Polypropylene
Crylon
Ti
Dralon
^ -, R
Orion
Cotton sateen
Glass filament
Glass combination
Glass texturized
Type Yarn**
Warp
CF
CF
CF
CF
CF
CF
CF
S
CF
CF
CF
Fill
CF
CF
CF
CF
CF
CF
CF
S
CF
S
T
Yarn Dernier
Warp
210
200
250
210
200
200
200
Fill
210
200
250
210
200
200
200
Thread Count,
threads /in.
Warp
74
80
76
81
77
76
76
95
54
48
46
Fill
68
76
66
69
63
71
62
58
56
22
24
Weave
Pattern
2x2 Twill
3x1 Twill
3x1 Twill
3x1 Twill
3x1 Twill
3x1 Twill
3x1 Twill
Satin
3x1 Twill
2x2 Twill
3x1 Twill
Fabric
Thickness,
Mils
9.4
10.8
9.1
12.4
10.2
9.8
8.7
24.1
9.6
24.6
16.5
Fabric
Weight
oz/yd^
4.1
4.0
3.9
4.6
5.1
4.4
4.3
10
9
16.5
14
* Creslan acrylic, Amer. Cyanamid; Dacron polyester, DuPont; Crylon acrylic, Crylon S.A.(Fr);
Dralon , Farberfabriken Bayer (W.Ger.); Orion* acrylic, DuPont.
** CF = continuous filament; S = staple; T = texturized.
-------
Table 3. EFFECT OF RELATIVE HUMIDITY ON OUTLET DUST CONCENTRATION AND EFFICIENCY
Bag
Description
Nylon
Creslan
R
Dacron
Polypropylene
CrylorR
Dralon
OrlonR
Cotton
Filament glass
Combination glass
Texturized glass
Outlet Dust Concentration,
grains/ 1000 ft3
Relative Humidity, %
20
130
168
34
36
148
26
12
0.04
148.1
10.4
63.8
30
148
177
32
32
89
24
7.5
0.2
135.9
10.3
40.2
40
61
100
13.1
35
56
17
6.9
0.0
106.4
2.2
19.9
50
4.4
37
1.9
7.0
13
0.8
3.9
0.0
25.4
0.1
6.0
60
0.02
3.1
Q.7
2.7
1.3
0.6
0.8
0.0
9.1
0.1
1.1
Efficiency, weight %
Relative Humiditv. %
20
95.62
94.47
98.86
98.80
95.12
99.11
99.59
99.99+
95.00
99.65
97.84
30
95.02
94.02
98.95
98.96
97.29
99.23
99.75
99.99+
95.31
99.66
98.66
40
98.02
96.35
99.12
98.85
98.14
99.43
99.78
99.99+
96.55
99.92
99.31
50
99.86
98.78
99.94
99.78
99.56
99.98
99.87
99.99+
99.18
99.99+
99.80
60
99.99+
99.90
99.98
99.91
99.96
99.99+
99.97
99.99+
99.71
99.99+
99.96
Note: Inlet dust concentration C. = 3.0 gr/ft , fly ash, 4.0 ym median diameter.
-------
by Werle5. These two parameters affect the agglomeration
characteristics of particles and their adherence to the
fabric. Therefore, as relative humidity was varied over a
wide range, temperature was only varied over a limited
ambient range that simulated anticipated environmental con-
ditions within asbestos plants.
The gas flow rate was varied in order to optimize the
air-to-cloth ratio for the fixed filter area that will be
used. In order to vary the air-to-cloth ratio from 0.46 to
1.22 m3/min/m2 (1.5-4.0 cfm/ft2), the gas flow rate was
varied from 0.655-1.740 m3/min (23.1-61.5 cfm) for two bags.
This variation in flow rate affects the velocity at the face
of the filter. The major collection mechanisms are dif-
fusion to the fabric filter at low velocity and inertial
impaction and interception at high velocity. Therefore, the
flow rate variations in effect test the relative effective-
ness of these collection mechanisms.
Asbestos processing rarely results in a constant flow
rate through the filter. As the filter cake builds up, the
pressure drop increases until the bags are cleaned, then the
flow rate decreases again in a regular cycle. Stafford and
Ettinger6 reported that, as the filter becomes loaded, the
efficiency increases for velocities less than 20 fpm; but
for velocities greater than 100 fpm, the efficiency initially
increases and later decreases. These efficiency fluctua-
tions occur in the periods between each cleaning cycle and
were studied.
Industrial plants operate under a wide range of dust
loadings at the filter. Therefore, the experimental appara-
tus was made to be capable of varying the dust loading from
10 to 45 g/m (4.4-19.7 gr/ft3) over the entire range of
air-to-cloth ratios in order to simulate the mid-range of
actual conditions. Stenhouse7 reported that, as dust load
14
-------
increases, the efficiency gradually: (1) decreases for
particle diameters of 40 to 45 ym, (2) peaks at about 0.7 g
for particles of 10 to 15 ym, and (3) peaks at about 7.0 g
for particles of 0.0 to 5.0 ym. Therefore, this variation of
efficiency with total dust load was investigated for a fabric
filter collecting asbestos.
In Phase I of the program, the user's survey showed that
cyclone-baghouse and scrubber-baghouse combinations are used
by only 4.4% and 1.1%, respectively, of the plants questioned.
The cyclone-baghouse combination, which is the most prevalent
type of combination used, was to be investigated. The cy-
clone is used as a pre-cleaner and removes the major fraction
of particles greater than 20 ym in effective diameter, so
that the baghouse is not loaded as quickly with large fibers.
However, when it was found during testing that increased
dust loadings marginally increased the collection efficiency,
this option was suspended.
Wet processes such as scrubbers and impinger devices are
more often used as primary collection devices rather than as
pre-cleaners for baghouses. Some wet collectors are being
successfully used to remove asbestos particles above 5 ym;
however, an EPA study8 found the Pentapure to be very ineffi-
cient for sub-micron particles from a gray iron foundry. For
systems such as these, it is generally difficult to follow
them with a fabric filter after cleaner because the high
moisture content following the wet collector would cause
blinding of the fabric filter.
OPTIONS WITHIN THE BAGHOUSE
The effect of weave on filtering efficiency was studied
by Draemel9 in a single compartment test baghouse using a
fly ash test dust having a mass median diameter of 3.7 ym
at an inlet concentration of 6.86 g/m (3.0 gr/ft ). Fabrics
tested were of Dacron with a 76 x 63 thread count made with
15
-------
both continuous filament and staple yarns in six different
weaves. Results are shown in Table 4. For both types of
yarn, the 3x2 twill was the most efficient (> 99.9% and
95.3%).
Fiber denier is a means of expressing fiber weight by
expressing the weight in grams of 9,000 m of fiber. Gener-
ally, smaller fibers produce higher efficiencies and lower
pressure drops.
The type of thread used is either staple or filament.
Dick10 stated that natural fibers are generally used in a
staple (spun) form, while synthetics are used as filaments
or artificial staple form (by spinning short or chopped
filaments). In the study by Draemel mentioned above, the
relative efficiencies of filament and staple Dacron was
studied. The filament yarn group showed higher efficiencies
(98-99%) than did the staple yarn due to the smaller free
areas in the filament weaves.
Many types of fibers are commercially available in the
form of filter bags. Natural fibers such as cotton and wool,
and synthetics such as Dacron polyester, nylon, and Nomex
nylon are most commonly used for asbestos. The type of fiber
determines the temperature, abrasion, and chemical resis-
tance of the bag as were reported for the most common fabrics
in the Phase I report.
Thread count is determined by the number of threads in
the warp (lengthwise) and fill (widthwise) directions.
Increasing the thread count decreases the pore size or free
area, and hence improves the efficiency of the fabric.
Spaite and Walsh11 have shown that a small change in the
thread count can have a significant effect on the fabric's
efficiency.
Fabric texture is dependent on the finish of the fabric.
Fabrics may be napped to expose more surface area for
16
-------
Table 4. EFFICIENCIES FOR CLOTHS OF DIFFERENT WEAVES
Weave
3x1 Twill
3x2 Twill
2x2 Twill
Plain
Satin
Crowfoot
3x1 Twill
3x2 Twill
2x2 Twill
Plain
Satin
Crowfoot
Yarn Type
Filament
it
it
it
it
M
Staple
M
ii
M
M
ii
Free Area
0.001
0.0
0.0
0.002
0.0
0.0
0.139
0.130
0.172
0.139
0.155
0.169
Outlet Concentrations
R/103m3
112.00
4.80
8.67
8.44
15.80
6.66
805.00
323.00
760.00
428.00
977.00
444.00
(Rr/103ft3)
(49.10)
(2.10)
(3-79)
(3.69)
(6.89)
(2.91)
(352.00)
(141.00)
(332.00)
(187.00)
(427.00)
(194.00)
Weight Efficiencies
%
98.36
99.93
99.87
99.88
99.77
99.90
88.27
95.30
88.93
93.77
85.77
93.53
Note: All fabrics have 76 x 63 thread count.
Filament yarns are 250/50 warp and fill Dacron.
Staple yarns are 250 equivalent denier warp and fill Dacron.
-------
collection, therefore increasing efficiency. Napped fabrics
are harder to clean, but Dick states that they are useful
for light dust loads at low pressure drops, and high air-to-
cloth ratios. The physical properties of the bag are deter-
mined by the fabric used. The tensile strength should be at
least 9.0-17.9 kg/cm (50-100 Ibs/in.) for abrasion resistance
and dimensional stability9. Dimensional stability may be a
problem with synthetics, some of which may either stretch
with weight or shrink at high temperature, thus changing
fabric porosity and permeability. Wear rate is largely
dependent on the abrasion resistance. Abrasion causes either
yarn failure (surface abrasion) or intrayarn (fiber to fiber)
abrasion. Generally, filament fibers are more resistant than
the staple form. The permeability or resistance of the fab-
ric is generally determined by the fiber and the weave.
Pressure is equal to permeability times linear velocity by
Darcy's Law. Other physical and chemical properties of the
fabric determine its applicability; however, for use on as-
bestos bearing dusts, these are generally not of concern.
A bag manufacturer's quality control and method of con-
struction can be important. Both the reliability and special
features of the bag can be influenced by quality control.
This variable cannot be easily studied or quantified.
Fabric pre-treatment is discussed by Billings and
Wilder12. Since asbestos floats are usually used for this
pre-treatment, the benefits of pre-treatment are inherently
achieved in asbestos collection. The bridging of the fabric
pores by asbestos fibers does increase efficiency.
Bag seams may be either stitched or bonded. Stitched
bags have the disadvantage of an uneven velocity distribution
across the seam which may cause unequal filtration. There-
fore, the seam length should be minimized. Bonded seams are
less of a problem but should still be minimized.
18
-------
It was decided to select six bag fabrics for the actual
study, subject to the statistical experimental design. Three
each of natural and synthetic fabrics were chosen. In each
case, the weave chosen was that most likely to be the most
efficient rather than the most economical, subject to the
fabrics' commercial availability.
Physical arrangement of the bags is usually determined
by maintenance considerations. Bags or envelopes are arranged
to facilitate inspection and replacement. As only one or two
bags, or several bags in series, were used in this project
within the small experimental baghouse available at IITRI,
the rearrangement of bags to improve performance was beyond
the scope of this project.
Bag dimensions and shape may also be optimized to improve
efficiency. Some manufacturers use tube shaped bags while
others use envelopes. Dimensions also vary with the manu-
facturer within the general limit of a length to diameter
ratio of 30:1. For the particular baghouse used in this pro-
ject, only one size and shape of bag can be used. Therefore,
the bag was a tubular one of 12.7 cm D x 178.0 cm L
(5 in. D x 70 in. L).
Installation of baffles to evenly distribute air flow
and to cause initial inertial separation is sometimes util-
ized in industry. As the authors used a two bag Y-shaped
entry system with an evenly distributed dust loading from
the dust feeder, this option was not necessary.
Air-to-cloth ratio was tested within the experimental
09 2
ranges of 0.46-1.22 m /min/m (1.5-4.0 cfm/ft ) for mechani-
cal shaking. The typical range of air-to-cloth ratios for
asbestos was found to be less than 0.92 m /min/m (3.0 cfm/ft )
in Phase I and 0.77-0.92 m3/min/m2 (2.5-3.0 cfm/ft2) by
Strauss. Rozovsky13 stated that the preferable ratio was
0.61-0.74 m3/min/m2 (2.0-2.5 cfm/ft ), while the economical
19
-------
ratio was 0.92 m3/min/m2 (3.0 cfm/ft ). Therefore, it was
supposed that the optimum air-to-cloth ratio would fall with-
in the range to be experimentally tested. - •''
There are many bag cleaning methods commercially avail-
able including mechanical shakers, reverse air, and pulse
jet. These three methods are the most commonly used methods;
however, the pulse jet method requires special equipment
which is not available on a small scale. It was initially
intended that a reverse air mechanism using a traversing blow
ring would be used; however, this was found to be commercially
unavailable in the size required and infeasible to construct.
Mechanical shaking is the most common method used in the as-
bestos industry and was studied.
During the cleaning cycle, bags are taken out of line,
shaken, and then put back into line. Goldfield11*, working
with asbestos as a test dust, reported that a photometer
placed on the outlet of the baghouse indicated a marked
surge in the dust concentration after the bags had been
shaken. He reported a period of two to three minutes before
steady state exit concentrations were achieved.
Goldfield11* and Dennis15 noted that different materials
gave characteristic effluent dust concentration vs. time
curves. Synthetic cloths gave higher peaks and took longer
to reach a low value. This is thought to be due to the bet-
ter cake release and lower adherence characteristics of
synthetics. The study by Goldfield was lacking in quantita-
tive data. However, it did indicate the necessity to study
this cycle in emissions in relation to the total emission
for the options of recycling and pre-caking of the filters.
The effect of varying the shake rate and duration on the
minimum filter drag was studied by Billings and Wilder12.
Minimum drag decreases with increasing shake duration, thus
decreasing the pressure drop and increasing the filter
20
-------
velocity. This temporarily decreases filter efficiency which
increases again as the filter cake builds up. This increase
continues until the next shaking cycle is initiated.
For this study, the variables of shake amplitude, fre-
quency, duration, and interval of the mechanical shaker were
to be investigated to as great a degree as determined by the
statistical design of the experiments.
Dual bags are used by some manufacturers. These bags
consist of two fabric filters, one within the other. This
type of dual bag was not readily available for the size of
baghouse to be used; therefore, this option was not investi-
gated.
OPTIONS AFTER THE BAGHOUSE
Use of a second baghouse in series with the first has
not, to our knowledge, been investigated previously for econ-
omic reasons. The efficiency of one baghouse has generally
been deemed sufficient without further cleaning of the gas
stream. However, this option, which is similar in effect to
that of the dual bag, was investigated. This option was
feasible because of the small scale of the laboratory experi-
ment. Two bags, or two sets of two bags, can be connected in
series with the same overall effect as having two baghouses
in series.
A high cost option would be to place an electrostatic
precipitator in series after the baghouse. The purpose would
be to collect the very fine submicron particles not collected
by the baghouse. This option is especially attractive in that
electrostatic precipitators are more effective if subjected
to a light dust loading. Another possibility would be to
precede the baghouse with an electrostatic precipitator so
that the fabric filter could take advantage of the electro-
static charging of the uncollected fibers. This option was
not investigated during this program due to economic limitations
21
-------
Efficiency of a newly cleaned filter bag is greatly
decreased until the filter cake rebuilds. This is the basis
of the concept of pre-caking a filter before use. Generally
in industry, one section is cleaned as a unit and then put
back on line at lowered efficiency. This practice tends to
keep the efficiency of the entire collector somewhat uniform,
but not at its optimum. This option could be studied with
the same apparatus and testing as needed for the two bag-
houses in series option. Therefore, the authors investi-
gated the improved performance expected by recycling the
exhaust from a bag in which the filter cake was rebuilding.
Thus, the following reduced list of options (see Table 5)
affecting baghouse performance was actually investigated. The
initial experiment design was based on this reduced list
of options. Limitations imposed by theoretical, apparatus,
operational, and time considerations necessitated consider-
able modification to this initial design as experiments
progressed.
Ranges of the values of the variables were chosen in
accordance with the literature and the industrial user's
survey as stated above. After consultation with the EPA16,
several ranges were adjusted, i.e., the range of the shaking
variables was modified to increase the probability of dust
cake build-up. Other variable's values were modified
slightly because of design considerations. The number of
levels for each variable was limited to three by the statis-
tical design in order to limit the tests in the experimental
test plan to a reasonable number. Upper and lower values
were chosen for each variable, and then the middle value was
determined by the average of the logarithms of the extreme
values. Levels of the variables actually tested are given
in Table 6.
22
-------
Table 5. REDUCED LIST OF OPTIONS FOR BAGHOUSE
A. Options Before the Baghouse
1. Changing the fibers surface properties by altering
the relative humidity of the system.
2. Optimize the air stream in terms of:
a. flow rate
b. flow rate fluctuations
3. Optimize the dust loading in terms of:
a. fluctuations
b. total dust load
4. Use of cyclone to remove major fraction.
B. Options Within the Baghouse
1. Bag construction factors:
a. fabric weave
b. fabric denier
c. type of thread (staple or filament)
^d. type of fiber
e. thread count
f. fabric texture
g. physical properties (e.g., tensile strength,
wear rate)
2. Improve baghouse design:
a. optimize air-to-cloth ratio
b. study mechanical shaking in terms of amplitude,
rate, duration, and frequency.
C. Options After the Baghouse
1. Use of a second baghouse.
2. Develop a method to recycle exhaust from newly
cleaned bags through a caked bag.
23
-------
Table 6. FINAL LIST OF OPTIONS FOR BAGHOUSE
A. Options Before the Baghouse
1. Type of waste collected:
a. asbestos cement processing
b. raw fiber asbestos
2. Relative humidity:
a. 20% (modified to 30% or ambient in Phase II)
b. 40%
c. 80% (modified to 60% in Phase II)
3. Optimize total dust loading.
a. 10 g/m3,
b. 21 g/mj*
c. 45 g/m
4. Use of cyclone to remove major fraction (found unpro-
fitable in Phase II).
a. cyclone
b. without cyclone
B. Options Within the Baghouse
1. Bag construction.
a. cotton sateen, 96 x 60 thread count, 9.7 oz/yd2
b. napped cotton, 98 x 60 thread count, 8.7 oz/yd^
c. cotton twill, 73 x 60 thread count, 7.4 oz/yd?
d. Dacron twill, 75 x 71 thread count, 5.8 oz/yd
e. Dacron twill, 64 x 51 thread count, 8.5 oz/yd2
f. Nomex twill, 95 x 60 thread count, 5.2 oz/yd2
2. Improve baghouse design:
a. optimize air-to-cloth ratio (dependent on flow
rate for constant filter area, i.e., number
of bags)
(1) 0.46 nu/min/m2, (1.5 cfm/ft?)
(2) 0.76 m,/min/m, (2.5 cfm/ft,)
(3) 1.22 mj/min/n/ (4.0 cfm/ftz)
b. study mechanical shaking
(1) amplitude
(a) 0.875 cm
(b) 1.750 cm
(c) 3.500 cm
(2) frequency
(a) 1.0 cps
(b) 2.2 cps
(c) 5.0 cps
(3) duration
(a) 20 sec
(b) 40 sec
(c) 80 sec
(4) time between shake cycles
(a) 16.0 min
(b) 42.0 min
(c) 120.0 min
3. Bag series:
a. in series
b. not in series
24
-------
The Phase I experimental design was modified insofai
as testing order so that all tests with the same filter fab-
ric and waste type could be made in order of increasing
humidity. Thus, the stabilized set of bags would be moved
as little as possible in order to maximize dust cake reten-
tion. Due to feeding problems with the raw asbestos fiber
waste, only two fabrics were tested using this waste mater-
ial. Raw asbestos fiber waste is representative of the
asbestos milling process and asbestos fabric industry.
Installations which perform these processes are very few in
number compared to the number of facilities generating asbestos
cement waste.
As no further testing was conducted with the fibrous
waste, the number of tests in Phase II was also greatly
reduced. Due to time limitations and the wide scatter of
data up to that point, the studies of the second fabric,
Nomex, were not conducted in Phase II. Loss of this data
is not a major problem in that, because of expense, Nomex is
used only on high temperature gas streams. High temperatures
are not typical of asbestos industry.
After Phase II of the test program, a critique and re-
evaluation of the test program was undertaken. This was
decided upon in consultation with the EPA16, due to the fact
that the counting error and other factors contributing to
error were, up to that time, of the same order of magnitude
as the differences between the results of different tests.
Therefore, it was decided to delay all further testing and
electron microscope work until these problems were consid-
ered more fully. The critique and reevaluation were carried
out with the cooperation of Mr. Richard Gerber of Aerospace
Corp.
Feasibility of alternate sampling and sizing methods
were considered for application to fibers. Those considered
25
-------
were: (1) optical microscope, (2) electron microscope,
;3) Royco and other light scattering techniques, (4) Anderson
impactor and other inertial techniques, (5) mobility analyzer,
and (6) techniques using the Condensation Nuclei Counter (CNC)
including the diffusion battery. It was decided to optically
count all fibers, but to analyze only those greater than 5 or '
6 ym in length as they are most clearly viewed and measured,
thus producing the least statistical error. Experiments with
increased sampling time were also to be conducted to deter-
mine the effect on data reliability. Electron microscope
counting was suspended until statistical questions could be
resolved. Royco total number of particle concentrations for
equivalent diameters would be recorded in as many size ranges
as possible. The Royco traces would only be suitable for
comparison, as total particles are measured for an equivalent
diameter dependent on fiber orientation. It was also decided
to employ intermittant sampling with the CNC as an additional
corroborating tool.
The possible program alternatives, from total cancell-
ation or suspension of the program to drastic modification
of the test series, were considered. It was decided to use
new bags for each test and to stabilize them until both the
pressure drop and the Royco traces were constant. Only
cotton sateen bags would be tested at an air-to-cloth ratio
of 0.92 m3/min/m2 (3.0:1 cfm/ft2), the highest ratio in com-
32 ?
mon use and as near to 1.22 m /min/m (4.0:1 cfm/ft) as
could be sustained for extended stabilization. The dust
3 3
loading would be 45 g/m (19.7 gr/ft ) and the humidity at
ambient levels, in accordance with the statistical analysis
of Phase III. However, the dust loading was later reduced
to 21 g/m (9.2 gr/ft ) to enable continuous 24 hour oper-
ation for increased stabilization.
In order to allow time for the increased stabilization
of new bags for each test, Phase III testing was reduced to
26
-------
a total of eight tests. Extreme conditions of the shaking
variables were to be tested in a search for order of magni-
tude changes in outlet concentrations. It was decided to
conduct four tests at both the lowest dust caking conditions
(16 min cycle with 80 sec duration) and the highest dust
caking conditions (120 min cycle with 20 sec duration) . The
four tests were of the 0.875 and 3.500 cm (0.344 and 1.378 in.)
shake amplitudes with both 1 and 5 cps shake frequencies. If
order of magnitude or statistically significant differences
in outlet concentration were detected in this modified
Phase III plan, then a Phase IV test would be run. This
would consist of one stabilization and testing series on the
double filtration system using the optimized system as the
primary filter and a pre-stabilized bag as the secondary
filter. If significant differences were not detected, the
remainder of the program was to be devoted to improving
sampling and sizing techniques. The cyclone pre-cleaner
option was suspended due to time limitations and the fact
that Phase II analysis showed a correlation between increased
dust loading and decreased outlet concentration.
27
-------
SECTION 5
EXPERIMENTAL APPARATUS
The baghouse used for the project was a No. 1 Model
70-BC Assembled Intermittent Wheelabrator Dustube Dust
Collector manufactured by Wheelabrator-Frye, Corp. This
baghouse has a capacity of twelve filter bags. The bags
are of a nominal size of 12.7 cm D x 178.0 cm L (5 in. D x
70 in. L). For most tests, only two filter bags were used at a
time. For this arrangement, a special Y-shaped inlet adaptor
with two conical hoppers was constructed and installed. The
purpose of the adaptor is to maintain similar flow conditions
in the inlet plenum as would exist for all twelve bags. The
remaining holes for bags were closed with gasketing, sealant,
and metal plates to ensure no flow through them.
The manual shaker mechanism was adapted to automatic
mechanical operation. A 1/4 hp, 3,450 rpm motor with a
variable transmission producing 0-675 rpm was mounted on the
rear of the baghouse. The motor drive was then connected
to the shaking lever by a cam with three eccentric positions
producing shaking amplitudes of 0.875, 1.750, and 3.500 cm
(0.344, 0.689, and 1.378 in.). The transmission was cali-
brated by stroboscope with nearly linear results to the
values given in Table 7. The desired shaking frequencies
of 1, 2.2, and 5 cps are equivalent to settings 1, 3, and 8.
An automatic timing system based on a 4-pole, 16-min
adjustable cammed timer and a single-pole or 120-min adjust-
able cammed timer was constructed. A fused 220 V AC line
28
-------
Table 7. SHAKER ASSEMBLY MOTOR CALIBRATION
Gear Setting
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Motor Speed (rpm)
58*
93*
132
150
192
230
266
306
344
380
417
453
490
524
558
590
621
655
685
700
* Extrapolated
29
-------
was used to power the timers, as well as the timer switches
and relays used to control the power supply to various 110 V
outlets (see Figure 1, Schematic of Automatic Timer System).
The timer system is set for a specific shaking duration
(preceded and followed by one minute of settling) and time
between shakes, and controls the operation of the main
blower, an optional auxiliary blower, the dust feeder, sam-
pler, shaker, and pressure transducer and chart.
Filter fabrics chosen for experimental study included
three cottons, two Dacrons, and one Nomex. Cotton is the
most commonly used fabric for asbestos control in industry
and has been found by Dennis15 to have dramatically lower
outlet concentrations over a 30 min cycle than does Dacron
for fly ash. The three cotton fabrics chosen were a sateen
(the most commonly used), a napped fabric (more efficient
in some applications) , and the only twill commercially avail-
able. Draemel's9 work indicated that, for fly ash, a
3 x 1 or 3 x 2 twill with as high a thread count as possible
would be the most efficient. Dacron fabrics were chosen
as they are the second most commonly used fabric in the as-
bestos industry. Nomex is the most common fabric for high
temperature asbestos applications. The filter fabrics and
their characteristics are given in Table 8.
The original dust feeder used was an IITRI-built screw
feeder modified with a more powerful variable speed motor
capable of breaking large pieces (^3 cm [1.18 in.]) of as-
bestos cement. However, it was found that this feeder would
jam after periods longer than 30 min. Vibra-Screw, Inc.
then graciously offered to lend a SCR-20 feeder for the
duration of the project. This feeder maintains a constant,
linearly variable feed rate over the required range of
5-60 g/min (2.2-26.2 gr/ft ) using asbestos cement waste
sifted through a No. 4 mesh. The calibration curve of the
30
-------
1
110 V
HO >
v v v — — ~
30 Amp
Fuse
' AC
C
Switch
\(Normally
Open For
I6min. Cycle)
>v Timer \
) Motor 1 >
y42,l20min.(;
/AC
(
30 Amp
Fuse
L Timer 1
\ Switch,
(Normally
Closed)
Relay 1 (
Coil V
Timer 2
\Switch 2
(Normally
Closed}
Relay 3 (
Coil V
•
Relay 1
\ Switch
(Normally
Closed)
) Timer v
Motor 2 f
ISmin. f
Relay 3
\Switch 1
(Normally
Closed)
^Dust (
) Feeder V
Timer 2
\Switch 1
(Normally
Closed)
Relay 2 (
Coil V
Relay 3
\Switch 2
(Normally
Closed)
J"impler /
itlet >
plugged r
iring f
iiaoilization)
Relay 2
\Switch 1
(Normally
Closed)
\3lower (
J ' ^
Timer 2
\Switch 3
(Normally
Open)
Relay 4 (
Coil I
Relay 2
\Switch 2
(Normally
Closed)
^ Blower 2 /^
NOutlet I
y(Normally I
/ UnpluggedJV,^
Relay 4
\ Switch
(Normally
Open)
J Shaker f
Relay 5
\Switch 2
(Normally
Open)
) Pressure
Transducer
Timer 2
\Switch 4
(Normally
Open)
Relay 5 (
Coil V
Relay 5
\Switch 1
(Normally
Open)
A Pressure
J Chart
Figure 1. Schematic of automatic timer system
-------
was used to power the timers, as well as the timer switches
and relays used to control the power supply to various 110 V
outlets (see Figure 1, Schematic of Automatic Timer System).
The timer system is set for a specific shaking duration
(preceded and followed by one minute of settling) and time
between shakes, and controls the operation of the main
blower, an optional auxiliary blower, the dust feeder, sam-
pler, shaker, and pressure transducer and chart.
Filter fabrics chosen for experimental study included
three cottons, two Dacrons, and one Nomex. Cotton is the
most commonly used fabric for asbestos control in industry
and has been found by Dennis15 to have dramatically lower
outlet concentrations over a 30 min cycle than does Dacron
for fly ash. The three cotton fabrics chosen were a sateen
(the most commonly used), a napped fabric (more efficient
in some applications), and the only twill commercially avail-
able. Draemel's9 work indicated that, for fly ash, a
3 x 1 or 3 x 2 twill with as high a thread count as possible
would be the most efficient. Dacron fabrics were chosen
as they are the second most commonly used fabric in the as-
bestos industry. Nomex is the most common fabric for high
temperature asbestos applications. The filter fabrics and
their characteristics are given in Table 8.
The original dust feeder used was an IITRI-built screw
feeder modified with a more powerful variable speed motor
capable of breaking large pieces (^3 cm [1.18 in.]) of as-
bestos cement. However, it was found that this feeder would
jam after periods longer than 30 min. Vibra-Screw, Inc.
then graciously offered to lend a SCR-20 feeder for the
duration of the project. This feeder maintains a constant,
linearly variable feed rate over the required range of
3
5-60 g/min (2.2-26.2 gr/ft ) using asbestos cement waste
sifted through a No. 4 mesh. The calibration curve of the
30
-------
'
110 \
110 \
1
v v v
30 Amp
Fuse
/ AC
c
Switch
\ (Normally
Open For
I6min. Cycle)
\Timer >
) Motor 1 >
y42,l20min>
/AC
30 Amp
Fuse
A A A.
i Timer 1
\Switch
(Normally
Closed)
Relay 1 (
Coil I
L Timer 2
\Switch 2
(Normally
Closed)
Relay 3 (
Coil I
•
. Relay 1
\Switch
(Normally
Closed)
^\ Timer >
) Motor 2 >
J I6min. f
Relay 3
\Switch
(Normally
Closed)
)Dust [
Feeder V
Timer 2
\Switch 1
(Normally
Closed)
Relay 2 (
Coil I
Relay 3
\Switch 2
(Normally
Closed)
-^ Sampler (
\Outlet >
KUnplugged r
J During r
Stabilization)
Relay 2
\Switch 1
(Normally
Closed)
\Blower f
Timer 2
\Switch 3
(Normally
Open)
Relay 4 (
Coil I
Relay 2
\Switch 2
(Normally
Closed)
^\ Blower 2/^
\Outlet (
/Normally I
_^/ UnpluggedjN^
Relay 4
\ Switch
(Normally
Open)
J Shaker \_
Relay 5
\Switch 2
(Normally
Open)
^Pressure
yTransducer
k Timer 2
\Switch 4
(Normally
Open)
Relay 5 (
CO,, ^
Relay 5
\Switch 1
(Normally
Open)
A Pressure
J Chart
Figure 1. Schematic of automatic timer system
-------
Table 8. FILTER BAG CHARACTERISTICS*
LJ
NJ
Fabric
Warp Yarns
Fill Yarns
Thread Count
Weight
Permeability
(cfm/ft2 @ 1/2 in. H20)
Finish
Weave
Price per Bag
#101-00
Cotton
Spun
Spun
96 x 60
9.7
15-20
Woven
Sateen
4 x 1
$2.85
#101-10
Cotton
Spun
Spun
98 x 60
8.7
10-15
Desized &
Napped
4x1
$3.60
#102-00
Cotton
Spun
Spun
73 x 60
7.4
14.5
Woven
Twill
2 x 1
$3.95
#736-50
Dacron
Filament
Spun
75 x 71
5.8
15-25
Woven
Twill
3x1
$3.65
#757-52
Dacron
Spun
Spun
64 x 51
8.5
30-40
Woven
Twill
2x2
$3.60
#340-50
Nomex
Filament
Spun
95 x 60
5.2
20-35
Woven
Twill
2 x 1
$8.30
* All filter bags 12.7 cm D x 178.0 cm L (5 in. D x 70 in. L) and supplied by W.W. Criswell
Co., Division Wheelabrator-Frye Inc.
-------
Vibra-Screw SCR-20 feeder is given in Figure 2. The dust
feeder hopper has a capacity of 56.6 £ (2 ft3), allowing
unattended operation at low feed rates for long periods.
It was discovered that the raw asbestos fiber could not
be fed through the IITRI screw feeder or the Vibra-Screw
SCR-20 dust feeder. This material has a very low bulk
density of 0.117 g/cc (7.28 lb/ft3). This low bulk density
results in its tendency to mat and form a composite struc-
ture which gives it extremely bad flow properties. Even
with the vibrations applied by the SCR-20, the raw fiber
would not feed evenly into the screw feeder. Therefore, a
manually fed, steeply inclined vibrating trough feeder had
to be used to feed the fibrous asbestos. This method was
not nearly as accurate as the SCR-20 for asbestos cement
waste, but supplied average mass concentrations over a
shaking cycle of adequate accuracy and repeatability to
test for major differences in efficiency caused by waste
type.
A velocity of 300 m/min (^1,000 fpm) must be maintained
in order to keep the asbestos cement waste suspended in the
inlet duct. Velocities as high as 1,200 m/min (4,000 fpm)
are commonly used in industry. (This is only necessary to
keep non-sifted waste suspended.) To accomplish suspension
32
for the lowest air-to-cloth ratio of 0.46 m /min/m
(1.5 cfm/ft2), an inlet duct of ^5 cm diameter (2 in.
schedule 40) PVC pipe was used on the inlet and at least
two bags were used simultaneously. The inlet ducting was
made as straight as possible to minimize frictional losses
in the pipe. The downstream ducting was made of four inch
diameter stove pipe to reduce friction loss. A 745.7 w
(1 hp), 3,450 rpm centrifugal blower was used as the primary
suction fan. This blower, equipped with a sliding damper
to control the flow, was capable of maintaining the system
33
-------
LO
C
•H
s
"So
-------
at an air-to-cloth ratio of 1.22 m3/min/m2 (4.0 cfm/ft2) at
moderate pressure drops (<. 7.6 cm H20 [3.0 in. H2OJ) across
the filter bag. An auxiliary Tornado 447.4 w (3/5 hp)
blower was attached to the inlet of the air stream in order
to achieve and maintain the flow necessary to stabilize six
bags at an air-to-cloth ratio of 0.76 m3/min/m2 (2.5 cfm/ft2).
This stabilization method for treating two bags of each of
three fabrics simultaneously was used during Phase I of the
experimental plan. Velocity measurements were made by pitot
tube in both the inlet and outlet ducts.
The relative humidity was raised by injecting steam
into the inlet flow. When the steam was added after the as-
bestos waste, a moist slurry was formed which blocked the
duct at the 80% RH level. The injection point was then
changed to before the dust addition. The 2070 RH level was
often below ambient conditions, so for the second experi-
mental phase, the relative humidity levels were changed
from 20, 40, and 80% to 30, 40, and 60%. Wet and dry bulb
thermometers were used to measure relative humidity. When
located in the inlet duct after the addition of cement
waste and steam, agglomerated moist dust was impacted on
the bulbs of the thermometers making the readings inaccurate
representations of gas flow conditions. Therefore, the wet
and dry bulb thermometers were moved to the clean side of
the baghouse.
The pressure drop across the fabric filter is measured
by a 0-25 mm Hg (0-1 in. Hg) differential pressure trans-
ducer. The transducer output is recorded on a Leeds and
Northrup Speedomax G variable input chart recorder. Due to
the high chart speed (^2.5 cm/min [ \,1 in./mini) and the
desire to conserve chart paper, only the pressure drop at
the beginning and end of each filtering cycle is measured
and recorded. The period recorded extends from approximately
35
-------
one minute before the ore-shake settling period to approxi-
mately one and one-half minutes after the recommencement of
filtering after the post-shake settling period.
Particulate sampling was conducted by isokinetic sam-
pling onto a filter and by Royco light scattering. Isokine-
tic sampling was done using a modified EPA Method 5 sampling
train consisting of a short (<_ 38.1 cm [£ 15 in. including
nozzle]) probe and S-type pitot tube with the filter holder
connected directly to the sampling meter box. The isokine-
tic sampling port was a 3 x 1-1/2 in. port in the vertical
outlet section at the EPA required distances (eight diameters
downstream and five diameters upstream) from changes of
direction. The filter used was a Millipore (MF) filter of
0.8 vim pore size. A Pvoyco Model 245 particle counter with
Module 510 display was used to measure total particles in
5 3
concentrations less than or equal to 10 particles/ft in
the overall size range of 0.3-5.0 pm and in sub-ranges of
0.3-0.6, 0.6-1.5, 1.5-3.0, and 3.0-5.0 pm. A 1-100:1 porous
tube diluter (see Figure 3) was constructed and installed
so that the smaller size ranges could be used without going
off-scale for the instrument. The 0-1 milliamp output for the
various size sub-ranges was then attached to a LSE Model
M24 4-channel chart recorder for Phases III and IV testing.
A schematic of the major features of the experimental bag-
house apparatus is shown in Figure 4.
A separate, non-shaking, two bag baghouse was designed
and constructed of 1.9 cm (3/4 in.) plywood (see Figure 5)
for the Phase IV bag series system. This second baghouse
was installed in the bag series configuration using collap-
sible ^10.2 cm (4 in.) D ducting for the connections and a
section of ^10.2 cm (4 in.) D stove pipe for a sampling
section. The operating conditions chosen for the first,
shakeable baghouse were the optimum conditions from Phases I,
II, and III. Two new bags were installed in the first
36
-------
LO
—I L
-
<®l
r
?'!"
i-i
.1
i
®
- —
ft
Rotameters
0.1-1.0 scfh
1.0-10.0 scfh
10.0-100.0
Manometer: -0.1-0.0-7
£- Bleed Valve
H20
--
Pump /%
^ in
" ' ' " """ ""Kji1
_~ir
—
• *"— f*—
- , i
1 1
r '
•
•"•-"--j— - *
"1 "•*
Sample Inlet
_ ^ _^
r
i
i
i
u,
]
(
f
\ '
i '
j i
j
i
Copper
s
Absolute
Oil Filter
Tube
^/Porous
:::::
to
Royco
Figure 3. Porous tube diluter
Scale: 1/4" = I1
-------
Baghouse•
Pressure Chart Recorder
Pressure Transducer
00
Scole* I"-I'
9'21/4"
-Outlet Duct (4"D. Stove Pipe)
__/yV-^^-Sampling Port 11/2x3
IT
Dry Bulb
Thermometer
Double
Flange For
Cyclone
Insertion
Mfl
OOoo
Exhaust Fan
Manometer
For Inlet
Velocity
To Exhaust Hood
Timer
'Assembly
Port For
Pitot Tube
a Sampling
Inlet Duct
(2"D. P.V.C.)
From Dust Feeder
8 Steam Inlet
-Sampling Metering Box
3
.from Dust Feeder
/"in
Exhaust Hood
To 220V
Outlet
Figure 4. Baghouse testing apparatus
-------
FROM PRIMARY
BAGHOUSE
I i \
( I. / \
/ . >' 1
C. j_~_...""'-^~ ""
TO EXHAUST
FAN
-------
baghouse, while those previously stabilized in Phase III
(minus the dust shaken loose during moving and installation)
were installed in the second (non-shakeable) baghouse. Pres-
sure taps connected to a U-tube manometer were installed in
the second baghouse to measure pressure drop across the
fabric.
40
-------
SECTION 6
EXPERIMENTAL PROCEDURE
SAMPLE ANALYSIS TECHNIQUES
Optical Microscope Counts
The optical microscope slides were prepared and counted
by the procedures of the Joint AIHA-AGCIH Aerosol Hazards
Evaluation Committee17. Slides were prepared using the
recommended counting medium on a one-to-one by volume solu-
tion of dimethyl phthalate and diethyl oxylate with 50 mg
of membrane filter material added per milliliter of solu-
tion. Fibers greater than or equal to 1.5 ym in length
were counted and sized using phase contrast at 400-500X
magnification.
Full field counts were continued until at least 100
fibers were viewed. The minimum number of fields viewed
was 20, unless at least 1,000 fibers greater than 1.5 ym
and 100 fibers greater than 5.0 ym or 6.0 ym were viewed in
fewer fields. The data reported are in terms of fibers of
length 21 1-5 1-™ or >^ 5.0 or 6.0 ym; however, the larger
numbers of fibers in the smaller size ranges dominate the
concentration values for the smaller ranges. The exact
counting magnification and technique varied somewhat from
one experimental phase to the next so that the results of
several phases are not exactly comparable. The highest
repeatability and thus reliability of results was achieved
in the last two phases. A minimum of duplicate counts of
the same filter were made in Phases II to IV.
41
-------
Electron Microscope Counts
The Phase I electron microscope grids of samples were
prepared and counted in the following manner. The electron
microscope grids were prepared by punching out a 3.05 mm
diameter portion from near the center of the sample and
placing it on a carbon coated 200 mesh grid. The grid with
the sample was then placed in a slow acetone wash to dissolve
the membrane filter. Six grids from three samples were
processed simultaneously. Fibers were then sized and counted
using the electron microscope at a nominal (photograph)
magnification of 33,OOOX. This is equivalent to a magnifi-
cation of 26,400X at the viewing screen. The fibers greater
than 0.06 ym on each grid were counted and sized until 100
fibers were viewed in at least 20 fields or until 100 fields
were viewed (unless there were fewer than 100 viewable
fields).
Calculation of Outlet Concentrations
The number concentrations of fibers per cubic meter
for each size range (>_ 0.6 ym, ^.1.5 ym, and ^5.0 or 6.0 ym)
were calculated by the following equation
no. of fibers _ no. of fibers counted
.
m of air no. of fields viewed
. effective filter area , 1
area of each field viewed sampled volume
The effective filter area for the filter holder used is
7.39 x 10"3 m2 (11.46 in.2); the field area was 5.72 x
10~8 m2 (8.86 x 10"5 in.2) for Phases I and II; 10.18 x
_o o A 9
10 m (1.58 x 10'^ in. ) for Phases III and IV for the
optical microscope; and 10.57 x 10~12 m2 (1.64 x 10~8 in.2)
for Phase I for the electron microscope.
42
-------
PHASE I TESTING
The Phase I experimental plan tested the operating
parameters of fabric, relative humidity, and waste type at
a constant air-to-cloth ratio, dust loading, and shaking
cycle. The air-to-cloth ratio and dust loading chosen were
the median values of 0.76 m3/min/m2 (2.5 cfm/ft2) and
21 g/m (9.2 gr/ft ), respectively. After discussions with
the EPA16, the maximum dust cake retention values of the
shaking variables of amplitude (0.875 cm [0.344 in.]),
frequency (1 cps) , and duration (20 sec) were used. These
values were used in the hope of maximizing collection effi-
ciency by maximizing dust cake. The 16 minute shake cycle
was chosen to maximize the number of shake cycles in a given
time period. A one minute settling period was used before
and after each shake.
Stabilization of two bags each of three fabrics was
conducted simultaneously at ambient humidity. The bags
were stabilized for a minimum of eight hours or until the
trace of pressure drop across the fabric was consistently
repeated. Each set of two bags was then restabilized under
the same conditions for four cycles to replace the filter
cake that was loosened or partially removed during the
moving of the bags for storage and reinstallation. Sampling
of the emissions from the baghouse was then conducted for an
additional four cycles. This restabilization and sampling
process was then conducted for the 40% and 80% relative
humidity levels in the order of increasing humidity. The
entire series for the six fabrics at ambient, 40%, and 80%
relative humidity was done for the asbestos cement waste
using the IITRI screw feeder. Analytical results are pre-
sented in Table 9. Ambient humidity levels were sometimes
above the desired 20% on any given day and were sometimes
as high as 40%. At the 80% humidity level, saturation of
the fabric occurs as some of the moisture condenses on the
43
-------
Table 9. PHASE I RESULTS FOR ASBESTOS CEMENT WASTE*
Test
Sample
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Filter
Fabric
Cotton Sateen
96 x 60.
9.7 oz/yd2
Napped Cotton
98 x 60 „
8.7 oz/yd2
Nomex
95 x 60 .
5.2 oz/yd2
Dacron
75 x 71 0
5.8 oz.ydz
Cotton Twill
73 x 60 9
7.4 oz/ydz
Dacron
64 x 51
8.5 oz/yd2
Humidity
Range
(% R.H.)
20-30
15-25
35-45
75-85**
30-40
35-45
65-75
75_85**
30-40
35-45
75_85**
25-35
35-45
75-85**
20-40
35-45
75_85**
30-45
35-45
75-85**
Sampled
Volume
(ft3)
13.92
11.31
ioroo
10.09
12.87
8.02
7.72
4.43
10.46
4.64
7.04
5.64
2.63
14.43
6.46
11.90
7.96
6.20
5.00
6.52
Outlet Concentration
(No. of fiber s/m3)
> 0.06 vim
2.15 x lOa
5.57 x lOo
13.62 x 10ft
3.49 x 10a
0.37 x 10ft
13.88 x lOo
3.99 x 10ft
4.45 x 10°
1.35 x 10ft
1.06 x lOo
7.51 x 10°
9.77 x 10ft
31.67 x 10ft
4.34 x 10°
28.64 x lOo
2.75 x 10ft
9.36 x 10°
28.36 x lOn
10.75 x 10ft
13.05 x 10°
> 1.50 urn
1.18 x lof
1.12 x 10?
2.28 x 10?
3.31 x 10°
5.40 x 10?
9.06 x 10?
8.14 x 10?
14.35 x 10°
8.71 x 10?
19.43 x 10?
15.56 x 10°
20.60 x 10ft
10.83 x 10?
3.02 x 10°
4.83 x 10ft
3.24 x 10?
4.67 x 10°
5.95 x 10?
53.76 x 10?
3.77 x 10°
> 6.00 ym
0.95 x lo|
0.97 x 10?
1.74 x 10?
1.72 x 10°
2.70 x ID?
3.10 x 10?
3.28 x 10?
6.27 x 10°
3.02 x ID?
5.10 x 10?
6.04 x 10°
4.36 x 10ft
3.12 x 10?
1.09 x 10°
1.48 x 10ft
0.99 x 10?
1.29 x 10°
1.47 x ID?
3.28 x 10?
0.73 x 10°
9 3
* All tests at air:cloth ratio of ^2.5:1 (cfm/ft ), dust loading ^22 g/m .
** At high humidity, plugging of inlet at dust feeder inlet occurs due to super-
saturation..
-------
clean side of the baghouse. Under these conditions, the
pressure drop builds to a value much higher than for the
lower humidity levels, due to either blockage of the fabric
or increased dust cake. The dust cake then falls when it
becomes too saturated to be supported. This causes an
abrupt decrease in filtering efficiency.
Due to the physical and flow characteristics of raw
asbestos fibrous waste, the manually fed vibrating trough
feeder had to be used for the second part of Phase I.
Therefore, only two fabrics were tested with the fibrous
asbestos waste, cotton sateen, and Dacron #1, manufactured
of filament by spun fibers. Each set of two bags was
stabilized separately for a minimum of eight hours at
ambient humidity and then was restabilized and tested as
before at each humidity level in increasing order. When
the bags were removed, it was found that the dust cake
bridged across the entire bag. The results of the electron
and optical microscope analyses are given in Table 10.
Because of feeding difficulties and the fact that raw
fibrous asbestos is used in fewer industrial applications,
raw fibrous waste was not studied in the following phases.
It can be seen from Tables 9 and 10 that cotton sateen
and cotton twill fabrics are the most efficient of the
natural, low temperature fabrics. Because of its economy,
availability, performance, and wide usage, cotton sateen
was chosen for further study. For high temperature appli-
cations, Nomex was to be studied in Phase II because of its
performance relative to the other synthetic, high tempera-
ture fabrics -- especially in the electron microscope size
range.
Obvious trends between relative humidity, fabric type,
and humidity are nearly indeterminable from the data. Thus,
the accuracy of sampling and analysis and of the values of
45
-------
Table 10. PHASE I RESULTS FOR FIBROUS ASBESTOS WASTE*
Test
Sample
Number
1
2
3
4
5
6
Filter
Fabric
Cotton Sateen
96 x 60 „
9.7 oz/yd2
Dacron (fxs)
75 x 71 .
5.8 oz/yd2
Humidity
Range
(% R.H.)
20-30
35-45
40-80**
25-35
35-45
75-80**
Sampled
Volume
(ft3)
7.61
7.32
6.40
7.19
7.28
6.16
Outlet Concentration
(No. of fiber s/mj)
>. 0.06 ym
4.86 x 10o
5.06 x 10«
1.50 x 10°
7.00 x 10o
4.24 x IDs
5.27 x 10°
>,1.50 ym
3.68 x 10?
4.14 x 10?
4.13 x 10°
3.68 x 10*?
3.10 x 10?
5.62 x 10b
> 6.00 ym
1.05 x 10£
1.06 x 10?
1.14 x 10°
0.89 x 105
0.77 x 10?
1.26 x 10°
* All tests at air:cloth ratio of ^2.5:1 (cfm/ft2) and dust loading ^22 g/m .
** At high humidity ranges, blockage of inlet occurs (not as severe as with
cement waste).
-------
the experimental variables was questioned. As counting
error appeared to be the largest source of error due to
inherent problems in the procedure, it was decided to
attempt to refine the optical microscope technique and to
suspend the electron microscope analysis.
PHASE II TESTING
Phase II tested the effects of relative humidity, air-
to-cloth ratio, and dust loading for the two fabrics of
cotton sateen and Nomex, chosen for low and high temperature
applications from Phase I. Relative humidity was kept at
levels of 30, 40, and 60% with the usual accuracy of + 5%.
The air-to-cloth ratio was to be fixed at the values of
0.46, 0.76, and 1.22 m3/min/m2 (1.5, 2.5, and 4.0 cfm/ft2);
however, as pressure drop across the fabric rose to over
12.7 cm H90 (5 in. H90), the air-to-cloth ratio of
32 ?
1.22 m /min/m (4.0 cfm/ft ) could not be maintained. Dust
loadings were 10, 21, or 45 g/m3 (4.4, 9.2, or 19.7 gr/ft3)
for the desired air-to-cloth ratios. A prestabilized set
of bags from Phase I was used with restabilization for
four 16 minute cycles before each test. A one-half fractional
experimental design of 27 combinations for cotton sateen and
Nomex for cement waste was planned. The tests for both
fabrics were to be conducted in random order of the combina-
tions to be tested; however, the cotton sateen series was
performed first with internal random order.
The Phase II optical microscope analysis was begun
immediately after the completion of testing for the cotton
sateen series. This was done to insure that any inaccuracies
in sample counting encountered in Phase I would not be
continued into the Nomex series. The data and results for
the cotton sateen series are presented in Table 11. The
initial series of optical counts resulted in a high varia-
bility (+ 50%) in the data for the replicate tests
47
-------
Table 11. DATA AND RESULTS FOR COTTON SATEEN, PHASE II
Sample
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Relative
Humidity
(%)
55-65
35-45
35-45
35-45
55-65
25-35
25-35
35-45
35-45
25-35
55-65
35-45
55-65
55-65
55-65
25-36
25-35
25-35
35-45
Air: Cloth
Ratio
(cfm/ft2)
1.5:1
1.5:1
1.5:1
1.5:1
2.1:1
1.5:1
3.92:12
3.92:12
3.23:12
2.5:1
2.95:12
2.5:1
1.5:1
2.95:12
2.5:1
2.5:1
2.50:12
2.11:12
2.23:13
Dust
Loading
(K/m3)
45
21
21
10
45
45
10
21
55. 7*
21
28. 55
45
10
13. 66
21
10
72. 0^
85.3"
11. 26
Sampled
Volume
(ft3)
4.768
4.278
4.284
3.937
3.537
6.298
8.927
9.590
8.155
5.552
9.937
6.333
4.685
7.100
6.835
5.329
9.483
8.968
5.145
Number of Fibers in
20 fields1^ 400X
> 1.5 urn £. 6.0 urn
#1
350
508
365
232
360
391
427
307
466
178
550
481
482
470
638
555
582
844
1372
#?
375
362
463
509
543
628
524
755
577
767
964
932
1029
1073
1358
1027
1418
1374
943
#1
74
80
44
59
83
67
57
41
22
31
73
74
77
52
82
102
61
83
114
£2
98
40
68
67
107
105
76
64
95
118
84
78
70
67
67
58
76
81
88
Number of Fiber s/cc
> 1 . 5 urn > 6 . 0 urn
#1
16.722
27.051
19.409
13.424
23.186
14.143
10.896
7.292
13.017
7.303
12.608
17.302
23.436
15.080
21.264
23.725
13.981
21.439
60.747
n
17.916
19.276
24.620
29.451
34.972
22.715
13.372
17.934
16.118
31.470
22.099
33.521
50.033
34.427
45.260
43.901
34.063
34.902
41.752
7^1
3.536
4.260
2.340
3.414
5.346
2.423
1.454
1.211
0.614
1.272
1.674
2.662
3.744
1.668
2.733
4.360
1.465
2.108
5.048
n
4.682
2.130
3.616
3.877
6.891
3.798
1.939
1.520
2.654
4.842
3.021
2.806
3.404
2.150
2.233
2.479
1.826
2.058
3.896
00
area is equal to 5.723
x lO-3 m2 (11.46 in.2).
x 10"8 m2 (8.86 x 10"5 in.2);
Field
7.386
Design of air:cloth ratio = 4.0:1 cfm/ft2.
Design air:cloth ratio = 2.5:1 cfm/ft .
Design dust loading = 45 g/m3 (19.7 gr/ft3).
Design dust loading = 21 g/m3 (9.2 gr/ft3).
Design dust loading = 10 g/m (4.4 gr ft ).
effective filter area is
-------
(i.e., 2 and 3). Therefore, it was decided to perform dupli-
cate optical counts of the Phase II samples. Also, electron
microscope analysis of samples was suspended until the error
could be assigned to sampling or analysis. Both sets of
data are given in Table 11. The duplicate samples counted
were from the same filter sample, but from different wedge
shaped segments of the filter. It can be seen that the
counting error is quite large, i.e., at least + 50% in both
size ranges and often even greater in the 1.5 ym size range.
Some trends are discernible from the results for cotton
sateen bags:
1. The outlet concentration generally decreases with
increasing air-to-cloth ratio.
2. No trends of effects on outlet concentration by
relative humidity or dust loading are readily
apparent.
It should be noted that the OSHA standard of 2 fibers/cc
appears to be exceeded in several cases. Due to the high
variability in the results for the cotton sateen series,
the Nomex series of tests was not performed.
As the concentration data by membrane filter from
Phase I samples were suspect, both a Condensation Nuclei
Counter (CNC) and a Royco Model 245 light scattering monitor
with Module 510 display were tested as outlet concentration
sampling methods. The Royco, which has a total range of
0-100,000 particles/ft3 in sizes 0.3-5.0 ym, was operative
the last few tests of the cotton sateen series and showed
that definite increases in emissions occurred immediately
after cleaning.
PHASE III TESTING
Phase III tests of the mechanical shaking variables
investigated the effects of the high dust caking conditions
(120 min, 20 sec) and the lowest dust caking conditions
49
-------
(16 min, 80 sec) of shaking interval and duration. Extreme
values of the shaking variables of amplitude (0.875 and
3.500 cm [0.344 and 1.378 in.]) and frequency (1 and 5 cps)
were tested at each dust caking condition. All stabilization
and testing was conducted at an air-to-cloth ratio of
19 9 3
0.92 nT/min/nr (3.0:1 cfm/ftr), dust loading of 21 g/m
(9.2 gr/ft3), and ambient humidity (40-90%). A new pair of
cotton sateen bags were run until the pressure drop and
Royco particle count indicated that the cake build-up and
emissions had stabilized for each set of conditions. Bag-
house operation was 24 hours per day, five days per week.
Optical microscope analysis results are presented in
Table 12. Duplicate counts were performed in all cases,
and in those in which the variability was high, a third
count was made. In most cases, the duplicates were in
reasonable agreement. Several changes in sampling and
counting technique may have resulted in this improvement
in analytical reproducibility of results over those of the
previous phases. They are the use of a more accurate dry
gas meter, denser fiber loading, counting experience, change
in magnification from 400X to 500X, change in cut size from
6.0 ym to 5.0 jam, and use of a tally counter.
Some trends can be observed from the data:
1. The outlet concentrations for the 120 min cycle
with 20 sec shake are all lower than those for
16 min cycle with 80 sec shake.
2. The high level of shaking amplitude produces lower
outlet concentrations in both cases.
3. The high level of either shaking amplitude or
frequency reduces the pressure drop across the
fabric.
The Royco traces of total particle concentrations sub-
stantiate these observations. The peak in outlet concen-
tration occurs during the shaking and settling period when
50
-------
Table 12. PHASE III FIBER COUNTS*
Sample
Number
1
2
3
4
5
6
7
8
Shake Cycle
Interval
(min)
120
16
16
16
16
120
120
120
Duration
(sec)
20
80
80
80
80
20
20
20
Amplitude
(cm)
0.875
3.500
0.875
3.500
0.875
0.875
3.500
3.500
Frequency
(cps)
1
5
5
1
1
5
1
5
Sampled
Volume**
(ft3)
286.6
169.0
149. 6
170.6
143. 8
344.0
362.2
362.8
No. of Fibers > 1.5 Um/cm
Count No.
1
1.1259
5.0280
17.1133
4.6736
8.1092
3.2037
1.4636
2.4165
2
3.5993
5.5615
10.0717
6.6152
7 . 6284
3.0821
2.2320
2.9695
3
3.0918
-
7.7378
-
-
-
-
-
No. of Fibers > 5.0 Um/cm
Count No.
1
0.3181
1.9670
8.7649
1.7063
2.3924
0.7395
0.2758
0.3365
2
1.3189
1.8882
1.9002
1.3160
1.4663
0.5956
0.3370
0.3247
3
0.5064
-
1.2269
- -
-
-
-
-
Stable
Ap
(in. H20)
7.5
0.6
2.0
1,5
2.8
1.3
1.3
1.0
2 3
* All runs at 3.0:1 cfm/ft air-to-cloth ratio, 21 g/m dust loading, and ambient humidity (40-90%).
** 120 min. cycles sampled for 1-2 hr cycle, 16 min cycles sampled for 4-16 min cycles.
-------
the concentration is high but the flow is low and when the
air flow starts after shaking. During the remainder of the
cycle, the outlet concentration remains at a low, stable
level. Thus, the two hour cycle has a lower total emission
than an equal time period of successive 16 min cycles. Also,
the peak in outlet concentration when the blower starts
after a 3.500 cm (1.378 in.) shake is visibly lower than
after a 0.875 cm (0.344 in.) shake for particles ^0.3 ym.
Examples of Royco traces for several shaking conditions for
particles >_ 1.5 ym are given in Figure 6.
PHASE IV TESTING
Phase IV tested the feasibility and effectiveness of a
bag series system, A separate non-shaking, two bag baghouse
was installed as the second baghouse in the bag series
configuration. The operating conditions chosen for the
first shakeable baghouse were the optimum shaking conditions
of 120 min cycle, 20 sec, 3.500 cm (1.378 in.), and 1 cps
shake from Phase III. Two new bags were installed in the
first baghouse, while those previously stabilized in
Phase III for the same shaking conditions (minus the dust
shaken loose during moving and installation) were installed
in the second (non-shakeable) baghouse.
The results of the optical analyses of samples completed
after the completion of 24, 70, and 164 hours of operation
are given in Table 13. These results show that the outlet
concentrations for the bag series system are slightly
higher than those for the most efficient single bag systems
in Phase III. The mechanism that could be causing this
slight increase in outlet concentration which is of marginal
statistical significance is that the filtered air from the
first bag is freeing dust from the cake of the second bag
due to air velocity while not rebuilding the cake by dust
loading. Thus, it would appear that two baghouses in series
52
-------
A|
0.6 -
0.0
1.0 -.
§0.8-
A|
0.6-
0.2
10
20 30
Time from Beginning of Cycle (min)
(a)
Time from Beginning of Cycle (min)
(c)
40
100
120
Oft f) ,
u .u
0
*\j 1.0 _
14-1
^ 0.8 .
7|0.6-
S 0.4.
u
4J
S n 7
fl] U . / '
m
° n n .
(
10 20 30 40 100 120
Time from Beginning of Cycle (min)
(b)
\
\
' \- ______ /
} 10 20 30 40 100 120
Figure 6. Phase III Royco Particle Counter traces: (a) 120 min cycle, 20 sec, 0.875 cm, 5 cps shake;
(b) 120 min cycle, 20 sec, 3.500 cm, 5 cps shake; (c) 120 min cycle, 20 sec, 3.500 cm, 1 cps shake
-------
Table 13. PHASE IV FIBER COUNTS*
Sample
Number
1
2
3
Operating
Hours to
End of
Sampling
24
70
164
Sample
Volume**
(ft3)
328.1
248.5
412.6
No. of Fibers
£.1.5 ym/cm3
Count No.
1
1.8702
4.2941
3.5959
2
3.0754
4.8472
3.9890
No. of Fibers
i. 5 . 0 ym/cm3
Count No.
1
0.3169
0.8554
0.5193
2
0.3278
0.7317
0.5628
Stable Ap
Across Bag #1
(in. H20)
1.50
2.20
2.65
Stable Ap
Across Bag #2
(in. H20)
0.55
0.55
0.47
t_n
All runs at 0.92 m3/min/m2 (3.0:1 cfm/ft2) air-to-cloth ratio, 21 g/m3
(9.2 gr/ft3) dust loading, ambient relative humidity (60-90%), and shaking
variables of 120 min, 20 sec, 3.500 cm (1.378 in.), and 1 cps.
** All samples taken for 1-2 hr cycle.
-------
are not a viable approach to reducing emissions. However,
it was observed that during the initial hours of stabili-
zation, the Royco trace for particles greater than 0.3 ym
at 100:1 dilution on the ^3,530,000 par tides/m3
o
(100,000 particles/ft ) scale remained on scale. This was
in contrast to all Phase III experiments with a single bag-
house and indicated that a steady state low outlet concen-
tration was being emitted from the second bag at the
beginning of the bag series run. The peak observed when
air flow was resumed after cleaning was very low in the
case of the initial stabilization of the bag series. There-
fore, the bag series system may prove valuable for reducing
high emissions during stabilization of a section of new bags
by recycling flow through a previously stabilized section
of the baghouse.
55
-------
SECTION 7
DISCUSSION OF RESULTS OF TESTING BASED ON
STATISTICAL ANALYSIS OF DATA
INTRODUCTION
An experiment employing four phases was conducted to
investigate the effects of nine factors on the exit side
concentration of asbestos fibers emitted from a baghouse.
These factors are:
Type of waste
Humidity
Bag fabric
Air-to-cloth ratio
Dust loading
Bag-shaking amplitude
Bag-shaking frequency
Joint effect of duration and period of bag-shaking
The use of one bag or two bags in a series
Measurements of exit concentrations were taken after a
24-hour stabilization period. The type of waste, bag fabric,
amplitude, joint effect of duration and period of bag-shaking,
and number of bags were shown to significantly affect exit
concentration.
The discussion of the statistical analyses and their
results is divided into six major sections. The first
56
-------
section is a brief presentation of the statistical methods
used in analyzing the data. The next four sections discuss
each phase of the experiment and its results. These sections
first discuss the experimental design; then the data are
presented followed by a discussion of the results of the
statistical analyses. The last part of each section summar-
izes the findings for the particular phase. The sixth and
final section summarizes the findings for all four phases
and recommends efficient baghouse operation alternatives.
METHODS OF ANALYSIS
Two of the nine factors considered in this study are
qualitative in nature: asbestos and bag cloth. The quali-
tative distinction among the levels of these two factors
can be represented numerically by five new variables. The
values of these new variables and their corresponding factor
levels are given in Tables 14 and 15. The general rule is
that if a factor has N mutually exclusive states, then N-l
appropriately coded variables with associated coefficients
can represent any possible pattern of differences among.the
states with respect to a quantitative property such as mean
concentration. These five new variables were used to investi-
gate the effects of various bag fabrics and asbestos waste.
Analysis of the quantitative factors employed both
their actual values and their common logarithms. The common
logarithm is often applied to variables which are inherently
positive and which have a large range. They provide an
opportunity for more complex relationships between outlet
concentration and the factors to be investigated.
The values of the outlet concentrations of asbestos
fibers were also transformed to their common logarithms,
because they too are inherently positive. More importantly,
the distribution of these concentrations was log-normal.
By taking the log of the concentrations, a well-developed
57
-------
Table 14. NUMERIC CODING OF WASTE TYPE
Waste Type
Asbestos Cement
Asbestos Fibers
Coded Variable
Zl
+1
-1
Table 15. NUMERIC CODING OF BAG TYPE
Bag Type
Cotton Sateen
Napped Cotton
Nomex
Dacron No. 1
Cotton Twill
Dacron No . 2
Coded Variables
Z31
+1
0
0
0
0
0
Z32
0
+1
0
0
0
0
Z33
0
0
+1
0
0
0
Z34
0
0
0
+1
0
0
Z35
0
0
0
0
+1
0
58
-------
field of statistical theory based on normal variates could
be applied.
The size ranges reported in the data are greater than
0.06 ym in Phase I, greater than 1.5 ym in all phases,
greater than 6.0 ym in Phases I and II, and greater than
5.0 ym in Phases III and IV. Because of the much greater
numbers of fibers in the smaller size ranges of the outlet
concentration, the smaller fibers dominate the concentration
values in each size range.
The effects of the factors on outlet concentration were
analyzed three ways:
1. Correlation coefficients were computed between
outlet concentration or its log transform, and
the factors or their log transforms.
2. 90% confidence intervals were constructed about
the geometric* means of outlet concentration for
various factor values.
3. Relations between the log of fiber concentration
and the factors were investigated using stepwise
linear regression.
The specific methods of analysis used in each phase
are stated in their respective sections. Appendix A
presents a more detailed discussion of these statistical
methods.
DISCUSSION OF PHASE I RESULTS
The main purpose of Phase I was to determine the effects
of different bag fabrics and asbestos waste on the outlet
concentration of asbestos fibers. Both the asbestos cement
waste and raw asbestos waste were obtained from Johns-Manville,
* Geometric rather than arithmetic means were computed since
the logs of concentration were more normally distributed
than the observed concentrations. This further enhanced
the appropriateness of the t-distribution s application
for computing confidence intervals.
59
-------
Waukegan, Illinois, from their baghouse control equipment
of the Transite asbestos cement pipe process. All combina-
tions of humidity and bag fabric were used with asbestos
cement waste. However, the unsuitability of the SCR-20 dust
feeder to feed raw asbestos waste necessitated manual feeding.
This consumed more time than anticipated, thus limiting the
fabrics exposed to raw asbestos waste to cotton sateen and
Dacron No. 1. This limitation required the analysis of
Phase I data to be carried out on two separate subsamples.
Subsample 1 Results
The factors or independent variables investigated
using the first subsample of data are presented in Table 16
along with their respective values or levels. All six bag
fabrics, but only one type of asbestos waste (cement waste),
were included.
The data base generated by Phase I testing is given in
Table 17. In this phase of testing, the outlet concen-
trations of fibers greater than 6.0 ym, 1.5 ym, and 0.06 ym
were the dependent variables.
Only the first twenty tests in Table 17 were included
in Subsample 1. These tests used asbestos cement waste only.
Correlations Between Subsample 1 Variables -
Correlation coefficients, r, for Subsample 1 are given
in Table 18. The range within which the associated probab-
ility, P, falls is indicated in conjunction with each
correlation. P is the probability that a correlation as
large in magnitude as r would occur due only to sampling
error. If P is greater than 0.10, the correlation is not
considered to be statistically significant.
The correlation analysis of Subsample 1 indicates that
cotton sateen is the best fabric for reducing outlet con-
centration of asbestos fibers greater than 1.5 ym. For
60
-------
Table 16. THE INDEPENDENT VARIABLES AND THEIR DESIRED
LEVELS FOR SUBSAMPLE 1 OF PHASE I
Variable
X-,, Waste Type
X£, Humidity
X3, Bag Type
X^, Air-to-Cloth Ratio
Xc, Dust Loading
Xg, Amplitude of Shake Cycle
Xy, Frequency of Shake Cycle
Xo, Period and Duration of
0 Shake Cycle
Xg, Number of Bags
Level
1.
1.
2.
3.
1.
2.
3.
4.
5.
6.
1.
1.
1.
1.
1.
1.
Asbestos Cement
20%
40%
80%
Cotton Sateen
Napped Cotton
Nomex
Dacron No. 1
Cotton Twill
Dacron No . 2
0.76 m3/min/m2 (2.5 cfm/ft2)
22 g/m3 (9.6 gr/ft3)
0.875 cm (0.344 in.)
1.0 cps
Period - 16 min
Duration - 20 sec
1
61
-------
Table 17. DATA BASE FOR PHASE I
Test
No. .
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Comb.
No.
1
1
2
3
4
4
5
6
7
7
8
9
10
11
12
13
14
15
15
16
17
18
19
20
21
22
Waste Type
Zl
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
-1
-1
-1
-1
-1
-1
Humidity
x2
25
20
40
80
35
40
70
80
35
40
80
30
40
80
30
40
80
37
40
80
25
40
60
30
40
77
Codec;
Z31
1
1
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
1
0
0
0
Z32
0
0
0
0
1
1
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Bag Variables*
Z33
0
0
0
0
0
0
0
0
1
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Z34
0
0
0
0
0
0
0
0
0
0
0
1
1
1
0
0
0
0
0
0
0
0
0
1
1
1
Z35
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
1
0
0
0
0
0
0
0
0
0
Outlet Concentration
(No. of fibers/cm3)
>. 0.06 urn
215
557
1362
349
37
1388
399
445
135
106
751
977
3167
434
2864
275
936
2836
1075
1305
486
506
150
700
424
527
2. 1 • 5 ym
1.18
1.12
2.28
3.31
5.40
9.06
8.14
14.35
8.71
19.43
15.56
20.60
10.83
3.02
4.83
3.24
4.67
5.95
53.76
3.77
3.68
4.14
4.13
3.68
3.10
5.62
>. 6 . 0 ym
0.95
0.97
1.74
1.72
2.70
3.10
3.28
6.27
3.02
5.10
6.04
4.36
3.12
1.09
1.48
0.99
1.29
1.47
3.28
0.73
1.05
1.06
1.14
0.89
0.77
1.26
ON
fO
*
Variable coding given in Tables 14 and 15.
-------
Table IS. CORRELATIONS BETWEEN PHASE I VARIABLES
FOR SUBSAMPLE 1 (N = 20)
Concentration of
Fibers >.
0.06 ym
.
1 . 50 ym
6.00 ym
L°£lO Concentration
of Fibers >
0.06 ym
1.50 ym
6.00 ym
Independent Variable
Z31 First Bag Var.
Z32 Second Bag Var.
Z33 Third Bag Var.
Z^ Fourth Bag Var.
Z35 Fifth Bag Var.
X2 Humidity
Z31 First Bag Var.
Z32 Second Bag Var.
Z33 Third Bag Var.
2^34 Fourth Bag Var.
Z35 Fifth Bag Var.
X2 Humidity
Z31 First Bag Var.
Z32 Second Bag Var.
Z33 Third Bag Var.
Z3, Fourth Bag Var.
Z35 Fifth Bag Var.
X2 Humidity
Z31 First Bag Var.
Z32 Second Bag Var.
Z33 Third Bag Var.
Z.,^ Fourth Bag Var.
Z35 Fifth Bag Var.
Z2 Log]n Humidity
Z31 First Bag Var.
Z32 Second Bag Var.
Z33 Third Bag Var.
Z^ Fourth Bag Var.
Z35 Fifth Bag Var.
Z2 Log-iQ Humidity
Zoi First Bag Var.
*J J-
Z32 Second Bag Var.
Z33 Third Bag Var.
ZOA Fourth Bag Var.
Z35 Fifth Bag Var.
Z2 Log1Q Humidity
Correlation
r
-0.194
-0.223
-0.294
+0.247
+0.171
-0.198
-0.346
-0.031
+0.168
+0.055
-0.208
-0.073
-0.386
+0.360
+0.525
+0.056
-0.348
+0.175
-0.079
-0.281
-0.361
+0.235
+0.161
+0 . 044
-0.664
+0.172
+0 . 349
+0.148
-0.179
+0.186
-0.393
+0.403
+0.483
+0.087
-0.357
+0.135
Probability
P
*
*
*
*
*
Vv
*
*
*
*
*
*
0.05 < P < 0.1
*
0.01 < P < 0.02
*
*
*
*
*
*
*
*
*
0.001 < P < 0.005
*
*
*
*
*
0.05 < P < 0.10
0.05 < P < 0.10
0.02 < P < 0.05
*
*
*
* Not significant at 0.10 probability level.
63
-------
outlet concentrations of fibers greater than 0.06 ym, no
bag fabric had any significant correlation. Relative humid-
ity did not significantly correlate with outlet concentration
or its log regardless of fiber lengths.
Geometric Means of Outlet Concentration for Subsample 1 -
Table 19 gives the geometric mean and 90% confidence
limits for the data of Subsample 1. The results are given
separately for each of the three fiber sizes. The geometric
means and 90% confidence limits of outlet concentrations
are plotted in Figures 7, 8, and 9.
A confidence interval which overlaps the mean of another
indicates that no significant difference exists between the
two respective bag fabrics; the significance level being
0.10. The reader, however, is cautioned against making
multiple pairwise comparisons and drawing conclusions about
the significance of differences among all bag fabrics, since
the probability of making correct inferences among all pair-
wise comparisons is much lower than the 0.10 significance
level for making one pairwise comparison. The plots of the
geometric means and their confidence intervals for the six
bag fabrics are presented only to provide the reader with
a visual representation of the experimental results and
magnitude of experimental error. The regression analysis
provides a statistically sound basis for determining which
bag fabrics are significantly different, and these results
are given in the next section.
Figure 7 shows that, while Nomex, napped cotton, and
cotton sateen had the lowest outlet concentrations, the
confidence intervals about the geometric means overlap,
indicating that the means are not significantly different
from one another. Figure 8 shows that cotton sateen had
the lowest geometric mean outlet concentration for fibers
greater than 1.5 ym of any fabric. Dacron No. 1 and cotton
64
-------
Table 19. GEOMETRIC MEANS AND 90% CONFIDENCE LIMITS
FOR PHASE I - SUBSAMPLE 1
Type of Bag
Fabric
Cotton Sateen
Napped Cotton
Nomex
Dacron No. 1
Cotton Twill
Dacron No. 2
N
4
4
3
3
3
3
Outlet Concentration
(Number of fibers/cm3)
Fibers >. 0.06 vim
Geometric
Mean
489
309
221
1104
904
1585
Confidence Limits
Lower
211
61
52
284
184
789
Upper
1132
1567
942
4295
4436
3184
Fibers > 1.5 ym
Geometric
Mean
1.78
8.69
13.80
8.77
4.18
10.64
Confidence Limits
Lower
1.02
5.68
7.85
2.33
3.09
1.55
Upper
3.11
13.30
24.27
33.04
5.64
73.28
Fibers >. 6.0 )am
Geometric
Mean
1.29
3.62
4.53
2.45
1.24
1.52
Confidence Limits
Lower
0.90
2.43
2.77
0.92
0.94
0.55
Upper
1.85
5.41
7.40
6.56
1.63
4.23
Ui
-------
c CD
O 4-)
•H
4-J
}-i
4-1
0)
O
O
o
10
9-.
QJ
J3
M
0)
•rH
i-l Q)
^ •§ •
d S Q
0 S io8
o
6
O
O
Cotton
Sateen
Napped
Cotton
Nomex
Dacron
Cotton
Twill
Dacron #2
Figure 7. Estimates of the geometric mean and their 90% confidence intervals
for outlet concentration of asbestos fibers greater than 0.06 utn by
type of bag - Phase I - Subsample 1
-------
C
o
•H
ctf
4->
C
QJ
O
C
o
4-J
0)
r-l
4J
O
3.2xl0
7-
M
Q)
4-1
-------
00
p
o
•H
4-1
CO
J-l
4J 3.2xl0
6
C QJ
CD ,O
O -i-l
C >-H
O
CJ M-l
O
4-1
QJ !^
r-l
-------
twill bag produced the second lowest geometric mean outlet
concentration for outlet concentration of fibers greater
than 1.5 ym. Figure 9 shows that both cotton sateen and
cotton twill have the lowest geometric mean outlet concen-
trations of fibers greater than 6.0 ym. Thus, the best
overall bag fabric for reducing outlet concentration of
asbestos fibers is cotton sateen. This result supports that
of the correlation analysis.
Regression Analysis of Subsample 1
The regression model considered for Subsample 1 data
included the numeric coding of bag fabric and the log of
humidity, the dependent variables being the log of outlet
concentration for the three fiber lengths. None of these
factors were significantly related to outlet concentration
of fibers greater than 0.06 vim. However, bag fabric was
significantly related to outlet concentration of fibers
that were greater than 1.5 ym and 6.0 urn. The regression
statistics and equations for these two fiber lengths are
presented in Tables 20 and 21. The computed geometric means
of outlet concentration from these relations are given in
Table 22. It is apparent from this table that, overall,
the cotton sateen fabric produced the lowest outlet con-
centration of asbestos fibers. Again, this supports the
correlation and confidence interval results.
Subsample 2 Results
Only two bag fabrics, but both types of asbestos waste
and three humidity levels, are included in the analysis of
the second subsample of data from Phase I. The data
included in this subsample are tests 1-4, 12-14, and 21-26
in Table 17. Table 23 presents the values of the nine
factors employed.
69
-------
Table 20. RESULTS OF REGRESSION ANALYSIS OF SUBSAMPLE 1
PHASE I FOR FIBERS GREATER THAN 1.5 ym
Dependent Variable:
Data Base:
Degree of Determination:
Residual Standard Deviation;
Variables of Significance
Constant Term:
Y2 ~ Log^Q of Outlet
Concentration of
fibers ^.1-5 ym
Phase I - Subsample 1
54.87o
0.301
'31
:32
First coded bag variable
Fifth coded bag variable
Regression
Coefficient
+1.007
-0.757
-0.385
Standard
Error
0.172
0.193
Prob.
Level
0.0004
0.0591
Variables Not Significant (P > 0,10)
Zo2 , Second coded bag variable
Zo3. Third coded bag variable
Z^^, Fourth coded bag variable
^2, Log,Q of humidity
Regression Equation
Y2 = 1.007 - 0.757Z31 - 0.385Z32
70
-------
Table 21. RESULTS OF REGRESSION ANALYSIS OF SUBSAMPLE 1
PHASE I FOR FIBERS GREATER THAN 6.0 ym
Dependent Variable:
Data Base:
Degree of Determination:
of Outlet
Concentration of
fibers >_ 6.0 ym
Phase I - Subsample 1
60.1%
Residual Standard Deviation: 0.199
Variables of Significance
Constant Term:
J32
Second coded bag variable
oo, Third coded bag variable
'32
Fourth coded bag variable
Regression
Coefficient
+0.126
+0.433
+0.530
+0.264
Variables Not Significant (P > 0.10)
J31
First coded bag variable
Z35, Fifth coded bag variable
Z2, Log1Q of humidity
Regression Equation
Y = 0.12.6 + 0.433Z32 + 0.530Z33 + 0-264Z34
Standard
Error
Prob.
Level
0.178 0.0021
0.131 0.0010
0.131 0.0587
71
-------
Table 22. COMPUTED GEOMETRIC MEANS OF
OUTLET CONCENTRATION
Type of Bag
Cotton Sateen
Napped Cotton
Nomex
Dacron No. 1
Cotton Twill
Dacron No . 2
Computed Geometric Mean of Outlet Concentration
(Number of fibers/cm^)
Fibers >. 0 . 06 urn
585
585
585
585
585
585
Fibers >. 1.5 ym
1.78
10.16
10.16
10.16
4.18
10.16
Fibers >. 6 .0 ym
1.34
3.62
4.53
2.46
1.34
1.34
-------
Table 23. THE INDEPENDENT VARIABLES AND THEIR DESIRED
LEVELS FOR SUBSAMPLE 2 OF PHASE I
Variable
X-, , Waste Type
X~ , Humidity
X3, Bag Type
X, , Air-to-Cloth Ratio
X5, Dust Loading
X,-, Amplitude of Shake Cycle
X^ , Frequency of Shake Cycle
Xn, Period and Duration of
0 Shake Cycle
XQ, Number of Bags
Level
1.
2.
1.
2.
3.
1.
2,
1.
1.
1.
1.
1.
1.
Asbestos Cement
Raw Asbestos Fiber
20%
40%
80%
Cotton Sateen
Dacron No. 1
0.76 m3/min/m2 (2.5 cfm/ft2)
22 g/m3 (9.6 gr/ft3)
0.875 cm (0.344 in.)
1.0 cps
Period - 16 min
Duration - 20 sec
1
73
-------
Correlations Between Subsample 2 Variables -
The correlations of the outlet concentration or log
outlet concentration and the independent varibles in
Subsample 2 are given in Table 24. Since only two bag
fabrics, cotton sateen and Dacron No. 1, were used, only
one coded variable, Z31> for bag type was necessary. The
cotton sateen bag was coded as 1 and the Dacron No. 1 bag
as 0.
There are no significant correlations between these
three independent variables and the outlet concentration
and log outlet concentration of fibers greater than 0.06 ym.
This is in agreement with Subsample 1 results.
For outlet concentration and log outlet concentration
of fibers greater than 1.5 urn, only the type of bag had a
significant correlation. The correlations are both nega-
tive, indicating that cotton sateen tended to have a signifi-
cantly lower outlet concentration and log outlet concentration
than that of Dacron No. 1. This is again in agreement with
Subsample 1 results. Both humidity and type of waste were
not significantly correlated with outlet concentration or
log outlet concentration for this set of fiber lengths.
Waste type is the only correlation of significance for
fibers greater than 6.0 ym. Both bag type and humidity are
not correlated with outlet concentration or log outlet
concentration. The significant correlation of waste type
with both outlet concentration and log outlet concentration
is positive, indicating that asbestos cement waste tended
to have a higher outlet concentration and log outlet con-
centration than that of raw asbestos fibers.
Geometric Means of Outlet Concentration for Subsample 2 -
The geometric means and their 90% confidence limits
for Subsample 2 data are presented in Table 25. The means
are given separately for bag fabric and type of waste.
74
-------
Table 24. CORRELATIONS BETWEEN PHASE I VARIABLES
FOR SUBSAMPLE 2 (N = 13)
Concentration of
Fibers >.
0.06 ym
1.50 ym
6.00 ym
LogiQ Concentration
of Fibers >.
0.06 ym
1.50 ym
6.00 ym
Independent Variable
Zl
Z31
x2
Zl
Z31
x2
Zl
Z31
x2
Waste Type
Bag Type
Humidity
Waste Type
Bag Type
Humidity
Waste Type
Bag Type
Humidity
-
Zl
Z31
z2
Z1
zn
Z2
Zl
Z31
z2
Waste Type
Bag Type
Log^Q Humidity
Waste Type
Bag Type
Log1Q Humidity
Waste Type
Bag Type
Log10 Humidity
Correlation
r
+0.357
-0.342
-0.172
+0.197
-0.492
-0.100
+0.479
-0.338
-0.088
+0.329
-0.416
-0.178
-0.083
-0.569
+0.222
+0.524
-0.250
+0.071
Probability
P
*
*
*
*
0.05 < P < 1.
*
0.05 < P < 0.
*
*
*
*
*
*
0.02 < P < 0.
*
0.02 < P < 0.
*
*
10
10
05
05
Not
significant at 0.10 probability level.
75
-------
Table 25. GEOIIETRIC 1IEANS AND 90% CONFIDENCE LIMITS
FOR PHASE I - SUBSAMPLE 2
Type of
Bag Fabric
Cotton Sateen
Dacron No. 1
Type of
Waste
Cement
Raw Fiber
N
7
6
7
6
Outlet Concentration
(Number of Fibers/cm3)
Fibers > 0.6 pm
Geometric
Mean
414
771
693
424
Confidence Limits
Lower
245
412
354
273
Upper
700
1442
1353
657
Fibers > 1.5 ym
Geometric
Mean
2.51
5.92
3.52
3.99
Confidence Limits
Lower
1.65
3.12
1.58
3.39
Upper
3.82
11.23
7.83
4.70
Fibers > 6.0 ym
Geometric
Mean
1.20
1.53
1.70
1.02
Confidence Limits
Lower
0.099
0.085
1.10
0.88
Upper
1.45
2.74
2.63
1.17
-------
These means and their confidence intervals are plotted in
Figures 10-13. These plots show that cotton sateen produced
significantly lower fiber concentrations for fiber lengths
greater than 1.5 ym than did Dacron No. 1, and that cement
waste had higher outlet concentrations of fibers greater
t-han 6.0 ym than did raw asbestos fiber waste. No signifi-
cant effects were detected for the other fiber lengths.
These findings support the correlation results.
Regression Analysis^ of Subs ample 2
The regression analysis for Subsample 2 of Phase I
data included the type of waste along with bag type (cotton
sateen and Dacron No. 1) and the log of humidity as candidate
variables. Again, none of the candidate variables were
related to the log outlet concentration of fibers greater
than 0.06 ym. For fibers greater than 1.5 ym, the type of
bag significantly affected the log of outlet concentration.
The computed geometric mean of outlet concentration for
these fibers was 2.51 fibers per cubic centimeter for the
cotton sateen bag and 5.92 fibers per cubic centimeter for
the Dacron No. 1 bag. These means were computed from the
regression equation in Table 27. For fibers greater than
6.0 ym, only the type of waste significantly affected the
log of outlet concentration. For this fiber size, the
geometric mean of outlet concentration generated by cement
waste is 1.70 fibers per cubic centimeter, and that generated
by raw asbestos fiber is 1.02 fibers per cubic centimeter.
Again, the log of humidity did not significantly affect
outlet concentration beyond experimental error. The
regression statistics and equations are given in Tables 26
and 27.
Conclusions from Phase I Results
The results of Phase I all support the use of cotton
sateen bag fabric for reducing the outlet concentration of
77
-------
0)
01
4-J
(0
C/3
o
4J
4-)
O
O
o
V4
o
ttf
Q
1200"
•H e
4J U
ca -^
M 03
C QJ 600-
-------
E
o
U CO
(8 Vl
V-l 01
4J ,0
C -i-l
01 IH
a
C MJ
o o
u
-------
1600
800
o
•H
4-J
cd
0)
a
a
o
a
6
o
en
5-i
a»
JO
•r-l
M-l
O
0)
1
c
400
200
0
o
Cement
Waste
Raw Asbestos
Fibers
Figure 12. Estimates of the geometric mean and their 90%
confidence intervals for outlet concentration of asbestos
fibers greater than 0.06 ym by type of waste -
Phase I - Subsample 2
80
-------
o
u
CO
0)
XI
•H
ro
C E
0 O
•rt ---.
4J CO
Ct) }-l
M 0)
4-1 XI
C -rl
1. 5 ym
Fibers ^ 6 . 0 Pm
Figure 13. Estimates of the geometric mean and their 90%
confidence intervals for outlet concentration of asbestos
fibers by type of waste - Phase I - Subsample 2
81
-------
Table 26. RESULTS OF REGRESSION ANALYSIS OF SUBSAMPLE 2
PHASE I FOR FIBERS GREATER THAN 1.5 ym
Dependent Variable:
Data Base:
Degree of Determination:
Residual Standard Deviation:
Var i ab1es of S igni ficance
Constant Term:
ZOTi Coded bag variable
°f Outlet
Concentration of
fibers >_ 1.5 ym
Phase I - Subsample 2
32.4%
0.292
Regression
Coefficient
+0.773
-0.373
Variables Not Significant (P > 0.10)
Z-, , Coded waste variable
Z0 , Log-, Q of humidity
Regression Equation
Yo = 0.773 - 0.373Z
Standard
Error
0.162
Prob.
Level
0.0438
31
82
-------
Table 27. RESULTS OF REGRESSION ANALYSIS OF SUBSAMPLE 2 -
PHASE I FOR FIBERS GREATER THAN 6.0 ym
Dependent Variable:
Data Base:
Degree of Determination:
_, LoglO of Outlet
Concentration of
fibers >_ 6.0 ym
Phase I - Subsample 2
27.5%
Residual Standard Deviation: 0.197
Variables of Significance
Constant Term:
Z-, , Coded waste variable
Regression
Coefficient
0.118
0.112
Variables Not Significant (P > 0.10)
ZOT, Coded bag variable
Zj, ^°SIQ °f humidity
Regression Equation
Y3 = 0.118 + 0.112Z].
Standard
Error
0.055
Prob.
Level
0.0662
83
-------
asbestos fibers. For fibers greater than 0.06 ym, cotton
sateen is not significantly different from other bag fabrics.
For fibers greater than 1.5 ym, cotton sateen had outlet
concentrations significantly lower than any of the other
bags tested. And for fibers greater than 6.0 ym, cotton
sateen had outlet concentrations significantly lower than
napped cotton, Nomex, and Dacron No. 1, and performed as
well as cotton twill or Dacron No. 2.
Humidity had no effect on outlet concentration greater
than experimental error, but type of waste significantly
affected the outlet concentration of fibers greater than
6.0 ym.
For fibers greater than 6.0 ym, the raw asbestos
fiber waste had a significantly lower outlet concentration
than that of asbestos cement waste.
That cotton sateen allows the lowest outlet concen-
tration is not surprising. This fabric has one of the
fullest weaves (4 x 1) and highest thread counts (96 x 60)
of the fabrics tested, thus reducing pore size. Also, the
fabric is manufactured from all spun fibers, so that the
fabric fabriles also act to effectively reduce the pore
size.
That humidity has no effect on outlet concentration can
be attributed to two factors. First, the experimental error
was very high in this phase, thus possibly masking any
effects. Second, at the highest level of relative humidity
(75-85%), both settling of agglomerated dust in the inlet
duct and blinding of the fabric by moisture were observed.
The conclusion that raw asbestos fiber waste had a
lower outlet concentration in the greater than 6.0 ym range
is reasonable in that the longer fibers expected in the raw
asbestos fiber waste would be collected more easily than
those fibers shortened by the asbestos cement process. If
84
-------
the raw asbestos fiber had been from one of the longer
length grades of fiber, such as those used in textiles, etc.,
a correlation would also be expected between waste type and
outlet concentration for the smaller fibers. However, the
raw asbestos fibers used were also of the shorter length
grades used in the asbestos cement industry.
DISCUSSION OF PHASE II RESULTS
Phase II examined effects of relative humidity, air-
to-cloth ratio, and dust loading. Phase I demonstrated
that a bag made of cotton sateen was more efficient overall
than other bag fabrics. Thus, this bag fabric was used
throughout the remaining tests. Phase I also demonstrated
that the outlet concentration for asbestos cement waste was
higher than that for fibrous asbestos waste under similar
baghouse conditions. : Since asbestos cement waste has higher
outlet concentrations, it is a larger scale problem indus-
trially; and since fibrous asbestos waste could not be fed
by the SCR-20 dust feeder, only asbestos cement waste was
used for Phase II and the remaining phases of the study.
The independent variables are given in Table 28. Desired
and actual levels of the independent variables differed;
however, these differences did not significantly affect the
validity of statistical analysis.
Outlet concentrations of asbestos fibers were measured
for two different size ranges (>_ 1,5 ym and >_ 6.0 ym) in
Phase II. The data base generated by this phase of testing
is given in Table 29. A total of nineteen tests were con-
ducted. Tests 2 and 3, and 17 and 18 were treated as
replicate tests. Due to the poor precision in counting
the fibers, two counts of the concentration of fibers for
each test and size range were made. Although these second
estimates are not complete replicates of each test, they
were treated as replicates since the variation in the
85
-------
Table 28. PHASE II INDEPENDENT VARIABLES AND
THEIR DESIRED LEVELS
Variable
X-,, Waste Type
Xj > Humidity
X3> Bag Type
X^, Air-to-Cloth Ratio
Xr, Dust Loading
Xg, Amplitude of Shake Cycle
Xy , Frequency of Shake Cycle
Xg, Period and Duration of
Shake Cycle
XQ , Number of Bags
Level
1.
1.
2.
3.
1.
1.
2.
3.
1.
2.
3.
1.
1.
1.
1.
Asbestos Cement
30%
40%
60%
Cotton Sateen
0.46 m^/min/m? (1.5
0.76 m^/min/m, (2.5
1.22 mj/min/m (4.0
10 g/m3, (4.4 gr/ft3,)
21 g/m^ (9,2 gr/ftj)
45 g/nT (19.7 gr/ft3
0.875 cm (0.344 in.)
1 cps
Period - 16 min
Duration - 20 sec
1
cfm/ft?)
cfm/fto)
cfm/fO
j
86
-------
Table 29. DATA BASE OF PHASE II
Test No.
1
2
3
4
5
6
10
11
i A
12
13
T /
14
-i r-
15
* X"
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
-
Comb. No.
1
1
2
2
2*
2*
3
3
4
4
5
5
6
6
7
7
8
8
9
9
10
10
11
11
12
12
13
13
14
14
15
15
16
16
16*
16*
17
17
:X2
60
60
40
40
40
40
40
40
60
60
30
30
30
30
40
40
40
40
30
30
60
60
40
40
6.0
60
60
60
60
60
30
30
30
30
30
30
40
40
X4
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5
2.5
2.5
1.5
1.5
3.9
3.9
4.0
4.0
3.2
3.2
2.5
2.5
3.0
3.0
2.5
2.5
1.5
1.5
3.0
3.0
2.5
2.5
2.5
2.5
2.5
2.5
2.1
2.1
2.2
2.2
XS
45
45
21
21
21
21
10
10
45
45
45
45
10
10
21
21
56
56
21
21
28
28
45
45
10
10
14
14
21
21
10
10
72
72
85
85
11
11
Outlet Concentration
(No. fibers/m3 x 106)
> 1.5 ym
16.72
17.92
27.05
19-28
19.41
24.62
13.42
29.45
23.19
34.97
14.14
22.71
10.90
13.37
7.29
17 . 93
13.02
16.12
7.30
31.47
12.61
22.10
17.30
33.52
23 . 44
50.03
15.08
34.43
21.26
45.26
23.72
43.90
13.98
34.06
21.44
34.90
60.75
41.75
> 6.0 pm
3.54
4.68
4.26
2.13
2.34
3 62
*y • v *—
3.41
3 88
5.35
6.89
2.42
3 80
-j • \J \J
1.45
1 94
1.21
1.52
0.61
2.65
1.27
4.84
1.67
3.02
2.66
2.81
3.74
3.40
1.67
2.15
2.73
2.23
4.36
2.48
1.47
1.83
2.11
2.06
5.05
3.90
* Full replicates
X2 - Percent Humidity, X4 - Air-to-Cloth Ratio (cfm/ft )
X5 - Dust Loading (g/m3)
Fixed Variables
X-L - Asbestos Cement Waste, X-j - Cotton Sateen Bag
X6 - Amplitude 0.875 cm (0.344 in.)
X7 - Frequency 1.0 cps
Xg - Period - 16 min.
Duration - 20 sec.
87
-------
counting is on the same order of magnitude as the complete
replicates. Thus, there are a total of thirty-eight obser-
vations of outlet concentrations for seventeen combinations
of the independent variables.
Correlations Between Phase II Variables
Correlation coefficients, r, are given in Table 30 for
outlet concentrations in each of the two size ranges and
their common log transforms when paired with the specified
independent variables. The range within which the associated
probability, P, falls is indicated in conjunction with each
correlation.
There are no significant correlations between dust
loading and outlet concentration or its transform, and
between relative humidity and outlet concentration and its
transform except for the outlet concentrations of fibers
greater than 6.0 ym. The probability that this correlation
is not significant is approximately 0.10.
The most significant correlation for all fiber sizes
considered is between the outlet concentration or its log
and the air-to-cloth ratio. This correlation is strongest
for the log transformed variables for fiber sizes greater
than 6.0 ym. The sign of this correlation is negative,
indicating that increasing air-to-cloth ratio in the range
addressed by this study decreases the outlet concentration
of asbestos fibers.
Regression Analysis of Phase II
The candidate variables for predicting the outlet con-
centration of asbestos fibers for each of the two size
ranges were:
Percent humidity
Air-to-cloth ratio
Dust loading
88
-------
Table 30. CORRELATIONS BETWEEN PHASE II VARIABLES
Dependent
Variable
Cone . of fibers >_ 1.5 ym
Cone . of fibers >_ 1.5 ym
Cone, of fibers >^ 1.5 ym
Cone, of fibers > 6.0 ym
Cone, of fibers > 6.0 ym
Cone . of fibers >_ 6.0 ym
Log,Q Cone, of fibers > 1.5 ytn
Log-,,, Cone, of fibers > 1.5 ym
LognQ Cone, of fibers > 1.5 ym
Log-,Q Cone, of fibers > 6.0 ym
LogiQ Cone, of fibers > 6.0 ym
Log-i,, Cone, of fibers > 6.0 ym
independent Variable
X2 Percent Humidity
X, Air-to-Cloth Ratio
Xc Dust Loading
X2 Percent Humidity
X4 Air-to-Cloth Ratio
X,- Dust Loading
Z~ L°Sir) °^ Percent Humidity
Z, Log-,,, of Air-to-Cloth Ratio
Zc Lo§in °^ Dust Loading
Z«' Log,0 of Percent Humidity
Z, Log-|0 of Air-to-Cloth Ratio
Z, Log-,Q of Dust Loading
(N = 38)
Correlation
r
+ .122
-.266
-.141
+ .275
-.458
-.142
+ .156
-.281
-.134
+ .258
-.517
-.162
Probability
P
*
.10 > P > .05
*
.10 > P > .05
.005 > P > .001
*
*
.10 > P > .05
*
*
P < .001
*
00
* Not significant at 0.10 probability level
-------
Separate analyses were performed for each fiber size range.
The analyses attempted to relate the independent variables
and their common log transformations to the log of outlet
concentrations. The log transforms provided the best fit
to the data. Pertinent statistics for the equations of
fibers of the two size ranges are given in Tables 31 and 32.
The only significant variable is air-to-cloth ratio which
tends to decrease the outlet concentration of asbestos
fibers when it is increased. The equation for fibers greater
than 1.5 ym is given in Table 31. By taking the antilog of
this equation, the relation between outlet concentration
and air-to-cloth ratio is
30.9
outlet concentration =
(air-to-cloth ratio)0'430
This relation, along with the geometric mean and 90% con-
fidence intervals for each level of air-to-cloth ratio, is
plotted in Figure 14. From this plot, it is strikingly
apparent that there is a large variation in the replicate
counts for a given combination of independent variables.
However, the fit of the regression line is good, given the
variance of measurement; the F value for the lack of fit
being 1.43 with an associated probability of 0.22.
The equation relating air-to-cloth ratio and outlet
concentration of asbestos fibers of a size greater than
6.0 ym is given in Table 32. This equation can be trans-
formed by taking its antilog. The resulting equation is then
4.98
outlet concentration =
(air-to-cloth ratio)0'784
This relation, along with the geometric means and 9070 con-
fidence intervals for each level of air-to-cloth ratio, are
plotted in Figure 15. The geometric means were calculated
by taking the antilog of the arithmetic mean of log outlet
concentration. Again, the variation in the data for given
90
-------
Table 31. RESULTS OF REGRESSION ANALYSIS OF PHASE II
FOR FIBERS GREATER THAN 1.5 pm
Dependent Variable:
Data Base:
Degree of Determination:
Residual Standard Deviation;
Independent Variables
Constant Term:
Y, = LoglQ °f Outlet
Concentration of
fibers >L 1-5 ym
All tests shown in Table 29
7.9%
0.212
1.490
Variables of Significance
Z,, Log,Q of air-to-cloth ratio
Regression Standard Prob.
Coefficient Error Level
-0.430
0.245
0.084
Variables Not Significant (P > 0.10)
Z^, Login of relative humidity
Z,-,
of dust loading
Regression Equation
YX = 1.490 - 0.430Z4
91
-------
Table 32, RESULTS OF REGRESSION ANALYSIS OF PHASE II
FOR FIBERS GREATER THAN 6.0 ym
Dependent Variable:
Data Base:
Degree of Determination:
Residual Standard Deviation:
Independent Variables
Constant Term:
Variables of Significance
Y2 =
of Outlet
Concentration of
fibers > 6.0 ym
All tests shown in Table 29
26.72%
0.187 .
0.697
Regression
Coefficient
Z^, LO§XO air-to-cloth ratio -0.784
Variables Hot Significant (P > 0.10)
Z2> Log1Q of relative humidity (P = 0.122)
Z,-, Log,n of dust loading
Regression Equation
Yo = 0.697 - 0.784Z,
Standard
Error
0.216
Prob.
Level
0.001
92
-------
60,0
30.0
c
0
•H
4J
cfl
^
c
(U
o
c
0
u
4J
QJ
H
4J
^^
6
u
\
en
^
cu
•H
4-1
M-)
0
(U
rQ
S
3
C
15.0
7.5
1.5 2
Air-to-Cloth R.atio
(cfm/ft'
Figure 14. Geometric means, 907o confidence intervals, and the
regression line for outlet concentration of fibers greater
than 1.5 ym by air-to-cloth ratio - Phase II
93
-------
Cj
-------
values of air-to-cloth ratio is high. However, the 1.64 F
value and 0.14 probability level for lack of fit demonstrate
that the fit of the equation to the observations is not
inappropriate.
The percentage of the total variance in the Phase II
data base that can be attributed to experimental error is
80% for fibers greater than 1.5 ym and 59% for fibers greater
than 6.0 urn. The experimental error was computed from the
full replicate tests and the replicate counts made for each
experimental run.
Conclusions^ from Phase II Results
Air-to-cloth ratio has been shown to be a significant
variable to control for reducing the concentration of emitted
asbestos fibers from a baghouse. For air-to-cloth ratios
ranging from 0.46 to 1.22 m /min/m (1.5 to 4.0 cfm/ft2) , an
air-to-cloth ratio of 1.22 m3/min/m2 (4.0 cfm/ft2) tends to
be the best value for reducing outlet concentrations of
asbestos fibers.
The relative humidity and dust loading did not demon-
strate an effect on outlet concentration. However, since
the variance accounted for by these variables had to be
greater than experimental error, these variables cannot be
ruled out as factors that affect the outlet concentration.
The experimental error is quite high and may have masked
the effects of these variables. New techniques are necessary
to precisely calculate the outlet concentration of fibers.
For fiber sizes greater than 6.0 ym, the estimates of outlet
concentration became more precise. For these fiber sizes,
relative humidity had a positive correlation with outlet
concentration that was slightly less than the 0.10 signifi-
cance level. This means that humidity may decrease the
outlet concentration of asbestos fibers. Although such a
95
-------
relation is not definitive, from a practical standpoint,
low humidity should be maintained for baghouse operations,
if possible.
Thus, Phase II results established that for air-to-
3 2
cloth ratios within the range of 0.46 to 1.22 m /min/m
(1.5 to 4.0 cfm/ft2), the higher air-to-cloth ratio decreases
the outlet concentration of asbestos fibers from baghouses.
And secondly, the results indicate that a low humidity may
decrea-se outlet concentration particularly for fibers greater
than 6.0 ym.
The demonstrated effect of outlet concentration decreasing
with increasing air-to-cloth ratio is not the normally
expected result; however, that effect has been extrapolated
by some workers. In fact, ultra-high air-to-cloth ratios
have been suggested as a possible means of increasing bag-
house collection efficiency. It seems that, in the low range
of air-to-cloth ratios studied, the highest air-to-cloth ratio
32 2
of 1.22 m /min/m (4.0 cfm/ft ) increases the rate of dust
cake build-up in a manner that increases collection efficiency
for asbestos cement waste.
A possible mechanism for the decrease in outlet concen-
tration with reduced humidity may be forwarded from the experi-
mental operating difficulties of Phases I and II. At very
high relative humidity levels (75-8570) , agglomeration of
fibers due to condensation occurs to such a degree that much
of the dust loading is collected in the inlet duct. Also,
blinding of the fabric by moisture is evident, thus increasing
the pressure drop. Assuming that this agglomeration still
takes place at moderate levels of relative humidity, it can
be postulated that many of the larger fibers grow to such
size as to be settled in the gravity settling chamber and
thus never reach the filter fabric. Therefore, the filtering
dust cake build-up is less rapid and the dust cake consists
96
-------
of smaller fibers, some of them incapable of bridging the
fabric pores, and the outlet concentration of fibers increases
The lack of a significant effect on outlet concentration
by dust loading may be explained in the following manner. At
relatively high dust loadings, the filtering dust cake build-
up is so rapid regardless of the level of dust loading, that
the outlet concentration is decreased to such a degree that
it is independent of the dust loading. Also, at the higher
levels of relative humidity, a large fraction of the dust
loading may be removed by gravity separation of the agglom-
erated fibers.
DISCUSSION OF PHASE III RESULTS
Phase III investigated the effects of shake frequency,
amplitude, and the joint variation of period and duration of
the mechanical shaking cycle on outlet concentration of as-
bestos. The air-to-cloth ratio, dust loading, and relative
humidity were all held constant throughout this phase. All
combinations of the independent variables were tested.
Actual levels of the independent variables and observed
concentrations of asbestos emissions are given in Table 33.
As in Phase II, the bag fabric was cotton sateen, and the
waste was from an asbestos cement plant. The outlet concen-
tration of asbestos fibers was measured for two different
size ranges (>_ 1.5 ym and >_ 5.0 ym) . (The size range of
>_ 5.0 ym was substituted for that of >_ 6.0 ym after the post
Phase II critique in order to extend the limits of the more
accurate larger size range to the limits of those fibers
clearly and easily viewed. This size range also corresponds
with those of the standard OSHA and AIHA methods.) A total
of eight combinations of the shaking variables were examined.
Estimates of outlet concentration were made two or three
times for each combination by duplicate counts. These
independent estimates of outlet concentration were treated as
97
-------
replicates to assess the fit of the regression equations and
estimate the percent of overall variation that can be
attributed to experimental error. Thus, there are eighteen
tests listed in Table 33 comprised of eight different com-
binations of the independent variables.
The period and duration were treated as a single
variable with only two levels. The first level represents
a period of 120 minutes and a duration of 20 seconds. The
second level represents a period of 16 minutes and a dura-
tion of 80 seconds. These levels and those of amplitude
and frequency were combined to form a full factorial design.
Correlations Between Phase III Variables
Correlation coefficients, r, are given in Table 34 for
each of the two outlet concentrations and their common log
transforms. The range within which the associated probabil-
ity, P, falls is indicated in conjunction with each correla-
tion.
There are no significant correlations between the
dependent variables and amplitude or frequency of the shake
cycle. The correlations between all the dependent variables
and the joint period and duration levels are significant.
These correlations indicate that an increase in period with
a corresponding decrease in duration reduces the outlet
concentration of asbestos fibers.
Regression Analysis of Phase III
The candidate variables for predicting the emission
concentration of asbestos fibers for each of the two size
ranges were:
Amplitude
Frequency
Period
98
-------
Table 33. DATA BASE FOR PHASE III
Test
No.
1
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Comb.
No.
}
}
1
2
2
3
3
3
4
4
5
5
6
6
7
7
8
8
X8
120
120
120
16
16
16
16
16
16
16
16
16
120
120
120
120
120
120
X6
0.875
0.875
0.875
3.500
3.500
0.875
0.875
0.875
3.500
3.500
0.875
0.875
0.875
0.875
3.500
3.500
3.500
3.500
X7
1
1
1
5
5
5
5
5
1
1
1
1
5
5
1
1
5
5
Outlet Concentration
(No. of fibers/cm3)
> 1.5 ym
1.1259
3.5993
3.0918
5.0280
5.5615
17.1133
10.0717
7.7378
4.6736
6.6152
8.1092
7.6284
3.2037
3.0821
1.4636
2.2320
2.4165
2.9695
> 5.0 ym
0.3181
1.3189
0.5064
1.9670
1.8882
8.7649
1.9002
1.2269
1.7063
1.3160
2.3924
1.4663
0.7395
0.5956
0.2758
0.3370
0.3365
0.3247
X6 -
X -
Period and duration setting
a period of 120 min dictates a duration of 20 sec
a period of 16 min dictates a duration of 80 sec
Amplitude (cm)
Frequency (cps)
Fixed Variables
X-, - Asbestos cement waste
X2 - 40-90% humidity
X~ - Cotton sateen bag
- 0.92 m3/min/m2 (3.0 cfm/ft ) air-to-cloth ratio
X5 - 21 g/m3 (9.2 gr/ft3) dust loading
XQ - No second bag
99
-------
Table 34. CORRELATIONS BETWEEN PHASE III VARIABLES
Dependent
Variable
Cone, of fibers > 1.5 ym
Cone, of fibers > 1.5 ym
Cone, of fibers > 1.5 ym
Cone . of fibers > 1.5 ym
Cone, of fibers > 5.0 ym
Cone . of fibers > 5.0 ym
Cone, of fibers > 5.0 ym
Cone, of fibers >_ 5.0 ym
Log-, 0 Cone . of fibers > 1.5 ym
Log,0 Cone, of fibers _> 1.5 ym
Log,Q Cone, of fibers >_ 1.5 ym
Log,n Cone, of fibers >_ 1.5 ym
Log,0 Cone, of fibers >_ 5.0 ym
Log, 0 Cone, of fibers ;> 5.0 ym
Log,0 Cone, of fibers >_ 5.0 ym
Login Cone, of fibers >_ 5.0 ym
Independent Variable
X, Amplitude
X^ Frequency
Xg Period
Xg Duration
Xg Amplitude
Xy Frequency
Xg Period
Xq Duration
Zg Log-, Q Amplitude
Z-, Log,Q Frequency
ZQ Log, n Period
o X, \j
Zg Log-,Q Duration
Zg Log, « Amplitude
Z-, Log, Q Frequency
ZQ Log, n Period
O J- \J
Zg Log,Q Duration
(N = 18)
Correlation
r
-0.343
+0.275
-0.727
+0.727
-0.239
+0.239
-0.528
+0.528
-0.276
+0.290
-0.833
+0.833
-0.274
+0.193
-0.819
+0.819
Probability
P
*
*
P < 0.01
P < 0.01
*
0.02 < P < 0.01
0.01 < P < 0.01
*
*
P < 0.01
P < 0.01
*
P < 0.01
P < 0.01
o
o
* Not significant at 0.10 probability level
-------
of the shake cycle. The period of the cycle was used in lieu
of the duration of the cycle. Both of these variables had a
perfect inverse correlation. Thus, the effects of period on
the outlet concentration are the effects of both period and
duration jointly.
Separate analyses have been performed for each set of
fiber lengths. The common log transformation of the variables
provided the best fit to the data. The pertinent statistics
for each equation are given in Tables 35 and 36.
The equation for the relation between outlet concen-
tration of asbestos fibers greater than 1.5 ym and the mechan-
ical shaking variables is given in Table 35. By taking the
antilog of this equation, the relation becomes
39 90
concentration = -j^.^u
(period)0'557 • (amplitude)0'270
Thus, as period and amplitude are increased, the outlet con-
centration is reduced. However, since period and duration
were varied jointly, an increase in period must be accom-
panied by a corresponding decrease in duration for this
relation to hold. Plots of this equation with the corre-
sponding 90% confidence intervals about the geometric means
at various amplitudes and period levels are given in
Figures 16 and 17.
The equation for the relation between outlet concen-
tration of asbestos fibers greater than 5.0 ym and the mechan-
ical shaking variables is given in Table 36. By taking the
antilog of this equation, the relation becomes
18.64
concentration = A 736 • ~, -> •_ j >.0.360
(period) (amplitude)
This equation is functionally the same as the equation for
fibers greater than 1.5 ym. Plots of this equation with the
corresponding 90% confidence intervals about the geometric
101
-------
Table 35. RESULTS OF REGRESSION ANALYSIS OF PHASE III
FOR FIBERS GREATER THAN 1.5 ym
Dependent Variable: Y-, = Logio of Outlet
Concentration of
fibers >^ 1.5 ym
Data Base: All tests shown in Table 33
Degree of Determination: 77%
Residual Standard Deviation: 0.1538
Regression Standard Prob.
Variables of Significance Coefficient Error Level
Constant Term: +1.600
Zg, Log1Q of period -0.557 0.083 <0.001
Zg, Log1Q of amplitude -0.260 0.121 0.040
Variables Not Significant (P > 0.10)
Z7> Log1Q frequency
Regression Equation
Y-, = 1.600 - 0.260Z, - 0.557Z,n
1 D 1U
102
-------
Table 36. RESULTS OF REGRESSION ANALYSIS OF PHASE III
FOR FIBERS GREATER THAN 5.0 urn
Dependent Variable:
Data Base:
Degree of Determination:
Residual Standard Deviation:
of Outlet
Concentration of
fibers >_ 5.0 um
All tests shown in Table 33
74.5%
0.2178
Variables of Significance
Constant Term:
Zg, Log1Q of period
Zg, Login of amplitude
Regression
Coefficient
1.270
-0.736
-0.360
Variables Not Significant (P > 0.10)
Z7, Log10 of Frequency
Regression Equation
Y9 = 1.270 - 0.360Z, - 0.736Zg
Standard
Error
0.1173
0.1716
Prob.
Level
<0.0001
0.0511
103
-------
Amplitude = 3.50 cm
8.0-
4.2
C oo
o B
•rt O
•M -•-.
nj co
M l-i
JJ d) r\ /-. -
C rO 2.0
O) -H
O <4-l
c
o •w
U 0
u ^
CD (U
1-1 -9
^ 1 1.0-
P P
o c
v v
0.8-
0.4-
•V
^w^
v. Fibers >
T" VN 1 . 5 urn
1
C)
j
1\
N
\
\
N
N
N
\
O Fibers >
6.0 pm
1 III
' III
20 40 60 80 100 120
Shake Period (min)
Figure 16. Geometric means, 90% confidence intervals, and the
regression lines for outlet concentration of asbestos fibers
by shake period for an amplitude = 3.50 cm - Phase III
104
-------
Amplitude = 0.875 cm
10.0--
8.0,.
^ 4.0
d co
0 0
•H CJ
4-J ^-
nj CQ
•M Q)
d ,n
a) -H
O 4-1
o «w 2.0-
u o
4-* V-l
CU 0)
rH ,Q
JJ g
0 -S
1.0.
0.8-
0.4.
«
<
«
\
X.
^s
\
^^
\
) x
\ f
\ s
\
\
\
\
fli
\ I
\
•^
\
N
•
J 1 1 1 i_i
M^B
Fibers >
) 1 . 5 pm
!=
D
Fibers > I
6.0 ym
'
20
40
60 80 100 120
Shake Period (min)
Figure 17. Geometric means, 9070 confidence intervals, and the
regression lines for outlet concentration of asbestos fibers
by shake period for an amplitude = 0.875 cm - Phase III
105
-------
means at various amplitude and period levels are also shown
in Figures 16 and 17.
The percentage of the total variance in the data base
that can be attributed to experimental error is 27?0 for
fibers greater than 1.5 ym and 39% for fibers greater than
5.0 ym. The experimental error for Phase III was computed
from replicate counts of outlet concentration. No full
replicate experimental runs were made in this phase.
Conclusions from Phase III Results
Both the correlation and regression analysis establish
the joint variation of period and duration as the most
important mechanical shaking factor affecting outlet concen-
tration for both asbestos fiber sizes. Outlet concentration
was the least when a long period (120 min) and a short
duration (20 sec) were used. The regression analysis also
indicated that amplitude may have a significant effect on
outlet concentration. The higher amplitude (3.5 cm) had an
overall lower outlet concentration for both fiber sizes.
However, amplitude was not by itself significantly corre-
lated with the outlet concentrations.
This anomaly occurred because the values of period and
amplitude were not completely uncorrelated in the experiment.
Since period was included in the equation, one can consider
that amplitude is a significant factor in reducing outlet
concentration when the effects of period and duration are
controlled.
The percent of variance attributed to experimental
error was less in Phase III than Phase II. However, this
percentage was between 25 and 4070, and a method for reduction
of this error would be useful in any follow-up work.
As substantiated by the Royco traces of total parti-
cles (see Figure 6), the outlet concentration of fibers is
106
-------
reduced by using a combination of long period and short
duration shaking. The long period allows the filtering dust
cake to remain intact and at high efficiency for longer
periods of operation than do several short periods. The
short duration shake of 20 sec allows removal of the major
portion of the dust cake in order to reduce pressure drop3
while minimizing the disturbance of the filtering efficiency
of the dust cake. The higher levels of either shaking
amplitude or frequency reduce the maximum pressure drop
during a cycle by increasing dust release. However, it
seems that the greater amplitude better maintains the fil-
tering characteristics of the dust cake than does the
greater frequency. Thus, amplitude becomes important as a
means of reducing pressure drop for the long period and short
duration cycle while maintaining efficiency of collection.
DISCUSSION OF PHASE IV RESULTS
Phase IV investigated the effects of placing a second
bag in series with the original bag on the concentration of
asbestos emissions. The two bags were made of cotton sateen
fabric. The other variables were held constant (see Table 37)
As a secondary investigation, the stabilization period
was varied for this phase of experimentation. Prior to this,
a 24-hour period of stabilization had been used. The outlet
concentrations of asbestos fibers were measured for two
different size ranges (>_ 1.5 ym and >_ 5.0 ym) . The data
generated by this phase consisted of three combinations of
tests; one combination for the three different periods of
stabilization time. First, the effect of stabilization time
on the outlet concentration of asbestos fibers was investi-
gated. Then the effects of a bag in series was assessed.
The data generated from Phase IV were compared with
that of combinations No. 7 and No. 8 of Phase III (refer to
Table 33) to assess the effects of using a double bag
107
-------
Table 37. PHASE IV INDEPENDENT VARIABLES AND
THEIR DESIRED LEVELS
Variable
X-,, Waste Type
2 '
3 '
X4>
x5.
X6'
xy,
Xg,
Humidity
Bag Type
Air-to-Cloth Ratio
Dust Loading
Amplitude of Shake Cycle
Frequency of Shake Cycle
Period and Duration of
Shake Cycle
Xg, Number of Bags
X10'
Stabilization Period
Level
1.
1.
1.
1.
1.
1.
1.
1.
1.
2.
1.
2.
3.
Asbestos Cement
60-907o
Cotton Sateen
0.92 m3/min/m2 (3.0 cfm/ft2)
21 g/tn3 (9.2 gr/ft3)
3.5 cm (1.378 in.)
1.0 cps
Period - 120 min
Duration - 20 sec
1
2
24 hours
70 hours
164 hours
108
-------
arrangement over that of a single bag. Combinations No. 7
and No. 8 were found appropriate for comparison since the
independent variables, other than number of bags, humidity,
and frequency of shake cycle, were identical. Phases I, II,
and III established that the latter two variables did not
affect outlet concentration beyond the level of experimental
error.
Stabilization Period Results
The data base for Phase IV is given in Table 38. Two
estimates of outlet concentration were made for each com-
bination. The mean and 907=> confidence limits for the out-
let concentration for each stabilization period for both
fibers greater than 1.5 ym and 5.0 ym are given in Table 39
and plotted in Figures 18 and 19. The confidence intervals
overlap the geometric means for all three stabilization
times for fibers greater than 1.5 ym, thus indicating that
there is no significant effect of stabilization time on
outlet concentration for fibers greater than 1.5 ym beyond
the experimental error of this phase.
For fibers greater than 5.0 ym, the lowest stabilization
time (24 hours) produced a confidence interval that does not
overlap the means of the other stabilization levels. The
geometric mean of the outlet concentration for this stabili-
zation time, 0.322, is lower than that of the other two
stabilization times. Thus, for fibers greater than 5.0 jam
with all other conditions being constant, the outlet concen-
tration is significantly lower for a stabilization time of
24 hours as compared with stabilization times of 70 and
164 hours. No other conclusions other than this should be
drawn from these results.
Results of Two Bags in Series
Since frequency was not established as a significant
variable on outlet concentration during Phase III,
109
-------
Table 38. DATA BASE FOR PHASE IV
Test
No.
1
2
3
4
5
6
Comb.
No.
1
1
2
2
3
3
Stabilization
Period (hrs)
24
24
70
70
164
164
Outlet Concentration
(No. of fibers/cm^)
>. 1.5 ym
1.8702
3.0754
4.2941
4.8472
3.5959
3.9890
>. 6.0 urn
0.3169
0.3278
0.8554
0.7317
0.5193
0.5628
Fixed Variables
X -
- Asbestos cement waste
60-90% relative humidity
X~ - Cotton sateen bag
X^ - 0.91 m3/min/m2 (3.0 cfm/ft2) air-to-cloth ratio
X5 - 21 g/m3 (9.2 gr/ft3) dust loading
X, - 3.5 cm (1.378 in.) amplitude of shake cycle
Xy - 1.0 cps frequency of shaking
Xg - Shake period 120 min; duration 20 sec
XQ - Two bags in series
110
-------
Table 39. GEOMETRIC MEAN AND 90% CONFIDENCE LIMITS OF OUTLET
CONCENTRATION FOR DIFFERENT STABILIZATION PERIODS
Stabilization
Period
24
70
164
Outlet Concentration*
Fibers > 1.5 ym
N
2
2
2
Geometric
Mean
2.632
4.562
3.787
Confidence Limits
Lower
0.449
3.112
2.730
Upper
11.523
6.688
5.255
Fibers > 5.0 ym
N
2
2
2
Geometric
Mean
0.322
0.791
0.541
Confidence Limits
Lower
0.290
0.483
0.419
Upper
0.359
1.295
0.697
* No. of fibers/cm .
-------
9.6.
c
o
•H
•U
to
M
4J
C
-------
•H
4J
-------
combinations No. 7 and No. 8 were pooled to calculate the
mean and 90% confidence intervals of outlet concentration of
a baghouse with only one bag. These two combinations are
similar to the combinations of variables tested for a bag-
house with two bags in series. Thus, the mean and 9070 con-
fidence intervals of Phase III combinations No. 7 and No. 8
are used to compare the effects of using one or two bags.
Since stabilization time did not have an effect on out-
let concentration greater than the experimental error, the
outlet concentration of the three stabilization levels were
pooled to measure bag series effect for fibers greater than
1.5 ym. However, only the two estimates of outlet concen-
tration for a stabilization of 24 hours were pooled to com-
pute the geometric mean and 90% confidence limits of outlet
concentration for fibers greater than 5.0 ym. This is the
same stabilization period that was used in Phase III.
Again, the log transforms of outlet concentration are
r
employed to compare the effects of bag series. The geometric
means and their 90% confidence intervals for these tests
are given in Table 40 and plotted in Figure 20. For fibers
greater than 5.0 ym in length, the confidence interval for
the single bag tests overlaps the geometric mean of the
tests using two bags in series. This indicates that for
fibers greater than 5.0 ym, the difference between using
two bags in series or just a single bag is not significantly
greater than the experimental error of the tests.
However, for fibers greater than 1.5 ym in length, the
confidence intervals for both one and two bags do not over-
lap each other's geometric means. Thus, the effect of using
one or two bags is significantly different for fiber lengths
greater than 1.5 ym. The better alternative in this case is
using a single bag rather than two bags in series since the
mean outlet concentration for a single bag was lower than for
the two bag arrangement.
114
-------
Table 40. GEOMETRIC MEAN AND 90% CONFIDENCE LIMITS OF OUTLET
CONCENTRATION FOR ONE AND TWO BAG BAGHOUSES
Number
of Bags
1
2
Outlet Concentration*
Fibers > 1.5 ym
N
4
6
Geometric
Mean
2.200
3.460
Confidence Limits
Lower
1.551
2.618
Upper
3.122
4.573
Fibers > 5.0 ym
N
4
2
Geometric
Mean
.317
.322
Confidence Limits
Lower
.284
.290
Upper
.355
.352
!
No. of fibers/cm .
-------
O
•H
4J
4J
C
0)
O
-u
OJ
r-l
•»->
P
O
I
en
0)
•H
Mi
III
O
0)
6.4- •
3.2 --
1.6--
0.8 ..
0.4--
0.2 ..
Fibers >1.5 ym
Fibers >5.0 ym
Number of Bags
Figure 20. Estimates of geometric mean and their 90% confidence
limits for outlet concentration of asbestos fibers by
stabilization period - Phase IV
116
-------
Assuming that the inlet concentration to the second
baghouse in Phase IV is the same as the outlet concentration
of the single baghouse in Phase III for the same operating
conditions, it would appear that the second baghouse cannot
maintain the filtering dust cake efficiency when challenged
with such low loadings. As stated previously, the filtering
dust cake of the second baghouse may not even be capable of
maintaining the same filtering characteristics when the inlet
loading is very low. However, when the first baghouse is
being stabilized and the inlet loading to the second one
is higher, the series outlet concentration is lower than
that of a single baghouse.
SUMMARY OF THE RESULTS OBTAINED FROM PHASES I THROUGH IV
The four phases of the experimental program have estab-
lished the individual effects of factors that can be mani-
pulated by the users and manufacturers of baghouses employed
to reduce the outlet concentration of asbestos fibers from
enclosed sources. Of the factors listed in Table 6, the
type of bag, air-to-cloth ratio, shake amplitude, and the
shake cycle period and duration jointly, significantly affect
the outlet concentration of asbestos fibers. To reduce the
outlet concentration of asbestos fibers, the levels of these
variables considered in this study should be set as follows:
Bag type - cotton sateen
• Air-to-cloth ratio - 1.22 m3/min/m2 (4.0 cfm/ft )
Amplitude of shake cycle - 3.5 cm(1.378 in.)
• Period and duration of shake cycle - 120 min and
20 sec
These recommendations are based on the results of the individ-
ual effects of the levels of factors considered and degree of
precision of the measurement of outlet concentration attained.
117
-------
SECTION 8
REFERENCES
1. C&E News. P. 8, October 6, 1975.
2. Timbrell, V. (Paper presented at AIHA Conference,
Miami, Florida. June 1974).
3. Strauss, W. Industrial Gas Cleaning. Pergammon Press,
New York, 1966.
4. Durham and Harrington, NAPCA, PHS, USDEW, AICHE, 63rd
Annual Meeting. Chicago, Illinois. November 1970.
5. Werle, D.K. Fabric Filters in Pollution Control-
Fundamentals and Applications. IITRI-C8196-14.
6. Stafford, R. and Ettinger, H.J. Filter Efficiency as
a Function of Particles Size and Velocity. Atmospheric
Environment. £(5):353-362, 1972.
7. Stenhouse, J.I.T. The Behavior of Fibrous Filters in
High Inertia Systems. Filtration and Separation.
£(4):429, 1972.
8. Cooper, D.W. Pentapure Impinger Evaluation. Report
No. EPA-650/2-75-024-a, March 1975.
9. Draemel, D.C. Relationship Between Fabric Structure and
Filtration Performance in Dust Filtration. Report
No. EPA-R2-73-288, July 1973.
10. Dick, G.A. Fabric Filters. Canadian Mining Journal.
October 1970.
11. Spaite, P.W. and Walsh, G.W. Effect of Fabric Structure
on Filter Performance. AIHA Journal. 24_:357, 1963.
12. Billings, C.E. and Wilder, J. Handbook of Fabric Filter
Technology. Vol. 1, Fabric Filter Systems Study, NAPCA.
December 1970.
118
-------
Say
15. Dennis R Collection Efficiency as a Function of
Particle Size Shape, and Density: Theory and Experience
Proceedings: Symposium on the Use of Fabric Filters for
the Control of Submicron Particulates, November 1974.
and °eStreich' D'K' T^ephone communications,
17. Joint AIHA-ACGIH Aerosol Hazards Evaluation Committee
Recommended Procedures for Sampling and Counting
Asbestos Fibers. AIHA Journal. 36(2) : 83-90, ' February
iy / j . J
119
-------
APPENDIX A
METHODS OF ANALYSIS
Statistical Methods
The effects of the independent variables or their
appropriate transformations on the dependent variable, outlet
concentration of asbestos fibers, or its appropriate trans-
formation, have been analyzed in three ways. Correlation
coefficients have been computed between the independent and
dependent variables. Ninety percent confidence intervals
have been constructed about the mean of concentration levels
of particular variables such as bag fabric and type of as-
bestos waste. Regression equations have been developed for
variables in each of the experimental design phases except
Phase IV.
The regression equations were developed by the stepwise
least-squares method. Prior to the development of the actual
regression equations for each phase, a mathematical model
was in each instance formulated expressing the way in which
the relevant independent variables might be functionally
related to the dependent variable. The general form of the
mathematical models employed was a linear expression of the
independent variables or their appropriate transformations.
The terms in each model were candidates for inclusion in the
fitted regression equation.
The general model takes the form
Y = b0X0 + b1X1 + . . . + bNXN + e - Y + e
121
-------
where Y stands for the observed values of the dependent
A
variable or its appropriate transformation, Y stands for the
corresponding values of the dependent variable or its trans-
formation from the expression involving the X's and the b's;
the X's are the values of the independent variables or their
transformations, the b's are the coefficients to be estimated
from the data, and e represents the differences between the
observed and the computed values of the dependent variables
due to residual variation in the observations.
The set of data for each model for each phase was
analyzed by computer (Univac 1108) for the purpose of
selecting the terms (X's) to appear in the equation and
computing the values of the regression coefficients (b's) and
other relevant statistics. The computer program used, a
modification of BMD-02R, performs stepwise multiple
regression -- i.e., the equation is built up, term by term,
by introducing at each step that candidate term which will
result in the greatest reduction in the sum of squared
deviations between the observed values of the dependent
variable and the values computed from the resulting regres-
sion equation. A cutoff point for this process can be set
by the analyst through the choice of a critical "F" value.
The F value associated with the coefficient of a term (b's)
in a regression equation is the square of the ratio of the
coefficient to its standard error. In other words, no
candidate term is introduced into the equation unless the
value of the coefficient of that term is a specified multiple
of its standard error. This excludes from the equation
terms with coefficients of a magnitude that could readily
arise due merely to the inevitable residual variation
between measurements. Thus, no term (X's) is considered to
have a significant effect on the dependent variable unless
the variation resulting from different values of this term is
greater than that resulting from measurement variation.
122
-------
By inserting various values of the terms in the final
equation, the effects on the dependent variable become
readily apparent. For those variables represented by terms
that did not enter into the final equation, their effect
on the dependent variable is equal to or less than the
multiple of the F level and residual variation due to measure-
ment and, thus, are considered to have relatively little
effect on the dependent variable.
The 907o confidence intervals constructed for the parti-
cular dependent variables are based on the t-distribution.
The interval spanned indicates with a 90% probability where
the population mean of the dependent variable might lie.
When an interval overlaps the mean of a different level of
a variable, the population mean of the outlet concentration
of asbestos for each of these two levels are not signifi-
cantly different.
The correlation coefficients computed are standard
Pearson Product correlations. Correlations vary from -1 to
+1 with both -1 and +1 indicating perfect correlation between
two variables and a zero indicating no correlation at all.
The probability that a correlation is significantly dif-
ferent from zero can be computed. As correlations tend to
-1, the relation between two variables tends to be more
inversely related. As correlations tend to +1, the relation
between two variables tends to be more directly related. A
knowledge of the correlations between the independent varia-
bles of this study and the observed concentrations of asbestos
emissions provides a straightforward method for assessing the
type and strength of the effects of the independent variables
but does not provide a functional relation between them.
For a relation between the independent variables and the
emissions of asbestos fibers, a regression equation must be
constructed.
123
-------
From the results of these three types of analyses, the
affects of the ten baghouse variables on the outlet con-
centration of asbestos fibers can be determined.
124
-------
T REPORT NO.
qPA-600/2-76-065
TECHNICAL REPORT DATA
fPfcflftr read lu&iiclions on the reverse before completing)
12.
4. TITLE AND SUBTITLE
Assessment of Particle Control Technology for
Enclosed Asbestos Sources--Phase II
5. REPORT DATE
March 1976
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Paul C. Siebert, Thomas C. Ripley, and
Colin F. Harwood
8. PERFORMING ORG/
9. PERFORMING ORGANIZATION NAME AND ADDRESS
IIT Research Institute
10 West 35th Street
Chicago, Illinois 60616
3. RECIPIENT'S ACCESSION-NO.
REPORT NO.
10. PROGRAM ELEMENT NO.
1AB015; RCAP 21AFA-006
11. CONTRACT/GRANT NO.
68-02-1353
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Pai^k, NC 27711
13. TYPE OF REPORT AND PERIOD COVEMED
Phase II Final: 6/74-6/75
14. SPONSORING AGENCY CODE
EPA-ORD
is. SUPPLEMENTARY NOTES Deport EPA-650/2 - 74-088 was the first report of this series. EPA
project officer for this report is D. K.Oestreich, Mail Drop 62, Ext 2547.
16. ABSTRACT
repO]it gives results of an experimental study to optimize control of
emissions of asbestos fibers using a baghouse. Baghouse operating parameters
found to be statistically significant in reducing asbestos emissions were: bag fabric,
waste type, air-to-cloth ratio, relative humidity, period between shakes and
duration of shaking, and shaking amplitude. Values of these operating parameters
are recommended for industry usage to significantly reduce outlet concentrations of
asbestos. These operating conditions resulted in pressure drops across the fabric
filter that were quite reasonable (= or < 2.0 in. H2O). The most economical
alternatives of cotton sateen bags, high air-to-cloth ratio, and low pressure drop
operating conditions were found to be among the most significant in reducing asbestos
emissions. Among the recommendations are: an air-to-cloth ratio of 1.22 cu m/min/
sq m (4.0 cfm/sq ft), a combination of period between shakes of 120 min with a
shaking duration of 20 sec, and a shaking amplitude of 3. 500 cm.
KEY WORDS AND DOCUMHNT ANALYSIS
DESCRIPTORS
Air Pollution
Asbestos
Fibers
Dust
Dust Collectors
Assessment
Measurement
Air Filters
Fabrics
Filters
b.IDENTIFIERS/OPEN ENDED TERMS
Air Pollution Control
Stationary Sources
Enclosed Sources
Particulate
Baghouses
Fabric Filters
c. COS AT I Held/Group
13B
11E,08G
11G
13A
14B
HE
2. DISTRIBUTION STATEMENT
Unlimited
19. StCURITY CLASS (This Report)
Unclassified
PAGE-S
20. SECURITY CLASS (This page)
Unclassified
22. PHIC1
EPA Form 2220-i (9-73)
125
------- |