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

-------